diff options
author | Schoen <schoepas@deher1m1598.emea.adsint.biz> | 2017-04-03 08:52:05 +0300 |
---|---|---|
committer | Schoen <schoepas@deher1m1598.emea.adsint.biz> | 2017-04-03 09:47:02 +0300 |
commit | 90778901c9ea1e16d5907981e91bceba25ff207d (patch) | |
tree | 00bf16ee403c3ab1b507a5ec90b098a22e050f7c | |
parent | 76eebd9379a5dad519ff01cf215fbe3db6de931e (diff) | |
parent | 3bf0026bec6f839ae0f7c40bde845299ce22e8de (diff) |
Merge branch 'master' into cycles_disney_brdfcycles_disney_brdf
760 files changed, 23549 insertions, 13520 deletions
diff --git a/CMakeLists.txt b/CMakeLists.txt index 092de982b9d..3a91e1b655a 100644 --- a/CMakeLists.txt +++ b/CMakeLists.txt @@ -445,6 +445,7 @@ option(WITH_BOOST "Enable features depending on boost" ON) # Unit testsing option(WITH_GTESTS "Enable GTest unit testing" OFF) +option(WITH_OPENGL_TESTS "Enable OpenGL related unit testing (Experimental)" OFF) # Documentation @@ -518,18 +519,20 @@ endif() option(WITH_LEGACY_DEPSGRAPH "Build Blender with legacy dependency graph" ON) mark_as_advanced(WITH_LEGACY_DEPSGRAPH) -# Use hardcoded paths or find_package to find externals -option(WITH_WINDOWS_FIND_MODULES "Use find_package to locate libraries" OFF) -mark_as_advanced(WITH_WINDOWS_FIND_MODULES) +if(WIN32) + # Use hardcoded paths or find_package to find externals + option(WITH_WINDOWS_FIND_MODULES "Use find_package to locate libraries" OFF) + mark_as_advanced(WITH_WINDOWS_FIND_MODULES) -option(WITH_WINDOWS_CODESIGN "Use signtool to sign the final binary." OFF) -mark_as_advanced(WITH_WINDOWS_CODESIGN) + option(WITH_WINDOWS_CODESIGN "Use signtool to sign the final binary." OFF) + mark_as_advanced(WITH_WINDOWS_CODESIGN) -set(WINDOWS_CODESIGN_PFX CACHE FILEPATH "Path to pfx file to use for codesigning.") -mark_as_advanced(WINDOWS_CODESIGN_PFX) + set(WINDOWS_CODESIGN_PFX CACHE FILEPATH "Path to pfx file to use for codesigning.") + mark_as_advanced(WINDOWS_CODESIGN_PFX) -set(WINDOWS_CODESIGN_PFX_PASSWORD CACHE STRING "password for pfx file used for codesigning.") -mark_as_advanced(WINDOWS_CODESIGN_PFX_PASSWORD) + set(WINDOWS_CODESIGN_PFX_PASSWORD CACHE STRING "password for pfx file used for codesigning.") + mark_as_advanced(WINDOWS_CODESIGN_PFX_PASSWORD) +endif() # avoid using again option_defaults_clear() @@ -924,7 +927,7 @@ if(WITH_X11) if(WITH_X11_ALPHA) find_library(X11_Xrender_LIB Xrender ${X11_LIB_SEARCH_PATH}) mark_as_advanced(X11_Xrender_LIB) - if (X11_Xrender_LIB) + if(X11_Xrender_LIB) list(APPEND PLATFORM_LINKLIBS ${X11_Xrender_LIB}) else() set(WITH_X11_ALPHA OFF) diff --git a/GNUmakefile b/GNUmakefile index 1fda1a25a92..86964e68873 100644 --- a/GNUmakefile +++ b/GNUmakefile @@ -1,4 +1,4 @@ -# -*- mode: gnumakefile; tab-width: 8; indent-tabs-mode: t; -*- +# -*- mode: gnumakefile; tab-width: 4; indent-tabs-mode: t; -*- # vim: tabstop=4 # # ##### BEGIN GPL LICENSE BLOCK ##### @@ -113,7 +113,7 @@ CMAKE_CONFIG = cmake $(BUILD_CMAKE_ARGS) \ # X11 spesific ifdef DISPLAY CMAKE_CONFIG_TOOL = cmake-gui -else +else CMAKE_CONFIG_TOOL = ccmake endif @@ -127,7 +127,7 @@ all: .FORCE # # if test ! -f $(BUILD_DIR)/CMakeCache.txt ; then \ # # $(CMAKE_CONFIG); \ # # fi - + # # do this always incase of failed initial build, could be smarter here... @$(CMAKE_CONFIG) diff --git a/build_files/build_environment/install_deps.sh b/build_files/build_environment/install_deps.sh index ee49f83b38d..ed1b9a3e62a 100755 --- a/build_files/build_environment/install_deps.sh +++ b/build_files/build_environment/install_deps.sh @@ -4252,7 +4252,7 @@ print_info() { PRINT " $_3" _buildargs="$_buildargs $_1 $_2 $_3" if [ -d $INST/osl ]; then - _1="-D CYCLES_OSL=$INST/osl" + _1="-D OSL_ROOT_DIR=$INST/osl" PRINT " $_1" _buildargs="$_buildargs $_1" fi diff --git a/build_files/buildbot/slave_compile.py b/build_files/buildbot/slave_compile.py index 860dd1174cf..c8d69c38644 100644 --- a/build_files/buildbot/slave_compile.py +++ b/build_files/buildbot/slave_compile.py @@ -91,7 +91,6 @@ if 'cmake' in builder: elif builder.startswith('win32'): bits = 32 cmake_options.extend(['-G', 'Visual Studio 12 2013']) - cmake_extra_options.append('-DCUDA_NVCC_EXECUTABLE:FILEPATH=C:/Program Files/NVIDIA GPU Computing Toolkit/CUDA/v8.0/bin/nvcc.exe') elif builder.startswith('linux'): tokens = builder.split("_") @@ -111,8 +110,6 @@ if 'cmake' in builder: cuda_chroot_name = 'buildbot_' + deb_name + '_x86_64' targets = ['player', 'blender', 'cuda'] - cmake_extra_options.append('-DCUDA_NVCC_EXECUTABLE=/usr/local/cuda-8.0/bin/nvcc') - cmake_options.append("-C" + os.path.join(blender_dir, cmake_config_file)) # Prepare CMake options needed to configure cuda binaries compilation. diff --git a/build_files/cmake/buildinfo.cmake b/build_files/cmake/buildinfo.cmake index 1f0a283ba0e..a43b99f5d1e 100644 --- a/build_files/cmake/buildinfo.cmake +++ b/build_files/cmake/buildinfo.cmake @@ -56,7 +56,7 @@ if(EXISTS ${SOURCE_DIR}/.git) string(REGEX REPLACE "[\r\n]+" ";" _git_contains_branches "${_git_contains_branches}") string(REGEX REPLACE ";[ \t]+" ";" _git_contains_branches "${_git_contains_branches}") foreach(_branch ${_git_contains_branches}) - if (NOT "${_branch}" MATCHES "\\(HEAD.*") + if(NOT "${_branch}" MATCHES "\\(HEAD.*") set(MY_WC_BRANCH "${_branch}") break() endif() diff --git a/build_files/cmake/macros.cmake b/build_files/cmake/macros.cmake index 09428953a46..f9992ee92b9 100644 --- a/build_files/cmake/macros.cmake +++ b/build_files/cmake/macros.cmake @@ -1574,24 +1574,24 @@ macro(openmp_delayload endmacro() MACRO(WINDOWS_SIGN_TARGET target) - if (WITH_WINDOWS_CODESIGN) - if (!SIGNTOOL_EXE) + if(WITH_WINDOWS_CODESIGN) + if(!SIGNTOOL_EXE) error("Codesigning is enabled, but signtool is not found") else() - if (WINDOWS_CODESIGN_PFX_PASSWORD) + if(WINDOWS_CODESIGN_PFX_PASSWORD) set(CODESIGNPASSWORD /p ${WINDOWS_CODESIGN_PFX_PASSWORD}) else() - if ($ENV{PFXPASSWORD}) + if($ENV{PFXPASSWORD}) set(CODESIGNPASSWORD /p $ENV{PFXPASSWORD}) else() - message( FATAL_ERROR "WITH_WINDOWS_CODESIGN is on but WINDOWS_CODESIGN_PFX_PASSWORD not set, and environment variable PFXPASSWORD not found, unable to sign code.") + message(FATAL_ERROR "WITH_WINDOWS_CODESIGN is on but WINDOWS_CODESIGN_PFX_PASSWORD not set, and environment variable PFXPASSWORD not found, unable to sign code.") endif() endif() add_custom_command(TARGET ${target} - POST_BUILD - COMMAND ${SIGNTOOL_EXE} sign /f ${WINDOWS_CODESIGN_PFX} ${CODESIGNPASSWORD} $<TARGET_FILE:${target}> - VERBATIM - ) + POST_BUILD + COMMAND ${SIGNTOOL_EXE} sign /f ${WINDOWS_CODESIGN_PFX} ${CODESIGNPASSWORD} $<TARGET_FILE:${target}> + VERBATIM + ) endif() endif() ENDMACRO() diff --git a/build_files/cmake/packaging.cmake b/build_files/cmake/packaging.cmake index c7063ed6772..5b894d96f95 100644 --- a/build_files/cmake/packaging.cmake +++ b/build_files/cmake/packaging.cmake @@ -1,5 +1,7 @@ -set(PROJECT_DESCRIPTION "Blender is a very fast and versatile 3D modeller/renderer.") -set(PROJECT_COPYRIGHT "Copyright (C) 2001-2012 Blender Foundation") +string(TIMESTAMP CURRENT_YEAR "%Y") + +set(PROJECT_DESCRIPTION "Blender is the free and open source 3D creation suite software.") +set(PROJECT_COPYRIGHT "Copyright (C) 2001-${CURRENT_YEAR} Blender Foundation") set(PROJECT_CONTACT "foundation@blender.org") set(PROJECT_VENDOR "Blender Foundation") @@ -38,8 +40,8 @@ unset(MY_WC_HASH) # Force Package Name execute_process(COMMAND date "+%Y%m%d" OUTPUT_VARIABLE CPACK_DATE OUTPUT_STRIP_TRAILING_WHITESPACE) string(TOLOWER ${PROJECT_NAME} PROJECT_NAME_LOWER) -if (MSVC) - if ("${CMAKE_SIZEOF_VOID_P}" EQUAL "8") +if(MSVC) + if("${CMAKE_SIZEOF_VOID_P}" EQUAL "8") set(PACKAGE_ARCH windows64) else() set(PACKAGE_ARCH windows32) @@ -48,7 +50,7 @@ else(MSVC) set(PACKAGE_ARCH ${CMAKE_SYSTEM_PROCESSOR}) endif() -if (CPACK_OVERRIDE_PACKAGENAME) +if(CPACK_OVERRIDE_PACKAGENAME) set(CPACK_PACKAGE_FILE_NAME ${CPACK_OVERRIDE_PACKAGENAME}-${PACKAGE_ARCH}) else() set(CPACK_PACKAGE_FILE_NAME ${PROJECT_NAME_LOWER}-${MAJOR_VERSION}.${MINOR_VERSION}.${PATCH_VERSION}-git${CPACK_DATE}.${BUILD_REV}-${PACKAGE_ARCH}) @@ -135,4 +137,3 @@ unset(MINOR_VERSION) unset(PATCH_VERSION) unset(BUILD_REV) - diff --git a/build_files/cmake/platform/platform_win32_msvc.cmake b/build_files/cmake/platform/platform_win32_msvc.cmake index 45a44596e61..3ea93a96290 100644 --- a/build_files/cmake/platform/platform_win32_msvc.cmake +++ b/build_files/cmake/platform/platform_win32_msvc.cmake @@ -33,7 +33,7 @@ endmacro() macro(windows_find_package package_name ) if(WITH_WINDOWS_FIND_MODULES) - find_package( ${package_name}) + find_package(${package_name}) endif(WITH_WINDOWS_FIND_MODULES) endmacro() diff --git a/doc/python_api/rst/bge.texture.rst b/doc/python_api/rst/bge.texture.rst index 49f6c4469a4..3028ee653f8 100644 --- a/doc/python_api/rst/bge.texture.rst +++ b/doc/python_api/rst/bge.texture.rst @@ -681,7 +681,7 @@ Image classes .. attribute:: zbuff - Use depth component of render as grey scale color - suitable for texture source. + Use depth component of render as grayscale color - suitable for texture source. :type: bool @@ -817,7 +817,7 @@ Image classes .. attribute:: zbuff - Use depth component of viewport as grey scale color - suitable for texture source. + Use depth component of viewport as grayscale color - suitable for texture source. :type: bool @@ -1260,8 +1260,8 @@ Filter classes .. class:: FilterGray - Filter for gray scale effect. - Proportions of R, G and B contributions in the output gray scale are 28:151:77. + Filter for grayscale effect. + Proportions of R, G and B contributions in the output grayscale are 28:151:77. .. attribute:: previous diff --git a/doc/python_api/sphinx_doc_gen.py b/doc/python_api/sphinx_doc_gen.py index ec3131ca19e..47bb323e574 100644 --- a/doc/python_api/sphinx_doc_gen.py +++ b/doc/python_api/sphinx_doc_gen.py @@ -427,9 +427,9 @@ if BLENDER_REVISION != "Unknown": BLENDER_VERSION_DOTS += " " + BLENDER_REVISION # '2.62.1 SHA1' BLENDER_VERSION_PATH = "_".join(blender_version_strings) # '2_62_1' -if bpy.app.version_cycle == "release": - BLENDER_VERSION_PATH = "%s%s_release" % ("_".join(blender_version_strings[:2]), - bpy.app.version_char) # '2_62_release' +if bpy.app.version_cycle in {"rc", "release"}: + # '2_62a_release' + BLENDER_VERSION_PATH = "%s%s_release" % ("_".join(blender_version_strings[:2]), bpy.app.version_char) # --------------------------DOWNLOADABLE FILES---------------------------------- diff --git a/doc/python_api/sphinx_doc_update.py b/doc/python_api/sphinx_doc_update.py index 3d48c1145e1..561e58dec66 100755 --- a/doc/python_api/sphinx_doc_update.py +++ b/doc/python_api/sphinx_doc_update.py @@ -96,6 +96,11 @@ def main(): rsync_base = "rsync://%s@%s:%s" % (args.user, args.rsync_server, args.rsync_root) + blenver = blenver_zip = "" + api_name = "" + branch = "" + is_release = False + # I) Update local mirror using rsync. rsync_mirror_cmd = ("rsync", "--delete-after", "-avzz", rsync_base, args.mirror_dir) subprocess.run(rsync_mirror_cmd, env=dict(os.environ, RSYNC_PASSWORD=args.password)) @@ -108,19 +113,24 @@ def main(): subprocess.run(doc_gen_cmd) # III) Get Blender version info. - blenver = blenver_zip = "" getver_file = os.path.join(tmp_dir, "blendver.txt") getver_script = ("" "import sys, bpy\n" "with open(sys.argv[-1], 'w') as f:\n" - " f.write('%d_%d%s_release\\n' % (bpy.app.version[0], bpy.app.version[1], bpy.app.version_char)\n" - " if bpy.app.version_cycle in {'rc', 'release'} else '%d_%d_%d\\n' % bpy.app.version)\n" - " f.write('%d_%d_%d' % bpy.app.version)\n") + " is_release = bpy.app.version_cycle in {'rc', 'release'}\n" + " branch = bpy.app.build_branch.split()[0].decode()\n" + " f.write('%d\\n' % is_release)\n" + " f.write('%s\\n' % branch)\n" + " f.write('%d.%d%s\\n' % (bpy.app.version[0], bpy.app.version[1], bpy.app.version_char)\n" + " if is_release else '%s\\n' % branch)\n" + " f.write('%d_%d%s_release' % (bpy.app.version[0], bpy.app.version[1], bpy.app.version_char)\n" + " if is_release else '%d_%d_%d' % bpy.app.version)\n") get_ver_cmd = (args.blender, "--background", "-noaudio", "--factory-startup", "--python-exit-code", "1", "--python-expr", getver_script, "--", getver_file) subprocess.run(get_ver_cmd) with open(getver_file) as f: - blenver, blenver_zip = f.read().split("\n") + is_release, branch, blenver, blenver_zip = f.read().split("\n") + is_release = bool(int(is_release)) os.remove(getver_file) # IV) Build doc. @@ -132,7 +142,7 @@ def main(): os.chdir(curr_dir) # V) Cleanup existing matching dir in server mirror (if any), and copy new doc. - api_name = "blender_python_api_%s" % blenver + api_name = blenver api_dir = os.path.join(args.mirror_dir, api_name) if os.path.exists(api_dir): shutil.rmtree(api_dir) @@ -150,19 +160,15 @@ def main(): os.rename(zip_path, os.path.join(api_dir, "%s.zip" % zip_name)) # VII) Create symlinks and html redirects. - #~ os.symlink(os.path.join(DEFAULT_SYMLINK_ROOT, api_name, "contents.html"), os.path.join(api_dir, "index.html")) os.symlink("./contents.html", os.path.join(api_dir, "index.html")) - if blenver.endswith("release"): - symlink = os.path.join(args.mirror_dir, "blender_python_api_current") + if is_release: + symlink = os.path.join(args.mirror_dir, "current") os.remove(symlink) os.symlink("./%s" % api_name, symlink) with open(os.path.join(args.mirror_dir, "250PythonDoc/index.html"), 'w') as f: f.write("<html><head><title>Redirecting...</title><meta http-equiv=\"REFRESH\"" "content=\"0;url=../%s/\"></head><body>Redirecting...</body></html>" % api_name) - else: - symlink = os.path.join(args.mirror_dir, "blender_python_api_master") - os.remove(symlink) - os.symlink("./%s" % api_name, symlink) + elif branch == "master": with open(os.path.join(args.mirror_dir, "blender_python_api/index.html"), 'w') as f: f.write("<html><head><title>Redirecting...</title><meta http-equiv=\"REFRESH\"" "content=\"0;url=../%s/\"></head><body>Redirecting...</body></html>" % api_name) diff --git a/extern/clew/README.blender b/extern/clew/README.blender index 8355a96db8e..127b6bdac63 100644 --- a/extern/clew/README.blender +++ b/extern/clew/README.blender @@ -1,5 +1,5 @@ Project: OpenCL Wrangler URL: https://github.com/OpenCLWrangler/clew License: Apache 2.0 -Upstream version: 309a653 +Upstream version: 27a6867 Local modifications: None diff --git a/extern/clew/include/clew.h b/extern/clew/include/clew.h index 2a583c81599..0bbb95b9839 100644 --- a/extern/clew/include/clew.h +++ b/extern/clew/include/clew.h @@ -369,7 +369,7 @@ typedef unsigned int cl_GLenum; #endif /* Define basic vector types */ -/* WOrkaround for ppc64el platform: conflicts with bool from C++. */ +/* Workaround for ppc64el platform: conflicts with bool from C++. */ #if defined( __VEC__ ) && !(defined(__PPC64__) && defined(__LITTLE_ENDIAN__)) #include <altivec.h> /* may be omitted depending on compiler. AltiVec spec provides no way to detect whether the header is required. */ typedef vector unsigned char __cl_uchar16; @@ -2765,11 +2765,40 @@ CLEW_FUN_EXPORT PFNCLGETGLCONTEXTINFOKHR __clewGetGLContextInfoKH #define CL_DEVICE_GPU_OVERLAP_NV 0x4004 #define CL_DEVICE_KERNEL_EXEC_TIMEOUT_NV 0x4005 #define CL_DEVICE_INTEGRATED_MEMORY_NV 0x4006 +#define CL_DEVICE_ATTRIBUTE_ASYNC_ENGINE_COUNT_NV 0x4007 +#define CL_DEVICE_PCI_BUS_ID_NV 0x4008 +#define CL_DEVICE_PCI_SLOT_ID_NV 0x4009 /********************************* * cl_amd_device_attribute_query * *********************************/ #define CL_DEVICE_PROFILING_TIMER_OFFSET_AMD 0x4036 +#define CL_DEVICE_TOPOLOGY_AMD 0x4037 +#define CL_DEVICE_BOARD_NAME_AMD 0x4038 +#define CL_DEVICE_GLOBAL_FREE_MEMORY_AMD 0x4039 +#define CL_DEVICE_SIMD_PER_COMPUTE_UNIT_AMD 0x4040 +#define CL_DEVICE_SIMD_WIDTH_AMD 0x4041 +#define CL_DEVICE_SIMD_INSTRUCTION_WIDTH_AMD 0x4042 +#define CL_DEVICE_WAVEFRONT_WIDTH_AMD 0x4043 +#define CL_DEVICE_GLOBAL_MEM_CHANNELS_AMD 0x4044 +#define CL_DEVICE_GLOBAL_MEM_CHANNEL_BANKS_AMD 0x4045 +#define CL_DEVICE_GLOBAL_MEM_CHANNEL_BANK_WIDTH_AMD 0x4046 +#define CL_DEVICE_LOCAL_MEM_SIZE_PER_COMPUTE_UNIT_AMD 0x4047 +#define CL_DEVICE_LOCAL_MEM_BANKS_AMD 0x4048 +#define CL_DEVICE_THREAD_TRACE_SUPPORTED_AMD 0x4049 +#define CL_DEVICE_GFXIP_MAJOR_AMD 0x404A +#define CL_DEVICE_GFXIP_MINOR_AMD 0x404B +#define CL_DEVICE_AVAILABLE_ASYNC_QUEUES_AMD 0x404C + +#ifndef CL_DEVICE_TOPOLOGY_TYPE_PCIE_AMD +#define CL_DEVICE_TOPOLOGY_TYPE_PCIE_AMD 1 + +typedef union +{ + struct { cl_uint type; cl_uint data[5]; } raw; + struct { cl_uint type; cl_char unused[17]; cl_char bus; cl_char device; cl_char function; } pcie; +} cl_device_topology_amd; +#endif /********************************* * cl_arm_printf extension diff --git a/extern/clew/src/clew.c b/extern/clew/src/clew.c index 3717c74f1a0..dbdddfc0332 100644 --- a/extern/clew/src/clew.c +++ b/extern/clew/src/clew.c @@ -15,7 +15,7 @@ typedef HMODULE CLEW_DYNLIB_HANDLE; - #define CLEW_DYNLIB_OPEN LoadLibrary + #define CLEW_DYNLIB_OPEN LoadLibraryA #define CLEW_DYNLIB_CLOSE FreeLibrary #define CLEW_DYNLIB_IMPORT GetProcAddress #else @@ -223,7 +223,7 @@ int clewInit() __clewSetCommandQueueProperty = (PFNCLSETCOMMANDQUEUEPROPERTY )CLEW_DYNLIB_IMPORT(module, "clSetCommandQueueProperty"); #endif __clewCreateBuffer = (PFNCLCREATEBUFFER )CLEW_DYNLIB_IMPORT(module, "clCreateBuffer"); - __clewCreateSubBuffer = (PFNCLCREATESUBBUFFER )CLEW_DYNLIB_IMPORT(module, "clCreateBuffer"); + __clewCreateSubBuffer = (PFNCLCREATESUBBUFFER )CLEW_DYNLIB_IMPORT(module, "clCreateSubBuffer"); __clewCreateImage = (PFNCLCREATEIMAGE )CLEW_DYNLIB_IMPORT(module, "clCreateImage"); __clewRetainMemObject = (PFNCLRETAINMEMOBJECT )CLEW_DYNLIB_IMPORT(module, "clRetainMemObject"); __clewReleaseMemObject = (PFNCLRELEASEMEMOBJECT )CLEW_DYNLIB_IMPORT(module, "clReleaseMemObject"); diff --git a/extern/cuew/include/cuew.h b/extern/cuew/include/cuew.h index 19087117667..4cce29d38ab 100644 --- a/extern/cuew/include/cuew.h +++ b/extern/cuew/include/cuew.h @@ -114,7 +114,7 @@ extern "C" { #define cuGLGetDevices cuGLGetDevices_v2 /* Types. */ -#if defined(__x86_64) || defined(AMD64) || defined(_M_AMD64) +#if defined(__x86_64) || defined(AMD64) || defined(_M_AMD64) || defined (__aarch64__) typedef unsigned long long CUdeviceptr; #else typedef unsigned int CUdeviceptr; diff --git a/intern/CMakeLists.txt b/intern/CMakeLists.txt index 9a5476772ab..bfe230250ae 100644 --- a/intern/CMakeLists.txt +++ b/intern/CMakeLists.txt @@ -34,7 +34,7 @@ add_subdirectory(mikktspace) add_subdirectory(glew-mx) add_subdirectory(eigen) -if (WITH_GAMEENGINE_DECKLINK) +if(WITH_GAMEENGINE_DECKLINK) add_subdirectory(decklink) endif() @@ -62,7 +62,7 @@ if(WITH_IK_ITASC) add_subdirectory(itasc) endif() -if(WITH_IK_SOLVER OR WITH_GAMEENGINE OR WITH_MOD_BOOLEAN) +if(WITH_GAMEENGINE) add_subdirectory(moto) endif() diff --git a/intern/atomic/atomic_ops.h b/intern/atomic/atomic_ops.h index 1107deddf94..1e9528f9ed9 100644 --- a/intern/atomic/atomic_ops.h +++ b/intern/atomic/atomic_ops.h @@ -101,11 +101,11 @@ ATOMIC_INLINE size_t atomic_fetch_and_add_z(size_t *p, size_t x); ATOMIC_INLINE size_t atomic_fetch_and_sub_z(size_t *p, size_t x); ATOMIC_INLINE size_t atomic_cas_z(size_t *v, size_t old, size_t _new); -ATOMIC_INLINE unsigned atomic_add_and_fetch_u(unsigned *p, unsigned x); -ATOMIC_INLINE unsigned atomic_sub_and_fetch_u(unsigned *p, unsigned x); -ATOMIC_INLINE unsigned atomic_fetch_and_add_u(unsigned *p, unsigned x); -ATOMIC_INLINE unsigned atomic_fetch_and_sub_u(unsigned *p, unsigned x); -ATOMIC_INLINE unsigned atomic_cas_u(unsigned *v, unsigned old, unsigned _new); +ATOMIC_INLINE unsigned int atomic_add_and_fetch_u(unsigned int *p, unsigned int x); +ATOMIC_INLINE unsigned int atomic_sub_and_fetch_u(unsigned int *p, unsigned int x); +ATOMIC_INLINE unsigned int atomic_fetch_and_add_u(unsigned int *p, unsigned int x); +ATOMIC_INLINE unsigned int atomic_fetch_and_sub_u(unsigned int *p, unsigned int x); +ATOMIC_INLINE unsigned int atomic_cas_u(unsigned int *v, unsigned int old, unsigned int _new); /* WARNING! Float 'atomics' are really faked ones, those are actually closer to some kind of spinlock-sync'ed operation, * which means they are only efficient if collisions are highly unlikely (i.e. if probability of two threads diff --git a/intern/atomic/intern/atomic_ops_ext.h b/intern/atomic/intern/atomic_ops_ext.h index 8421aa72192..b72c94563fc 100644 --- a/intern/atomic/intern/atomic_ops_ext.h +++ b/intern/atomic/intern/atomic_ops_ext.h @@ -113,58 +113,58 @@ ATOMIC_INLINE size_t atomic_cas_z(size_t *v, size_t old, size_t _new) /******************************************************************************/ /* unsigned operations. */ -ATOMIC_INLINE unsigned atomic_add_and_fetch_u(unsigned *p, unsigned x) +ATOMIC_INLINE unsigned int atomic_add_and_fetch_u(unsigned int *p, unsigned int x) { - assert(sizeof(unsigned) == LG_SIZEOF_INT); + assert(sizeof(unsigned int) == LG_SIZEOF_INT); #if (LG_SIZEOF_INT == 8) - return (unsigned)atomic_add_and_fetch_uint64((uint64_t *)p, (uint64_t)x); + return (unsigned int)atomic_add_and_fetch_uint64((uint64_t *)p, (uint64_t)x); #elif (LG_SIZEOF_INT == 4) - return (unsigned)atomic_add_and_fetch_uint32((uint32_t *)p, (uint32_t)x); + return (unsigned int)atomic_add_and_fetch_uint32((uint32_t *)p, (uint32_t)x); #endif } -ATOMIC_INLINE unsigned atomic_sub_and_fetch_u(unsigned *p, unsigned x) +ATOMIC_INLINE unsigned int atomic_sub_and_fetch_u(unsigned int *p, unsigned int x) { - assert(sizeof(unsigned) == LG_SIZEOF_INT); + assert(sizeof(unsigned int) == LG_SIZEOF_INT); #if (LG_SIZEOF_INT == 8) - return (unsigned)atomic_add_and_fetch_uint64((uint64_t *)p, (uint64_t)-((int64_t)x)); + return (unsigned int)atomic_add_and_fetch_uint64((uint64_t *)p, (uint64_t)-((int64_t)x)); #elif (LG_SIZEOF_INT == 4) - return (unsigned)atomic_add_and_fetch_uint32((uint32_t *)p, (uint32_t)-((int32_t)x)); + return (unsigned int)atomic_add_and_fetch_uint32((uint32_t *)p, (uint32_t)-((int32_t)x)); #endif } -ATOMIC_INLINE unsigned atomic_fetch_and_add_u(unsigned *p, unsigned x) +ATOMIC_INLINE unsigned int atomic_fetch_and_add_u(unsigned int *p, unsigned int x) { - assert(sizeof(unsigned) == LG_SIZEOF_INT); + assert(sizeof(unsigned int) == LG_SIZEOF_INT); #if (LG_SIZEOF_INT == 8) - return (unsigned)atomic_fetch_and_add_uint64((uint64_t *)p, (uint64_t)x); + return (unsigned int)atomic_fetch_and_add_uint64((uint64_t *)p, (uint64_t)x); #elif (LG_SIZEOF_INT == 4) - return (unsigned)atomic_fetch_and_add_uint32((uint32_t *)p, (uint32_t)x); + return (unsigned int)atomic_fetch_and_add_uint32((uint32_t *)p, (uint32_t)x); #endif } -ATOMIC_INLINE unsigned atomic_fetch_and_sub_u(unsigned *p, unsigned x) +ATOMIC_INLINE unsigned int atomic_fetch_and_sub_u(unsigned int *p, unsigned int x) { - assert(sizeof(unsigned) == LG_SIZEOF_INT); + assert(sizeof(unsigned int) == LG_SIZEOF_INT); #if (LG_SIZEOF_INT == 8) - return (unsigned)atomic_fetch_and_add_uint64((uint64_t *)p, (uint64_t)-((int64_t)x)); + return (unsigned int)atomic_fetch_and_add_uint64((uint64_t *)p, (uint64_t)-((int64_t)x)); #elif (LG_SIZEOF_INT == 4) - return (unsigned)atomic_fetch_and_add_uint32((uint32_t *)p, (uint32_t)-((int32_t)x)); + return (unsigned int)atomic_fetch_and_add_uint32((uint32_t *)p, (uint32_t)-((int32_t)x)); #endif } -ATOMIC_INLINE unsigned atomic_cas_u(unsigned *v, unsigned old, unsigned _new) +ATOMIC_INLINE unsigned int atomic_cas_u(unsigned int *v, unsigned int old, unsigned int _new) { - assert(sizeof(unsigned) == LG_SIZEOF_INT); + assert(sizeof(unsigned int) == LG_SIZEOF_INT); #if (LG_SIZEOF_INT == 8) - return (unsigned)atomic_cas_uint64((uint64_t *)v, (uint64_t)old, (uint64_t)_new); + return (unsigned int)atomic_cas_uint64((uint64_t *)v, (uint64_t)old, (uint64_t)_new); #elif (LG_SIZEOF_INT == 4) - return (unsigned)atomic_cas_uint32((uint32_t *)v, (uint32_t)old, (uint32_t)_new); + return (unsigned int)atomic_cas_uint32((uint32_t *)v, (uint32_t)old, (uint32_t)_new); #endif } diff --git a/intern/audaspace/intern/AUD_SoftwareDevice.cpp b/intern/audaspace/intern/AUD_SoftwareDevice.cpp index 15594d340be..f9d65aa2363 100644 --- a/intern/audaspace/intern/AUD_SoftwareDevice.cpp +++ b/intern/audaspace/intern/AUD_SoftwareDevice.cpp @@ -365,6 +365,7 @@ bool AUD_SoftwareDevice::AUD_SoftwareHandle::seek(float position) if(!m_status) return false; + m_pitch->setPitch(m_user_pitch); m_reader->seek((int)(position * m_reader->getSpecs().rate)); if(m_status == AUD_STATUS_STOPPED) diff --git a/intern/cycles/CMakeLists.txt b/intern/cycles/CMakeLists.txt index 79c1c3e3e82..806a8660e8c 100644 --- a/intern/cycles/CMakeLists.txt +++ b/intern/cycles/CMakeLists.txt @@ -22,6 +22,7 @@ if(WITH_CYCLES_NATIVE_ONLY) -DWITH_KERNEL_NATIVE ) set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -march=native") + set(CYCLES_KERNEL_FLAGS "-march=native") elseif(NOT WITH_CPU_SSE) set(CXX_HAS_SSE FALSE) set(CXX_HAS_AVX FALSE) @@ -59,10 +60,13 @@ elseif(WIN32 AND MSVC) set(CMAKE_CXX_FLAGS_RELEASE "${CMAKE_CXX_FLAGS_RELEASE} /Ox") set(CMAKE_CXX_FLAGS_RELWITHDEBINFO "${CMAKE_CXX_FLAGS_RELWITHDEBINFO} /Ox") set(CMAKE_CXX_FLAGS_MINSIZEREL "${CMAKE_CXX_FLAGS_MINSIZEREL} /Ox") + + set(CYCLES_KERNEL_FLAGS "/fp:fast -D_CRT_SECURE_NO_WARNINGS /GS-") elseif(CMAKE_COMPILER_IS_GNUCC) check_cxx_compiler_flag(-msse CXX_HAS_SSE) check_cxx_compiler_flag(-mavx CXX_HAS_AVX) check_cxx_compiler_flag(-mavx2 CXX_HAS_AVX2) + set(CYCLES_KERNEL_FLAGS "-ffast-math") if(CXX_HAS_SSE) set(CYCLES_SSE2_KERNEL_FLAGS "-ffast-math -msse -msse2 -mfpmath=sse") set(CYCLES_SSE3_KERNEL_FLAGS "-ffast-math -msse -msse2 -msse3 -mssse3 -mfpmath=sse") @@ -74,10 +78,12 @@ elseif(CMAKE_COMPILER_IS_GNUCC) if(CXX_HAS_AVX2) set(CYCLES_AVX2_KERNEL_FLAGS "-ffast-math -msse -msse2 -msse3 -mssse3 -msse4.1 -mavx -mavx2 -mfma -mlzcnt -mbmi -mbmi2 -mf16c -mfpmath=sse") endif() + set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -ffast-math -fno-finite-math-only") elseif(CMAKE_CXX_COMPILER_ID MATCHES "Clang") check_cxx_compiler_flag(-msse CXX_HAS_SSE) check_cxx_compiler_flag(-mavx CXX_HAS_AVX) check_cxx_compiler_flag(-mavx2 CXX_HAS_AVX2) + set(CYCLES_KERNEL_FLAGS "-ffast-math") if(CXX_HAS_SSE) set(CYCLES_SSE2_KERNEL_FLAGS "-ffast-math -msse -msse2") set(CYCLES_SSE3_KERNEL_FLAGS "-ffast-math -msse -msse2 -msse3 -mssse3") @@ -89,6 +95,7 @@ elseif(CMAKE_CXX_COMPILER_ID MATCHES "Clang") if(CXX_HAS_AVX2) set(CYCLES_AVX2_KERNEL_FLAGS "-ffast-math -msse -msse2 -msse3 -mssse3 -msse4.1 -mavx -mavx2 -mfma -mlzcnt -mbmi -mbmi2 -mf16c") endif() + set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -ffast-math -fno-finite-math-only") endif() if(CXX_HAS_SSE) diff --git a/intern/cycles/app/CMakeLists.txt b/intern/cycles/app/CMakeLists.txt index 8cd499b7ca6..aabb8f63640 100644 --- a/intern/cycles/app/CMakeLists.txt +++ b/intern/cycles/app/CMakeLists.txt @@ -1,14 +1,6 @@ set(INC - . - ../bvh - ../device - ../graph - ../kernel - ../kernel/svm - ../render - ../subd - ../util + .. ) set(INC_SYS ) diff --git a/intern/cycles/app/cycles_server.cpp b/intern/cycles/app/cycles_server.cpp index 4ef9cd070bb..e65b9d769e4 100644 --- a/intern/cycles/app/cycles_server.cpp +++ b/intern/cycles/app/cycles_server.cpp @@ -16,15 +16,15 @@ #include <stdio.h> -#include "device.h" - -#include "util_args.h" -#include "util_foreach.h" -#include "util_path.h" -#include "util_stats.h" -#include "util_string.h" -#include "util_task.h" -#include "util_logging.h" +#include "device/device.h" + +#include "util/util_args.h" +#include "util/util_foreach.h" +#include "util/util_path.h" +#include "util/util_stats.h" +#include "util/util_string.h" +#include "util/util_task.h" +#include "util/util_logging.h" using namespace ccl; diff --git a/intern/cycles/app/cycles_standalone.cpp b/intern/cycles/app/cycles_standalone.cpp index 9816d614a7c..0cd249f0d84 100644 --- a/intern/cycles/app/cycles_standalone.cpp +++ b/intern/cycles/app/cycles_standalone.cpp @@ -16,29 +16,29 @@ #include <stdio.h> -#include "buffers.h" -#include "camera.h" -#include "device.h" -#include "scene.h" -#include "session.h" -#include "integrator.h" - -#include "util_args.h" -#include "util_foreach.h" -#include "util_function.h" -#include "util_logging.h" -#include "util_path.h" -#include "util_progress.h" -#include "util_string.h" -#include "util_time.h" -#include "util_transform.h" -#include "util_version.h" +#include "render/buffers.h" +#include "render/camera.h" +#include "device/device.h" +#include "render/scene.h" +#include "render/session.h" +#include "render/integrator.h" + +#include "util/util_args.h" +#include "util/util_foreach.h" +#include "util/util_function.h" +#include "util/util_logging.h" +#include "util/util_path.h" +#include "util/util_progress.h" +#include "util/util_string.h" +#include "util/util_time.h" +#include "util/util_transform.h" +#include "util/util_version.h" #ifdef WITH_CYCLES_STANDALONE_GUI -#include "util_view.h" +#include "util/util_view.h" #endif -#include "cycles_xml.h" +#include "app/cycles_xml.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/app/cycles_xml.cpp b/intern/cycles/app/cycles_xml.cpp index 35a30ae683f..04f00ef0e10 100644 --- a/intern/cycles/app/cycles_xml.cpp +++ b/intern/cycles/app/cycles_xml.cpp @@ -20,31 +20,31 @@ #include <algorithm> #include <iterator> -#include "node_xml.h" - -#include "background.h" -#include "camera.h" -#include "film.h" -#include "graph.h" -#include "integrator.h" -#include "light.h" -#include "mesh.h" -#include "nodes.h" -#include "object.h" -#include "osl.h" -#include "shader.h" -#include "scene.h" - -#include "subd_patch.h" -#include "subd_split.h" - -#include "util_debug.h" -#include "util_foreach.h" -#include "util_path.h" -#include "util_transform.h" -#include "util_xml.h" - -#include "cycles_xml.h" +#include "graph/node_xml.h" + +#include "render/background.h" +#include "render/camera.h" +#include "render/film.h" +#include "render/graph.h" +#include "render/integrator.h" +#include "render/light.h" +#include "render/mesh.h" +#include "render/nodes.h" +#include "render/object.h" +#include "render/osl.h" +#include "render/shader.h" +#include "render/scene.h" + +#include "subd/subd_patch.h" +#include "subd/subd_split.h" + +#include "util/util_debug.h" +#include "util/util_foreach.h" +#include "util/util_path.h" +#include "util/util_transform.h" +#include "util/util_xml.h" + +#include "app/cycles_xml.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/blender/CMakeLists.txt b/intern/cycles/blender/CMakeLists.txt index b57502b3b14..ae4977aaed0 100644 --- a/intern/cycles/blender/CMakeLists.txt +++ b/intern/cycles/blender/CMakeLists.txt @@ -1,12 +1,6 @@ set(INC - ../graph - ../render - ../device - ../kernel - ../kernel/svm - ../util - ../subd + .. ../../glew-mx ../../guardedalloc ../../mikktspace diff --git a/intern/cycles/blender/addon/__init__.py b/intern/cycles/blender/addon/__init__.py index 235d19e91e8..eb792af7264 100644 --- a/intern/cycles/blender/addon/__init__.py +++ b/intern/cycles/blender/addon/__init__.py @@ -107,7 +107,13 @@ def engine_exit(): engine.exit() +classes = ( + CyclesRender, +) + + def register(): + from bpy.utils import register_class from . import ui from . import properties from . import presets @@ -122,12 +128,15 @@ def register(): properties.register() ui.register() presets.register() - bpy.utils.register_module(__name__) + + for cls in classes: + register_class(cls) bpy.app.handlers.version_update.append(version_update.do_versions) def unregister(): + from bpy.utils import unregister_class from . import ui from . import properties from . import presets @@ -138,4 +147,6 @@ def unregister(): ui.unregister() properties.unregister() presets.unregister() - bpy.utils.unregister_module(__name__) + + for cls in classes: + unregister_class(cls) diff --git a/intern/cycles/blender/addon/engine.py b/intern/cycles/blender/addon/engine.py index c8c9ef58c52..ab57dd44bdb 100644 --- a/intern/cycles/blender/addon/engine.py +++ b/intern/cycles/blender/addon/engine.py @@ -50,6 +50,24 @@ def _workaround_buggy_drivers(): _cycles.opencl_disable() +def _configure_argument_parser(): + import argparse + parser = argparse.ArgumentParser(description="Cycles Addon argument parser") + parser.add_argument("--cycles-resumable-num-chunks", + help="Number of chunks to split sample range into", + default=None) + parser.add_argument("--cycles-resumable-current-chunk", + help="Current chunk of samples range to render", + default=None) + parser.add_argument("--cycles-resumable-start-chunk", + help="Start chunk to render", + default=None) + parser.add_argument("--cycles-resumable-end-chunk", + help="End chunk to render", + default=None) + return parser + + def _parse_command_line(): import sys @@ -57,25 +75,22 @@ def _parse_command_line(): if "--" not in argv: return - argv = argv[argv.index("--") + 1:] - - num_resumable_chunks = None - current_resumable_chunk = None - - # TODO(sergey): Add some nice error prints if argument is not used properly. - idx = 0 - while idx < len(argv) - 1: - arg = argv[idx] - if arg == '--cycles-resumable-num-chunks': - num_resumable_chunks = int(argv[idx + 1]) - elif arg == '--cycles-resumable-current-chunk': - current_resumable_chunk = int(argv[idx + 1]) - idx += 1 + parser = _configure_argument_parser() + args, unknown = parser.parse_known_args(argv[argv.index("--") + 1:]) - if num_resumable_chunks is not None and current_resumable_chunk is not None: - import _cycles - _cycles.set_resumable_chunks(num_resumable_chunks, - current_resumable_chunk) + if args.cycles_resumable_num_chunks is not None: + if args.cycles_resumable_current_chunk is not None: + import _cycles + _cycles.set_resumable_chunk( + int(args.cycles_resumable_num_chunks), + int(args.cycles_resumable_current_chunk)) + elif args.cycles_resumable_start_chunk is not None and \ + args.cycles_resumable_end_chunk: + import _cycles + _cycles.set_resumable_chunk_range( + int(args.cycles_resumable_num_chunks), + int(args.cycles_resumable_start_chunk), + int(args.cycles_resumable_end_chunk)) def init(): diff --git a/intern/cycles/blender/addon/presets.py b/intern/cycles/blender/addon/presets.py index f97b51b629d..82c4ffc6e50 100644 --- a/intern/cycles/blender/addon/presets.py +++ b/intern/cycles/blender/addon/presets.py @@ -82,12 +82,23 @@ class AddPresetSampling(AddPresetBase, Operator): preset_subdir = "cycles/sampling" +classes = ( + AddPresetIntegrator, + AddPresetSampling, +) + + def register(): - pass + from bpy.utils import register_class + for cls in classes: + register_class(cls) def unregister(): - pass + from bpy.utils import unregister_class + for cls in classes: + unregister_class(cls) + if __name__ == "__main__": register() diff --git a/intern/cycles/blender/addon/properties.py b/intern/cycles/blender/addon/properties.py index 5c51f9afc28..cbf469b3a89 100644 --- a/intern/cycles/blender/addon/properties.py +++ b/intern/cycles/blender/addon/properties.py @@ -665,8 +665,10 @@ class CyclesRenderSettings(bpy.types.PropertyGroup): cls.debug_use_cpu_sse3 = BoolProperty(name="SSE3", default=True) cls.debug_use_cpu_sse2 = BoolProperty(name="SSE2", default=True) cls.debug_use_qbvh = BoolProperty(name="QBVH", default=True) + cls.debug_use_cpu_split_kernel = BoolProperty(name="Split Kernel", default=False) cls.debug_use_cuda_adaptive_compile = BoolProperty(name="Adaptive Compile", default=False) + cls.debug_use_cuda_split_kernel = BoolProperty(name="Split Kernel", default=False) cls.debug_opencl_kernel_type = EnumProperty( name="OpenCL Kernel Type", @@ -693,6 +695,8 @@ class CyclesRenderSettings(bpy.types.PropertyGroup): update=devices_update_callback ) + cls.debug_opencl_kernel_single_program = BoolProperty(name="Single Program", default=False, update=devices_update_callback); + cls.debug_use_opencl_debug = BoolProperty(name="Debug OpenCL", default=False) @classmethod @@ -1092,6 +1096,12 @@ class CyclesObjectSettings(bpy.types.PropertyGroup): default=1.0, ) + cls.is_shadow_catcher = BoolProperty( + name="Shadow Catcher", + description="Only render shadows on this object, for compositing renders into real footage", + default=False, + ) + @classmethod def unregister(cls): del bpy.types.Object.cycles diff --git a/intern/cycles/blender/addon/ui.py b/intern/cycles/blender/addon/ui.py index 44af5f7efed..2b50d272be8 100644 --- a/intern/cycles/blender/addon/ui.py +++ b/intern/cycles/blender/addon/ui.py @@ -86,12 +86,10 @@ def use_sample_all_lights(context): return cscene.sample_all_lights_direct or cscene.sample_all_lights_indirect -def show_device_selection(context): - type = get_device_type(context) - if type == 'NETWORK': +def show_device_active(context): + cscene = context.scene.cycles + if cscene.device != 'GPU': return True - if not type in {'CUDA', 'OPENCL'}: - return False return context.user_preferences.addons[__package__].preferences.has_active_device() @@ -186,9 +184,6 @@ class CyclesRender_PT_sampling(CyclesButtonsPanel, Panel): sub.label(text="AA Samples:") sub.prop(cscene, "aa_samples", text="Render") sub.prop(cscene, "preview_aa_samples", text="Preview") - sub.separator() - sub.prop(cscene, "sample_all_lights_direct") - sub.prop(cscene, "sample_all_lights_indirect") col = split.column() sub = col.column(align=True) @@ -205,6 +200,10 @@ class CyclesRender_PT_sampling(CyclesButtonsPanel, Panel): sub.prop(cscene, "subsurface_samples", text="Subsurface") sub.prop(cscene, "volume_samples", text="Volume") + col = layout.column(align=True) + col.prop(cscene, "sample_all_lights_direct") + col.prop(cscene, "sample_all_lights_indirect") + if not (use_opencl(context) and cscene.feature_set != 'EXPERIMENTAL'): layout.row().prop(cscene, "sampling_pattern", text="Pattern") @@ -270,7 +269,7 @@ class CyclesRender_PT_geometry(CyclesButtonsPanel, Panel): row = col.row() row.prop(ccscene, "minimum_width", text="Min Pixels") - row.prop(ccscene, "maximum_width", text="Max Ext.") + row.prop(ccscene, "maximum_width", text="Max Extension") class CyclesRender_PT_light_paths(CyclesButtonsPanel, Panel): @@ -788,6 +787,8 @@ class CyclesObject_PT_cycles_settings(CyclesButtonsPanel, Panel): if ob.type != 'LAMP': flow.prop(visibility, "shadow") + layout.prop(cob, "is_shadow_catcher") + col = layout.column() col.label(text="Performance:") row = col.row() @@ -1518,15 +1519,18 @@ class CyclesRender_PT_debug(CyclesButtonsPanel, Panel): row.prop(cscene, "debug_use_cpu_avx", toggle=True) row.prop(cscene, "debug_use_cpu_avx2", toggle=True) col.prop(cscene, "debug_use_qbvh") + col.prop(cscene, "debug_use_cpu_split_kernel") col = layout.column() col.label('CUDA Flags:') col.prop(cscene, "debug_use_cuda_adaptive_compile") + col.prop(cscene, "debug_use_cuda_split_kernel") col = layout.column() col.label('OpenCL Flags:') col.prop(cscene, "debug_opencl_kernel_type", text="Kernel") col.prop(cscene, "debug_opencl_device_type", text="Device") + col.prop(cscene, "debug_opencl_kernel_single_program", text="Single Program") col.prop(cscene, "debug_use_opencl_debug", text="Debug") @@ -1633,7 +1637,7 @@ def draw_device(self, context): split = layout.split(percentage=1/3) split.label("Device:") row = split.row() - row.active = show_device_selection(context) + row.active = show_device_active(context) row.prop(cscene, "device", text="") if engine.with_osl() and use_cpu(context): @@ -1712,17 +1716,75 @@ def get_panels(): return panels + +classes = ( + CYCLES_MT_sampling_presets, + CYCLES_MT_integrator_presets, + CyclesRender_PT_sampling, + CyclesRender_PT_geometry, + CyclesRender_PT_light_paths, + CyclesRender_PT_motion_blur, + CyclesRender_PT_film, + CyclesRender_PT_performance, + CyclesRender_PT_layer_options, + CyclesRender_PT_layer_passes, + CyclesRender_PT_views, + Cycles_PT_post_processing, + CyclesCamera_PT_dof, + Cycles_PT_context_material, + CyclesObject_PT_motion_blur, + CyclesObject_PT_cycles_settings, + CYCLES_OT_use_shading_nodes, + CyclesLamp_PT_preview, + CyclesLamp_PT_lamp, + CyclesLamp_PT_nodes, + CyclesLamp_PT_spot, + CyclesWorld_PT_preview, + CyclesWorld_PT_surface, + CyclesWorld_PT_volume, + CyclesWorld_PT_ambient_occlusion, + CyclesWorld_PT_mist, + CyclesWorld_PT_ray_visibility, + CyclesWorld_PT_settings, + CyclesMaterial_PT_preview, + CyclesMaterial_PT_surface, + CyclesMaterial_PT_volume, + CyclesMaterial_PT_displacement, + CyclesMaterial_PT_settings, + CyclesTexture_PT_context, + CyclesTexture_PT_node, + CyclesTexture_PT_mapping, + CyclesTexture_PT_colors, + CyclesParticle_PT_textures, + CyclesRender_PT_bake, + CyclesRender_PT_debug, + CyclesParticle_PT_CurveSettings, + CyclesScene_PT_simplify, +) + + def register(): + from bpy.utils import register_class + bpy.types.RENDER_PT_render.append(draw_device) bpy.types.VIEW3D_HT_header.append(draw_pause) for panel in get_panels(): panel.COMPAT_ENGINES.add('CYCLES') + for cls in classes: + register_class(cls) + + def unregister(): + from bpy.utils import unregister_class + bpy.types.RENDER_PT_render.remove(draw_device) bpy.types.VIEW3D_HT_header.remove(draw_pause) for panel in get_panels(): if 'CYCLES' in panel.COMPAT_ENGINES: panel.COMPAT_ENGINES.remove('CYCLES') + + for cls in classes: + unregister_class(cls) diff --git a/intern/cycles/blender/blender_camera.cpp b/intern/cycles/blender/blender_camera.cpp index f02fc553908..40d6b25f2b7 100644 --- a/intern/cycles/blender/blender_camera.cpp +++ b/intern/cycles/blender/blender_camera.cpp @@ -14,13 +14,13 @@ * limitations under the License. */ -#include "camera.h" -#include "scene.h" +#include "render/camera.h" +#include "render/scene.h" -#include "blender_sync.h" -#include "blender_util.h" +#include "blender/blender_sync.h" +#include "blender/blender_util.h" -#include "util_logging.h" +#include "util/util_logging.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/blender/blender_curves.cpp b/intern/cycles/blender/blender_curves.cpp index e42ff5d72a6..6fa038e8bf0 100644 --- a/intern/cycles/blender/blender_curves.cpp +++ b/intern/cycles/blender/blender_curves.cpp @@ -14,18 +14,18 @@ * limitations under the License. */ -#include "attribute.h" -#include "camera.h" -#include "curves.h" -#include "mesh.h" -#include "object.h" -#include "scene.h" +#include "render/attribute.h" +#include "render/camera.h" +#include "render/curves.h" +#include "render/mesh.h" +#include "render/object.h" +#include "render/scene.h" -#include "blender_sync.h" -#include "blender_util.h" +#include "blender/blender_sync.h" +#include "blender/blender_util.h" -#include "util_foreach.h" -#include "util_logging.h" +#include "util/util_foreach.h" +#include "util/util_logging.h" CCL_NAMESPACE_BEGIN @@ -411,6 +411,7 @@ static void ExportCurveTrianglePlanes(Mesh *mesh, ParticleCurveData *CData, } } + mesh->resize_mesh(mesh->verts.size(), mesh->triangles.size()); mesh->attributes.remove(ATTR_STD_VERTEX_NORMAL); mesh->attributes.remove(ATTR_STD_FACE_NORMAL); mesh->add_face_normals(); @@ -434,8 +435,8 @@ static void ExportCurveTriangleGeometry(Mesh *mesh, if(CData->curve_keynum[curve] <= 1 || CData->curve_length[curve] == 0.0f) continue; - numverts += (CData->curve_keynum[curve] - 2)*2*resolution + resolution; - numtris += (CData->curve_keynum[curve] - 2)*resolution; + numverts += (CData->curve_keynum[curve] - 1)*resolution + resolution; + numtris += (CData->curve_keynum[curve] - 1)*2*resolution; } } @@ -545,6 +546,7 @@ static void ExportCurveTriangleGeometry(Mesh *mesh, } } + mesh->resize_mesh(mesh->verts.size(), mesh->triangles.size()); mesh->attributes.remove(ATTR_STD_VERTEX_NORMAL); mesh->attributes.remove(ATTR_STD_FACE_NORMAL); mesh->add_face_normals(); @@ -890,7 +892,7 @@ void BlenderSync::sync_curves(Mesh *mesh, } /* obtain general settings */ - bool use_curves = scene->curve_system_manager->use_curves; + const bool use_curves = scene->curve_system_manager->use_curves; if(!(use_curves && b_ob.mode() != b_ob.mode_PARTICLE_EDIT)) { if(!motion) @@ -898,11 +900,11 @@ void BlenderSync::sync_curves(Mesh *mesh, return; } - int primitive = scene->curve_system_manager->primitive; - int triangle_method = scene->curve_system_manager->triangle_method; - int resolution = scene->curve_system_manager->resolution; - size_t vert_num = mesh->verts.size(); - size_t tri_num = mesh->num_triangles(); + const int primitive = scene->curve_system_manager->primitive; + const int triangle_method = scene->curve_system_manager->triangle_method; + const int resolution = scene->curve_system_manager->resolution; + const size_t vert_num = mesh->verts.size(); + const size_t tri_num = mesh->num_triangles(); int used_res = 1; /* extract particle hair data - should be combined with connecting to mesh later*/ diff --git a/intern/cycles/blender/blender_logging.cpp b/intern/cycles/blender/blender_logging.cpp index f4f86929168..d0f82e37662 100644 --- a/intern/cycles/blender/blender_logging.cpp +++ b/intern/cycles/blender/blender_logging.cpp @@ -14,8 +14,8 @@ * limitations under the License. */ -#include "CCL_api.h" -#include "util_logging.h" +#include "blender/CCL_api.h" +#include "util/util_logging.h" void CCL_init_logging(const char *argv0) { diff --git a/intern/cycles/blender/blender_mesh.cpp b/intern/cycles/blender/blender_mesh.cpp index 85117cfff7b..e0e89cec65c 100644 --- a/intern/cycles/blender/blender_mesh.cpp +++ b/intern/cycles/blender/blender_mesh.cpp @@ -15,21 +15,22 @@ */ -#include "mesh.h" -#include "object.h" -#include "scene.h" -#include "camera.h" +#include "render/mesh.h" +#include "render/object.h" +#include "render/scene.h" +#include "render/camera.h" -#include "blender_sync.h" -#include "blender_session.h" -#include "blender_util.h" +#include "blender/blender_sync.h" +#include "blender/blender_session.h" +#include "blender/blender_util.h" -#include "subd_patch.h" -#include "subd_split.h" +#include "subd/subd_patch.h" +#include "subd/subd_split.h" -#include "util_foreach.h" -#include "util_logging.h" -#include "util_math.h" +#include "util/util_algorithm.h" +#include "util/util_foreach.h" +#include "util/util_logging.h" +#include "util/util_math.h" #include "mikktspace.h" @@ -525,69 +526,177 @@ static void attr_create_uv_map(Scene *scene, } /* Create vertex pointiness attributes. */ + +/* Compare vertices by sum of their coordinates. */ +class VertexAverageComparator { +public: + VertexAverageComparator(const array<float3>& verts) + : verts_(verts) { + } + + bool operator()(const int& vert_idx_a, const int& vert_idx_b) + { + const float3 &vert_a = verts_[vert_idx_a]; + const float3 &vert_b = verts_[vert_idx_b]; + if(vert_a == vert_b) { + /* Special case for doubles, so we ensure ordering. */ + return vert_idx_a > vert_idx_b; + } + const float x1 = vert_a.x + vert_a.y + vert_a.z; + const float x2 = vert_b.x + vert_b.y + vert_b.z; + return x1 < x2; + } + +protected: + const array<float3>& verts_; +}; + static void attr_create_pointiness(Scene *scene, Mesh *mesh, BL::Mesh& b_mesh, bool subdivision) { - if(mesh->need_attribute(scene, ATTR_STD_POINTINESS)) { - const int numverts = b_mesh.vertices.length(); - AttributeSet& attributes = (subdivision)? mesh->subd_attributes: mesh->attributes; - Attribute *attr = attributes.add(ATTR_STD_POINTINESS); - float *data = attr->data_float(); - int *counter = new int[numverts]; - float *raw_data = new float[numverts]; - float3 *edge_accum = new float3[numverts]; - - /* Calculate pointiness using single ring neighborhood. */ - memset(counter, 0, sizeof(int) * numverts); - memset(raw_data, 0, sizeof(float) * numverts); - memset(edge_accum, 0, sizeof(float3) * numverts); - BL::Mesh::edges_iterator e; - int i = 0; - for(b_mesh.edges.begin(e); e != b_mesh.edges.end(); ++e, ++i) { - int v0 = b_mesh.edges[i].vertices()[0], - v1 = b_mesh.edges[i].vertices()[1]; - float3 co0 = get_float3(b_mesh.vertices[v0].co()), - co1 = get_float3(b_mesh.vertices[v1].co()); - float3 edge = normalize(co1 - co0); - edge_accum[v0] += edge; - edge_accum[v1] += -edge; - ++counter[v0]; - ++counter[v1]; - } - i = 0; - BL::Mesh::vertices_iterator v; - for(b_mesh.vertices.begin(v); v != b_mesh.vertices.end(); ++v, ++i) { - if(counter[i] > 0) { - float3 normal = get_float3(b_mesh.vertices[i].normal()); - float angle = safe_acosf(dot(normal, edge_accum[i] / counter[i])); - raw_data[i] = angle * M_1_PI_F; + if(!mesh->need_attribute(scene, ATTR_STD_POINTINESS)) { + return; + } + const int num_verts = b_mesh.vertices.length(); + /* STEP 1: Find out duplicated vertices and point duplicates to a single + * original vertex. + */ + vector<int> sorted_vert_indeices(num_verts); + for(int vert_index = 0; vert_index < num_verts; ++vert_index) { + sorted_vert_indeices[vert_index] = vert_index; + } + VertexAverageComparator compare(mesh->verts); + sort(sorted_vert_indeices.begin(), sorted_vert_indeices.end(), compare); + /* This array stores index of the original vertex for the given vertex + * index. + */ + vector<int> vert_orig_index(num_verts); + for(int sorted_vert_index = 0; + sorted_vert_index < num_verts; + ++sorted_vert_index) + { + const int vert_index = sorted_vert_indeices[sorted_vert_index]; + const float3 &vert_co = mesh->verts[vert_index]; + bool found = false; + for(int other_sorted_vert_index = sorted_vert_index + 1; + other_sorted_vert_index < num_verts; + ++other_sorted_vert_index) + { + const int other_vert_index = + sorted_vert_indeices[other_sorted_vert_index]; + const float3 &other_vert_co = mesh->verts[other_vert_index]; + /* We are too far away now, we wouldn't have duplicate. */ + if((other_vert_co.x + other_vert_co.y + other_vert_co.z) - + (vert_co.x + vert_co.y + vert_co.z) > 3 * FLT_EPSILON) + { + break; } - else { - raw_data[i] = 0.0f; + /* Found duplicate. */ + if(len_squared(other_vert_co - vert_co) < FLT_EPSILON) { + found = true; + vert_orig_index[vert_index] = other_vert_index; + break; } } - - /* Blur vertices to approximate 2 ring neighborhood. */ - memset(counter, 0, sizeof(int) * numverts); - memcpy(data, raw_data, sizeof(float) * numverts); - i = 0; - for(b_mesh.edges.begin(e); e != b_mesh.edges.end(); ++e, ++i) { - int v0 = b_mesh.edges[i].vertices()[0], - v1 = b_mesh.edges[i].vertices()[1]; - data[v0] += raw_data[v1]; - data[v1] += raw_data[v0]; - ++counter[v0]; - ++counter[v1]; + if(!found) { + vert_orig_index[vert_index] = vert_index; + } + } + /* Make sure we always points to the very first orig vertex. */ + for(int vert_index = 0; vert_index < num_verts; ++vert_index) { + int orig_index = vert_orig_index[vert_index]; + while(orig_index != vert_orig_index[orig_index]) { + orig_index = vert_orig_index[orig_index]; } - for(i = 0; i < numverts; ++i) { - data[i] /= counter[i] + 1; + vert_orig_index[vert_index] = orig_index; + } + sorted_vert_indeices.free_memory(); + /* STEP 2: Calculate vertex normals taking into account their possible + * duplicates which gets "welded" together. + */ + vector<float3> vert_normal(num_verts, make_float3(0.0f, 0.0f, 0.0f)); + /* First we accumulate all vertex normals in the original index. */ + for(int vert_index = 0; vert_index < num_verts; ++vert_index) { + const float3 normal = get_float3(b_mesh.vertices[vert_index].normal()); + const int orig_index = vert_orig_index[vert_index]; + vert_normal[orig_index] += normal; + } + /* Then we normalize the accumulated result and flush it to all duplicates + * as well. + */ + for(int vert_index = 0; vert_index < num_verts; ++vert_index) { + const int orig_index = vert_orig_index[vert_index]; + vert_normal[vert_index] = normalize(vert_normal[orig_index]); + } + /* STEP 3: Calculate pointiness using single ring neighborhood. */ + vector<int> counter(num_verts, 0); + vector<float> raw_data(num_verts, 0.0f); + vector<float3> edge_accum(num_verts, make_float3(0.0f, 0.0f, 0.0f)); + BL::Mesh::edges_iterator e; + EdgeMap visited_edges; + int edge_index = 0; + memset(&counter[0], 0, sizeof(int) * counter.size()); + for(b_mesh.edges.begin(e); e != b_mesh.edges.end(); ++e, ++edge_index) { + const int v0 = vert_orig_index[b_mesh.edges[edge_index].vertices()[0]], + v1 = vert_orig_index[b_mesh.edges[edge_index].vertices()[1]]; + if(visited_edges.exists(v0, v1)) { + continue; + } + visited_edges.insert(v0, v1); + float3 co0 = get_float3(b_mesh.vertices[v0].co()), + co1 = get_float3(b_mesh.vertices[v1].co()); + float3 edge = normalize(co1 - co0); + edge_accum[v0] += edge; + edge_accum[v1] += -edge; + ++counter[v0]; + ++counter[v1]; + } + for(int vert_index = 0; vert_index < num_verts; ++vert_index) { + const int orig_index = vert_orig_index[vert_index]; + if(orig_index != vert_index) { + /* Skip duplicates, they'll be overwritten later on. */ + continue; + } + if(counter[vert_index] > 0) { + const float3 normal = vert_normal[vert_index]; + const float angle = + safe_acosf(dot(normal, + edge_accum[vert_index] / counter[vert_index])); + raw_data[vert_index] = angle * M_1_PI_F; + } + else { + raw_data[vert_index] = 0.0f; } - - delete [] counter; - delete [] raw_data; - delete [] edge_accum; + } + /* STEP 3: Blur vertices to approximate 2 ring neighborhood. */ + AttributeSet& attributes = (subdivision)? mesh->subd_attributes: mesh->attributes; + Attribute *attr = attributes.add(ATTR_STD_POINTINESS); + float *data = attr->data_float(); + memcpy(data, &raw_data[0], sizeof(float) * raw_data.size()); + memset(&counter[0], 0, sizeof(int) * counter.size()); + edge_index = 0; + visited_edges.clear(); + for(b_mesh.edges.begin(e); e != b_mesh.edges.end(); ++e, ++edge_index) { + const int v0 = vert_orig_index[b_mesh.edges[edge_index].vertices()[0]], + v1 = vert_orig_index[b_mesh.edges[edge_index].vertices()[1]]; + if(visited_edges.exists(v0, v1)) { + continue; + } + visited_edges.insert(v0, v1); + data[v0] += raw_data[v1]; + data[v1] += raw_data[v0]; + ++counter[v0]; + ++counter[v1]; + } + for(int vert_index = 0; vert_index < num_verts; ++vert_index) { + data[vert_index] /= counter[vert_index] + 1; + } + /* STEP 4: Copy attribute to the duplicated vertices. */ + for(int vert_index = 0; vert_index < num_verts; ++vert_index) { + const int orig_index = vert_orig_index[vert_index]; + data[vert_index] = data[orig_index]; } } @@ -656,9 +765,6 @@ static void create_mesh(Scene *scene, generated[i++] = get_float3(v->undeformed_co())*size - loc; } - /* Create needed vertex attributes. */ - attr_create_pointiness(scene, mesh, b_mesh, subdivision); - /* create faces */ vector<int> nverts(numfaces); vector<int> face_flags(numfaces, FACE_FLAG_NONE); @@ -671,6 +777,15 @@ static void create_mesh(Scene *scene, int shader = clamp(f->material_index(), 0, used_shaders.size()-1); bool smooth = f->use_smooth() || use_loop_normals; + if(use_loop_normals) { + BL::Array<float, 12> loop_normals = f->split_normals(); + for(int i = 0; i < n; i++) { + N[vi[i]] = make_float3(loop_normals[i * 3], + loop_normals[i * 3 + 1], + loop_normals[i * 3 + 2]); + } + } + /* Create triangles. * * NOTE: Autosmooth is already taken care about. @@ -704,7 +819,7 @@ static void create_mesh(Scene *scene, int shader = clamp(p->material_index(), 0, used_shaders.size()-1); bool smooth = p->use_smooth() || use_loop_normals; - vi.reserve(n); + vi.resize(n); for(int i = 0; i < n; i++) { /* NOTE: Autosmooth is already taken care about. */ vi[i] = b_mesh.loops[p->loop_start() + i].vertex_index(); @@ -718,6 +833,7 @@ static void create_mesh(Scene *scene, /* Create all needed attributes. * The calculate functions will check whether they're needed or not. */ + attr_create_pointiness(scene, mesh, b_mesh, subdivision); attr_create_vertex_color(scene, mesh, b_mesh, nverts, face_flags, subdivision); attr_create_uv_map(scene, mesh, b_mesh, nverts, face_flags, subdivision, subdivide_uvs); @@ -1178,4 +1294,3 @@ void BlenderSync::sync_mesh_motion(BL::Object& b_ob, } CCL_NAMESPACE_END - diff --git a/intern/cycles/blender/blender_object.cpp b/intern/cycles/blender/blender_object.cpp index 637cf7abda8..d05699236cc 100644 --- a/intern/cycles/blender/blender_object.cpp +++ b/intern/cycles/blender/blender_object.cpp @@ -14,24 +14,24 @@ * limitations under the License. */ -#include "camera.h" -#include "integrator.h" -#include "graph.h" -#include "light.h" -#include "mesh.h" -#include "object.h" -#include "scene.h" -#include "nodes.h" -#include "particles.h" -#include "shader.h" - -#include "blender_object_cull.h" -#include "blender_sync.h" -#include "blender_util.h" - -#include "util_foreach.h" -#include "util_hash.h" -#include "util_logging.h" +#include "render/camera.h" +#include "render/integrator.h" +#include "render/graph.h" +#include "render/light.h" +#include "render/mesh.h" +#include "render/object.h" +#include "render/scene.h" +#include "render/nodes.h" +#include "render/particles.h" +#include "render/shader.h" + +#include "blender/blender_object_cull.h" +#include "blender/blender_sync.h" +#include "blender/blender_util.h" + +#include "util/util_foreach.h" +#include "util/util_hash.h" +#include "util/util_logging.h" CCL_NAMESPACE_BEGIN @@ -343,6 +343,13 @@ Object *BlenderSync::sync_object(BL::Object& b_parent, object_updated = true; } + PointerRNA cobject = RNA_pointer_get(&b_ob.ptr, "cycles"); + bool is_shadow_catcher = get_boolean(cobject, "is_shadow_catcher"); + if(is_shadow_catcher != object->is_shadow_catcher) { + object->is_shadow_catcher = is_shadow_catcher; + object_updated = true; + } + /* object sync * transform comparison should not be needed, but duplis don't work perfect * in the depsgraph and may not signal changes, so this is a workaround */ diff --git a/intern/cycles/blender/blender_object_cull.cpp b/intern/cycles/blender/blender_object_cull.cpp index 08918dd1a49..0333c027f70 100644 --- a/intern/cycles/blender/blender_object_cull.cpp +++ b/intern/cycles/blender/blender_object_cull.cpp @@ -16,9 +16,9 @@ #include <cstdlib> -#include "camera.h" +#include "render/camera.h" -#include "blender_object_cull.h" +#include "blender/blender_object_cull.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/blender/blender_object_cull.h b/intern/cycles/blender/blender_object_cull.h index b6f0ca5cd31..2147877a860 100644 --- a/intern/cycles/blender/blender_object_cull.h +++ b/intern/cycles/blender/blender_object_cull.h @@ -17,8 +17,8 @@ #ifndef __BLENDER_OBJECT_CULL_H__ #define __BLENDER_OBJECT_CULL_H__ -#include "blender_sync.h" -#include "util_types.h" +#include "blender/blender_sync.h" +#include "util/util_types.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/blender/blender_particles.cpp b/intern/cycles/blender/blender_particles.cpp index dd2900a8d5b..00f8cb3cf1b 100644 --- a/intern/cycles/blender/blender_particles.cpp +++ b/intern/cycles/blender/blender_particles.cpp @@ -14,14 +14,14 @@ * limitations under the License. */ -#include "mesh.h" -#include "object.h" -#include "particles.h" +#include "render/mesh.h" +#include "render/object.h" +#include "render/particles.h" -#include "blender_sync.h" -#include "blender_util.h" +#include "blender/blender_sync.h" +#include "blender/blender_util.h" -#include "util_foreach.h" +#include "util/util_foreach.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/blender/blender_python.cpp b/intern/cycles/blender/blender_python.cpp index 438abc49f88..d509e9de981 100644 --- a/intern/cycles/blender/blender_python.cpp +++ b/intern/cycles/blender/blender_python.cpp @@ -16,21 +16,21 @@ #include <Python.h> -#include "CCL_api.h" +#include "blender/CCL_api.h" -#include "blender_sync.h" -#include "blender_session.h" +#include "blender/blender_sync.h" +#include "blender/blender_session.h" -#include "util_foreach.h" -#include "util_logging.h" -#include "util_md5.h" -#include "util_opengl.h" -#include "util_path.h" -#include "util_string.h" -#include "util_types.h" +#include "util/util_foreach.h" +#include "util/util_logging.h" +#include "util/util_md5.h" +#include "util/util_opengl.h" +#include "util/util_path.h" +#include "util/util_string.h" +#include "util/util_types.h" #ifdef WITH_OSL -#include "osl.h" +#include "render/osl.h" #include <OSL/oslquery.h> #include <OSL/oslconfig.h> @@ -67,8 +67,10 @@ bool debug_flags_sync_from_scene(BL::Scene b_scene) flags.cpu.sse3 = get_boolean(cscene, "debug_use_cpu_sse3"); flags.cpu.sse2 = get_boolean(cscene, "debug_use_cpu_sse2"); flags.cpu.qbvh = get_boolean(cscene, "debug_use_qbvh"); + flags.cpu.split_kernel = get_boolean(cscene, "debug_use_cpu_split_kernel"); /* Synchronize CUDA flags. */ flags.cuda.adaptive_compile = get_boolean(cscene, "debug_use_cuda_adaptive_compile"); + flags.cuda.split_kernel = get_boolean(cscene, "debug_use_cuda_split_kernel"); /* Synchronize OpenCL kernel type. */ switch(get_enum(cscene, "debug_opencl_kernel_type")) { case 0: @@ -104,6 +106,7 @@ bool debug_flags_sync_from_scene(BL::Scene b_scene) } /* Synchronize other OpenCL flags. */ flags.opencl.debug = get_boolean(cscene, "debug_use_opencl_debug"); + flags.opencl.single_program = get_boolean(cscene, "debug_opencl_kernel_single_program"); return flags.opencl.device_type != opencl_device_type || flags.opencl.kernel_type != opencl_kernel_type; } @@ -641,7 +644,7 @@ static PyObject *debug_flags_reset_func(PyObject * /*self*/, PyObject * /*args*/ Py_RETURN_NONE; } -static PyObject *set_resumable_chunks_func(PyObject * /*self*/, PyObject *args) +static PyObject *set_resumable_chunk_func(PyObject * /*self*/, PyObject *args) { int num_resumable_chunks, current_resumable_chunk; if(!PyArg_ParseTuple(args, "ii", @@ -676,6 +679,53 @@ static PyObject *set_resumable_chunks_func(PyObject * /*self*/, PyObject *args) Py_RETURN_NONE; } +static PyObject *set_resumable_chunk_range_func(PyObject * /*self*/, PyObject *args) +{ + int num_chunks, start_chunk, end_chunk; + if(!PyArg_ParseTuple(args, "iii", + &num_chunks, + &start_chunk, + &end_chunk)) { + Py_RETURN_NONE; + } + + if(num_chunks <= 0) { + fprintf(stderr, "Cycles: Bad value for number of resumable chunks.\n"); + abort(); + Py_RETURN_NONE; + } + if(start_chunk < 1 || start_chunk > num_chunks) { + fprintf(stderr, "Cycles: Bad value for start chunk number.\n"); + abort(); + Py_RETURN_NONE; + } + if(end_chunk < 1 || end_chunk > num_chunks) { + fprintf(stderr, "Cycles: Bad value for start chunk number.\n"); + abort(); + Py_RETURN_NONE; + } + if(start_chunk > end_chunk) { + fprintf(stderr, "Cycles: End chunk should be higher than start one.\n"); + abort(); + Py_RETURN_NONE; + } + + VLOG(1) << "Initialized resumable render: " + << "num_resumable_chunks=" << num_chunks << ", " + << "start_resumable_chunk=" << start_chunk + << "end_resumable_chunk=" << end_chunk; + BlenderSession::num_resumable_chunks = num_chunks; + BlenderSession::start_resumable_chunk = start_chunk; + BlenderSession::end_resumable_chunk = end_chunk; + + printf("Cycles: Will render chunks %d to %d of %d\n", + start_chunk, + end_chunk, + num_chunks); + + Py_RETURN_NONE; +} + static PyObject *get_device_types_func(PyObject * /*self*/, PyObject * /*args*/) { vector<DeviceInfo>& devices = Device::available_devices(); @@ -715,7 +765,8 @@ static PyMethodDef methods[] = { {"debug_flags_reset", debug_flags_reset_func, METH_NOARGS, ""}, /* Resumable render */ - {"set_resumable_chunks", set_resumable_chunks_func, METH_VARARGS, ""}, + {"set_resumable_chunk", set_resumable_chunk_func, METH_VARARGS, ""}, + {"set_resumable_chunk_range", set_resumable_chunk_range_func, METH_VARARGS, ""}, /* Compute Device selection */ {"get_device_types", get_device_types_func, METH_VARARGS, ""}, diff --git a/intern/cycles/blender/blender_session.cpp b/intern/cycles/blender/blender_session.cpp index 2f30cbd961f..26f9bccd95d 100644 --- a/intern/cycles/blender/blender_session.cpp +++ b/intern/cycles/blender/blender_session.cpp @@ -16,36 +16,38 @@ #include <stdlib.h> -#include "background.h" -#include "buffers.h" -#include "camera.h" -#include "device.h" -#include "integrator.h" -#include "film.h" -#include "light.h" -#include "mesh.h" -#include "object.h" -#include "scene.h" -#include "session.h" -#include "shader.h" - -#include "util_color.h" -#include "util_foreach.h" -#include "util_function.h" -#include "util_hash.h" -#include "util_logging.h" -#include "util_progress.h" -#include "util_time.h" - -#include "blender_sync.h" -#include "blender_session.h" -#include "blender_util.h" +#include "render/background.h" +#include "render/buffers.h" +#include "render/camera.h" +#include "device/device.h" +#include "render/integrator.h" +#include "render/film.h" +#include "render/light.h" +#include "render/mesh.h" +#include "render/object.h" +#include "render/scene.h" +#include "render/session.h" +#include "render/shader.h" + +#include "util/util_color.h" +#include "util/util_foreach.h" +#include "util/util_function.h" +#include "util/util_hash.h" +#include "util/util_logging.h" +#include "util/util_progress.h" +#include "util/util_time.h" + +#include "blender/blender_sync.h" +#include "blender/blender_session.h" +#include "blender/blender_util.h" CCL_NAMESPACE_BEGIN bool BlenderSession::headless = false; int BlenderSession::num_resumable_chunks = 0; int BlenderSession::current_resumable_chunk = 0; +int BlenderSession::start_resumable_chunk = 0; +int BlenderSession::end_resumable_chunk = 0; BlenderSession::BlenderSession(BL::RenderEngine& b_engine, BL::UserPreferences& b_userpref, @@ -68,6 +70,7 @@ BlenderSession::BlenderSession(BL::RenderEngine& b_engine, background = true; last_redraw_time = 0.0; start_resize_time = 0.0; + last_status_time = 0.0; } BlenderSession::BlenderSession(BL::RenderEngine& b_engine, @@ -93,6 +96,7 @@ BlenderSession::BlenderSession(BL::RenderEngine& b_engine, background = false; last_redraw_time = 0.0; start_resize_time = 0.0; + last_status_time = 0.0; } BlenderSession::~BlenderSession() @@ -989,10 +993,14 @@ void BlenderSession::update_status_progress() if(substatus.size() > 0) status += " | " + substatus; - if(status != last_status) { + double current_time = time_dt(); + /* When rendering in a window, redraw the status at least once per second to keep the elapsed and remaining time up-to-date. + * For headless rendering, only report when something significant changes to keep the console output readable. */ + if(status != last_status || (!headless && (current_time - last_status_time) > 1.0)) { b_engine.update_stats("", (timestatus + scene + status).c_str()); b_engine.update_memory_stats(mem_used, mem_peak); last_status = status; + last_status_time = current_time; } if(progress != last_progress) { b_engine.update_progress(progress); @@ -1342,9 +1350,21 @@ void BlenderSession::update_resumable_tile_manager(int num_samples) return; } - int num_samples_per_chunk = (int)ceilf((float)num_samples / num_resumable_chunks); - int range_start_sample = num_samples_per_chunk * (current_resumable_chunk - 1); - int range_num_samples = num_samples_per_chunk; + const int num_samples_per_chunk = (int)ceilf((float)num_samples / num_resumable_chunks); + + int range_start_sample, range_num_samples; + if(current_resumable_chunk != 0) { + /* Single chunk rendering. */ + range_start_sample = num_samples_per_chunk * (current_resumable_chunk - 1); + range_num_samples = num_samples_per_chunk; + } + else { + /* Ranged-chunks. */ + const int num_chunks = end_resumable_chunk - start_resumable_chunk + 1; + range_start_sample = num_samples_per_chunk * (start_resumable_chunk - 1); + range_num_samples = num_chunks * num_samples_per_chunk; + } + /* Make sure we don't overshoot. */ if(range_start_sample + range_num_samples > num_samples) { range_num_samples = num_samples - range_num_samples; } diff --git a/intern/cycles/blender/blender_session.h b/intern/cycles/blender/blender_session.h index 82fe218b4ce..22b21a18f2e 100644 --- a/intern/cycles/blender/blender_session.h +++ b/intern/cycles/blender/blender_session.h @@ -17,12 +17,12 @@ #ifndef __BLENDER_SESSION_H__ #define __BLENDER_SESSION_H__ -#include "device.h" -#include "scene.h" -#include "session.h" -#include "bake.h" +#include "device/device.h" +#include "render/scene.h" +#include "render/session.h" +#include "render/bake.h" -#include "util_vector.h" +#include "util/util_vector.h" CCL_NAMESPACE_BEGIN @@ -113,6 +113,7 @@ public: string last_status; string last_error; float last_progress; + double last_status_time; int width, height; double start_resize_time; @@ -137,6 +138,10 @@ public: /* Current resumable chunk index to render. */ static int current_resumable_chunk; + /* Alternative to single-chunk rendering to render a range of chunks. */ + static int start_resumable_chunk; + static int end_resumable_chunk; + protected: void do_write_update_render_result(BL::RenderResult& b_rr, BL::RenderLayer& b_rlay, diff --git a/intern/cycles/blender/blender_shader.cpp b/intern/cycles/blender/blender_shader.cpp index c9d4236a7f2..1ec82445b20 100644 --- a/intern/cycles/blender/blender_shader.cpp +++ b/intern/cycles/blender/blender_shader.cpp @@ -14,20 +14,21 @@ * limitations under the License. */ -#include "background.h" -#include "graph.h" -#include "light.h" -#include "nodes.h" -#include "osl.h" -#include "scene.h" -#include "shader.h" - -#include "blender_texture.h" -#include "blender_sync.h" -#include "blender_util.h" - -#include "util_debug.h" -#include "util_string.h" +#include "render/background.h" +#include "render/graph.h" +#include "render/light.h" +#include "render/nodes.h" +#include "render/osl.h" +#include "render/scene.h" +#include "render/shader.h" + +#include "blender/blender_texture.h" +#include "blender/blender_sync.h" +#include "blender/blender_util.h" + +#include "util/util_debug.h" +#include "util/util_string.h" +#include "util/util_task.h" CCL_NAMESPACE_BEGIN @@ -622,7 +623,8 @@ static ShaderNode *add_node(Scene *scene, bool is_builtin = b_image.packed_file() || b_image.source() == BL::Image::source_GENERATED || b_image.source() == BL::Image::source_MOVIE || - b_engine.is_preview(); + (b_engine.is_preview() && + b_image.source() != BL::Image::source_SEQUENCE); if(is_builtin) { /* for builtin images we're using image datablock name to find an image to @@ -675,7 +677,8 @@ static ShaderNode *add_node(Scene *scene, bool is_builtin = b_image.packed_file() || b_image.source() == BL::Image::source_GENERATED || b_image.source() == BL::Image::source_MOVIE || - b_engine.is_preview(); + (b_engine.is_preview() && + b_image.source() != BL::Image::source_SEQUENCE); if(is_builtin) { int scene_frame = b_scene.frame_current(); @@ -1168,6 +1171,13 @@ static void add_nodes(Scene *scene, /* Sync Materials */ +void BlenderSync::sync_materials_simpligy(Shader *shader) +{ + ShaderGraph *graph = shader->graph; + graph->simplify(scene); + shader->tag_update(scene); +} + void BlenderSync::sync_materials(bool update_all) { shader_map.set_default(scene->default_surface); @@ -1175,6 +1185,8 @@ void BlenderSync::sync_materials(bool update_all) /* material loop */ BL::BlendData::materials_iterator b_mat; + TaskPool pool; + for(b_data.materials.begin(b_mat); b_mat != b_data.materials.end(); ++b_mat) { Shader *shader; @@ -1210,9 +1222,31 @@ void BlenderSync::sync_materials(bool update_all) shader->displacement_method = (experimental) ? get_displacement_method(cmat) : DISPLACE_BUMP; shader->set_graph(graph); - shader->tag_update(scene); + + /* By simplifying the shader graph as soon as possible, some + * redundant shader nodes might be removed which prevents loading + * unnecessary attributes later. + * + * However, since graph simplification also accounts for e.g. mix + * weight, this would cause frequent expensive resyncs in interactive + * sessions, so for those sessions optimization is only performed + * right before compiling. + */ + if(!preview) { + pool.push(function_bind(&BlenderSync::sync_materials_simpligy, + this, + shader)); + } + else { + /* NOTE: Update tagging can access links which are being + * optimized out. + */ + shader->tag_update(scene); + } } } + + pool.wait_work(); } /* Sync World */ diff --git a/intern/cycles/blender/blender_sync.cpp b/intern/cycles/blender/blender_sync.cpp index d8043105cd8..3b071bf0e7d 100644 --- a/intern/cycles/blender/blender_sync.cpp +++ b/intern/cycles/blender/blender_sync.cpp @@ -14,29 +14,29 @@ * limitations under the License. */ -#include "background.h" -#include "camera.h" -#include "film.h" -#include "graph.h" -#include "integrator.h" -#include "light.h" -#include "mesh.h" -#include "nodes.h" -#include "object.h" -#include "scene.h" -#include "shader.h" -#include "curves.h" - -#include "device.h" - -#include "blender_sync.h" -#include "blender_session.h" -#include "blender_util.h" - -#include "util_debug.h" -#include "util_foreach.h" -#include "util_opengl.h" -#include "util_hash.h" +#include "render/background.h" +#include "render/camera.h" +#include "render/film.h" +#include "render/graph.h" +#include "render/integrator.h" +#include "render/light.h" +#include "render/mesh.h" +#include "render/nodes.h" +#include "render/object.h" +#include "render/scene.h" +#include "render/shader.h" +#include "render/curves.h" + +#include "device/device.h" + +#include "blender/blender_sync.h" +#include "blender/blender_session.h" +#include "blender/blender_util.h" + +#include "util/util_debug.h" +#include "util/util_foreach.h" +#include "util/util_opengl.h" +#include "util/util_hash.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/blender/blender_sync.h b/intern/cycles/blender/blender_sync.h index 6984cbda259..62e2f8f563a 100644 --- a/intern/cycles/blender/blender_sync.h +++ b/intern/cycles/blender/blender_sync.h @@ -22,15 +22,15 @@ #include "RNA_access.h" #include "RNA_blender_cpp.h" -#include "blender_util.h" +#include "blender/blender_util.h" -#include "scene.h" -#include "session.h" +#include "render/scene.h" +#include "render/session.h" -#include "util_map.h" -#include "util_set.h" -#include "util_transform.h" -#include "util_vector.h" +#include "util/util_map.h" +#include "util/util_set.h" +#include "util/util_transform.h" +#include "util/util_vector.h" CCL_NAMESPACE_BEGIN @@ -96,6 +96,7 @@ public: private: /* sync */ void sync_lamps(bool update_all); + void sync_materials_simpligy(Shader *shader); void sync_materials(bool update_all); void sync_objects(BL::SpaceView3D& b_v3d, float motion_time = 0.0f); void sync_motion(BL::RenderSettings& b_render, diff --git a/intern/cycles/blender/blender_texture.cpp b/intern/cycles/blender/blender_texture.cpp index 3807e683c7c..b2e27b76189 100644 --- a/intern/cycles/blender/blender_texture.cpp +++ b/intern/cycles/blender/blender_texture.cpp @@ -14,7 +14,7 @@ * limitations under the License. */ -#include "blender_texture.h" +#include "blender/blender_texture.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/blender/blender_texture.h b/intern/cycles/blender/blender_texture.h index ad96f9db8ed..734231a85ec 100644 --- a/intern/cycles/blender/blender_texture.h +++ b/intern/cycles/blender/blender_texture.h @@ -18,7 +18,7 @@ #define __BLENDER_TEXTURE_H__ #include <stdlib.h> -#include "blender_sync.h" +#include "blender/blender_sync.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/blender/blender_util.h b/intern/cycles/blender/blender_util.h index b67834cdea3..4d575330520 100644 --- a/intern/cycles/blender/blender_util.h +++ b/intern/cycles/blender/blender_util.h @@ -17,14 +17,15 @@ #ifndef __BLENDER_UTIL_H__ #define __BLENDER_UTIL_H__ -#include "mesh.h" +#include "render/mesh.h" -#include "util_map.h" -#include "util_path.h" -#include "util_set.h" -#include "util_transform.h" -#include "util_types.h" -#include "util_vector.h" +#include "util/util_algorithm.h" +#include "util/util_map.h" +#include "util/util_path.h" +#include "util/util_set.h" +#include "util/util_transform.h" +#include "util/util_types.h" +#include "util/util_vector.h" /* Hacks to hook into Blender API * todo: clean this up ... */ @@ -78,7 +79,7 @@ static inline BL::Mesh object_to_mesh(BL::BlendData& data, me.calc_normals_split(); } else { - me.split_faces(); + me.split_faces(false); } } if(subdivision_type == Mesh::SUBDIVISION_NONE) { @@ -786,6 +787,35 @@ struct ParticleSystemKey { } }; +class EdgeMap { +public: + EdgeMap() { + } + + void clear() { + edges_.clear(); + } + + void insert(int v0, int v1) { + get_sorted_verts(v0, v1); + edges_.insert(std::pair<int, int>(v0, v1)); + } + + bool exists(int v0, int v1) { + get_sorted_verts(v0, v1); + return edges_.find(std::pair<int, int>(v0, v1)) != edges_.end(); + } + +protected: + void get_sorted_verts(int& v0, int& v1) { + if(v0 > v1) { + swap(v0, v1); + } + } + + set< std::pair<int, int> > edges_; +}; + CCL_NAMESPACE_END #endif /* __BLENDER_UTIL_H__ */ diff --git a/intern/cycles/bvh/CMakeLists.txt b/intern/cycles/bvh/CMakeLists.txt index 92e48f0d87f..4701d75350a 100644 --- a/intern/cycles/bvh/CMakeLists.txt +++ b/intern/cycles/bvh/CMakeLists.txt @@ -1,12 +1,6 @@ set(INC - . - ../graph - ../kernel - ../kernel/svm - ../render - ../util - ../device + .. ) set(INC_SYS diff --git a/intern/cycles/bvh/bvh.cpp b/intern/cycles/bvh/bvh.cpp index 874a4246d1d..58348d16746 100644 --- a/intern/cycles/bvh/bvh.cpp +++ b/intern/cycles/bvh/bvh.cpp @@ -15,25 +15,25 @@ * limitations under the License. */ -#include "mesh.h" -#include "object.h" -#include "scene.h" -#include "curves.h" - -#include "bvh.h" -#include "bvh_build.h" -#include "bvh_node.h" -#include "bvh_params.h" -#include "bvh_unaligned.h" - -#include "util_debug.h" -#include "util_foreach.h" -#include "util_logging.h" -#include "util_map.h" -#include "util_progress.h" -#include "util_system.h" -#include "util_types.h" -#include "util_math.h" +#include "render/mesh.h" +#include "render/object.h" +#include "render/scene.h" +#include "render/curves.h" + +#include "bvh/bvh.h" +#include "bvh/bvh_build.h" +#include "bvh/bvh_node.h" +#include "bvh/bvh_params.h" +#include "bvh/bvh_unaligned.h" + +#include "util/util_debug.h" +#include "util/util_foreach.h" +#include "util/util_logging.h" +#include "util/util_map.h" +#include "util/util_progress.h" +#include "util/util_system.h" +#include "util/util_types.h" +#include "util/util_math.h" CCL_NAMESPACE_BEGIN @@ -67,7 +67,7 @@ BVH *BVH::create(const BVHParams& params, const vector<Object*>& objects) if(params.use_qbvh) return new QBVH(params, objects); else - return new RegularBVH(params, objects); + return new BinaryBVH(params, objects); } /* Building */ @@ -81,6 +81,7 @@ void BVH::build(Progress& progress) pack.prim_type, pack.prim_index, pack.prim_object, + pack.prim_time, params, progress); BVHNode *root = bvh_build.run(); @@ -256,6 +257,10 @@ void BVH::pack_instances(size_t nodes_size, size_t leaf_nodes_size) pack.leaf_nodes.resize(leaf_nodes_size); pack.object_node.resize(objects.size()); + if(params.num_motion_curve_steps > 0 || params.num_motion_triangle_steps > 0) { + pack.prim_time.resize(prim_index_size); + } + int *pack_prim_index = (pack.prim_index.size())? &pack.prim_index[0]: NULL; int *pack_prim_type = (pack.prim_type.size())? &pack.prim_type[0]: NULL; int *pack_prim_object = (pack.prim_object.size())? &pack.prim_object[0]: NULL; @@ -264,6 +269,7 @@ void BVH::pack_instances(size_t nodes_size, size_t leaf_nodes_size) uint *pack_prim_tri_index = (pack.prim_tri_index.size())? &pack.prim_tri_index[0]: NULL; int4 *pack_nodes = (pack.nodes.size())? &pack.nodes[0]: NULL; int4 *pack_leaf_nodes = (pack.leaf_nodes.size())? &pack.leaf_nodes[0]: NULL; + float2 *pack_prim_time = (pack.prim_time.size())? &pack.prim_time[0]: NULL; /* merge */ foreach(Object *ob, objects) { @@ -309,6 +315,7 @@ void BVH::pack_instances(size_t nodes_size, size_t leaf_nodes_size) int *bvh_prim_type = &bvh->pack.prim_type[0]; uint *bvh_prim_visibility = &bvh->pack.prim_visibility[0]; uint *bvh_prim_tri_index = &bvh->pack.prim_tri_index[0]; + float2 *bvh_prim_time = bvh->pack.prim_time.size()? &bvh->pack.prim_time[0]: NULL; for(size_t i = 0; i < bvh_prim_index_size; i++) { if(bvh->pack.prim_type[i] & PRIMITIVE_ALL_CURVE) { @@ -324,6 +331,9 @@ void BVH::pack_instances(size_t nodes_size, size_t leaf_nodes_size) pack_prim_type[pack_prim_index_offset] = bvh_prim_type[i]; pack_prim_visibility[pack_prim_index_offset] = bvh_prim_visibility[i]; pack_prim_object[pack_prim_index_offset] = 0; // unused for instances + if(bvh_prim_time != NULL) { + pack_prim_time[pack_prim_index_offset] = bvh_prim_time[i]; + } pack_prim_index_offset++; } } @@ -414,64 +424,64 @@ static bool node_bvh_is_unaligned(const BVHNode *node) { const BVHNode *node0 = node->get_child(0), *node1 = node->get_child(1); - return node0->is_unaligned() || node1->is_unaligned(); + return node0->is_unaligned || node1->is_unaligned; } -RegularBVH::RegularBVH(const BVHParams& params_, const vector<Object*>& objects_) +BinaryBVH::BinaryBVH(const BVHParams& params_, const vector<Object*>& objects_) : BVH(params_, objects_) { } -void RegularBVH::pack_leaf(const BVHStackEntry& e, - const LeafNode *leaf) +void BinaryBVH::pack_leaf(const BVHStackEntry& e, + const LeafNode *leaf) { assert(e.idx + BVH_NODE_LEAF_SIZE <= pack.leaf_nodes.size()); float4 data[BVH_NODE_LEAF_SIZE]; memset(data, 0, sizeof(data)); - if(leaf->num_triangles() == 1 && pack.prim_index[leaf->m_lo] == -1) { + if(leaf->num_triangles() == 1 && pack.prim_index[leaf->lo] == -1) { /* object */ - data[0].x = __int_as_float(~(leaf->m_lo)); + data[0].x = __int_as_float(~(leaf->lo)); data[0].y = __int_as_float(0); } else { /* triangle */ - data[0].x = __int_as_float(leaf->m_lo); - data[0].y = __int_as_float(leaf->m_hi); + data[0].x = __int_as_float(leaf->lo); + data[0].y = __int_as_float(leaf->hi); } - data[0].z = __uint_as_float(leaf->m_visibility); + data[0].z = __uint_as_float(leaf->visibility); if(leaf->num_triangles() != 0) { - data[0].w = __uint_as_float(pack.prim_type[leaf->m_lo]); + data[0].w = __uint_as_float(pack.prim_type[leaf->lo]); } memcpy(&pack.leaf_nodes[e.idx], data, sizeof(float4)*BVH_NODE_LEAF_SIZE); } -void RegularBVH::pack_inner(const BVHStackEntry& e, - const BVHStackEntry& e0, - const BVHStackEntry& e1) +void BinaryBVH::pack_inner(const BVHStackEntry& e, + const BVHStackEntry& e0, + const BVHStackEntry& e1) { - if(e0.node->is_unaligned() || e1.node->is_unaligned()) { + if(e0.node->is_unaligned || e1.node->is_unaligned) { pack_unaligned_inner(e, e0, e1); } else { pack_aligned_inner(e, e0, e1); } } -void RegularBVH::pack_aligned_inner(const BVHStackEntry& e, - const BVHStackEntry& e0, - const BVHStackEntry& e1) +void BinaryBVH::pack_aligned_inner(const BVHStackEntry& e, + const BVHStackEntry& e0, + const BVHStackEntry& e1) { pack_aligned_node(e.idx, - e0.node->m_bounds, e1.node->m_bounds, + e0.node->bounds, e1.node->bounds, e0.encodeIdx(), e1.encodeIdx(), - e0.node->m_visibility, e1.node->m_visibility); + e0.node->visibility, e1.node->visibility); } -void RegularBVH::pack_aligned_node(int idx, - const BoundBox& b0, - const BoundBox& b1, - int c0, int c1, - uint visibility0, uint visibility1) +void BinaryBVH::pack_aligned_node(int idx, + const BoundBox& b0, + const BoundBox& b1, + int c0, int c1, + uint visibility0, uint visibility1) { assert(idx + BVH_NODE_SIZE <= pack.nodes.size()); assert(c0 < 0 || c0 < pack.nodes.size()); @@ -498,26 +508,26 @@ void RegularBVH::pack_aligned_node(int idx, memcpy(&pack.nodes[idx], data, sizeof(int4)*BVH_NODE_SIZE); } -void RegularBVH::pack_unaligned_inner(const BVHStackEntry& e, - const BVHStackEntry& e0, - const BVHStackEntry& e1) +void BinaryBVH::pack_unaligned_inner(const BVHStackEntry& e, + const BVHStackEntry& e0, + const BVHStackEntry& e1) { pack_unaligned_node(e.idx, e0.node->get_aligned_space(), e1.node->get_aligned_space(), - e0.node->m_bounds, - e1.node->m_bounds, + e0.node->bounds, + e1.node->bounds, e0.encodeIdx(), e1.encodeIdx(), - e0.node->m_visibility, e1.node->m_visibility); + e0.node->visibility, e1.node->visibility); } -void RegularBVH::pack_unaligned_node(int idx, - const Transform& aligned_space0, - const Transform& aligned_space1, - const BoundBox& bounds0, - const BoundBox& bounds1, - int c0, int c1, - uint visibility0, uint visibility1) +void BinaryBVH::pack_unaligned_node(int idx, + const Transform& aligned_space0, + const Transform& aligned_space1, + const BoundBox& bounds0, + const BoundBox& bounds1, + int c0, int c1, + uint visibility0, uint visibility1) { assert(idx + BVH_UNALIGNED_NODE_SIZE <= pack.nodes.size()); assert(c0 < 0 || c0 < pack.nodes.size()); @@ -543,7 +553,7 @@ void RegularBVH::pack_unaligned_node(int idx, memcpy(&pack.nodes[idx], data, sizeof(float4)*BVH_UNALIGNED_NODE_SIZE); } -void RegularBVH::pack_nodes(const BVHNode *root) +void BinaryBVH::pack_nodes(const BVHNode *root) { const size_t num_nodes = root->getSubtreeSize(BVH_STAT_NODE_COUNT); const size_t num_leaf_nodes = root->getSubtreeSize(BVH_STAT_LEAF_COUNT); @@ -620,7 +630,7 @@ void RegularBVH::pack_nodes(const BVHNode *root) pack.root_index = (root->is_leaf())? -1: 0; } -void RegularBVH::refit_nodes() +void BinaryBVH::refit_nodes() { assert(!params.top_level); @@ -629,7 +639,7 @@ void RegularBVH::refit_nodes() refit_node(0, (pack.root_index == -1)? true: false, bbox, visibility); } -void RegularBVH::refit_node(int idx, bool leaf, BoundBox& bbox, uint& visibility) +void BinaryBVH::refit_node(int idx, bool leaf, BoundBox& bbox, uint& visibility) { if(leaf) { assert(idx + BVH_NODE_LEAF_SIZE <= pack.leaf_nodes.size()); @@ -759,18 +769,18 @@ static bool node_qbvh_is_unaligned(const BVHNode *node) *node1 = node->get_child(1); bool has_unaligned = false; if(node0->is_leaf()) { - has_unaligned |= node0->is_unaligned(); + has_unaligned |= node0->is_unaligned; } else { - has_unaligned |= node0->get_child(0)->is_unaligned(); - has_unaligned |= node0->get_child(1)->is_unaligned(); + has_unaligned |= node0->get_child(0)->is_unaligned; + has_unaligned |= node0->get_child(1)->is_unaligned; } if(node1->is_leaf()) { - has_unaligned |= node1->is_unaligned(); + has_unaligned |= node1->is_unaligned; } else { - has_unaligned |= node1->get_child(0)->is_unaligned(); - has_unaligned |= node1->get_child(1)->is_unaligned(); + has_unaligned |= node1->get_child(0)->is_unaligned; + has_unaligned |= node1->get_child(1)->is_unaligned; } return has_unaligned; } @@ -785,19 +795,19 @@ void QBVH::pack_leaf(const BVHStackEntry& e, const LeafNode *leaf) { float4 data[BVH_QNODE_LEAF_SIZE]; memset(data, 0, sizeof(data)); - if(leaf->num_triangles() == 1 && pack.prim_index[leaf->m_lo] == -1) { + if(leaf->num_triangles() == 1 && pack.prim_index[leaf->lo] == -1) { /* object */ - data[0].x = __int_as_float(~(leaf->m_lo)); + data[0].x = __int_as_float(~(leaf->lo)); data[0].y = __int_as_float(0); } else { /* triangle */ - data[0].x = __int_as_float(leaf->m_lo); - data[0].y = __int_as_float(leaf->m_hi); + data[0].x = __int_as_float(leaf->lo); + data[0].y = __int_as_float(leaf->hi); } - data[0].z = __uint_as_float(leaf->m_visibility); + data[0].z = __uint_as_float(leaf->visibility); if(leaf->num_triangles() != 0) { - data[0].w = __uint_as_float(pack.prim_type[leaf->m_lo]); + data[0].w = __uint_as_float(pack.prim_type[leaf->lo]); } memcpy(&pack.leaf_nodes[e.idx], data, sizeof(float4)*BVH_QNODE_LEAF_SIZE); @@ -813,7 +823,7 @@ void QBVH::pack_inner(const BVHStackEntry& e, */ if(params.use_unaligned_nodes) { for(int i = 0; i < num; i++) { - if(en[i].node->is_unaligned()) { + if(en[i].node->is_unaligned) { has_unaligned = true; break; } @@ -838,15 +848,15 @@ void QBVH::pack_aligned_inner(const BVHStackEntry& e, BoundBox bounds[4]; int child[4]; for(int i = 0; i < num; ++i) { - bounds[i] = en[i].node->m_bounds; + bounds[i] = en[i].node->bounds; child[i] = en[i].encodeIdx(); } pack_aligned_node(e.idx, bounds, child, - e.node->m_visibility, - e.node->m_time_from, - e.node->m_time_to, + e.node->visibility, + e.node->time_from, + e.node->time_to, num); } @@ -907,16 +917,16 @@ void QBVH::pack_unaligned_inner(const BVHStackEntry& e, int child[4]; for(int i = 0; i < num; ++i) { aligned_space[i] = en[i].node->get_aligned_space(); - bounds[i] = en[i].node->m_bounds; + bounds[i] = en[i].node->bounds; child[i] = en[i].encodeIdx(); } pack_unaligned_node(e.idx, aligned_space, bounds, child, - e.node->m_visibility, - e.node->m_time_from, - e.node->m_time_to, + e.node->visibility, + e.node->time_from, + e.node->time_to, num); } diff --git a/intern/cycles/bvh/bvh.h b/intern/cycles/bvh/bvh.h index 35f4d305883..60bc62ee6e4 100644 --- a/intern/cycles/bvh/bvh.h +++ b/intern/cycles/bvh/bvh.h @@ -18,10 +18,10 @@ #ifndef __BVH_H__ #define __BVH_H__ -#include "bvh_params.h" +#include "bvh/bvh_params.h" -#include "util_types.h" -#include "util_vector.h" +#include "util/util_types.h" +#include "util/util_vector.h" CCL_NAMESPACE_BEGIN @@ -68,6 +68,8 @@ struct PackedBVH { array<int> prim_index; /* mapping from BVH primitive index, to the object id of that primitive. */ array<int> prim_object; + /* Time range of BVH primitive. */ + array<float2> prim_time; /* index of the root node. */ int root_index; @@ -108,15 +110,15 @@ protected: virtual void refit_nodes() = 0; }; -/* Regular BVH +/* Binary BVH * * Typical BVH with each node having two children. */ -class RegularBVH : public BVH { +class BinaryBVH : public BVH { protected: /* constructor */ friend class BVH; - RegularBVH(const BVHParams& params, const vector<Object*>& objects); + BinaryBVH(const BVHParams& params, const vector<Object*>& objects); /* pack */ void pack_nodes(const BVHNode *root); diff --git a/intern/cycles/bvh/bvh_binning.cpp b/intern/cycles/bvh/bvh_binning.cpp index 5ddd7349f7b..3226008f511 100644 --- a/intern/cycles/bvh/bvh_binning.cpp +++ b/intern/cycles/bvh/bvh_binning.cpp @@ -19,11 +19,11 @@ #include <stdlib.h> -#include "bvh_binning.h" +#include "bvh/bvh_binning.h" -#include "util_algorithm.h" -#include "util_boundbox.h" -#include "util_types.h" +#include "util/util_algorithm.h" +#include "util/util_boundbox.h" +#include "util/util_types.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/bvh/bvh_binning.h b/intern/cycles/bvh/bvh_binning.h index 52955f70151..285f9c56a62 100644 --- a/intern/cycles/bvh/bvh_binning.h +++ b/intern/cycles/bvh/bvh_binning.h @@ -18,10 +18,10 @@ #ifndef __BVH_BINNING_H__ #define __BVH_BINNING_H__ -#include "bvh_params.h" -#include "bvh_unaligned.h" +#include "bvh/bvh_params.h" +#include "bvh/bvh_unaligned.h" -#include "util_types.h" +#include "util/util_types.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/bvh/bvh_build.cpp b/intern/cycles/bvh/bvh_build.cpp index a2f8b33cb0b..95c71b54da0 100644 --- a/intern/cycles/bvh/bvh_build.cpp +++ b/intern/cycles/bvh/bvh_build.cpp @@ -15,26 +15,26 @@ * limitations under the License. */ -#include "bvh_binning.h" -#include "bvh_build.h" -#include "bvh_node.h" -#include "bvh_params.h" +#include "bvh/bvh_binning.h" +#include "bvh/bvh_build.h" +#include "bvh/bvh_node.h" +#include "bvh/bvh_params.h" #include "bvh_split.h" -#include "mesh.h" -#include "object.h" -#include "scene.h" -#include "curves.h" - -#include "util_algorithm.h" -#include "util_debug.h" -#include "util_foreach.h" -#include "util_logging.h" -#include "util_progress.h" -#include "util_stack_allocator.h" -#include "util_simd.h" -#include "util_time.h" -#include "util_queue.h" +#include "render/mesh.h" +#include "render/object.h" +#include "render/scene.h" +#include "render/curves.h" + +#include "util/util_algorithm.h" +#include "util/util_debug.h" +#include "util/util_foreach.h" +#include "util/util_logging.h" +#include "util/util_progress.h" +#include "util/util_stack_allocator.h" +#include "util/util_simd.h" +#include "util/util_time.h" +#include "util/util_queue.h" CCL_NAMESPACE_BEGIN @@ -93,12 +93,14 @@ BVHBuild::BVHBuild(const vector<Object*>& objects_, array<int>& prim_type_, array<int>& prim_index_, array<int>& prim_object_, + array<float2>& prim_time_, const BVHParams& params_, Progress& progress_) : objects(objects_), prim_type(prim_type_), prim_index(prim_index_), prim_object(prim_object_), + prim_time(prim_time_), params(params_), progress(progress_), progress_start_time(0.0), @@ -465,6 +467,9 @@ BVHNode* BVHBuild::run() } spatial_free_index = 0; + need_prim_time = params.num_motion_curve_steps > 0 || + params.num_motion_triangle_steps > 0; + /* init progress updates */ double build_start_time; build_start_time = progress_start_time = time_dt(); @@ -475,6 +480,12 @@ BVHNode* BVHBuild::run() prim_type.resize(references.size()); prim_index.resize(references.size()); prim_object.resize(references.size()); + if(need_prim_time) { + prim_time.resize(references.size()); + } + else { + prim_time.resize(0); + } /* build recursively */ BVHNode *rootnode; @@ -849,11 +860,14 @@ BVHNode *BVHBuild::create_object_leaf_nodes(const BVHReference *ref, int start, prim_type[start] = ref->prim_type(); prim_index[start] = ref->prim_index(); prim_object[start] = ref->prim_object(); + if(need_prim_time) { + prim_time[start] = make_float2(ref->time_from(), ref->time_to()); + } uint visibility = objects[ref->prim_object()]->visibility; BVHNode *leaf_node = new LeafNode(ref->bounds(), visibility, start, start+1); - leaf_node->m_time_from = ref->time_from(); - leaf_node->m_time_to = ref->time_to(); + leaf_node->time_from = ref->time_from(); + leaf_node->time_to = ref->time_to(); return leaf_node; } else { @@ -862,12 +876,12 @@ BVHNode *BVHBuild::create_object_leaf_nodes(const BVHReference *ref, int start, BVHNode *leaf1 = create_object_leaf_nodes(ref+mid, start+mid, num-mid); BoundBox bounds = BoundBox::empty; - bounds.grow(leaf0->m_bounds); - bounds.grow(leaf1->m_bounds); + bounds.grow(leaf0->bounds); + bounds.grow(leaf1->bounds); BVHNode *inner_node = new InnerNode(bounds, leaf0, leaf1); - inner_node->m_time_from = min(leaf0->m_time_from, leaf1->m_time_from); - inner_node->m_time_to = max(leaf0->m_time_to, leaf1->m_time_to); + inner_node->time_from = min(leaf0->time_from, leaf1->time_from); + inner_node->time_to = max(leaf0->time_to, leaf1->time_to); return inner_node; } } @@ -891,11 +905,13 @@ BVHNode* BVHBuild::create_leaf_node(const BVHRange& range, * can not control. */ typedef StackAllocator<256, int> LeafStackAllocator; + typedef StackAllocator<256, float2> LeafTimeStackAllocator; typedef StackAllocator<256, BVHReference> LeafReferenceStackAllocator; vector<int, LeafStackAllocator> p_type[PRIMITIVE_NUM_TOTAL]; vector<int, LeafStackAllocator> p_index[PRIMITIVE_NUM_TOTAL]; vector<int, LeafStackAllocator> p_object[PRIMITIVE_NUM_TOTAL]; + vector<float2, LeafTimeStackAllocator> p_time[PRIMITIVE_NUM_TOTAL]; vector<BVHReference, LeafReferenceStackAllocator> p_ref[PRIMITIVE_NUM_TOTAL]; /* TODO(sergey): In theory we should be able to store references. */ @@ -918,6 +934,8 @@ BVHNode* BVHBuild::create_leaf_node(const BVHRange& range, p_type[type_index].push_back(ref.prim_type()); p_index[type_index].push_back(ref.prim_index()); p_object[type_index].push_back(ref.prim_object()); + p_time[type_index].push_back(make_float2(ref.time_from(), + ref.time_to())); bounds[type_index].grow(ref.bounds()); visibility[type_index] |= objects[ref.prim_object()]->visibility; @@ -947,9 +965,13 @@ BVHNode* BVHBuild::create_leaf_node(const BVHRange& range, vector<int, LeafStackAllocator> local_prim_type, local_prim_index, local_prim_object; + vector<float2, LeafTimeStackAllocator> local_prim_time; local_prim_type.resize(num_new_prims); local_prim_index.resize(num_new_prims); local_prim_object.resize(num_new_prims); + if(need_prim_time) { + local_prim_time.resize(num_new_prims); + } for(int i = 0; i < PRIMITIVE_NUM_TOTAL; ++i) { int num = (int)p_type[i].size(); if(num != 0) { @@ -962,6 +984,9 @@ BVHNode* BVHBuild::create_leaf_node(const BVHRange& range, local_prim_type[index] = p_type[i][j]; local_prim_index[index] = p_index[i][j]; local_prim_object[index] = p_object[i][j]; + if(need_prim_time) { + local_prim_time[index] = p_time[i][j]; + } if(params.use_unaligned_nodes && !alignment_found) { alignment_found = unaligned_heuristic.compute_aligned_space(p_ref[i][j], @@ -979,19 +1004,19 @@ BVHNode* BVHBuild::create_leaf_node(const BVHRange& range, time_from = min(time_from, ref.time_from()); time_to = max(time_to, ref.time_to()); } - leaf_node->m_time_from = time_from; - leaf_node->m_time_to = time_to; + leaf_node->time_from = time_from; + leaf_node->time_to = time_to; } if(alignment_found) { /* Need to recalculate leaf bounds with new alignment. */ - leaf_node->m_bounds = BoundBox::empty; + leaf_node->bounds = BoundBox::empty; for(int j = 0; j < num; ++j) { const BVHReference &ref = p_ref[i][j]; BoundBox ref_bounds = unaligned_heuristic.compute_aligned_prim_boundbox( ref, aligned_space); - leaf_node->m_bounds.grow(ref_bounds); + leaf_node->bounds.grow(ref_bounds); } /* Set alignment space. */ leaf_node->set_aligned_space(aligned_space); @@ -1028,11 +1053,17 @@ BVHNode* BVHBuild::create_leaf_node(const BVHRange& range, prim_type.reserve(reserve); prim_index.reserve(reserve); prim_object.reserve(reserve); + if(need_prim_time) { + prim_time.reserve(reserve); + } } prim_type.resize(range_end); prim_index.resize(range_end); prim_object.resize(range_end); + if(need_prim_time) { + prim_time.resize(range_end); + } } spatial_spin_lock.unlock(); @@ -1041,6 +1072,9 @@ BVHNode* BVHBuild::create_leaf_node(const BVHRange& range, memcpy(&prim_type[start_index], &local_prim_type[0], new_leaf_data_size); memcpy(&prim_index[start_index], &local_prim_index[0], new_leaf_data_size); memcpy(&prim_object[start_index], &local_prim_object[0], new_leaf_data_size); + if(need_prim_time) { + memcpy(&prim_time[start_index], &local_prim_time[0], sizeof(float2)*num_new_leaf_data); + } } } else { @@ -1053,6 +1087,9 @@ BVHNode* BVHBuild::create_leaf_node(const BVHRange& range, memcpy(&prim_type[start_index], &local_prim_type[0], new_leaf_data_size); memcpy(&prim_index[start_index], &local_prim_index[0], new_leaf_data_size); memcpy(&prim_object[start_index], &local_prim_object[0], new_leaf_data_size); + if(need_prim_time) { + memcpy(&prim_time[start_index], &local_prim_time[0], sizeof(float2)*num_new_leaf_data); + } } } @@ -1062,8 +1099,8 @@ BVHNode* BVHBuild::create_leaf_node(const BVHRange& range, */ for(int i = 0; i < num_leaves; ++i) { LeafNode *leaf = (LeafNode *)leaves[i]; - leaf->m_lo += start_index; - leaf->m_hi += start_index; + leaf->lo += start_index; + leaf->hi += start_index; } /* Create leaf node for object. */ @@ -1092,17 +1129,17 @@ BVHNode* BVHBuild::create_leaf_node(const BVHRange& range, return new InnerNode(range.bounds(), leaves[0], leaves[1]); } else if(num_leaves == 3) { - BoundBox inner_bounds = merge(leaves[1]->m_bounds, leaves[2]->m_bounds); + BoundBox inner_bounds = merge(leaves[1]->bounds, leaves[2]->bounds); BVHNode *inner = new InnerNode(inner_bounds, leaves[1], leaves[2]); return new InnerNode(range.bounds(), leaves[0], inner); } else { /* Should be doing more branches if more primitive types added. */ assert(num_leaves <= 5); - BoundBox inner_bounds_a = merge(leaves[0]->m_bounds, leaves[1]->m_bounds); - BoundBox inner_bounds_b = merge(leaves[2]->m_bounds, leaves[3]->m_bounds); + BoundBox inner_bounds_a = merge(leaves[0]->bounds, leaves[1]->bounds); + BoundBox inner_bounds_b = merge(leaves[2]->bounds, leaves[3]->bounds); BVHNode *inner_a = new InnerNode(inner_bounds_a, leaves[0], leaves[1]); BVHNode *inner_b = new InnerNode(inner_bounds_b, leaves[2], leaves[3]); - BoundBox inner_bounds_c = merge(inner_a->m_bounds, inner_b->m_bounds); + BoundBox inner_bounds_c = merge(inner_a->bounds, inner_b->bounds); BVHNode *inner_c = new InnerNode(inner_bounds_c, inner_a, inner_b); if(num_leaves == 5) { return new InnerNode(range.bounds(), inner_c, leaves[4]); @@ -1137,8 +1174,8 @@ void BVHBuild::rotate(BVHNode *node, int max_depth) rotate(parent->children[c], max_depth-1); /* compute current area of all children */ - BoundBox bounds0 = parent->children[0]->m_bounds; - BoundBox bounds1 = parent->children[1]->m_bounds; + BoundBox bounds0 = parent->children[0]->bounds; + BoundBox bounds1 = parent->children[1]->bounds; float area0 = bounds0.half_area(); float area1 = bounds1.half_area(); @@ -1158,8 +1195,8 @@ void BVHBuild::rotate(BVHNode *node, int max_depth) BoundBox& other = (c == 0)? bounds1: bounds0; /* transpose child bounds */ - BoundBox target0 = child->children[0]->m_bounds; - BoundBox target1 = child->children[1]->m_bounds; + BoundBox target0 = child->children[0]->bounds; + BoundBox target1 = child->children[1]->bounds; /* compute cost for both possible swaps */ float cost0 = merge(other, target1).half_area() - child_area[c]; @@ -1191,7 +1228,7 @@ void BVHBuild::rotate(BVHNode *node, int max_depth) InnerNode *child = (InnerNode*)parent->children[best_child]; swap(parent->children[best_other], child->children[best_target]); - child->m_bounds = merge(child->children[0]->m_bounds, child->children[1]->m_bounds); + child->bounds = merge(child->children[0]->bounds, child->children[1]->bounds); } CCL_NAMESPACE_END diff --git a/intern/cycles/bvh/bvh_build.h b/intern/cycles/bvh/bvh_build.h index ee3cde66a2f..5733708050d 100644 --- a/intern/cycles/bvh/bvh_build.h +++ b/intern/cycles/bvh/bvh_build.h @@ -20,13 +20,13 @@ #include <float.h> -#include "bvh.h" -#include "bvh_binning.h" -#include "bvh_unaligned.h" +#include "bvh/bvh.h" +#include "bvh/bvh_binning.h" +#include "bvh/bvh_unaligned.h" -#include "util_boundbox.h" -#include "util_task.h" -#include "util_vector.h" +#include "util/util_boundbox.h" +#include "util/util_task.h" +#include "util/util_vector.h" CCL_NAMESPACE_BEGIN @@ -48,6 +48,7 @@ public: array<int>& prim_type, array<int>& prim_index, array<int>& prim_object, + array<float2>& prim_time, const BVHParams& params, Progress& progress); ~BVHBuild(); @@ -112,6 +113,9 @@ protected: array<int>& prim_type; array<int>& prim_index; array<int>& prim_object; + array<float2>& prim_time; + + bool need_prim_time; /* Build parameters. */ BVHParams params; diff --git a/intern/cycles/bvh/bvh_node.cpp b/intern/cycles/bvh/bvh_node.cpp index 67580e1bc7b..4f788c66797 100644 --- a/intern/cycles/bvh/bvh_node.cpp +++ b/intern/cycles/bvh/bvh_node.cpp @@ -15,12 +15,12 @@ * limitations under the License. */ -#include "bvh.h" -#include "bvh_build.h" -#include "bvh_node.h" +#include "bvh/bvh.h" +#include "bvh/bvh_build.h" +#include "bvh/bvh_node.h" -#include "util_debug.h" -#include "util_vector.h" +#include "util/util_debug.h" +#include "util/util_vector.h" CCL_NAMESPACE_BEGIN @@ -62,12 +62,12 @@ int BVHNode::getSubtreeSize(BVH_STAT stat) const } return cnt; case BVH_STAT_ALIGNED_COUNT: - if(!is_unaligned()) { + if(!is_unaligned) { cnt = 1; } break; case BVH_STAT_UNALIGNED_COUNT: - if(is_unaligned()) { + if(is_unaligned) { cnt = 1; } break; @@ -75,7 +75,7 @@ int BVHNode::getSubtreeSize(BVH_STAT stat) const if(!is_leaf()) { bool has_unaligned = false; for(int j = 0; j < num_children(); j++) { - has_unaligned |= get_child(j)->is_unaligned(); + has_unaligned |= get_child(j)->is_unaligned; } cnt += has_unaligned? 0: 1; } @@ -84,7 +84,7 @@ int BVHNode::getSubtreeSize(BVH_STAT stat) const if(!is_leaf()) { bool has_unaligned = false; for(int j = 0; j < num_children(); j++) { - has_unaligned |= get_child(j)->is_unaligned(); + has_unaligned |= get_child(j)->is_unaligned; } cnt += has_unaligned? 1: 0; } @@ -95,12 +95,12 @@ int BVHNode::getSubtreeSize(BVH_STAT stat) const for(int i = 0; i < num_children(); i++) { BVHNode *node = get_child(i); if(node->is_leaf()) { - has_unaligned |= node->is_unaligned(); + has_unaligned |= node->is_unaligned; } else { for(int j = 0; j < node->num_children(); j++) { cnt += node->get_child(j)->getSubtreeSize(stat); - has_unaligned |= node->get_child(j)->is_unaligned(); + has_unaligned |= node->get_child(j)->is_unaligned; } } } @@ -113,12 +113,12 @@ int BVHNode::getSubtreeSize(BVH_STAT stat) const for(int i = 0; i < num_children(); i++) { BVHNode *node = get_child(i); if(node->is_leaf()) { - has_unaligned |= node->is_unaligned(); + has_unaligned |= node->is_unaligned; } else { for(int j = 0; j < node->num_children(); j++) { cnt += node->get_child(j)->getSubtreeSize(stat); - has_unaligned |= node->get_child(j)->is_unaligned(); + has_unaligned |= node->get_child(j)->is_unaligned; } } } @@ -126,10 +126,10 @@ int BVHNode::getSubtreeSize(BVH_STAT stat) const } return cnt; case BVH_STAT_ALIGNED_LEAF_COUNT: - cnt = (is_leaf() && !is_unaligned()) ? 1 : 0; + cnt = (is_leaf() && !is_unaligned) ? 1 : 0; break; case BVH_STAT_UNALIGNED_LEAF_COUNT: - cnt = (is_leaf() && is_unaligned()) ? 1 : 0; + cnt = (is_leaf() && is_unaligned) ? 1 : 0; break; default: assert(0); /* unknown mode */ @@ -157,7 +157,7 @@ float BVHNode::computeSubtreeSAHCost(const BVHParams& p, float probability) cons for(int i = 0; i < num_children(); i++) { BVHNode *child = get_child(i); - SAH += child->computeSubtreeSAHCost(p, probability * child->m_bounds.safe_area()/m_bounds.safe_area()); + SAH += child->computeSubtreeSAHCost(p, probability * child->bounds.safe_area()/bounds.safe_area()); } return SAH; @@ -165,15 +165,15 @@ float BVHNode::computeSubtreeSAHCost(const BVHParams& p, float probability) cons uint BVHNode::update_visibility() { - if(!is_leaf() && m_visibility == 0) { + if(!is_leaf() && visibility == 0) { InnerNode *inner = (InnerNode*)this; BVHNode *child0 = inner->children[0]; BVHNode *child1 = inner->children[1]; - m_visibility = child0->update_visibility()|child1->update_visibility(); + visibility = child0->update_visibility()|child1->update_visibility(); } - return m_visibility; + return visibility; } void BVHNode::update_time() @@ -184,8 +184,8 @@ void BVHNode::update_time() BVHNode *child1 = inner->children[1]; child0->update_time(); child1->update_time(); - m_time_from = min(child0->m_time_from, child1->m_time_from); - m_time_to = max(child0->m_time_to, child1->m_time_to); + time_from = min(child0->time_from, child1->time_from); + time_to = max(child0->time_to, child1->time_to); } } @@ -209,7 +209,7 @@ void LeafNode::print(int depth) const for(int i = 0; i < depth; i++) printf(" "); - printf("leaf node %d to %d\n", m_lo, m_hi); + printf("leaf node %d to %d\n", lo, hi); } CCL_NAMESPACE_END diff --git a/intern/cycles/bvh/bvh_node.h b/intern/cycles/bvh/bvh_node.h index 090c426de56..60511b4b012 100644 --- a/intern/cycles/bvh/bvh_node.h +++ b/intern/cycles/bvh/bvh_node.h @@ -18,9 +18,9 @@ #ifndef __BVH_NODE_H__ #define __BVH_NODE_H__ -#include "util_boundbox.h" -#include "util_debug.h" -#include "util_types.h" +#include "util/util_boundbox.h" +#include "util/util_debug.h" +#include "util/util_types.h" CCL_NAMESPACE_BEGIN @@ -46,16 +46,16 @@ class BVHParams; class BVHNode { public: - BVHNode() : m_is_unaligned(false), - m_aligned_space(NULL), - m_time_from(0.0f), - m_time_to(1.0f) + BVHNode() : is_unaligned(false), + aligned_space(NULL), + time_from(0.0f), + time_to(1.0f) { } virtual ~BVHNode() { - delete m_aligned_space; + delete aligned_space; } virtual bool is_leaf() const = 0; @@ -63,30 +63,26 @@ public: virtual BVHNode *get_child(int i) const = 0; virtual int num_triangles() const { return 0; } virtual void print(int depth = 0) const = 0; - bool is_unaligned() const { return m_is_unaligned; } inline void set_aligned_space(const Transform& aligned_space) { - m_is_unaligned = true; - if(m_aligned_space == NULL) { - m_aligned_space = new Transform(aligned_space); + is_unaligned = true; + if(this->aligned_space == NULL) { + this->aligned_space = new Transform(aligned_space); } else { - *m_aligned_space = aligned_space; + *this->aligned_space = aligned_space; } } inline Transform get_aligned_space() const { - if(m_aligned_space == NULL) { + if(aligned_space == NULL) { return transform_identity(); } - return *m_aligned_space; + return *aligned_space; } - BoundBox m_bounds; - uint m_visibility; - // Subtree functions int getSubtreeSize(BVH_STAT stat=BVH_STAT_NODE_COUNT) const; float computeSubtreeSAHCost(const BVHParams& p, float probability = 1.0f) const; @@ -95,13 +91,18 @@ public: uint update_visibility(); void update_time(); - bool m_is_unaligned; + // Properties. + BoundBox bounds; + uint visibility; + + bool is_unaligned; - // TODO(sergey): Can be stored as 3x3 matrix, but better to have some - // utilities and type defines in util_transform first. - Transform *m_aligned_space; + /* TODO(sergey): Can be stored as 3x3 matrix, but better to have some + * utilities and type defines in util_transform first. + */ + Transform *aligned_space; - float m_time_from, m_time_to; + float time_from, time_to; }; class InnerNode : public BVHNode @@ -111,20 +112,20 @@ public: BVHNode* child0, BVHNode* child1) { - m_bounds = bounds; + this->bounds = bounds; children[0] = child0; children[1] = child1; if(child0 && child1) - m_visibility = child0->m_visibility|child1->m_visibility; + visibility = child0->visibility|child1->visibility; else - m_visibility = 0; /* happens on build cancel */ + visibility = 0; /* happens on build cancel */ } explicit InnerNode(const BoundBox& bounds) { - m_bounds = bounds; - m_visibility = 0; + this->bounds = bounds; + visibility = 0; children[0] = NULL; children[1] = NULL; } @@ -140,12 +141,12 @@ public: class LeafNode : public BVHNode { public: - LeafNode(const BoundBox& bounds, uint visibility, int lo, int hi) + LeafNode(const BoundBox& bounds, uint visibility, int lo, int hi) + : lo(lo), + hi(hi) { - m_bounds = bounds; - m_visibility = visibility; - m_lo = lo; - m_hi = hi; + this->bounds = bounds; + this->visibility = visibility; } LeafNode(const LeafNode& s) @@ -157,14 +158,13 @@ public: bool is_leaf() const { return true; } int num_children() const { return 0; } BVHNode *get_child(int) const { return NULL; } - int num_triangles() const { return m_hi - m_lo; } + int num_triangles() const { return hi - lo; } void print(int depth) const; - int m_lo; - int m_hi; + int lo; + int hi; }; CCL_NAMESPACE_END #endif /* __BVH_NODE_H__ */ - diff --git a/intern/cycles/bvh/bvh_params.h b/intern/cycles/bvh/bvh_params.h index 65f9da1c194..9795a7a4350 100644 --- a/intern/cycles/bvh/bvh_params.h +++ b/intern/cycles/bvh/bvh_params.h @@ -18,9 +18,9 @@ #ifndef __BVH_PARAMS_H__ #define __BVH_PARAMS_H__ -#include "util_boundbox.h" +#include "util/util_boundbox.h" -#include "kernel_types.h" +#include "kernel/kernel_types.h" CCL_NAMESPACE_BEGIN @@ -104,6 +104,7 @@ public: primitive_mask = PRIMITIVE_ALL; num_motion_curve_steps = 0; + num_motion_triangle_steps = 0; } /* SAH costs */ diff --git a/intern/cycles/bvh/bvh_sort.cpp b/intern/cycles/bvh/bvh_sort.cpp index e5bcf9995bf..d29629c0279 100644 --- a/intern/cycles/bvh/bvh_sort.cpp +++ b/intern/cycles/bvh/bvh_sort.cpp @@ -15,12 +15,12 @@ * limitations under the License. */ -#include "bvh_build.h" -#include "bvh_sort.h" +#include "bvh/bvh_build.h" +#include "bvh/bvh_sort.h" -#include "util_algorithm.h" -#include "util_debug.h" -#include "util_task.h" +#include "util/util_algorithm.h" +#include "util/util_debug.h" +#include "util/util_task.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/bvh/bvh_split.cpp b/intern/cycles/bvh/bvh_split.cpp index d0d5fbe5a7a..b10d69a495d 100644 --- a/intern/cycles/bvh/bvh_split.cpp +++ b/intern/cycles/bvh/bvh_split.cpp @@ -15,14 +15,14 @@ * limitations under the License. */ -#include "bvh_build.h" -#include "bvh_split.h" -#include "bvh_sort.h" +#include "bvh/bvh_build.h" +#include "bvh/bvh_split.h" +#include "bvh/bvh_sort.h" -#include "mesh.h" -#include "object.h" +#include "render/mesh.h" +#include "render/object.h" -#include "util_algorithm.h" +#include "util/util_algorithm.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/bvh/bvh_split.h b/intern/cycles/bvh/bvh_split.h index dbdb51f1a5b..a874a118b99 100644 --- a/intern/cycles/bvh/bvh_split.h +++ b/intern/cycles/bvh/bvh_split.h @@ -18,8 +18,8 @@ #ifndef __BVH_SPLIT_H__ #define __BVH_SPLIT_H__ -#include "bvh_build.h" -#include "bvh_params.h" +#include "bvh/bvh_build.h" +#include "bvh/bvh_params.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/bvh/bvh_unaligned.cpp b/intern/cycles/bvh/bvh_unaligned.cpp index a876c670914..ef227d20ea9 100644 --- a/intern/cycles/bvh/bvh_unaligned.cpp +++ b/intern/cycles/bvh/bvh_unaligned.cpp @@ -15,17 +15,17 @@ */ -#include "bvh_unaligned.h" +#include "bvh/bvh_unaligned.h" -#include "mesh.h" -#include "object.h" +#include "render/mesh.h" +#include "render/object.h" -#include "bvh_binning.h" +#include "bvh/bvh_binning.h" #include "bvh_params.h" -#include "util_boundbox.h" -#include "util_debug.h" -#include "util_transform.h" +#include "util/util_boundbox.h" +#include "util/util_debug.h" +#include "util/util_transform.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/bvh/bvh_unaligned.h b/intern/cycles/bvh/bvh_unaligned.h index 4d0872f4a39..f41bae79e2b 100644 --- a/intern/cycles/bvh/bvh_unaligned.h +++ b/intern/cycles/bvh/bvh_unaligned.h @@ -17,7 +17,7 @@ #ifndef __BVH_UNALIGNED_H__ #define __BVH_UNALIGNED_H__ -#include "util_vector.h" +#include "util/util_vector.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/device/CMakeLists.txt b/intern/cycles/device/CMakeLists.txt index 966ff5e52ba..6ef2aa1caad 100644 --- a/intern/cycles/device/CMakeLists.txt +++ b/intern/cycles/device/CMakeLists.txt @@ -1,12 +1,6 @@ set(INC - . - ../graph - ../kernel - ../kernel/svm - ../kernel/osl - ../util - ../render + .. ../../glew-mx ) @@ -33,6 +27,7 @@ set(SRC device_cuda.cpp device_multi.cpp device_opencl.cpp + device_split_kernel.cpp device_task.cpp ) @@ -56,6 +51,7 @@ set(SRC_HEADERS device_memory.h device_intern.h device_network.h + device_split_kernel.h device_task.h ) diff --git a/intern/cycles/device/device.cpp b/intern/cycles/device/device.cpp index 31c99f49d6d..968af447e29 100644 --- a/intern/cycles/device/device.cpp +++ b/intern/cycles/device/device.cpp @@ -17,18 +17,18 @@ #include <stdlib.h> #include <string.h> -#include "device.h" -#include "device_intern.h" - -#include "util_debug.h" -#include "util_foreach.h" -#include "util_half.h" -#include "util_math.h" -#include "util_opengl.h" -#include "util_time.h" -#include "util_types.h" -#include "util_vector.h" -#include "util_string.h" +#include "device/device.h" +#include "device/device_intern.h" + +#include "util/util_debug.h" +#include "util/util_foreach.h" +#include "util/util_half.h" +#include "util/util_math.h" +#include "util/util_opengl.h" +#include "util/util_time.h" +#include "util/util_types.h" +#include "util/util_vector.h" +#include "util/util_string.h" CCL_NAMESPACE_BEGIN @@ -48,11 +48,11 @@ std::ostream& operator <<(std::ostream &os, os << "Max nodes group: " << requested_features.max_nodes_group << std::endl; /* TODO(sergey): Decode bitflag into list of names. */ os << "Nodes features: " << requested_features.nodes_features << std::endl; - os << "Use hair: " + os << "Use Hair: " << string_from_bool(requested_features.use_hair) << std::endl; - os << "Use object motion: " + os << "Use Object Motion: " << string_from_bool(requested_features.use_object_motion) << std::endl; - os << "Use camera motion: " + os << "Use Camera Motion: " << string_from_bool(requested_features.use_camera_motion) << std::endl; os << "Use Baking: " << string_from_bool(requested_features.use_baking) << std::endl; @@ -80,7 +80,7 @@ Device::~Device() void Device::pixels_alloc(device_memory& mem) { - mem_alloc(mem, MEM_READ_WRITE); + mem_alloc("pixels", mem, MEM_READ_WRITE); } void Device::pixels_copy_from(device_memory& mem, int y, int w, int h) diff --git a/intern/cycles/device/device.h b/intern/cycles/device/device.h index ccee25ae34e..ac06e561795 100644 --- a/intern/cycles/device/device.h +++ b/intern/cycles/device/device.h @@ -19,15 +19,15 @@ #include <stdlib.h> -#include "device_memory.h" -#include "device_task.h" +#include "device/device_memory.h" +#include "device/device_task.h" -#include "util_list.h" -#include "util_stats.h" -#include "util_string.h" -#include "util_thread.h" -#include "util_types.h" -#include "util_vector.h" +#include "util/util_list.h" +#include "util/util_stats.h" +#include "util/util_string.h" +#include "util/util_thread.h" +#include "util/util_types.h" +#include "util/util_vector.h" CCL_NAMESPACE_BEGIN @@ -121,6 +121,9 @@ public: /* Use Transparent shadows */ bool use_transparent; + /* Use various shadow tricks, such as shadow catcher. */ + bool use_shadow_tricks; + DeviceRequestedFeatures() { /* TODO(sergey): Find more meaningful defaults. */ @@ -137,6 +140,7 @@ public: use_integrator_branched = false; use_patch_evaluation = false; use_transparent = false; + use_shadow_tricks = false; } bool modified(const DeviceRequestedFeatures& requested_features) @@ -153,7 +157,8 @@ public: use_volume == requested_features.use_volume && use_integrator_branched == requested_features.use_integrator_branched && use_patch_evaluation == requested_features.use_patch_evaluation && - use_transparent == requested_features.use_transparent); + use_transparent == requested_features.use_transparent && + use_shadow_tricks == requested_features.use_shadow_tricks); } /* Convert the requested features structure to a build options, @@ -194,9 +199,12 @@ public: if(!use_patch_evaluation) { build_options += " -D__NO_PATCH_EVAL__"; } - if(!use_transparent) { + if(!use_transparent && !use_volume) { build_options += " -D__NO_TRANSPARENT__"; } + if(!use_shadow_tricks) { + build_options += " -D__NO_SHADOW_TRICKS__"; + } return build_options; } }; @@ -228,13 +236,21 @@ public: DeviceInfo info; virtual const string& error_message() { return error_msg; } bool have_error() { return !error_message().empty(); } + virtual void set_error(const string& error) + { + if(!have_error()) { + error_msg = error; + } + fprintf(stderr, "%s\n", error.c_str()); + fflush(stderr); + } virtual bool show_samples() const { return false; } /* statistics */ Stats &stats; /* regular memory */ - virtual void mem_alloc(device_memory& mem, MemoryType type) = 0; + virtual void mem_alloc(const char *name, device_memory& mem, MemoryType type) = 0; virtual void mem_copy_to(device_memory& mem) = 0; virtual void mem_copy_from(device_memory& mem, int y, int w, int h, int elem) = 0; diff --git a/intern/cycles/device/device_cpu.cpp b/intern/cycles/device/device_cpu.cpp index c8e001ec2fd..2761d9488ca 100644 --- a/intern/cycles/device/device_cpu.cpp +++ b/intern/cycles/device/device_cpu.cpp @@ -20,36 +20,124 @@ /* So ImathMath is included before our kernel_cpu_compat. */ #ifdef WITH_OSL /* So no context pollution happens from indirectly included windows.h */ -# include "util_windows.h" +# include "util/util_windows.h" # include <OSL/oslexec.h> #endif -#include "device.h" -#include "device_intern.h" +#include "device/device.h" +#include "device/device_intern.h" +#include "device/device_split_kernel.h" -#include "kernel.h" -#include "kernel_compat_cpu.h" -#include "kernel_types.h" -#include "kernel_globals.h" +#include "kernel/kernel.h" +#include "kernel/kernel_compat_cpu.h" +#include "kernel/kernel_types.h" +#include "kernel/split/kernel_split_data.h" +#include "kernel/kernel_globals.h" -#include "osl_shader.h" -#include "osl_globals.h" +#include "kernel/osl/osl_shader.h" +#include "kernel/osl/osl_globals.h" -#include "buffers.h" +#include "render/buffers.h" -#include "util_debug.h" -#include "util_foreach.h" -#include "util_function.h" -#include "util_logging.h" -#include "util_opengl.h" -#include "util_progress.h" -#include "util_system.h" -#include "util_thread.h" +#include "util/util_debug.h" +#include "util/util_foreach.h" +#include "util/util_function.h" +#include "util/util_logging.h" +#include "util/util_map.h" +#include "util/util_opengl.h" +#include "util/util_progress.h" +#include "util/util_system.h" +#include "util/util_thread.h" CCL_NAMESPACE_BEGIN +class CPUDevice; + +class CPUSplitKernel : public DeviceSplitKernel { + CPUDevice *device; +public: + explicit CPUSplitKernel(CPUDevice *device); + + virtual bool enqueue_split_kernel_data_init(const KernelDimensions& dim, + RenderTile& rtile, + int num_global_elements, + device_memory& kernel_globals, + device_memory& kernel_data_, + device_memory& split_data, + device_memory& ray_state, + device_memory& queue_index, + device_memory& use_queues_flag, + device_memory& work_pool_wgs); + + virtual SplitKernelFunction* get_split_kernel_function(string kernel_name, const DeviceRequestedFeatures&); + virtual int2 split_kernel_local_size(); + virtual int2 split_kernel_global_size(device_memory& kg, device_memory& data, DeviceTask *task); + virtual uint64_t state_buffer_size(device_memory& kg, device_memory& data, size_t num_threads); +}; + class CPUDevice : public Device { + static unordered_map<string, void*> kernel_functions; + + static void register_kernel_function(const char* name, void* func) + { + kernel_functions[name] = func; + } + + static const char* get_arch_name() + { +#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_AVX2 + if(system_cpu_support_avx2()) { + return "cpu_avx2"; + } + else +#endif +#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_AVX + if(system_cpu_support_avx()) { + return "cpu_avx"; + } + else +#endif +#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_SSE41 + if(system_cpu_support_sse41()) { + return "cpu_sse41"; + } + else +#endif +#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_SSE3 + if(system_cpu_support_sse3()) { + return "cpu_sse3"; + } + else +#endif +#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_SSE2 + if(system_cpu_support_sse2()) { + return "cpu_sse2"; + } + else +#endif + { + return "cpu"; + } + } + + template<typename F> + static F get_kernel_function(string name) + { + name = string("kernel_") + get_arch_name() + "_" + name; + + unordered_map<string, void*>::iterator it = kernel_functions.find(name); + + if(it == kernel_functions.end()) { + assert(!"kernel function not found"); + return NULL; + } + + return (F)it->second; + } + + friend class CPUSplitKernel; + public: TaskPool task_pool; KernelGlobals kernel_globals; @@ -57,10 +145,15 @@ public: #ifdef WITH_OSL OSLGlobals osl_globals; #endif + + bool use_split_kernel; + + DeviceRequestedFeatures requested_features; CPUDevice(DeviceInfo& info, Stats &stats, bool background) : Device(info, stats, background) { + #ifdef WITH_OSL kernel_globals.osl = &osl_globals; #endif @@ -105,6 +198,28 @@ public: { VLOG(1) << "Will be using regular kernels."; } + + use_split_kernel = DebugFlags().cpu.split_kernel; + if(use_split_kernel) { + VLOG(1) << "Will be using split kernel."; + } + + kernel_cpu_register_functions(register_kernel_function); +#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_SSE2 + kernel_cpu_sse2_register_functions(register_kernel_function); +#endif +#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_SSE3 + kernel_cpu_sse3_register_functions(register_kernel_function); +#endif +#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_SSE41 + kernel_cpu_sse41_register_functions(register_kernel_function); +#endif +#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_AVX + kernel_cpu_avx_register_functions(register_kernel_function); +#endif +#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_AVX2 + kernel_cpu_avx2_register_functions(register_kernel_function); +#endif } ~CPUDevice() @@ -117,9 +232,20 @@ public: return (TaskScheduler::num_threads() == 1); } - void mem_alloc(device_memory& mem, MemoryType /*type*/) + void mem_alloc(const char *name, device_memory& mem, MemoryType /*type*/) { + if(name) { + VLOG(1) << "Buffer allocate: " << name << ", " + << string_human_readable_number(mem.memory_size()) << " bytes. (" + << string_human_readable_size(mem.memory_size()) << ")"; + } + mem.device_pointer = mem.data_pointer; + + if(!mem.device_pointer) { + mem.device_pointer = (device_ptr)malloc(mem.memory_size()); + } + mem.device_size = mem.memory_size(); stats.mem_alloc(mem.device_size); } @@ -144,6 +270,10 @@ public: void mem_free(device_memory& mem) { if(mem.device_pointer) { + if(!mem.data_pointer) { + free((void*)mem.device_pointer); + } + mem.device_pointer = 0; stats.mem_free(mem.device_size); mem.device_size = 0; @@ -196,8 +326,14 @@ public: void thread_run(DeviceTask *task) { - if(task->type == DeviceTask::PATH_TRACE) - thread_path_trace(*task); + if(task->type == DeviceTask::PATH_TRACE) { + if(!use_split_kernel) { + thread_path_trace(*task); + } + else { + thread_path_trace_split(*task); + } + } else if(task->type == DeviceTask::FILM_CONVERT) thread_film_convert(*task); else if(task->type == DeviceTask::SHADER) @@ -258,7 +394,7 @@ public: { path_trace_kernel = kernel_cpu_path_trace; } - + while(task.acquire_tile(this, tile)) { float *render_buffer = (float*)tile.buffer; uint *rng_state = (uint*)tile.rng_state; @@ -294,6 +430,49 @@ public: thread_kernel_globals_free(&kg); } + void thread_path_trace_split(DeviceTask& task) + { + if(task_pool.canceled()) { + if(task.need_finish_queue == false) + return; + } + + RenderTile tile; + + CPUSplitKernel split_kernel(this); + + /* allocate buffer for kernel globals */ + device_memory kgbuffer; + kgbuffer.resize(sizeof(KernelGlobals)); + mem_alloc("kernel_globals", kgbuffer, MEM_READ_WRITE); + + KernelGlobals *kg = (KernelGlobals*)kgbuffer.device_pointer; + *kg = thread_kernel_globals_init(); + + requested_features.max_closure = MAX_CLOSURE; + if(!split_kernel.load_kernels(requested_features)) { + thread_kernel_globals_free((KernelGlobals*)kgbuffer.device_pointer); + mem_free(kgbuffer); + + return; + } + + while(task.acquire_tile(this, tile)) { + device_memory data; + split_kernel.path_trace(&task, tile, kgbuffer, data); + + task.release_tile(tile); + + if(task_pool.canceled()) { + if(task.need_finish_queue == false) + break; + } + } + + thread_kernel_globals_free((KernelGlobals*)kgbuffer.device_pointer); + mem_free(kgbuffer); + } + void thread_film_convert(DeviceTask& task) { float sample_scale = 1.0f/(task.sample + 1); @@ -501,6 +680,10 @@ protected: inline void thread_kernel_globals_free(KernelGlobals *kg) { + if(kg == NULL) { + return; + } + if(kg->transparent_shadow_intersections != NULL) { free(kg->transparent_shadow_intersections); } @@ -515,8 +698,175 @@ protected: OSLShader::thread_free(kg); #endif } + + virtual bool load_kernels(DeviceRequestedFeatures& requested_features_) { + requested_features = requested_features_; + + return true; + } +}; + +/* split kernel */ + +class CPUSplitKernelFunction : public SplitKernelFunction { +public: + CPUDevice* device; + void (*func)(KernelGlobals *kg, KernelData *data); + + CPUSplitKernelFunction(CPUDevice* device) : device(device), func(NULL) {} + ~CPUSplitKernelFunction() {} + + virtual bool enqueue(const KernelDimensions& dim, device_memory& kernel_globals, device_memory& data) + { + if(!func) { + return false; + } + + KernelGlobals *kg = (KernelGlobals*)kernel_globals.device_pointer; + kg->global_size = make_int2(dim.global_size[0], dim.global_size[1]); + + for(int y = 0; y < dim.global_size[1]; y++) { + for(int x = 0; x < dim.global_size[0]; x++) { + kg->global_id = make_int2(x, y); + + func(kg, (KernelData*)data.device_pointer); + } + } + + return true; + } }; +CPUSplitKernel::CPUSplitKernel(CPUDevice *device) : DeviceSplitKernel(device), device(device) +{ +} + +bool CPUSplitKernel::enqueue_split_kernel_data_init(const KernelDimensions& dim, + RenderTile& rtile, + int num_global_elements, + device_memory& kernel_globals, + device_memory& data, + device_memory& split_data, + device_memory& ray_state, + device_memory& queue_index, + device_memory& use_queues_flags, + device_memory& work_pool_wgs) +{ + typedef void(*data_init_t)(KernelGlobals *kg, + ccl_constant KernelData *data, + ccl_global void *split_data_buffer, + int num_elements, + ccl_global char *ray_state, + ccl_global uint *rng_state, + int start_sample, + int end_sample, + int sx, int sy, int sw, int sh, int offset, int stride, + ccl_global int *Queue_index, + int queuesize, + ccl_global char *use_queues_flag, + ccl_global unsigned int *work_pool_wgs, + unsigned int num_samples, + ccl_global float *buffer); + + data_init_t data_init; + +#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_AVX2 + if(system_cpu_support_avx2()) { + data_init = kernel_cpu_avx2_data_init; + } + else +#endif +#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_AVX + if(system_cpu_support_avx()) { + data_init = kernel_cpu_avx_data_init; + } + else +#endif +#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_SSE41 + if(system_cpu_support_sse41()) { + data_init = kernel_cpu_sse41_data_init; + } + else +#endif +#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_SSE3 + if(system_cpu_support_sse3()) { + data_init = kernel_cpu_sse3_data_init; + } + else +#endif +#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_SSE2 + if(system_cpu_support_sse2()) { + data_init = kernel_cpu_sse2_data_init; + } + else +#endif + { + data_init = kernel_cpu_data_init; + } + + KernelGlobals *kg = (KernelGlobals*)kernel_globals.device_pointer; + kg->global_size = make_int2(dim.global_size[0], dim.global_size[1]); + + for(int y = 0; y < dim.global_size[1]; y++) { + for(int x = 0; x < dim.global_size[0]; x++) { + kg->global_id = make_int2(x, y); + + data_init((KernelGlobals*)kernel_globals.device_pointer, + (KernelData*)data.device_pointer, + (void*)split_data.device_pointer, + num_global_elements, + (char*)ray_state.device_pointer, + (uint*)rtile.rng_state, + rtile.start_sample, + rtile.start_sample + rtile.num_samples, + rtile.x, + rtile.y, + rtile.w, + rtile.h, + rtile.offset, + rtile.stride, + (int*)queue_index.device_pointer, + dim.global_size[0] * dim.global_size[1], + (char*)use_queues_flags.device_pointer, + (uint*)work_pool_wgs.device_pointer, + rtile.num_samples, + (float*)rtile.buffer); + } + } + + return true; +} + +SplitKernelFunction* CPUSplitKernel::get_split_kernel_function(string kernel_name, const DeviceRequestedFeatures&) +{ + CPUSplitKernelFunction *kernel = new CPUSplitKernelFunction(device); + + kernel->func = device->get_kernel_function<void(*)(KernelGlobals*, KernelData*)>(kernel_name); + if(!kernel->func) { + delete kernel; + return NULL; + } + + return kernel; +} + +int2 CPUSplitKernel::split_kernel_local_size() +{ + return make_int2(1, 1); +} + +int2 CPUSplitKernel::split_kernel_global_size(device_memory& /*kg*/, device_memory& /*data*/, DeviceTask * /*task*/) { + return make_int2(64, 1); +} + +uint64_t CPUSplitKernel::state_buffer_size(device_memory& kernel_globals, device_memory& /*data*/, size_t num_threads) { + KernelGlobals *kg = (KernelGlobals*)kernel_globals.device_pointer; + + return split_data_buffer_size(kg, num_threads); +} + +unordered_map<string, void*> CPUDevice::kernel_functions; + Device *device_cpu_create(DeviceInfo& info, Stats &stats, bool background) { return new CPUDevice(info, stats, background); diff --git a/intern/cycles/device/device_cuda.cpp b/intern/cycles/device/device_cuda.cpp index dafac6dfcb3..606494f08ed 100644 --- a/intern/cycles/device/device_cuda.cpp +++ b/intern/cycles/device/device_cuda.cpp @@ -15,32 +15,36 @@ */ #include <climits> +#include <limits.h> #include <stdio.h> #include <stdlib.h> #include <string.h> -#include "device.h" -#include "device_intern.h" +#include "device/device.h" +#include "device/device_intern.h" +#include "device/device_split_kernel.h" -#include "buffers.h" +#include "render/buffers.h" #ifdef WITH_CUDA_DYNLOAD # include "cuew.h" #else -# include "util_opengl.h" +# include "util/util_opengl.h" # include <cuda.h> # include <cudaGL.h> #endif -#include "util_debug.h" -#include "util_logging.h" -#include "util_map.h" -#include "util_md5.h" -#include "util_opengl.h" -#include "util_path.h" -#include "util_string.h" -#include "util_system.h" -#include "util_types.h" -#include "util_time.h" +#include "util/util_debug.h" +#include "util/util_logging.h" +#include "util/util_map.h" +#include "util/util_md5.h" +#include "util/util_opengl.h" +#include "util/util_path.h" +#include "util/util_string.h" +#include "util/util_system.h" +#include "util/util_types.h" +#include "util/util_time.h" + +#include "kernel/split/kernel_split_data_types.h" CCL_NAMESPACE_BEGIN @@ -78,6 +82,31 @@ int cuewCompilerVersion(void) } /* namespace */ #endif /* WITH_CUDA_DYNLOAD */ +class CUDADevice; + +class CUDASplitKernel : public DeviceSplitKernel { + CUDADevice *device; +public: + explicit CUDASplitKernel(CUDADevice *device); + + virtual uint64_t state_buffer_size(device_memory& kg, device_memory& data, size_t num_threads); + + virtual bool enqueue_split_kernel_data_init(const KernelDimensions& dim, + RenderTile& rtile, + int num_global_elements, + device_memory& kernel_globals, + device_memory& kernel_data_, + device_memory& split_data, + device_memory& ray_state, + device_memory& queue_index, + device_memory& use_queues_flag, + device_memory& work_pool_wgs); + + virtual SplitKernelFunction* get_split_kernel_function(string kernel_name, const DeviceRequestedFeatures&); + virtual int2 split_kernel_local_size(); + virtual int2 split_kernel_global_size(device_memory& kg, device_memory& data, DeviceTask *task); +}; + class CUDADevice : public Device { public: @@ -258,16 +287,21 @@ public: return DebugFlags().cuda.adaptive_compile; } + bool use_split_kernel() + { + return DebugFlags().cuda.split_kernel; + } + /* Common NVCC flags which stays the same regardless of shading model, * kernel sources md5 and only depends on compiler or compilation settings. */ string compile_kernel_get_common_cflags( - const DeviceRequestedFeatures& requested_features) + const DeviceRequestedFeatures& requested_features, bool split=false) { const int cuda_version = cuewCompilerVersion(); const int machine = system_cpu_bits(); - const string kernel_path = path_get("kernel"); - const string include = kernel_path; + const string source_path = path_get("source"); + const string include_path = source_path; string cflags = string_printf("-m%d " "--ptxas-options=\"-v\" " "--use_fast_math " @@ -276,7 +310,7 @@ public: "-I\"%s\"", machine, cuda_version, - include.c_str()); + include_path.c_str()); if(use_adaptive_compilation()) { cflags += " " + requested_features.get_build_options(); } @@ -287,6 +321,11 @@ public: #ifdef WITH_CYCLES_DEBUG cflags += " -D__KERNEL_DEBUG__"; #endif + + if(split) { + cflags += " -D__SPLIT__"; + } + return cflags; } @@ -306,21 +345,21 @@ public: cuda_error_message("CUDA nvcc compiler version could not be parsed."); return false; } - if(cuda_version < 75) { + if(cuda_version < 80) { printf("Unsupported CUDA version %d.%d detected, " - "you need CUDA 7.5 or newer.\n", + "you need CUDA 8.0 or newer.\n", major, minor); return false; } - else if(cuda_version != 75 && cuda_version != 80) { + else if(cuda_version != 80) { printf("CUDA version %d.%d detected, build may succeed but only " - "CUDA 7.5 and 8.0 are officially supported.\n", + "CUDA 8.0 is officially supported.\n", major, minor); } return true; } - string compile_kernel(const DeviceRequestedFeatures& requested_features) + string compile_kernel(const DeviceRequestedFeatures& requested_features, bool split=false) { /* Compute cubin name. */ int major, minor; @@ -329,7 +368,8 @@ public: /* Attempt to use kernel provided with Blender. */ if(!use_adaptive_compilation()) { - const string cubin = path_get(string_printf("lib/kernel_sm_%d%d.cubin", + const string cubin = path_get(string_printf(split ? "lib/kernel_split_sm_%d%d.cubin" + : "lib/kernel_sm_%d%d.cubin", major, minor)); VLOG(1) << "Testing for pre-compiled kernel " << cubin << "."; if(path_exists(cubin)) { @@ -339,18 +379,19 @@ public: } const string common_cflags = - compile_kernel_get_common_cflags(requested_features); + compile_kernel_get_common_cflags(requested_features, split); /* Try to use locally compiled kernel. */ - const string kernel_path = path_get("kernel"); - const string kernel_md5 = path_files_md5_hash(kernel_path); + const string source_path = path_get("source"); + const string kernel_md5 = path_files_md5_hash(source_path); /* We include cflags into md5 so changing cuda toolkit or changing other * compiler command line arguments makes sure cubin gets re-built. */ const string cubin_md5 = util_md5_string(kernel_md5 + common_cflags); - const string cubin_file = string_printf("cycles_kernel_sm%d%d_%s.cubin", + const string cubin_file = string_printf(split ? "cycles_kernel_split_sm%d%d_%s.cubin" + : "cycles_kernel_sm%d%d_%s.cubin", major, minor, cubin_md5.c_str()); const string cubin = path_cache_get(path_join("kernels", cubin_file)); @@ -383,9 +424,10 @@ public: return ""; } const char *nvcc = cuewCompilerPath(); - const string kernel = path_join(kernel_path, - path_join("kernels", - path_join("cuda", "kernel.cu"))); + const string kernel = path_join( + path_join(source_path, "kernel"), + path_join("kernels", + path_join("cuda", split ? "kernel_split.cu" : "kernel.cu"))); double starttime = time_dt(); printf("Compiling CUDA kernel ...\n"); @@ -433,7 +475,7 @@ public: return false; /* get kernel */ - string cubin = compile_kernel(requested_features); + string cubin = compile_kernel(requested_features, use_split_kernel()); if(cubin == "") return false; @@ -466,8 +508,14 @@ public: } } - void mem_alloc(device_memory& mem, MemoryType /*type*/) + void mem_alloc(const char *name, device_memory& mem, MemoryType /*type*/) { + if(name) { + VLOG(1) << "Buffer allocate: " << name << ", " + << string_human_readable_number(mem.memory_size()) << " bytes. (" + << string_human_readable_size(mem.memory_size()) << ")"; + } + cuda_push_context(); CUdeviceptr device_pointer; size_t size = mem.memory_size(); @@ -504,7 +552,9 @@ public: void mem_zero(device_memory& mem) { - memset((void*)mem.data_pointer, 0, mem.memory_size()); + if(mem.data_pointer) { + memset((void*)mem.data_pointer, 0, mem.memory_size()); + } cuda_push_context(); if(mem.device_pointer) @@ -617,7 +667,7 @@ public: /* Data Storage */ if(interpolation == INTERPOLATION_NONE) { if(has_bindless_textures) { - mem_alloc(mem, MEM_READ_ONLY); + mem_alloc(NULL, mem, MEM_READ_ONLY); mem_copy_to(mem); cuda_push_context(); @@ -641,7 +691,7 @@ public: cuda_pop_context(); } else { - mem_alloc(mem, MEM_READ_ONLY); + mem_alloc(NULL, mem, MEM_READ_ONLY); mem_copy_to(mem); cuda_push_context(); @@ -1258,25 +1308,48 @@ public: /* Upload Bindless Mapping */ load_bindless_mapping(); - /* keep rendering tiles until done */ - while(task->acquire_tile(this, tile)) { - int start_sample = tile.start_sample; - int end_sample = tile.start_sample + tile.num_samples; + if(!use_split_kernel()) { + /* keep rendering tiles until done */ + while(task->acquire_tile(this, tile)) { + int start_sample = tile.start_sample; + int end_sample = tile.start_sample + tile.num_samples; - for(int sample = start_sample; sample < end_sample; sample++) { - if(task->get_cancel()) { - if(task->need_finish_queue == false) - break; - } + for(int sample = start_sample; sample < end_sample; sample++) { + if(task->get_cancel()) { + if(task->need_finish_queue == false) + break; + } - path_trace(tile, sample, branched); + path_trace(tile, sample, branched); - tile.sample = sample + 1; + tile.sample = sample + 1; - task->update_progress(&tile, tile.w*tile.h); + task->update_progress(&tile, tile.w*tile.h); + } + + task->release_tile(tile); + } + } + else { + DeviceRequestedFeatures requested_features; + if(!use_adaptive_compilation()) { + requested_features.max_closure = 64; } - task->release_tile(tile); + CUDASplitKernel split_kernel(this); + split_kernel.load_kernels(requested_features); + + while(task->acquire_tile(this, tile)) { + device_memory void_buffer; + split_kernel.path_trace(task, tile, void_buffer, void_buffer); + + task->release_tile(tile); + + if(task->get_cancel()) { + if(task->need_finish_queue == false) + break; + } + } } } else if(task->type == DeviceTask::SHADER) { @@ -1329,8 +1402,223 @@ public: { task_pool.cancel(); } + + friend class CUDASplitKernelFunction; + friend class CUDASplitKernel; +}; + +/* redefine the cuda_assert macro so it can be used outside of the CUDADevice class + * now that the definition of that class is complete + */ +#undef cuda_assert +#define cuda_assert(stmt) \ + { \ + CUresult result = stmt; \ + \ + if(result != CUDA_SUCCESS) { \ + string message = string_printf("CUDA error: %s in %s", cuewErrorString(result), #stmt); \ + if(device->error_msg == "") \ + device->error_msg = message; \ + fprintf(stderr, "%s\n", message.c_str()); \ + /*cuda_abort();*/ \ + device->cuda_error_documentation(); \ + } \ + } (void)0 + +/* split kernel */ + +class CUDASplitKernelFunction : public SplitKernelFunction{ + CUDADevice* device; + CUfunction func; +public: + CUDASplitKernelFunction(CUDADevice *device, CUfunction func) : device(device), func(func) {} + + /* enqueue the kernel, returns false if there is an error */ + bool enqueue(const KernelDimensions &dim, device_memory &/*kg*/, device_memory &/*data*/) + { + return enqueue(dim, NULL); + } + + /* enqueue the kernel, returns false if there is an error */ + bool enqueue(const KernelDimensions &dim, void *args[]) + { + device->cuda_push_context(); + + if(device->have_error()) + return false; + + /* we ignore dim.local_size for now, as this is faster */ + int threads_per_block; + cuda_assert(cuFuncGetAttribute(&threads_per_block, CU_FUNC_ATTRIBUTE_MAX_THREADS_PER_BLOCK, func)); + + int xthreads = (int)sqrt(threads_per_block); + int ythreads = (int)sqrt(threads_per_block); + + int xblocks = (dim.global_size[0] + xthreads - 1)/xthreads; + int yblocks = (dim.global_size[1] + ythreads - 1)/ythreads; + + cuda_assert(cuFuncSetCacheConfig(func, CU_FUNC_CACHE_PREFER_L1)); + + cuda_assert(cuLaunchKernel(func, + xblocks , yblocks, 1, /* blocks */ + xthreads, ythreads, 1, /* threads */ + 0, 0, args, 0)); + + device->cuda_pop_context(); + + return !device->have_error(); + } }; +CUDASplitKernel::CUDASplitKernel(CUDADevice *device) : DeviceSplitKernel(device), device(device) +{ +} + +uint64_t CUDASplitKernel::state_buffer_size(device_memory& /*kg*/, device_memory& /*data*/, size_t num_threads) +{ + device_vector<uint64_t> size_buffer; + size_buffer.resize(1); + device->mem_alloc(NULL, size_buffer, MEM_READ_WRITE); + + device->cuda_push_context(); + + uint threads = num_threads; + CUdeviceptr d_size = device->cuda_device_ptr(size_buffer.device_pointer); + + struct args_t { + uint* num_threads; + CUdeviceptr* size; + }; + + args_t args = { + &threads, + &d_size + }; + + CUfunction state_buffer_size; + cuda_assert(cuModuleGetFunction(&state_buffer_size, device->cuModule, "kernel_cuda_state_buffer_size")); + + cuda_assert(cuLaunchKernel(state_buffer_size, + 1, 1, 1, + 1, 1, 1, + 0, 0, (void**)&args, 0)); + + device->cuda_pop_context(); + + device->mem_copy_from(size_buffer, 0, 1, 1, sizeof(uint64_t)); + device->mem_free(size_buffer); + + return *size_buffer.get_data(); +} + +bool CUDASplitKernel::enqueue_split_kernel_data_init(const KernelDimensions& dim, + RenderTile& rtile, + int num_global_elements, + device_memory& /*kernel_globals*/, + device_memory& /*kernel_data*/, + device_memory& split_data, + device_memory& ray_state, + device_memory& queue_index, + device_memory& use_queues_flag, + device_memory& work_pool_wgs) +{ + device->cuda_push_context(); + + CUdeviceptr d_split_data = device->cuda_device_ptr(split_data.device_pointer); + CUdeviceptr d_ray_state = device->cuda_device_ptr(ray_state.device_pointer); + CUdeviceptr d_queue_index = device->cuda_device_ptr(queue_index.device_pointer); + CUdeviceptr d_use_queues_flag = device->cuda_device_ptr(use_queues_flag.device_pointer); + CUdeviceptr d_work_pool_wgs = device->cuda_device_ptr(work_pool_wgs.device_pointer); + + CUdeviceptr d_rng_state = device->cuda_device_ptr(rtile.rng_state); + CUdeviceptr d_buffer = device->cuda_device_ptr(rtile.buffer); + + int end_sample = rtile.start_sample + rtile.num_samples; + int queue_size = dim.global_size[0] * dim.global_size[1]; + + struct args_t { + CUdeviceptr* split_data_buffer; + int* num_elements; + CUdeviceptr* ray_state; + CUdeviceptr* rng_state; + int* start_sample; + int* end_sample; + int* sx; + int* sy; + int* sw; + int* sh; + int* offset; + int* stride; + CUdeviceptr* queue_index; + int* queuesize; + CUdeviceptr* use_queues_flag; + CUdeviceptr* work_pool_wgs; + int* num_samples; + CUdeviceptr* buffer; + }; + + args_t args = { + &d_split_data, + &num_global_elements, + &d_ray_state, + &d_rng_state, + &rtile.start_sample, + &end_sample, + &rtile.x, + &rtile.y, + &rtile.w, + &rtile.h, + &rtile.offset, + &rtile.stride, + &d_queue_index, + &queue_size, + &d_use_queues_flag, + &d_work_pool_wgs, + &rtile.num_samples, + &d_buffer + }; + + CUfunction data_init; + cuda_assert(cuModuleGetFunction(&data_init, device->cuModule, "kernel_cuda_path_trace_data_init")); + if(device->have_error()) { + return false; + } + + CUDASplitKernelFunction(device, data_init).enqueue(dim, (void**)&args); + + device->cuda_pop_context(); + + return !device->have_error(); +} + +SplitKernelFunction* CUDASplitKernel::get_split_kernel_function(string kernel_name, const DeviceRequestedFeatures&) +{ + CUfunction func; + + device->cuda_push_context(); + + cuda_assert(cuModuleGetFunction(&func, device->cuModule, (string("kernel_cuda_") + kernel_name).data())); + if(device->have_error()) { + device->cuda_error_message(string_printf("kernel \"kernel_cuda_%s\" not found in module", kernel_name.data())); + return NULL; + } + + device->cuda_pop_context(); + + return new CUDASplitKernelFunction(device, func); +} + +int2 CUDASplitKernel::split_kernel_local_size() +{ + return make_int2(32, 1); +} + +int2 CUDASplitKernel::split_kernel_global_size(device_memory& /*kg*/, device_memory& /*data*/, DeviceTask */*task*/) +{ + /* TODO(mai): implement something here to detect ideal work size */ + return make_int2(256, 256); +} + bool device_cuda_init(void) { #ifdef WITH_CUDA_DYNLOAD diff --git a/intern/cycles/device/device_memory.h b/intern/cycles/device/device_memory.h index 5b5b4dc6802..4b10514a9d2 100644 --- a/intern/cycles/device/device_memory.h +++ b/intern/cycles/device/device_memory.h @@ -28,10 +28,10 @@ * other devices this is a pointer to device memory, where we will copy memory * to and from. */ -#include "util_debug.h" -#include "util_half.h" -#include "util_types.h" -#include "util_vector.h" +#include "util/util_debug.h" +#include "util/util_half.h" +#include "util/util_types.h" +#include "util/util_vector.h" CCL_NAMESPACE_BEGIN @@ -48,7 +48,8 @@ enum DataType { TYPE_UINT, TYPE_INT, TYPE_FLOAT, - TYPE_HALF + TYPE_HALF, + TYPE_UINT64, }; static inline size_t datatype_size(DataType datatype) @@ -59,6 +60,7 @@ static inline size_t datatype_size(DataType datatype) case TYPE_UINT: return sizeof(uint); case TYPE_INT: return sizeof(int); case TYPE_HALF: return sizeof(half); + case TYPE_UINT64: return sizeof(uint64_t); default: return 0; } } @@ -160,6 +162,11 @@ template<> struct device_type_traits<half4> { static const int num_elements = 4; }; +template<> struct device_type_traits<uint64_t> { + static const DataType data_type = TYPE_UINT64; + static const int num_elements = 1; +}; + /* Device Memory */ class device_memory @@ -180,10 +187,27 @@ public: /* device pointer */ device_ptr device_pointer; -protected: - device_memory() {} + device_memory() + { + data_type = device_type_traits<uchar>::data_type; + data_elements = device_type_traits<uchar>::num_elements; + data_pointer = 0; + data_size = 0; + device_size = 0; + data_width = 0; + data_height = 0; + data_depth = 0; + device_pointer = 0; + } virtual ~device_memory() { assert(!device_pointer); } + void resize(size_t size) + { + data_size = size; + data_width = size; + } + +protected: /* no copying */ device_memory(const device_memory&); device_memory& operator = (const device_memory&); @@ -198,16 +222,8 @@ public: { data_type = device_type_traits<T>::data_type; data_elements = device_type_traits<T>::num_elements; - data_pointer = 0; - data_size = 0; - device_size = 0; - data_width = 0; - data_height = 0; - data_depth = 0; assert(data_elements > 0); - - device_pointer = 0; } virtual ~device_vector() {} @@ -266,6 +282,7 @@ public: data_height = 0; data_depth = 0; data_size = 0; + device_pointer = 0; } size_t size() diff --git a/intern/cycles/device/device_multi.cpp b/intern/cycles/device/device_multi.cpp index 31b800640d3..624260a81c8 100644 --- a/intern/cycles/device/device_multi.cpp +++ b/intern/cycles/device/device_multi.cpp @@ -17,17 +17,17 @@ #include <stdlib.h> #include <sstream> -#include "device.h" -#include "device_intern.h" -#include "device_network.h" +#include "device/device.h" +#include "device/device_intern.h" +#include "device/device_network.h" -#include "buffers.h" +#include "render/buffers.h" -#include "util_foreach.h" -#include "util_list.h" -#include "util_logging.h" -#include "util_map.h" -#include "util_time.h" +#include "util/util_foreach.h" +#include "util/util_list.h" +#include "util/util_logging.h" +#include "util/util_map.h" +#include "util/util_time.h" CCL_NAMESPACE_BEGIN @@ -106,11 +106,11 @@ public: return true; } - void mem_alloc(device_memory& mem, MemoryType type) + void mem_alloc(const char *name, device_memory& mem, MemoryType type) { foreach(SubDevice& sub, devices) { mem.device_pointer = 0; - sub.device->mem_alloc(mem, type); + sub.device->mem_alloc(name, mem, type); sub.ptr_map[unique_ptr] = mem.device_pointer; } @@ -162,6 +162,7 @@ public: void mem_free(device_memory& mem) { device_ptr tmp = mem.device_pointer; + stats.mem_free(mem.device_size); foreach(SubDevice& sub, devices) { mem.device_pointer = sub.ptr_map[tmp]; @@ -170,7 +171,6 @@ public: } mem.device_pointer = 0; - stats.mem_free(mem.device_size); } void const_copy_to(const char *name, void *host, size_t size) @@ -202,6 +202,7 @@ public: void tex_free(device_memory& mem) { device_ptr tmp = mem.device_pointer; + stats.mem_free(mem.device_size); foreach(SubDevice& sub, devices) { mem.device_pointer = sub.ptr_map[tmp]; @@ -210,7 +211,6 @@ public: } mem.device_pointer = 0; - stats.mem_free(mem.device_size); } void pixels_alloc(device_memory& mem) diff --git a/intern/cycles/device/device_network.cpp b/intern/cycles/device/device_network.cpp index 53eef6cf199..66758954f44 100644 --- a/intern/cycles/device/device_network.cpp +++ b/intern/cycles/device/device_network.cpp @@ -14,12 +14,12 @@ * limitations under the License. */ -#include "device.h" -#include "device_intern.h" -#include "device_network.h" +#include "device/device.h" +#include "device/device_intern.h" +#include "device/device_network.h" -#include "util_foreach.h" -#include "util_logging.h" +#include "util/util_foreach.h" +#include "util/util_logging.h" #if defined(WITH_NETWORK) @@ -87,8 +87,14 @@ public: snd.write(); } - void mem_alloc(device_memory& mem, MemoryType type) + void mem_alloc(const char *name, device_memory& mem, MemoryType type) { + if(name) { + VLOG(1) << "Buffer allocate: " << name << ", " + << string_human_readable_number(mem.memory_size()) << " bytes. (" + << string_human_readable_size(mem.memory_size()) << ")"; + } + thread_scoped_lock lock(rpc_lock); mem.device_pointer = ++mem_counter; @@ -481,7 +487,7 @@ protected: mem.data_pointer = 0; /* perform the allocation on the actual device */ - device->mem_alloc(mem, type); + device->mem_alloc(NULL, mem, type); /* store a mapping to/from client_pointer and real device pointer */ pointer_mapping_insert(client_pointer, mem.device_pointer); diff --git a/intern/cycles/device/device_network.h b/intern/cycles/device/device_network.h index d28cfe3121f..a5d24c66018 100644 --- a/intern/cycles/device/device_network.h +++ b/intern/cycles/device/device_network.h @@ -33,12 +33,12 @@ #include <sstream> #include <deque> -#include "buffers.h" +#include "render/buffers.h" -#include "util_foreach.h" -#include "util_list.h" -#include "util_map.h" -#include "util_string.h" +#include "util/util_foreach.h" +#include "util/util_list.h" +#include "util/util_map.h" +#include "util/util_string.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/device/device_opencl.cpp b/intern/cycles/device/device_opencl.cpp index ba94c592a5f..edd2047debc 100644 --- a/intern/cycles/device/device_opencl.cpp +++ b/intern/cycles/device/device_opencl.cpp @@ -16,12 +16,12 @@ #ifdef WITH_OPENCL -#include "opencl/opencl.h" +#include "device/opencl/opencl.h" -#include "device_intern.h" +#include "device/device_intern.h" -#include "util_foreach.h" -#include "util_logging.h" +#include "util/util_foreach.h" +#include "util/util_logging.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/device/device_split_kernel.cpp b/intern/cycles/device/device_split_kernel.cpp new file mode 100644 index 00000000000..ae462a560b7 --- /dev/null +++ b/intern/cycles/device/device_split_kernel.cpp @@ -0,0 +1,306 @@ +/* + * Copyright 2011-2016 Blender Foundation + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#include "device/device_split_kernel.h" + +#include "kernel/kernel_types.h" +#include "kernel/split/kernel_split_data_types.h" + +#include "util/util_time.h" + +CCL_NAMESPACE_BEGIN + +static const double alpha = 0.1; /* alpha for rolling average */ + +DeviceSplitKernel::DeviceSplitKernel(Device *device) : device(device) +{ + current_max_closure = -1; + first_tile = true; + + avg_time_per_sample = 0.0; + + kernel_path_init = NULL; + kernel_scene_intersect = NULL; + kernel_lamp_emission = NULL; + kernel_do_volume = NULL; + kernel_queue_enqueue = NULL; + kernel_indirect_background = NULL; + kernel_shader_eval = NULL; + kernel_holdout_emission_blurring_pathtermination_ao = NULL; + kernel_subsurface_scatter = NULL; + kernel_direct_lighting = NULL; + kernel_shadow_blocked_ao = NULL; + kernel_shadow_blocked_dl = NULL; + kernel_next_iteration_setup = NULL; + kernel_indirect_subsurface = NULL; + kernel_buffer_update = NULL; +} + +DeviceSplitKernel::~DeviceSplitKernel() +{ + device->mem_free(split_data); + device->mem_free(ray_state); + device->mem_free(use_queues_flag); + device->mem_free(queue_index); + device->mem_free(work_pool_wgs); + + delete kernel_path_init; + delete kernel_scene_intersect; + delete kernel_lamp_emission; + delete kernel_do_volume; + delete kernel_queue_enqueue; + delete kernel_indirect_background; + delete kernel_shader_eval; + delete kernel_holdout_emission_blurring_pathtermination_ao; + delete kernel_subsurface_scatter; + delete kernel_direct_lighting; + delete kernel_shadow_blocked_ao; + delete kernel_shadow_blocked_dl; + delete kernel_next_iteration_setup; + delete kernel_indirect_subsurface; + delete kernel_buffer_update; +} + +bool DeviceSplitKernel::load_kernels(const DeviceRequestedFeatures& requested_features) +{ +#define LOAD_KERNEL(name) \ + kernel_##name = get_split_kernel_function(#name, requested_features); \ + if(!kernel_##name) { \ + return false; \ + } + + LOAD_KERNEL(path_init); + LOAD_KERNEL(scene_intersect); + LOAD_KERNEL(lamp_emission); + LOAD_KERNEL(do_volume); + LOAD_KERNEL(queue_enqueue); + LOAD_KERNEL(indirect_background); + LOAD_KERNEL(shader_eval); + LOAD_KERNEL(holdout_emission_blurring_pathtermination_ao); + LOAD_KERNEL(subsurface_scatter); + LOAD_KERNEL(direct_lighting); + LOAD_KERNEL(shadow_blocked_ao); + LOAD_KERNEL(shadow_blocked_dl); + LOAD_KERNEL(next_iteration_setup); + LOAD_KERNEL(indirect_subsurface); + LOAD_KERNEL(buffer_update); + +#undef LOAD_KERNEL + + current_max_closure = requested_features.max_closure; + + return true; +} + +size_t DeviceSplitKernel::max_elements_for_max_buffer_size(device_memory& kg, device_memory& data, uint64_t max_buffer_size) +{ + uint64_t size_per_element = state_buffer_size(kg, data, 1024) / 1024; + return max_buffer_size / size_per_element; +} + +bool DeviceSplitKernel::path_trace(DeviceTask *task, + RenderTile& tile, + device_memory& kgbuffer, + device_memory& kernel_data) +{ + if(device->have_error()) { + return false; + } + + /* Get local size */ + size_t local_size[2]; + { + int2 lsize = split_kernel_local_size(); + local_size[0] = lsize[0]; + local_size[1] = lsize[1]; + } + + /* Set gloabl size */ + size_t global_size[2]; + { + int2 gsize = split_kernel_global_size(kgbuffer, kernel_data, task); + + /* Make sure that set work size is a multiple of local + * work size dimensions. + */ + global_size[0] = round_up(gsize[0], local_size[0]); + global_size[1] = round_up(gsize[1], local_size[1]); + } + + /* Number of elements in the global state buffer */ + int num_global_elements = global_size[0] * global_size[1]; + assert(num_global_elements % WORK_POOL_SIZE == 0); + + /* Allocate all required global memory once. */ + if(first_tile) { + first_tile = false; + + /* Calculate max groups */ + + /* Denotes the maximum work groups possible w.r.t. current requested tile size. */ + unsigned int max_work_groups = num_global_elements / WORK_POOL_SIZE + 1; + + /* Allocate work_pool_wgs memory. */ + work_pool_wgs.resize(max_work_groups * sizeof(unsigned int)); + device->mem_alloc("work_pool_wgs", work_pool_wgs, MEM_READ_WRITE); + + queue_index.resize(NUM_QUEUES * sizeof(int)); + device->mem_alloc("queue_index", queue_index, MEM_READ_WRITE); + + use_queues_flag.resize(sizeof(char)); + device->mem_alloc("use_queues_flag", use_queues_flag, MEM_READ_WRITE); + + ray_state.resize(num_global_elements); + device->mem_alloc("ray_state", ray_state, MEM_READ_WRITE); + + split_data.resize(state_buffer_size(kgbuffer, kernel_data, num_global_elements)); + device->mem_alloc("split_data", split_data, MEM_READ_WRITE); + } + +#define ENQUEUE_SPLIT_KERNEL(name, global_size, local_size) \ + if(device->have_error()) { \ + return false; \ + } \ + if(!kernel_##name->enqueue(KernelDimensions(global_size, local_size), kgbuffer, kernel_data)) { \ + return false; \ + } + + tile.sample = tile.start_sample; + + /* for exponential increase between tile updates */ + int time_multiplier = 1; + + while(tile.sample < tile.start_sample + tile.num_samples) { + /* to keep track of how long it takes to run a number of samples */ + double start_time = time_dt(); + + /* initial guess to start rolling average */ + const int initial_num_samples = 1; + /* approx number of samples per second */ + int samples_per_second = (avg_time_per_sample > 0.0) ? + int(double(time_multiplier) / avg_time_per_sample) + 1 : initial_num_samples; + + RenderTile subtile = tile; + subtile.start_sample = tile.sample; + subtile.num_samples = min(samples_per_second, tile.start_sample + tile.num_samples - tile.sample); + + if(device->have_error()) { + return false; + } + + /* reset state memory here as global size for data_init + * kernel might not be large enough to do in kernel + */ + device->mem_zero(work_pool_wgs); + device->mem_zero(split_data); + device->mem_zero(ray_state); + + if(!enqueue_split_kernel_data_init(KernelDimensions(global_size, local_size), + subtile, + num_global_elements, + kgbuffer, + kernel_data, + split_data, + ray_state, + queue_index, + use_queues_flag, + work_pool_wgs)) + { + return false; + } + + ENQUEUE_SPLIT_KERNEL(path_init, global_size, local_size); + + bool activeRaysAvailable = true; + + while(activeRaysAvailable) { + /* Do path-iteration in host [Enqueue Path-iteration kernels. */ + for(int PathIter = 0; PathIter < 16; PathIter++) { + ENQUEUE_SPLIT_KERNEL(scene_intersect, global_size, local_size); + ENQUEUE_SPLIT_KERNEL(lamp_emission, global_size, local_size); + ENQUEUE_SPLIT_KERNEL(do_volume, global_size, local_size); + ENQUEUE_SPLIT_KERNEL(queue_enqueue, global_size, local_size); + ENQUEUE_SPLIT_KERNEL(indirect_background, global_size, local_size); + ENQUEUE_SPLIT_KERNEL(shader_eval, global_size, local_size); + ENQUEUE_SPLIT_KERNEL(holdout_emission_blurring_pathtermination_ao, global_size, local_size); + ENQUEUE_SPLIT_KERNEL(subsurface_scatter, global_size, local_size); + ENQUEUE_SPLIT_KERNEL(direct_lighting, global_size, local_size); + ENQUEUE_SPLIT_KERNEL(shadow_blocked_ao, global_size, local_size); + ENQUEUE_SPLIT_KERNEL(shadow_blocked_dl, global_size, local_size); + ENQUEUE_SPLIT_KERNEL(next_iteration_setup, global_size, local_size); + ENQUEUE_SPLIT_KERNEL(indirect_subsurface, global_size, local_size); + ENQUEUE_SPLIT_KERNEL(queue_enqueue, global_size, local_size); + ENQUEUE_SPLIT_KERNEL(buffer_update, global_size, local_size); + + if(task->get_cancel()) { + return true; + } + } + + /* Decide if we should exit path-iteration in host. */ + device->mem_copy_from(ray_state, 0, global_size[0] * global_size[1] * sizeof(char), 1, 1); + + activeRaysAvailable = false; + + for(int rayStateIter = 0; rayStateIter < global_size[0] * global_size[1]; ++rayStateIter) { + int8_t state = ray_state.get_data()[rayStateIter]; + + if(state != RAY_INACTIVE) { + if(state == RAY_INVALID) { + /* Something went wrong, abort to avoid looping endlessly. */ + device->set_error("Split kernel error: invalid ray state"); + return false; + } + + /* Not all rays are RAY_INACTIVE. */ + activeRaysAvailable = true; + break; + } + } + + if(task->get_cancel()) { + return true; + } + } + + double time_per_sample = ((time_dt()-start_time) / subtile.num_samples); + + if(avg_time_per_sample == 0.0) { + /* start rolling average */ + avg_time_per_sample = time_per_sample; + } + else { + avg_time_per_sample = alpha*time_per_sample + (1.0-alpha)*avg_time_per_sample; + } + +#undef ENQUEUE_SPLIT_KERNEL + + tile.sample += subtile.num_samples; + task->update_progress(&tile, tile.w*tile.h*subtile.num_samples); + + time_multiplier = min(time_multiplier << 1, 10); + + if(task->get_cancel()) { + return true; + } + } + + return true; +} + +CCL_NAMESPACE_END + + diff --git a/intern/cycles/device/device_split_kernel.h b/intern/cycles/device/device_split_kernel.h new file mode 100644 index 00000000000..15a94953a11 --- /dev/null +++ b/intern/cycles/device/device_split_kernel.h @@ -0,0 +1,132 @@ +/* + * Copyright 2011-2016 Blender Foundation + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#ifndef __DEVICE_SPLIT_KERNEL_H__ +#define __DEVICE_SPLIT_KERNEL_H__ + +#include "device/device.h" +#include "render/buffers.h" + +CCL_NAMESPACE_BEGIN + +/* When allocate global memory in chunks. We may not be able to + * allocate exactly "CL_DEVICE_MAX_MEM_ALLOC_SIZE" bytes in chunks; + * Since some bytes may be needed for aligning chunks of memory; + * This is the amount of memory that we dedicate for that purpose. + */ +#define DATA_ALLOCATION_MEM_FACTOR 5000000 //5MB + +/* Types used for split kernel */ + +class KernelDimensions { +public: + size_t global_size[2]; + size_t local_size[2]; + + KernelDimensions(size_t global_size_[2], size_t local_size_[2]) + { + memcpy(global_size, global_size_, sizeof(global_size)); + memcpy(local_size, local_size_, sizeof(local_size)); + } +}; + +class SplitKernelFunction { +public: + virtual ~SplitKernelFunction() {} + + /* enqueue the kernel, returns false if there is an error */ + virtual bool enqueue(const KernelDimensions& dim, device_memory& kg, device_memory& data) = 0; +}; + +class DeviceSplitKernel { +private: + Device *device; + + SplitKernelFunction *kernel_path_init; + SplitKernelFunction *kernel_scene_intersect; + SplitKernelFunction *kernel_lamp_emission; + SplitKernelFunction *kernel_do_volume; + SplitKernelFunction *kernel_queue_enqueue; + SplitKernelFunction *kernel_indirect_background; + SplitKernelFunction *kernel_shader_eval; + SplitKernelFunction *kernel_holdout_emission_blurring_pathtermination_ao; + SplitKernelFunction *kernel_subsurface_scatter; + SplitKernelFunction *kernel_direct_lighting; + SplitKernelFunction *kernel_shadow_blocked_ao; + SplitKernelFunction *kernel_shadow_blocked_dl; + SplitKernelFunction *kernel_next_iteration_setup; + SplitKernelFunction *kernel_indirect_subsurface; + SplitKernelFunction *kernel_buffer_update; + + /* Global memory variables [porting]; These memory is used for + * co-operation between different kernels; Data written by one + * kernel will be available to another kernel via this global + * memory. + */ + device_memory split_data; + device_vector<uchar> ray_state; + device_memory queue_index; /* Array of size num_queues * sizeof(int) that tracks the size of each queue. */ + + /* Flag to make sceneintersect and lampemission kernel use queues. */ + device_memory use_queues_flag; + + /* Approximate time it takes to complete one sample */ + double avg_time_per_sample; + + /* Work pool with respect to each work group. */ + device_memory work_pool_wgs; + + /* clos_max value for which the kernels have been loaded currently. */ + int current_max_closure; + + /* Marked True in constructor and marked false at the end of path_trace(). */ + bool first_tile; + +public: + explicit DeviceSplitKernel(Device* device); + virtual ~DeviceSplitKernel(); + + bool load_kernels(const DeviceRequestedFeatures& requested_features); + bool path_trace(DeviceTask *task, + RenderTile& rtile, + device_memory& kgbuffer, + device_memory& kernel_data); + + virtual uint64_t state_buffer_size(device_memory& kg, device_memory& data, size_t num_threads) = 0; + size_t max_elements_for_max_buffer_size(device_memory& kg, device_memory& data, uint64_t max_buffer_size); + + virtual bool enqueue_split_kernel_data_init(const KernelDimensions& dim, + RenderTile& rtile, + int num_global_elements, + device_memory& kernel_globals, + device_memory& kernel_data_, + device_memory& split_data, + device_memory& ray_state, + device_memory& queue_index, + device_memory& use_queues_flag, + device_memory& work_pool_wgs) = 0; + + virtual SplitKernelFunction* get_split_kernel_function(string kernel_name, const DeviceRequestedFeatures&) = 0; + virtual int2 split_kernel_local_size() = 0; + virtual int2 split_kernel_global_size(device_memory& kg, device_memory& data, DeviceTask *task) = 0; +}; + +CCL_NAMESPACE_END + +#endif /* __DEVICE_SPLIT_KERNEL_H__ */ + + + diff --git a/intern/cycles/device/device_task.cpp b/intern/cycles/device/device_task.cpp index 48d18035c13..ca303365627 100644 --- a/intern/cycles/device/device_task.cpp +++ b/intern/cycles/device/device_task.cpp @@ -17,12 +17,12 @@ #include <stdlib.h> #include <string.h> -#include "device_task.h" +#include "device/device_task.h" -#include "buffers.h" +#include "render/buffers.h" -#include "util_algorithm.h" -#include "util_time.h" +#include "util/util_algorithm.h" +#include "util/util_time.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/device/device_task.h b/intern/cycles/device/device_task.h index 8bd54c3d2b0..feee89fd6e4 100644 --- a/intern/cycles/device/device_task.h +++ b/intern/cycles/device/device_task.h @@ -17,11 +17,11 @@ #ifndef __DEVICE_TASK_H__ #define __DEVICE_TASK_H__ -#include "device_memory.h" +#include "device/device_memory.h" -#include "util_function.h" -#include "util_list.h" -#include "util_task.h" +#include "util/util_function.h" +#include "util/util_list.h" +#include "util/util_task.h" CCL_NAMESPACE_BEGIN @@ -51,6 +51,8 @@ public: int shader_filter; int shader_x, shader_w; + int passes_size; + explicit DeviceTask(Type type = PATH_TRACE); int get_subtask_count(int num, int max_size = 0); diff --git a/intern/cycles/device/opencl/opencl.h b/intern/cycles/device/opencl/opencl.h index 4023ba89a10..764216d0dfa 100644 --- a/intern/cycles/device/opencl/opencl.h +++ b/intern/cycles/device/opencl/opencl.h @@ -16,39 +16,39 @@ #ifdef WITH_OPENCL -#include "device.h" +#include "device/device.h" -#include "util_map.h" -#include "util_param.h" -#include "util_string.h" +#include "util/util_map.h" +#include "util/util_param.h" +#include "util/util_string.h" #include "clew.h" CCL_NAMESPACE_BEGIN -#define CL_MEM_PTR(p) ((cl_mem)(uintptr_t)(p)) - -/* Macro declarations used with split kernel */ - -/* Macro to enable/disable work-stealing */ -#define __WORK_STEALING__ - -#define SPLIT_KERNEL_LOCAL_SIZE_X 64 -#define SPLIT_KERNEL_LOCAL_SIZE_Y 1 +/* Define CYCLES_DISABLE_DRIVER_WORKAROUNDS to disable workaounds for testing */ +#ifndef CYCLES_DISABLE_DRIVER_WORKAROUNDS +/* Work around AMD driver hangs by ensuring each command is finished before doing anything else. */ +# undef clEnqueueNDRangeKernel +# define clEnqueueNDRangeKernel(a, b, c, d, e, f, g, h, i) \ + clFinish(a); \ + CLEW_GET_FUN(__clewEnqueueNDRangeKernel)(a, b, c, d, e, f, g, h, i); \ + clFinish(a); + +# undef clEnqueueWriteBuffer +# define clEnqueueWriteBuffer(a, b, c, d, e, f, g, h, i) \ + clFinish(a); \ + CLEW_GET_FUN(__clewEnqueueWriteBuffer)(a, b, c, d, e, f, g, h, i); \ + clFinish(a); + +# undef clEnqueueReadBuffer +# define clEnqueueReadBuffer(a, b, c, d, e, f, g, h, i) \ + clFinish(a); \ + CLEW_GET_FUN(__clewEnqueueReadBuffer)(a, b, c, d, e, f, g, h, i); \ + clFinish(a); +#endif /* CYCLES_DISABLE_DRIVER_WORKAROUNDS */ -/* This value may be tuned according to the scene we are rendering. - * - * Modifying PATH_ITER_INC_FACTOR value proportional to number of expected - * ray-bounces will improve performance. - */ -#define PATH_ITER_INC_FACTOR 8 - -/* When allocate global memory in chunks. We may not be able to - * allocate exactly "CL_DEVICE_MAX_MEM_ALLOC_SIZE" bytes in chunks; - * Since some bytes may be needed for aligning chunks of memory; - * This is the amount of memory that we dedicate for that purpose. - */ -#define DATA_ALLOCATION_MEM_FACTOR 5000000 //5MB +#define CL_MEM_PTR(p) ((cl_mem)(uintptr_t)(p)) struct OpenCLPlatformDevice { OpenCLPlatformDevice(cl_platform_id platform_id, @@ -90,6 +90,54 @@ public: cl_device_id device_id); static void get_usable_devices(vector<OpenCLPlatformDevice> *usable_devices, bool force_all = false); + static bool use_single_program(); + + /* ** Some handy shortcuts to low level cl*GetInfo() functions. ** */ + + /* Platform information. */ + static bool get_num_platforms(cl_uint *num_platforms, cl_int *error = NULL); + static cl_uint get_num_platforms(); + + static bool get_platforms(vector<cl_platform_id> *platform_ids, + cl_int *error = NULL); + static vector<cl_platform_id> get_platforms(); + + static bool get_platform_name(cl_platform_id platform_id, + string *platform_name); + static string get_platform_name(cl_platform_id platform_id); + + static bool get_num_platform_devices(cl_platform_id platform_id, + cl_device_type device_type, + cl_uint *num_devices, + cl_int *error = NULL); + static cl_uint get_num_platform_devices(cl_platform_id platform_id, + cl_device_type device_type); + + static bool get_platform_devices(cl_platform_id platform_id, + cl_device_type device_type, + vector<cl_device_id> *device_ids, + cl_int* error = NULL); + static vector<cl_device_id> get_platform_devices(cl_platform_id platform_id, + cl_device_type device_type); + + /* Device information. */ + static bool get_device_name(cl_device_id device_id, + string *device_name, + cl_int* error = NULL); + + static string get_device_name(cl_device_id device_id); + + static bool get_device_type(cl_device_id device_id, + cl_device_type *device_type, + cl_int* error = NULL); + static cl_device_type get_device_type(cl_device_id device_id); + + /* Get somewhat more readable device name. + * Main difference is AMD OpenCL here which only gives code name + * for the regular device name. This will give more sane device + * name using some extensions. + */ + static string get_readable_device_name(cl_device_id device_id); }; /* Thread safe cache for contexts and programs. @@ -248,6 +296,7 @@ public: bool device_initialized; string platform_name; + string device_name; bool opencl_error(cl_int err); void opencl_error(const string& message); @@ -266,10 +315,10 @@ public: /* Has to be implemented by the real device classes. * The base device will then load all these programs. */ - virtual void load_kernels(const DeviceRequestedFeatures& requested_features, + virtual bool load_kernels(const DeviceRequestedFeatures& requested_features, vector<OpenCLProgram*> &programs) = 0; - void mem_alloc(device_memory& mem, MemoryType type); + void mem_alloc(const char *name, device_memory& mem, MemoryType type); void mem_copy_to(device_memory& mem); void mem_copy_from(device_memory& mem, int y, int w, int h, int elem); void mem_zero(device_memory& mem); @@ -326,16 +375,39 @@ protected: class ArgumentWrapper { public: - ArgumentWrapper() : size(0), pointer(NULL) {} - template <typename T> + ArgumentWrapper() : size(0), pointer(NULL) + { + } + + ArgumentWrapper(device_memory& argument) : size(sizeof(void*)), + pointer((void*)(&argument.device_pointer)) + { + } + + template<typename T> + ArgumentWrapper(device_vector<T>& argument) : size(sizeof(void*)), + pointer((void*)(&argument.device_pointer)) + { + } + + template<typename T> ArgumentWrapper(T& argument) : size(sizeof(argument)), - pointer(&argument) { } + pointer(&argument) + { + } + ArgumentWrapper(int argument) : size(sizeof(int)), int_value(argument), - pointer(&int_value) { } + pointer(&int_value) + { + } + ArgumentWrapper(float argument) : size(sizeof(float)), float_value(argument), - pointer(&float_value) { } + pointer(&float_value) + { + } + size_t size; int int_value; float float_value; diff --git a/intern/cycles/device/opencl/opencl_base.cpp b/intern/cycles/device/opencl/opencl_base.cpp index a2b900312e7..52d0662a8e3 100644 --- a/intern/cycles/device/opencl/opencl_base.cpp +++ b/intern/cycles/device/opencl/opencl_base.cpp @@ -16,15 +16,15 @@ #ifdef WITH_OPENCL -#include "opencl.h" +#include "device/opencl/opencl.h" -#include "kernel_types.h" +#include "kernel/kernel_types.h" -#include "util_foreach.h" -#include "util_logging.h" -#include "util_md5.h" -#include "util_path.h" -#include "util_time.h" +#include "util/util_foreach.h" +#include "util/util_logging.h" +#include "util/util_md5.h" +#include "util/util_path.h" +#include "util/util_time.h" CCL_NAMESPACE_BEGIN @@ -82,9 +82,10 @@ OpenCLDeviceBase::OpenCLDeviceBase(DeviceInfo& info, Stats &stats, bool backgrou cpPlatform = platform_device.platform_id; cdDevice = platform_device.device_id; platform_name = platform_device.platform_name; + device_name = platform_device.device_name; VLOG(2) << "Creating new Cycles device for OpenCL platform " << platform_name << ", device " - << platform_device.device_name << "."; + << device_name << "."; { /* try to use cached context */ @@ -113,12 +114,16 @@ OpenCLDeviceBase::OpenCLDeviceBase(DeviceInfo& info, Stats &stats, bool backgrou } cqCommandQueue = clCreateCommandQueue(cxContext, cdDevice, 0, &ciErr); - if(opencl_error(ciErr)) + if(opencl_error(ciErr)) { + opencl_error("OpenCL: Error creating command queue"); return; + } null_mem = (device_ptr)clCreateBuffer(cxContext, CL_MEM_READ_ONLY, 1, NULL, &ciErr); - if(opencl_error(ciErr)) + if(opencl_error(ciErr)) { + opencl_error("OpenCL: Error creating memory buffer for NULL"); return; + } fprintf(stderr, "Device init success\n"); device_initialized = true; @@ -147,10 +152,8 @@ OpenCLDeviceBase::~OpenCLDeviceBase() void CL_CALLBACK OpenCLDeviceBase::context_notify_callback(const char *err_info, const void * /*private_info*/, size_t /*cb*/, void *user_data) { - char name[256]; - clGetDeviceInfo((cl_device_id)user_data, CL_DEVICE_NAME, sizeof(name), &name, NULL); - - fprintf(stderr, "OpenCL error (%s): %s\n", name, err_info); + string device_name = OpenCLInfo::get_device_name((cl_device_id)user_data); + fprintf(stderr, "OpenCL error (%s): %s\n", device_name.c_str(), err_info); } bool OpenCLDeviceBase::opencl_version_check() @@ -191,6 +194,8 @@ string OpenCLDeviceBase::device_md5_hash(string kernel_custom_build_options) bool OpenCLDeviceBase::load_kernels(const DeviceRequestedFeatures& requested_features) { + VLOG(2) << "Loading kernels for platform " << platform_name + << ", device " << device_name << "."; /* Verify if device was initialized. */ if(!device_initialized) { fprintf(stderr, "OpenCL: failed to initialize device.\n"); @@ -206,11 +211,14 @@ bool OpenCLDeviceBase::load_kernels(const DeviceRequestedFeatures& requested_fea base_program.add_kernel(ustring("convert_to_half_float")); base_program.add_kernel(ustring("shader")); base_program.add_kernel(ustring("bake")); + base_program.add_kernel(ustring("zero_buffer")); vector<OpenCLProgram*> programs; programs.push_back(&base_program); /* Call actual class to fill the vector with its programs. */ - load_kernels(requested_features, programs); + if(!load_kernels(requested_features, programs)) { + return false; + } /* Parallel compilation is supported by Cycles, but currently all OpenCL frameworks * serialize the calls internally, so it's not much use right now. @@ -242,8 +250,14 @@ bool OpenCLDeviceBase::load_kernels(const DeviceRequestedFeatures& requested_fea return true; } -void OpenCLDeviceBase::mem_alloc(device_memory& mem, MemoryType type) +void OpenCLDeviceBase::mem_alloc(const char *name, device_memory& mem, MemoryType type) { + if(name) { + VLOG(1) << "Buffer allocate: " << name << ", " + << string_human_readable_number(mem.memory_size()) << " bytes. (" + << string_human_readable_size(mem.memory_size()) << ")"; + } + size_t size = mem.memory_size(); cl_mem_flags mem_flag; @@ -311,8 +325,61 @@ void OpenCLDeviceBase::mem_copy_from(device_memory& mem, int y, int w, int h, in void OpenCLDeviceBase::mem_zero(device_memory& mem) { if(mem.device_pointer) { - memset((void*)mem.data_pointer, 0, mem.memory_size()); - mem_copy_to(mem); + if(base_program.is_loaded()) { + cl_kernel ckZeroBuffer = base_program(ustring("zero_buffer")); + + size_t global_size[] = {1024, 1024}; + size_t num_threads = global_size[0] * global_size[1]; + + cl_mem d_buffer = CL_MEM_PTR(mem.device_pointer); + cl_ulong d_offset = 0; + cl_ulong d_size = 0; + + while(d_offset < mem.memory_size()) { + d_size = std::min<cl_ulong>(num_threads*sizeof(float4), mem.memory_size() - d_offset); + + kernel_set_args(ckZeroBuffer, 0, d_buffer, d_size, d_offset); + + ciErr = clEnqueueNDRangeKernel(cqCommandQueue, + ckZeroBuffer, + 2, + NULL, + global_size, + NULL, + 0, + NULL, + NULL); + opencl_assert_err(ciErr, "clEnqueueNDRangeKernel"); + + d_offset += d_size; + } + } + + if(mem.data_pointer) { + memset((void*)mem.data_pointer, 0, mem.memory_size()); + } + + if(!base_program.is_loaded()) { + void* zero = (void*)mem.data_pointer; + + if(!mem.data_pointer) { + zero = util_aligned_malloc(mem.memory_size(), 16); + memset(zero, 0, mem.memory_size()); + } + + opencl_assert(clEnqueueWriteBuffer(cqCommandQueue, + CL_MEM_PTR(mem.device_pointer), + CL_TRUE, + 0, + mem.memory_size(), + zero, + 0, + NULL, NULL)); + + if(!mem.data_pointer) { + util_aligned_free(zero); + } + } } } @@ -337,7 +404,7 @@ void OpenCLDeviceBase::const_copy_to(const char *name, void *host, size_t size) device_vector<uchar> *data = new device_vector<uchar>(); data->copy((uchar*)host, size); - mem_alloc(*data, MEM_READ_ONLY); + mem_alloc(name, *data, MEM_READ_ONLY); i = const_mem_map.insert(ConstMemMap::value_type(name, data)).first; } else { @@ -356,7 +423,7 @@ void OpenCLDeviceBase::tex_alloc(const char *name, VLOG(1) << "Texture allocate: " << name << ", " << string_human_readable_number(mem.memory_size()) << " bytes. (" << string_human_readable_size(mem.memory_size()) << ")"; - mem_alloc(mem, MEM_READ_ONLY); + mem_alloc(NULL, mem, MEM_READ_ONLY); mem_copy_to(mem); assert(mem_map.find(name) == mem_map.end()); mem_map.insert(MemMap::value_type(name, mem.device_pointer)); @@ -460,7 +527,7 @@ void OpenCLDeviceBase::film_convert(DeviceTask& task, device_ptr buffer, device_ #define KERNEL_TEX(type, ttype, name) \ set_kernel_arg_mem(ckFilmConvertKernel, &start_arg_index, #name); -#include "kernel_textures.h" +#include "kernel/kernel_textures.h" #undef KERNEL_TEX start_arg_index += kernel_set_args(ckFilmConvertKernel, @@ -511,7 +578,7 @@ void OpenCLDeviceBase::shader(DeviceTask& task) #define KERNEL_TEX(type, ttype, name) \ set_kernel_arg_mem(kernel, &start_arg_index, #name); -#include "kernel_textures.h" +#include "kernel/kernel_textures.h" #undef KERNEL_TEX start_arg_index += kernel_set_args(kernel, diff --git a/intern/cycles/device/opencl/opencl_mega.cpp b/intern/cycles/device/opencl/opencl_mega.cpp index 6ea7619e022..a2fd1d71156 100644 --- a/intern/cycles/device/opencl/opencl_mega.cpp +++ b/intern/cycles/device/opencl/opencl_mega.cpp @@ -16,15 +16,15 @@ #ifdef WITH_OPENCL -#include "opencl.h" +#include "device/opencl/opencl.h" -#include "buffers.h" +#include "render/buffers.h" -#include "kernel_types.h" +#include "kernel/kernel_types.h" -#include "util_md5.h" -#include "util_path.h" -#include "util_time.h" +#include "util/util_md5.h" +#include "util/util_path.h" +#include "util/util_time.h" CCL_NAMESPACE_BEGIN @@ -43,11 +43,12 @@ public: return true; } - virtual void load_kernels(const DeviceRequestedFeatures& /*requested_features*/, + virtual bool load_kernels(const DeviceRequestedFeatures& /*requested_features*/, vector<OpenCLProgram*> &programs) { path_trace_program.add_kernel(ustring("path_trace")); programs.push_back(&path_trace_program); + return true; } ~OpenCLDeviceMegaKernel() @@ -83,7 +84,7 @@ public: #define KERNEL_TEX(type, ttype, name) \ set_kernel_arg_mem(ckPathTraceKernel, &start_arg_index, #name); -#include "kernel_textures.h" +#include "kernel/kernel_textures.h" #undef KERNEL_TEX start_arg_index += kernel_set_args(ckPathTraceKernel, diff --git a/intern/cycles/device/opencl/opencl_split.cpp b/intern/cycles/device/opencl/opencl_split.cpp index 3c3c2150128..579dbc84f53 100644 --- a/intern/cycles/device/opencl/opencl_split.cpp +++ b/intern/cycles/device/opencl/opencl_split.cpp @@ -16,1290 +16,359 @@ #ifdef WITH_OPENCL -#include "opencl.h" +#include "device/opencl/opencl.h" -#include "buffers.h" +#include "render/buffers.h" -#include "kernel_types.h" +#include "kernel/kernel_types.h" +#include "kernel/split/kernel_split_data_types.h" -#include "util_md5.h" -#include "util_path.h" -#include "util_time.h" +#include "device/device_split_kernel.h" + +#include "util/util_logging.h" +#include "util/util_md5.h" +#include "util/util_path.h" +#include "util/util_time.h" CCL_NAMESPACE_BEGIN -/* TODO(sergey): This is to keep tile split on OpenCL level working - * for now, since without this view-port render does not work as it - * should. - * - * Ideally it'll be done on the higher level, but we need to get ready - * for merge rather soon, so let's keep split logic private here in - * the file. - */ -class SplitRenderTile : public RenderTile { -public: - SplitRenderTile() - : RenderTile(), - buffer_offset_x(0), - buffer_offset_y(0), - rng_state_offset_x(0), - rng_state_offset_y(0), - buffer_rng_state_stride(0) {} - - explicit SplitRenderTile(RenderTile& tile) - : RenderTile(), - buffer_offset_x(0), - buffer_offset_y(0), - rng_state_offset_x(0), - rng_state_offset_y(0), - buffer_rng_state_stride(0) - { - x = tile.x; - y = tile.y; - w = tile.w; - h = tile.h; - start_sample = tile.start_sample; - num_samples = tile.num_samples; - sample = tile.sample; - resolution = tile.resolution; - offset = tile.offset; - stride = tile.stride; - buffer = tile.buffer; - rng_state = tile.rng_state; - buffers = tile.buffers; +class OpenCLSplitKernel; + +static string get_build_options(OpenCLDeviceBase *device, const DeviceRequestedFeatures& requested_features) +{ + string build_options = "-D__SPLIT_KERNEL__ "; + build_options += requested_features.get_build_options(); + + /* Set compute device build option. */ + cl_device_type device_type; + OpenCLInfo::get_device_type(device->cdDevice, &device_type, &device->ciErr); + assert(device->ciErr == CL_SUCCESS); + if(device_type == CL_DEVICE_TYPE_GPU) { + build_options += " -D__COMPUTE_DEVICE_GPU__"; } - /* Split kernel is device global memory constrained; - * hence split kernel cant render big tile size's in - * one go. If the user sets a big tile size (big tile size - * is a term relative to the available device global memory), - * we split the tile further and then call path_trace on - * each of those split tiles. The following variables declared, - * assist in achieving that purpose - */ - int buffer_offset_x; - int buffer_offset_y; - int rng_state_offset_x; - int rng_state_offset_y; - int buffer_rng_state_stride; -}; + return build_options; +} /* OpenCLDeviceSplitKernel's declaration/definition. */ class OpenCLDeviceSplitKernel : public OpenCLDeviceBase { public: - /* Kernel declaration. */ + DeviceSplitKernel *split_kernel; OpenCLProgram program_data_init; - OpenCLProgram program_scene_intersect; - OpenCLProgram program_lamp_emission; - OpenCLProgram program_queue_enqueue; - OpenCLProgram program_background_buffer_update; - OpenCLProgram program_shader_eval; - OpenCLProgram program_holdout_emission_blurring_pathtermination_ao; - OpenCLProgram program_direct_lighting; - OpenCLProgram program_shadow_blocked; - OpenCLProgram program_next_iteration_setup; - OpenCLProgram program_sum_all_radiance; - - /* Global memory variables [porting]; These memory is used for - * co-operation between different kernels; Data written by one - * kernel will be available to another kernel via this global - * memory. - */ - cl_mem rng_coop; - cl_mem throughput_coop; - cl_mem L_transparent_coop; - cl_mem PathRadiance_coop; - cl_mem Ray_coop; - cl_mem PathState_coop; - cl_mem Intersection_coop; - cl_mem kgbuffer; /* KernelGlobals buffer. */ - - /* Global buffers for ShaderData. */ - cl_mem sd; /* ShaderData used in the main path-iteration loop. */ - cl_mem sd_DL_shadow; /* ShaderData used in Direct Lighting and - * shadow_blocked kernel. - */ - - /* Global memory required for shadow blocked and accum_radiance. */ - cl_mem BSDFEval_coop; - cl_mem ISLamp_coop; - cl_mem LightRay_coop; - cl_mem AOAlpha_coop; - cl_mem AOBSDF_coop; - cl_mem AOLightRay_coop; - cl_mem Intersection_coop_shadow; - -#ifdef WITH_CYCLES_DEBUG - /* DebugData memory */ - cl_mem debugdata_coop; -#endif - - /* Global state array that tracks ray state. */ - cl_mem ray_state; - - /* Per sample buffers. */ - cl_mem per_sample_output_buffers; - - /* Denotes which sample each ray is being processed for. */ - cl_mem work_array; - - /* Queue */ - cl_mem Queue_data; /* Array of size queuesize * num_queues * sizeof(int). */ - cl_mem Queue_index; /* Array of size num_queues * sizeof(int); - * Tracks the size of each queue. - */ - - /* Flag to make sceneintersect and lampemission kernel use queues. */ - cl_mem use_queues_flag; - - /* Amount of memory in output buffer associated with one pixel/thread. */ - size_t per_thread_output_buffer_size; - - /* Total allocatable available device memory. */ - size_t total_allocatable_memory; - - /* host version of ray_state; Used in checking host path-iteration - * termination. - */ - char *hostRayStateArray; - - /* Number of path-iterations to be done in one shot. */ - unsigned int PathIteration_times; - -#ifdef __WORK_STEALING__ - /* Work pool with respect to each work group. */ - cl_mem work_pool_wgs; - - /* Denotes the maximum work groups possible w.r.t. current tile size. */ - unsigned int max_work_groups; -#endif - - /* clos_max value for which the kernels have been loaded currently. */ - int current_max_closure; - - /* Marked True in constructor and marked false at the end of path_trace(). */ - bool first_tile; - - OpenCLDeviceSplitKernel(DeviceInfo& info, Stats &stats, bool background_) - : OpenCLDeviceBase(info, stats, background_) - { - background = background_; - - /* Initialize cl_mem variables. */ - kgbuffer = NULL; - sd = NULL; - sd_DL_shadow = NULL; - - rng_coop = NULL; - throughput_coop = NULL; - L_transparent_coop = NULL; - PathRadiance_coop = NULL; - Ray_coop = NULL; - PathState_coop = NULL; - Intersection_coop = NULL; - ray_state = NULL; - - AOAlpha_coop = NULL; - AOBSDF_coop = NULL; - AOLightRay_coop = NULL; - BSDFEval_coop = NULL; - ISLamp_coop = NULL; - LightRay_coop = NULL; - Intersection_coop_shadow = NULL; - -#ifdef WITH_CYCLES_DEBUG - debugdata_coop = NULL; -#endif - - work_array = NULL; - - /* Queue. */ - Queue_data = NULL; - Queue_index = NULL; - use_queues_flag = NULL; - - per_sample_output_buffers = NULL; - - per_thread_output_buffer_size = 0; - hostRayStateArray = NULL; - PathIteration_times = PATH_ITER_INC_FACTOR; -#ifdef __WORK_STEALING__ - work_pool_wgs = NULL; - max_work_groups = 0; -#endif - current_max_closure = -1; - first_tile = true; - - /* Get device's maximum memory that can be allocated. */ - ciErr = clGetDeviceInfo(cdDevice, - CL_DEVICE_MAX_MEM_ALLOC_SIZE, - sizeof(size_t), - &total_allocatable_memory, - NULL); - assert(ciErr == CL_SUCCESS); - if(platform_name == "AMD Accelerated Parallel Processing") { - /* This value is tweak-able; AMD platform does not seem to - * give maximum performance when all of CL_DEVICE_MAX_MEM_ALLOC_SIZE - * is considered for further computation. - */ - total_allocatable_memory /= 2; - } - } + OpenCLProgram program_state_buffer_size; - virtual bool show_samples() const { - return false; - } + OpenCLDeviceSplitKernel(DeviceInfo& info, Stats &stats, bool background_); - /* Split kernel utility functions. */ - size_t get_tex_size(const char *tex_name) + ~OpenCLDeviceSplitKernel() { - cl_mem ptr; - size_t ret_size = 0; - MemMap::iterator i = mem_map.find(tex_name); - if(i != mem_map.end()) { - ptr = CL_MEM_PTR(i->second); - ciErr = clGetMemObjectInfo(ptr, - CL_MEM_SIZE, - sizeof(ret_size), - &ret_size, - NULL); - assert(ciErr == CL_SUCCESS); - } - return ret_size; + task_pool.stop(); + + /* Release kernels */ + program_data_init.release(); + + delete split_kernel; } - size_t get_shader_data_size(size_t max_closure) + virtual bool load_kernels(const DeviceRequestedFeatures& requested_features, + vector<OpenCLDeviceBase::OpenCLProgram*> &programs) { - /* ShaderData size with variable size ShaderClosure array */ - return sizeof(ShaderData) - (sizeof(ShaderClosure) * (MAX_CLOSURE - max_closure)); + bool single_program = OpenCLInfo::use_single_program(); + program_data_init = OpenCLDeviceBase::OpenCLProgram(this, + single_program ? "split" : "split_data_init", + single_program ? "kernel_split.cl" : "kernel_data_init.cl", + get_build_options(this, requested_features)); + + program_data_init.add_kernel(ustring("path_trace_data_init")); + programs.push_back(&program_data_init); + + program_state_buffer_size = OpenCLDeviceBase::OpenCLProgram(this, + single_program ? "split" : "split_state_buffer_size", + single_program ? "kernel_split.cl" : "kernel_state_buffer_size.cl", + get_build_options(this, requested_features)); + program_state_buffer_size.add_kernel(ustring("path_trace_state_buffer_size")); + programs.push_back(&program_state_buffer_size); + + return split_kernel->load_kernels(requested_features); } - /* Returns size of KernelGlobals structure associated with OpenCL. */ - size_t get_KernelGlobals_size() + void thread_run(DeviceTask *task) { - /* Copy dummy KernelGlobals related to OpenCL from kernel_globals.h to - * fetch its size. - */ - typedef struct KernelGlobals { - ccl_constant KernelData *data; + if(task->type == DeviceTask::FILM_CONVERT) { + film_convert(*task, task->buffer, task->rgba_byte, task->rgba_half); + } + else if(task->type == DeviceTask::SHADER) { + shader(*task); + } + else if(task->type == DeviceTask::PATH_TRACE) { + RenderTile tile; + + /* Copy dummy KernelGlobals related to OpenCL from kernel_globals.h to + * fetch its size. + */ + typedef struct KernelGlobals { + ccl_constant KernelData *data; #define KERNEL_TEX(type, ttype, name) \ - ccl_global type *name; -#include "kernel_textures.h" + ccl_global type *name; +#include "kernel/kernel_textures.h" #undef KERNEL_TEX - void *sd_input; - void *isect_shadow; - } KernelGlobals; + SplitData split_data; + SplitParams split_param_data; + } KernelGlobals; + + /* Allocate buffer for kernel globals */ + device_memory kgbuffer; + kgbuffer.resize(sizeof(KernelGlobals)); + mem_alloc("kernel_globals", kgbuffer, MEM_READ_WRITE); + + /* Keep rendering tiles until done. */ + while(task->acquire_tile(this, tile)) { + split_kernel->path_trace(task, + tile, + kgbuffer, + *const_mem_map["__data"]); + + /* Complete kernel execution before release tile. */ + /* This helps in multi-device render; + * The device that reaches the critical-section function + * release_tile waits (stalling other devices from entering + * release_tile) for all kernels to complete. If device1 (a + * slow-render device) reaches release_tile first then it would + * stall device2 (a fast-render device) from proceeding to render + * next tile. + */ + clFinish(cqCommandQueue); - return sizeof(KernelGlobals); + task->release_tile(tile); + } + + mem_free(kgbuffer); + } + } + +protected: + /* ** Those guys are for workign around some compiler-specific bugs ** */ + + string build_options_for_base_program( + const DeviceRequestedFeatures& requested_features) + { + return requested_features.get_build_options(); } - virtual void load_kernels(const DeviceRequestedFeatures& requested_features, - vector<OpenCLProgram*> &programs) + friend class OpenCLSplitKernel; + friend class OpenCLSplitKernelFunction; +}; + +class OpenCLSplitKernelFunction : public SplitKernelFunction { +public: + OpenCLDeviceSplitKernel* device; + OpenCLDeviceBase::OpenCLProgram program; + + OpenCLSplitKernelFunction(OpenCLDeviceSplitKernel* device) : device(device) {} + ~OpenCLSplitKernelFunction() { program.release(); } + + virtual bool enqueue(const KernelDimensions& dim, device_memory& kg, device_memory& data) { - string build_options = "-D__SPLIT_KERNEL__ "; -#ifdef __WORK_STEALING__ - build_options += "-D__WORK_STEALING__ "; -#endif - build_options += requested_features.get_build_options(); - - /* Set compute device build option. */ - cl_device_type device_type; - ciErr = clGetDeviceInfo(cdDevice, - CL_DEVICE_TYPE, - sizeof(cl_device_type), - &device_type, - NULL); - assert(ciErr == CL_SUCCESS); - if(device_type == CL_DEVICE_TYPE_GPU) { - build_options += " -D__COMPUTE_DEVICE_GPU__"; + device->kernel_set_args(program(), 0, kg, data); + + device->ciErr = clEnqueueNDRangeKernel(device->cqCommandQueue, + program(), + 2, + NULL, + dim.global_size, + dim.local_size, + 0, + NULL, + NULL); + + device->opencl_assert_err(device->ciErr, "clEnqueueNDRangeKernel"); + + if(device->ciErr != CL_SUCCESS) { + string message = string_printf("OpenCL error: %s in clEnqueueNDRangeKernel()", + clewErrorString(device->ciErr)); + device->opencl_error(message); + return false; } -#define GLUE(a, b) a ## b -#define LOAD_KERNEL(name) \ - do { \ - GLUE(program_, name) = OpenCLProgram(this, "split_" #name, "kernel_" #name ".cl", build_options); \ - GLUE(program_, name).add_kernel(ustring("path_trace_" #name)); \ - programs.push_back(&GLUE(program_, name)); \ - } while(false) - - LOAD_KERNEL(data_init); - LOAD_KERNEL(scene_intersect); - LOAD_KERNEL(lamp_emission); - LOAD_KERNEL(queue_enqueue); - LOAD_KERNEL(background_buffer_update); - LOAD_KERNEL(shader_eval); - LOAD_KERNEL(holdout_emission_blurring_pathtermination_ao); - LOAD_KERNEL(direct_lighting); - LOAD_KERNEL(shadow_blocked); - LOAD_KERNEL(next_iteration_setup); - LOAD_KERNEL(sum_all_radiance); - -#undef FIND_KERNEL -#undef GLUE - - current_max_closure = requested_features.max_closure; + return true; } +}; - ~OpenCLDeviceSplitKernel() +class OpenCLSplitKernel : public DeviceSplitKernel { + OpenCLDeviceSplitKernel *device; +public: + explicit OpenCLSplitKernel(OpenCLDeviceSplitKernel *device) : DeviceSplitKernel(device), device(device) { + } + + virtual SplitKernelFunction* get_split_kernel_function(string kernel_name, + const DeviceRequestedFeatures& requested_features) { - task_pool.stop(); + OpenCLSplitKernelFunction* kernel = new OpenCLSplitKernelFunction(device); - /* Release kernels */ - program_data_init.release(); - program_scene_intersect.release(); - program_lamp_emission.release(); - program_queue_enqueue.release(); - program_background_buffer_update.release(); - program_shader_eval.release(); - program_holdout_emission_blurring_pathtermination_ao.release(); - program_direct_lighting.release(); - program_shadow_blocked.release(); - program_next_iteration_setup.release(); - program_sum_all_radiance.release(); - - /* Release global memory */ - release_mem_object_safe(rng_coop); - release_mem_object_safe(throughput_coop); - release_mem_object_safe(L_transparent_coop); - release_mem_object_safe(PathRadiance_coop); - release_mem_object_safe(Ray_coop); - release_mem_object_safe(PathState_coop); - release_mem_object_safe(Intersection_coop); - release_mem_object_safe(kgbuffer); - release_mem_object_safe(sd); - release_mem_object_safe(sd_DL_shadow); - release_mem_object_safe(ray_state); - release_mem_object_safe(AOAlpha_coop); - release_mem_object_safe(AOBSDF_coop); - release_mem_object_safe(AOLightRay_coop); - release_mem_object_safe(BSDFEval_coop); - release_mem_object_safe(ISLamp_coop); - release_mem_object_safe(LightRay_coop); - release_mem_object_safe(Intersection_coop_shadow); -#ifdef WITH_CYCLES_DEBUG - release_mem_object_safe(debugdata_coop); -#endif - release_mem_object_safe(use_queues_flag); - release_mem_object_safe(Queue_data); - release_mem_object_safe(Queue_index); - release_mem_object_safe(work_array); -#ifdef __WORK_STEALING__ - release_mem_object_safe(work_pool_wgs); -#endif - release_mem_object_safe(per_sample_output_buffers); - - if(hostRayStateArray != NULL) { - free(hostRayStateArray); + bool single_program = OpenCLInfo::use_single_program(); + kernel->program = + OpenCLDeviceBase::OpenCLProgram(device, + single_program ? "split" : "split_" + kernel_name, + single_program ? "kernel_split.cl" : "kernel_" + kernel_name + ".cl", + get_build_options(device, requested_features)); + + kernel->program.add_kernel(ustring("path_trace_" + kernel_name)); + kernel->program.load(); + + if(!kernel->program.is_loaded()) { + delete kernel; + return NULL; } + + return kernel; } - void path_trace(DeviceTask *task, - SplitRenderTile& rtile, - int2 max_render_feasible_tile_size) + virtual uint64_t state_buffer_size(device_memory& kg, device_memory& data, size_t num_threads) { - /* cast arguments to cl types */ - cl_mem d_data = CL_MEM_PTR(const_mem_map["__data"]->device_pointer); - cl_mem d_buffer = CL_MEM_PTR(rtile.buffer); - cl_mem d_rng_state = CL_MEM_PTR(rtile.rng_state); - cl_int d_x = rtile.x; - cl_int d_y = rtile.y; - cl_int d_w = rtile.w; - cl_int d_h = rtile.h; - cl_int d_offset = rtile.offset; - cl_int d_stride = rtile.stride; - - /* Make sure that set render feasible tile size is a multiple of local - * work size dimensions. - */ - assert(max_render_feasible_tile_size.x % SPLIT_KERNEL_LOCAL_SIZE_X == 0); - assert(max_render_feasible_tile_size.y % SPLIT_KERNEL_LOCAL_SIZE_Y == 0); + device_vector<uint64_t> size_buffer; + size_buffer.resize(1); + device->mem_alloc(NULL, size_buffer, MEM_READ_WRITE); + + uint threads = num_threads; + device->kernel_set_args(device->program_state_buffer_size(), 0, kg, data, threads, size_buffer); + + size_t global_size = 64; + device->ciErr = clEnqueueNDRangeKernel(device->cqCommandQueue, + device->program_state_buffer_size(), + 1, + NULL, + &global_size, + NULL, + 0, + NULL, + NULL); + + device->opencl_assert_err(device->ciErr, "clEnqueueNDRangeKernel"); + + device->mem_copy_from(size_buffer, 0, 1, 1, sizeof(uint64_t)); + device->mem_free(size_buffer); + + if(device->ciErr != CL_SUCCESS) { + string message = string_printf("OpenCL error: %s in clEnqueueNDRangeKernel()", + clewErrorString(device->ciErr)); + device->opencl_error(message); + return 0; + } + + return *size_buffer.get_data(); + } - size_t global_size[2]; - size_t local_size[2] = {SPLIT_KERNEL_LOCAL_SIZE_X, - SPLIT_KERNEL_LOCAL_SIZE_Y}; + virtual bool enqueue_split_kernel_data_init(const KernelDimensions& dim, + RenderTile& rtile, + int num_global_elements, + device_memory& kernel_globals, + device_memory& kernel_data, + device_memory& split_data, + device_memory& ray_state, + device_memory& queue_index, + device_memory& use_queues_flag, + device_memory& work_pool_wgs + ) + { + cl_int dQueue_size = dim.global_size[0] * dim.global_size[1]; /* Set the range of samples to be processed for every ray in * path-regeneration logic. */ cl_int start_sample = rtile.start_sample; cl_int end_sample = rtile.start_sample + rtile.num_samples; - cl_int num_samples = rtile.num_samples; - -#ifdef __WORK_STEALING__ - global_size[0] = (((d_w - 1) / local_size[0]) + 1) * local_size[0]; - global_size[1] = (((d_h - 1) / local_size[1]) + 1) * local_size[1]; - unsigned int num_parallel_samples = 1; -#else - global_size[1] = (((d_h - 1) / local_size[1]) + 1) * local_size[1]; - unsigned int num_threads = max_render_feasible_tile_size.x * - max_render_feasible_tile_size.y; - unsigned int num_tile_columns_possible = num_threads / global_size[1]; - /* Estimate number of parallel samples that can be - * processed in parallel. - */ - unsigned int num_parallel_samples = min(num_tile_columns_possible / d_w, - rtile.num_samples); - /* Wavefront size in AMD is 64. - * TODO(sergey): What about other platforms? - */ - if(num_parallel_samples >= 64) { - /* TODO(sergey): Could use generic round-up here. */ - num_parallel_samples = (num_parallel_samples / 64) * 64; - } - assert(num_parallel_samples != 0); - - global_size[0] = d_w * num_parallel_samples; -#endif /* __WORK_STEALING__ */ - - assert(global_size[0] * global_size[1] <= - max_render_feasible_tile_size.x * max_render_feasible_tile_size.y); - - /* Allocate all required global memory once. */ - if(first_tile) { - size_t num_global_elements = max_render_feasible_tile_size.x * - max_render_feasible_tile_size.y; - /* TODO(sergey): This will actually over-allocate if - * particular kernel does not support multiclosure. - */ - size_t shaderdata_size = get_shader_data_size(current_max_closure); - -#ifdef __WORK_STEALING__ - /* Calculate max groups */ - size_t max_global_size[2]; - size_t tile_x = max_render_feasible_tile_size.x; - size_t tile_y = max_render_feasible_tile_size.y; - max_global_size[0] = (((tile_x - 1) / local_size[0]) + 1) * local_size[0]; - max_global_size[1] = (((tile_y - 1) / local_size[1]) + 1) * local_size[1]; - max_work_groups = (max_global_size[0] * max_global_size[1]) / - (local_size[0] * local_size[1]); - /* Allocate work_pool_wgs memory. */ - work_pool_wgs = mem_alloc(max_work_groups * sizeof(unsigned int)); -#endif /* __WORK_STEALING__ */ - - /* Allocate queue_index memory only once. */ - Queue_index = mem_alloc(NUM_QUEUES * sizeof(int)); - use_queues_flag = mem_alloc(sizeof(char)); - kgbuffer = mem_alloc(get_KernelGlobals_size()); - - /* Create global buffers for ShaderData. */ - sd = mem_alloc(num_global_elements * shaderdata_size); - sd_DL_shadow = mem_alloc(num_global_elements * 2 * shaderdata_size); - - /* Creation of global memory buffers which are shared among - * the kernels. - */ - rng_coop = mem_alloc(num_global_elements * sizeof(RNG)); - throughput_coop = mem_alloc(num_global_elements * sizeof(float3)); - L_transparent_coop = mem_alloc(num_global_elements * sizeof(float)); - PathRadiance_coop = mem_alloc(num_global_elements * sizeof(PathRadiance)); - Ray_coop = mem_alloc(num_global_elements * sizeof(Ray)); - PathState_coop = mem_alloc(num_global_elements * sizeof(PathState)); - Intersection_coop = mem_alloc(num_global_elements * sizeof(Intersection)); - AOAlpha_coop = mem_alloc(num_global_elements * sizeof(float3)); - AOBSDF_coop = mem_alloc(num_global_elements * sizeof(float3)); - AOLightRay_coop = mem_alloc(num_global_elements * sizeof(Ray)); - BSDFEval_coop = mem_alloc(num_global_elements * sizeof(BsdfEval)); - ISLamp_coop = mem_alloc(num_global_elements * sizeof(int)); - LightRay_coop = mem_alloc(num_global_elements * sizeof(Ray)); - Intersection_coop_shadow = mem_alloc(2 * num_global_elements * sizeof(Intersection)); - -#ifdef WITH_CYCLES_DEBUG - debugdata_coop = mem_alloc(num_global_elements * sizeof(DebugData)); -#endif - - ray_state = mem_alloc(num_global_elements * sizeof(char)); - - hostRayStateArray = (char *)calloc(num_global_elements, sizeof(char)); - assert(hostRayStateArray != NULL && "Can't create hostRayStateArray memory"); - - Queue_data = mem_alloc(num_global_elements * (NUM_QUEUES * sizeof(int)+sizeof(int))); - work_array = mem_alloc(num_global_elements * sizeof(unsigned int)); - per_sample_output_buffers = mem_alloc(num_global_elements * - per_thread_output_buffer_size); - } - - cl_int dQueue_size = global_size[0] * global_size[1]; cl_uint start_arg_index = - kernel_set_args(program_data_init(), + device->kernel_set_args(device->program_data_init(), 0, - kgbuffer, - sd_DL_shadow, - d_data, - per_sample_output_buffers, - d_rng_state, - rng_coop, - throughput_coop, - L_transparent_coop, - PathRadiance_coop, - Ray_coop, - PathState_coop, - Intersection_coop_shadow, - ray_state); + kernel_globals, + kernel_data, + split_data, + num_global_elements, + ray_state, + rtile.rng_state); /* TODO(sergey): Avoid map lookup here. */ #define KERNEL_TEX(type, ttype, name) \ - set_kernel_arg_mem(program_data_init(), &start_arg_index, #name); -#include "kernel_textures.h" + device->set_kernel_arg_mem(device->program_data_init(), &start_arg_index, #name); +#include "kernel/kernel_textures.h" #undef KERNEL_TEX start_arg_index += - kernel_set_args(program_data_init(), + device->kernel_set_args(device->program_data_init(), start_arg_index, start_sample, - d_x, - d_y, - d_w, - d_h, - d_offset, - d_stride, - rtile.rng_state_offset_x, - rtile.rng_state_offset_y, - rtile.buffer_rng_state_stride, - Queue_data, - Queue_index, + end_sample, + rtile.x, + rtile.y, + rtile.w, + rtile.h, + rtile.offset, + rtile.stride, + queue_index, dQueue_size, use_queues_flag, - work_array, -#ifdef __WORK_STEALING__ work_pool_wgs, - num_samples, -#endif -#ifdef WITH_CYCLES_DEBUG - debugdata_coop, -#endif - num_parallel_samples); - - kernel_set_args(program_scene_intersect(), - 0, - kgbuffer, - d_data, - rng_coop, - Ray_coop, - PathState_coop, - Intersection_coop, - ray_state, - d_w, - d_h, - Queue_data, - Queue_index, - dQueue_size, - use_queues_flag, -#ifdef WITH_CYCLES_DEBUG - debugdata_coop, -#endif - num_parallel_samples); - - kernel_set_args(program_lamp_emission(), - 0, - kgbuffer, - d_data, - throughput_coop, - PathRadiance_coop, - Ray_coop, - PathState_coop, - Intersection_coop, - ray_state, - d_w, - d_h, - Queue_data, - Queue_index, - dQueue_size, - use_queues_flag, - num_parallel_samples); - - kernel_set_args(program_queue_enqueue(), - 0, - Queue_data, - Queue_index, - ray_state, - dQueue_size); - - kernel_set_args(program_background_buffer_update(), - 0, - kgbuffer, - d_data, - per_sample_output_buffers, - d_rng_state, - rng_coop, - throughput_coop, - PathRadiance_coop, - Ray_coop, - PathState_coop, - L_transparent_coop, - ray_state, - d_w, - d_h, - d_x, - d_y, - d_stride, - rtile.rng_state_offset_x, - rtile.rng_state_offset_y, - rtile.buffer_rng_state_stride, - work_array, - Queue_data, - Queue_index, - dQueue_size, - end_sample, - start_sample, -#ifdef __WORK_STEALING__ - work_pool_wgs, - num_samples, -#endif -#ifdef WITH_CYCLES_DEBUG - debugdata_coop, -#endif - num_parallel_samples); - - kernel_set_args(program_shader_eval(), - 0, - kgbuffer, - d_data, - sd, - rng_coop, - Ray_coop, - PathState_coop, - Intersection_coop, - ray_state, - Queue_data, - Queue_index, - dQueue_size); - - kernel_set_args(program_holdout_emission_blurring_pathtermination_ao(), - 0, - kgbuffer, - d_data, - sd, - per_sample_output_buffers, - rng_coop, - throughput_coop, - L_transparent_coop, - PathRadiance_coop, - PathState_coop, - Intersection_coop, - AOAlpha_coop, - AOBSDF_coop, - AOLightRay_coop, - d_w, - d_h, - d_x, - d_y, - d_stride, - ray_state, - work_array, - Queue_data, - Queue_index, - dQueue_size, -#ifdef __WORK_STEALING__ - start_sample, -#endif - num_parallel_samples); - - kernel_set_args(program_direct_lighting(), - 0, - kgbuffer, - d_data, - sd, - rng_coop, - PathState_coop, - ISLamp_coop, - LightRay_coop, - BSDFEval_coop, - ray_state, - Queue_data, - Queue_index, - dQueue_size); - - kernel_set_args(program_shadow_blocked(), - 0, - kgbuffer, - d_data, - PathState_coop, - LightRay_coop, - AOLightRay_coop, - ray_state, - Queue_data, - Queue_index, - dQueue_size); - - kernel_set_args(program_next_iteration_setup(), - 0, - kgbuffer, - d_data, - sd, - rng_coop, - throughput_coop, - PathRadiance_coop, - Ray_coop, - PathState_coop, - LightRay_coop, - ISLamp_coop, - BSDFEval_coop, - AOLightRay_coop, - AOBSDF_coop, - AOAlpha_coop, - ray_state, - Queue_data, - Queue_index, - dQueue_size, - use_queues_flag); - - kernel_set_args(program_sum_all_radiance(), - 0, - d_data, - d_buffer, - per_sample_output_buffers, - num_parallel_samples, - d_w, - d_h, - d_stride, - rtile.buffer_offset_x, - rtile.buffer_offset_y, - rtile.buffer_rng_state_stride, - start_sample); - - /* Macro for Enqueuing split kernels. */ -#define GLUE(a, b) a ## b -#define ENQUEUE_SPLIT_KERNEL(kernelName, globalSize, localSize) \ - { \ - ciErr = clEnqueueNDRangeKernel(cqCommandQueue, \ - GLUE(program_, \ - kernelName)(), \ - 2, \ - NULL, \ - globalSize, \ - localSize, \ - 0, \ - NULL, \ - NULL); \ - opencl_assert_err(ciErr, "clEnqueueNDRangeKernel"); \ - if(ciErr != CL_SUCCESS) { \ - string message = string_printf("OpenCL error: %s in clEnqueueNDRangeKernel()", \ - clewErrorString(ciErr)); \ - opencl_error(message); \ - return; \ - } \ - } (void) 0 + rtile.num_samples, + rtile.buffer); /* Enqueue ckPathTraceKernel_data_init kernel. */ - ENQUEUE_SPLIT_KERNEL(data_init, global_size, local_size); - bool activeRaysAvailable = true; - - /* Record number of time host intervention has been made */ - unsigned int numHostIntervention = 0; - unsigned int numNextPathIterTimes = PathIteration_times; - bool canceled = false; - while(activeRaysAvailable) { - /* Twice the global work size of other kernels for - * ckPathTraceKernel_shadow_blocked_direct_lighting. */ - size_t global_size_shadow_blocked[2]; - global_size_shadow_blocked[0] = global_size[0] * 2; - global_size_shadow_blocked[1] = global_size[1]; - - /* Do path-iteration in host [Enqueue Path-iteration kernels. */ - for(int PathIter = 0; PathIter < PathIteration_times; PathIter++) { - ENQUEUE_SPLIT_KERNEL(scene_intersect, global_size, local_size); - ENQUEUE_SPLIT_KERNEL(lamp_emission, global_size, local_size); - ENQUEUE_SPLIT_KERNEL(queue_enqueue, global_size, local_size); - ENQUEUE_SPLIT_KERNEL(background_buffer_update, global_size, local_size); - ENQUEUE_SPLIT_KERNEL(shader_eval, global_size, local_size); - ENQUEUE_SPLIT_KERNEL(holdout_emission_blurring_pathtermination_ao, global_size, local_size); - ENQUEUE_SPLIT_KERNEL(direct_lighting, global_size, local_size); - ENQUEUE_SPLIT_KERNEL(shadow_blocked, global_size_shadow_blocked, local_size); - ENQUEUE_SPLIT_KERNEL(next_iteration_setup, global_size, local_size); - - if(task->get_cancel()) { - canceled = true; - break; - } - } - - /* Read ray-state into Host memory to decide if we should exit - * path-iteration in host. - */ - ciErr = clEnqueueReadBuffer(cqCommandQueue, - ray_state, - CL_TRUE, - 0, - global_size[0] * global_size[1] * sizeof(char), - hostRayStateArray, - 0, - NULL, - NULL); - assert(ciErr == CL_SUCCESS); - - activeRaysAvailable = false; - - for(int rayStateIter = 0; - rayStateIter < global_size[0] * global_size[1]; - ++rayStateIter) - { - if(int8_t(hostRayStateArray[rayStateIter]) != RAY_INACTIVE) { - /* Not all rays are RAY_INACTIVE. */ - activeRaysAvailable = true; - break; - } - } - - if(activeRaysAvailable) { - numHostIntervention++; - PathIteration_times = PATH_ITER_INC_FACTOR; - /* Host intervention done before all rays become RAY_INACTIVE; - * Set do more initial iterations for the next tile. - */ - numNextPathIterTimes += PATH_ITER_INC_FACTOR; - } - - if(task->get_cancel()) { - canceled = true; - break; - } - } - - /* Execute SumALLRadiance kernel to accumulate radiance calculated in - * per_sample_output_buffers into RenderTile's output buffer. - */ - if(!canceled) { - size_t sum_all_radiance_local_size[2] = {16, 16}; - size_t sum_all_radiance_global_size[2]; - sum_all_radiance_global_size[0] = - (((d_w - 1) / sum_all_radiance_local_size[0]) + 1) * - sum_all_radiance_local_size[0]; - sum_all_radiance_global_size[1] = - (((d_h - 1) / sum_all_radiance_local_size[1]) + 1) * - sum_all_radiance_local_size[1]; - ENQUEUE_SPLIT_KERNEL(sum_all_radiance, - sum_all_radiance_global_size, - sum_all_radiance_local_size); - } - -#undef ENQUEUE_SPLIT_KERNEL -#undef GLUE - - if(numHostIntervention == 0) { - /* This means that we are executing kernel more than required - * Must avoid this for the next sample/tile. - */ - PathIteration_times = ((numNextPathIterTimes - PATH_ITER_INC_FACTOR) <= 0) ? - PATH_ITER_INC_FACTOR : numNextPathIterTimes - PATH_ITER_INC_FACTOR; - } - else { - /* Number of path-iterations done for this tile is set as - * Initial path-iteration times for the next tile - */ - PathIteration_times = numNextPathIterTimes; - } - - first_tile = false; - } - - /* Calculates the amount of memory that has to be always - * allocated in order for the split kernel to function. - * This memory is tile/scene-property invariant (meaning, - * the value returned by this function does not depend - * on the user set tile size or scene properties. - */ - size_t get_invariable_mem_allocated() - { - size_t total_invariable_mem_allocated = 0; - size_t KernelGlobals_size = 0; - - KernelGlobals_size = get_KernelGlobals_size(); - - total_invariable_mem_allocated += KernelGlobals_size; /* KernelGlobals size */ - total_invariable_mem_allocated += NUM_QUEUES * sizeof(unsigned int); /* Queue index size */ - total_invariable_mem_allocated += sizeof(char); /* use_queues_flag size */ - - return total_invariable_mem_allocated; - } - - /* Calculate the memory that has-to-be/has-been allocated for - * the split kernel to function. - */ - size_t get_tile_specific_mem_allocated(const int2 tile_size) - { - size_t tile_specific_mem_allocated = 0; - - /* Get required tile info */ - unsigned int user_set_tile_w = tile_size.x; - unsigned int user_set_tile_h = tile_size.y; - -#ifdef __WORK_STEALING__ - /* Calculate memory to be allocated for work_pools in - * case of work_stealing. - */ - size_t max_global_size[2]; - size_t max_num_work_pools = 0; - max_global_size[0] = - (((user_set_tile_w - 1) / SPLIT_KERNEL_LOCAL_SIZE_X) + 1) * - SPLIT_KERNEL_LOCAL_SIZE_X; - max_global_size[1] = - (((user_set_tile_h - 1) / SPLIT_KERNEL_LOCAL_SIZE_Y) + 1) * - SPLIT_KERNEL_LOCAL_SIZE_Y; - max_num_work_pools = - (max_global_size[0] * max_global_size[1]) / - (SPLIT_KERNEL_LOCAL_SIZE_X * SPLIT_KERNEL_LOCAL_SIZE_Y); - tile_specific_mem_allocated += max_num_work_pools * sizeof(unsigned int); -#endif - - tile_specific_mem_allocated += - user_set_tile_w * user_set_tile_h * per_thread_output_buffer_size; - tile_specific_mem_allocated += - user_set_tile_w * user_set_tile_h * sizeof(RNG); - - return tile_specific_mem_allocated; - } - - /* Calculates the texture memories and KernelData (d_data) memory - * that has been allocated. - */ - size_t get_scene_specific_mem_allocated(cl_mem d_data) - { - size_t scene_specific_mem_allocated = 0; - /* Calculate texture memories. */ -#define KERNEL_TEX(type, ttype, name) \ - scene_specific_mem_allocated += get_tex_size(#name); -#include "kernel_textures.h" -#undef KERNEL_TEX - size_t d_data_size; - ciErr = clGetMemObjectInfo(d_data, - CL_MEM_SIZE, - sizeof(d_data_size), - &d_data_size, - NULL); - assert(ciErr == CL_SUCCESS && "Can't get d_data mem object info"); - scene_specific_mem_allocated += d_data_size; - return scene_specific_mem_allocated; - } - - /* Calculate the memory required for one thread in split kernel. */ - size_t get_per_thread_memory() - { - size_t shaderdata_size = 0; - /* TODO(sergey): This will actually over-allocate if - * particular kernel does not support multiclosure. - */ - shaderdata_size = get_shader_data_size(current_max_closure); - size_t retval = sizeof(RNG) - + sizeof(float3) /* Throughput size */ - + sizeof(float) /* L transparent size */ - + sizeof(char) /* Ray state size */ - + sizeof(unsigned int) /* Work element size */ - + sizeof(int) /* ISLamp_size */ - + sizeof(PathRadiance) + sizeof(Ray) + sizeof(PathState) - + sizeof(Intersection) /* Overall isect */ - + sizeof(Intersection) /* Instersection_coop_AO */ - + sizeof(Intersection) /* Intersection coop DL */ - + shaderdata_size /* Overall ShaderData */ - + (shaderdata_size * 2) /* ShaderData : DL and shadow */ - + sizeof(Ray) + sizeof(BsdfEval) - + sizeof(float3) /* AOAlpha size */ - + sizeof(float3) /* AOBSDF size */ - + sizeof(Ray) - + (sizeof(int) * NUM_QUEUES) - + per_thread_output_buffer_size; - return retval; - } - - /* Considers the total memory available in the device and - * and returns the maximum global work size possible. - */ - size_t get_feasible_global_work_size(int2 tile_size, cl_mem d_data) - { - /* Calculate invariably allocated memory. */ - size_t invariable_mem_allocated = get_invariable_mem_allocated(); - /* Calculate tile specific allocated memory. */ - size_t tile_specific_mem_allocated = - get_tile_specific_mem_allocated(tile_size); - /* Calculate scene specific allocated memory. */ - size_t scene_specific_mem_allocated = - get_scene_specific_mem_allocated(d_data); - /* Calculate total memory available for the threads in global work size. */ - size_t available_memory = total_allocatable_memory - - invariable_mem_allocated - - tile_specific_mem_allocated - - scene_specific_mem_allocated - - DATA_ALLOCATION_MEM_FACTOR; - size_t per_thread_memory_required = get_per_thread_memory(); - return (available_memory / per_thread_memory_required); - } - - /* Checks if the device has enough memory to render the whole tile; - * If not, we should split single tile into multiple tiles of small size - * and process them all. - */ - bool need_to_split_tile(unsigned int d_w, - unsigned int d_h, - int2 max_render_feasible_tile_size) - { - size_t global_size_estimate[2]; - /* TODO(sergey): Such round-ups are in quite few places, need to replace - * them with an utility macro. - */ - global_size_estimate[0] = - (((d_w - 1) / SPLIT_KERNEL_LOCAL_SIZE_X) + 1) * - SPLIT_KERNEL_LOCAL_SIZE_X; - global_size_estimate[1] = - (((d_h - 1) / SPLIT_KERNEL_LOCAL_SIZE_Y) + 1) * - SPLIT_KERNEL_LOCAL_SIZE_Y; - if((global_size_estimate[0] * global_size_estimate[1]) > - (max_render_feasible_tile_size.x * max_render_feasible_tile_size.y)) - { - return true; - } - else { + device->ciErr = clEnqueueNDRangeKernel(device->cqCommandQueue, + device->program_data_init(), + 2, + NULL, + dim.global_size, + dim.local_size, + 0, + NULL, + NULL); + + device->opencl_assert_err(device->ciErr, "clEnqueueNDRangeKernel"); + + if(device->ciErr != CL_SUCCESS) { + string message = string_printf("OpenCL error: %s in clEnqueueNDRangeKernel()", + clewErrorString(device->ciErr)); + device->opencl_error(message); return false; } - } - /* Considers the scene properties, global memory available in the device - * and returns a rectanglular tile dimension (approx the maximum) - * that should render on split kernel. - */ - int2 get_max_render_feasible_tile_size(size_t feasible_global_work_size) - { - int2 max_render_feasible_tile_size; - int square_root_val = (int)sqrt(feasible_global_work_size); - max_render_feasible_tile_size.x = square_root_val; - max_render_feasible_tile_size.y = square_root_val; - /* Ciel round-off max_render_feasible_tile_size. */ - int2 ceil_render_feasible_tile_size; - ceil_render_feasible_tile_size.x = - (((max_render_feasible_tile_size.x - 1) / SPLIT_KERNEL_LOCAL_SIZE_X) + 1) * - SPLIT_KERNEL_LOCAL_SIZE_X; - ceil_render_feasible_tile_size.y = - (((max_render_feasible_tile_size.y - 1) / SPLIT_KERNEL_LOCAL_SIZE_Y) + 1) * - SPLIT_KERNEL_LOCAL_SIZE_Y; - if(ceil_render_feasible_tile_size.x * ceil_render_feasible_tile_size.y <= - feasible_global_work_size) - { - return ceil_render_feasible_tile_size; - } - /* Floor round-off max_render_feasible_tile_size. */ - int2 floor_render_feasible_tile_size; - floor_render_feasible_tile_size.x = - (max_render_feasible_tile_size.x / SPLIT_KERNEL_LOCAL_SIZE_X) * - SPLIT_KERNEL_LOCAL_SIZE_X; - floor_render_feasible_tile_size.y = - (max_render_feasible_tile_size.y / SPLIT_KERNEL_LOCAL_SIZE_Y) * - SPLIT_KERNEL_LOCAL_SIZE_Y; - return floor_render_feasible_tile_size; + return true; } - /* Try splitting the current tile into multiple smaller - * almost-square-tiles. - */ - int2 get_split_tile_size(RenderTile rtile, - int2 max_render_feasible_tile_size) + virtual int2 split_kernel_local_size() { - int2 split_tile_size; - int num_global_threads = max_render_feasible_tile_size.x * - max_render_feasible_tile_size.y; - int d_w = rtile.w; - int d_h = rtile.h; - /* Ceil round off d_w and d_h */ - d_w = (((d_w - 1) / SPLIT_KERNEL_LOCAL_SIZE_X) + 1) * - SPLIT_KERNEL_LOCAL_SIZE_X; - d_h = (((d_h - 1) / SPLIT_KERNEL_LOCAL_SIZE_Y) + 1) * - SPLIT_KERNEL_LOCAL_SIZE_Y; - while(d_w * d_h > num_global_threads) { - /* Halve the longer dimension. */ - if(d_w >= d_h) { - d_w = d_w / 2; - d_w = (((d_w - 1) / SPLIT_KERNEL_LOCAL_SIZE_X) + 1) * - SPLIT_KERNEL_LOCAL_SIZE_X; - } - else { - d_h = d_h / 2; - d_h = (((d_h - 1) / SPLIT_KERNEL_LOCAL_SIZE_Y) + 1) * - SPLIT_KERNEL_LOCAL_SIZE_Y; - } - } - split_tile_size.x = d_w; - split_tile_size.y = d_h; - return split_tile_size; + return make_int2(64, 1); } - /* Splits existing tile into multiple tiles of tile size split_tile_size. */ - vector<SplitRenderTile> split_tiles(RenderTile rtile, int2 split_tile_size) + virtual int2 split_kernel_global_size(device_memory& kg, device_memory& data, DeviceTask */*task*/) { - vector<SplitRenderTile> to_path_trace_rtile; - int d_w = rtile.w; - int d_h = rtile.h; - int num_tiles_x = (((d_w - 1) / split_tile_size.x) + 1); - int num_tiles_y = (((d_h - 1) / split_tile_size.y) + 1); - /* Buffer and rng_state offset calc. */ - size_t offset_index = rtile.offset + (rtile.x + rtile.y * rtile.stride); - size_t offset_x = offset_index % rtile.stride; - size_t offset_y = offset_index / rtile.stride; - /* Resize to_path_trace_rtile. */ - to_path_trace_rtile.resize(num_tiles_x * num_tiles_y); - for(int tile_iter_y = 0; tile_iter_y < num_tiles_y; tile_iter_y++) { - for(int tile_iter_x = 0; tile_iter_x < num_tiles_x; tile_iter_x++) { - int rtile_index = tile_iter_y * num_tiles_x + tile_iter_x; - to_path_trace_rtile[rtile_index].rng_state_offset_x = offset_x + tile_iter_x * split_tile_size.x; - to_path_trace_rtile[rtile_index].rng_state_offset_y = offset_y + tile_iter_y * split_tile_size.y; - to_path_trace_rtile[rtile_index].buffer_offset_x = offset_x + tile_iter_x * split_tile_size.x; - to_path_trace_rtile[rtile_index].buffer_offset_y = offset_y + tile_iter_y * split_tile_size.y; - to_path_trace_rtile[rtile_index].start_sample = rtile.start_sample; - to_path_trace_rtile[rtile_index].num_samples = rtile.num_samples; - to_path_trace_rtile[rtile_index].sample = rtile.sample; - to_path_trace_rtile[rtile_index].resolution = rtile.resolution; - to_path_trace_rtile[rtile_index].offset = rtile.offset; - to_path_trace_rtile[rtile_index].buffers = rtile.buffers; - to_path_trace_rtile[rtile_index].buffer = rtile.buffer; - to_path_trace_rtile[rtile_index].rng_state = rtile.rng_state; - to_path_trace_rtile[rtile_index].x = rtile.x + (tile_iter_x * split_tile_size.x); - to_path_trace_rtile[rtile_index].y = rtile.y + (tile_iter_y * split_tile_size.y); - to_path_trace_rtile[rtile_index].buffer_rng_state_stride = rtile.stride; - /* Fill width and height of the new render tile. */ - to_path_trace_rtile[rtile_index].w = (tile_iter_x == (num_tiles_x - 1)) ? - (d_w - (tile_iter_x * split_tile_size.x)) /* Border tile */ - : split_tile_size.x; - to_path_trace_rtile[rtile_index].h = (tile_iter_y == (num_tiles_y - 1)) ? - (d_h - (tile_iter_y * split_tile_size.y)) /* Border tile */ - : split_tile_size.y; - to_path_trace_rtile[rtile_index].stride = to_path_trace_rtile[rtile_index].w; - } + cl_device_type type = OpenCLInfo::get_device_type(device->cdDevice); + /* Use small global size on CPU devices as it seems to be much faster. */ + if(type == CL_DEVICE_TYPE_CPU) { + VLOG(1) << "Global size: (64, 64)."; + return make_int2(64, 64); } - return to_path_trace_rtile; - } - void thread_run(DeviceTask *task) - { - if(task->type == DeviceTask::FILM_CONVERT) { - film_convert(*task, task->buffer, task->rgba_byte, task->rgba_half); - } - else if(task->type == DeviceTask::SHADER) { - shader(*task); - } - else if(task->type == DeviceTask::PATH_TRACE) { - RenderTile tile; - bool initialize_data_and_check_render_feasibility = false; - bool need_to_split_tiles_further = false; - int2 max_render_feasible_tile_size; - size_t feasible_global_work_size; - const int2 tile_size = task->requested_tile_size; - /* Keep rendering tiles until done. */ - while(task->acquire_tile(this, tile)) { - if(!initialize_data_and_check_render_feasibility) { - /* Initialize data. */ - /* Calculate per_thread_output_buffer_size. */ - size_t output_buffer_size = 0; - ciErr = clGetMemObjectInfo((cl_mem)tile.buffer, - CL_MEM_SIZE, - sizeof(output_buffer_size), - &output_buffer_size, - NULL); - assert(ciErr == CL_SUCCESS && "Can't get tile.buffer mem object info"); - /* This value is different when running on AMD and NV. */ - if(background) { - /* In offline render the number of buffer elements - * associated with tile.buffer is the current tile size. - */ - per_thread_output_buffer_size = - output_buffer_size / (tile.w * tile.h); - } - else { - /* interactive rendering, unlike offline render, the number of buffer elements - * associated with tile.buffer is the entire viewport size. - */ - per_thread_output_buffer_size = - output_buffer_size / (tile.buffers->params.width * - tile.buffers->params.height); - } - /* Check render feasibility. */ - feasible_global_work_size = get_feasible_global_work_size( - tile_size, - CL_MEM_PTR(const_mem_map["__data"]->device_pointer)); - max_render_feasible_tile_size = - get_max_render_feasible_tile_size( - feasible_global_work_size); - need_to_split_tiles_further = - need_to_split_tile(tile_size.x, - tile_size.y, - max_render_feasible_tile_size); - initialize_data_and_check_render_feasibility = true; - } - if(need_to_split_tiles_further) { - int2 split_tile_size = - get_split_tile_size(tile, - max_render_feasible_tile_size); - vector<SplitRenderTile> to_path_trace_render_tiles = - split_tiles(tile, split_tile_size); - /* Print message to console */ - if(background && (to_path_trace_render_tiles.size() > 1)) { - fprintf(stderr, "Message : Tiles need to be split " - "further inside path trace (due to insufficient " - "device-global-memory for split kernel to " - "function) \n" - "The current tile of dimensions %dx%d is split " - "into tiles of dimension %dx%d for render \n", - tile.w, tile.h, - split_tile_size.x, - split_tile_size.y); - } - /* Process all split tiles. */ - for(int tile_iter = 0; - tile_iter < to_path_trace_render_tiles.size(); - ++tile_iter) - { - path_trace(task, - to_path_trace_render_tiles[tile_iter], - max_render_feasible_tile_size); - } - } - else { - /* No splitting required; process the entire tile at once. */ - /* Render feasible tile size is user-set-tile-size itself. */ - max_render_feasible_tile_size.x = - (((tile_size.x - 1) / SPLIT_KERNEL_LOCAL_SIZE_X) + 1) * - SPLIT_KERNEL_LOCAL_SIZE_X; - max_render_feasible_tile_size.y = - (((tile_size.y - 1) / SPLIT_KERNEL_LOCAL_SIZE_Y) + 1) * - SPLIT_KERNEL_LOCAL_SIZE_Y; - /* buffer_rng_state_stride is stride itself. */ - SplitRenderTile split_tile(tile); - split_tile.buffer_rng_state_stride = tile.stride; - path_trace(task, split_tile, max_render_feasible_tile_size); - } - tile.sample = tile.start_sample + tile.num_samples; + cl_ulong max_buffer_size; + clGetDeviceInfo(device->cdDevice, CL_DEVICE_MAX_MEM_ALLOC_SIZE, sizeof(cl_ulong), &max_buffer_size, NULL); + VLOG(1) << "Maximum device allocation size: " + << string_human_readable_number(max_buffer_size) << " bytes. (" + << string_human_readable_size(max_buffer_size) << ")."; - /* Complete kernel execution before release tile. */ - /* This helps in multi-device render; - * The device that reaches the critical-section function - * release_tile waits (stalling other devices from entering - * release_tile) for all kernels to complete. If device1 (a - * slow-render device) reaches release_tile first then it would - * stall device2 (a fast-render device) from proceeding to render - * next tile. - */ - clFinish(cqCommandQueue); - - task->release_tile(tile); - } - } - } - -protected: - cl_mem mem_alloc(size_t bufsize, cl_mem_flags mem_flag = CL_MEM_READ_WRITE) - { - cl_mem ptr; - assert(bufsize != 0); - ptr = clCreateBuffer(cxContext, mem_flag, bufsize, NULL, &ciErr); - opencl_assert_err(ciErr, "clCreateBuffer"); - return ptr; + size_t num_elements = max_elements_for_max_buffer_size(kg, data, max_buffer_size / 2); + int2 global_size = make_int2(round_down((int)sqrt(num_elements), 64), (int)sqrt(num_elements)); + VLOG(1) << "Global size: " << global_size << "."; + return global_size; } +}; - /* ** Those guys are for workign around some compiler-specific bugs ** */ +OpenCLDeviceSplitKernel::OpenCLDeviceSplitKernel(DeviceInfo& info, Stats &stats, bool background_) +: OpenCLDeviceBase(info, stats, background_) +{ + split_kernel = new OpenCLSplitKernel(this); - string build_options_for_base_program( - const DeviceRequestedFeatures& requested_features) - { - return requested_features.get_build_options(); - } -}; + background = background_; +} Device *opencl_create_split_device(DeviceInfo& info, Stats& stats, bool background) { diff --git a/intern/cycles/device/opencl/opencl_util.cpp b/intern/cycles/device/opencl/opencl_util.cpp index 82e1640e508..8128fcee09b 100644 --- a/intern/cycles/device/opencl/opencl_util.cpp +++ b/intern/cycles/device/opencl/opencl_util.cpp @@ -16,11 +16,12 @@ #ifdef WITH_OPENCL -#include "opencl.h" +#include "device/opencl/opencl.h" -#include "util_logging.h" -#include "util_path.h" -#include "util_time.h" +#include "util/util_logging.h" +#include "util/util_md5.h" +#include "util/util_path.h" +#include "util/util_time.h" using std::cerr; using std::endl; @@ -234,7 +235,7 @@ string OpenCLCache::get_kernel_md5() thread_scoped_lock lock(self.kernel_md5_lock); if(self.kernel_md5.empty()) { - self.kernel_md5 = path_files_md5_hash(path_get("kernel")); + self.kernel_md5 = path_files_md5_hash(path_get("source")); } return self.kernel_md5; } @@ -309,6 +310,8 @@ bool OpenCLDeviceBase::OpenCLProgram::build_kernel(const string *debug_src) string build_options; build_options = device->kernel_build_options(debug_src) + kernel_build_options; + VLOG(1) << "Build options passed to clBuildProgram: '" + << build_options << "'."; cl_int ciErr = clBuildProgram(program, 0, NULL, build_options.c_str(), NULL, NULL); /* show warnings even if build is successful */ @@ -336,12 +339,13 @@ bool OpenCLDeviceBase::OpenCLProgram::build_kernel(const string *debug_src) bool OpenCLDeviceBase::OpenCLProgram::compile_kernel(const string *debug_src) { - string source = "#include \"kernels/opencl/" + kernel_file + "\" // " + OpenCLCache::get_kernel_md5() + "\n"; + string source = "#include \"kernel/kernels/opencl/" + kernel_file + "\"\n"; /* We compile kernels consisting of many files. unfortunately OpenCL * kernel caches do not seem to recognize changes in included files. * so we force recompile on changes by adding the md5 hash of all files. */ - source = path_source_replace_includes(source, path_get("kernel")); + source = path_source_replace_includes(source, path_get("source")); + source += "\n// " + util_md5_string(source) + "\n"; if(debug_src) { path_write_text(*debug_src, source); @@ -352,10 +356,10 @@ bool OpenCLDeviceBase::OpenCLProgram::compile_kernel(const string *debug_src) cl_int ciErr; program = clCreateProgramWithSource(device->cxContext, - 1, - &source_str, - &source_len, - &ciErr); + 1, + &source_str, + &source_len, + &ciErr); if(ciErr != CL_SUCCESS) { add_error(string("OpenCL program creation failed: ") + clewErrorString(ciErr)); @@ -438,7 +442,11 @@ void OpenCLDeviceBase::OpenCLProgram::load() if(!program) { add_log(string("OpenCL program ") + program_name + " not found in cache.", true); - string basename = "cycles_kernel_" + program_name + "_" + device_md5 + "_" + OpenCLCache::get_kernel_md5(); + /* need to create source to get md5 */ + string source = "#include \"kernel/kernels/opencl/" + kernel_file + "\"\n"; + source = path_source_replace_includes(source, path_get("source")); + + string basename = "cycles_kernel_" + program_name + "_" + device_md5 + "_" + util_md5_string(source); basename = path_cache_get(path_join("kernels", basename)); string clbin = basename + ".clbin"; @@ -544,6 +552,11 @@ bool OpenCLInfo::use_debug() return DebugFlags().opencl.debug; } +bool OpenCLInfo::use_single_program() +{ + return DebugFlags().opencl.single_program; +} + bool OpenCLInfo::kernel_use_advanced_shading(const string& platform) { /* keep this in sync with kernel_types.h! */ @@ -587,11 +600,20 @@ bool OpenCLInfo::device_supported(const string& platform_name, const cl_device_id device_id) { cl_device_type device_type; - clGetDeviceInfo(device_id, - CL_DEVICE_TYPE, - sizeof(cl_device_type), - &device_type, - NULL); + if(!get_device_type(device_id, &device_type)) { + return false; + } + string device_name; + if(!get_device_name(device_id, &device_name)) { + return false; + } + /* It is possible tyo have Iris GPU on AMD/Apple OpenCL framework + * (aka, it will not be on Intel framework). This isn't supported + * and needs an explicit blacklist. + */ + if(strstr(device_name.c_str(), "Iris")) { + return false; + } if(platform_name == "AMD Accelerated Parallel Processing" && device_type == CL_DEVICE_TYPE_GPU) { @@ -705,39 +727,30 @@ void OpenCLInfo::get_usable_devices(vector<OpenCLPlatformDevice> *usable_devices return; } + cl_int error; vector<cl_device_id> device_ids; - cl_uint num_devices = 0; vector<cl_platform_id> platform_ids; - cl_uint num_platforms = 0; - /* Get devices. */ - if(clGetPlatformIDs(0, NULL, &num_platforms) != CL_SUCCESS || - num_platforms == 0) - { - FIRST_VLOG(2) << "No OpenCL platforms were found."; + /* Get platforms. */ + if(!get_platforms(&platform_ids, &error)) { + FIRST_VLOG(2) << "Error fetching platforms:" + << string(clewErrorString(error)); first_time = false; return; } - platform_ids.resize(num_platforms); - if(clGetPlatformIDs(num_platforms, &platform_ids[0], NULL) != CL_SUCCESS) { - FIRST_VLOG(2) << "Failed to fetch platform IDs from the driver.."; + if(platform_ids.size() == 0) { + FIRST_VLOG(2) << "No OpenCL platforms were found."; first_time = false; return; } /* Devices are numbered consecutively across platforms. */ - for(int platform = 0; platform < num_platforms; platform++) { + for(int platform = 0; platform < platform_ids.size(); platform++) { cl_platform_id platform_id = platform_ids[platform]; - char pname[256]; - if(clGetPlatformInfo(platform_id, - CL_PLATFORM_NAME, - sizeof(pname), - &pname, - NULL) != CL_SUCCESS) - { + string platform_name; + if(!get_platform_name(platform_id, &platform_name)) { FIRST_VLOG(2) << "Failed to get platform name, ignoring."; continue; } - string platform_name = pname; FIRST_VLOG(2) << "Enumerating devices for platform " << platform_name << "."; if(!platform_version_check(platform_id)) { @@ -745,39 +758,28 @@ void OpenCLInfo::get_usable_devices(vector<OpenCLPlatformDevice> *usable_devices << " due to too old compiler version."; continue; } - num_devices = 0; - cl_int ciErr; - if((ciErr = clGetDeviceIDs(platform_id, - device_type, - 0, - NULL, - &num_devices)) != CL_SUCCESS || num_devices == 0) + if(!get_platform_devices(platform_id, + device_type, + &device_ids, + &error)) { FIRST_VLOG(2) << "Ignoring platform " << platform_name - << ", failed to fetch number of devices: " << string(clewErrorString(ciErr)); + << ", failed to fetch of devices: " + << string(clewErrorString(error)); continue; } - device_ids.resize(num_devices); - if(clGetDeviceIDs(platform_id, - device_type, - num_devices, - &device_ids[0], - NULL) != CL_SUCCESS) - { + if(device_ids.size() == 0) { FIRST_VLOG(2) << "Ignoring platform " << platform_name - << ", failed to fetch devices list."; + << ", it has no devices."; continue; } - for(int num = 0; num < num_devices; num++) { - cl_device_id device_id = device_ids[num]; - char device_name[1024] = "\0"; - if(clGetDeviceInfo(device_id, - CL_DEVICE_NAME, - sizeof(device_name), - &device_name, - NULL) != CL_SUCCESS) - { - FIRST_VLOG(2) << "Failed to fetch device name, ignoring."; + for(int num = 0; num < device_ids.size(); num++) { + const cl_device_id device_id = device_ids[num]; + string device_name; + if(!get_device_name(device_id, &device_name, &error)) { + FIRST_VLOG(2) << "Failed to fetch device name: " + << string(clewErrorString(error)) + << ", ignoring."; continue; } if(!device_version_check(device_id)) { @@ -789,24 +791,28 @@ void OpenCLInfo::get_usable_devices(vector<OpenCLPlatformDevice> *usable_devices device_supported(platform_name, device_id)) { cl_device_type device_type; - if(clGetDeviceInfo(device_id, - CL_DEVICE_TYPE, - sizeof(cl_device_type), - &device_type, - NULL) != CL_SUCCESS) - { + if(!get_device_type(device_id, &device_type, &error)) { FIRST_VLOG(2) << "Ignoring device " << device_name - << ", failed to fetch device type."; + << ", failed to fetch device type:" + << string(clewErrorString(error)); continue; } - FIRST_VLOG(2) << "Adding new device " << device_name << "."; + string readable_device_name = + get_readable_device_name(device_id); + if(readable_device_name != device_name) { + FIRST_VLOG(2) << "Using more readable device name: " + << readable_device_name; + } + FIRST_VLOG(2) << "Adding new device " + << readable_device_name << "."; string hardware_id = get_hardware_id(platform_name, device_id); - usable_devices->push_back(OpenCLPlatformDevice(platform_id, - platform_name, - device_id, - device_type, - device_name, - hardware_id)); + usable_devices->push_back(OpenCLPlatformDevice( + platform_id, + platform_name, + device_id, + device_type, + readable_device_name, + hardware_id)); } else { FIRST_VLOG(2) << "Ignoring device " << device_name @@ -817,6 +823,252 @@ void OpenCLInfo::get_usable_devices(vector<OpenCLPlatformDevice> *usable_devices first_time = false; } +bool OpenCLInfo::get_platforms(vector<cl_platform_id> *platform_ids, + cl_int *error) +{ + /* Reset from possible previous state. */ + platform_ids->resize(0); + cl_uint num_platforms; + if(!get_num_platforms(&num_platforms, error)) { + return false; + } + /* Get actual platforms. */ + cl_int err; + platform_ids->resize(num_platforms); + if((err = clGetPlatformIDs(num_platforms, + &platform_ids->at(0), + NULL)) != CL_SUCCESS) { + if(error != NULL) { + *error = err; + } + return false; + } + if(error != NULL) { + *error = CL_SUCCESS; + } + return true; +} + +vector<cl_platform_id> OpenCLInfo::get_platforms() +{ + vector<cl_platform_id> platform_ids; + get_platforms(&platform_ids); + return platform_ids; +} + +bool OpenCLInfo::get_num_platforms(cl_uint *num_platforms, cl_int *error) +{ + cl_int err; + if((err = clGetPlatformIDs(0, NULL, num_platforms)) != CL_SUCCESS) { + if(error != NULL) { + *error = err; + } + *num_platforms = 0; + return false; + } + if(error != NULL) { + *error = CL_SUCCESS; + } + return true; +} + +cl_uint OpenCLInfo::get_num_platforms() +{ + cl_uint num_platforms; + if(!get_num_platforms(&num_platforms)) { + return 0; + } + return num_platforms; +} + +bool OpenCLInfo::get_platform_name(cl_platform_id platform_id, + string *platform_name) +{ + char buffer[256]; + if(clGetPlatformInfo(platform_id, + CL_PLATFORM_NAME, + sizeof(buffer), + &buffer, + NULL) != CL_SUCCESS) + { + *platform_name = ""; + return false; + } + *platform_name = buffer; + return true; +} + +string OpenCLInfo::get_platform_name(cl_platform_id platform_id) +{ + string platform_name; + if (!get_platform_name(platform_id, &platform_name)) { + return ""; + } + return platform_name; +} + +bool OpenCLInfo::get_num_platform_devices(cl_platform_id platform_id, + cl_device_type device_type, + cl_uint *num_devices, + cl_int *error) +{ + cl_int err; + if((err = clGetDeviceIDs(platform_id, + device_type, + 0, + NULL, + num_devices)) != CL_SUCCESS) + { + if(error != NULL) { + *error = err; + } + *num_devices = 0; + return false; + } + if(error != NULL) { + *error = CL_SUCCESS; + } + return true; +} + +cl_uint OpenCLInfo::get_num_platform_devices(cl_platform_id platform_id, + cl_device_type device_type) +{ + cl_uint num_devices; + if(!get_num_platform_devices(platform_id, + device_type, + &num_devices)) + { + return 0; + } + return num_devices; +} + +bool OpenCLInfo::get_platform_devices(cl_platform_id platform_id, + cl_device_type device_type, + vector<cl_device_id> *device_ids, + cl_int* error) +{ + /* Reset from possible previous state. */ + device_ids->resize(0); + /* Get number of devices to pre-allocate memory. */ + cl_uint num_devices; + if(!get_num_platform_devices(platform_id, + device_type, + &num_devices, + error)) + { + return false; + } + /* Get actual device list. */ + device_ids->resize(num_devices); + cl_int err; + if((err = clGetDeviceIDs(platform_id, + device_type, + num_devices, + &device_ids->at(0), + NULL)) != CL_SUCCESS) + { + if(error != NULL) { + *error = err; + } + return false; + } + if(error != NULL) { + *error = CL_SUCCESS; + } + return true; +} + +vector<cl_device_id> OpenCLInfo::get_platform_devices(cl_platform_id platform_id, + cl_device_type device_type) +{ + vector<cl_device_id> devices; + get_platform_devices(platform_id, device_type, &devices); + return devices; +} + +bool OpenCLInfo::get_device_name(cl_device_id device_id, + string *device_name, + cl_int* error) +{ + char buffer[1024]; + cl_int err; + if((err = clGetDeviceInfo(device_id, + CL_DEVICE_NAME, + sizeof(buffer), + &buffer, + NULL)) != CL_SUCCESS) + { + if(error != NULL) { + *error = err; + } + *device_name = ""; + return false; + } + if(error != NULL) { + *error = CL_SUCCESS; + } + *device_name = buffer; + return true; +} + +string OpenCLInfo::get_device_name(cl_device_id device_id) +{ + string device_name; + if(!get_device_name(device_id, &device_name)) { + return ""; + } + return device_name; +} + +bool OpenCLInfo::get_device_type(cl_device_id device_id, + cl_device_type *device_type, + cl_int* error) +{ + cl_int err; + if((err = clGetDeviceInfo(device_id, + CL_DEVICE_TYPE, + sizeof(cl_device_type), + device_type, + NULL)) != CL_SUCCESS) + { + if(error != NULL) { + *error = err; + } + *device_type = 0; + return false; + } + if(error != NULL) { + *error = CL_SUCCESS; + } + return true; +} + +cl_device_type OpenCLInfo::get_device_type(cl_device_id device_id) +{ + cl_device_type device_type; + if(!get_device_type(device_id, &device_type)) { + return 0; + } + return device_type; +} + +string OpenCLInfo::get_readable_device_name(cl_device_id device_id) +{ + char board_name[1024]; + if(clGetDeviceInfo(device_id, + CL_DEVICE_BOARD_NAME_AMD, + sizeof(board_name), + &board_name, + NULL) == CL_SUCCESS) + { + return board_name; + } + /* Fallback to standard device name API. */ + return get_device_name(device_id); +} + CCL_NAMESPACE_END #endif diff --git a/intern/cycles/graph/CMakeLists.txt b/intern/cycles/graph/CMakeLists.txt index 4ea18728f1c..e70a18137bd 100644 --- a/intern/cycles/graph/CMakeLists.txt +++ b/intern/cycles/graph/CMakeLists.txt @@ -1,7 +1,6 @@ set(INC - . - ../util + .. ) set(SRC diff --git a/intern/cycles/graph/node.cpp b/intern/cycles/graph/node.cpp index 3c228a716d5..10d91a1e4ef 100644 --- a/intern/cycles/graph/node.cpp +++ b/intern/cycles/graph/node.cpp @@ -14,12 +14,12 @@ * limitations under the License. */ -#include "node.h" -#include "node_type.h" +#include "graph/node.h" +#include "graph/node_type.h" -#include "util_foreach.h" -#include "util_param.h" -#include "util_transform.h" +#include "util/util_foreach.h" +#include "util/util_param.h" +#include "util/util_transform.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/graph/node.h b/intern/cycles/graph/node.h index 64410f4539b..53425f5faf1 100644 --- a/intern/cycles/graph/node.h +++ b/intern/cycles/graph/node.h @@ -16,11 +16,11 @@ #pragma once -#include "node_type.h" +#include "graph/node_type.h" -#include "util_map.h" -#include "util_param.h" -#include "util_vector.h" +#include "util/util_map.h" +#include "util/util_param.h" +#include "util/util_vector.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/graph/node_enum.h b/intern/cycles/graph/node_enum.h index 2bae531c036..4e40c294f4f 100644 --- a/intern/cycles/graph/node_enum.h +++ b/intern/cycles/graph/node_enum.h @@ -16,8 +16,8 @@ #pragma once -#include "util_map.h" -#include "util_param.h" +#include "util/util_map.h" +#include "util/util_param.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/graph/node_type.cpp b/intern/cycles/graph/node_type.cpp index 5b98de778ad..a3a8fa5f382 100644 --- a/intern/cycles/graph/node_type.cpp +++ b/intern/cycles/graph/node_type.cpp @@ -14,9 +14,9 @@ * limitations under the License. */ -#include "node_type.h" -#include "util_foreach.h" -#include "util_transform.h" +#include "graph/node_type.h" +#include "util/util_foreach.h" +#include "util/util_transform.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/graph/node_type.h b/intern/cycles/graph/node_type.h index 1fb135f6d22..7d46e31ce24 100644 --- a/intern/cycles/graph/node_type.h +++ b/intern/cycles/graph/node_type.h @@ -16,12 +16,12 @@ #pragma once -#include "node_enum.h" +#include "graph/node_enum.h" -#include "util_map.h" -#include "util_param.h" -#include "util_string.h" -#include "util_vector.h" +#include "util/util_map.h" +#include "util/util_param.h" +#include "util/util_string.h" +#include "util/util_vector.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/graph/node_xml.cpp b/intern/cycles/graph/node_xml.cpp index 590e09645ed..aad2740ffc0 100644 --- a/intern/cycles/graph/node_xml.cpp +++ b/intern/cycles/graph/node_xml.cpp @@ -14,11 +14,11 @@ * limitations under the License. */ -#include "node_xml.h" +#include "graph/node_xml.h" -#include "util_foreach.h" -#include "util_string.h" -#include "util_transform.h" +#include "util/util_foreach.h" +#include "util/util_string.h" +#include "util/util_transform.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/graph/node_xml.h b/intern/cycles/graph/node_xml.h index 7494c5e6e55..63e80bf79f2 100644 --- a/intern/cycles/graph/node_xml.h +++ b/intern/cycles/graph/node_xml.h @@ -16,11 +16,11 @@ #pragma once -#include "node.h" +#include "graph/node.h" -#include "util_map.h" -#include "util_string.h" -#include "util_xml.h" +#include "util/util_map.h" +#include "util/util_string.h" +#include "util/util_xml.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/kernel/CMakeLists.txt b/intern/cycles/kernel/CMakeLists.txt index 5f3ceb0f864..c3772dfa2d8 100644 --- a/intern/cycles/kernel/CMakeLists.txt +++ b/intern/cycles/kernel/CMakeLists.txt @@ -1,10 +1,7 @@ remove_extra_strict_flags() set(INC - . - ../util - osl - svm + .. ) set(INC_SYS @@ -13,19 +10,28 @@ set(INC_SYS set(SRC kernels/cpu/kernel.cpp + kernels/cpu/kernel_split.cpp kernels/opencl/kernel.cl + kernels/opencl/kernel_state_buffer_size.cl + kernels/opencl/kernel_split.cl kernels/opencl/kernel_data_init.cl + kernels/opencl/kernel_path_init.cl kernels/opencl/kernel_queue_enqueue.cl kernels/opencl/kernel_scene_intersect.cl kernels/opencl/kernel_lamp_emission.cl - kernels/opencl/kernel_background_buffer_update.cl + kernels/opencl/kernel_do_volume.cl + kernels/opencl/kernel_indirect_background.cl kernels/opencl/kernel_shader_eval.cl kernels/opencl/kernel_holdout_emission_blurring_pathtermination_ao.cl + kernels/opencl/kernel_subsurface_scatter.cl kernels/opencl/kernel_direct_lighting.cl - kernels/opencl/kernel_shadow_blocked.cl + kernels/opencl/kernel_shadow_blocked_ao.cl + kernels/opencl/kernel_shadow_blocked_dl.cl kernels/opencl/kernel_next_iteration_setup.cl - kernels/opencl/kernel_sum_all_radiance.cl + kernels/opencl/kernel_indirect_subsurface.cl + kernels/opencl/kernel_buffer_update.cl kernels/cuda/kernel.cu + kernels/cuda/kernel_split.cu ) set(SRC_BVH_HEADERS @@ -68,6 +74,7 @@ set(SRC_HEADERS kernel_path_common.h kernel_path_state.h kernel_path_surface.h + kernel_path_subsurface.h kernel_path_volume.h kernel_projection.h kernel_queues.h @@ -88,6 +95,10 @@ set(SRC_KERNELS_CPU_HEADERS kernels/cpu/kernel_cpu_image.h ) +set(SRC_KERNELS_CUDA_HEADERS + kernels/cuda/kernel_config.h +) + set(SRC_CLOSURE_HEADERS closure/alloc.h closure/bsdf.h @@ -184,6 +195,7 @@ set(SRC_UTIL_HEADERS ../util/util_hash.h ../util/util_math.h ../util/util_math_fast.h + ../util/util_math_intersect.h ../util/util_static_assert.h ../util/util_transform.h ../util/util_texture.h @@ -191,17 +203,25 @@ set(SRC_UTIL_HEADERS ) set(SRC_SPLIT_HEADERS - split/kernel_background_buffer_update.h + split/kernel_buffer_update.h split/kernel_data_init.h split/kernel_direct_lighting.h + split/kernel_do_volume.h split/kernel_holdout_emission_blurring_pathtermination_ao.h + split/kernel_indirect_background.h + split/kernel_indirect_subsurface.h split/kernel_lamp_emission.h split/kernel_next_iteration_setup.h + split/kernel_path_init.h + split/kernel_queue_enqueue.h split/kernel_scene_intersect.h split/kernel_shader_eval.h - split/kernel_shadow_blocked.h + split/kernel_shadow_blocked_ao.h + split/kernel_shadow_blocked_dl.h split/kernel_split_common.h - split/kernel_sum_all_radiance.h + split/kernel_split_data.h + split/kernel_split_data_types.h + split/kernel_subsurface_scatter.h ) # CUDA module @@ -229,8 +249,9 @@ if(WITH_CYCLES_CUDA_BINARIES) endif() # build for each arch - set(cuda_sources kernels/cuda/kernel.cu + set(cuda_sources kernels/cuda/kernel.cu kernels/cuda/kernel_split.cu ${SRC_HEADERS} + ${SRC_KERNELS_CUDA_HEADERS} ${SRC_BVH_HEADERS} ${SRC_SVM_HEADERS} ${SRC_GEOM_HEADERS} @@ -239,15 +260,22 @@ if(WITH_CYCLES_CUDA_BINARIES) ) set(cuda_cubins) - macro(CYCLES_CUDA_KERNEL_ADD arch experimental) - if(${experimental}) - set(cuda_extra_flags "-D__KERNEL_EXPERIMENTAL__") - set(cuda_cubin kernel_experimental_${arch}.cubin) + macro(CYCLES_CUDA_KERNEL_ADD arch split experimental) + if(${split}) + set(cuda_extra_flags "-D__SPLIT__") + set(cuda_cubin kernel_split) else() set(cuda_extra_flags "") - set(cuda_cubin kernel_${arch}.cubin) + set(cuda_cubin kernel) endif() + if(${experimental}) + set(cuda_extra_flags ${cuda_extra_flags} -D__KERNEL_EXPERIMENTAL__) + set(cuda_cubin ${cuda_cubin}_experimental) + endif() + + set(cuda_cubin ${cuda_cubin}_${arch}.cubin) + if(WITH_CYCLES_DEBUG) set(cuda_debug_flags "-D__KERNEL_DEBUG__") else() @@ -260,13 +288,19 @@ if(WITH_CYCLES_CUDA_BINARIES) set(cuda_version_flags "-D__KERNEL_CUDA_VERSION__=${cuda_nvcc_version}") set(cuda_math_flags "--use_fast_math") + if(split) + set(cuda_kernel_src "/kernels/cuda/kernel_split.cu") + else() + set(cuda_kernel_src "/kernels/cuda/kernel.cu") + endif() + add_custom_command( OUTPUT ${cuda_cubin} COMMAND ${cuda_nvcc_command} -arch=${arch} ${CUDA_NVCC_FLAGS} -m${CUDA_BITS} - --cubin ${CMAKE_CURRENT_SOURCE_DIR}/kernels/cuda/kernel.cu + --cubin ${CMAKE_CURRENT_SOURCE_DIR}${cuda_kernel_src} -o ${CMAKE_CURRENT_BINARY_DIR}/${cuda_cubin} --ptxas-options="-v" ${cuda_arch_flags} @@ -274,8 +308,7 @@ if(WITH_CYCLES_CUDA_BINARIES) ${cuda_math_flags} ${cuda_extra_flags} ${cuda_debug_flags} - -I${CMAKE_CURRENT_SOURCE_DIR}/../util - -I${CMAKE_CURRENT_SOURCE_DIR}/svm + -I${CMAKE_CURRENT_SOURCE_DIR}/.. -DCCL_NAMESPACE_BEGIN= -DCCL_NAMESPACE_END= -DNVCC @@ -293,7 +326,12 @@ if(WITH_CYCLES_CUDA_BINARIES) foreach(arch ${CYCLES_CUDA_BINARIES_ARCH}) # Compile regular kernel - CYCLES_CUDA_KERNEL_ADD(${arch} FALSE) + CYCLES_CUDA_KERNEL_ADD(${arch} FALSE FALSE) + + if(WITH_CYCLES_CUDA_SPLIT_KERNEL_BINARIES) + # Compile split kernel + CYCLES_CUDA_KERNEL_ADD(${arch} TRUE FALSE) + endif() endforeach() add_custom_target(cycles_kernel_cuda ALL DEPENDS ${cuda_cubins}) @@ -311,36 +349,50 @@ endif() include_directories(${INC}) include_directories(SYSTEM ${INC_SYS}) +set_source_files_properties(kernels/cpu/kernel.cpp PROPERTIES COMPILE_FLAGS "${CYCLES_KERNEL_FLAGS}") +set_source_files_properties(kernels/cpu/kernel_split.cpp PROPERTIES COMPILE_FLAGS "${CYCLES_KERNEL_FLAGS}") + if(CXX_HAS_SSE) list(APPEND SRC kernels/cpu/kernel_sse2.cpp kernels/cpu/kernel_sse3.cpp kernels/cpu/kernel_sse41.cpp + kernels/cpu/kernel_split_sse2.cpp + kernels/cpu/kernel_split_sse3.cpp + kernels/cpu/kernel_split_sse41.cpp ) set_source_files_properties(kernels/cpu/kernel_sse2.cpp PROPERTIES COMPILE_FLAGS "${CYCLES_SSE2_KERNEL_FLAGS}") set_source_files_properties(kernels/cpu/kernel_sse3.cpp PROPERTIES COMPILE_FLAGS "${CYCLES_SSE3_KERNEL_FLAGS}") set_source_files_properties(kernels/cpu/kernel_sse41.cpp PROPERTIES COMPILE_FLAGS "${CYCLES_SSE41_KERNEL_FLAGS}") + set_source_files_properties(kernels/cpu/kernel_split_sse2.cpp PROPERTIES COMPILE_FLAGS "${CYCLES_SSE2_KERNEL_FLAGS}") + set_source_files_properties(kernels/cpu/kernel_split_sse3.cpp PROPERTIES COMPILE_FLAGS "${CYCLES_SSE3_KERNEL_FLAGS}") + set_source_files_properties(kernels/cpu/kernel_split_sse41.cpp PROPERTIES COMPILE_FLAGS "${CYCLES_SSE41_KERNEL_FLAGS}") endif() if(CXX_HAS_AVX) list(APPEND SRC kernels/cpu/kernel_avx.cpp + kernels/cpu/kernel_split_avx.cpp ) set_source_files_properties(kernels/cpu/kernel_avx.cpp PROPERTIES COMPILE_FLAGS "${CYCLES_AVX_KERNEL_FLAGS}") + set_source_files_properties(kernels/cpu/kernel_split_avx.cpp PROPERTIES COMPILE_FLAGS "${CYCLES_AVX_KERNEL_FLAGS}") endif() if(CXX_HAS_AVX2) list(APPEND SRC kernels/cpu/kernel_avx2.cpp + kernels/cpu/kernel_split_avx2.cpp ) set_source_files_properties(kernels/cpu/kernel_avx2.cpp PROPERTIES COMPILE_FLAGS "${CYCLES_AVX2_KERNEL_FLAGS}") + set_source_files_properties(kernels/cpu/kernel_split_avx2.cpp PROPERTIES COMPILE_FLAGS "${CYCLES_AVX2_KERNEL_FLAGS}") endif() add_library(cycles_kernel ${SRC} ${SRC_HEADERS} ${SRC_KERNELS_CPU_HEADERS} + ${SRC_KERNELS_CUDA_HEADERS} ${SRC_BVH_HEADERS} ${SRC_CLOSURE_HEADERS} ${SRC_SVM_HEADERS} @@ -362,24 +414,33 @@ endif() #add_custom_target(cycles_kernel_preprocess ALL DEPENDS ${KERNEL_PREPROCESSED}) #delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "${KERNEL_PREPROCESSED}" ${CYCLES_INSTALL_PATH}/kernel) -delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "kernels/opencl/kernel.cl" ${CYCLES_INSTALL_PATH}/kernel/kernels/opencl) -delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "kernels/opencl/kernel_data_init.cl" ${CYCLES_INSTALL_PATH}/kernel/kernels/opencl) -delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "kernels/opencl/kernel_queue_enqueue.cl" ${CYCLES_INSTALL_PATH}/kernel/kernels/opencl) -delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "kernels/opencl/kernel_scene_intersect.cl" ${CYCLES_INSTALL_PATH}/kernel/kernels/opencl) -delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "kernels/opencl/kernel_lamp_emission.cl" ${CYCLES_INSTALL_PATH}/kernel/kernels/opencl) -delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "kernels/opencl/kernel_background_buffer_update.cl" ${CYCLES_INSTALL_PATH}/kernel/kernels/opencl) -delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "kernels/opencl/kernel_shader_eval.cl" ${CYCLES_INSTALL_PATH}/kernel/kernels/opencl) -delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "kernels/opencl/kernel_holdout_emission_blurring_pathtermination_ao.cl" ${CYCLES_INSTALL_PATH}/kernel/kernels/opencl) -delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "kernels/opencl/kernel_direct_lighting.cl" ${CYCLES_INSTALL_PATH}/kernel/kernels/opencl) -delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "kernels/opencl/kernel_shadow_blocked.cl" ${CYCLES_INSTALL_PATH}/kernel/kernels/opencl) -delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "kernels/opencl/kernel_next_iteration_setup.cl" ${CYCLES_INSTALL_PATH}/kernel/kernels/opencl) -delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "kernels/opencl/kernel_sum_all_radiance.cl" ${CYCLES_INSTALL_PATH}/kernel/kernels/opencl) -delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "kernels/cuda/kernel.cu" ${CYCLES_INSTALL_PATH}/kernel/kernels/cuda) -delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "${SRC_HEADERS}" ${CYCLES_INSTALL_PATH}/kernel) -delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "${SRC_BVH_HEADERS}" ${CYCLES_INSTALL_PATH}/kernel/bvh) -delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "${SRC_CLOSURE_HEADERS}" ${CYCLES_INSTALL_PATH}/kernel/closure) -delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "${SRC_SVM_HEADERS}" ${CYCLES_INSTALL_PATH}/kernel/svm) -delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "${SRC_GEOM_HEADERS}" ${CYCLES_INSTALL_PATH}/kernel/geom) -delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "${SRC_UTIL_HEADERS}" ${CYCLES_INSTALL_PATH}/kernel) -delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "${SRC_SPLIT_HEADERS}" ${CYCLES_INSTALL_PATH}/kernel/split) +delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "kernels/opencl/kernel.cl" ${CYCLES_INSTALL_PATH}/source/kernel/kernels/opencl) +delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "kernels/opencl/kernel_state_buffer_size.cl" ${CYCLES_INSTALL_PATH}/source/kernel/kernels/opencl) +delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "kernels/opencl/kernel_split.cl" ${CYCLES_INSTALL_PATH}/source/kernel/kernels/opencl) +delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "kernels/opencl/kernel_data_init.cl" ${CYCLES_INSTALL_PATH}/source/kernel/kernels/opencl) +delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "kernels/opencl/kernel_path_init.cl" ${CYCLES_INSTALL_PATH}/source/kernel/kernels/opencl) +delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "kernels/opencl/kernel_queue_enqueue.cl" ${CYCLES_INSTALL_PATH}/source/kernel/kernels/opencl) +delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "kernels/opencl/kernel_scene_intersect.cl" ${CYCLES_INSTALL_PATH}/source/kernel/kernels/opencl) +delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "kernels/opencl/kernel_lamp_emission.cl" ${CYCLES_INSTALL_PATH}/source/kernel/kernels/opencl) +delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "kernels/opencl/kernel_do_volume.cl" ${CYCLES_INSTALL_PATH}/source/kernel/kernels/opencl) +delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "kernels/opencl/kernel_indirect_background.cl" ${CYCLES_INSTALL_PATH}/source/kernel/kernels/opencl) +delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "kernels/opencl/kernel_shader_eval.cl" ${CYCLES_INSTALL_PATH}/source/kernel/kernels/opencl) +delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "kernels/opencl/kernel_holdout_emission_blurring_pathtermination_ao.cl" ${CYCLES_INSTALL_PATH}/source/kernel/kernels/opencl) +delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "kernels/opencl/kernel_subsurface_scatter.cl" ${CYCLES_INSTALL_PATH}/source/kernel/kernels/opencl) +delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "kernels/opencl/kernel_direct_lighting.cl" ${CYCLES_INSTALL_PATH}/source/kernel/kernels/opencl) +delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "kernels/opencl/kernel_shadow_blocked_ao.cl" ${CYCLES_INSTALL_PATH}/source/kernel/kernels/opencl) +delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "kernels/opencl/kernel_shadow_blocked_dl.cl" ${CYCLES_INSTALL_PATH}/source/kernel/kernels/opencl) +delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "kernels/opencl/kernel_next_iteration_setup.cl" ${CYCLES_INSTALL_PATH}/source/kernel/kernels/opencl) +delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "kernels/opencl/kernel_indirect_subsurface.cl" ${CYCLES_INSTALL_PATH}/source/kernel/kernels/opencl) +delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "kernels/opencl/kernel_buffer_update.cl" ${CYCLES_INSTALL_PATH}/source/kernel/kernels/opencl) +delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "kernels/cuda/kernel.cu" ${CYCLES_INSTALL_PATH}/source/kernel/kernels/cuda) +delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "kernels/cuda/kernel_split.cu" ${CYCLES_INSTALL_PATH}/source/kernel/kernels/cuda) +delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "${SRC_HEADERS}" ${CYCLES_INSTALL_PATH}/source/kernel) +delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "${SRC_KERNELS_CUDA_HEADERS}" ${CYCLES_INSTALL_PATH}/source/kernel/kernels/cuda) +delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "${SRC_BVH_HEADERS}" ${CYCLES_INSTALL_PATH}/source/kernel/bvh) +delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "${SRC_CLOSURE_HEADERS}" ${CYCLES_INSTALL_PATH}/source/kernel/closure) +delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "${SRC_SVM_HEADERS}" ${CYCLES_INSTALL_PATH}/source/kernel/svm) +delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "${SRC_GEOM_HEADERS}" ${CYCLES_INSTALL_PATH}/source/kernel/geom) +delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "${SRC_UTIL_HEADERS}" ${CYCLES_INSTALL_PATH}/source/util) +delayed_install(${CMAKE_CURRENT_SOURCE_DIR} "${SRC_SPLIT_HEADERS}" ${CYCLES_INSTALL_PATH}/source/kernel/split) diff --git a/intern/cycles/kernel/bvh/bvh.h b/intern/cycles/kernel/bvh/bvh.h index 36798982653..85741016b25 100644 --- a/intern/cycles/kernel/bvh/bvh.h +++ b/intern/cycles/kernel/bvh/bvh.h @@ -27,43 +27,43 @@ CCL_NAMESPACE_BEGIN -#include "bvh_types.h" +#include "kernel/bvh/bvh_types.h" /* Common QBVH functions. */ #ifdef __QBVH__ -# include "qbvh_nodes.h" +# include "kernel/bvh/qbvh_nodes.h" #endif /* Regular BVH traversal */ -#include "bvh_nodes.h" +#include "kernel/bvh/bvh_nodes.h" #define BVH_FUNCTION_NAME bvh_intersect #define BVH_FUNCTION_FEATURES 0 -#include "bvh_traversal.h" +#include "kernel/bvh/bvh_traversal.h" #if defined(__INSTANCING__) # define BVH_FUNCTION_NAME bvh_intersect_instancing # define BVH_FUNCTION_FEATURES BVH_INSTANCING -# include "bvh_traversal.h" +# include "kernel/bvh/bvh_traversal.h" #endif #if defined(__HAIR__) # define BVH_FUNCTION_NAME bvh_intersect_hair # define BVH_FUNCTION_FEATURES BVH_INSTANCING|BVH_HAIR|BVH_HAIR_MINIMUM_WIDTH -# include "bvh_traversal.h" +# include "kernel/bvh/bvh_traversal.h" #endif #if defined(__OBJECT_MOTION__) # define BVH_FUNCTION_NAME bvh_intersect_motion # define BVH_FUNCTION_FEATURES BVH_INSTANCING|BVH_MOTION -# include "bvh_traversal.h" +# include "kernel/bvh/bvh_traversal.h" #endif #if defined(__HAIR__) && defined(__OBJECT_MOTION__) # define BVH_FUNCTION_NAME bvh_intersect_hair_motion # define BVH_FUNCTION_FEATURES BVH_INSTANCING|BVH_HAIR|BVH_HAIR_MINIMUM_WIDTH|BVH_MOTION -# include "bvh_traversal.h" +# include "kernel/bvh/bvh_traversal.h" #endif /* Subsurface scattering BVH traversal */ @@ -71,12 +71,12 @@ CCL_NAMESPACE_BEGIN #if defined(__SUBSURFACE__) # define BVH_FUNCTION_NAME bvh_intersect_subsurface # define BVH_FUNCTION_FEATURES BVH_HAIR -# include "bvh_subsurface.h" +# include "kernel/bvh/bvh_subsurface.h" # if defined(__OBJECT_MOTION__) # define BVH_FUNCTION_NAME bvh_intersect_subsurface_motion # define BVH_FUNCTION_FEATURES BVH_MOTION|BVH_HAIR -# include "bvh_subsurface.h" +# include "kernel/bvh/bvh_subsurface.h" # endif #endif /* __SUBSURFACE__ */ @@ -85,18 +85,18 @@ CCL_NAMESPACE_BEGIN #if defined(__VOLUME__) # define BVH_FUNCTION_NAME bvh_intersect_volume # define BVH_FUNCTION_FEATURES BVH_HAIR -# include "bvh_volume.h" +# include "kernel/bvh/bvh_volume.h" # if defined(__INSTANCING__) # define BVH_FUNCTION_NAME bvh_intersect_volume_instancing # define BVH_FUNCTION_FEATURES BVH_INSTANCING|BVH_HAIR -# include "bvh_volume.h" +# include "kernel/bvh/bvh_volume.h" # endif # if defined(__OBJECT_MOTION__) # define BVH_FUNCTION_NAME bvh_intersect_volume_motion # define BVH_FUNCTION_FEATURES BVH_INSTANCING|BVH_MOTION|BVH_HAIR -# include "bvh_volume.h" +# include "kernel/bvh/bvh_volume.h" # endif #endif /* __VOLUME__ */ @@ -105,30 +105,30 @@ CCL_NAMESPACE_BEGIN #if defined(__SHADOW_RECORD_ALL__) # define BVH_FUNCTION_NAME bvh_intersect_shadow_all # define BVH_FUNCTION_FEATURES 0 -# include "bvh_shadow_all.h" +# include "kernel/bvh/bvh_shadow_all.h" # if defined(__INSTANCING__) # define BVH_FUNCTION_NAME bvh_intersect_shadow_all_instancing # define BVH_FUNCTION_FEATURES BVH_INSTANCING -# include "bvh_shadow_all.h" +# include "kernel/bvh/bvh_shadow_all.h" # endif # if defined(__HAIR__) # define BVH_FUNCTION_NAME bvh_intersect_shadow_all_hair # define BVH_FUNCTION_FEATURES BVH_INSTANCING|BVH_HAIR -# include "bvh_shadow_all.h" +# include "kernel/bvh/bvh_shadow_all.h" # endif # if defined(__OBJECT_MOTION__) # define BVH_FUNCTION_NAME bvh_intersect_shadow_all_motion # define BVH_FUNCTION_FEATURES BVH_INSTANCING|BVH_MOTION -# include "bvh_shadow_all.h" +# include "kernel/bvh/bvh_shadow_all.h" # endif # if defined(__HAIR__) && defined(__OBJECT_MOTION__) # define BVH_FUNCTION_NAME bvh_intersect_shadow_all_hair_motion # define BVH_FUNCTION_FEATURES BVH_INSTANCING|BVH_HAIR|BVH_MOTION -# include "bvh_shadow_all.h" +# include "kernel/bvh/bvh_shadow_all.h" # endif #endif /* __SHADOW_RECORD_ALL__ */ @@ -137,18 +137,18 @@ CCL_NAMESPACE_BEGIN #if defined(__VOLUME_RECORD_ALL__) # define BVH_FUNCTION_NAME bvh_intersect_volume_all # define BVH_FUNCTION_FEATURES BVH_HAIR -# include "bvh_volume_all.h" +# include "kernel/bvh/bvh_volume_all.h" # if defined(__INSTANCING__) # define BVH_FUNCTION_NAME bvh_intersect_volume_all_instancing # define BVH_FUNCTION_FEATURES BVH_INSTANCING|BVH_HAIR -# include "bvh_volume_all.h" +# include "kernel/bvh/bvh_volume_all.h" # endif # if defined(__OBJECT_MOTION__) # define BVH_FUNCTION_NAME bvh_intersect_volume_all_motion # define BVH_FUNCTION_FEATURES BVH_INSTANCING|BVH_MOTION|BVH_HAIR -# include "bvh_volume_all.h" +# include "kernel/bvh/bvh_volume_all.h" # endif #endif /* __VOLUME_RECORD_ALL__ */ @@ -202,8 +202,9 @@ ccl_device_intersect bool scene_intersect(KernelGlobals *kg, } #ifdef __SUBSURFACE__ +/* Note: ray is passed by value to work around a possible CUDA compiler bug. */ ccl_device_intersect void scene_intersect_subsurface(KernelGlobals *kg, - const Ray *ray, + const Ray ray, SubsurfaceIntersection *ss_isect, int subsurface_object, uint *lcg_state, @@ -212,7 +213,7 @@ ccl_device_intersect void scene_intersect_subsurface(KernelGlobals *kg, #ifdef __OBJECT_MOTION__ if(kernel_data.bvh.have_motion) { return bvh_intersect_subsurface_motion(kg, - ray, + &ray, ss_isect, subsurface_object, lcg_state, @@ -220,7 +221,7 @@ ccl_device_intersect void scene_intersect_subsurface(KernelGlobals *kg, } #endif /* __OBJECT_MOTION__ */ return bvh_intersect_subsurface(kg, - ray, + &ray, ss_isect, subsurface_object, lcg_state, @@ -229,30 +230,63 @@ ccl_device_intersect void scene_intersect_subsurface(KernelGlobals *kg, #endif #ifdef __SHADOW_RECORD_ALL__ -ccl_device_intersect bool scene_intersect_shadow_all(KernelGlobals *kg, const Ray *ray, Intersection *isect, uint max_hits, uint *num_hits) +ccl_device_intersect bool scene_intersect_shadow_all(KernelGlobals *kg, + const Ray *ray, + Intersection *isect, + int skip_object, + uint max_hits, + uint *num_hits) { # ifdef __OBJECT_MOTION__ if(kernel_data.bvh.have_motion) { # ifdef __HAIR__ - if(kernel_data.bvh.have_curves) - return bvh_intersect_shadow_all_hair_motion(kg, ray, isect, max_hits, num_hits); + if(kernel_data.bvh.have_curves) { + return bvh_intersect_shadow_all_hair_motion(kg, + ray, + isect, + skip_object, + max_hits, + num_hits); + } # endif /* __HAIR__ */ - return bvh_intersect_shadow_all_motion(kg, ray, isect, max_hits, num_hits); + return bvh_intersect_shadow_all_motion(kg, + ray, + isect, + skip_object, + max_hits, + num_hits); } # endif /* __OBJECT_MOTION__ */ # ifdef __HAIR__ - if(kernel_data.bvh.have_curves) - return bvh_intersect_shadow_all_hair(kg, ray, isect, max_hits, num_hits); + if(kernel_data.bvh.have_curves) { + return bvh_intersect_shadow_all_hair(kg, + ray, + isect, + skip_object, + max_hits, + num_hits); + } # endif /* __HAIR__ */ # ifdef __INSTANCING__ - if(kernel_data.bvh.have_instancing) - return bvh_intersect_shadow_all_instancing(kg, ray, isect, max_hits, num_hits); + if(kernel_data.bvh.have_instancing) { + return bvh_intersect_shadow_all_instancing(kg, + ray, + isect, + skip_object, + max_hits, + num_hits); + } # endif /* __INSTANCING__ */ - return bvh_intersect_shadow_all(kg, ray, isect, max_hits, num_hits); + return bvh_intersect_shadow_all(kg, + ray, + isect, + skip_object, + max_hits, + num_hits); } #endif /* __SHADOW_RECORD_ALL__ */ @@ -357,7 +391,7 @@ ccl_device_inline float3 ray_offset(float3 P, float3 Ng) #endif } -#if defined(__SHADOW_RECORD_ALL__) || defined (__VOLUME_RECORD_ALL__) +#if defined(__VOLUME_RECORD_ALL__) || (defined(__SHADOW_RECORD_ALL__) && defined(__KERNEL_CPU__)) /* ToDo: Move to another file? */ ccl_device int intersections_compare(const void *a, const void *b) { @@ -373,5 +407,28 @@ ccl_device int intersections_compare(const void *a, const void *b) } #endif -CCL_NAMESPACE_END +#if defined(__SHADOW_RECORD_ALL__) +ccl_device_inline void sort_intersections(Intersection *hits, uint num_hits) +{ +#ifdef __KERNEL_GPU__ + /* Use bubble sort which has more friendly memory pattern on GPU. */ + bool swapped; + do { + swapped = false; + for(int j = 0; j < num_hits - 1; ++j) { + if(hits[j].t > hits[j + 1].t) { + struct Intersection tmp = hits[j]; + hits[j] = hits[j + 1]; + hits[j + 1] = tmp; + swapped = true; + } + } + --num_hits; + } while(swapped); +#else + qsort(hits, num_hits, sizeof(Intersection), intersections_compare); +#endif +} +#endif /* __SHADOW_RECORD_ALL__ | __VOLUME_RECORD_ALL__ */ +CCL_NAMESPACE_END diff --git a/intern/cycles/kernel/bvh/bvh_nodes.h b/intern/cycles/kernel/bvh/bvh_nodes.h index 726bef1794c..74a9ebf14e4 100644 --- a/intern/cycles/kernel/bvh/bvh_nodes.h +++ b/intern/cycles/kernel/bvh/bvh_nodes.h @@ -17,8 +17,8 @@ // TODO(sergey): Look into avoid use of full Transform and use 3x3 matrix and // 3-vector which might be faster. ccl_device_forceinline Transform bvh_unaligned_node_fetch_space(KernelGlobals *kg, - int node_addr, - int child) + int node_addr, + int child) { Transform space; const int child_addr = node_addr + child * 3; @@ -31,12 +31,12 @@ ccl_device_forceinline Transform bvh_unaligned_node_fetch_space(KernelGlobals *k #if !defined(__KERNEL_SSE2__) ccl_device_forceinline int bvh_aligned_node_intersect(KernelGlobals *kg, - const float3 P, - const float3 idir, - const float t, - const int node_addr, - const uint visibility, - float dist[2]) + const float3 P, + const float3 idir, + const float t, + const int node_addr, + const uint visibility, + float dist[2]) { /* fetch node data */ @@ -78,14 +78,14 @@ ccl_device_forceinline int bvh_aligned_node_intersect(KernelGlobals *kg, } ccl_device_forceinline int bvh_aligned_node_intersect_robust(KernelGlobals *kg, - const float3 P, - const float3 idir, - const float t, - const float difl, - const float extmax, - const int node_addr, - const uint visibility, - float dist[2]) + const float3 P, + const float3 idir, + const float t, + const float difl, + const float extmax, + const int node_addr, + const uint visibility, + float dist[2]) { /* fetch node data */ @@ -203,13 +203,13 @@ ccl_device_forceinline bool bvh_unaligned_node_intersect_child_robust( } ccl_device_forceinline int bvh_unaligned_node_intersect(KernelGlobals *kg, - const float3 P, - const float3 dir, - const float3 idir, - const float t, - const int node_addr, - const uint visibility, - float dist[2]) + const float3 P, + const float3 dir, + const float3 idir, + const float t, + const int node_addr, + const uint visibility, + float dist[2]) { int mask = 0; float4 cnodes = kernel_tex_fetch(__bvh_nodes, node_addr+0); @@ -233,15 +233,15 @@ ccl_device_forceinline int bvh_unaligned_node_intersect(KernelGlobals *kg, } ccl_device_forceinline int bvh_unaligned_node_intersect_robust(KernelGlobals *kg, - const float3 P, - const float3 dir, - const float3 idir, - const float t, - const float difl, - const float extmax, - const int node_addr, - const uint visibility, - float dist[2]) + const float3 P, + const float3 dir, + const float3 idir, + const float t, + const float difl, + const float extmax, + const int node_addr, + const uint visibility, + float dist[2]) { int mask = 0; float4 cnodes = kernel_tex_fetch(__bvh_nodes, node_addr+0); @@ -265,13 +265,13 @@ ccl_device_forceinline int bvh_unaligned_node_intersect_robust(KernelGlobals *kg } ccl_device_forceinline int bvh_node_intersect(KernelGlobals *kg, - const float3 P, - const float3 dir, - const float3 idir, - const float t, - const int node_addr, - const uint visibility, - float dist[2]) + const float3 P, + const float3 dir, + const float3 idir, + const float t, + const int node_addr, + const uint visibility, + float dist[2]) { float4 node = kernel_tex_fetch(__bvh_nodes, node_addr); if(__float_as_uint(node.x) & PATH_RAY_NODE_UNALIGNED) { @@ -296,15 +296,15 @@ ccl_device_forceinline int bvh_node_intersect(KernelGlobals *kg, } ccl_device_forceinline int bvh_node_intersect_robust(KernelGlobals *kg, - const float3 P, - const float3 dir, - const float3 idir, - const float t, - const float difl, - const float extmax, - const int node_addr, - const uint visibility, - float dist[2]) + const float3 P, + const float3 dir, + const float3 idir, + const float t, + const float difl, + const float extmax, + const int node_addr, + const uint visibility, + float dist[2]) { float4 node = kernel_tex_fetch(__bvh_nodes, node_addr); if(__float_as_uint(node.x) & PATH_RAY_NODE_UNALIGNED) { @@ -442,19 +442,19 @@ ccl_device_forceinline int bvh_aligned_node_intersect_robust( } ccl_device_forceinline int bvh_unaligned_node_intersect(KernelGlobals *kg, - const float3 P, - const float3 dir, - const ssef& isect_near, - const ssef& isect_far, - const int node_addr, - const uint visibility, - float dist[2]) + const float3 P, + const float3 dir, + const ssef& isect_near, + const ssef& isect_far, + const int node_addr, + const uint visibility, + float dist[2]) { Transform space0 = bvh_unaligned_node_fetch_space(kg, node_addr, 0); Transform space1 = bvh_unaligned_node_fetch_space(kg, node_addr, 1); float3 aligned_dir0 = transform_direction(&space0, dir), - aligned_dir1 = transform_direction(&space1, dir);; + aligned_dir1 = transform_direction(&space1, dir); float3 aligned_P0 = transform_point(&space0, P), aligned_P1 = transform_point(&space1, P); float3 nrdir0 = -bvh_inverse_direction(aligned_dir0), @@ -503,20 +503,20 @@ ccl_device_forceinline int bvh_unaligned_node_intersect(KernelGlobals *kg, } ccl_device_forceinline int bvh_unaligned_node_intersect_robust(KernelGlobals *kg, - const float3 P, - const float3 dir, - const ssef& isect_near, - const ssef& isect_far, - const float difl, - const int node_addr, - const uint visibility, - float dist[2]) + const float3 P, + const float3 dir, + const ssef& isect_near, + const ssef& isect_far, + const float difl, + const int node_addr, + const uint visibility, + float dist[2]) { Transform space0 = bvh_unaligned_node_fetch_space(kg, node_addr, 0); Transform space1 = bvh_unaligned_node_fetch_space(kg, node_addr, 1); float3 aligned_dir0 = transform_direction(&space0, dir), - aligned_dir1 = transform_direction(&space1, dir);; + aligned_dir1 = transform_direction(&space1, dir); float3 aligned_P0 = transform_point(&space0, P), aligned_P1 = transform_point(&space1, P); float3 nrdir0 = -bvh_inverse_direction(aligned_dir0), @@ -574,17 +574,17 @@ ccl_device_forceinline int bvh_unaligned_node_intersect_robust(KernelGlobals *kg } ccl_device_forceinline int bvh_node_intersect(KernelGlobals *kg, - const float3& P, - const float3& dir, - const ssef& isect_near, - const ssef& isect_far, - const ssef& tsplat, - const ssef Psplat[3], - const ssef idirsplat[3], - const shuffle_swap_t shufflexyz[3], - const int node_addr, - const uint visibility, - float dist[2]) + const float3& P, + const float3& dir, + const ssef& isect_near, + const ssef& isect_far, + const ssef& tsplat, + const ssef Psplat[3], + const ssef idirsplat[3], + const shuffle_swap_t shufflexyz[3], + const int node_addr, + const uint visibility, + float dist[2]) { float4 node = kernel_tex_fetch(__bvh_nodes, node_addr); if(__float_as_uint(node.x) & PATH_RAY_NODE_UNALIGNED) { @@ -612,19 +612,19 @@ ccl_device_forceinline int bvh_node_intersect(KernelGlobals *kg, } ccl_device_forceinline int bvh_node_intersect_robust(KernelGlobals *kg, - const float3& P, - const float3& dir, - const ssef& isect_near, - const ssef& isect_far, - const ssef& tsplat, - const ssef Psplat[3], - const ssef idirsplat[3], - const shuffle_swap_t shufflexyz[3], - const float difl, - const float extmax, - const int node_addr, - const uint visibility, - float dist[2]) + const float3& P, + const float3& dir, + const ssef& isect_near, + const ssef& isect_far, + const ssef& tsplat, + const ssef Psplat[3], + const ssef idirsplat[3], + const shuffle_swap_t shufflexyz[3], + const float difl, + const float extmax, + const int node_addr, + const uint visibility, + float dist[2]) { float4 node = kernel_tex_fetch(__bvh_nodes, node_addr); if(__float_as_uint(node.x) & PATH_RAY_NODE_UNALIGNED) { diff --git a/intern/cycles/kernel/bvh/bvh_shadow_all.h b/intern/cycles/kernel/bvh/bvh_shadow_all.h index df33a86bb18..267e098f912 100644 --- a/intern/cycles/kernel/bvh/bvh_shadow_all.h +++ b/intern/cycles/kernel/bvh/bvh_shadow_all.h @@ -18,7 +18,7 @@ */ #ifdef __QBVH__ -# include "qbvh_shadow_all.h" +# include "kernel/bvh/qbvh_shadow_all.h" #endif #if BVH_FEATURE(BVH_HAIR) @@ -45,6 +45,7 @@ ccl_device_inline bool BVH_FUNCTION_FULL_NAME(BVH)(KernelGlobals *kg, const Ray *ray, Intersection *isect_array, + const int skip_object, const uint max_hits, uint *num_hits) { @@ -100,9 +101,6 @@ bool BVH_FUNCTION_FULL_NAME(BVH)(KernelGlobals *kg, gen_idirsplat_swap(pn, shuf_identity, shuf_swap, idir, idirsplat, shufflexyz); #endif /* __KERNEL_SSE2__ */ - IsectPrecalc isect_precalc; - triangle_intersect_precalc(dir, &isect_precalc); - /* traversal loop */ do { do { @@ -189,6 +187,16 @@ bool BVH_FUNCTION_FULL_NAME(BVH)(KernelGlobals *kg, while(prim_addr < prim_addr2) { kernel_assert((kernel_tex_fetch(__prim_type, prim_addr) & PRIMITIVE_ALL) == p_type); +#ifdef __SHADOW_TRICKS__ + uint tri_object = (object == OBJECT_NONE) + ? kernel_tex_fetch(__prim_object, prim_addr) + : object; + if(tri_object == skip_object) { + ++prim_addr; + continue; + } +#endif + bool hit; /* todo: specialized intersect functions which don't fill in @@ -198,9 +206,9 @@ bool BVH_FUNCTION_FULL_NAME(BVH)(KernelGlobals *kg, switch(p_type) { case PRIMITIVE_TRIANGLE: { hit = triangle_intersect(kg, - &isect_precalc, isect_array, P, + dir, PATH_RAY_SHADOW, object, prim_addr); @@ -309,12 +317,11 @@ bool BVH_FUNCTION_FULL_NAME(BVH)(KernelGlobals *kg, object = kernel_tex_fetch(__prim_object, -prim_addr-1); # if BVH_FEATURE(BVH_MOTION) - bvh_instance_motion_push(kg, object, ray, &P, &dir, &idir, &isect_t, &ob_itfm); + isect_t = bvh_instance_motion_push(kg, object, ray, &P, &dir, &idir, isect_t, &ob_itfm); # else - bvh_instance_push(kg, object, ray, &P, &dir, &idir, &isect_t); + isect_t = bvh_instance_push(kg, object, ray, &P, &dir, &idir, isect_t); # endif - triangle_intersect_precalc(dir, &isect_precalc); num_hits_in_instance = 0; isect_array->t = isect_t; @@ -354,22 +361,17 @@ bool BVH_FUNCTION_FULL_NAME(BVH)(KernelGlobals *kg, bvh_instance_pop_factor(kg, object, ray, &P, &dir, &idir, &t_fac); # endif - triangle_intersect_precalc(dir, &isect_precalc); - /* scale isect->t to adjust for instancing */ for(int i = 0; i < num_hits_in_instance; i++) { (isect_array-i-1)->t *= t_fac; } } else { - float ignore_t = FLT_MAX; - # if BVH_FEATURE(BVH_MOTION) - bvh_instance_motion_pop(kg, object, ray, &P, &dir, &idir, &ignore_t, &ob_itfm); + bvh_instance_motion_pop(kg, object, ray, &P, &dir, &idir, FLT_MAX, &ob_itfm); # else - bvh_instance_pop(kg, object, ray, &P, &dir, &idir, &ignore_t); + bvh_instance_pop(kg, object, ray, &P, &dir, &idir, FLT_MAX); # endif - triangle_intersect_precalc(dir, &isect_precalc); } isect_t = tmax; @@ -400,6 +402,7 @@ bool BVH_FUNCTION_FULL_NAME(BVH)(KernelGlobals *kg, ccl_device_inline bool BVH_FUNCTION_NAME(KernelGlobals *kg, const Ray *ray, Intersection *isect_array, + const int skip_object, const uint max_hits, uint *num_hits) { @@ -408,6 +411,7 @@ ccl_device_inline bool BVH_FUNCTION_NAME(KernelGlobals *kg, return BVH_FUNCTION_FULL_NAME(QBVH)(kg, ray, isect_array, + skip_object, max_hits, num_hits); } @@ -418,6 +422,7 @@ ccl_device_inline bool BVH_FUNCTION_NAME(KernelGlobals *kg, return BVH_FUNCTION_FULL_NAME(BVH)(kg, ray, isect_array, + skip_object, max_hits, num_hits); } diff --git a/intern/cycles/kernel/bvh/bvh_subsurface.h b/intern/cycles/kernel/bvh/bvh_subsurface.h index 889bbca21e2..bda7e34907a 100644 --- a/intern/cycles/kernel/bvh/bvh_subsurface.h +++ b/intern/cycles/kernel/bvh/bvh_subsurface.h @@ -18,7 +18,7 @@ */ #ifdef __QBVH__ -# include "qbvh_subsurface.h" +# include "kernel/bvh/qbvh_subsurface.h" #endif #if BVH_FEATURE(BVH_HAIR) @@ -75,16 +75,16 @@ void BVH_FUNCTION_FULL_NAME(BVH)(KernelGlobals *kg, if(!(object_flag & SD_OBJECT_TRANSFORM_APPLIED)) { #if BVH_FEATURE(BVH_MOTION) Transform ob_itfm; - bvh_instance_motion_push(kg, - subsurface_object, - ray, - &P, - &dir, - &idir, - &isect_t, - &ob_itfm); + isect_t = bvh_instance_motion_push(kg, + subsurface_object, + ray, + &P, + &dir, + &idir, + isect_t, + &ob_itfm); #else - bvh_instance_push(kg, subsurface_object, ray, &P, &dir, &idir, &isect_t); + isect_t = bvh_instance_push(kg, subsurface_object, ray, &P, &dir, &idir, isect_t); #endif object = subsurface_object; } @@ -109,9 +109,6 @@ void BVH_FUNCTION_FULL_NAME(BVH)(KernelGlobals *kg, gen_idirsplat_swap(pn, shuf_identity, shuf_swap, idir, idirsplat, shufflexyz); #endif - IsectPrecalc isect_precalc; - triangle_intersect_precalc(dir, &isect_precalc); - /* traversal loop */ do { do { @@ -197,9 +194,9 @@ void BVH_FUNCTION_FULL_NAME(BVH)(KernelGlobals *kg, for(; prim_addr < prim_addr2; prim_addr++) { kernel_assert(kernel_tex_fetch(__prim_type, prim_addr) == type); triangle_intersect_subsurface(kg, - &isect_precalc, ss_isect, P, + dir, object, prim_addr, isect_t, diff --git a/intern/cycles/kernel/bvh/bvh_traversal.h b/intern/cycles/kernel/bvh/bvh_traversal.h index 80c8f31473a..c58d3b0316c 100644 --- a/intern/cycles/kernel/bvh/bvh_traversal.h +++ b/intern/cycles/kernel/bvh/bvh_traversal.h @@ -18,7 +18,7 @@ */ #ifdef __QBVH__ -# include "qbvh_traversal.h" +# include "kernel/bvh/qbvh_traversal.h" #endif #if BVH_FEATURE(BVH_HAIR) @@ -104,9 +104,6 @@ ccl_device_noinline bool BVH_FUNCTION_FULL_NAME(BVH)(KernelGlobals *kg, gen_idirsplat_swap(pn, shuf_identity, shuf_swap, idir, idirsplat, shufflexyz); #endif - IsectPrecalc isect_precalc; - triangle_intersect_precalc(dir, &isect_precalc); - /* traversal loop */ do { do { @@ -238,9 +235,9 @@ ccl_device_noinline bool BVH_FUNCTION_FULL_NAME(BVH)(KernelGlobals *kg, BVH_DEBUG_NEXT_INTERSECTION(); kernel_assert(kernel_tex_fetch(__prim_type, prim_addr) == type); if(triangle_intersect(kg, - &isect_precalc, isect, P, + dir, visibility, object, prim_addr)) @@ -354,11 +351,10 @@ ccl_device_noinline bool BVH_FUNCTION_FULL_NAME(BVH)(KernelGlobals *kg, object = kernel_tex_fetch(__prim_object, -prim_addr-1); # if BVH_FEATURE(BVH_MOTION) - bvh_instance_motion_push(kg, object, ray, &P, &dir, &idir, &isect->t, &ob_itfm); + isect->t = bvh_instance_motion_push(kg, object, ray, &P, &dir, &idir, isect->t, &ob_itfm); # else - bvh_instance_push(kg, object, ray, &P, &dir, &idir, &isect->t); + isect->t = bvh_instance_push(kg, object, ray, &P, &dir, &idir, isect->t); # endif - triangle_intersect_precalc(dir, &isect_precalc); # if defined(__KERNEL_SSE2__) Psplat[0] = ssef(P.x); @@ -391,11 +387,10 @@ ccl_device_noinline bool BVH_FUNCTION_FULL_NAME(BVH)(KernelGlobals *kg, /* instance pop */ # if BVH_FEATURE(BVH_MOTION) - bvh_instance_motion_pop(kg, object, ray, &P, &dir, &idir, &isect->t, &ob_itfm); + isect->t = bvh_instance_motion_pop(kg, object, ray, &P, &dir, &idir, isect->t, &ob_itfm); # else - bvh_instance_pop(kg, object, ray, &P, &dir, &idir, &isect->t); + isect->t = bvh_instance_pop(kg, object, ray, &P, &dir, &idir, isect->t); # endif - triangle_intersect_precalc(dir, &isect_precalc); # if defined(__KERNEL_SSE2__) Psplat[0] = ssef(P.x); diff --git a/intern/cycles/kernel/bvh/bvh_volume.h b/intern/cycles/kernel/bvh/bvh_volume.h index 57e5b8d736d..764aaee44a1 100644 --- a/intern/cycles/kernel/bvh/bvh_volume.h +++ b/intern/cycles/kernel/bvh/bvh_volume.h @@ -18,7 +18,7 @@ */ #ifdef __QBVH__ -# include "qbvh_volume.h" +# include "kernel/bvh/qbvh_volume.h" #endif #if BVH_FEATURE(BVH_HAIR) @@ -97,9 +97,6 @@ bool BVH_FUNCTION_FULL_NAME(BVH)(KernelGlobals *kg, gen_idirsplat_swap(pn, shuf_identity, shuf_swap, idir, idirsplat, shufflexyz); #endif - IsectPrecalc isect_precalc; - triangle_intersect_precalc(dir, &isect_precalc); - /* traversal loop */ do { do { @@ -194,9 +191,9 @@ bool BVH_FUNCTION_FULL_NAME(BVH)(KernelGlobals *kg, continue; } triangle_intersect(kg, - &isect_precalc, isect, P, + dir, visibility, object, prim_addr); @@ -238,13 +235,11 @@ bool BVH_FUNCTION_FULL_NAME(BVH)(KernelGlobals *kg, int object_flag = kernel_tex_fetch(__object_flag, object); if(object_flag & SD_OBJECT_HAS_VOLUME) { # if BVH_FEATURE(BVH_MOTION) - bvh_instance_motion_push(kg, object, ray, &P, &dir, &idir, &isect->t, &ob_itfm); + isect->t = bvh_instance_motion_push(kg, object, ray, &P, &dir, &idir, isect->t, &ob_itfm); # else - bvh_instance_push(kg, object, ray, &P, &dir, &idir, &isect->t); + isect->t = bvh_instance_push(kg, object, ray, &P, &dir, &idir, isect->t); # endif - triangle_intersect_precalc(dir, &isect_precalc); - # if defined(__KERNEL_SSE2__) Psplat[0] = ssef(P.x); Psplat[1] = ssef(P.y); @@ -281,13 +276,11 @@ bool BVH_FUNCTION_FULL_NAME(BVH)(KernelGlobals *kg, /* instance pop */ # if BVH_FEATURE(BVH_MOTION) - bvh_instance_motion_pop(kg, object, ray, &P, &dir, &idir, &isect->t, &ob_itfm); + isect->t = bvh_instance_motion_pop(kg, object, ray, &P, &dir, &idir, isect->t, &ob_itfm); # else - bvh_instance_pop(kg, object, ray, &P, &dir, &idir, &isect->t); + isect->t = bvh_instance_pop(kg, object, ray, &P, &dir, &idir, isect->t); # endif - triangle_intersect_precalc(dir, &isect_precalc); - # if defined(__KERNEL_SSE2__) Psplat[0] = ssef(P.x); Psplat[1] = ssef(P.y); diff --git a/intern/cycles/kernel/bvh/bvh_volume_all.h b/intern/cycles/kernel/bvh/bvh_volume_all.h index 5a1accebaa0..04ec334e54d 100644 --- a/intern/cycles/kernel/bvh/bvh_volume_all.h +++ b/intern/cycles/kernel/bvh/bvh_volume_all.h @@ -18,7 +18,7 @@ */ #ifdef __QBVH__ -# include "qbvh_volume_all.h" +# include "kernel/bvh/qbvh_volume_all.h" #endif #if BVH_FEATURE(BVH_HAIR) @@ -101,9 +101,6 @@ uint BVH_FUNCTION_FULL_NAME(BVH)(KernelGlobals *kg, gen_idirsplat_swap(pn, shuf_identity, shuf_swap, idir, idirsplat, shufflexyz); #endif /* __KERNEL_SSE2__ */ - IsectPrecalc isect_precalc; - triangle_intersect_precalc(dir, &isect_precalc); - /* traversal loop */ do { do { @@ -199,9 +196,9 @@ uint BVH_FUNCTION_FULL_NAME(BVH)(KernelGlobals *kg, continue; } hit = triangle_intersect(kg, - &isect_precalc, isect_array, P, + dir, visibility, object, prim_addr); @@ -288,14 +285,12 @@ uint BVH_FUNCTION_FULL_NAME(BVH)(KernelGlobals *kg, object = kernel_tex_fetch(__prim_object, -prim_addr-1); int object_flag = kernel_tex_fetch(__object_flag, object); if(object_flag & SD_OBJECT_HAS_VOLUME) { - # if BVH_FEATURE(BVH_MOTION) - bvh_instance_motion_push(kg, object, ray, &P, &dir, &idir, &isect_t, &ob_itfm); + isect_t = bvh_instance_motion_push(kg, object, ray, &P, &dir, &idir, isect_t, &ob_itfm); # else - bvh_instance_push(kg, object, ray, &P, &dir, &idir, &isect_t); + isect_t = bvh_instance_push(kg, object, ray, &P, &dir, &idir, isect_t); # endif - triangle_intersect_precalc(dir, &isect_precalc); num_hits_in_instance = 0; isect_array->t = isect_t; @@ -341,20 +336,17 @@ uint BVH_FUNCTION_FULL_NAME(BVH)(KernelGlobals *kg, # else bvh_instance_pop_factor(kg, object, ray, &P, &dir, &idir, &t_fac); # endif - triangle_intersect_precalc(dir, &isect_precalc); /* Scale isect->t to adjust for instancing. */ for(int i = 0; i < num_hits_in_instance; i++) { (isect_array-i-1)->t *= t_fac; } } else { - float ignore_t = FLT_MAX; # if BVH_FEATURE(BVH_MOTION) - bvh_instance_motion_pop(kg, object, ray, &P, &dir, &idir, &ignore_t, &ob_itfm); + bvh_instance_motion_pop(kg, object, ray, &P, &dir, &idir, FLT_MAX, &ob_itfm); # else - bvh_instance_pop(kg, object, ray, &P, &dir, &idir, &ignore_t); + bvh_instance_pop(kg, object, ray, &P, &dir, &idir, FLT_MAX); # endif - triangle_intersect_precalc(dir, &isect_precalc); } isect_t = tmax; diff --git a/intern/cycles/kernel/bvh/qbvh_shadow_all.h b/intern/cycles/kernel/bvh/qbvh_shadow_all.h index 607295f9ed5..ce474438f2c 100644 --- a/intern/cycles/kernel/bvh/qbvh_shadow_all.h +++ b/intern/cycles/kernel/bvh/qbvh_shadow_all.h @@ -33,6 +33,7 @@ ccl_device bool BVH_FUNCTION_FULL_NAME(QBVH)(KernelGlobals *kg, const Ray *ray, Intersection *isect_array, + const int skip_object, const uint max_hits, uint *num_hits) { @@ -96,15 +97,13 @@ ccl_device bool BVH_FUNCTION_FULL_NAME(QBVH)(KernelGlobals *kg, &near_x, &near_y, &near_z, &far_x, &far_y, &far_z); - IsectPrecalc isect_precalc; - triangle_intersect_precalc(dir, &isect_precalc); - /* Traversal loop. */ do { do { /* Traverse internal nodes. */ while(node_addr >= 0 && node_addr != ENTRYPOINT_SENTINEL) { float4 inodes = kernel_tex_fetch(__bvh_nodes, node_addr+0); + (void)inodes; if(false #ifdef __VISIBILITY_FLAG__ @@ -270,6 +269,16 @@ ccl_device bool BVH_FUNCTION_FULL_NAME(QBVH)(KernelGlobals *kg, while(prim_addr < prim_addr2) { kernel_assert((kernel_tex_fetch(__prim_type, prim_addr) & PRIMITIVE_ALL) == p_type); +#ifdef __SHADOW_TRICKS__ + uint tri_object = (object == OBJECT_NONE) + ? kernel_tex_fetch(__prim_object, prim_addr) + : object; + if(tri_object == skip_object) { + ++prim_addr; + continue; + } +#endif + bool hit; /* todo: specialized intersect functions which don't fill in @@ -279,9 +288,9 @@ ccl_device bool BVH_FUNCTION_FULL_NAME(QBVH)(KernelGlobals *kg, switch(p_type) { case PRIMITIVE_TRIANGLE: { hit = triangle_intersect(kg, - &isect_precalc, isect_array, P, + dir, PATH_RAY_SHADOW, object, prim_addr); @@ -390,9 +399,9 @@ ccl_device bool BVH_FUNCTION_FULL_NAME(QBVH)(KernelGlobals *kg, object = kernel_tex_fetch(__prim_object, -prim_addr-1); # if BVH_FEATURE(BVH_MOTION) - bvh_instance_motion_push(kg, object, ray, &P, &dir, &idir, &isect_t, &ob_itfm); + isect_t = bvh_instance_motion_push(kg, object, ray, &P, &dir, &idir, isect_t, &ob_itfm); # else - bvh_instance_push(kg, object, ray, &P, &dir, &idir, &isect_t); + isect_t = bvh_instance_push(kg, object, ray, &P, &dir, &idir, isect_t); # endif num_hits_in_instance = 0; @@ -414,8 +423,6 @@ ccl_device bool BVH_FUNCTION_FULL_NAME(QBVH)(KernelGlobals *kg, org4 = sse3f(ssef(P.x), ssef(P.y), ssef(P.z)); # endif - triangle_intersect_precalc(dir, &isect_precalc); - ++stack_ptr; kernel_assert(stack_ptr < BVH_QSTACK_SIZE); traversal_stack[stack_ptr].addr = ENTRYPOINT_SENTINEL; @@ -445,11 +452,10 @@ ccl_device bool BVH_FUNCTION_FULL_NAME(QBVH)(KernelGlobals *kg, } } else { - float ignore_t = FLT_MAX; # if BVH_FEATURE(BVH_MOTION) - bvh_instance_motion_pop(kg, object, ray, &P, &dir, &idir, &ignore_t, &ob_itfm); + bvh_instance_motion_pop(kg, object, ray, &P, &dir, &idir, FLT_MAX, &ob_itfm); # else - bvh_instance_pop(kg, object, ray, &P, &dir, &idir, &ignore_t); + bvh_instance_pop(kg, object, ray, &P, &dir, &idir, FLT_MAX); # endif } @@ -472,8 +478,6 @@ ccl_device bool BVH_FUNCTION_FULL_NAME(QBVH)(KernelGlobals *kg, org4 = sse3f(ssef(P.x), ssef(P.y), ssef(P.z)); # endif - triangle_intersect_precalc(dir, &isect_precalc); - object = OBJECT_NONE; node_addr = traversal_stack[stack_ptr].addr; --stack_ptr; diff --git a/intern/cycles/kernel/bvh/qbvh_subsurface.h b/intern/cycles/kernel/bvh/qbvh_subsurface.h index 84dc4003133..be7658d11d7 100644 --- a/intern/cycles/kernel/bvh/qbvh_subsurface.h +++ b/intern/cycles/kernel/bvh/qbvh_subsurface.h @@ -64,16 +64,16 @@ ccl_device void BVH_FUNCTION_FULL_NAME(QBVH)(KernelGlobals *kg, if(!(object_flag & SD_OBJECT_TRANSFORM_APPLIED)) { #if BVH_FEATURE(BVH_MOTION) Transform ob_itfm; - bvh_instance_motion_push(kg, - subsurface_object, - ray, - &P, - &dir, - &idir, - &isect_t, - &ob_itfm); + isect_t = bvh_instance_motion_push(kg, + subsurface_object, + ray, + &P, + &dir, + &idir, + isect_t, + &ob_itfm); #else - bvh_instance_push(kg, subsurface_object, ray, &P, &dir, &idir, &isect_t); + isect_t = bvh_instance_push(kg, subsurface_object, ray, &P, &dir, &idir, isect_t); #endif object = subsurface_object; } @@ -105,9 +105,6 @@ ccl_device void BVH_FUNCTION_FULL_NAME(QBVH)(KernelGlobals *kg, &near_x, &near_y, &near_z, &far_x, &far_y, &far_z); - IsectPrecalc isect_precalc; - triangle_intersect_precalc(dir, &isect_precalc); - /* Traversal loop. */ do { do { @@ -253,9 +250,9 @@ ccl_device void BVH_FUNCTION_FULL_NAME(QBVH)(KernelGlobals *kg, for(; prim_addr < prim_addr2; prim_addr++) { kernel_assert(kernel_tex_fetch(__prim_type, prim_addr) == type); triangle_intersect_subsurface(kg, - &isect_precalc, ss_isect, P, + dir, object, prim_addr, isect_t, diff --git a/intern/cycles/kernel/bvh/qbvh_traversal.h b/intern/cycles/kernel/bvh/qbvh_traversal.h index 10ae7bee852..fca75a1d416 100644 --- a/intern/cycles/kernel/bvh/qbvh_traversal.h +++ b/intern/cycles/kernel/bvh/qbvh_traversal.h @@ -106,15 +106,13 @@ ccl_device bool BVH_FUNCTION_FULL_NAME(QBVH)(KernelGlobals *kg, &near_x, &near_y, &near_z, &far_x, &far_y, &far_z); - IsectPrecalc isect_precalc; - triangle_intersect_precalc(dir, &isect_precalc); - /* Traversal loop. */ do { do { /* Traverse internal nodes. */ while(node_addr >= 0 && node_addr != ENTRYPOINT_SENTINEL) { float4 inodes = kernel_tex_fetch(__bvh_nodes, node_addr+0); + (void)inodes; if(UNLIKELY(node_dist > isect->t) #if BVH_FEATURE(BVH_MOTION) @@ -122,8 +120,9 @@ ccl_device bool BVH_FUNCTION_FULL_NAME(QBVH)(KernelGlobals *kg, || UNLIKELY(ray->time > inodes.z) #endif #ifdef __VISIBILITY_FLAG__ - || (__float_as_uint(inodes.x) & visibility) == 0) + || (__float_as_uint(inodes.x) & visibility) == 0 #endif + ) { /* Pop. */ node_addr = traversal_stack[stack_ptr].addr; @@ -333,9 +332,9 @@ ccl_device bool BVH_FUNCTION_FULL_NAME(QBVH)(KernelGlobals *kg, BVH_DEBUG_NEXT_INTERSECTION(); kernel_assert(kernel_tex_fetch(__prim_type, prim_addr) == type); if(triangle_intersect(kg, - &isect_precalc, isect, P, + dir, visibility, object, prim_addr)) { @@ -447,8 +446,6 @@ ccl_device bool BVH_FUNCTION_FULL_NAME(QBVH)(KernelGlobals *kg, org4 = sse3f(ssef(P.x), ssef(P.y), ssef(P.z)); # endif - triangle_intersect_precalc(dir, &isect_precalc); - ++stack_ptr; kernel_assert(stack_ptr < BVH_QSTACK_SIZE); traversal_stack[stack_ptr].addr = ENTRYPOINT_SENTINEL; @@ -468,9 +465,9 @@ ccl_device bool BVH_FUNCTION_FULL_NAME(QBVH)(KernelGlobals *kg, /* Instance pop. */ # if BVH_FEATURE(BVH_MOTION) - bvh_instance_motion_pop(kg, object, ray, &P, &dir, &idir, &isect->t, &ob_itfm); + isect->t = bvh_instance_motion_pop(kg, object, ray, &P, &dir, &idir, isect->t, &ob_itfm); # else - bvh_instance_pop(kg, object, ray, &P, &dir, &idir, &isect->t); + isect->t = bvh_instance_pop(kg, object, ray, &P, &dir, &idir, isect->t); # endif qbvh_near_far_idx_calc(idir, @@ -489,8 +486,6 @@ ccl_device bool BVH_FUNCTION_FULL_NAME(QBVH)(KernelGlobals *kg, org4 = sse3f(ssef(P.x), ssef(P.y), ssef(P.z)); # endif - triangle_intersect_precalc(dir, &isect_precalc); - object = OBJECT_NONE; node_addr = traversal_stack[stack_ptr].addr; node_dist = traversal_stack[stack_ptr].dist; diff --git a/intern/cycles/kernel/bvh/qbvh_volume.h b/intern/cycles/kernel/bvh/qbvh_volume.h index dc6627e2dbb..192ce009524 100644 --- a/intern/cycles/kernel/bvh/qbvh_volume.h +++ b/intern/cycles/kernel/bvh/qbvh_volume.h @@ -91,9 +91,6 @@ ccl_device bool BVH_FUNCTION_FULL_NAME(QBVH)(KernelGlobals *kg, &near_x, &near_y, &near_z, &far_x, &far_y, &far_z); - IsectPrecalc isect_precalc; - triangle_intersect_precalc(dir, &isect_precalc); - /* Traversal loop. */ do { do { @@ -266,7 +263,7 @@ ccl_device bool BVH_FUNCTION_FULL_NAME(QBVH)(KernelGlobals *kg, continue; } /* Intersect ray against primitive. */ - triangle_intersect(kg, &isect_precalc, isect, P, visibility, object, prim_addr); + triangle_intersect(kg, isect, P, dir, visibility, object, prim_addr); } break; } @@ -295,9 +292,9 @@ ccl_device bool BVH_FUNCTION_FULL_NAME(QBVH)(KernelGlobals *kg, int object_flag = kernel_tex_fetch(__object_flag, object); if(object_flag & SD_OBJECT_HAS_VOLUME) { # if BVH_FEATURE(BVH_MOTION) - bvh_instance_motion_push(kg, object, ray, &P, &dir, &idir, &isect->t, &ob_itfm); + isect->t = bvh_instance_motion_push(kg, object, ray, &P, &dir, &idir, isect->t, &ob_itfm); # else - bvh_instance_push(kg, object, ray, &P, &dir, &idir, &isect->t); + isect->t = bvh_instance_push(kg, object, ray, &P, &dir, &idir, isect->t); # endif qbvh_near_far_idx_calc(idir, @@ -316,8 +313,6 @@ ccl_device bool BVH_FUNCTION_FULL_NAME(QBVH)(KernelGlobals *kg, org4 = sse3f(ssef(P.x), ssef(P.y), ssef(P.z)); # endif - triangle_intersect_precalc(dir, &isect_precalc); - ++stack_ptr; kernel_assert(stack_ptr < BVH_QSTACK_SIZE); traversal_stack[stack_ptr].addr = ENTRYPOINT_SENTINEL; @@ -341,9 +336,9 @@ ccl_device bool BVH_FUNCTION_FULL_NAME(QBVH)(KernelGlobals *kg, /* Instance pop. */ # if BVH_FEATURE(BVH_MOTION) - bvh_instance_motion_pop(kg, object, ray, &P, &dir, &idir, &isect->t, &ob_itfm); + isect->t = bvh_instance_motion_pop(kg, object, ray, &P, &dir, &idir, isect->t, &ob_itfm); # else - bvh_instance_pop(kg, object, ray, &P, &dir, &idir, &isect->t); + isect->t = bvh_instance_pop(kg, object, ray, &P, &dir, &idir, isect->t); # endif qbvh_near_far_idx_calc(idir, @@ -362,8 +357,6 @@ ccl_device bool BVH_FUNCTION_FULL_NAME(QBVH)(KernelGlobals *kg, org4 = sse3f(ssef(P.x), ssef(P.y), ssef(P.z)); # endif - triangle_intersect_precalc(dir, &isect_precalc); - object = OBJECT_NONE; node_addr = traversal_stack[stack_ptr].addr; --stack_ptr; diff --git a/intern/cycles/kernel/bvh/qbvh_volume_all.h b/intern/cycles/kernel/bvh/qbvh_volume_all.h index ff1fa92af6e..ac5f58a9a51 100644 --- a/intern/cycles/kernel/bvh/qbvh_volume_all.h +++ b/intern/cycles/kernel/bvh/qbvh_volume_all.h @@ -95,9 +95,6 @@ ccl_device uint BVH_FUNCTION_FULL_NAME(QBVH)(KernelGlobals *kg, &near_x, &near_y, &near_z, &far_x, &far_y, &far_z); - IsectPrecalc isect_precalc; - triangle_intersect_precalc(dir, &isect_precalc); - /* Traversal loop. */ do { do { @@ -271,7 +268,7 @@ ccl_device uint BVH_FUNCTION_FULL_NAME(QBVH)(KernelGlobals *kg, continue; } /* Intersect ray against primitive. */ - hit = triangle_intersect(kg, &isect_precalc, isect_array, P, visibility, object, prim_addr); + hit = triangle_intersect(kg, isect_array, P, dir, visibility, object, prim_addr); if(hit) { /* Move on to next entry in intersections array. */ isect_array++; @@ -346,9 +343,9 @@ ccl_device uint BVH_FUNCTION_FULL_NAME(QBVH)(KernelGlobals *kg, int object_flag = kernel_tex_fetch(__object_flag, object); if(object_flag & SD_OBJECT_HAS_VOLUME) { # if BVH_FEATURE(BVH_MOTION) - bvh_instance_motion_push(kg, object, ray, &P, &dir, &idir, &isect_t, &ob_itfm); + isect_t = bvh_instance_motion_push(kg, object, ray, &P, &dir, &idir, isect_t, &ob_itfm); # else - bvh_instance_push(kg, object, ray, &P, &dir, &idir, &isect_t); + isect_t = bvh_instance_push(kg, object, ray, &P, &dir, &idir, isect_t); # endif qbvh_near_far_idx_calc(idir, @@ -367,7 +364,6 @@ ccl_device uint BVH_FUNCTION_FULL_NAME(QBVH)(KernelGlobals *kg, org4 = sse3f(ssef(P.x), ssef(P.y), ssef(P.z)); # endif - triangle_intersect_precalc(dir, &isect_precalc); num_hits_in_instance = 0; isect_array->t = isect_t; @@ -406,11 +402,10 @@ ccl_device uint BVH_FUNCTION_FULL_NAME(QBVH)(KernelGlobals *kg, } } else { - float ignore_t = FLT_MAX; # if BVH_FEATURE(BVH_MOTION) - bvh_instance_motion_pop(kg, object, ray, &P, &dir, &idir, &ignore_t, &ob_itfm); + bvh_instance_motion_pop(kg, object, ray, &P, &dir, &idir, FLT_MAX, &ob_itfm); # else - bvh_instance_pop(kg, object, ray, &P, &dir, &idir, &ignore_t); + bvh_instance_pop(kg, object, ray, &P, &dir, &idir, FLT_MAX); # endif } @@ -433,8 +428,6 @@ ccl_device uint BVH_FUNCTION_FULL_NAME(QBVH)(KernelGlobals *kg, org4 = sse3f(ssef(P.x), ssef(P.y), ssef(P.z)); # endif - triangle_intersect_precalc(dir, &isect_precalc); - object = OBJECT_NONE; node_addr = traversal_stack[stack_ptr].addr; --stack_ptr; diff --git a/intern/cycles/kernel/closure/alloc.h b/intern/cycles/kernel/closure/alloc.h index b7abc1ec507..e799855a65e 100644 --- a/intern/cycles/kernel/closure/alloc.h +++ b/intern/cycles/kernel/closure/alloc.h @@ -20,17 +20,17 @@ ccl_device ShaderClosure *closure_alloc(ShaderData *sd, int size, ClosureType ty { kernel_assert(size <= sizeof(ShaderClosure)); - int num_closure = ccl_fetch(sd, num_closure); - int num_closure_extra = ccl_fetch(sd, num_closure_extra); + int num_closure = sd->num_closure; + int num_closure_extra = sd->num_closure_extra; if(num_closure + num_closure_extra >= MAX_CLOSURE) return NULL; - ShaderClosure *sc = &ccl_fetch(sd, closure)[num_closure]; + ShaderClosure *sc = &sd->closure[num_closure]; sc->type = type; sc->weight = weight; - ccl_fetch(sd, num_closure)++; + sd->num_closure++; return sc; } @@ -44,25 +44,25 @@ ccl_device ccl_addr_space void *closure_alloc_extra(ShaderData *sd, int size) * This lets us keep the same fast array iteration over closures, as we * found linked list iteration and iteration with skipping to be slower. */ int num_extra = ((size + sizeof(ShaderClosure) - 1) / sizeof(ShaderClosure)); - int num_closure = ccl_fetch(sd, num_closure); - int num_closure_extra = ccl_fetch(sd, num_closure_extra) + num_extra; + int num_closure = sd->num_closure; + int num_closure_extra = sd->num_closure_extra + num_extra; if(num_closure + num_closure_extra > MAX_CLOSURE) { /* Remove previous closure. */ - ccl_fetch(sd, num_closure)--; - ccl_fetch(sd, num_closure_extra)++; + sd->num_closure--; + sd->num_closure_extra++; return NULL; } - ccl_fetch(sd, num_closure_extra) = num_closure_extra; - return (ccl_addr_space void*)(ccl_fetch(sd, closure) + MAX_CLOSURE - num_closure_extra); + sd->num_closure_extra = num_closure_extra; + return (ccl_addr_space void*)(sd->closure + MAX_CLOSURE - num_closure_extra); } ccl_device_inline ShaderClosure *bsdf_alloc(ShaderData *sd, int size, float3 weight) { ShaderClosure *sc = closure_alloc(sd, size, CLOSURE_NONE_ID, weight); - if(!sc) + if(sc == NULL) return NULL; float sample_weight = fabsf(average(weight)); diff --git a/intern/cycles/kernel/closure/bsdf.h b/intern/cycles/kernel/closure/bsdf.h index 04f9e711c7e..0302fa9b43e 100644 --- a/intern/cycles/kernel/closure/bsdf.h +++ b/intern/cycles/kernel/closure/bsdf.h @@ -14,77 +14,77 @@ * limitations under the License. */ -#include "../closure/bsdf_ashikhmin_velvet.h" -#include "../closure/bsdf_diffuse.h" -#include "../closure/bsdf_oren_nayar.h" -#include "../closure/bsdf_phong_ramp.h" -#include "../closure/bsdf_diffuse_ramp.h" -#include "../closure/bsdf_microfacet.h" -#include "../closure/bsdf_microfacet_multi.h" -#include "../closure/bsdf_reflection.h" -#include "../closure/bsdf_refraction.h" -#include "../closure/bsdf_transparent.h" -#include "../closure/bsdf_ashikhmin_shirley.h" -#include "../closure/bsdf_toon.h" -#include "../closure/bsdf_hair.h" -#include "../closure/bsdf_principled_diffuse.h" -#include "../closure/bsdf_principled_sheen.h" +#include "kernel/closure/bsdf_ashikhmin_velvet.h" +#include "kernel/closure/bsdf_diffuse.h" +#include "kernel/closure/bsdf_oren_nayar.h" +#include "kernel/closure/bsdf_phong_ramp.h" +#include "kernel/closure/bsdf_diffuse_ramp.h" +#include "kernel/closure/bsdf_microfacet.h" +#include "kernel/closure/bsdf_microfacet_multi.h" +#include "kernel/closure/bsdf_reflection.h" +#include "kernel/closure/bsdf_refraction.h" +#include "kernel/closure/bsdf_transparent.h" +#include "kernel/closure/bsdf_ashikhmin_shirley.h" +#include "kernel/closure/bsdf_toon.h" +#include "kernel/closure/bsdf_hair.h" +#include "kernel/closure/bsdf_principled_diffuse.h" +#include "kernel/closure/bsdf_principled_sheen.h" #ifdef __SUBSURFACE__ -# include "../closure/bssrdf.h" +# include "kernel/closure/bssrdf.h" #endif #ifdef __VOLUME__ -# include "../closure/volume.h" +# include "kernel/closure/volume.h" #endif CCL_NAMESPACE_BEGIN ccl_device_forceinline int bsdf_sample(KernelGlobals *kg, - ShaderData *sd, - const ShaderClosure *sc, - float randu, - float randv, - float3 *eval, - float3 *omega_in, - differential3 *domega_in, - float *pdf) + ShaderData *sd, + const ShaderClosure *sc, + float randu, + float randv, + float3 *eval, + float3 *omega_in, + differential3 *domega_in, + float *pdf) { int label; switch(sc->type) { case CLOSURE_BSDF_DIFFUSE_ID: case CLOSURE_BSDF_BSSRDF_ID: - label = bsdf_diffuse_sample(sc, ccl_fetch(sd, Ng), ccl_fetch(sd, I), ccl_fetch(sd, dI).dx, ccl_fetch(sd, dI).dy, randu, randv, + label = bsdf_diffuse_sample(sc, sd->Ng, sd->I, sd->dI.dx, sd->dI.dy, randu, randv, eval, omega_in, &domega_in->dx, &domega_in->dy, pdf); break; #ifdef __SVM__ case CLOSURE_BSDF_OREN_NAYAR_ID: - label = bsdf_oren_nayar_sample(sc, ccl_fetch(sd, Ng), ccl_fetch(sd, I), ccl_fetch(sd, dI).dx, ccl_fetch(sd, dI).dy, randu, randv, + label = bsdf_oren_nayar_sample(sc, sd->Ng, sd->I, sd->dI.dx, sd->dI.dy, randu, randv, eval, omega_in, &domega_in->dx, &domega_in->dy, pdf); break; #ifdef __OSL__ case CLOSURE_BSDF_PHONG_RAMP_ID: - label = bsdf_phong_ramp_sample(sc, ccl_fetch(sd, Ng), ccl_fetch(sd, I), ccl_fetch(sd, dI).dx, ccl_fetch(sd, dI).dy, randu, randv, + label = bsdf_phong_ramp_sample(sc, sd->Ng, sd->I, sd->dI.dx, sd->dI.dy, randu, randv, eval, omega_in, &domega_in->dx, &domega_in->dy, pdf); break; case CLOSURE_BSDF_DIFFUSE_RAMP_ID: - label = bsdf_diffuse_ramp_sample(sc, ccl_fetch(sd, Ng), ccl_fetch(sd, I), ccl_fetch(sd, dI).dx, ccl_fetch(sd, dI).dy, randu, randv, + label = bsdf_diffuse_ramp_sample(sc, sd->Ng, sd->I, sd->dI.dx, sd->dI.dy, randu, randv, eval, omega_in, &domega_in->dx, &domega_in->dy, pdf); break; #endif case CLOSURE_BSDF_TRANSLUCENT_ID: - label = bsdf_translucent_sample(sc, ccl_fetch(sd, Ng), ccl_fetch(sd, I), ccl_fetch(sd, dI).dx, ccl_fetch(sd, dI).dy, randu, randv, + label = bsdf_translucent_sample(sc, sd->Ng, sd->I, sd->dI.dx, sd->dI.dy, randu, randv, eval, omega_in, &domega_in->dx, &domega_in->dy, pdf); break; case CLOSURE_BSDF_REFLECTION_ID: - label = bsdf_reflection_sample(sc, ccl_fetch(sd, Ng), ccl_fetch(sd, I), ccl_fetch(sd, dI).dx, ccl_fetch(sd, dI).dy, randu, randv, + label = bsdf_reflection_sample(sc, sd->Ng, sd->I, sd->dI.dx, sd->dI.dy, randu, randv, eval, omega_in, &domega_in->dx, &domega_in->dy, pdf); break; case CLOSURE_BSDF_REFRACTION_ID: - label = bsdf_refraction_sample(sc, ccl_fetch(sd, Ng), ccl_fetch(sd, I), ccl_fetch(sd, dI).dx, ccl_fetch(sd, dI).dy, randu, randv, + label = bsdf_refraction_sample(sc, sd->Ng, sd->I, sd->dI.dx, sd->dI.dy, randu, randv, eval, omega_in, &domega_in->dx, &domega_in->dy, pdf); break; case CLOSURE_BSDF_TRANSPARENT_ID: - label = bsdf_transparent_sample(sc, ccl_fetch(sd, Ng), ccl_fetch(sd, I), ccl_fetch(sd, dI).dx, ccl_fetch(sd, dI).dy, randu, randv, + label = bsdf_transparent_sample(sc, sd->Ng, sd->I, sd->dI.dx, sd->dI.dy, randu, randv, eval, omega_in, &domega_in->dx, &domega_in->dy, pdf); break; case CLOSURE_BSDF_MICROFACET_GGX_ID: @@ -93,63 +93,63 @@ ccl_device_forceinline int bsdf_sample(KernelGlobals *kg, case CLOSURE_BSDF_MICROFACET_GGX_ANISO_ID: case CLOSURE_BSDF_MICROFACET_GGX_ANISO_FRESNEL_ID: case CLOSURE_BSDF_MICROFACET_GGX_REFRACTION_ID: - label = bsdf_microfacet_ggx_sample(kg, sc, ccl_fetch(sd, Ng), ccl_fetch(sd, I), ccl_fetch(sd, dI).dx, ccl_fetch(sd, dI).dy, randu, randv, + label = bsdf_microfacet_ggx_sample(kg, sc, sd->Ng, sd->I, sd->dI.dx, sd->dI.dy, randu, randv, eval, omega_in, &domega_in->dx, &domega_in->dy, pdf); break; case CLOSURE_BSDF_MICROFACET_MULTI_GGX_ID: case CLOSURE_BSDF_MICROFACET_MULTI_GGX_FRESNEL_ID: - label = bsdf_microfacet_multi_ggx_sample(kg, sc, ccl_fetch(sd, Ng), ccl_fetch(sd, I), ccl_fetch(sd, dI).dx, ccl_fetch(sd, dI).dy, randu, randv, - eval, omega_in, &domega_in->dx, &domega_in->dy, pdf, &ccl_fetch(sd, lcg_state)); + label = bsdf_microfacet_multi_ggx_sample(kg, sc, sd->Ng, sd->I, sd->dI.dx, sd->dI.dy, randu, randv, + eval, omega_in, &domega_in->dx, &domega_in->dy, pdf, &sd->lcg_state); break; case CLOSURE_BSDF_MICROFACET_MULTI_GGX_GLASS_ID: case CLOSURE_BSDF_MICROFACET_MULTI_GGX_GLASS_FRESNEL_ID: - label = bsdf_microfacet_multi_ggx_glass_sample(kg, sc, ccl_fetch(sd, Ng), ccl_fetch(sd, I), ccl_fetch(sd, dI).dx, ccl_fetch(sd, dI).dy, randu, randv, - eval, omega_in, &domega_in->dx, &domega_in->dy, pdf, &ccl_fetch(sd, lcg_state)); + label = bsdf_microfacet_multi_ggx_glass_sample(kg, sc, sd->Ng, sd->I, sd->dI.dx, sd->dI.dy, randu, randv, + eval, omega_in, &domega_in->dx, &domega_in->dy, pdf, &sd->lcg_state); break; case CLOSURE_BSDF_MICROFACET_BECKMANN_ID: case CLOSURE_BSDF_MICROFACET_BECKMANN_ANISO_ID: case CLOSURE_BSDF_MICROFACET_BECKMANN_REFRACTION_ID: - label = bsdf_microfacet_beckmann_sample(kg, sc, ccl_fetch(sd, Ng), ccl_fetch(sd, I), ccl_fetch(sd, dI).dx, ccl_fetch(sd, dI).dy, randu, randv, + label = bsdf_microfacet_beckmann_sample(kg, sc, sd->Ng, sd->I, sd->dI.dx, sd->dI.dy, randu, randv, eval, omega_in, &domega_in->dx, &domega_in->dy, pdf); break; case CLOSURE_BSDF_ASHIKHMIN_SHIRLEY_ID: case CLOSURE_BSDF_ASHIKHMIN_SHIRLEY_ANISO_ID: - label = bsdf_ashikhmin_shirley_sample(sc, ccl_fetch(sd, Ng), ccl_fetch(sd, I), ccl_fetch(sd, dI).dx, ccl_fetch(sd, dI).dy, randu, randv, + label = bsdf_ashikhmin_shirley_sample(sc, sd->Ng, sd->I, sd->dI.dx, sd->dI.dy, randu, randv, eval, omega_in, &domega_in->dx, &domega_in->dy, pdf); break; case CLOSURE_BSDF_ASHIKHMIN_VELVET_ID: - label = bsdf_ashikhmin_velvet_sample(sc, ccl_fetch(sd, Ng), ccl_fetch(sd, I), ccl_fetch(sd, dI).dx, ccl_fetch(sd, dI).dy, randu, randv, + label = bsdf_ashikhmin_velvet_sample(sc, sd->Ng, sd->I, sd->dI.dx, sd->dI.dy, randu, randv, eval, omega_in, &domega_in->dx, &domega_in->dy, pdf); break; case CLOSURE_BSDF_DIFFUSE_TOON_ID: - label = bsdf_diffuse_toon_sample(sc, ccl_fetch(sd, Ng), ccl_fetch(sd, I), ccl_fetch(sd, dI).dx, ccl_fetch(sd, dI).dy, randu, randv, + label = bsdf_diffuse_toon_sample(sc, sd->Ng, sd->I, sd->dI.dx, sd->dI.dy, randu, randv, eval, omega_in, &domega_in->dx, &domega_in->dy, pdf); break; case CLOSURE_BSDF_GLOSSY_TOON_ID: - label = bsdf_glossy_toon_sample(sc, ccl_fetch(sd, Ng), ccl_fetch(sd, I), ccl_fetch(sd, dI).dx, ccl_fetch(sd, dI).dy, randu, randv, + label = bsdf_glossy_toon_sample(sc, sd->Ng, sd->I, sd->dI.dx, sd->dI.dy, randu, randv, eval, omega_in, &domega_in->dx, &domega_in->dy, pdf); break; case CLOSURE_BSDF_HAIR_REFLECTION_ID: - label = bsdf_hair_reflection_sample(sc, ccl_fetch(sd, Ng), ccl_fetch(sd, I), ccl_fetch(sd, dI).dx, ccl_fetch(sd, dI).dy, randu, randv, + label = bsdf_hair_reflection_sample(sc, sd->Ng, sd->I, sd->dI.dx, sd->dI.dy, randu, randv, eval, omega_in, &domega_in->dx, &domega_in->dy, pdf); break; case CLOSURE_BSDF_HAIR_TRANSMISSION_ID: - label = bsdf_hair_transmission_sample(sc, ccl_fetch(sd, Ng), ccl_fetch(sd, I), ccl_fetch(sd, dI).dx, ccl_fetch(sd, dI).dy, randu, randv, + label = bsdf_hair_transmission_sample(sc, sd->Ng, sd->I, sd->dI.dx, sd->dI.dy, randu, randv, eval, omega_in, &domega_in->dx, &domega_in->dy, pdf); break; case CLOSURE_BSDF_PRINCIPLED_DIFFUSE_ID: case CLOSURE_BSDF_BSSRDF_PRINCIPLED_ID: - label = bsdf_principled_diffuse_sample(sc, ccl_fetch(sd, Ng), ccl_fetch(sd, I), ccl_fetch(sd, dI).dx, ccl_fetch(sd, dI).dy, randu, randv, + label = bsdf_principled_diffuse_sample(sc, sd->Ng, sd->I, sd->dI.dx, sd->dI.dy, randu, randv, eval, omega_in, &domega_in->dx, &domega_in->dy, pdf); break; case CLOSURE_BSDF_PRINCIPLED_SHEEN_ID: - label = bsdf_principled_sheen_sample(sc, ccl_fetch(sd, Ng), ccl_fetch(sd, I), ccl_fetch(sd, dI).dx, ccl_fetch(sd, dI).dy, randu, randv, + label = bsdf_principled_sheen_sample(sc, sd->Ng, sd->I, sd->dI.dx, sd->dI.dy, randu, randv, eval, omega_in, &domega_in->dx, &domega_in->dy, pdf); break; #endif #ifdef __VOLUME__ case CLOSURE_VOLUME_HENYEY_GREENSTEIN_ID: - label = volume_henyey_greenstein_sample(sc, ccl_fetch(sd, I), ccl_fetch(sd, dI).dx, ccl_fetch(sd, dI).dy, randu, randv, eval, omega_in, &domega_in->dx, &domega_in->dy, pdf); + label = volume_henyey_greenstein_sample(sc, sd->I, sd->dI.dx, sd->dI.dy, randu, randv, eval, omega_in, &domega_in->dx, &domega_in->dy, pdf); break; #endif default: @@ -173,35 +173,35 @@ float3 bsdf_eval(KernelGlobals *kg, { float3 eval; - if(dot(ccl_fetch(sd, Ng), omega_in) >= 0.0f) { + if(dot(sd->Ng, omega_in) >= 0.0f) { switch(sc->type) { case CLOSURE_BSDF_DIFFUSE_ID: case CLOSURE_BSDF_BSSRDF_ID: - eval = bsdf_diffuse_eval_reflect(sc, ccl_fetch(sd, I), omega_in, pdf); + eval = bsdf_diffuse_eval_reflect(sc, sd->I, omega_in, pdf); break; #ifdef __SVM__ case CLOSURE_BSDF_OREN_NAYAR_ID: - eval = bsdf_oren_nayar_eval_reflect(sc, ccl_fetch(sd, I), omega_in, pdf); + eval = bsdf_oren_nayar_eval_reflect(sc, sd->I, omega_in, pdf); break; #ifdef __OSL__ case CLOSURE_BSDF_PHONG_RAMP_ID: - eval = bsdf_phong_ramp_eval_reflect(sc, ccl_fetch(sd, I), omega_in, pdf); + eval = bsdf_phong_ramp_eval_reflect(sc, sd->I, omega_in, pdf); break; case CLOSURE_BSDF_DIFFUSE_RAMP_ID: - eval = bsdf_diffuse_ramp_eval_reflect(sc, ccl_fetch(sd, I), omega_in, pdf); + eval = bsdf_diffuse_ramp_eval_reflect(sc, sd->I, omega_in, pdf); break; #endif case CLOSURE_BSDF_TRANSLUCENT_ID: - eval = bsdf_translucent_eval_reflect(sc, ccl_fetch(sd, I), omega_in, pdf); + eval = bsdf_translucent_eval_reflect(sc, sd->I, omega_in, pdf); break; case CLOSURE_BSDF_REFLECTION_ID: - eval = bsdf_reflection_eval_reflect(sc, ccl_fetch(sd, I), omega_in, pdf); + eval = bsdf_reflection_eval_reflect(sc, sd->I, omega_in, pdf); break; case CLOSURE_BSDF_REFRACTION_ID: - eval = bsdf_refraction_eval_reflect(sc, ccl_fetch(sd, I), omega_in, pdf); + eval = bsdf_refraction_eval_reflect(sc, sd->I, omega_in, pdf); break; case CLOSURE_BSDF_TRANSPARENT_ID: - eval = bsdf_transparent_eval_reflect(sc, ccl_fetch(sd, I), omega_in, pdf); + eval = bsdf_transparent_eval_reflect(sc, sd->I, omega_in, pdf); break; case CLOSURE_BSDF_MICROFACET_GGX_ID: case CLOSURE_BSDF_MICROFACET_GGX_FRESNEL_ID: @@ -209,51 +209,51 @@ float3 bsdf_eval(KernelGlobals *kg, case CLOSURE_BSDF_MICROFACET_GGX_ANISO_ID: case CLOSURE_BSDF_MICROFACET_GGX_ANISO_FRESNEL_ID: case CLOSURE_BSDF_MICROFACET_GGX_REFRACTION_ID: - eval = bsdf_microfacet_ggx_eval_reflect(sc, ccl_fetch(sd, I), omega_in, pdf); + eval = bsdf_microfacet_ggx_eval_reflect(sc, sd->I, omega_in, pdf); break; case CLOSURE_BSDF_MICROFACET_MULTI_GGX_ID: case CLOSURE_BSDF_MICROFACET_MULTI_GGX_FRESNEL_ID: - eval = bsdf_microfacet_multi_ggx_eval_reflect(sc, ccl_fetch(sd, I), omega_in, pdf, &ccl_fetch(sd, lcg_state)); + eval = bsdf_microfacet_multi_ggx_eval_reflect(sc, sd->I, omega_in, pdf, &sd->lcg_state); break; case CLOSURE_BSDF_MICROFACET_MULTI_GGX_GLASS_ID: case CLOSURE_BSDF_MICROFACET_MULTI_GGX_GLASS_FRESNEL_ID: - eval = bsdf_microfacet_multi_ggx_glass_eval_reflect(sc, ccl_fetch(sd, I), omega_in, pdf, &ccl_fetch(sd, lcg_state)); + eval = bsdf_microfacet_multi_ggx_glass_eval_reflect(sc, sd->I, omega_in, pdf, &sd->lcg_state); break; case CLOSURE_BSDF_MICROFACET_BECKMANN_ID: case CLOSURE_BSDF_MICROFACET_BECKMANN_ANISO_ID: case CLOSURE_BSDF_MICROFACET_BECKMANN_REFRACTION_ID: - eval = bsdf_microfacet_beckmann_eval_reflect(sc, ccl_fetch(sd, I), omega_in, pdf); + eval = bsdf_microfacet_beckmann_eval_reflect(sc, sd->I, omega_in, pdf); break; case CLOSURE_BSDF_ASHIKHMIN_SHIRLEY_ID: case CLOSURE_BSDF_ASHIKHMIN_SHIRLEY_ANISO_ID: - eval = bsdf_ashikhmin_shirley_eval_reflect(sc, ccl_fetch(sd, I), omega_in, pdf); + eval = bsdf_ashikhmin_shirley_eval_reflect(sc, sd->I, omega_in, pdf); break; case CLOSURE_BSDF_ASHIKHMIN_VELVET_ID: - eval = bsdf_ashikhmin_velvet_eval_reflect(sc, ccl_fetch(sd, I), omega_in, pdf); + eval = bsdf_ashikhmin_velvet_eval_reflect(sc, sd->I, omega_in, pdf); break; case CLOSURE_BSDF_DIFFUSE_TOON_ID: - eval = bsdf_diffuse_toon_eval_reflect(sc, ccl_fetch(sd, I), omega_in, pdf); + eval = bsdf_diffuse_toon_eval_reflect(sc, sd->I, omega_in, pdf); break; case CLOSURE_BSDF_GLOSSY_TOON_ID: - eval = bsdf_glossy_toon_eval_reflect(sc, ccl_fetch(sd, I), omega_in, pdf); + eval = bsdf_glossy_toon_eval_reflect(sc, sd->I, omega_in, pdf); break; case CLOSURE_BSDF_HAIR_REFLECTION_ID: - eval = bsdf_hair_reflection_eval_reflect(sc, ccl_fetch(sd, I), omega_in, pdf); + eval = bsdf_hair_reflection_eval_reflect(sc, sd->I, omega_in, pdf); break; case CLOSURE_BSDF_HAIR_TRANSMISSION_ID: - eval = bsdf_hair_transmission_eval_reflect(sc, ccl_fetch(sd, I), omega_in, pdf); + eval = bsdf_hair_transmission_eval_reflect(sc, sd->I, omega_in, pdf); break; case CLOSURE_BSDF_PRINCIPLED_DIFFUSE_ID: case CLOSURE_BSDF_BSSRDF_PRINCIPLED_ID: - eval = bsdf_principled_diffuse_eval_reflect(sc, ccl_fetch(sd, I), omega_in, pdf); + eval = bsdf_principled_diffuse_eval_reflect(sc, sd->I, omega_in, pdf); break; case CLOSURE_BSDF_PRINCIPLED_SHEEN_ID: - eval = bsdf_principled_sheen_eval_reflect(sc, ccl_fetch(sd, I), omega_in, pdf); + eval = bsdf_principled_sheen_eval_reflect(sc, sd->I, omega_in, pdf); break; #endif #ifdef __VOLUME__ case CLOSURE_VOLUME_HENYEY_GREENSTEIN_ID: - eval = volume_henyey_greenstein_eval_phase(sc, ccl_fetch(sd, I), omega_in, pdf); + eval = volume_henyey_greenstein_eval_phase(sc, sd->I, omega_in, pdf); break; #endif default: @@ -265,23 +265,23 @@ float3 bsdf_eval(KernelGlobals *kg, switch(sc->type) { case CLOSURE_BSDF_DIFFUSE_ID: case CLOSURE_BSDF_BSSRDF_ID: - eval = bsdf_diffuse_eval_transmit(sc, ccl_fetch(sd, I), omega_in, pdf); + eval = bsdf_diffuse_eval_transmit(sc, sd->I, omega_in, pdf); break; #ifdef __SVM__ case CLOSURE_BSDF_OREN_NAYAR_ID: - eval = bsdf_oren_nayar_eval_transmit(sc, ccl_fetch(sd, I), omega_in, pdf); + eval = bsdf_oren_nayar_eval_transmit(sc, sd->I, omega_in, pdf); break; case CLOSURE_BSDF_TRANSLUCENT_ID: - eval = bsdf_translucent_eval_transmit(sc, ccl_fetch(sd, I), omega_in, pdf); + eval = bsdf_translucent_eval_transmit(sc, sd->I, omega_in, pdf); break; case CLOSURE_BSDF_REFLECTION_ID: - eval = bsdf_reflection_eval_transmit(sc, ccl_fetch(sd, I), omega_in, pdf); + eval = bsdf_reflection_eval_transmit(sc, sd->I, omega_in, pdf); break; case CLOSURE_BSDF_REFRACTION_ID: - eval = bsdf_refraction_eval_transmit(sc, ccl_fetch(sd, I), omega_in, pdf); + eval = bsdf_refraction_eval_transmit(sc, sd->I, omega_in, pdf); break; case CLOSURE_BSDF_TRANSPARENT_ID: - eval = bsdf_transparent_eval_transmit(sc, ccl_fetch(sd, I), omega_in, pdf); + eval = bsdf_transparent_eval_transmit(sc, sd->I, omega_in, pdf); break; case CLOSURE_BSDF_MICROFACET_GGX_ID: case CLOSURE_BSDF_MICROFACET_GGX_FRESNEL_ID: @@ -289,51 +289,51 @@ float3 bsdf_eval(KernelGlobals *kg, case CLOSURE_BSDF_MICROFACET_GGX_ANISO_ID: case CLOSURE_BSDF_MICROFACET_GGX_ANISO_FRESNEL_ID: case CLOSURE_BSDF_MICROFACET_GGX_REFRACTION_ID: - eval = bsdf_microfacet_ggx_eval_transmit(sc, ccl_fetch(sd, I), omega_in, pdf); + eval = bsdf_microfacet_ggx_eval_transmit(sc, sd->I, omega_in, pdf); break; case CLOSURE_BSDF_MICROFACET_MULTI_GGX_ID: case CLOSURE_BSDF_MICROFACET_MULTI_GGX_FRESNEL_ID: - eval = bsdf_microfacet_multi_ggx_eval_transmit(sc, ccl_fetch(sd, I), omega_in, pdf, &ccl_fetch(sd, lcg_state)); + eval = bsdf_microfacet_multi_ggx_eval_transmit(sc, sd->I, omega_in, pdf, &sd->lcg_state); break; case CLOSURE_BSDF_MICROFACET_MULTI_GGX_GLASS_ID: case CLOSURE_BSDF_MICROFACET_MULTI_GGX_GLASS_FRESNEL_ID: - eval = bsdf_microfacet_multi_ggx_glass_eval_transmit(sc, ccl_fetch(sd, I), omega_in, pdf, &ccl_fetch(sd, lcg_state)); + eval = bsdf_microfacet_multi_ggx_glass_eval_transmit(sc, sd->I, omega_in, pdf, &sd->lcg_state); break; case CLOSURE_BSDF_MICROFACET_BECKMANN_ID: case CLOSURE_BSDF_MICROFACET_BECKMANN_ANISO_ID: case CLOSURE_BSDF_MICROFACET_BECKMANN_REFRACTION_ID: - eval = bsdf_microfacet_beckmann_eval_transmit(sc, ccl_fetch(sd, I), omega_in, pdf); + eval = bsdf_microfacet_beckmann_eval_transmit(sc, sd->I, omega_in, pdf); break; case CLOSURE_BSDF_ASHIKHMIN_SHIRLEY_ID: case CLOSURE_BSDF_ASHIKHMIN_SHIRLEY_ANISO_ID: - eval = bsdf_ashikhmin_shirley_eval_transmit(sc, ccl_fetch(sd, I), omega_in, pdf); + eval = bsdf_ashikhmin_shirley_eval_transmit(sc, sd->I, omega_in, pdf); break; case CLOSURE_BSDF_ASHIKHMIN_VELVET_ID: - eval = bsdf_ashikhmin_velvet_eval_transmit(sc, ccl_fetch(sd, I), omega_in, pdf); + eval = bsdf_ashikhmin_velvet_eval_transmit(sc, sd->I, omega_in, pdf); break; case CLOSURE_BSDF_DIFFUSE_TOON_ID: - eval = bsdf_diffuse_toon_eval_transmit(sc, ccl_fetch(sd, I), omega_in, pdf); + eval = bsdf_diffuse_toon_eval_transmit(sc, sd->I, omega_in, pdf); break; case CLOSURE_BSDF_GLOSSY_TOON_ID: - eval = bsdf_glossy_toon_eval_transmit(sc, ccl_fetch(sd, I), omega_in, pdf); + eval = bsdf_glossy_toon_eval_transmit(sc, sd->I, omega_in, pdf); break; case CLOSURE_BSDF_HAIR_REFLECTION_ID: - eval = bsdf_hair_reflection_eval_transmit(sc, ccl_fetch(sd, I), omega_in, pdf); + eval = bsdf_hair_reflection_eval_transmit(sc, sd->I, omega_in, pdf); break; case CLOSURE_BSDF_HAIR_TRANSMISSION_ID: - eval = bsdf_hair_transmission_eval_transmit(sc, ccl_fetch(sd, I), omega_in, pdf); + eval = bsdf_hair_transmission_eval_transmit(sc, sd->I, omega_in, pdf); break; case CLOSURE_BSDF_PRINCIPLED_DIFFUSE_ID: case CLOSURE_BSDF_BSSRDF_PRINCIPLED_ID: - eval = bsdf_principled_diffuse_eval_transmit(sc, ccl_fetch(sd, I), omega_in, pdf); + eval = bsdf_principled_diffuse_eval_transmit(sc, sd->I, omega_in, pdf); break; case CLOSURE_BSDF_PRINCIPLED_SHEEN_ID: - eval = bsdf_principled_sheen_eval_transmit(sc, ccl_fetch(sd, I), omega_in, pdf); + eval = bsdf_principled_sheen_eval_transmit(sc, sd->I, omega_in, pdf); break; #endif #ifdef __VOLUME__ case CLOSURE_VOLUME_HENYEY_GREENSTEIN_ID: - eval = volume_henyey_greenstein_eval_phase(sc, ccl_fetch(sd, I), omega_in, pdf); + eval = volume_henyey_greenstein_eval_phase(sc, sd->I, omega_in, pdf); break; #endif default: diff --git a/intern/cycles/kernel/closure/bsdf_microfacet.h b/intern/cycles/kernel/closure/bsdf_microfacet.h index b74e8ab97cf..58f6140970d 100644 --- a/intern/cycles/kernel/closure/bsdf_microfacet.h +++ b/intern/cycles/kernel/closure/bsdf_microfacet.h @@ -327,7 +327,7 @@ ccl_device bool bsdf_microfacet_merge(const ShaderClosure *a, const ShaderClosur (bsdf_a->alpha_y == bsdf_b->alpha_y) && (isequal_float3(bsdf_a->T, bsdf_b->T)) && (bsdf_a->ior == bsdf_b->ior) && - ((!bsdf_a->extra && !bsdf_b->extra) || + ((bsdf_a->extra == NULL && bsdf_b->extra == NULL) || ((bsdf_a->extra && bsdf_b->extra) && (isequal_float3(bsdf_a->extra->color, bsdf_b->extra->color)))); } diff --git a/intern/cycles/kernel/closure/bsdf_microfacet_multi.h b/intern/cycles/kernel/closure/bsdf_microfacet_multi.h index a49b0717a3d..57f1e733ee7 100644 --- a/intern/cycles/kernel/closure/bsdf_microfacet_multi.h +++ b/intern/cycles/kernel/closure/bsdf_microfacet_multi.h @@ -43,7 +43,7 @@ ccl_device_forceinline float D_ggx_aniso(const float3 wm, const float2 alpha) ccl_device_forceinline float2 mf_sampleP22_11(const float cosI, const float2 randU) { if(cosI > 0.9999f || cosI < 1e-6f) { - const float r = sqrtf(randU.x / (1.0f - randU.x)); + const float r = sqrtf(randU.x / max(1.0f - randU.x, 1e-7f)); const float phi = M_2PI_F * randU.y; return make_float2(r*cosf(phi), r*sinf(phi)); } @@ -83,7 +83,7 @@ ccl_device_forceinline float3 mf_sample_vndf(const float3 wi, const float2 alpha const float3 wi_11 = normalize(make_float3(alpha.x*wi.x, alpha.y*wi.y, wi.z)); const float2 slope_11 = mf_sampleP22_11(wi_11.z, randU); - const float2 cossin_phi = normalize(make_float2(wi_11.x, wi_11.y)); + const float3 cossin_phi = safe_normalize(make_float3(wi_11.x, wi_11.y, 0.0f)); const float slope_x = alpha.x*(cossin_phi.x * slope_11.x - cossin_phi.y * slope_11.y); const float slope_y = alpha.y*(cossin_phi.y * slope_11.x + cossin_phi.x * slope_11.y); @@ -313,18 +313,18 @@ ccl_device_forceinline float mf_glass_pdf(const float3 wi, const float3 wo, cons #define MF_PHASE_FUNCTION glass #define MF_MULTI_GLASS -#include "bsdf_microfacet_multi_impl.h" +#include "kernel/closure/bsdf_microfacet_multi_impl.h" /* The diffuse phase function is not implemented as a node yet. */ #if 0 #define MF_PHASE_FUNCTION diffuse #define MF_MULTI_DIFFUSE -#include "bsdf_microfacet_multi_impl.h" +#include "kernel/closure/bsdf_microfacet_multi_impl.h" #endif #define MF_PHASE_FUNCTION glossy #define MF_MULTI_GLOSSY -#include "bsdf_microfacet_multi_impl.h" +#include "kernel/closure/bsdf_microfacet_multi_impl.h" ccl_device void bsdf_microfacet_multi_ggx_blur(ShaderClosure *sc, float roughness) { diff --git a/intern/cycles/kernel/geom/geom.h b/intern/cycles/kernel/geom/geom.h index 6838e26c242..c623e3490fd 100644 --- a/intern/cycles/kernel/geom/geom.h +++ b/intern/cycles/kernel/geom/geom.h @@ -14,19 +14,19 @@ * limitations under the License. */ -#include "geom_attribute.h" -#include "geom_object.h" +#include "kernel/geom/geom_attribute.h" +#include "kernel/geom/geom_object.h" #ifdef __PATCH_EVAL__ -# include "geom_patch.h" +# include "kernel/geom/geom_patch.h" #endif -#include "geom_triangle.h" -#include "geom_subd_triangle.h" -#include "geom_triangle_intersect.h" -#include "geom_motion_triangle.h" -#include "geom_motion_triangle_intersect.h" -#include "geom_motion_triangle_shader.h" -#include "geom_motion_curve.h" -#include "geom_curve.h" -#include "geom_volume.h" -#include "geom_primitive.h" +#include "kernel/geom/geom_triangle.h" +#include "kernel/geom/geom_subd_triangle.h" +#include "kernel/geom/geom_triangle_intersect.h" +#include "kernel/geom/geom_motion_triangle.h" +#include "kernel/geom/geom_motion_triangle_intersect.h" +#include "kernel/geom/geom_motion_triangle_shader.h" +#include "kernel/geom/geom_motion_curve.h" +#include "kernel/geom/geom_curve.h" +#include "kernel/geom/geom_volume.h" +#include "kernel/geom/geom_primitive.h" diff --git a/intern/cycles/kernel/geom/geom_attribute.h b/intern/cycles/kernel/geom/geom_attribute.h index 08ccee56335..cc62192ef21 100644 --- a/intern/cycles/kernel/geom/geom_attribute.h +++ b/intern/cycles/kernel/geom/geom_attribute.h @@ -30,7 +30,7 @@ ccl_device_inline uint subd_triangle_patch(KernelGlobals *kg, const ShaderData * ccl_device_inline uint attribute_primitive_type(KernelGlobals *kg, const ShaderData *sd) { #ifdef __HAIR__ - if(ccl_fetch(sd, type) & PRIMITIVE_ALL_CURVE) { + if(sd->type & PRIMITIVE_ALL_CURVE) { return ATTR_PRIM_CURVE; } else @@ -53,12 +53,12 @@ ccl_device_inline AttributeDescriptor attribute_not_found() ccl_device_inline AttributeDescriptor find_attribute(KernelGlobals *kg, const ShaderData *sd, uint id) { - if(ccl_fetch(sd, object) == PRIM_NONE) { + if(sd->object == PRIM_NONE) { return attribute_not_found(); } /* for SVM, find attribute by unique id */ - uint attr_offset = ccl_fetch(sd, object)*kernel_data.bvh.attributes_map_stride; + uint attr_offset = sd->object*kernel_data.bvh.attributes_map_stride; attr_offset += attribute_primitive_type(kg, sd); uint4 attr_map = kernel_tex_fetch(__attributes_map, attr_offset); @@ -73,7 +73,7 @@ ccl_device_inline AttributeDescriptor find_attribute(KernelGlobals *kg, const Sh AttributeDescriptor desc; desc.element = (AttributeElement)attr_map.y; - if(ccl_fetch(sd, prim) == PRIM_NONE && + if(sd->prim == PRIM_NONE && desc.element != ATTR_ELEMENT_MESH && desc.element != ATTR_ELEMENT_VOXEL && desc.element != ATTR_ELEMENT_OBJECT) diff --git a/intern/cycles/kernel/geom/geom_curve.h b/intern/cycles/kernel/geom/geom_curve.h index 9de335403ce..bb33b91847e 100644 --- a/intern/cycles/kernel/geom/geom_curve.h +++ b/intern/cycles/kernel/geom/geom_curve.h @@ -22,6 +22,12 @@ CCL_NAMESPACE_BEGIN #ifdef __HAIR__ +#if defined(__KERNEL_CUDA__) && (__CUDA_ARCH__ < 300) +# define ccl_device_curveintersect ccl_device +#else +# define ccl_device_curveintersect ccl_device_forceinline +#endif + /* Reading attributes on various curve elements */ ccl_device float curve_attribute_float(KernelGlobals *kg, const ShaderData *sd, const AttributeDescriptor desc, float *dx, float *dy) @@ -32,22 +38,22 @@ ccl_device float curve_attribute_float(KernelGlobals *kg, const ShaderData *sd, if(dy) *dy = 0.0f; #endif - return kernel_tex_fetch(__attributes_float, desc.offset + ccl_fetch(sd, prim)); + return kernel_tex_fetch(__attributes_float, desc.offset + sd->prim); } else if(desc.element == ATTR_ELEMENT_CURVE_KEY || desc.element == ATTR_ELEMENT_CURVE_KEY_MOTION) { - float4 curvedata = kernel_tex_fetch(__curves, ccl_fetch(sd, prim)); - int k0 = __float_as_int(curvedata.x) + PRIMITIVE_UNPACK_SEGMENT(ccl_fetch(sd, type)); + float4 curvedata = kernel_tex_fetch(__curves, sd->prim); + int k0 = __float_as_int(curvedata.x) + PRIMITIVE_UNPACK_SEGMENT(sd->type); int k1 = k0 + 1; float f0 = kernel_tex_fetch(__attributes_float, desc.offset + k0); float f1 = kernel_tex_fetch(__attributes_float, desc.offset + k1); #ifdef __RAY_DIFFERENTIALS__ - if(dx) *dx = ccl_fetch(sd, du).dx*(f1 - f0); + if(dx) *dx = sd->du.dx*(f1 - f0); if(dy) *dy = 0.0f; #endif - return (1.0f - ccl_fetch(sd, u))*f0 + ccl_fetch(sd, u)*f1; + return (1.0f - sd->u)*f0 + sd->u*f1; } else { #ifdef __RAY_DIFFERENTIALS__ @@ -71,22 +77,22 @@ ccl_device float3 curve_attribute_float3(KernelGlobals *kg, const ShaderData *sd if(dy) *dy = make_float3(0.0f, 0.0f, 0.0f); #endif - return float4_to_float3(kernel_tex_fetch(__attributes_float3, desc.offset + ccl_fetch(sd, prim))); + return float4_to_float3(kernel_tex_fetch(__attributes_float3, desc.offset + sd->prim)); } else if(desc.element == ATTR_ELEMENT_CURVE_KEY || desc.element == ATTR_ELEMENT_CURVE_KEY_MOTION) { - float4 curvedata = kernel_tex_fetch(__curves, ccl_fetch(sd, prim)); - int k0 = __float_as_int(curvedata.x) + PRIMITIVE_UNPACK_SEGMENT(ccl_fetch(sd, type)); + float4 curvedata = kernel_tex_fetch(__curves, sd->prim); + int k0 = __float_as_int(curvedata.x) + PRIMITIVE_UNPACK_SEGMENT(sd->type); int k1 = k0 + 1; float3 f0 = float4_to_float3(kernel_tex_fetch(__attributes_float3, desc.offset + k0)); float3 f1 = float4_to_float3(kernel_tex_fetch(__attributes_float3, desc.offset + k1)); #ifdef __RAY_DIFFERENTIALS__ - if(dx) *dx = ccl_fetch(sd, du).dx*(f1 - f0); + if(dx) *dx = sd->du.dx*(f1 - f0); if(dy) *dy = make_float3(0.0f, 0.0f, 0.0f); #endif - return (1.0f - ccl_fetch(sd, u))*f0 + ccl_fetch(sd, u)*f1; + return (1.0f - sd->u)*f0 + sd->u*f1; } else { #ifdef __RAY_DIFFERENTIALS__ @@ -104,22 +110,22 @@ ccl_device float curve_thickness(KernelGlobals *kg, ShaderData *sd) { float r = 0.0f; - if(ccl_fetch(sd, type) & PRIMITIVE_ALL_CURVE) { - float4 curvedata = kernel_tex_fetch(__curves, ccl_fetch(sd, prim)); - int k0 = __float_as_int(curvedata.x) + PRIMITIVE_UNPACK_SEGMENT(ccl_fetch(sd, type)); + if(sd->type & PRIMITIVE_ALL_CURVE) { + float4 curvedata = kernel_tex_fetch(__curves, sd->prim); + int k0 = __float_as_int(curvedata.x) + PRIMITIVE_UNPACK_SEGMENT(sd->type); int k1 = k0 + 1; float4 P_curve[2]; - if(ccl_fetch(sd, type) & PRIMITIVE_CURVE) { + if(sd->type & PRIMITIVE_CURVE) { P_curve[0]= kernel_tex_fetch(__curve_keys, k0); P_curve[1]= kernel_tex_fetch(__curve_keys, k1); } else { - motion_curve_keys(kg, ccl_fetch(sd, object), ccl_fetch(sd, prim), ccl_fetch(sd, time), k0, k1, P_curve); + motion_curve_keys(kg, sd->object, sd->prim, sd->time, k0, k1, P_curve); } - r = (P_curve[1].w - P_curve[0].w) * ccl_fetch(sd, u) + P_curve[0].w; + r = (P_curve[1].w - P_curve[0].w) * sd->u + P_curve[0].w; } return r*2.0f; @@ -130,8 +136,8 @@ ccl_device float curve_thickness(KernelGlobals *kg, ShaderData *sd) ccl_device float3 curve_motion_center_location(KernelGlobals *kg, ShaderData *sd) { - float4 curvedata = kernel_tex_fetch(__curves, ccl_fetch(sd, prim)); - int k0 = __float_as_int(curvedata.x) + PRIMITIVE_UNPACK_SEGMENT(ccl_fetch(sd, type)); + float4 curvedata = kernel_tex_fetch(__curves, sd->prim); + int k0 = __float_as_int(curvedata.x) + PRIMITIVE_UNPACK_SEGMENT(sd->type); int k1 = k0 + 1; float4 P_curve[2]; @@ -139,7 +145,7 @@ ccl_device float3 curve_motion_center_location(KernelGlobals *kg, ShaderData *sd P_curve[0]= kernel_tex_fetch(__curve_keys, k0); P_curve[1]= kernel_tex_fetch(__curve_keys, k1); - return float4_to_float3(P_curve[1]) * ccl_fetch(sd, u) + float4_to_float3(P_curve[0]) * (1.0f - ccl_fetch(sd, u)); + return float4_to_float3(P_curve[1]) * sd->u + float4_to_float3(P_curve[0]) * (1.0f - sd->u); } /* Curve tangent normal */ @@ -148,14 +154,14 @@ ccl_device float3 curve_tangent_normal(KernelGlobals *kg, ShaderData *sd) { float3 tgN = make_float3(0.0f,0.0f,0.0f); - if(ccl_fetch(sd, type) & PRIMITIVE_ALL_CURVE) { + if(sd->type & PRIMITIVE_ALL_CURVE) { - tgN = -(-ccl_fetch(sd, I) - ccl_fetch(sd, dPdu) * (dot(ccl_fetch(sd, dPdu),-ccl_fetch(sd, I)) / len_squared(ccl_fetch(sd, dPdu)))); + tgN = -(-sd->I - sd->dPdu * (dot(sd->dPdu,-sd->I) / len_squared(sd->dPdu))); tgN = normalize(tgN); /* need to find suitable scaled gd for corrected normal */ #if 0 - tgN = normalize(tgN - gd * ccl_fetch(sd, dPdu)); + tgN = normalize(tgN - gd * sd->dPdu); #endif } @@ -222,13 +228,22 @@ ccl_device_inline ssef transform_point_T3(const ssef t[3], const ssef &a) #ifdef __KERNEL_SSE2__ /* Pass P and dir by reference to aligned vector */ -ccl_device_forceinline bool bvh_cardinal_curve_intersect(KernelGlobals *kg, Intersection *isect, +ccl_device_curveintersect bool bvh_cardinal_curve_intersect(KernelGlobals *kg, Intersection *isect, const float3 &P, const float3 &dir, uint visibility, int object, int curveAddr, float time, int type, uint *lcg_state, float difl, float extmax) #else -ccl_device_forceinline bool bvh_cardinal_curve_intersect(KernelGlobals *kg, Intersection *isect, +ccl_device_curveintersect bool bvh_cardinal_curve_intersect(KernelGlobals *kg, Intersection *isect, float3 P, float3 dir, uint visibility, int object, int curveAddr, float time,int type, uint *lcg_state, float difl, float extmax) #endif { + const bool is_curve_primitive = (type & PRIMITIVE_CURVE); + + if(!is_curve_primitive && kernel_data.bvh.use_bvh_steps) { + const float2 prim_time = kernel_tex_fetch(__prim_time, curveAddr); + if(time < prim_time.x || time > prim_time.y) { + return false; + } + } + int segment = PRIMITIVE_UNPACK_SEGMENT(type); float epsilon = 0.0f; float r_st, r_en; @@ -255,9 +270,9 @@ ccl_device_forceinline bool bvh_cardinal_curve_intersect(KernelGlobals *kg, Inte int ka = max(k0 - 1, v00.x); int kb = min(k1 + 1, v00.x + v00.y - 1); -#ifdef __KERNEL_AVX2__ +#if defined(__KERNEL_AVX2__) && (!defined(_MSC_VER) || _MSC_VER > 1800) avxf P_curve_0_1, P_curve_2_3; - if(type & PRIMITIVE_CURVE) { + if(is_curve_primitive) { P_curve_0_1 = _mm256_loadu2_m128(&kg->__curve_keys.data[k0].x, &kg->__curve_keys.data[ka].x); P_curve_2_3 = _mm256_loadu2_m128(&kg->__curve_keys.data[kb].x, &kg->__curve_keys.data[k1].x); } @@ -268,7 +283,7 @@ ccl_device_forceinline bool bvh_cardinal_curve_intersect(KernelGlobals *kg, Inte #else /* __KERNEL_AVX2__ */ ssef P_curve[4]; - if(type & PRIMITIVE_CURVE) { + if(is_curve_primitive) { P_curve[0] = load4f(&kg->__curve_keys.data[ka].x); P_curve[1] = load4f(&kg->__curve_keys.data[k0].x); P_curve[2] = load4f(&kg->__curve_keys.data[k1].x); @@ -290,7 +305,7 @@ ccl_device_forceinline bool bvh_cardinal_curve_intersect(KernelGlobals *kg, Inte ssef htfm1 = shuffle<1, 0, 1, 3>(load1f_first(extract<0>(d_ss)), vdir0); ssef htfm2 = shuffle<1, 3, 2, 3>(mul_shuf, vdir0); -#ifdef __KERNEL_AVX2__ +#if defined(__KERNEL_AVX2__) && (!defined(_MSC_VER) || _MSC_VER > 1800) const avxf vPP = _mm256_broadcast_ps(&P.m128); const avxf htfm00 = avxf(htfm0.m128, htfm0.m128); const avxf htfm11 = avxf(htfm1.m128, htfm1.m128); @@ -363,7 +378,7 @@ ccl_device_forceinline bool bvh_cardinal_curve_intersect(KernelGlobals *kg, Inte float4 P_curve[4]; - if(type & PRIMITIVE_CURVE) { + if(is_curve_primitive) { P_curve[0] = kernel_tex_fetch(__curve_keys, ka); P_curve[1] = kernel_tex_fetch(__curve_keys, k0); P_curve[2] = kernel_tex_fetch(__curve_keys, k1); @@ -679,7 +694,7 @@ ccl_device_forceinline bool bvh_cardinal_curve_intersect(KernelGlobals *kg, Inte return hit; } -ccl_device_forceinline bool bvh_curve_intersect(KernelGlobals *kg, Intersection *isect, +ccl_device_curveintersect bool bvh_curve_intersect(KernelGlobals *kg, Intersection *isect, float3 P, float3 direction, uint visibility, int object, int curveAddr, float time, int type, uint *lcg_state, float difl, float extmax) { /* define few macros to minimize code duplication for SSE */ @@ -689,6 +704,15 @@ ccl_device_forceinline bool bvh_curve_intersect(KernelGlobals *kg, Intersection # define dot3(x, y) dot(x, y) #endif + const bool is_curve_primitive = (type & PRIMITIVE_CURVE); + + if(!is_curve_primitive && kernel_data.bvh.use_bvh_steps) { + const float2 prim_time = kernel_tex_fetch(__prim_time, curveAddr); + if(time < prim_time.x || time > prim_time.y) { + return false; + } + } + int segment = PRIMITIVE_UNPACK_SEGMENT(type); /* curve Intersection check */ int flags = kernel_data.curve.curveflags; @@ -703,7 +727,7 @@ ccl_device_forceinline bool bvh_curve_intersect(KernelGlobals *kg, Intersection #ifndef __KERNEL_SSE2__ float4 P_curve[2]; - if(type & PRIMITIVE_CURVE) { + if(is_curve_primitive) { P_curve[0] = kernel_tex_fetch(__curve_keys, k0); P_curve[1] = kernel_tex_fetch(__curve_keys, k1); } @@ -738,7 +762,7 @@ ccl_device_forceinline bool bvh_curve_intersect(KernelGlobals *kg, Intersection #else ssef P_curve[2]; - if(type & PRIMITIVE_CURVE) { + if(is_curve_primitive) { P_curve[0] = load4f(&kg->__curve_keys.data[k0].x); P_curve[1] = load4f(&kg->__curve_keys.data[k1].x); } @@ -948,7 +972,7 @@ ccl_device_inline float3 bvh_curve_refine(KernelGlobals *kg, ShaderData *sd, con if(isect->object != OBJECT_NONE) { #ifdef __OBJECT_MOTION__ - Transform tfm = ccl_fetch(sd, ob_itfm); + Transform tfm = sd->ob_itfm; #else Transform tfm = object_fetch_transform(kg, isect->object, OBJECT_INVERSE_TRANSFORM); #endif @@ -961,7 +985,7 @@ ccl_device_inline float3 bvh_curve_refine(KernelGlobals *kg, ShaderData *sd, con int prim = kernel_tex_fetch(__prim_index, isect->prim); float4 v00 = kernel_tex_fetch(__curves, prim); - int k0 = __float_as_int(v00.x) + PRIMITIVE_UNPACK_SEGMENT(ccl_fetch(sd, type)); + int k0 = __float_as_int(v00.x) + PRIMITIVE_UNPACK_SEGMENT(sd->type); int k1 = k0 + 1; float3 tg; @@ -972,14 +996,14 @@ ccl_device_inline float3 bvh_curve_refine(KernelGlobals *kg, ShaderData *sd, con float4 P_curve[4]; - if(ccl_fetch(sd, type) & PRIMITIVE_CURVE) { + if(sd->type & PRIMITIVE_CURVE) { P_curve[0] = kernel_tex_fetch(__curve_keys, ka); P_curve[1] = kernel_tex_fetch(__curve_keys, k0); P_curve[2] = kernel_tex_fetch(__curve_keys, k1); P_curve[3] = kernel_tex_fetch(__curve_keys, kb); } else { - motion_cardinal_curve_keys(kg, ccl_fetch(sd, object), ccl_fetch(sd, prim), ccl_fetch(sd, time), ka, k0, k1, kb, P_curve); + motion_cardinal_curve_keys(kg, sd->object, sd->prim, sd->time, ka, k0, k1, kb, P_curve); } float3 p[4]; @@ -991,43 +1015,43 @@ ccl_device_inline float3 bvh_curve_refine(KernelGlobals *kg, ShaderData *sd, con P = P + D*t; #ifdef __UV__ - ccl_fetch(sd, u) = isect->u; - ccl_fetch(sd, v) = 0.0f; + sd->u = isect->u; + sd->v = 0.0f; #endif tg = normalize(curvetangent(isect->u, p[0], p[1], p[2], p[3])); if(kernel_data.curve.curveflags & CURVE_KN_RIBBONS) { - ccl_fetch(sd, Ng) = normalize(-(D - tg * (dot(tg, D)))); + sd->Ng = normalize(-(D - tg * (dot(tg, D)))); } else { /* direction from inside to surface of curve */ float3 p_curr = curvepoint(isect->u, p[0], p[1], p[2], p[3]); - ccl_fetch(sd, Ng) = normalize(P - p_curr); + sd->Ng = normalize(P - p_curr); /* adjustment for changing radius */ float gd = isect->v; if(gd != 0.0f) { - ccl_fetch(sd, Ng) = ccl_fetch(sd, Ng) - gd * tg; - ccl_fetch(sd, Ng) = normalize(ccl_fetch(sd, Ng)); + sd->Ng = sd->Ng - gd * tg; + sd->Ng = normalize(sd->Ng); } } /* todo: sometimes the normal is still so that this is detected as * backfacing even if cull backfaces is enabled */ - ccl_fetch(sd, N) = ccl_fetch(sd, Ng); + sd->N = sd->Ng; } else { float4 P_curve[2]; - if(ccl_fetch(sd, type) & PRIMITIVE_CURVE) { + if(sd->type & PRIMITIVE_CURVE) { P_curve[0]= kernel_tex_fetch(__curve_keys, k0); P_curve[1]= kernel_tex_fetch(__curve_keys, k1); } else { - motion_curve_keys(kg, ccl_fetch(sd, object), ccl_fetch(sd, prim), ccl_fetch(sd, time), k0, k1, P_curve); + motion_curve_keys(kg, sd->object, sd->prim, sd->time, k0, k1, P_curve); } float l = 1.0f; @@ -1038,39 +1062,39 @@ ccl_device_inline float3 bvh_curve_refine(KernelGlobals *kg, ShaderData *sd, con float3 dif = P - float4_to_float3(P_curve[0]); #ifdef __UV__ - ccl_fetch(sd, u) = dot(dif,tg)/l; - ccl_fetch(sd, v) = 0.0f; + sd->u = dot(dif,tg)/l; + sd->v = 0.0f; #endif if(flag & CURVE_KN_TRUETANGENTGNORMAL) { - ccl_fetch(sd, Ng) = -(D - tg * dot(tg, D)); - ccl_fetch(sd, Ng) = normalize(ccl_fetch(sd, Ng)); + sd->Ng = -(D - tg * dot(tg, D)); + sd->Ng = normalize(sd->Ng); } else { float gd = isect->v; /* direction from inside to surface of curve */ - ccl_fetch(sd, Ng) = (dif - tg * ccl_fetch(sd, u) * l) / (P_curve[0].w + ccl_fetch(sd, u) * l * gd); + sd->Ng = (dif - tg * sd->u * l) / (P_curve[0].w + sd->u * l * gd); /* adjustment for changing radius */ if(gd != 0.0f) { - ccl_fetch(sd, Ng) = ccl_fetch(sd, Ng) - gd * tg; - ccl_fetch(sd, Ng) = normalize(ccl_fetch(sd, Ng)); + sd->Ng = sd->Ng - gd * tg; + sd->Ng = normalize(sd->Ng); } } - ccl_fetch(sd, N) = ccl_fetch(sd, Ng); + sd->N = sd->Ng; } #ifdef __DPDU__ /* dPdu/dPdv */ - ccl_fetch(sd, dPdu) = tg; - ccl_fetch(sd, dPdv) = cross(tg, ccl_fetch(sd, Ng)); + sd->dPdu = tg; + sd->dPdv = cross(tg, sd->Ng); #endif if(isect->object != OBJECT_NONE) { #ifdef __OBJECT_MOTION__ - Transform tfm = ccl_fetch(sd, ob_tfm); + Transform tfm = sd->ob_tfm; #else Transform tfm = object_fetch_transform(kg, isect->object, OBJECT_TRANSFORM); #endif diff --git a/intern/cycles/kernel/geom/geom_motion_triangle_intersect.h b/intern/cycles/kernel/geom/geom_motion_triangle_intersect.h index d57d74ea882..f74995becf5 100644 --- a/intern/cycles/kernel/geom/geom_motion_triangle_intersect.h +++ b/intern/cycles/kernel/geom/geom_motion_triangle_intersect.h @@ -48,7 +48,7 @@ ccl_device_inline float3 motion_triangle_refine(KernelGlobals *kg, return P; } # ifdef __OBJECT_MOTION__ - Transform tfm = ccl_fetch(sd, ob_itfm); + Transform tfm = sd->ob_itfm; # else Transform tfm = object_fetch_transform(kg, isect->object, @@ -77,7 +77,7 @@ ccl_device_inline float3 motion_triangle_refine(KernelGlobals *kg, if(isect->object != OBJECT_NONE) { # ifdef __OBJECT_MOTION__ - Transform tfm = ccl_fetch(sd, ob_tfm); + Transform tfm = sd->ob_tfm; # else Transform tfm = object_fetch_transform(kg, isect->object, @@ -116,7 +116,7 @@ float3 motion_triangle_refine_subsurface(KernelGlobals *kg, # ifdef __INTERSECTION_REFINE__ if(isect->object != OBJECT_NONE) { # ifdef __OBJECT_MOTION__ - Transform tfm = ccl_fetch(sd, ob_itfm); + Transform tfm = sd->ob_itfm; # else Transform tfm = object_fetch_transform(kg, isect->object, @@ -144,7 +144,7 @@ float3 motion_triangle_refine_subsurface(KernelGlobals *kg, if(isect->object != OBJECT_NONE) { # ifdef __OBJECT_MOTION__ - Transform tfm = ccl_fetch(sd, ob_tfm); + Transform tfm = sd->ob_tfm; # else Transform tfm = object_fetch_transform(kg, isect->object, @@ -166,14 +166,15 @@ float3 motion_triangle_refine_subsurface(KernelGlobals *kg, * time and do a ray intersection with the resulting triangle. */ -ccl_device_inline bool motion_triangle_intersect(KernelGlobals *kg, - Intersection *isect, - float3 P, - float3 dir, - float time, - uint visibility, - int object, - int prim_addr) +ccl_device_inline bool motion_triangle_intersect( + KernelGlobals *kg, + Intersection *isect, + float3 P, + float3 dir, + float time, + uint visibility, + int object, + int prim_addr) { /* Primitive index for vertex location lookup. */ int prim = kernel_tex_fetch(__prim_index, prim_addr); @@ -185,11 +186,15 @@ ccl_device_inline bool motion_triangle_intersect(KernelGlobals *kg, motion_triangle_vertices(kg, fobject, prim, time, verts); /* Ray-triangle intersection, unoptimized. */ float t, u, v; - if(ray_triangle_intersect_uv(P, - dir, - isect->t, - verts[2], verts[0], verts[1], - &u, &v, &t)) + if(ray_triangle_intersect(P, + dir, + isect->t, +#if defined(__KERNEL_SSE2__) && defined(__KERNEL_SSE__) + (ssef*)verts, +#else + verts[0], verts[1], verts[2], +#endif + &u, &v, &t)) { #ifdef __VISIBILITY_FLAG__ /* Visibility flag test. we do it here under the assumption @@ -237,11 +242,15 @@ ccl_device_inline void motion_triangle_intersect_subsurface( motion_triangle_vertices(kg, fobject, prim, time, verts); /* Ray-triangle intersection, unoptimized. */ float t, u, v; - if(ray_triangle_intersect_uv(P, - dir, - tmax, - verts[2], verts[0], verts[1], - &u, &v, &t)) + if(ray_triangle_intersect(P, + dir, + tmax, +#if defined(__KERNEL_SSE2__) && defined(__KERNEL_SSE__) + (ssef*)verts, +#else + verts[0], verts[1], verts[2], +#endif + &u, &v, &t)) { for(int i = min(max_hits, ss_isect->num_hits) - 1; i >= 0; --i) { if(ss_isect->hits[i].t == t) { diff --git a/intern/cycles/kernel/geom/geom_motion_triangle_shader.h b/intern/cycles/kernel/geom/geom_motion_triangle_shader.h index 0e024a05db6..cb456056e20 100644 --- a/intern/cycles/kernel/geom/geom_motion_triangle_shader.h +++ b/intern/cycles/kernel/geom/geom_motion_triangle_shader.h @@ -39,26 +39,26 @@ ccl_device_noinline void motion_triangle_shader_setup(KernelGlobals *kg, bool subsurface) { /* Get shader. */ - ccl_fetch(sd, shader) = kernel_tex_fetch(__tri_shader, ccl_fetch(sd, prim)); + sd->shader = kernel_tex_fetch(__tri_shader, sd->prim); /* Get motion info. */ /* TODO(sergey): This logic is really similar to motion_triangle_vertices(), * can we de-duplicate something here? */ int numsteps, numverts; - object_motion_info(kg, ccl_fetch(sd, object), &numsteps, &numverts, NULL); + object_motion_info(kg, sd->object, &numsteps, &numverts, NULL); /* Figure out which steps we need to fetch and their interpolation factor. */ int maxstep = numsteps*2; - int step = min((int)(ccl_fetch(sd, time)*maxstep), maxstep-1); - float t = ccl_fetch(sd, time)*maxstep - step; + int step = min((int)(sd->time*maxstep), maxstep-1); + float t = sd->time*maxstep - step; /* Find attribute. */ AttributeElement elem; - int offset = find_attribute_motion(kg, ccl_fetch(sd, object), + int offset = find_attribute_motion(kg, sd->object, ATTR_STD_MOTION_VERTEX_POSITION, &elem); kernel_assert(offset != ATTR_STD_NOT_FOUND); /* Fetch vertex coordinates. */ float3 verts[3], next_verts[3]; - uint4 tri_vindex = kernel_tex_fetch(__tri_vindex, ccl_fetch(sd, prim)); + uint4 tri_vindex = kernel_tex_fetch(__tri_vindex, sd->prim); motion_triangle_verts_for_step(kg, tri_vindex, offset, numverts, numsteps, step, verts); motion_triangle_verts_for_step(kg, tri_vindex, offset, numverts, numsteps, step+1, next_verts); /* Interpolate between steps. */ @@ -68,7 +68,7 @@ ccl_device_noinline void motion_triangle_shader_setup(KernelGlobals *kg, /* Compute refined position. */ #ifdef __SUBSURFACE__ if(subsurface) { - ccl_fetch(sd, P) = motion_triangle_refine_subsurface(kg, + sd->P = motion_triangle_refine_subsurface(kg, sd, isect, ray, @@ -77,29 +77,29 @@ ccl_device_noinline void motion_triangle_shader_setup(KernelGlobals *kg, else #endif /* __SUBSURFACE__*/ { - ccl_fetch(sd, P) = motion_triangle_refine(kg, sd, isect, ray, verts); + sd->P = motion_triangle_refine(kg, sd, isect, ray, verts); } /* Compute face normal. */ float3 Ng; - if(ccl_fetch(sd, object_flag) & SD_OBJECT_NEGATIVE_SCALE_APPLIED) { + if(sd->object_flag & SD_OBJECT_NEGATIVE_SCALE_APPLIED) { Ng = normalize(cross(verts[2] - verts[0], verts[1] - verts[0])); } else { Ng = normalize(cross(verts[1] - verts[0], verts[2] - verts[0])); } - ccl_fetch(sd, Ng) = Ng; - ccl_fetch(sd, N) = Ng; + sd->Ng = Ng; + sd->N = Ng; /* Compute derivatives of P w.r.t. uv. */ #ifdef __DPDU__ - ccl_fetch(sd, dPdu) = (verts[0] - verts[2]); - ccl_fetch(sd, dPdv) = (verts[1] - verts[2]); + sd->dPdu = (verts[0] - verts[2]); + sd->dPdv = (verts[1] - verts[2]); #endif /* Compute smooth normal. */ - if(ccl_fetch(sd, shader) & SHADER_SMOOTH_NORMAL) { + if(sd->shader & SHADER_SMOOTH_NORMAL) { /* Find attribute. */ AttributeElement elem; int offset = find_attribute_motion(kg, - ccl_fetch(sd, object), + sd->object, ATTR_STD_MOTION_VERTEX_NORMAL, &elem); kernel_assert(offset != ATTR_STD_NOT_FOUND); @@ -112,10 +112,10 @@ ccl_device_noinline void motion_triangle_shader_setup(KernelGlobals *kg, normals[1] = (1.0f - t)*normals[1] + t*next_normals[1]; normals[2] = (1.0f - t)*normals[2] + t*next_normals[2]; /* Interpolate between vertices. */ - float u = ccl_fetch(sd, u); - float v = ccl_fetch(sd, v); + float u = sd->u; + float v = sd->v; float w = 1.0f - u - v; - ccl_fetch(sd, N) = (u*normals[0] + v*normals[1] + w*normals[2]); + sd->N = (u*normals[0] + v*normals[1] + w*normals[2]); } } diff --git a/intern/cycles/kernel/geom/geom_object.h b/intern/cycles/kernel/geom/geom_object.h index f51b2d18657..6ecdfe0173a 100644 --- a/intern/cycles/kernel/geom/geom_object.h +++ b/intern/cycles/kernel/geom/geom_object.h @@ -137,9 +137,9 @@ ccl_device_inline Transform object_fetch_transform_motion_test(KernelGlobals *kg ccl_device_inline void object_position_transform(KernelGlobals *kg, const ShaderData *sd, float3 *P) { #ifdef __OBJECT_MOTION__ - *P = transform_point_auto(&ccl_fetch(sd, ob_tfm), *P); + *P = transform_point_auto(&sd->ob_tfm, *P); #else - Transform tfm = object_fetch_transform(kg, ccl_fetch(sd, object), OBJECT_TRANSFORM); + Transform tfm = object_fetch_transform(kg, sd->object, OBJECT_TRANSFORM); *P = transform_point(&tfm, *P); #endif } @@ -149,9 +149,9 @@ ccl_device_inline void object_position_transform(KernelGlobals *kg, const Shader ccl_device_inline void object_inverse_position_transform(KernelGlobals *kg, const ShaderData *sd, float3 *P) { #ifdef __OBJECT_MOTION__ - *P = transform_point_auto(&ccl_fetch(sd, ob_itfm), *P); + *P = transform_point_auto(&sd->ob_itfm, *P); #else - Transform tfm = object_fetch_transform(kg, ccl_fetch(sd, object), OBJECT_INVERSE_TRANSFORM); + Transform tfm = object_fetch_transform(kg, sd->object, OBJECT_INVERSE_TRANSFORM); *P = transform_point(&tfm, *P); #endif } @@ -161,12 +161,12 @@ ccl_device_inline void object_inverse_position_transform(KernelGlobals *kg, cons ccl_device_inline void object_inverse_normal_transform(KernelGlobals *kg, const ShaderData *sd, float3 *N) { #ifdef __OBJECT_MOTION__ - if((ccl_fetch(sd, object) != OBJECT_NONE) || (ccl_fetch(sd, type) == PRIMITIVE_LAMP)) { - *N = normalize(transform_direction_transposed_auto(&ccl_fetch(sd, ob_tfm), *N)); + if((sd->object != OBJECT_NONE) || (sd->type == PRIMITIVE_LAMP)) { + *N = normalize(transform_direction_transposed_auto(&sd->ob_tfm, *N)); } #else - if(ccl_fetch(sd, object) != OBJECT_NONE) { - Transform tfm = object_fetch_transform(kg, ccl_fetch(sd, object), OBJECT_TRANSFORM); + if(sd->object != OBJECT_NONE) { + Transform tfm = object_fetch_transform(kg, sd->object, OBJECT_TRANSFORM); *N = normalize(transform_direction_transposed(&tfm, *N)); } #endif @@ -177,9 +177,9 @@ ccl_device_inline void object_inverse_normal_transform(KernelGlobals *kg, const ccl_device_inline void object_normal_transform(KernelGlobals *kg, const ShaderData *sd, float3 *N) { #ifdef __OBJECT_MOTION__ - *N = normalize(transform_direction_transposed_auto(&ccl_fetch(sd, ob_itfm), *N)); + *N = normalize(transform_direction_transposed_auto(&sd->ob_itfm, *N)); #else - Transform tfm = object_fetch_transform(kg, ccl_fetch(sd, object), OBJECT_INVERSE_TRANSFORM); + Transform tfm = object_fetch_transform(kg, sd->object, OBJECT_INVERSE_TRANSFORM); *N = normalize(transform_direction_transposed(&tfm, *N)); #endif } @@ -189,9 +189,9 @@ ccl_device_inline void object_normal_transform(KernelGlobals *kg, const ShaderDa ccl_device_inline void object_dir_transform(KernelGlobals *kg, const ShaderData *sd, float3 *D) { #ifdef __OBJECT_MOTION__ - *D = transform_direction_auto(&ccl_fetch(sd, ob_tfm), *D); + *D = transform_direction_auto(&sd->ob_tfm, *D); #else - Transform tfm = object_fetch_transform(kg, ccl_fetch(sd, object), OBJECT_TRANSFORM); + Transform tfm = object_fetch_transform(kg, sd->object, OBJECT_TRANSFORM); *D = transform_direction(&tfm, *D); #endif } @@ -201,9 +201,9 @@ ccl_device_inline void object_dir_transform(KernelGlobals *kg, const ShaderData ccl_device_inline void object_inverse_dir_transform(KernelGlobals *kg, const ShaderData *sd, float3 *D) { #ifdef __OBJECT_MOTION__ - *D = transform_direction_auto(&ccl_fetch(sd, ob_itfm), *D); + *D = transform_direction_auto(&sd->ob_itfm, *D); #else - Transform tfm = object_fetch_transform(kg, ccl_fetch(sd, object), OBJECT_INVERSE_TRANSFORM); + Transform tfm = object_fetch_transform(kg, sd->object, OBJECT_INVERSE_TRANSFORM); *D = transform_direction(&tfm, *D); #endif } @@ -212,13 +212,13 @@ ccl_device_inline void object_inverse_dir_transform(KernelGlobals *kg, const Sha ccl_device_inline float3 object_location(KernelGlobals *kg, const ShaderData *sd) { - if(ccl_fetch(sd, object) == OBJECT_NONE) + if(sd->object == OBJECT_NONE) return make_float3(0.0f, 0.0f, 0.0f); #ifdef __OBJECT_MOTION__ - return make_float3(ccl_fetch(sd, ob_tfm).x.w, ccl_fetch(sd, ob_tfm).y.w, ccl_fetch(sd, ob_tfm).z.w); + return make_float3(sd->ob_tfm.x.w, sd->ob_tfm.y.w, sd->ob_tfm.z.w); #else - Transform tfm = object_fetch_transform(kg, ccl_fetch(sd, object), OBJECT_TRANSFORM); + Transform tfm = object_fetch_transform(kg, sd->object, OBJECT_TRANSFORM); return make_float3(tfm.x.w, tfm.y.w, tfm.z.w); #endif } @@ -326,7 +326,7 @@ ccl_device_inline uint object_patch_map_offset(KernelGlobals *kg, int object) ccl_device int shader_pass_id(KernelGlobals *kg, const ShaderData *sd) { - return kernel_tex_fetch(__shader_flag, (ccl_fetch(sd, shader) & SHADER_MASK)*SHADER_SIZE + 1); + return kernel_tex_fetch(__shader_flag, (sd->shader & SHADER_MASK)*SHADER_SIZE + 1); } /* Particle data from which object was instanced */ @@ -425,7 +425,13 @@ ccl_device_inline float3 bvh_inverse_direction(float3 dir) /* Transform ray into object space to enter static object in BVH */ -ccl_device_inline void bvh_instance_push(KernelGlobals *kg, int object, const Ray *ray, float3 *P, float3 *dir, float3 *idir, ccl_addr_space float *t) +ccl_device_inline float bvh_instance_push(KernelGlobals *kg, + int object, + const Ray *ray, + float3 *P, + float3 *dir, + float3 *idir, + float t) { Transform tfm = object_fetch_transform(kg, object, OBJECT_INVERSE_TRANSFORM); @@ -435,8 +441,11 @@ ccl_device_inline void bvh_instance_push(KernelGlobals *kg, int object, const Ra *dir = bvh_clamp_direction(normalize_len(transform_direction(&tfm, ray->D), &len)); *idir = bvh_inverse_direction(*dir); - if(*t != FLT_MAX) - *t *= len; + if(t != FLT_MAX) { + t *= len; + } + + return t; } #ifdef __QBVH__ @@ -473,16 +482,24 @@ ccl_device_inline void qbvh_instance_push(KernelGlobals *kg, /* Transorm ray to exit static object in BVH */ -ccl_device_inline void bvh_instance_pop(KernelGlobals *kg, int object, const Ray *ray, float3 *P, float3 *dir, float3 *idir, ccl_addr_space float *t) +ccl_device_inline float bvh_instance_pop(KernelGlobals *kg, + int object, + const Ray *ray, + float3 *P, + float3 *dir, + float3 *idir, + float t) { - if(*t != FLT_MAX) { + if(t != FLT_MAX) { Transform tfm = object_fetch_transform(kg, object, OBJECT_INVERSE_TRANSFORM); - *t /= len(transform_direction(&tfm, ray->D)); + t /= len(transform_direction(&tfm, ray->D)); } *P = ray->P; *dir = bvh_clamp_direction(ray->D); *idir = bvh_inverse_direction(*dir); + + return t; } /* Same as above, but returns scale factor to apply to multiple intersection distances */ @@ -501,13 +518,13 @@ ccl_device_inline void bvh_instance_pop_factor(KernelGlobals *kg, int object, co #ifdef __OBJECT_MOTION__ /* Transform ray into object space to enter motion blurred object in BVH */ -ccl_device_inline void bvh_instance_motion_push(KernelGlobals *kg, +ccl_device_inline float bvh_instance_motion_push(KernelGlobals *kg, int object, const Ray *ray, float3 *P, float3 *dir, float3 *idir, - ccl_addr_space float *t, + float t, Transform *itfm) { object_fetch_transform_motion_test(kg, object, ray->time, itfm); @@ -518,8 +535,11 @@ ccl_device_inline void bvh_instance_motion_push(KernelGlobals *kg, *dir = bvh_clamp_direction(normalize_len(transform_direction(itfm, ray->D), &len)); *idir = bvh_inverse_direction(*dir); - if(*t != FLT_MAX) - *t *= len; + if(t != FLT_MAX) { + t *= len; + } + + return t; } #ifdef __QBVH__ @@ -557,22 +577,24 @@ ccl_device_inline void qbvh_instance_motion_push(KernelGlobals *kg, /* Transorm ray to exit motion blurred object in BVH */ -ccl_device_inline void bvh_instance_motion_pop(KernelGlobals *kg, - int object, - const Ray *ray, - float3 *P, - float3 *dir, - float3 *idir, - ccl_addr_space float *t, - Transform *itfm) -{ - if(*t != FLT_MAX) { - *t /= len(transform_direction(itfm, ray->D)); +ccl_device_inline float bvh_instance_motion_pop(KernelGlobals *kg, + int object, + const Ray *ray, + float3 *P, + float3 *dir, + float3 *idir, + float t, + Transform *itfm) +{ + if(t != FLT_MAX) { + t /= len(transform_direction(itfm, ray->D)); } *P = ray->P; *dir = bvh_clamp_direction(ray->D); *idir = bvh_inverse_direction(*dir); + + return t; } /* Same as above, but returns scale factor to apply to multiple intersection distances */ diff --git a/intern/cycles/kernel/geom/geom_patch.h b/intern/cycles/kernel/geom/geom_patch.h index 6a0ff5a4a04..5663b598508 100644 --- a/intern/cycles/kernel/geom/geom_patch.h +++ b/intern/cycles/kernel/geom/geom_patch.h @@ -267,7 +267,7 @@ ccl_device float patch_eval_float(KernelGlobals *kg, const ShaderData *sd, int o float weights_du[PATCH_MAX_CONTROL_VERTS]; float weights_dv[PATCH_MAX_CONTROL_VERTS]; - int num_control = patch_eval_control_verts(kg, ccl_fetch(sd, object), patch, u, v, channel, + int num_control = patch_eval_control_verts(kg, sd->object, patch, u, v, channel, indices, weights, weights_du, weights_dv); float val = 0.0f; @@ -294,7 +294,7 @@ ccl_device float3 patch_eval_float3(KernelGlobals *kg, const ShaderData *sd, int float weights_du[PATCH_MAX_CONTROL_VERTS]; float weights_dv[PATCH_MAX_CONTROL_VERTS]; - int num_control = patch_eval_control_verts(kg, ccl_fetch(sd, object), patch, u, v, channel, + int num_control = patch_eval_control_verts(kg, sd->object, patch, u, v, channel, indices, weights, weights_du, weights_dv); float3 val = make_float3(0.0f, 0.0f, 0.0f); @@ -321,7 +321,7 @@ ccl_device float3 patch_eval_uchar4(KernelGlobals *kg, const ShaderData *sd, int float weights_du[PATCH_MAX_CONTROL_VERTS]; float weights_dv[PATCH_MAX_CONTROL_VERTS]; - int num_control = patch_eval_control_verts(kg, ccl_fetch(sd, object), patch, u, v, channel, + int num_control = patch_eval_control_verts(kg, sd->object, patch, u, v, channel, indices, weights, weights_du, weights_dv); float3 val = make_float3(0.0f, 0.0f, 0.0f); diff --git a/intern/cycles/kernel/geom/geom_primitive.h b/intern/cycles/kernel/geom/geom_primitive.h index 8a73bb2f78b..90a9c2147cc 100644 --- a/intern/cycles/kernel/geom/geom_primitive.h +++ b/intern/cycles/kernel/geom/geom_primitive.h @@ -28,19 +28,19 @@ ccl_device_inline float primitive_attribute_float(KernelGlobals *kg, const AttributeDescriptor desc, float *dx, float *dy) { - if(ccl_fetch(sd, type) & PRIMITIVE_ALL_TRIANGLE) { + if(sd->type & PRIMITIVE_ALL_TRIANGLE) { if(subd_triangle_patch(kg, sd) == ~0) return triangle_attribute_float(kg, sd, desc, dx, dy); else return subd_triangle_attribute_float(kg, sd, desc, dx, dy); } #ifdef __HAIR__ - else if(ccl_fetch(sd, type) & PRIMITIVE_ALL_CURVE) { + else if(sd->type & PRIMITIVE_ALL_CURVE) { return curve_attribute_float(kg, sd, desc, dx, dy); } #endif #ifdef __VOLUME__ - else if(ccl_fetch(sd, object) != OBJECT_NONE && desc.element == ATTR_ELEMENT_VOXEL) { + else if(sd->object != OBJECT_NONE && desc.element == ATTR_ELEMENT_VOXEL) { return volume_attribute_float(kg, sd, desc, dx, dy); } #endif @@ -56,19 +56,19 @@ ccl_device_inline float3 primitive_attribute_float3(KernelGlobals *kg, const AttributeDescriptor desc, float3 *dx, float3 *dy) { - if(ccl_fetch(sd, type) & PRIMITIVE_ALL_TRIANGLE) { + if(sd->type & PRIMITIVE_ALL_TRIANGLE) { if(subd_triangle_patch(kg, sd) == ~0) return triangle_attribute_float3(kg, sd, desc, dx, dy); else return subd_triangle_attribute_float3(kg, sd, desc, dx, dy); } #ifdef __HAIR__ - else if(ccl_fetch(sd, type) & PRIMITIVE_ALL_CURVE) { + else if(sd->type & PRIMITIVE_ALL_CURVE) { return curve_attribute_float3(kg, sd, desc, dx, dy); } #endif #ifdef __VOLUME__ - else if(ccl_fetch(sd, object) != OBJECT_NONE && desc.element == ATTR_ELEMENT_VOXEL) { + else if(sd->object != OBJECT_NONE && desc.element == ATTR_ELEMENT_VOXEL) { return volume_attribute_float3(kg, sd, desc, dx, dy); } #endif @@ -118,9 +118,9 @@ ccl_device bool primitive_ptex(KernelGlobals *kg, ShaderData *sd, float2 *uv, in ccl_device float3 primitive_tangent(KernelGlobals *kg, ShaderData *sd) { #ifdef __HAIR__ - if(ccl_fetch(sd, type) & PRIMITIVE_ALL_CURVE) + if(sd->type & PRIMITIVE_ALL_CURVE) # ifdef __DPDU__ - return normalize(ccl_fetch(sd, dPdu)); + return normalize(sd->dPdu); # else return make_float3(0.0f, 0.0f, 0.0f); # endif @@ -133,12 +133,12 @@ ccl_device float3 primitive_tangent(KernelGlobals *kg, ShaderData *sd) float3 data = primitive_attribute_float3(kg, sd, desc, NULL, NULL); data = make_float3(-(data.y - 0.5f), (data.x - 0.5f), 0.0f); object_normal_transform(kg, sd, &data); - return cross(ccl_fetch(sd, N), normalize(cross(data, ccl_fetch(sd, N)))); + return cross(sd->N, normalize(cross(data, sd->N))); } else { /* otherwise use surface derivatives */ #ifdef __DPDU__ - return normalize(ccl_fetch(sd, dPdu)); + return normalize(sd->dPdu); #else return make_float3(0.0f, 0.0f, 0.0f); #endif @@ -153,17 +153,17 @@ ccl_device_inline float4 primitive_motion_vector(KernelGlobals *kg, ShaderData * float3 center; #ifdef __HAIR__ - bool is_curve_primitive = ccl_fetch(sd, type) & PRIMITIVE_ALL_CURVE; + bool is_curve_primitive = sd->type & PRIMITIVE_ALL_CURVE; if(is_curve_primitive) { center = curve_motion_center_location(kg, sd); - if(!(ccl_fetch(sd, object_flag) & SD_OBJECT_TRANSFORM_APPLIED)) { + if(!(sd->object_flag & SD_OBJECT_TRANSFORM_APPLIED)) { object_position_transform(kg, sd, ¢er); } } else #endif - center = ccl_fetch(sd, P); + center = sd->P; float3 motion_pre = center, motion_post = center; @@ -173,16 +173,16 @@ ccl_device_inline float4 primitive_motion_vector(KernelGlobals *kg, ShaderData * if(desc.offset != ATTR_STD_NOT_FOUND) { /* get motion info */ int numverts, numkeys; - object_motion_info(kg, ccl_fetch(sd, object), NULL, &numverts, &numkeys); + object_motion_info(kg, sd->object, NULL, &numverts, &numkeys); /* lookup attributes */ motion_pre = primitive_attribute_float3(kg, sd, desc, NULL, NULL); - desc.offset += (ccl_fetch(sd, type) & PRIMITIVE_ALL_TRIANGLE)? numverts: numkeys; + desc.offset += (sd->type & PRIMITIVE_ALL_TRIANGLE)? numverts: numkeys; motion_post = primitive_attribute_float3(kg, sd, desc, NULL, NULL); #ifdef __HAIR__ - if(is_curve_primitive && (ccl_fetch(sd, object_flag) & SD_OBJECT_HAS_VERTEX_MOTION) == 0) { + if(is_curve_primitive && (sd->object_flag & SD_OBJECT_HAS_VERTEX_MOTION) == 0) { object_position_transform(kg, sd, &motion_pre); object_position_transform(kg, sd, &motion_post); } @@ -193,10 +193,10 @@ ccl_device_inline float4 primitive_motion_vector(KernelGlobals *kg, ShaderData * * transformation was set match the world/object space of motion_pre/post */ Transform tfm; - tfm = object_fetch_vector_transform(kg, ccl_fetch(sd, object), OBJECT_VECTOR_MOTION_PRE); + tfm = object_fetch_vector_transform(kg, sd->object, OBJECT_VECTOR_MOTION_PRE); motion_pre = transform_point(&tfm, motion_pre); - tfm = object_fetch_vector_transform(kg, ccl_fetch(sd, object), OBJECT_VECTOR_MOTION_POST); + tfm = object_fetch_vector_transform(kg, sd->object, OBJECT_VECTOR_MOTION_POST); motion_post = transform_point(&tfm, motion_post); float3 motion_center; diff --git a/intern/cycles/kernel/geom/geom_subd_triangle.h b/intern/cycles/kernel/geom/geom_subd_triangle.h index 647840dc696..044e82f03d4 100644 --- a/intern/cycles/kernel/geom/geom_subd_triangle.h +++ b/intern/cycles/kernel/geom/geom_subd_triangle.h @@ -22,14 +22,14 @@ CCL_NAMESPACE_BEGIN ccl_device_inline uint subd_triangle_patch(KernelGlobals *kg, const ShaderData *sd) { - return (ccl_fetch(sd, prim) != PRIM_NONE) ? kernel_tex_fetch(__tri_patch, ccl_fetch(sd, prim)) : ~0; + return (sd->prim != PRIM_NONE) ? kernel_tex_fetch(__tri_patch, sd->prim) : ~0; } /* UV coords of triangle within patch */ ccl_device_inline void subd_triangle_patch_uv(KernelGlobals *kg, const ShaderData *sd, float2 uv[3]) { - uint4 tri_vindex = kernel_tex_fetch(__tri_vindex, ccl_fetch(sd, prim)); + uint4 tri_vindex = kernel_tex_fetch(__tri_vindex, sd->prim); uv[0] = kernel_tex_fetch(__tri_patch_uv, tri_vindex.x); uv[1] = kernel_tex_fetch(__tri_patch_uv, tri_vindex.y); @@ -110,7 +110,7 @@ ccl_device_noinline float subd_triangle_attribute_float(KernelGlobals *kg, const float2 dpdv = uv[1] - uv[2]; /* p is [s, t] */ - float2 p = dpdu * ccl_fetch(sd, u) + dpdv * ccl_fetch(sd, v) + uv[2]; + float2 p = dpdu * sd->u + dpdv * sd->v + uv[2]; float a, dads, dadt; a = patch_eval_float(kg, sd, desc.offset, patch, p.x, p.y, 0, &dads, &dadt); @@ -123,8 +123,8 @@ ccl_device_noinline float subd_triangle_attribute_float(KernelGlobals *kg, const float dtdv = dpdv.y; if(dx) { - float dudx = ccl_fetch(sd, du).dx; - float dvdx = ccl_fetch(sd, dv).dx; + float dudx = sd->du.dx; + float dvdx = sd->dv.dx; float dsdx = dsdu*dudx + dsdv*dvdx; float dtdx = dtdu*dudx + dtdv*dvdx; @@ -132,8 +132,8 @@ ccl_device_noinline float subd_triangle_attribute_float(KernelGlobals *kg, const *dx = dads*dsdx + dadt*dtdx; } if(dy) { - float dudy = ccl_fetch(sd, du).dy; - float dvdy = ccl_fetch(sd, dv).dy; + float dudy = sd->du.dy; + float dvdy = sd->dv.dy; float dsdy = dsdu*dudy + dsdv*dvdy; float dtdy = dtdu*dudy + dtdv*dvdy; @@ -174,11 +174,11 @@ ccl_device_noinline float subd_triangle_attribute_float(KernelGlobals *kg, const float c = mix(mix(f0, f1, uv[2].x), mix(f3, f2, uv[2].x), uv[2].y); #ifdef __RAY_DIFFERENTIALS__ - if(dx) *dx = ccl_fetch(sd, du).dx*a + ccl_fetch(sd, dv).dx*b - (ccl_fetch(sd, du).dx + ccl_fetch(sd, dv).dx)*c; - if(dy) *dy = ccl_fetch(sd, du).dy*a + ccl_fetch(sd, dv).dy*b - (ccl_fetch(sd, du).dy + ccl_fetch(sd, dv).dy)*c; + if(dx) *dx = sd->du.dx*a + sd->dv.dx*b - (sd->du.dx + sd->dv.dx)*c; + if(dy) *dy = sd->du.dy*a + sd->dv.dy*b - (sd->du.dy + sd->dv.dy)*c; #endif - return ccl_fetch(sd, u)*a + ccl_fetch(sd, v)*b + (1.0f - ccl_fetch(sd, u) - ccl_fetch(sd, v))*c; + return sd->u*a + sd->v*b + (1.0f - sd->u - sd->v)*c; } else if(desc.element == ATTR_ELEMENT_CORNER) { float2 uv[3]; @@ -202,11 +202,11 @@ ccl_device_noinline float subd_triangle_attribute_float(KernelGlobals *kg, const float c = mix(mix(f0, f1, uv[2].x), mix(f3, f2, uv[2].x), uv[2].y); #ifdef __RAY_DIFFERENTIALS__ - if(dx) *dx = ccl_fetch(sd, du).dx*a + ccl_fetch(sd, dv).dx*b - (ccl_fetch(sd, du).dx + ccl_fetch(sd, dv).dx)*c; - if(dy) *dy = ccl_fetch(sd, du).dy*a + ccl_fetch(sd, dv).dy*b - (ccl_fetch(sd, du).dy + ccl_fetch(sd, dv).dy)*c; + if(dx) *dx = sd->du.dx*a + sd->dv.dx*b - (sd->du.dx + sd->dv.dx)*c; + if(dy) *dy = sd->du.dy*a + sd->dv.dy*b - (sd->du.dy + sd->dv.dy)*c; #endif - return ccl_fetch(sd, u)*a + ccl_fetch(sd, v)*b + (1.0f - ccl_fetch(sd, u) - ccl_fetch(sd, v))*c; + return sd->u*a + sd->v*b + (1.0f - sd->u - sd->v)*c; } else { if(dx) *dx = 0.0f; @@ -229,7 +229,7 @@ ccl_device_noinline float3 subd_triangle_attribute_float3(KernelGlobals *kg, con float2 dpdv = uv[1] - uv[2]; /* p is [s, t] */ - float2 p = dpdu * ccl_fetch(sd, u) + dpdv * ccl_fetch(sd, v) + uv[2]; + float2 p = dpdu * sd->u + dpdv * sd->v + uv[2]; float3 a, dads, dadt; @@ -248,8 +248,8 @@ ccl_device_noinline float3 subd_triangle_attribute_float3(KernelGlobals *kg, con float dtdv = dpdv.y; if(dx) { - float dudx = ccl_fetch(sd, du).dx; - float dvdx = ccl_fetch(sd, dv).dx; + float dudx = sd->du.dx; + float dvdx = sd->dv.dx; float dsdx = dsdu*dudx + dsdv*dvdx; float dtdx = dtdu*dudx + dtdv*dvdx; @@ -257,8 +257,8 @@ ccl_device_noinline float3 subd_triangle_attribute_float3(KernelGlobals *kg, con *dx = dads*dsdx + dadt*dtdx; } if(dy) { - float dudy = ccl_fetch(sd, du).dy; - float dvdy = ccl_fetch(sd, dv).dy; + float dudy = sd->du.dy; + float dvdy = sd->dv.dy; float dsdy = dsdu*dudy + dsdv*dvdy; float dtdy = dtdu*dudy + dtdv*dvdy; @@ -299,11 +299,11 @@ ccl_device_noinline float3 subd_triangle_attribute_float3(KernelGlobals *kg, con float3 c = mix(mix(f0, f1, uv[2].x), mix(f3, f2, uv[2].x), uv[2].y); #ifdef __RAY_DIFFERENTIALS__ - if(dx) *dx = ccl_fetch(sd, du).dx*a + ccl_fetch(sd, dv).dx*b - (ccl_fetch(sd, du).dx + ccl_fetch(sd, dv).dx)*c; - if(dy) *dy = ccl_fetch(sd, du).dy*a + ccl_fetch(sd, dv).dy*b - (ccl_fetch(sd, du).dy + ccl_fetch(sd, dv).dy)*c; + if(dx) *dx = sd->du.dx*a + sd->dv.dx*b - (sd->du.dx + sd->dv.dx)*c; + if(dy) *dy = sd->du.dy*a + sd->dv.dy*b - (sd->du.dy + sd->dv.dy)*c; #endif - return ccl_fetch(sd, u)*a + ccl_fetch(sd, v)*b + (1.0f - ccl_fetch(sd, u) - ccl_fetch(sd, v))*c; + return sd->u*a + sd->v*b + (1.0f - sd->u - sd->v)*c; } else if(desc.element == ATTR_ELEMENT_CORNER || desc.element == ATTR_ELEMENT_CORNER_BYTE) { float2 uv[3]; @@ -337,11 +337,11 @@ ccl_device_noinline float3 subd_triangle_attribute_float3(KernelGlobals *kg, con float3 c = mix(mix(f0, f1, uv[2].x), mix(f3, f2, uv[2].x), uv[2].y); #ifdef __RAY_DIFFERENTIALS__ - if(dx) *dx = ccl_fetch(sd, du).dx*a + ccl_fetch(sd, dv).dx*b - (ccl_fetch(sd, du).dx + ccl_fetch(sd, dv).dx)*c; - if(dy) *dy = ccl_fetch(sd, du).dy*a + ccl_fetch(sd, dv).dy*b - (ccl_fetch(sd, du).dy + ccl_fetch(sd, dv).dy)*c; + if(dx) *dx = sd->du.dx*a + sd->dv.dx*b - (sd->du.dx + sd->dv.dx)*c; + if(dy) *dy = sd->du.dy*a + sd->dv.dy*b - (sd->du.dy + sd->dv.dy)*c; #endif - return ccl_fetch(sd, u)*a + ccl_fetch(sd, v)*b + (1.0f - ccl_fetch(sd, u) - ccl_fetch(sd, v))*c; + return sd->u*a + sd->v*b + (1.0f - sd->u - sd->v)*c; } else { if(dx) *dx = make_float3(0.0f, 0.0f, 0.0f); diff --git a/intern/cycles/kernel/geom/geom_triangle.h b/intern/cycles/kernel/geom/geom_triangle.h index 3229091bbb0..47778553b94 100644 --- a/intern/cycles/kernel/geom/geom_triangle.h +++ b/intern/cycles/kernel/geom/geom_triangle.h @@ -26,13 +26,13 @@ CCL_NAMESPACE_BEGIN ccl_device_inline float3 triangle_normal(KernelGlobals *kg, ShaderData *sd) { /* load triangle vertices */ - const uint4 tri_vindex = kernel_tex_fetch(__tri_vindex, ccl_fetch(sd, prim)); + const uint4 tri_vindex = kernel_tex_fetch(__tri_vindex, sd->prim); const float3 v0 = float4_to_float3(kernel_tex_fetch(__prim_tri_verts, tri_vindex.w+0)); const float3 v1 = float4_to_float3(kernel_tex_fetch(__prim_tri_verts, tri_vindex.w+1)); const float3 v2 = float4_to_float3(kernel_tex_fetch(__prim_tri_verts, tri_vindex.w+2)); /* return normal */ - if(ccl_fetch(sd, object_flag) & SD_OBJECT_NEGATIVE_SCALE_APPLIED) { + if(sd->object_flag & SD_OBJECT_NEGATIVE_SCALE_APPLIED) { return normalize(cross(v2 - v0, v1 - v0)); } else { @@ -110,34 +110,34 @@ ccl_device float triangle_attribute_float(KernelGlobals *kg, const ShaderData *s if(dx) *dx = 0.0f; if(dy) *dy = 0.0f; - return kernel_tex_fetch(__attributes_float, desc.offset + ccl_fetch(sd, prim)); + return kernel_tex_fetch(__attributes_float, desc.offset + sd->prim); } else if(desc.element == ATTR_ELEMENT_VERTEX || desc.element == ATTR_ELEMENT_VERTEX_MOTION) { - uint4 tri_vindex = kernel_tex_fetch(__tri_vindex, ccl_fetch(sd, prim)); + uint4 tri_vindex = kernel_tex_fetch(__tri_vindex, sd->prim); float f0 = kernel_tex_fetch(__attributes_float, desc.offset + tri_vindex.x); float f1 = kernel_tex_fetch(__attributes_float, desc.offset + tri_vindex.y); float f2 = kernel_tex_fetch(__attributes_float, desc.offset + tri_vindex.z); #ifdef __RAY_DIFFERENTIALS__ - if(dx) *dx = ccl_fetch(sd, du).dx*f0 + ccl_fetch(sd, dv).dx*f1 - (ccl_fetch(sd, du).dx + ccl_fetch(sd, dv).dx)*f2; - if(dy) *dy = ccl_fetch(sd, du).dy*f0 + ccl_fetch(sd, dv).dy*f1 - (ccl_fetch(sd, du).dy + ccl_fetch(sd, dv).dy)*f2; + if(dx) *dx = sd->du.dx*f0 + sd->dv.dx*f1 - (sd->du.dx + sd->dv.dx)*f2; + if(dy) *dy = sd->du.dy*f0 + sd->dv.dy*f1 - (sd->du.dy + sd->dv.dy)*f2; #endif - return ccl_fetch(sd, u)*f0 + ccl_fetch(sd, v)*f1 + (1.0f - ccl_fetch(sd, u) - ccl_fetch(sd, v))*f2; + return sd->u*f0 + sd->v*f1 + (1.0f - sd->u - sd->v)*f2; } else if(desc.element == ATTR_ELEMENT_CORNER) { - int tri = desc.offset + ccl_fetch(sd, prim)*3; + int tri = desc.offset + sd->prim*3; float f0 = kernel_tex_fetch(__attributes_float, tri + 0); float f1 = kernel_tex_fetch(__attributes_float, tri + 1); float f2 = kernel_tex_fetch(__attributes_float, tri + 2); #ifdef __RAY_DIFFERENTIALS__ - if(dx) *dx = ccl_fetch(sd, du).dx*f0 + ccl_fetch(sd, dv).dx*f1 - (ccl_fetch(sd, du).dx + ccl_fetch(sd, dv).dx)*f2; - if(dy) *dy = ccl_fetch(sd, du).dy*f0 + ccl_fetch(sd, dv).dy*f1 - (ccl_fetch(sd, du).dy + ccl_fetch(sd, dv).dy)*f2; + if(dx) *dx = sd->du.dx*f0 + sd->dv.dx*f1 - (sd->du.dx + sd->dv.dx)*f2; + if(dy) *dy = sd->du.dy*f0 + sd->dv.dy*f1 - (sd->du.dy + sd->dv.dy)*f2; #endif - return ccl_fetch(sd, u)*f0 + ccl_fetch(sd, v)*f1 + (1.0f - ccl_fetch(sd, u) - ccl_fetch(sd, v))*f2; + return sd->u*f0 + sd->v*f1 + (1.0f - sd->u - sd->v)*f2; } else { if(dx) *dx = 0.0f; @@ -153,24 +153,24 @@ ccl_device float3 triangle_attribute_float3(KernelGlobals *kg, const ShaderData if(dx) *dx = make_float3(0.0f, 0.0f, 0.0f); if(dy) *dy = make_float3(0.0f, 0.0f, 0.0f); - return float4_to_float3(kernel_tex_fetch(__attributes_float3, desc.offset + ccl_fetch(sd, prim))); + return float4_to_float3(kernel_tex_fetch(__attributes_float3, desc.offset + sd->prim)); } else if(desc.element == ATTR_ELEMENT_VERTEX || desc.element == ATTR_ELEMENT_VERTEX_MOTION) { - uint4 tri_vindex = kernel_tex_fetch(__tri_vindex, ccl_fetch(sd, prim)); + uint4 tri_vindex = kernel_tex_fetch(__tri_vindex, sd->prim); float3 f0 = float4_to_float3(kernel_tex_fetch(__attributes_float3, desc.offset + tri_vindex.x)); float3 f1 = float4_to_float3(kernel_tex_fetch(__attributes_float3, desc.offset + tri_vindex.y)); float3 f2 = float4_to_float3(kernel_tex_fetch(__attributes_float3, desc.offset + tri_vindex.z)); #ifdef __RAY_DIFFERENTIALS__ - if(dx) *dx = ccl_fetch(sd, du).dx*f0 + ccl_fetch(sd, dv).dx*f1 - (ccl_fetch(sd, du).dx + ccl_fetch(sd, dv).dx)*f2; - if(dy) *dy = ccl_fetch(sd, du).dy*f0 + ccl_fetch(sd, dv).dy*f1 - (ccl_fetch(sd, du).dy + ccl_fetch(sd, dv).dy)*f2; + if(dx) *dx = sd->du.dx*f0 + sd->dv.dx*f1 - (sd->du.dx + sd->dv.dx)*f2; + if(dy) *dy = sd->du.dy*f0 + sd->dv.dy*f1 - (sd->du.dy + sd->dv.dy)*f2; #endif - return ccl_fetch(sd, u)*f0 + ccl_fetch(sd, v)*f1 + (1.0f - ccl_fetch(sd, u) - ccl_fetch(sd, v))*f2; + return sd->u*f0 + sd->v*f1 + (1.0f - sd->u - sd->v)*f2; } else if(desc.element == ATTR_ELEMENT_CORNER || desc.element == ATTR_ELEMENT_CORNER_BYTE) { - int tri = desc.offset + ccl_fetch(sd, prim)*3; + int tri = desc.offset + sd->prim*3; float3 f0, f1, f2; if(desc.element == ATTR_ELEMENT_CORNER) { @@ -185,11 +185,11 @@ ccl_device float3 triangle_attribute_float3(KernelGlobals *kg, const ShaderData } #ifdef __RAY_DIFFERENTIALS__ - if(dx) *dx = ccl_fetch(sd, du).dx*f0 + ccl_fetch(sd, dv).dx*f1 - (ccl_fetch(sd, du).dx + ccl_fetch(sd, dv).dx)*f2; - if(dy) *dy = ccl_fetch(sd, du).dy*f0 + ccl_fetch(sd, dv).dy*f1 - (ccl_fetch(sd, du).dy + ccl_fetch(sd, dv).dy)*f2; + if(dx) *dx = sd->du.dx*f0 + sd->dv.dx*f1 - (sd->du.dx + sd->dv.dx)*f2; + if(dy) *dy = sd->du.dy*f0 + sd->dv.dy*f1 - (sd->du.dy + sd->dv.dy)*f2; #endif - return ccl_fetch(sd, u)*f0 + ccl_fetch(sd, v)*f1 + (1.0f - ccl_fetch(sd, u) - ccl_fetch(sd, v))*f2; + return sd->u*f0 + sd->v*f1 + (1.0f - sd->u - sd->v)*f2; } else { if(dx) *dx = make_float3(0.0f, 0.0f, 0.0f); diff --git a/intern/cycles/kernel/geom/geom_triangle_intersect.h b/intern/cycles/kernel/geom/geom_triangle_intersect.h index 4db121d94f4..804e74d7e37 100644 --- a/intern/cycles/kernel/geom/geom_triangle_intersect.h +++ b/intern/cycles/kernel/geom/geom_triangle_intersect.h @@ -22,232 +22,50 @@ CCL_NAMESPACE_BEGIN -/* Workaround stupidness of CUDA/OpenCL which doesn't allow to access indexed - * component of float3 value. - */ -#ifndef __KERNEL_CPU__ -# define IDX(vec, idx) \ - ((idx == 0) ? ((vec).x) : ( (idx == 1) ? ((vec).y) : ((vec).z) )) -#else -# define IDX(vec, idx) ((vec)[idx]) -#endif - -/* Ray-Triangle intersection for BVH traversal - * - * Sven Woop - * Watertight Ray/Triangle Intersection - * - * http://jcgt.org/published/0002/01/05/paper.pdf - */ - -/* Precalculated data for the ray->tri intersection. */ -typedef struct IsectPrecalc { - /* Maximal dimension kz, and orthogonal dimensions. */ - int kx, ky, kz; - - /* Shear constants. */ - float Sx, Sy, Sz; -} IsectPrecalc; - -#if (defined(__KERNEL_OPENCL_APPLE__)) || \ - (defined(__KERNEL_CUDA__) && (defined(i386) || defined(_M_IX86))) -ccl_device_noinline -#else -ccl_device_inline -#endif -void triangle_intersect_precalc(float3 dir, - IsectPrecalc *isect_precalc) -{ - /* Calculate dimension where the ray direction is maximal. */ -#ifndef __KERNEL_SSE__ - int kz = util_max_axis(make_float3(fabsf(dir.x), - fabsf(dir.y), - fabsf(dir.z))); - int kx = kz + 1; if(kx == 3) kx = 0; - int ky = kx + 1; if(ky == 3) ky = 0; -#else - int kx, ky, kz; - /* Avoiding mispredicted branch on direction. */ - kz = util_max_axis(fabs(dir)); - static const char inc_xaxis[] = {1, 2, 0, 55}; - static const char inc_yaxis[] = {2, 0, 1, 55}; - kx = inc_xaxis[kz]; - ky = inc_yaxis[kz]; -#endif - - float dir_kz = IDX(dir, kz); - - /* Swap kx and ky dimensions to preserve winding direction of triangles. */ - if(dir_kz < 0.0f) { - int tmp = kx; - kx = ky; - ky = tmp; - } - - /* Calculate the shear constants. */ - float inv_dir_z = 1.0f / dir_kz; - isect_precalc->Sx = IDX(dir, kx) * inv_dir_z; - isect_precalc->Sy = IDX(dir, ky) * inv_dir_z; - isect_precalc->Sz = inv_dir_z; - - /* Store the dimensions. */ - isect_precalc->kx = kx; - isect_precalc->ky = ky; - isect_precalc->kz = kz; -} - -/* TODO(sergey): Make it general utility function. */ -ccl_device_inline float xor_signmask(float x, int y) -{ - return __int_as_float(__float_as_int(x) ^ y); -} - ccl_device_inline bool triangle_intersect(KernelGlobals *kg, - const IsectPrecalc *isect_precalc, Intersection *isect, float3 P, + float3 dir, uint visibility, int object, int prim_addr) { - const int kx = isect_precalc->kx; - const int ky = isect_precalc->ky; - const int kz = isect_precalc->kz; - const float Sx = isect_precalc->Sx; - const float Sy = isect_precalc->Sy; - const float Sz = isect_precalc->Sz; - - /* Calculate vertices relative to ray origin. */ const uint tri_vindex = kernel_tex_fetch(__prim_tri_index, prim_addr); - -#if defined(__KERNEL_AVX2__) && defined(__KERNEL_SSE__) - const avxf avxf_P(P.m128, P.m128); - - const avxf tri_ab = kernel_tex_fetch_avxf(__prim_tri_verts, tri_vindex + 0); - const avxf tri_bc = kernel_tex_fetch_avxf(__prim_tri_verts, tri_vindex + 1); - - const avxf AB = tri_ab - avxf_P; - const avxf BC = tri_bc - avxf_P; - - const __m256i permute_mask = _mm256_set_epi32(0x3, kz, ky, kx, 0x3, kz, ky, kx); - - const avxf AB_k = shuffle(AB, permute_mask); - const avxf BC_k = shuffle(BC, permute_mask); - - /* Akz, Akz, Bkz, Bkz, Bkz, Bkz, Ckz, Ckz */ - const avxf ABBC_kz = shuffle<2>(AB_k, BC_k); - - /* Akx, Aky, Bkx, Bky, Bkx,Bky, Ckx, Cky */ - const avxf ABBC_kxy = shuffle<0,1,0,1>(AB_k, BC_k); - - const avxf Sxy(Sy, Sx, Sy, Sx); - - /* Ax, Ay, Bx, By, Bx, By, Cx, Cy */ - const avxf ABBC_xy = nmadd(ABBC_kz, Sxy, ABBC_kxy); - - float ABBC_kz_array[8]; - _mm256_storeu_ps((float*)&ABBC_kz_array, ABBC_kz); - - const float A_kz = ABBC_kz_array[0]; - const float B_kz = ABBC_kz_array[2]; - const float C_kz = ABBC_kz_array[6]; - - /* By, Bx, Cy, Cx, By, Bx, Ay, Ax */ - const avxf BCBA_yx = permute<3,2,7,6,3,2,1,0>(ABBC_xy); - - const avxf neg_mask(0,0,0,0,0x80000000, 0x80000000, 0x80000000, 0x80000000); - - /* W U V - * (AxBy-AyBx) (BxCy-ByCx) XX XX (BxBy-ByBx) (CxAy-CyAx) XX XX - */ - const avxf WUxxxxVxx_neg = _mm256_hsub_ps(ABBC_xy * BCBA_yx, neg_mask /* Dont care */); - - const avxf WUVWnegWUVW = permute<0,1,5,0,0,1,5,0>(WUxxxxVxx_neg) ^ neg_mask; - - /* Calculate scaled barycentric coordinates. */ - float WUVW_array[4]; - _mm_storeu_ps((float*)&WUVW_array, _mm256_castps256_ps128 (WUVWnegWUVW)); - - const float W = WUVW_array[0]; - const float U = WUVW_array[1]; - const float V = WUVW_array[2]; - - const int WUVW_mask = 0x7 & _mm256_movemask_ps(WUVWnegWUVW); - const int WUVW_zero = 0x7 & _mm256_movemask_ps(_mm256_cmp_ps(WUVWnegWUVW, - _mm256_setzero_ps(), 0)); - - if(!((WUVW_mask == 7) || (WUVW_mask == 0)) && ((WUVW_mask | WUVW_zero) != 7)) { - return false; - } +#if defined(__KERNEL_SSE2__) && defined(__KERNEL_SSE__) + const ssef *ssef_verts = (ssef*)&kg->__prim_tri_verts.data[tri_vindex]; #else const float4 tri_a = kernel_tex_fetch(__prim_tri_verts, tri_vindex+0), tri_b = kernel_tex_fetch(__prim_tri_verts, tri_vindex+1), tri_c = kernel_tex_fetch(__prim_tri_verts, tri_vindex+2); - const float3 A = make_float3(tri_a.x - P.x, tri_a.y - P.y, tri_a.z - P.z); - const float3 B = make_float3(tri_b.x - P.x, tri_b.y - P.y, tri_b.z - P.z); - const float3 C = make_float3(tri_c.x - P.x, tri_c.y - P.y, tri_c.z - P.z); - - const float A_kx = IDX(A, kx), A_ky = IDX(A, ky), A_kz = IDX(A, kz); - const float B_kx = IDX(B, kx), B_ky = IDX(B, ky), B_kz = IDX(B, kz); - const float C_kx = IDX(C, kx), C_ky = IDX(C, ky), C_kz = IDX(C, kz); - - /* Perform shear and scale of vertices. */ - const float Ax = A_kx - Sx * A_kz; - const float Ay = A_ky - Sy * A_kz; - const float Bx = B_kx - Sx * B_kz; - const float By = B_ky - Sy * B_kz; - const float Cx = C_kx - Sx * C_kz; - const float Cy = C_ky - Sy * C_kz; - - /* Calculate scaled barycentric coordinates. */ - float U = Cx * By - Cy * Bx; - float V = Ax * Cy - Ay * Cx; - float W = Bx * Ay - By * Ax; - if((U < 0.0f || V < 0.0f || W < 0.0f) && - (U > 0.0f || V > 0.0f || W > 0.0f)) - { - return false; - } #endif - - /* Calculate determinant. */ - float det = U + V + W; - if(UNLIKELY(det == 0.0f)) { - return false; - } - - /* Calculate scaled z-coordinates of vertices and use them to calculate - * the hit distance. - */ - const float T = (U * A_kz + V * B_kz + W * C_kz) * Sz; - const int sign_det = (__float_as_int(det) & 0x80000000); - const float sign_T = xor_signmask(T, sign_det); - if((sign_T < 0.0f) || - (sign_T > isect->t * xor_signmask(det, sign_det))) + float t, u, v; + if(ray_triangle_intersect(P, + dir, + isect->t, +#if defined(__KERNEL_SSE2__) && defined(__KERNEL_SSE__) + ssef_verts, +#else + float4_to_float3(tri_a), + float4_to_float3(tri_b), + float4_to_float3(tri_c), +#endif + &u, &v, &t)) { - return false; - } - #ifdef __VISIBILITY_FLAG__ - /* visibility flag test. we do it here under the assumption - * that most triangles are culled by node flags */ - if(kernel_tex_fetch(__prim_visibility, prim_addr) & visibility) + /* Visibility flag test. we do it here under the assumption + * that most triangles are culled by node flags. + */ + if(kernel_tex_fetch(__prim_visibility, prim_addr) & visibility) #endif - { -#ifdef __KERNEL_CUDA__ - if(A == B && B == C) { - return false; + { + isect->prim = prim_addr; + isect->object = object; + isect->type = PRIMITIVE_TRIANGLE; + isect->u = u; + isect->v = v; + isect->t = t; + return true; } -#endif - /* Normalize U, V, W, and T. */ - const float inv_det = 1.0f / det; - isect->prim = prim_addr; - isect->object = object; - isect->type = PRIMITIVE_TRIANGLE; - isect->u = U * inv_det; - isect->v = V * inv_det; - isect->t = T * inv_det; - return true; } return false; } @@ -260,138 +78,37 @@ ccl_device_inline bool triangle_intersect(KernelGlobals *kg, #ifdef __SUBSURFACE__ ccl_device_inline void triangle_intersect_subsurface( KernelGlobals *kg, - const IsectPrecalc *isect_precalc, SubsurfaceIntersection *ss_isect, float3 P, + float3 dir, int object, int prim_addr, float tmax, uint *lcg_state, int max_hits) { - const int kx = isect_precalc->kx; - const int ky = isect_precalc->ky; - const int kz = isect_precalc->kz; - const float Sx = isect_precalc->Sx; - const float Sy = isect_precalc->Sy; - const float Sz = isect_precalc->Sz; - - /* Calculate vertices relative to ray origin. */ const uint tri_vindex = kernel_tex_fetch(__prim_tri_index, prim_addr); - const float4 tri_a = kernel_tex_fetch(__prim_tri_verts, tri_vindex+0), - tri_b = kernel_tex_fetch(__prim_tri_verts, tri_vindex+1), - tri_c = kernel_tex_fetch(__prim_tri_verts, tri_vindex+2); - -#if defined(__KERNEL_AVX2__) && defined(__KERNEL_SSE__) - const avxf avxf_P(P.m128, P.m128); - - const avxf tri_ab = kernel_tex_fetch_avxf(__prim_tri_verts, tri_vindex + 0); - const avxf tri_bc = kernel_tex_fetch_avxf(__prim_tri_verts, tri_vindex + 1); - - const avxf AB = tri_ab - avxf_P; - const avxf BC = tri_bc - avxf_P; - - const __m256i permuteMask = _mm256_set_epi32(0x3, kz, ky, kx, 0x3, kz, ky, kx); - - const avxf AB_k = shuffle(AB, permuteMask); - const avxf BC_k = shuffle(BC, permuteMask); - - /* Akz, Akz, Bkz, Bkz, Bkz, Bkz, Ckz, Ckz */ - const avxf ABBC_kz = shuffle<2>(AB_k, BC_k); - - /* Akx, Aky, Bkx, Bky, Bkx,Bky, Ckx, Cky */ - const avxf ABBC_kxy = shuffle<0,1,0,1>(AB_k, BC_k); - - const avxf Sxy(Sy, Sx, Sy, Sx); - - /* Ax, Ay, Bx, By, Bx, By, Cx, Cy */ - const avxf ABBC_xy = nmadd(ABBC_kz, Sxy, ABBC_kxy); - - float ABBC_kz_array[8]; - _mm256_storeu_ps((float*)&ABBC_kz_array, ABBC_kz); - - const float A_kz = ABBC_kz_array[0]; - const float B_kz = ABBC_kz_array[2]; - const float C_kz = ABBC_kz_array[6]; - - /* By, Bx, Cy, Cx, By, Bx, Ay, Ax */ - const avxf BCBA_yx = permute<3,2,7,6,3,2,1,0>(ABBC_xy); - - const avxf negMask(0,0,0,0,0x80000000, 0x80000000, 0x80000000, 0x80000000); - - /* W U V - * (AxBy-AyBx) (BxCy-ByCx) XX XX (BxBy-ByBx) (CxAy-CyAx) XX XX - */ - const avxf WUxxxxVxx_neg = _mm256_hsub_ps(ABBC_xy * BCBA_yx, negMask /* Dont care */); - - const avxf WUVWnegWUVW = permute<0,1,5,0,0,1,5,0>(WUxxxxVxx_neg) ^ negMask; - - /* Calculate scaled barycentric coordinates. */ - float WUVW_array[4]; - _mm_storeu_ps((float*)&WUVW_array, _mm256_castps256_ps128 (WUVWnegWUVW)); - - const float W = WUVW_array[0]; - const float U = WUVW_array[1]; - const float V = WUVW_array[2]; - - const int WUVW_mask = 0x7 & _mm256_movemask_ps(WUVWnegWUVW); - const int WUVW_zero = 0x7 & _mm256_movemask_ps(_mm256_cmp_ps(WUVWnegWUVW, - _mm256_setzero_ps(), 0)); - - if(!((WUVW_mask == 7) || (WUVW_mask == 0)) && ((WUVW_mask | WUVW_zero) != 7)) { - return; - } +#if defined(__KERNEL_SSE2__) && defined(__KERNEL_SSE__) + const ssef *ssef_verts = (ssef*)&kg->__prim_tri_verts.data[tri_vindex]; #else - const float3 A = make_float3(tri_a.x - P.x, tri_a.y - P.y, tri_a.z - P.z); - const float3 B = make_float3(tri_b.x - P.x, tri_b.y - P.y, tri_b.z - P.z); - const float3 C = make_float3(tri_c.x - P.x, tri_c.y - P.y, tri_c.z - P.z); - - const float A_kx = IDX(A, kx), A_ky = IDX(A, ky), A_kz = IDX(A, kz); - const float B_kx = IDX(B, kx), B_ky = IDX(B, ky), B_kz = IDX(B, kz); - const float C_kx = IDX(C, kx), C_ky = IDX(C, ky), C_kz = IDX(C, kz); - - /* Perform shear and scale of vertices. */ - const float Ax = A_kx - Sx * A_kz; - const float Ay = A_ky - Sy * A_kz; - const float Bx = B_kx - Sx * B_kz; - const float By = B_ky - Sy * B_kz; - const float Cx = C_kx - Sx * C_kz; - const float Cy = C_ky - Sy * C_kz; - - /* Calculate scaled barycentric coordinates. */ - float U = Cx * By - Cy * Bx; - float V = Ax * Cy - Ay * Cx; - float W = Bx * Ay - By * Ax; - - if((U < 0.0f || V < 0.0f || W < 0.0f) && - (U > 0.0f || V > 0.0f || W > 0.0f)) - { - return; - } + const float3 tri_a = float4_to_float3(kernel_tex_fetch(__prim_tri_verts, tri_vindex+0)), + tri_b = float4_to_float3(kernel_tex_fetch(__prim_tri_verts, tri_vindex+1)), + tri_c = float4_to_float3(kernel_tex_fetch(__prim_tri_verts, tri_vindex+2)); #endif - - /* Calculate determinant. */ - float det = U + V + W; - if(UNLIKELY(det == 0.0f)) { - return; - } - - /* Calculate scaled z−coordinates of vertices and use them to calculate - * the hit distance. - */ - const int sign_det = (__float_as_int(det) & 0x80000000); - const float T = (U * A_kz + V * B_kz + W * C_kz) * Sz; - const float sign_T = xor_signmask(T, sign_det); - if((sign_T < 0.0f) || - (sign_T > tmax * xor_signmask(det, sign_det))) + float t, u, v; + if(!ray_triangle_intersect(P, + dir, + tmax, +#if defined(__KERNEL_SSE2__) && defined(__KERNEL_SSE__) + ssef_verts, +#else + tri_a, tri_b, tri_c, +#endif + &u, &v, &t)) { return; } - /* Normalize U, V, W, and T. */ - const float inv_det = 1.0f / det; - - const float t = T * inv_det; for(int i = min(max_hits, ss_isect->num_hits) - 1; i >= 0; --i) { if(ss_isect->hits[i].t == t) { return; @@ -418,18 +135,19 @@ ccl_device_inline void triangle_intersect_subsurface( isect->prim = prim_addr; isect->object = object; isect->type = PRIMITIVE_TRIANGLE; - isect->u = U * inv_det; - isect->v = V * inv_det; + isect->u = u; + isect->v = v; isect->t = t; /* Record geometric normal. */ - /* TODO(sergey): Use float4_to_float3() on just an edges. */ - const float3 v0 = float4_to_float3(tri_a); - const float3 v1 = float4_to_float3(tri_b); - const float3 v2 = float4_to_float3(tri_c); - ss_isect->Ng[hit] = normalize(cross(v1 - v0, v2 - v0)); -} +#if defined(__KERNEL_SSE2__) && defined(__KERNEL_SSE__) + const float3 tri_a = float4_to_float3(kernel_tex_fetch(__prim_tri_verts, tri_vindex+0)), + tri_b = float4_to_float3(kernel_tex_fetch(__prim_tri_verts, tri_vindex+1)), + tri_c = float4_to_float3(kernel_tex_fetch(__prim_tri_verts, tri_vindex+2)); #endif + ss_isect->Ng[hit] = normalize(cross(tri_b - tri_a, tri_c - tri_a)); +} +#endif /* __SUBSURFACE__ */ /* Refine triangle intersection to more precise hit point. For rays that travel * far the precision is often not so good, this reintersects the primitive from @@ -457,7 +175,7 @@ ccl_device_inline float3 triangle_refine(KernelGlobals *kg, return P; } # ifdef __OBJECT_MOTION__ - Transform tfm = ccl_fetch(sd, ob_itfm); + Transform tfm = sd->ob_itfm; # else Transform tfm = object_fetch_transform(kg, isect->object, OBJECT_INVERSE_TRANSFORM); # endif @@ -491,7 +209,7 @@ ccl_device_inline float3 triangle_refine(KernelGlobals *kg, if(isect->object != OBJECT_NONE) { # ifdef __OBJECT_MOTION__ - Transform tfm = ccl_fetch(sd, ob_tfm); + Transform tfm = sd->ob_tfm; # else Transform tfm = object_fetch_transform(kg, isect->object, OBJECT_TRANSFORM); # endif @@ -519,7 +237,7 @@ ccl_device_inline float3 triangle_refine_subsurface(KernelGlobals *kg, if(isect->object != OBJECT_NONE) { #ifdef __OBJECT_MOTION__ - Transform tfm = ccl_fetch(sd, ob_itfm); + Transform tfm = sd->ob_itfm; #else Transform tfm = object_fetch_transform(kg, isect->object, @@ -557,7 +275,7 @@ ccl_device_inline float3 triangle_refine_subsurface(KernelGlobals *kg, if(isect->object != OBJECT_NONE) { #ifdef __OBJECT_MOTION__ - Transform tfm = ccl_fetch(sd, ob_tfm); + Transform tfm = sd->ob_tfm; #else Transform tfm = object_fetch_transform(kg, isect->object, @@ -570,6 +288,4 @@ ccl_device_inline float3 triangle_refine_subsurface(KernelGlobals *kg, return P; } -#undef IDX - CCL_NAMESPACE_END diff --git a/intern/cycles/kernel/geom/geom_volume.h b/intern/cycles/kernel/geom/geom_volume.h index 03724c955be..1e0ef5201c9 100644 --- a/intern/cycles/kernel/geom/geom_volume.h +++ b/intern/cycles/kernel/geom/geom_volume.h @@ -64,7 +64,7 @@ ccl_device_inline float3 volume_normalized_position(KernelGlobals *kg, ccl_device float volume_attribute_float(KernelGlobals *kg, const ShaderData *sd, const AttributeDescriptor desc, float *dx, float *dy) { - float3 P = volume_normalized_position(kg, sd, ccl_fetch(sd, P)); + float3 P = volume_normalized_position(kg, sd, sd->P); #ifdef __KERNEL_CUDA__ # if __CUDA_ARCH__ >= 300 CUtexObject tex = kernel_tex_fetch(__bindless_mapping, desc.offset); @@ -91,7 +91,7 @@ ccl_device float volume_attribute_float(KernelGlobals *kg, const ShaderData *sd, ccl_device float3 volume_attribute_float3(KernelGlobals *kg, const ShaderData *sd, const AttributeDescriptor desc, float3 *dx, float3 *dy) { - float3 P = volume_normalized_position(kg, sd, ccl_fetch(sd, P)); + float3 P = volume_normalized_position(kg, sd, sd->P); #ifdef __KERNEL_CUDA__ # if __CUDA_ARCH__ >= 300 CUtexObject tex = kernel_tex_fetch(__bindless_mapping, desc.offset); diff --git a/intern/cycles/kernel/kernel.h b/intern/cycles/kernel/kernel.h index 9279a94c13a..06c0fb2fbca 100644 --- a/intern/cycles/kernel/kernel.h +++ b/intern/cycles/kernel/kernel.h @@ -19,7 +19,8 @@ /* CPU Kernel Interface */ -#include "util_types.h" +#include "util/util_types.h" +#include "kernel/kernel_types.h" CCL_NAMESPACE_BEGIN @@ -28,6 +29,7 @@ CCL_NAMESPACE_BEGIN #define KERNEL_FUNCTION_FULL_NAME(name) KERNEL_NAME_EVAL(KERNEL_ARCH, name) struct KernelGlobals; +struct KernelData; KernelGlobals *kernel_globals_create(); void kernel_globals_free(KernelGlobals *kg); @@ -46,31 +48,31 @@ void kernel_tex_copy(KernelGlobals *kg, ExtensionType extension = EXTENSION_REPEAT); #define KERNEL_ARCH cpu -#include "kernels/cpu/kernel_cpu.h" +#include "kernel/kernels/cpu/kernel_cpu.h" #ifdef WITH_CYCLES_OPTIMIZED_KERNEL_SSE2 # define KERNEL_ARCH cpu_sse2 -# include "kernels/cpu/kernel_cpu.h" +# include "kernel/kernels/cpu/kernel_cpu.h" #endif /* WITH_CYCLES_OPTIMIZED_KERNEL_SSE2 */ #ifdef WITH_CYCLES_OPTIMIZED_KERNEL_SSE3 # define KERNEL_ARCH cpu_sse3 -# include "kernels/cpu/kernel_cpu.h" +# include "kernel/kernels/cpu/kernel_cpu.h" #endif /* WITH_CYCLES_OPTIMIZED_KERNEL_SSE2 */ #ifdef WITH_CYCLES_OPTIMIZED_KERNEL_SSE41 # define KERNEL_ARCH cpu_sse41 -# include "kernels/cpu/kernel_cpu.h" +# include "kernel/kernels/cpu/kernel_cpu.h" #endif /* WITH_CYCLES_OPTIMIZED_KERNEL_SSE41 */ #ifdef WITH_CYCLES_OPTIMIZED_KERNEL_AVX # define KERNEL_ARCH cpu_avx -# include "kernels/cpu/kernel_cpu.h" +# include "kernel/kernels/cpu/kernel_cpu.h" #endif /* WITH_CYCLES_OPTIMIZED_KERNEL_AVX */ #ifdef WITH_CYCLES_OPTIMIZED_KERNEL_AVX2 # define KERNEL_ARCH cpu_avx2 -# include "kernels/cpu/kernel_cpu.h" +# include "kernel/kernels/cpu/kernel_cpu.h" #endif /* WITH_CYCLES_OPTIMIZED_KERNEL_AVX2 */ CCL_NAMESPACE_END diff --git a/intern/cycles/kernel/kernel_accumulate.h b/intern/cycles/kernel/kernel_accumulate.h index 6c3ee6b8098..823d30dde78 100644 --- a/intern/cycles/kernel/kernel_accumulate.h +++ b/intern/cycles/kernel/kernel_accumulate.h @@ -52,10 +52,17 @@ ccl_device_inline void bsdf_eval_init(BsdfEval *eval, ClosureType type, float3 v { eval->diffuse = value; } +#ifdef __SHADOW_TRICKS__ + eval->sum_no_mis = make_float3(0.0f, 0.0f, 0.0f); +#endif } -ccl_device_inline void bsdf_eval_accum(BsdfEval *eval, ClosureType type, float3 value) +ccl_device_inline void bsdf_eval_accum(BsdfEval *eval, ClosureType type, float3 value, float mis_weight) { +#ifdef __SHADOW_TRICKS__ + eval->sum_no_mis += value; +#endif + value *= mis_weight; #ifdef __PASSES__ if(eval->use_light_pass) { if(CLOSURE_IS_BSDF_DIFFUSE(type)) @@ -96,7 +103,7 @@ ccl_device_inline bool bsdf_eval_is_zero(BsdfEval *eval) } } -ccl_device_inline void bsdf_eval_mul(BsdfEval *eval, float value) +ccl_device_inline void bsdf_eval_mis(BsdfEval *eval, float value) { #ifdef __PASSES__ if(eval->use_light_pass) { @@ -115,8 +122,19 @@ ccl_device_inline void bsdf_eval_mul(BsdfEval *eval, float value) } } +ccl_device_inline void bsdf_eval_mul(BsdfEval *eval, float value) +{ +#ifdef __SHADOW_TRICKS__ + eval->sum_no_mis *= value; +#endif + bsdf_eval_mis(eval, value); +} + ccl_device_inline void bsdf_eval_mul3(BsdfEval *eval, float3 value) { +#ifdef __SHADOW_TRICKS__ + eval->sum_no_mis *= value; +#endif #ifdef __PASSES__ if(eval->use_light_pass) { eval->diffuse *= value; @@ -134,7 +152,7 @@ ccl_device_inline void bsdf_eval_mul3(BsdfEval *eval, float3 value) #endif } -ccl_device_inline float3 bsdf_eval_sum(BsdfEval *eval) +ccl_device_inline float3 bsdf_eval_sum(const BsdfEval *eval) { #ifdef __PASSES__ if(eval->use_light_pass) { @@ -198,6 +216,12 @@ ccl_device_inline void path_radiance_init(PathRadiance *L, int use_light_pass) { L->emission = make_float3(0.0f, 0.0f, 0.0f); } + +#ifdef __SHADOW_TRICKS__ + L->path_total = make_float3(0.0f, 0.0f, 0.0f); + L->path_total_shaded = make_float3(0.0f, 0.0f, 0.0f); + L->shadow_color = make_float3(0.0f, 0.0f, 0.0f); +#endif } ccl_device_inline void path_radiance_bsdf_bounce(PathRadiance *L, ccl_addr_space float3 *throughput, @@ -252,7 +276,12 @@ ccl_device_inline void path_radiance_accum_emission(PathRadiance *L, float3 thro } } -ccl_device_inline void path_radiance_accum_ao(PathRadiance *L, float3 throughput, float3 alpha, float3 bsdf, float3 ao, int bounce) +ccl_device_inline void path_radiance_accum_ao(PathRadiance *L, + float3 throughput, + float3 alpha, + float3 bsdf, + float3 ao, + int bounce) { #ifdef __PASSES__ if(L->use_light_pass) { @@ -271,6 +300,26 @@ ccl_device_inline void path_radiance_accum_ao(PathRadiance *L, float3 throughput { L->emission += throughput*bsdf*ao; } + +#ifdef __SHADOW_TRICKS__ + float3 light = throughput * bsdf; + L->path_total += light; + L->path_total_shaded += ao * light; +#endif +} + +ccl_device_inline void path_radiance_accum_total_ao( + PathRadiance *L, + float3 throughput, + float3 bsdf) +{ +#ifdef __SHADOW_TRICKS__ + L->path_total += throughput * bsdf; +#else + (void) L; + (void) throughput; + (void) bsdf; +#endif } ccl_device_inline void path_radiance_accum_light(PathRadiance *L, float3 throughput, BsdfEval *bsdf_eval, float3 shadow, float shadow_fac, int bounce, bool is_lamp) @@ -301,15 +350,38 @@ ccl_device_inline void path_radiance_accum_light(PathRadiance *L, float3 through { L->emission += throughput*bsdf_eval->diffuse*shadow; } + +#ifdef __SHADOW_TRICKS__ + float3 light = throughput * bsdf_eval->sum_no_mis; + L->path_total += light; + L->path_total_shaded += shadow * light; +#endif +} + +ccl_device_inline void path_radiance_accum_total_light( + PathRadiance *L, + float3 throughput, + const BsdfEval *bsdf_eval) +{ +#ifdef __SHADOW_TRICKS__ + L->path_total += throughput * bsdf_eval->sum_no_mis; +#else + (void) L; + (void) throughput; + (void) bsdf_eval; +#endif } -ccl_device_inline void path_radiance_accum_background(PathRadiance *L, float3 throughput, float3 value, int bounce) +ccl_device_inline void path_radiance_accum_background(PathRadiance *L, + ccl_addr_space PathState *state, + float3 throughput, + float3 value) { #ifdef __PASSES__ if(L->use_light_pass) { - if(bounce == 0) + if(state->bounce == 0) L->background += throughput*value; - else if(bounce == 1) + else if(state->bounce == 1) L->direct_emission += throughput*value; else L->indirect += throughput*value; @@ -319,6 +391,13 @@ ccl_device_inline void path_radiance_accum_background(PathRadiance *L, float3 th { L->emission += throughput*value; } + +#ifdef __SHADOW_TRICKS__ + L->path_total += throughput * value; + if(state->flag & PATH_RAY_SHADOW_CATCHER_ONLY) { + L->path_total_shaded += throughput * value; + } +#endif } ccl_device_inline void path_radiance_sum_indirect(PathRadiance *L) @@ -399,7 +478,7 @@ ccl_device_inline float3 path_radiance_clamp_and_sum(KernelGlobals *kg, PathRadi float sum = fabsf((L_sum).x) + fabsf((L_sum).y) + fabsf((L_sum).z); /* Reject invalid value */ - if(!isfinite(sum)) { + if(!isfinite_safe(sum)) { kernel_assert(!"Non-finite sum in path_radiance_clamp_and_sum!"); L_sum = make_float3(0.0f, 0.0f, 0.0f); @@ -468,7 +547,7 @@ ccl_device_inline float3 path_radiance_clamp_and_sum(KernelGlobals *kg, PathRadi /* Reject invalid value */ float sum = fabsf((L_sum).x) + fabsf((L_sum).y) + fabsf((L_sum).z); - if(!isfinite(sum)) { + if(!isfinite_safe(sum)) { kernel_assert(!"Non-finite final sum in path_radiance_clamp_and_sum!"); L_sum = make_float3(0.0f, 0.0f, 0.0f); } @@ -501,5 +580,34 @@ ccl_device_inline void path_radiance_accum_sample(PathRadiance *L, PathRadiance L->emission += L_sample->emission * fac; } -CCL_NAMESPACE_END +#ifdef __SHADOW_TRICKS__ +/* Calculate current shadow of the path. */ +ccl_device_inline float path_radiance_sum_shadow(const PathRadiance *L) +{ + float path_total = average(L->path_total); + float path_total_shaded = average(L->path_total_shaded); + if(path_total != 0.0f) { + return path_total_shaded / path_total; + } + return 1.0f; +} +/* Calculate final light sum and transparency for shadow catcher object. */ +ccl_device_inline float3 path_radiance_sum_shadowcatcher(KernelGlobals *kg, + const PathRadiance *L, + ccl_addr_space float* L_transparent) +{ + const float shadow = path_radiance_sum_shadow(L); + float3 L_sum; + if(kernel_data.background.transparent) { + *L_transparent = shadow; + L_sum = make_float3(0.0f, 0.0f, 0.0f); + } + else { + L_sum = L->shadow_color * shadow; + } + return L_sum; +} +#endif + +CCL_NAMESPACE_END diff --git a/intern/cycles/kernel/kernel_bake.h b/intern/cycles/kernel/kernel_bake.h index 5bcc57cdcdf..f18d145f7cf 100644 --- a/intern/cycles/kernel/kernel_bake.h +++ b/intern/cycles/kernel/kernel_bake.h @@ -54,7 +54,8 @@ ccl_device_inline void compute_light_pass(KernelGlobals *kg, float rbsdf = path_state_rng_1D(kg, &rng, &state, PRNG_BSDF); shader_eval_surface(kg, sd, &rng, &state, rbsdf, state.flag, SHADER_CONTEXT_MAIN); - /* TODO, disable the closures we won't need */ + /* TODO, disable more closures we don't need besides transparent */ + shader_bsdf_disable_transparency(kg, sd); #ifdef __BRANCHED_PATH__ if(!kernel_data.integrator.branched) { diff --git a/intern/cycles/kernel/kernel_camera.h b/intern/cycles/kernel/kernel_camera.h index dedac6b1465..0df5217d97a 100644 --- a/intern/cycles/kernel/kernel_camera.h +++ b/intern/cycles/kernel/kernel_camera.h @@ -457,7 +457,7 @@ ccl_device_inline float3 camera_world_to_ndc(KernelGlobals *kg, ShaderData *sd, { if(kernel_data.cam.type != CAMERA_PANORAMA) { /* perspective / ortho */ - if(ccl_fetch(sd, object) == PRIM_NONE && kernel_data.cam.type == CAMERA_PERSPECTIVE) + if(sd->object == PRIM_NONE && kernel_data.cam.type == CAMERA_PERSPECTIVE) P += camera_position(kg); Transform tfm = kernel_data.cam.worldtondc; @@ -467,7 +467,7 @@ ccl_device_inline float3 camera_world_to_ndc(KernelGlobals *kg, ShaderData *sd, /* panorama */ Transform tfm = kernel_data.cam.worldtocamera; - if(ccl_fetch(sd, object) != OBJECT_NONE) + if(sd->object != OBJECT_NONE) P = normalize(transform_point(&tfm, P)); else P = normalize(transform_direction(&tfm, P)); diff --git a/intern/cycles/kernel/kernel_compat_cpu.h b/intern/cycles/kernel/kernel_compat_cpu.h index 9d1f3bdc918..ae7c9b836c4 100644 --- a/intern/cycles/kernel/kernel_compat_cpu.h +++ b/intern/cycles/kernel/kernel_compat_cpu.h @@ -35,15 +35,24 @@ # define __NODES_FEATURES__ NODE_FEATURE_ALL #endif -#include "util_debug.h" -#include "util_math.h" -#include "util_simd.h" -#include "util_half.h" -#include "util_types.h" -#include "util_texture.h" +#include "util/util_debug.h" +#include "util/util_math.h" +#include "util/util_simd.h" +#include "util/util_half.h" +#include "util/util_types.h" +#include "util/util_texture.h" #define ccl_addr_space +#define ccl_local_id(d) 0 +#define ccl_global_id(d) (kg->global_id[d]) + +#define ccl_local_size(d) 1 +#define ccl_global_size(d) (kg->global_size[d]) + +#define ccl_group_id(d) ccl_global_id(d) +#define ccl_num_groups(d) ccl_global_size(d) + /* On x86_64, versions of glibc < 2.16 have an issue where expf is * much slower than the double version. This was fixed in glibc 2.16. */ @@ -78,9 +87,9 @@ template<typename T> struct texture { ccl_always_inline avxf fetch_avxf(const int index) { kernel_assert(index >= 0 && (index+1) < width); - ssef *ssefData = (ssef*)data; - ssef *ssefNodeData = &ssefData[index]; - return _mm256_loadu_ps((float *)ssefNodeData); + ssef *ssef_data = (ssef*)data; + ssef *ssef_node_data = &ssef_data[index]; + return _mm256_loadu_ps((float *)ssef_node_data); } #endif diff --git a/intern/cycles/kernel/kernel_compat_cuda.h b/intern/cycles/kernel/kernel_compat_cuda.h index e0c7b17c6a0..39e98c7dda6 100644 --- a/intern/cycles/kernel/kernel_compat_cuda.h +++ b/intern/cycles/kernel/kernel_compat_cuda.h @@ -38,7 +38,7 @@ #define ccl_device __device__ __inline__ # define ccl_device_forceinline __device__ __forceinline__ -#if (__KERNEL_CUDA_VERSION__ == 80) && (__CUDA_ARCH__ < 500) +#if __CUDA_ARCH__ < 500 # define ccl_device_inline __device__ __forceinline__ #else # define ccl_device_inline __device__ __inline__ @@ -46,6 +46,9 @@ #define ccl_device_noinline __device__ __noinline__ #define ccl_global #define ccl_constant +#define ccl_local __shared__ +#define ccl_local_param +#define ccl_private #define ccl_may_alias #define ccl_addr_space #define ccl_restrict __restrict__ @@ -57,8 +60,54 @@ /* Types */ -#include "util_half.h" -#include "util_types.h" +#include "util/util_half.h" +#include "util/util_types.h" + +/* Work item functions */ + +ccl_device_inline uint ccl_local_id(uint d) +{ + switch(d) { + case 0: return threadIdx.x; + case 1: return threadIdx.y; + case 2: return threadIdx.z; + default: return 0; + } +} + +#define ccl_global_id(d) (ccl_group_id(d) * ccl_local_size(d) + ccl_local_id(d)) + +ccl_device_inline uint ccl_local_size(uint d) +{ + switch(d) { + case 0: return blockDim.x; + case 1: return blockDim.y; + case 2: return blockDim.z; + default: return 0; + } +} + +#define ccl_global_size(d) (ccl_num_groups(d) * ccl_local_size(d)) + +ccl_device_inline uint ccl_group_id(uint d) +{ + switch(d) { + case 0: return blockIdx.x; + case 1: return blockIdx.y; + case 2: return blockIdx.z; + default: return 0; + } +} + +ccl_device_inline uint ccl_num_groups(uint d) +{ + switch(d) { + case 0: return gridDim.x; + case 1: return gridDim.y; + case 2: return gridDim.z; + default: return 0; + } +} /* Textures */ diff --git a/intern/cycles/kernel/kernel_compat_opencl.h b/intern/cycles/kernel/kernel_compat_opencl.h index f076e3a7d37..c2263ac0d49 100644 --- a/intern/cycles/kernel/kernel_compat_opencl.h +++ b/intern/cycles/kernel/kernel_compat_opencl.h @@ -39,6 +39,7 @@ #define ccl_constant __constant #define ccl_global __global #define ccl_local __local +#define ccl_local_param __local #define ccl_private __private #define ccl_restrict restrict #define ccl_align(n) __attribute__((aligned(n))) @@ -49,6 +50,15 @@ # define ccl_addr_space #endif +#define ccl_local_id(d) get_local_id(d) +#define ccl_global_id(d) get_global_id(d) + +#define ccl_local_size(d) get_local_size(d) +#define ccl_global_size(d) get_global_size(d) + +#define ccl_group_id(d) get_group_id(d) +#define ccl_num_groups(d) get_num_groups(d) + /* Selective nodes compilation. */ #ifndef __NODES_MAX_GROUP__ # define __NODES_MAX_GROUP__ NODE_GROUP_LEVEL_MAX @@ -133,8 +143,8 @@ /* define NULL */ #define NULL 0 -#include "util_half.h" -#include "util_types.h" +#include "util/util_half.h" +#include "util/util_types.h" #endif /* __KERNEL_COMPAT_OPENCL_H__ */ diff --git a/intern/cycles/kernel/kernel_emission.h b/intern/cycles/kernel/kernel_emission.h index 8c7c651a053..9e7d51f23f5 100644 --- a/intern/cycles/kernel/kernel_emission.h +++ b/intern/cycles/kernel/kernel_emission.h @@ -67,7 +67,7 @@ ccl_device_noinline float3 direct_emissive_eval(KernelGlobals *kg, ls->shader, ls->object, ls->prim, ls->u, ls->v, t, time, false, ls->lamp); - ls->Ng = ccl_fetch(emission_sd, Ng); + ls->Ng = emission_sd->Ng; /* no path flag, we're evaluating this for all closures. that's weak but * we'd have to do multiple evaluations otherwise */ @@ -76,7 +76,7 @@ ccl_device_noinline float3 direct_emissive_eval(KernelGlobals *kg, path_state_modify_bounce(state, false); /* evaluate emissive closure */ - if(ccl_fetch(emission_sd, flag) & SD_EMISSION) + if(emission_sd->flag & SD_EMISSION) eval = shader_emissive_eval(kg, emission_sd); else eval = make_float3(0.0f, 0.0f, 0.0f); @@ -112,7 +112,7 @@ ccl_device_noinline bool direct_emission(KernelGlobals *kg, -ls->D, dD, ls->t, - ccl_fetch(sd, time)); + sd->time); if(is_zero(light_eval)) return false; @@ -120,7 +120,7 @@ ccl_device_noinline bool direct_emission(KernelGlobals *kg, /* evaluate BSDF at shading point */ #ifdef __VOLUME__ - if(ccl_fetch(sd, prim) != PRIM_NONE) + if(sd->prim != PRIM_NONE) shader_bsdf_eval(kg, sd, ls->D, eval, ls->pdf, ls->shader & SHADER_USE_MIS); else { float bsdf_pdf; @@ -156,8 +156,13 @@ ccl_device_noinline bool direct_emission(KernelGlobals *kg, if(bsdf_eval_is_zero(eval)) return false; - if(kernel_data.integrator.light_inv_rr_threshold > 0.0f) { - float probability = max3(bsdf_eval_sum(eval)) * kernel_data.integrator.light_inv_rr_threshold; + if(kernel_data.integrator.light_inv_rr_threshold > 0.0f +#ifdef __SHADOW_TRICKS__ + && (state->flag & PATH_RAY_SHADOW_CATCHER) == 0 +#endif + ) + { + float probability = max3(fabs(bsdf_eval_sum(eval))) * kernel_data.integrator.light_inv_rr_threshold; if(probability < 1.0f) { if(rand_terminate >= probability) { return false; @@ -168,8 +173,8 @@ ccl_device_noinline bool direct_emission(KernelGlobals *kg, if(ls->shader & SHADER_CAST_SHADOW) { /* setup ray */ - bool transmit = (dot(ccl_fetch(sd, Ng), ls->D) < 0.0f); - ray->P = ray_offset(ccl_fetch(sd, P), (transmit)? -ccl_fetch(sd, Ng): ccl_fetch(sd, Ng)); + bool transmit = (dot(sd->Ng, ls->D) < 0.0f); + ray->P = ray_offset(sd->P, (transmit)? -sd->Ng: sd->Ng); if(ls->t == FLT_MAX) { /* distant light */ @@ -182,7 +187,7 @@ ccl_device_noinline bool direct_emission(KernelGlobals *kg, ray->D = normalize_len(ray->D, &ray->t); } - ray->dP = ccl_fetch(sd, dP); + ray->dP = sd->dP; ray->dD = differential3_zero(); } else { @@ -204,14 +209,14 @@ ccl_device_noinline float3 indirect_primitive_emission(KernelGlobals *kg, Shader float3 L = shader_emissive_eval(kg, sd); #ifdef __HAIR__ - if(!(path_flag & PATH_RAY_MIS_SKIP) && (ccl_fetch(sd, flag) & SD_USE_MIS) && (ccl_fetch(sd, type) & PRIMITIVE_ALL_TRIANGLE)) + if(!(path_flag & PATH_RAY_MIS_SKIP) && (sd->flag & SD_USE_MIS) && (sd->type & PRIMITIVE_ALL_TRIANGLE)) #else - if(!(path_flag & PATH_RAY_MIS_SKIP) && (ccl_fetch(sd, flag) & SD_USE_MIS)) + if(!(path_flag & PATH_RAY_MIS_SKIP) && (sd->flag & SD_USE_MIS)) #endif { /* multiple importance sampling, get triangle light pdf, * and compute weight with respect to BSDF pdf */ - float pdf = triangle_light_pdf(kg, ccl_fetch(sd, Ng), ccl_fetch(sd, I), t); + float pdf = triangle_light_pdf(kg, sd->Ng, sd->I, t); float mis_weight = power_heuristic(bsdf_pdf, pdf); return L*mis_weight; diff --git a/intern/cycles/kernel/kernel_globals.h b/intern/cycles/kernel/kernel_globals.h index 8e66a3a0340..c9c97ea977e 100644 --- a/intern/cycles/kernel/kernel_globals.h +++ b/intern/cycles/kernel/kernel_globals.h @@ -16,6 +16,9 @@ /* Constant Globals */ +#ifndef __KERNEL_GLOBALS_H__ +#define __KERNEL_GLOBALS_H__ + CCL_NAMESPACE_BEGIN /* On the CPU, we pass along the struct KernelGlobals to nearly everywhere in @@ -44,7 +47,7 @@ typedef struct KernelGlobals { # define KERNEL_TEX(type, ttype, name) ttype name; # define KERNEL_IMAGE_TEX(type, ttype, name) -# include "kernel_textures.h" +# include "kernel/kernel_textures.h" KernelData __data; @@ -64,6 +67,13 @@ typedef struct KernelGlobals { /* Storage for decoupled volume steps. */ VolumeStep *decoupled_volume_steps[2]; int decoupled_volume_steps_index; + + /* split kernel */ + SplitData split_data; + SplitParams split_param_data; + + int2 global_size; + int2 global_id; } KernelGlobals; #endif /* __KERNEL_CPU__ */ @@ -76,7 +86,10 @@ typedef struct KernelGlobals { #ifdef __KERNEL_CUDA__ __constant__ KernelData __data; -typedef struct KernelGlobals {} KernelGlobals; +typedef struct KernelGlobals { + /* NOTE: Keep the size in sync with SHADOW_STACK_MAX_HITS. */ + Intersection hits_stack[64]; +} KernelGlobals; # ifdef __KERNEL_CUDA_TEX_STORAGE__ # define KERNEL_TEX(type, ttype, name) ttype name; @@ -84,7 +97,7 @@ typedef struct KernelGlobals {} KernelGlobals; # define KERNEL_TEX(type, ttype, name) const __constant__ __device__ type *name; # endif # define KERNEL_IMAGE_TEX(type, ttype, name) ttype name; -# include "kernel_textures.h" +# include "kernel/kernel_textures.h" #endif /* __KERNEL_CUDA__ */ @@ -97,11 +110,11 @@ typedef ccl_addr_space struct KernelGlobals { # define KERNEL_TEX(type, ttype, name) \ ccl_global type *name; -# include "kernel_textures.h" +# include "kernel/kernel_textures.h" # ifdef __SPLIT_KERNEL__ - ShaderData *sd_input; - Intersection *isect_shadow; + SplitData split_data; + SplitParams split_param_data; # endif } KernelGlobals; @@ -143,3 +156,4 @@ ccl_device float lookup_table_read_2D(KernelGlobals *kg, float x, float y, int o CCL_NAMESPACE_END +#endif /* __KERNEL_GLOBALS_H__ */ diff --git a/intern/cycles/kernel/kernel_math.h b/intern/cycles/kernel/kernel_math.h index 9bee5603474..bd0e23b7705 100644 --- a/intern/cycles/kernel/kernel_math.h +++ b/intern/cycles/kernel/kernel_math.h @@ -17,11 +17,11 @@ #ifndef __KERNEL_MATH_H__ #define __KERNEL_MATH_H__ -#include "util_color.h" -#include "util_math.h" -#include "util_math_fast.h" -#include "util_texture.h" -#include "util_transform.h" +#include "util/util_color.h" +#include "util/util_math.h" +#include "util/util_math_fast.h" +#include "util/util_math_intersect.h" +#include "util/util_texture.h" +#include "util/util_transform.h" #endif /* __KERNEL_MATH_H__ */ - diff --git a/intern/cycles/kernel/kernel_passes.h b/intern/cycles/kernel/kernel_passes.h index 7aec47e4957..ed523696571 100644 --- a/intern/cycles/kernel/kernel_passes.h +++ b/intern/cycles/kernel/kernel_passes.h @@ -19,16 +19,16 @@ CCL_NAMESPACE_BEGIN ccl_device_inline void kernel_write_pass_float(ccl_global float *buffer, int sample, float value) { ccl_global float *buf = buffer; -#if defined(__SPLIT_KERNEL__) && defined(__WORK_STEALING__) +#if defined(__SPLIT_KERNEL__) atomic_add_and_fetch_float(buf, value); #else *buf = (sample == 0)? value: *buf + value; -#endif // __SPLIT_KERNEL__ && __WORK_STEALING__ +#endif /* __SPLIT_KERNEL__ */ } ccl_device_inline void kernel_write_pass_float3(ccl_global float *buffer, int sample, float3 value) { -#if defined(__SPLIT_KERNEL__) && defined(__WORK_STEALING__) +#if defined(__SPLIT_KERNEL__) ccl_global float *buf_x = buffer + 0; ccl_global float *buf_y = buffer + 1; ccl_global float *buf_z = buffer + 2; @@ -39,12 +39,12 @@ ccl_device_inline void kernel_write_pass_float3(ccl_global float *buffer, int sa #else ccl_global float3 *buf = (ccl_global float3*)buffer; *buf = (sample == 0)? value: *buf + value; -#endif // __SPLIT_KERNEL__ && __WORK_STEALING__ +#endif /* __SPLIT_KERNEL__ */ } ccl_device_inline void kernel_write_pass_float4(ccl_global float *buffer, int sample, float4 value) { -#if defined(__SPLIT_KERNEL__) && defined(__WORK_STEALING__) +#if defined(__SPLIT_KERNEL__) ccl_global float *buf_x = buffer + 0; ccl_global float *buf_y = buffer + 1; ccl_global float *buf_z = buffer + 2; @@ -57,7 +57,7 @@ ccl_device_inline void kernel_write_pass_float4(ccl_global float *buffer, int sa #else ccl_global float4 *buf = (ccl_global float4*)buffer; *buf = (sample == 0)? value: *buf + value; -#endif // __SPLIT_KERNEL__ && __WORK_STEALING__ +#endif /* __SPLIT_KERNEL__ */ } ccl_device_inline void kernel_write_data_passes(KernelGlobals *kg, ccl_global float *buffer, PathRadiance *L, @@ -75,18 +75,18 @@ ccl_device_inline void kernel_write_data_passes(KernelGlobals *kg, ccl_global fl return; if(!(path_flag & PATH_RAY_SINGLE_PASS_DONE)) { - if(!(ccl_fetch(sd, flag) & SD_TRANSPARENT) || + if(!(sd->flag & SD_TRANSPARENT) || kernel_data.film.pass_alpha_threshold == 0.0f || average(shader_bsdf_alpha(kg, sd)) >= kernel_data.film.pass_alpha_threshold) { if(sample == 0) { if(flag & PASS_DEPTH) { - float depth = camera_distance(kg, ccl_fetch(sd, P)); + float depth = camera_distance(kg, sd->P); kernel_write_pass_float(buffer + kernel_data.film.pass_depth, sample, depth); } if(flag & PASS_OBJECT_ID) { - float id = object_pass_id(kg, ccl_fetch(sd, object)); + float id = object_pass_id(kg, sd->object); kernel_write_pass_float(buffer + kernel_data.film.pass_object_id, sample, id); } if(flag & PASS_MATERIAL_ID) { @@ -96,7 +96,7 @@ ccl_device_inline void kernel_write_data_passes(KernelGlobals *kg, ccl_global fl } if(flag & PASS_NORMAL) { - float3 normal = ccl_fetch(sd, N); + float3 normal = sd->N; kernel_write_pass_float3(buffer + kernel_data.film.pass_normal, sample, normal); } if(flag & PASS_UV) { @@ -127,7 +127,7 @@ ccl_device_inline void kernel_write_data_passes(KernelGlobals *kg, ccl_global fl float mist_start = kernel_data.film.mist_start; float mist_inv_depth = kernel_data.film.mist_inv_depth; - float depth = camera_distance(kg, ccl_fetch(sd, P)); + float depth = camera_distance(kg, sd->P); float mist = saturate((depth - mist_start)*mist_inv_depth); /* falloff */ diff --git a/intern/cycles/kernel/kernel_path.h b/intern/cycles/kernel/kernel_path.h index f90701a8260..e7957042182 100644 --- a/intern/cycles/kernel/kernel_path.h +++ b/intern/cycles/kernel/kernel_path.h @@ -15,40 +15,41 @@ */ #ifdef __OSL__ -# include "osl_shader.h" +# include "kernel/osl/osl_shader.h" #endif -#include "kernel_random.h" -#include "kernel_projection.h" -#include "kernel_montecarlo.h" -#include "kernel_differential.h" -#include "kernel_camera.h" +#include "kernel/kernel_random.h" +#include "kernel/kernel_projection.h" +#include "kernel/kernel_montecarlo.h" +#include "kernel/kernel_differential.h" +#include "kernel/kernel_camera.h" -#include "geom/geom.h" -#include "bvh/bvh.h" +#include "kernel/geom/geom.h" +#include "kernel/bvh/bvh.h" -#include "kernel_accumulate.h" -#include "kernel_shader.h" -#include "kernel_light.h" -#include "kernel_passes.h" +#include "kernel/kernel_accumulate.h" +#include "kernel/kernel_shader.h" +#include "kernel/kernel_light.h" +#include "kernel/kernel_passes.h" #ifdef __SUBSURFACE__ -# include "kernel_subsurface.h" +# include "kernel/kernel_subsurface.h" #endif #ifdef __VOLUME__ -# include "kernel_volume.h" +# include "kernel/kernel_volume.h" #endif -#include "kernel_path_state.h" -#include "kernel_shadow.h" -#include "kernel_emission.h" -#include "kernel_path_common.h" -#include "kernel_path_surface.h" -#include "kernel_path_volume.h" +#include "kernel/kernel_path_state.h" +#include "kernel/kernel_shadow.h" +#include "kernel/kernel_emission.h" +#include "kernel/kernel_path_common.h" +#include "kernel/kernel_path_surface.h" +#include "kernel/kernel_path_volume.h" +#include "kernel/kernel_path_subsurface.h" #ifdef __KERNEL_DEBUG__ -# include "kernel_debug.h" +# include "kernel/kernel_debug.h" #endif CCL_NAMESPACE_BEGIN @@ -75,22 +76,25 @@ ccl_device_noinline void kernel_path_ao(KernelGlobals *kg, sample_cos_hemisphere(ao_N, bsdf_u, bsdf_v, &ao_D, &ao_pdf); - if(dot(ccl_fetch(sd, Ng), ao_D) > 0.0f && ao_pdf != 0.0f) { + if(dot(sd->Ng, ao_D) > 0.0f && ao_pdf != 0.0f) { Ray light_ray; float3 ao_shadow; - light_ray.P = ray_offset(ccl_fetch(sd, P), ccl_fetch(sd, Ng)); + light_ray.P = ray_offset(sd->P, sd->Ng); light_ray.D = ao_D; light_ray.t = kernel_data.background.ao_distance; #ifdef __OBJECT_MOTION__ - light_ray.time = ccl_fetch(sd, time); + light_ray.time = sd->time; #endif /* __OBJECT_MOTION__ */ - light_ray.dP = ccl_fetch(sd, dP); + light_ray.dP = sd->dP; light_ray.dD = differential3_zero(); if(!shadow_blocked(kg, emission_sd, state, &light_ray, &ao_shadow)) { path_radiance_accum_ao(L, throughput, ao_alpha, ao_bsdf, ao_shadow, state->bounce); } + else { + path_radiance_accum_total_ao(L, throughput, ao_bsdf); + } } } @@ -289,9 +293,9 @@ ccl_device void kernel_path_indirect(KernelGlobals *kg, /* sample background shader */ float3 L_background = indirect_background(kg, emission_sd, state, ray); path_radiance_accum_background(L, + state, throughput, - L_background, - state->bounce); + L_background); #endif /* __BACKGROUND__ */ break; @@ -311,6 +315,12 @@ ccl_device void kernel_path_indirect(KernelGlobals *kg, shader_merge_closures(sd); #endif /* __BRANCHED_PATH__ */ +#ifdef __SHADOW_TRICKS__ + if(!(sd->object_flag & SD_OBJECT_SHADOW_CATCHER)) { + state->flag &= ~PATH_RAY_SHADOW_CATCHER_ONLY; + } +#endif /* __SHADOW_TRICKS__ */ + /* blurring of bsdf after bounces, for rays that have a small likelihood * of following this particular path (diffuse, rough glossy) */ if(kernel_data.integrator.filter_glossy != FLT_MAX) { @@ -373,7 +383,7 @@ ccl_device void kernel_path_indirect(KernelGlobals *kg, /* do bssrdf scatter step if we picked a bssrdf closure */ if(sc) { - uint lcg_state = lcg_state_init(rng, state, 0x68bc21eb); + uint lcg_state = lcg_state_init(rng, state->rng_offset, state->sample, 0x68bc21eb); float bssrdf_u, bssrdf_v; path_state_rng_2D(kg, @@ -395,7 +405,8 @@ ccl_device void kernel_path_indirect(KernelGlobals *kg, #if defined(__EMISSION__) && defined(__BRANCHED_PATH__) if(kernel_data.integrator.use_direct_light) { - int all = kernel_data.integrator.sample_all_lights_indirect; + int all = (kernel_data.integrator.sample_all_lights_indirect) || + (state->flag & PATH_RAY_SHADOW_CATCHER); kernel_branched_path_surface_connect_light(kg, rng, sd, @@ -413,172 +424,6 @@ ccl_device void kernel_path_indirect(KernelGlobals *kg, } } -#ifdef __SUBSURFACE__ -# ifndef __KERNEL_CUDA__ -ccl_device -# else -ccl_device_inline -# endif -bool kernel_path_subsurface_scatter( - KernelGlobals *kg, - ShaderData *sd, - ShaderData *emission_sd, - PathRadiance *L, - PathState *state, - RNG *rng, - Ray *ray, - float3 *throughput, - SubsurfaceIndirectRays *ss_indirect) -{ - float bssrdf_probability; - ShaderClosure *sc = subsurface_scatter_pick_closure(kg, sd, &bssrdf_probability); - - /* modify throughput for picking bssrdf or bsdf */ - *throughput *= bssrdf_probability; - - /* do bssrdf scatter step if we picked a bssrdf closure */ - if(sc) { - /* We should never have two consecutive BSSRDF bounces, - * the second one should be converted to a diffuse BSDF to - * avoid this. - */ - kernel_assert(!ss_indirect->tracing); - - uint lcg_state = lcg_state_init(rng, state, 0x68bc21eb); - - SubsurfaceIntersection ss_isect; - float bssrdf_u, bssrdf_v; - path_state_rng_2D(kg, rng, state, PRNG_BSDF_U, &bssrdf_u, &bssrdf_v); - int num_hits = subsurface_scatter_multi_intersect(kg, - &ss_isect, - sd, - sc, - &lcg_state, - bssrdf_u, bssrdf_v, - false); -# ifdef __VOLUME__ - ss_indirect->need_update_volume_stack = - kernel_data.integrator.use_volumes && - ccl_fetch(sd, object_flag) & SD_OBJECT_INTERSECTS_VOLUME; -# endif /* __VOLUME__ */ - - /* compute lighting with the BSDF closure */ - for(int hit = 0; hit < num_hits; hit++) { - /* NOTE: We reuse the existing ShaderData, we assume the path - * integration loop stops when this function returns true. - */ - subsurface_scatter_multi_setup(kg, - &ss_isect, - hit, - sd, - state, - state->flag, - sc, - false); - - PathState *hit_state = &ss_indirect->state[ss_indirect->num_rays]; - Ray *hit_ray = &ss_indirect->rays[ss_indirect->num_rays]; - float3 *hit_tp = &ss_indirect->throughputs[ss_indirect->num_rays]; - PathRadiance *hit_L = &ss_indirect->L[ss_indirect->num_rays]; - - *hit_state = *state; - *hit_ray = *ray; - *hit_tp = *throughput; - - hit_state->rng_offset += PRNG_BOUNCE_NUM; - - path_radiance_init(hit_L, kernel_data.film.use_light_pass); - hit_L->direct_throughput = L->direct_throughput; - path_radiance_copy_indirect(hit_L, L); - - kernel_path_surface_connect_light(kg, rng, sd, emission_sd, *hit_tp, state, hit_L); - - if(kernel_path_surface_bounce(kg, - rng, - sd, - hit_tp, - hit_state, - hit_L, - hit_ray)) - { -# ifdef __LAMP_MIS__ - hit_state->ray_t = 0.0f; -# endif /* __LAMP_MIS__ */ - -# ifdef __VOLUME__ - if(ss_indirect->need_update_volume_stack) { - Ray volume_ray = *ray; - /* Setup ray from previous surface point to the new one. */ - volume_ray.D = normalize_len(hit_ray->P - volume_ray.P, - &volume_ray.t); - - kernel_volume_stack_update_for_subsurface( - kg, - emission_sd, - &volume_ray, - hit_state->volume_stack); - } -# endif /* __VOLUME__ */ - path_radiance_reset_indirect(L); - ss_indirect->num_rays++; - } - else { - path_radiance_accum_sample(L, hit_L, 1); - } - } - return true; - } - return false; -} - -ccl_device_inline void kernel_path_subsurface_init_indirect( - SubsurfaceIndirectRays *ss_indirect) -{ - ss_indirect->tracing = false; - ss_indirect->num_rays = 0; -} - -ccl_device void kernel_path_subsurface_accum_indirect( - SubsurfaceIndirectRays *ss_indirect, - PathRadiance *L) -{ - if(ss_indirect->tracing) { - path_radiance_sum_indirect(L); - path_radiance_accum_sample(&ss_indirect->direct_L, L, 1); - if(ss_indirect->num_rays == 0) { - *L = ss_indirect->direct_L; - } - } -} - -ccl_device void kernel_path_subsurface_setup_indirect( - KernelGlobals *kg, - SubsurfaceIndirectRays *ss_indirect, - PathState *state, - Ray *ray, - PathRadiance *L, - float3 *throughput) -{ - if(!ss_indirect->tracing) { - ss_indirect->direct_L = *L; - } - ss_indirect->tracing = true; - - /* Setup state, ray and throughput for indirect SSS rays. */ - ss_indirect->num_rays--; - - Ray *indirect_ray = &ss_indirect->rays[ss_indirect->num_rays]; - PathRadiance *indirect_L = &ss_indirect->L[ss_indirect->num_rays]; - - *state = ss_indirect->state[ss_indirect->num_rays]; - *ray = *indirect_ray; - *L = *indirect_L; - *throughput = ss_indirect->throughputs[ss_indirect->num_rays]; - - state->rng_offset += ss_indirect->num_rays * PRNG_BOUNCE_NUM; -} - -#endif /* __SUBSURFACE__ */ ccl_device_inline float4 kernel_path_integrate(KernelGlobals *kg, RNG *rng, @@ -631,7 +476,7 @@ ccl_device_inline float4 kernel_path_integrate(KernelGlobals *kg, } extmax = kernel_data.curve.maximum_width; - lcg_state = lcg_state_init(rng, &state, 0x51633e2d); + lcg_state = lcg_state_init(rng, state.rng_offset, state.sample, 0x51633e2d); } if(state.bounce > kernel_data.integrator.ao_bounces) { @@ -776,7 +621,7 @@ ccl_device_inline float4 kernel_path_integrate(KernelGlobals *kg, #ifdef __BACKGROUND__ /* sample background shader */ float3 L_background = indirect_background(kg, &emission_sd, &state, &ray); - path_radiance_accum_background(&L, throughput, L_background, state.bounce); + path_radiance_accum_background(&L, &state, throughput, L_background); #endif /* __BACKGROUND__ */ break; @@ -790,6 +635,21 @@ ccl_device_inline float4 kernel_path_integrate(KernelGlobals *kg, float rbsdf = path_state_rng_1D_for_decision(kg, rng, &state, PRNG_BSDF); shader_eval_surface(kg, &sd, rng, &state, rbsdf, state.flag, SHADER_CONTEXT_MAIN); +#ifdef __SHADOW_TRICKS__ + if((sd.object_flag & SD_OBJECT_SHADOW_CATCHER)) { + if(state.flag & PATH_RAY_CAMERA) { + state.flag |= (PATH_RAY_SHADOW_CATCHER | PATH_RAY_SHADOW_CATCHER_ONLY); + state.catcher_object = sd.object; + if(!kernel_data.background.transparent) { + L.shadow_color = indirect_background(kg, &emission_sd, &state, &ray); + } + } + } + else { + state.flag &= ~PATH_RAY_SHADOW_CATCHER_ONLY; + } +#endif /* __SHADOW_TRICKS__ */ + /* holdout */ #ifdef __HOLDOUT__ if(((sd.flag & SD_HOLDOUT) || @@ -907,7 +767,16 @@ ccl_device_inline float4 kernel_path_integrate(KernelGlobals *kg, } #endif /* __SUBSURFACE__ */ - float3 L_sum = path_radiance_clamp_and_sum(kg, &L); + float3 L_sum; +#ifdef __SHADOW_TRICKS__ + if(state.flag & PATH_RAY_SHADOW_CATCHER) { + L_sum = path_radiance_sum_shadowcatcher(kg, &L, &L_transparent); + } + else +#endif /* __SHADOW_TRICKS__ */ + { + L_sum = path_radiance_clamp_and_sum(kg, &L); + } kernel_write_light_passes(kg, buffer, &L, sample); diff --git a/intern/cycles/kernel/kernel_path_branched.h b/intern/cycles/kernel/kernel_path_branched.h index ff2b828795d..36fd6c95fe7 100644 --- a/intern/cycles/kernel/kernel_path_branched.h +++ b/intern/cycles/kernel/kernel_path_branched.h @@ -42,21 +42,25 @@ ccl_device_inline void kernel_branched_path_ao(KernelGlobals *kg, sample_cos_hemisphere(ao_N, bsdf_u, bsdf_v, &ao_D, &ao_pdf); - if(dot(ccl_fetch(sd, Ng), ao_D) > 0.0f && ao_pdf != 0.0f) { + if(dot(sd->Ng, ao_D) > 0.0f && ao_pdf != 0.0f) { Ray light_ray; float3 ao_shadow; - light_ray.P = ray_offset(ccl_fetch(sd, P), ccl_fetch(sd, Ng)); + light_ray.P = ray_offset(sd->P, sd->Ng); light_ray.D = ao_D; light_ray.t = kernel_data.background.ao_distance; #ifdef __OBJECT_MOTION__ - light_ray.time = ccl_fetch(sd, time); + light_ray.time = sd->time; #endif /* __OBJECT_MOTION__ */ - light_ray.dP = ccl_fetch(sd, dP); + light_ray.dP = sd->dP; light_ray.dD = differential3_zero(); - if(!shadow_blocked(kg, emission_sd, state, &light_ray, &ao_shadow)) + if(!shadow_blocked(kg, emission_sd, state, &light_ray, &ao_shadow)) { path_radiance_accum_ao(L, throughput*num_samples_inv, ao_alpha, ao_bsdf, ao_shadow, state->bounce); + } + else { + path_radiance_accum_total_ao(L, throughput*num_samples_inv, ao_bsdf); + } } } } @@ -67,8 +71,8 @@ ccl_device_noinline void kernel_branched_path_surface_indirect_light(KernelGloba RNG *rng, ShaderData *sd, ShaderData *indirect_sd, ShaderData *emission_sd, float3 throughput, float num_samples_adjust, PathState *state, PathRadiance *L) { - for(int i = 0; i < ccl_fetch(sd, num_closure); i++) { - const ShaderClosure *sc = &ccl_fetch(sd, closure)[i]; + for(int i = 0; i < sd->num_closure; i++) { + const ShaderClosure *sc = &sd->closure[i]; if(!CLOSURE_IS_BSDF(sc->type)) continue; @@ -140,14 +144,14 @@ ccl_device void kernel_branched_path_subsurface_scatter(KernelGlobals *kg, Ray *ray, float3 throughput) { - for(int i = 0; i < ccl_fetch(sd, num_closure); i++) { - ShaderClosure *sc = &ccl_fetch(sd, closure)[i]; + for(int i = 0; i < sd->num_closure; i++) { + ShaderClosure *sc = &sd->closure[i]; if(!CLOSURE_IS_BSSRDF(sc->type)) continue; /* set up random number generator */ - uint lcg_state = lcg_state_init(rng, state, 0x68bc21eb); + uint lcg_state = lcg_state_init(rng, state->rng_offset, state->sample, 0x68bc21eb); int num_samples = kernel_data.integrator.subsurface_samples; float num_samples_inv = 1.0f/num_samples; RNG bssrdf_rng = cmj_hash(*rng, i); @@ -169,7 +173,7 @@ ccl_device void kernel_branched_path_subsurface_scatter(KernelGlobals *kg, Ray volume_ray = *ray; bool need_update_volume_stack = kernel_data.integrator.use_volumes && - ccl_fetch(sd, object_flag) & SD_OBJECT_INTERSECTS_VOLUME; + sd->object_flag & SD_OBJECT_INTERSECTS_VOLUME; #endif /* __VOLUME__ */ /* compute lighting with the BSDF closure */ @@ -206,7 +210,8 @@ ccl_device void kernel_branched_path_subsurface_scatter(KernelGlobals *kg, #ifdef __EMISSION__ /* direct light */ if(kernel_data.integrator.use_direct_light) { - int all = kernel_data.integrator.sample_all_lights_direct; + int all = (kernel_data.integrator.sample_all_lights_direct) || + (state->flag & PATH_RAY_SHADOW_CATCHER); kernel_branched_path_surface_connect_light( kg, rng, @@ -280,7 +285,7 @@ ccl_device float4 kernel_branched_path_integrate(KernelGlobals *kg, RNG *rng, in } extmax = kernel_data.curve.maximum_width; - lcg_state = lcg_state_init(rng, &state, 0x51633e2d); + lcg_state = lcg_state_init(rng, state.rng_offset, state.sample, 0x51633e2d); } bool hit = scene_intersect(kg, ray, visibility, &isect, &lcg_state, difl, extmax); @@ -461,7 +466,7 @@ ccl_device float4 kernel_branched_path_integrate(KernelGlobals *kg, RNG *rng, in #ifdef __BACKGROUND__ /* sample background shader */ float3 L_background = indirect_background(kg, &emission_sd, &state, &ray); - path_radiance_accum_background(&L, throughput, L_background, state.bounce); + path_radiance_accum_background(&L, &state, throughput, L_background); #endif /* __BACKGROUND__ */ break; @@ -472,6 +477,21 @@ ccl_device float4 kernel_branched_path_integrate(KernelGlobals *kg, RNG *rng, in shader_eval_surface(kg, &sd, rng, &state, 0.0f, state.flag, SHADER_CONTEXT_MAIN); shader_merge_closures(&sd); +#ifdef __SHADOW_TRICKS__ + if((sd.object_flag & SD_OBJECT_SHADOW_CATCHER)) { + if(state.flag & PATH_RAY_CAMERA) { + state.flag |= (PATH_RAY_SHADOW_CATCHER | PATH_RAY_SHADOW_CATCHER_ONLY); + state.catcher_object = sd.object; + if(!kernel_data.background.transparent) { + L.shadow_color = indirect_background(kg, &emission_sd, &state, &ray); + } + } + } + else { + state.flag &= ~PATH_RAY_SHADOW_CATCHER_ONLY; + } +#endif /* __SHADOW_TRICKS__ */ + /* holdout */ #ifdef __HOLDOUT__ if((sd.flag & SD_HOLDOUT) || (sd.object_flag & SD_OBJECT_HOLDOUT_MASK)) { @@ -544,7 +564,8 @@ ccl_device float4 kernel_branched_path_integrate(KernelGlobals *kg, RNG *rng, in #ifdef __EMISSION__ /* direct light */ if(kernel_data.integrator.use_direct_light) { - int all = kernel_data.integrator.sample_all_lights_direct; + int all = (kernel_data.integrator.sample_all_lights_direct) || + (state.flag & PATH_RAY_SHADOW_CATCHER); kernel_branched_path_surface_connect_light(kg, rng, &sd, &emission_sd, &hit_state, throughput, 1.0f, &L, all); } @@ -581,7 +602,16 @@ ccl_device float4 kernel_branched_path_integrate(KernelGlobals *kg, RNG *rng, in #endif /* __VOLUME__ */ } - float3 L_sum = path_radiance_clamp_and_sum(kg, &L); + float3 L_sum; +#ifdef __SHADOW_TRICKS__ + if(state.flag & PATH_RAY_SHADOW_CATCHER) { + L_sum = path_radiance_sum_shadowcatcher(kg, &L, &L_transparent); + } + else +#endif /* __SHADOW_TRICKS__ */ + { + L_sum = path_radiance_clamp_and_sum(kg, &L); + } kernel_write_light_passes(kg, buffer, &L, sample); diff --git a/intern/cycles/kernel/kernel_path_common.h b/intern/cycles/kernel/kernel_path_common.h index 7b903556bf9..82f83deb595 100644 --- a/intern/cycles/kernel/kernel_path_common.h +++ b/intern/cycles/kernel/kernel_path_common.h @@ -14,7 +14,7 @@ * limitations under the License. */ -#include "util_hash.h" +#include "util/util_hash.h" CCL_NAMESPACE_BEGIN @@ -22,7 +22,7 @@ ccl_device_inline void kernel_path_trace_setup(KernelGlobals *kg, ccl_global uint *rng_state, int sample, int x, int y, - ccl_addr_space RNG *rng, + RNG *rng, ccl_addr_space Ray *ray) { float filter_u; diff --git a/intern/cycles/kernel/kernel_path_state.h b/intern/cycles/kernel/kernel_path_state.h index 661dc52fb31..c0cd2a63120 100644 --- a/intern/cycles/kernel/kernel_path_state.h +++ b/intern/cycles/kernel/kernel_path_state.h @@ -19,7 +19,7 @@ CCL_NAMESPACE_BEGIN ccl_device_inline void path_state_init(KernelGlobals *kg, ShaderData *stack_sd, ccl_addr_space PathState *state, - ccl_addr_space RNG *rng, + RNG *rng, int sample, ccl_addr_space Ray *ray) { @@ -54,6 +54,10 @@ ccl_device_inline void path_state_init(KernelGlobals *kg, state->volume_stack[0].shader = SHADER_NONE; } #endif + +#ifdef __SHADOW_TRICKS__ + state->catcher_object = OBJECT_NONE; +#endif } ccl_device_inline void path_state_next(KernelGlobals *kg, ccl_addr_space PathState *state, int label) diff --git a/intern/cycles/kernel/kernel_path_subsurface.h b/intern/cycles/kernel/kernel_path_subsurface.h new file mode 100644 index 00000000000..10b568ac3dd --- /dev/null +++ b/intern/cycles/kernel/kernel_path_subsurface.h @@ -0,0 +1,187 @@ +/* + * Copyright 2017 Blender Foundation + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +CCL_NAMESPACE_BEGIN + +#ifdef __SUBSURFACE__ +# ifndef __KERNEL_CUDA__ +ccl_device +# else +ccl_device_inline +# endif +bool kernel_path_subsurface_scatter( + KernelGlobals *kg, + ShaderData *sd, + ShaderData *emission_sd, + PathRadiance *L, + ccl_addr_space PathState *state, + RNG *rng, + ccl_addr_space Ray *ray, + ccl_addr_space float3 *throughput, + ccl_addr_space SubsurfaceIndirectRays *ss_indirect) +{ + float bssrdf_probability; + ShaderClosure *sc = subsurface_scatter_pick_closure(kg, sd, &bssrdf_probability); + + /* modify throughput for picking bssrdf or bsdf */ + *throughput *= bssrdf_probability; + + /* do bssrdf scatter step if we picked a bssrdf closure */ + if(sc) { + /* We should never have two consecutive BSSRDF bounces, + * the second one should be converted to a diffuse BSDF to + * avoid this. + */ + kernel_assert(!ss_indirect->tracing); + + uint lcg_state = lcg_state_init(rng, state->rng_offset, state->sample, 0x68bc21eb); + + SubsurfaceIntersection ss_isect; + float bssrdf_u, bssrdf_v; + path_state_rng_2D(kg, rng, state, PRNG_BSDF_U, &bssrdf_u, &bssrdf_v); + int num_hits = subsurface_scatter_multi_intersect(kg, + &ss_isect, + sd, + sc, + &lcg_state, + bssrdf_u, bssrdf_v, + false); +# ifdef __VOLUME__ + ss_indirect->need_update_volume_stack = + kernel_data.integrator.use_volumes && + sd->object_flag & SD_OBJECT_INTERSECTS_VOLUME; +# endif /* __VOLUME__ */ + + /* compute lighting with the BSDF closure */ + for(int hit = 0; hit < num_hits; hit++) { + /* NOTE: We reuse the existing ShaderData, we assume the path + * integration loop stops when this function returns true. + */ + subsurface_scatter_multi_setup(kg, + &ss_isect, + hit, + sd, + state, + state->flag, + sc, + false); + + ccl_addr_space PathState *hit_state = &ss_indirect->state[ss_indirect->num_rays]; + ccl_addr_space Ray *hit_ray = &ss_indirect->rays[ss_indirect->num_rays]; + ccl_addr_space float3 *hit_tp = &ss_indirect->throughputs[ss_indirect->num_rays]; + PathRadiance *hit_L = &ss_indirect->L[ss_indirect->num_rays]; + + *hit_state = *state; + *hit_ray = *ray; + *hit_tp = *throughput; + + hit_state->rng_offset += PRNG_BOUNCE_NUM; + + path_radiance_init(hit_L, kernel_data.film.use_light_pass); + hit_L->direct_throughput = L->direct_throughput; + path_radiance_copy_indirect(hit_L, L); + + kernel_path_surface_connect_light(kg, rng, sd, emission_sd, *hit_tp, state, hit_L); + + if(kernel_path_surface_bounce(kg, + rng, + sd, + hit_tp, + hit_state, + hit_L, + hit_ray)) + { +# ifdef __LAMP_MIS__ + hit_state->ray_t = 0.0f; +# endif /* __LAMP_MIS__ */ + +# ifdef __VOLUME__ + if(ss_indirect->need_update_volume_stack) { + Ray volume_ray = *ray; + /* Setup ray from previous surface point to the new one. */ + volume_ray.D = normalize_len(hit_ray->P - volume_ray.P, + &volume_ray.t); + + kernel_volume_stack_update_for_subsurface( + kg, + emission_sd, + &volume_ray, + hit_state->volume_stack); + } +# endif /* __VOLUME__ */ + path_radiance_reset_indirect(L); + ss_indirect->num_rays++; + } + else { + path_radiance_accum_sample(L, hit_L, 1); + } + } + return true; + } + return false; +} + +ccl_device_inline void kernel_path_subsurface_init_indirect( + ccl_addr_space SubsurfaceIndirectRays *ss_indirect) +{ + ss_indirect->tracing = false; + ss_indirect->num_rays = 0; +} + +ccl_device void kernel_path_subsurface_accum_indirect( + ccl_addr_space SubsurfaceIndirectRays *ss_indirect, + PathRadiance *L) +{ + if(ss_indirect->tracing) { + path_radiance_sum_indirect(L); + path_radiance_accum_sample(&ss_indirect->direct_L, L, 1); + if(ss_indirect->num_rays == 0) { + *L = ss_indirect->direct_L; + } + } +} + +ccl_device void kernel_path_subsurface_setup_indirect( + KernelGlobals *kg, + ccl_addr_space SubsurfaceIndirectRays *ss_indirect, + ccl_addr_space PathState *state, + ccl_addr_space Ray *ray, + PathRadiance *L, + ccl_addr_space float3 *throughput) +{ + if(!ss_indirect->tracing) { + ss_indirect->direct_L = *L; + } + ss_indirect->tracing = true; + + /* Setup state, ray and throughput for indirect SSS rays. */ + ss_indirect->num_rays--; + + ccl_addr_space Ray *indirect_ray = &ss_indirect->rays[ss_indirect->num_rays]; + PathRadiance *indirect_L = &ss_indirect->L[ss_indirect->num_rays]; + + *state = ss_indirect->state[ss_indirect->num_rays]; + *ray = *indirect_ray; + *L = *indirect_L; + *throughput = ss_indirect->throughputs[ss_indirect->num_rays]; + + state->rng_offset += ss_indirect->num_rays * PRNG_BOUNCE_NUM; +} + +#endif /* __SUBSURFACE__ */ + +CCL_NAMESPACE_END + diff --git a/intern/cycles/kernel/kernel_path_surface.h b/intern/cycles/kernel/kernel_path_surface.h index fea503d06e5..076c82f3853 100644 --- a/intern/cycles/kernel/kernel_path_surface.h +++ b/intern/cycles/kernel/kernel_path_surface.h @@ -16,16 +16,22 @@ CCL_NAMESPACE_BEGIN -#if defined(__BRANCHED_PATH__) || defined(__SUBSURFACE__) - +#if defined(__BRANCHED_PATH__) || defined(__SUBSURFACE__) || defined(__SHADOW_TRICKS__) /* branched path tracing: connect path directly to position on one or more lights and add it to L */ -ccl_device_noinline void kernel_branched_path_surface_connect_light(KernelGlobals *kg, RNG *rng, - ShaderData *sd, ShaderData *emission_sd, PathState *state, float3 throughput, - float num_samples_adjust, PathRadiance *L, int sample_all_lights) +ccl_device_noinline void kernel_branched_path_surface_connect_light( + KernelGlobals *kg, + RNG *rng, + ShaderData *sd, + ShaderData *emission_sd, + ccl_addr_space PathState *state, + float3 throughput, + float num_samples_adjust, + PathRadiance *L, + int sample_all_lights) { #ifdef __EMISSION__ /* sample illumination from lights to find path contribution */ - if(!(ccl_fetch(sd, flag) & SD_BSDF_HAS_EVAL)) + if(!(sd->flag & SD_BSDF_HAS_EVAL)) return; Ray light_ray; @@ -33,7 +39,7 @@ ccl_device_noinline void kernel_branched_path_surface_connect_light(KernelGlobal bool is_lamp; # ifdef __OBJECT_MOTION__ - light_ray.time = ccl_fetch(sd, time); + light_ray.time = sd->time; # endif if(sample_all_lights) { @@ -52,7 +58,7 @@ ccl_device_noinline void kernel_branched_path_surface_connect_light(KernelGlobal float terminate = path_branched_rng_light_termination(kg, &lamp_rng, state, j, num_samples); LightSample ls; - if(lamp_light_sample(kg, i, light_u, light_v, ccl_fetch(sd, P), &ls)) { + if(lamp_light_sample(kg, i, light_u, light_v, sd->P, &ls)) { /* The sampling probability returned by lamp_light_sample assumes that all lights were sampled. * However, this code only samples lamps, so if the scene also had mesh lights, the real probability is twice as high. */ if(kernel_data.integrator.pdf_triangles != 0.0f) @@ -66,6 +72,9 @@ ccl_device_noinline void kernel_branched_path_surface_connect_light(KernelGlobal /* accumulate */ path_radiance_accum_light(L, throughput*num_samples_inv, &L_light, shadow, num_samples_inv, state->bounce, is_lamp); } + else { + path_radiance_accum_total_light(L, throughput*num_samples_inv, &L_light); + } } } } @@ -87,7 +96,7 @@ ccl_device_noinline void kernel_branched_path_surface_connect_light(KernelGlobal light_t = 0.5f*light_t; LightSample ls; - if(light_sample(kg, light_t, light_u, light_v, ccl_fetch(sd, time), ccl_fetch(sd, P), state->bounce, &ls)) { + if(light_sample(kg, light_t, light_u, light_v, sd->time, sd->P, state->bounce, &ls)) { /* Same as above, probability needs to be corrected since the sampling was forced to select a mesh light. */ if(kernel_data.integrator.num_all_lights) ls.pdf *= 2.0f; @@ -100,6 +109,9 @@ ccl_device_noinline void kernel_branched_path_surface_connect_light(KernelGlobal /* accumulate */ path_radiance_accum_light(L, throughput*num_samples_inv, &L_light, shadow, num_samples_inv, state->bounce, is_lamp); } + else { + path_radiance_accum_total_light(L, throughput*num_samples_inv, &L_light); + } } } } @@ -113,7 +125,7 @@ ccl_device_noinline void kernel_branched_path_surface_connect_light(KernelGlobal float terminate = path_state_rng_light_termination(kg, rng, state); LightSample ls; - if(light_sample(kg, light_t, light_u, light_v, ccl_fetch(sd, time), ccl_fetch(sd, P), state->bounce, &ls)) { + if(light_sample(kg, light_t, light_u, light_v, sd->time, sd->P, state->bounce, &ls)) { /* sample random light */ if(direct_emission(kg, sd, emission_sd, &ls, state, &light_ray, &L_light, &is_lamp, terminate)) { /* trace shadow ray */ @@ -123,6 +135,9 @@ ccl_device_noinline void kernel_branched_path_surface_connect_light(KernelGlobal /* accumulate */ path_radiance_accum_light(L, throughput*num_samples_adjust, &L_light, shadow, num_samples_adjust, state->bounce, is_lamp); } + else { + path_radiance_accum_total_light(L, throughput*num_samples_adjust, &L_light); + } } } } @@ -130,9 +145,17 @@ ccl_device_noinline void kernel_branched_path_surface_connect_light(KernelGlobal } /* branched path tracing: bounce off or through surface to with new direction stored in ray */ -ccl_device bool kernel_branched_path_surface_bounce(KernelGlobals *kg, RNG *rng, - ShaderData *sd, const ShaderClosure *sc, int sample, int num_samples, - float3 *throughput, PathState *state, PathRadiance *L, Ray *ray) +ccl_device bool kernel_branched_path_surface_bounce( + KernelGlobals *kg, + RNG *rng, + ShaderData *sd, + const ShaderClosure *sc, + int sample, + int num_samples, + ccl_addr_space float3 *throughput, + ccl_addr_space PathState *state, + PathRadiance *L, + Ray *ray) { /* sample BSDF */ float bsdf_pdf; @@ -156,15 +179,15 @@ ccl_device bool kernel_branched_path_surface_bounce(KernelGlobals *kg, RNG *rng, path_state_next(kg, state, label); /* setup ray */ - ray->P = ray_offset(ccl_fetch(sd, P), (label & LABEL_TRANSMIT)? -ccl_fetch(sd, Ng): ccl_fetch(sd, Ng)); + ray->P = ray_offset(sd->P, (label & LABEL_TRANSMIT)? -sd->Ng: sd->Ng); ray->D = normalize(bsdf_omega_in); ray->t = FLT_MAX; #ifdef __RAY_DIFFERENTIALS__ - ray->dP = ccl_fetch(sd, dP); + ray->dP = sd->dP; ray->dD = bsdf_domega_in; #endif #ifdef __OBJECT_MOTION__ - ray->time = ccl_fetch(sd, time); + ray->time = sd->time; #endif #ifdef __VOLUME__ @@ -188,15 +211,29 @@ ccl_device bool kernel_branched_path_surface_bounce(KernelGlobals *kg, RNG *rng, #endif -#ifndef __SPLIT_KERNEL__ /* path tracing: connect path directly to position on a light and add it to L */ -ccl_device_inline void kernel_path_surface_connect_light(KernelGlobals *kg, ccl_addr_space RNG *rng, +ccl_device_inline void kernel_path_surface_connect_light(KernelGlobals *kg, RNG *rng, ShaderData *sd, ShaderData *emission_sd, float3 throughput, ccl_addr_space PathState *state, PathRadiance *L) { #ifdef __EMISSION__ - if(!(kernel_data.integrator.use_direct_light && (ccl_fetch(sd, flag) & SD_BSDF_HAS_EVAL))) + if(!(kernel_data.integrator.use_direct_light && (sd->flag & SD_BSDF_HAS_EVAL))) + return; + +#ifdef __SHADOW_TRICKS__ + if(state->flag & PATH_RAY_SHADOW_CATCHER) { + kernel_branched_path_surface_connect_light(kg, + rng, + sd, + emission_sd, + state, + throughput, + 1.0f, + L, + 1); return; + } +#endif /* sample illumination from lights to find path contribution */ float light_t = path_state_rng_1D(kg, rng, state, PRNG_LIGHT); @@ -208,11 +245,11 @@ ccl_device_inline void kernel_path_surface_connect_light(KernelGlobals *kg, ccl_ bool is_lamp; #ifdef __OBJECT_MOTION__ - light_ray.time = ccl_fetch(sd, time); + light_ray.time = sd->time; #endif LightSample ls; - if(light_sample(kg, light_t, light_u, light_v, ccl_fetch(sd, time), ccl_fetch(sd, P), state->bounce, &ls)) { + if(light_sample(kg, light_t, light_u, light_v, sd->time, sd->P, state->bounce, &ls)) { float terminate = path_state_rng_light_termination(kg, rng, state); if(direct_emission(kg, sd, emission_sd, &ls, state, &light_ray, &L_light, &is_lamp, terminate)) { /* trace shadow ray */ @@ -222,15 +259,17 @@ ccl_device_inline void kernel_path_surface_connect_light(KernelGlobals *kg, ccl_ /* accumulate */ path_radiance_accum_light(L, throughput, &L_light, shadow, 1.0f, state->bounce, is_lamp); } + else { + path_radiance_accum_total_light(L, throughput, &L_light); + } } } #endif } -#endif /* path tracing: bounce off or through surface to with new direction stored in ray */ ccl_device bool kernel_path_surface_bounce(KernelGlobals *kg, - ccl_addr_space RNG *rng, + RNG *rng, ShaderData *sd, ccl_addr_space float3 *throughput, ccl_addr_space PathState *state, @@ -238,7 +277,7 @@ ccl_device bool kernel_path_surface_bounce(KernelGlobals *kg, ccl_addr_space Ray *ray) { /* no BSDF? we can stop here */ - if(ccl_fetch(sd, flag) & SD_BSDF) { + if(sd->flag & SD_BSDF) { /* sample BSDF */ float bsdf_pdf; BsdfEval bsdf_eval; @@ -270,16 +309,16 @@ ccl_device bool kernel_path_surface_bounce(KernelGlobals *kg, path_state_next(kg, state, label); /* setup ray */ - ray->P = ray_offset(ccl_fetch(sd, P), (label & LABEL_TRANSMIT)? -ccl_fetch(sd, Ng): ccl_fetch(sd, Ng)); + ray->P = ray_offset(sd->P, (label & LABEL_TRANSMIT)? -sd->Ng: sd->Ng); ray->D = normalize(bsdf_omega_in); if(state->bounce == 0) - ray->t -= ccl_fetch(sd, ray_length); /* clipping works through transparent */ + ray->t -= sd->ray_length; /* clipping works through transparent */ else ray->t = FLT_MAX; #ifdef __RAY_DIFFERENTIALS__ - ray->dP = ccl_fetch(sd, dP); + ray->dP = sd->dP; ray->dD = bsdf_domega_in; #endif @@ -291,21 +330,21 @@ ccl_device bool kernel_path_surface_bounce(KernelGlobals *kg, return true; } #ifdef __VOLUME__ - else if(ccl_fetch(sd, flag) & SD_HAS_ONLY_VOLUME) { + else if(sd->flag & SD_HAS_ONLY_VOLUME) { /* no surface shader but have a volume shader? act transparent */ /* update path state, count as transparent */ path_state_next(kg, state, LABEL_TRANSPARENT); if(state->bounce == 0) - ray->t -= ccl_fetch(sd, ray_length); /* clipping works through transparent */ + ray->t -= sd->ray_length; /* clipping works through transparent */ else ray->t = FLT_MAX; /* setup ray position, direction stays unchanged */ - ray->P = ray_offset(ccl_fetch(sd, P), -ccl_fetch(sd, Ng)); + ray->P = ray_offset(sd->P, -sd->Ng); #ifdef __RAY_DIFFERENTIALS__ - ray->dP = ccl_fetch(sd, dP); + ray->dP = sd->dP; #endif /* enter/exit volume */ diff --git a/intern/cycles/kernel/kernel_path_volume.h b/intern/cycles/kernel/kernel_path_volume.h index 3d3b7385d8b..371f2c1c7cb 100644 --- a/intern/cycles/kernel/kernel_path_volume.h +++ b/intern/cycles/kernel/kernel_path_volume.h @@ -24,7 +24,7 @@ ccl_device_inline void kernel_path_volume_connect_light( ShaderData *sd, ShaderData *emission_sd, float3 throughput, - PathState *state, + ccl_addr_space PathState *state, PathRadiance *L) { #ifdef __EMISSION__ @@ -59,7 +59,7 @@ ccl_device_inline void kernel_path_volume_connect_light( } } } -#endif +#endif /* __EMISSION__ */ } #ifdef __KERNEL_GPU__ @@ -67,8 +67,14 @@ ccl_device_noinline #else ccl_device #endif -bool kernel_path_volume_bounce(KernelGlobals *kg, RNG *rng, - ShaderData *sd, float3 *throughput, PathState *state, PathRadiance *L, Ray *ray) +bool kernel_path_volume_bounce( + KernelGlobals *kg, + RNG *rng, + ShaderData *sd, + ccl_addr_space float3 *throughput, + ccl_addr_space PathState *state, + PathRadiance *L, + ccl_addr_space Ray *ray) { /* sample phase function */ float phase_pdf; @@ -111,9 +117,18 @@ bool kernel_path_volume_bounce(KernelGlobals *kg, RNG *rng, return true; } -ccl_device void kernel_branched_path_volume_connect_light(KernelGlobals *kg, RNG *rng, - ShaderData *sd, ShaderData *emission_sd, float3 throughput, PathState *state, PathRadiance *L, - bool sample_all_lights, Ray *ray, const VolumeSegment *segment) +#ifndef __SPLIT_KERNEL__ +ccl_device void kernel_branched_path_volume_connect_light( + KernelGlobals *kg, + RNG *rng, + ShaderData *sd, + ShaderData *emission_sd, + float3 throughput, + ccl_addr_space PathState *state, + PathRadiance *L, + bool sample_all_lights, + Ray *ray, + const VolumeSegment *segment) { #ifdef __EMISSION__ if(!kernel_data.integrator.use_direct_light) @@ -261,10 +276,11 @@ ccl_device void kernel_branched_path_volume_connect_light(KernelGlobals *kg, RNG } } } -#endif +#endif /* __EMISSION__ */ } +#endif /* __SPLIT_KERNEL__ */ -#endif +#endif /* __VOLUME_SCATTER__ */ CCL_NAMESPACE_END diff --git a/intern/cycles/kernel/kernel_queues.h b/intern/cycles/kernel/kernel_queues.h index cf5614b8a86..96bc636d5ac 100644 --- a/intern/cycles/kernel/kernel_queues.h +++ b/intern/cycles/kernel/kernel_queues.h @@ -17,12 +17,15 @@ #ifndef __KERNEL_QUEUE_H__ #define __KERNEL_QUEUE_H__ +CCL_NAMESPACE_BEGIN + /* * Queue utility functions for split kernel */ - +#ifdef __KERNEL_OPENCL__ #pragma OPENCL EXTENSION cl_khr_global_int32_base_atomics : enable #pragma OPENCL EXTENSION cl_khr_local_int32_base_atomics : enable +#endif /* * Enqueue ray index into the queue @@ -35,7 +38,8 @@ ccl_device void enqueue_ray_index( ccl_global int *queue_index) /* Array of size num_queues; Used for atomic increment. */ { /* This thread's queue index. */ - int my_queue_index = atomic_inc(&queue_index[queue_number]) + (queue_number * queue_size); + int my_queue_index = atomic_fetch_and_inc_uint32((ccl_global uint*)&queue_index[queue_number]) + + (queue_number * queue_size); queues[my_queue_index] = ray_index; } @@ -47,6 +51,7 @@ ccl_device void enqueue_ray_index( * is no more ray to allocate to other threads. */ ccl_device int get_ray_index( + KernelGlobals *kg, int thread_index, /* Global thread index. */ int queue_number, /* Queue to operate on. */ ccl_global int *queues, /* Buffer of all queues. */ @@ -68,24 +73,25 @@ ccl_device void enqueue_ray_index_local( int queue_number, /* Queue in which to enqueue ray index. */ char enqueue_flag, /* True for threads whose ray index has to be enqueued. */ int queuesize, /* queue size. */ - ccl_local unsigned int *local_queue_atomics, /* To to local queue atomics. */ + ccl_local_param unsigned int *local_queue_atomics, /* To to local queue atomics. */ ccl_global int *Queue_data, /* Queues. */ ccl_global int *Queue_index) /* To do global queue atomics. */ { - int lidx = get_local_id(1) * get_local_size(0) + get_local_id(0); + int lidx = ccl_local_id(1) * ccl_local_size(0) + ccl_local_id(0); /* Get local queue id .*/ unsigned int lqidx; if(enqueue_flag) { - lqidx = atomic_inc(local_queue_atomics); + lqidx = atomic_fetch_and_inc_uint32(local_queue_atomics); } - barrier(CLK_LOCAL_MEM_FENCE); + ccl_barrier(CCL_LOCAL_MEM_FENCE); /* Get global queue offset. */ if(lidx == 0) { - *local_queue_atomics = atomic_add(&Queue_index[queue_number], *local_queue_atomics); + *local_queue_atomics = atomic_fetch_and_add_uint32((ccl_global uint*)&Queue_index[queue_number], + *local_queue_atomics); } - barrier(CLK_LOCAL_MEM_FENCE); + ccl_barrier(CCL_LOCAL_MEM_FENCE); /* Get global queue index and enqueue ray. */ if(enqueue_flag) { @@ -96,19 +102,19 @@ ccl_device void enqueue_ray_index_local( ccl_device unsigned int get_local_queue_index( int queue_number, /* Queue in which to enqueue the ray; -1 if no queue */ - ccl_local unsigned int *local_queue_atomics) + ccl_local_param unsigned int *local_queue_atomics) { - int my_lqidx = atomic_inc(&local_queue_atomics[queue_number]); + int my_lqidx = atomic_fetch_and_inc_uint32(&local_queue_atomics[queue_number]); return my_lqidx; } ccl_device unsigned int get_global_per_queue_offset( int queue_number, - ccl_local unsigned int *local_queue_atomics, + ccl_local_param unsigned int *local_queue_atomics, ccl_global int* global_queue_atomics) { - unsigned int queue_offset = atomic_add(&global_queue_atomics[queue_number], - local_queue_atomics[queue_number]); + unsigned int queue_offset = atomic_fetch_and_add_uint32((ccl_global uint*)&global_queue_atomics[queue_number], + local_queue_atomics[queue_number]); return queue_offset; } @@ -116,10 +122,12 @@ ccl_device unsigned int get_global_queue_index( int queue_number, int queuesize, unsigned int lqidx, - ccl_local unsigned int * global_per_queue_offset) + ccl_local_param unsigned int * global_per_queue_offset) { int my_gqidx = queuesize * queue_number + lqidx + global_per_queue_offset[queue_number]; return my_gqidx; } +CCL_NAMESPACE_END + #endif // __KERNEL_QUEUE_H__ diff --git a/intern/cycles/kernel/kernel_random.h b/intern/cycles/kernel/kernel_random.h index e773753396f..d4f0caff5de 100644 --- a/intern/cycles/kernel/kernel_random.h +++ b/intern/cycles/kernel/kernel_random.h @@ -14,7 +14,7 @@ * limitations under the License. */ -#include "kernel_jitter.h" +#include "kernel/kernel_jitter.h" CCL_NAMESPACE_BEGIN @@ -98,7 +98,7 @@ ccl_device uint sobol_lookup(const uint m, const uint frame, const uint ex, cons return index; } -ccl_device_forceinline float path_rng_1D(KernelGlobals *kg, ccl_addr_space RNG *rng, int sample, int num_samples, int dimension) +ccl_device_forceinline float path_rng_1D(KernelGlobals *kg, RNG *rng, int sample, int num_samples, int dimension) { #ifdef __CMJ__ if(kernel_data.integrator.sampling_pattern == SAMPLING_PATTERN_CMJ) { @@ -130,7 +130,7 @@ ccl_device_forceinline float path_rng_1D(KernelGlobals *kg, ccl_addr_space RNG * #endif } -ccl_device_forceinline void path_rng_2D(KernelGlobals *kg, ccl_addr_space RNG *rng, int sample, int num_samples, int dimension, float *fx, float *fy) +ccl_device_forceinline void path_rng_2D(KernelGlobals *kg, RNG *rng, int sample, int num_samples, int dimension, float *fx, float *fy) { #ifdef __CMJ__ if(kernel_data.integrator.sampling_pattern == SAMPLING_PATTERN_CMJ) { @@ -147,7 +147,7 @@ ccl_device_forceinline void path_rng_2D(KernelGlobals *kg, ccl_addr_space RNG *r } } -ccl_device_inline void path_rng_init(KernelGlobals *kg, ccl_global uint *rng_state, int sample, int num_samples, ccl_addr_space RNG *rng, int x, int y, float *fx, float *fy) +ccl_device_inline void path_rng_init(KernelGlobals *kg, ccl_global uint *rng_state, int sample, int num_samples, RNG *rng, int x, int y, float *fx, float *fy) { #ifdef __SOBOL_FULL_SCREEN__ uint px, py; @@ -191,14 +191,14 @@ ccl_device void path_rng_end(KernelGlobals *kg, ccl_global uint *rng_state, RNG /* Linear Congruential Generator */ -ccl_device_forceinline float path_rng_1D(KernelGlobals *kg, RNG& rng, int sample, int num_samples, int dimension) +ccl_device_forceinline float path_rng_1D(KernelGlobals *kg, RNG *rng, int sample, int num_samples, int dimension) { /* implicit mod 2^32 */ - rng = (1103515245*(rng) + 12345); - return (float)rng * (1.0f/(float)0xFFFFFFFF); + *rng = (1103515245*(*rng) + 12345); + return (float)*rng * (1.0f/(float)0xFFFFFFFF); } -ccl_device_inline void path_rng_2D(KernelGlobals *kg, RNG& rng, int sample, int num_samples, int dimension, float *fx, float *fy) +ccl_device_inline void path_rng_2D(KernelGlobals *kg, RNG *rng, int sample, int num_samples, int dimension, float *fx, float *fy) { *fx = path_rng_1D(kg, rng, sample, num_samples, dimension); *fy = path_rng_1D(kg, rng, sample, num_samples, dimension + 1); @@ -259,12 +259,12 @@ ccl_device uint lcg_init(uint seed) * For branches in the path we must be careful not to reuse the same number * in a sequence and offset accordingly. */ -ccl_device_inline float path_state_rng_1D(KernelGlobals *kg, ccl_addr_space RNG *rng, const ccl_addr_space PathState *state, int dimension) +ccl_device_inline float path_state_rng_1D(KernelGlobals *kg, RNG *rng, const ccl_addr_space PathState *state, int dimension) { return path_rng_1D(kg, rng, state->sample, state->num_samples, state->rng_offset + dimension); } -ccl_device_inline float path_state_rng_1D_for_decision(KernelGlobals *kg, ccl_addr_space RNG *rng, const ccl_addr_space PathState *state, int dimension) +ccl_device_inline float path_state_rng_1D_for_decision(KernelGlobals *kg, RNG *rng, const ccl_addr_space PathState *state, int dimension) { /* the rng_offset is not increased for transparent bounces. if we do then * fully transparent objects can become subtly visible by the different @@ -277,29 +277,29 @@ ccl_device_inline float path_state_rng_1D_for_decision(KernelGlobals *kg, ccl_ad return path_rng_1D(kg, rng, state->sample, state->num_samples, rng_offset + dimension); } -ccl_device_inline void path_state_rng_2D(KernelGlobals *kg, ccl_addr_space RNG *rng, const ccl_addr_space PathState *state, int dimension, float *fx, float *fy) +ccl_device_inline void path_state_rng_2D(KernelGlobals *kg, RNG *rng, const ccl_addr_space PathState *state, int dimension, float *fx, float *fy) { path_rng_2D(kg, rng, state->sample, state->num_samples, state->rng_offset + dimension, fx, fy); } -ccl_device_inline float path_branched_rng_1D(KernelGlobals *kg, ccl_addr_space RNG *rng, const PathState *state, int branch, int num_branches, int dimension) +ccl_device_inline float path_branched_rng_1D(KernelGlobals *kg, RNG *rng, const ccl_addr_space PathState *state, int branch, int num_branches, int dimension) { return path_rng_1D(kg, rng, state->sample*num_branches + branch, state->num_samples*num_branches, state->rng_offset + dimension); } -ccl_device_inline float path_branched_rng_1D_for_decision(KernelGlobals *kg, ccl_addr_space RNG *rng, const PathState *state, int branch, int num_branches, int dimension) +ccl_device_inline float path_branched_rng_1D_for_decision(KernelGlobals *kg, RNG *rng, const ccl_addr_space PathState *state, int branch, int num_branches, int dimension) { int rng_offset = state->rng_offset + state->transparent_bounce*PRNG_BOUNCE_NUM; return path_rng_1D(kg, rng, state->sample*num_branches + branch, state->num_samples*num_branches, rng_offset + dimension); } -ccl_device_inline void path_branched_rng_2D(KernelGlobals *kg, ccl_addr_space RNG *rng, const PathState *state, int branch, int num_branches, int dimension, float *fx, float *fy) +ccl_device_inline void path_branched_rng_2D(KernelGlobals *kg, RNG *rng, const ccl_addr_space PathState *state, int branch, int num_branches, int dimension, float *fx, float *fy) { path_rng_2D(kg, rng, state->sample*num_branches + branch, state->num_samples*num_branches, state->rng_offset + dimension, fx, fy); } /* Utitility functions to get light termination value, since it might not be needed in many cases. */ -ccl_device_inline float path_state_rng_light_termination(KernelGlobals *kg, ccl_addr_space RNG *rng, const ccl_addr_space PathState *state) +ccl_device_inline float path_state_rng_light_termination(KernelGlobals *kg, RNG *rng, const ccl_addr_space PathState *state) { if(kernel_data.integrator.light_inv_rr_threshold > 0.0f) { return path_state_rng_1D_for_decision(kg, rng, state, PRNG_LIGHT_TERMINATE); @@ -307,7 +307,7 @@ ccl_device_inline float path_state_rng_light_termination(KernelGlobals *kg, ccl_ return 0.0f; } -ccl_device_inline float path_branched_rng_light_termination(KernelGlobals *kg, ccl_addr_space RNG *rng, const PathState *state, int branch, int num_branches) +ccl_device_inline float path_branched_rng_light_termination(KernelGlobals *kg, RNG *rng, const ccl_addr_space PathState *state, int branch, int num_branches) { if(kernel_data.integrator.light_inv_rr_threshold > 0.0f) { return path_branched_rng_1D_for_decision(kg, rng, state, branch, num_branches, PRNG_LIGHT_TERMINATE); @@ -315,7 +315,7 @@ ccl_device_inline float path_branched_rng_light_termination(KernelGlobals *kg, c return 0.0f; } -ccl_device_inline void path_state_branch(PathState *state, int branch, int num_branches) +ccl_device_inline void path_state_branch(ccl_addr_space PathState *state, int branch, int num_branches) { /* path is splitting into a branch, adjust so that each branch * still gets a unique sample from the same sequence */ @@ -324,18 +324,9 @@ ccl_device_inline void path_state_branch(PathState *state, int branch, int num_b state->num_samples = state->num_samples*num_branches; } -ccl_device_inline uint lcg_state_init(RNG *rng, const PathState *state, uint scramble) -{ - return lcg_init(*rng + state->rng_offset + state->sample*scramble); -} - -/* TODO(sergey): For until we can use generic address space from OpenCL 2.0. */ - -ccl_device_inline uint lcg_state_init_addrspace(ccl_addr_space RNG *rng, - const ccl_addr_space PathState *state, - uint scramble) +ccl_device_inline uint lcg_state_init(RNG *rng, int rng_offset, int sample, uint scramble) { - return lcg_init(*rng + state->rng_offset + state->sample*scramble); + return lcg_init(*rng + rng_offset + sample*scramble); } ccl_device float lcg_step_float_addrspace(ccl_addr_space uint *rng) diff --git a/intern/cycles/kernel/kernel_shader.h b/intern/cycles/kernel/kernel_shader.h index d0826e5e879..8c0c5e90a3e 100644 --- a/intern/cycles/kernel/kernel_shader.h +++ b/intern/cycles/kernel/kernel_shader.h @@ -24,12 +24,12 @@ * */ -#include "closure/alloc.h" -#include "closure/bsdf_util.h" -#include "closure/bsdf.h" -#include "closure/emissive.h" +#include "kernel/closure/alloc.h" +#include "kernel/closure/bsdf_util.h" +#include "kernel/closure/bsdf.h" +#include "kernel/closure/emissive.h" -#include "svm/svm.h" +#include "kernel/svm/svm.h" CCL_NAMESPACE_BEGIN @@ -38,13 +38,13 @@ CCL_NAMESPACE_BEGIN #ifdef __OBJECT_MOTION__ ccl_device void shader_setup_object_transforms(KernelGlobals *kg, ShaderData *sd, float time) { - if(ccl_fetch(sd, object_flag) & SD_OBJECT_MOTION) { - ccl_fetch(sd, ob_tfm) = object_fetch_transform_motion(kg, ccl_fetch(sd, object), time); - ccl_fetch(sd, ob_itfm) = transform_quick_inverse(ccl_fetch(sd, ob_tfm)); + if(sd->object_flag & SD_OBJECT_MOTION) { + sd->ob_tfm = object_fetch_transform_motion(kg, sd->object, time); + sd->ob_itfm = transform_quick_inverse(sd->ob_tfm); } else { - ccl_fetch(sd, ob_tfm) = object_fetch_transform(kg, ccl_fetch(sd, object), OBJECT_TRANSFORM); - ccl_fetch(sd, ob_itfm) = object_fetch_transform(kg, ccl_fetch(sd, object), OBJECT_INVERSE_TRANSFORM); + sd->ob_tfm = object_fetch_transform(kg, sd->object, OBJECT_TRANSFORM); + sd->ob_itfm = object_fetch_transform(kg, sd->object, OBJECT_INVERSE_TRANSFORM); } } #endif @@ -55,55 +55,55 @@ ccl_device_noinline void shader_setup_from_ray(KernelGlobals *kg, const Ray *ray) { #ifdef __INSTANCING__ - ccl_fetch(sd, object) = (isect->object == PRIM_NONE)? kernel_tex_fetch(__prim_object, isect->prim): isect->object; + sd->object = (isect->object == PRIM_NONE)? kernel_tex_fetch(__prim_object, isect->prim): isect->object; #endif - ccl_fetch(sd, type) = isect->type; - ccl_fetch(sd, flag) = 0; - ccl_fetch(sd, object_flag) = kernel_tex_fetch(__object_flag, - ccl_fetch(sd, object)); + sd->type = isect->type; + sd->flag = 0; + sd->object_flag = kernel_tex_fetch(__object_flag, + sd->object); /* matrices and time */ #ifdef __OBJECT_MOTION__ shader_setup_object_transforms(kg, sd, ray->time); - ccl_fetch(sd, time) = ray->time; + sd->time = ray->time; #endif - ccl_fetch(sd, prim) = kernel_tex_fetch(__prim_index, isect->prim); - ccl_fetch(sd, ray_length) = isect->t; + sd->prim = kernel_tex_fetch(__prim_index, isect->prim); + sd->ray_length = isect->t; #ifdef __UV__ - ccl_fetch(sd, u) = isect->u; - ccl_fetch(sd, v) = isect->v; + sd->u = isect->u; + sd->v = isect->v; #endif #ifdef __HAIR__ - if(ccl_fetch(sd, type) & PRIMITIVE_ALL_CURVE) { + if(sd->type & PRIMITIVE_ALL_CURVE) { /* curve */ - float4 curvedata = kernel_tex_fetch(__curves, ccl_fetch(sd, prim)); + float4 curvedata = kernel_tex_fetch(__curves, sd->prim); - ccl_fetch(sd, shader) = __float_as_int(curvedata.z); - ccl_fetch(sd, P) = bvh_curve_refine(kg, sd, isect, ray); + sd->shader = __float_as_int(curvedata.z); + sd->P = bvh_curve_refine(kg, sd, isect, ray); } else #endif - if(ccl_fetch(sd, type) & PRIMITIVE_TRIANGLE) { + if(sd->type & PRIMITIVE_TRIANGLE) { /* static triangle */ float3 Ng = triangle_normal(kg, sd); - ccl_fetch(sd, shader) = kernel_tex_fetch(__tri_shader, ccl_fetch(sd, prim)); + sd->shader = kernel_tex_fetch(__tri_shader, sd->prim); /* vectors */ - ccl_fetch(sd, P) = triangle_refine(kg, sd, isect, ray); - ccl_fetch(sd, Ng) = Ng; - ccl_fetch(sd, N) = Ng; + sd->P = triangle_refine(kg, sd, isect, ray); + sd->Ng = Ng; + sd->N = Ng; /* smooth normal */ - if(ccl_fetch(sd, shader) & SHADER_SMOOTH_NORMAL) - ccl_fetch(sd, N) = triangle_smooth_normal(kg, ccl_fetch(sd, prim), ccl_fetch(sd, u), ccl_fetch(sd, v)); + if(sd->shader & SHADER_SMOOTH_NORMAL) + sd->N = triangle_smooth_normal(kg, sd->prim, sd->u, sd->v); #ifdef __DPDU__ /* dPdu/dPdv */ - triangle_dPdudv(kg, ccl_fetch(sd, prim), &ccl_fetch(sd, dPdu), &ccl_fetch(sd, dPdv)); + triangle_dPdudv(kg, sd->prim, &sd->dPdu, &sd->dPdv); #endif } else { @@ -111,40 +111,40 @@ ccl_device_noinline void shader_setup_from_ray(KernelGlobals *kg, motion_triangle_shader_setup(kg, sd, isect, ray, false); } - ccl_fetch(sd, I) = -ray->D; + sd->I = -ray->D; - ccl_fetch(sd, flag) |= kernel_tex_fetch(__shader_flag, (ccl_fetch(sd, shader) & SHADER_MASK)*SHADER_SIZE); + sd->flag |= kernel_tex_fetch(__shader_flag, (sd->shader & SHADER_MASK)*SHADER_SIZE); #ifdef __INSTANCING__ if(isect->object != OBJECT_NONE) { /* instance transform */ - object_normal_transform_auto(kg, sd, &ccl_fetch(sd, N)); - object_normal_transform_auto(kg, sd, &ccl_fetch(sd, Ng)); + object_normal_transform_auto(kg, sd, &sd->N); + object_normal_transform_auto(kg, sd, &sd->Ng); # ifdef __DPDU__ - object_dir_transform_auto(kg, sd, &ccl_fetch(sd, dPdu)); - object_dir_transform_auto(kg, sd, &ccl_fetch(sd, dPdv)); + object_dir_transform_auto(kg, sd, &sd->dPdu); + object_dir_transform_auto(kg, sd, &sd->dPdv); # endif } #endif /* backfacing test */ - bool backfacing = (dot(ccl_fetch(sd, Ng), ccl_fetch(sd, I)) < 0.0f); + bool backfacing = (dot(sd->Ng, sd->I) < 0.0f); if(backfacing) { - ccl_fetch(sd, flag) |= SD_BACKFACING; - ccl_fetch(sd, Ng) = -ccl_fetch(sd, Ng); - ccl_fetch(sd, N) = -ccl_fetch(sd, N); + sd->flag |= SD_BACKFACING; + sd->Ng = -sd->Ng; + sd->N = -sd->N; #ifdef __DPDU__ - ccl_fetch(sd, dPdu) = -ccl_fetch(sd, dPdu); - ccl_fetch(sd, dPdv) = -ccl_fetch(sd, dPdv); + sd->dPdu = -sd->dPdu; + sd->dPdv = -sd->dPdv; #endif } #ifdef __RAY_DIFFERENTIALS__ /* differentials */ - differential_transfer(&ccl_fetch(sd, dP), ray->dP, ray->D, ray->dD, ccl_fetch(sd, Ng), isect->t); - differential_incoming(&ccl_fetch(sd, dI), ray->dD); - differential_dudv(&ccl_fetch(sd, du), &ccl_fetch(sd, dv), ccl_fetch(sd, dPdu), ccl_fetch(sd, dPdv), ccl_fetch(sd, dP), ccl_fetch(sd, Ng)); + differential_transfer(&sd->dP, ray->dP, ray->D, ray->dD, sd->Ng, isect->t); + differential_incoming(&sd->dI, ray->dD); + differential_dudv(&sd->du, &sd->dv, sd->dPdu, sd->dPdv, sd->dP, sd->Ng); #endif } @@ -203,11 +203,11 @@ void shader_setup_from_subsurface( # ifdef __INSTANCING__ if(isect->object != OBJECT_NONE) { /* instance transform */ - object_normal_transform(kg, sd, &sd->N); - object_normal_transform(kg, sd, &sd->Ng); + object_normal_transform_auto(kg, sd, &sd->N); + object_normal_transform_auto(kg, sd, &sd->Ng); # ifdef __DPDU__ - object_dir_transform(kg, sd, &sd->dPdu); - object_dir_transform(kg, sd, &sd->dPdv); + object_dir_transform_auto(kg, sd, &sd->dPdu); + object_dir_transform_auto(kg, sd, &sd->dPdv); # endif } # endif @@ -249,106 +249,106 @@ ccl_device_inline void shader_setup_from_sample(KernelGlobals *kg, int lamp) { /* vectors */ - ccl_fetch(sd, P) = P; - ccl_fetch(sd, N) = Ng; - ccl_fetch(sd, Ng) = Ng; - ccl_fetch(sd, I) = I; - ccl_fetch(sd, shader) = shader; + sd->P = P; + sd->N = Ng; + sd->Ng = Ng; + sd->I = I; + sd->shader = shader; if(prim != PRIM_NONE) - ccl_fetch(sd, type) = PRIMITIVE_TRIANGLE; + sd->type = PRIMITIVE_TRIANGLE; else if(lamp != LAMP_NONE) - ccl_fetch(sd, type) = PRIMITIVE_LAMP; + sd->type = PRIMITIVE_LAMP; else - ccl_fetch(sd, type) = PRIMITIVE_NONE; + sd->type = PRIMITIVE_NONE; /* primitive */ #ifdef __INSTANCING__ - ccl_fetch(sd, object) = object; + sd->object = object; #endif /* currently no access to bvh prim index for strand sd->prim*/ - ccl_fetch(sd, prim) = prim; + sd->prim = prim; #ifdef __UV__ - ccl_fetch(sd, u) = u; - ccl_fetch(sd, v) = v; + sd->u = u; + sd->v = v; #endif - ccl_fetch(sd, ray_length) = t; + sd->ray_length = t; - ccl_fetch(sd, flag) = kernel_tex_fetch(__shader_flag, (ccl_fetch(sd, shader) & SHADER_MASK)*SHADER_SIZE); - ccl_fetch(sd, object_flag) = 0; - if(ccl_fetch(sd, object) != OBJECT_NONE) { - ccl_fetch(sd, object_flag) |= kernel_tex_fetch(__object_flag, - ccl_fetch(sd, object)); + sd->flag = kernel_tex_fetch(__shader_flag, (sd->shader & SHADER_MASK)*SHADER_SIZE); + sd->object_flag = 0; + if(sd->object != OBJECT_NONE) { + sd->object_flag |= kernel_tex_fetch(__object_flag, + sd->object); #ifdef __OBJECT_MOTION__ shader_setup_object_transforms(kg, sd, time); - ccl_fetch(sd, time) = time; + sd->time = time; } else if(lamp != LAMP_NONE) { - ccl_fetch(sd, ob_tfm) = lamp_fetch_transform(kg, lamp, false); - ccl_fetch(sd, ob_itfm) = lamp_fetch_transform(kg, lamp, true); + sd->ob_tfm = lamp_fetch_transform(kg, lamp, false); + sd->ob_itfm = lamp_fetch_transform(kg, lamp, true); #endif } /* transform into world space */ if(object_space) { - object_position_transform_auto(kg, sd, &ccl_fetch(sd, P)); - object_normal_transform_auto(kg, sd, &ccl_fetch(sd, Ng)); - ccl_fetch(sd, N) = ccl_fetch(sd, Ng); - object_dir_transform_auto(kg, sd, &ccl_fetch(sd, I)); + object_position_transform_auto(kg, sd, &sd->P); + object_normal_transform_auto(kg, sd, &sd->Ng); + sd->N = sd->Ng; + object_dir_transform_auto(kg, sd, &sd->I); } - if(ccl_fetch(sd, type) & PRIMITIVE_TRIANGLE) { + if(sd->type & PRIMITIVE_TRIANGLE) { /* smooth normal */ - if(ccl_fetch(sd, shader) & SHADER_SMOOTH_NORMAL) { - ccl_fetch(sd, N) = triangle_smooth_normal(kg, ccl_fetch(sd, prim), ccl_fetch(sd, u), ccl_fetch(sd, v)); + if(sd->shader & SHADER_SMOOTH_NORMAL) { + sd->N = triangle_smooth_normal(kg, sd->prim, sd->u, sd->v); #ifdef __INSTANCING__ - if(!(ccl_fetch(sd, object_flag) & SD_OBJECT_TRANSFORM_APPLIED)) { - object_normal_transform_auto(kg, sd, &ccl_fetch(sd, N)); + if(!(sd->object_flag & SD_OBJECT_TRANSFORM_APPLIED)) { + object_normal_transform_auto(kg, sd, &sd->N); } #endif } /* dPdu/dPdv */ #ifdef __DPDU__ - triangle_dPdudv(kg, ccl_fetch(sd, prim), &ccl_fetch(sd, dPdu), &ccl_fetch(sd, dPdv)); + triangle_dPdudv(kg, sd->prim, &sd->dPdu, &sd->dPdv); # ifdef __INSTANCING__ - if(!(ccl_fetch(sd, object_flag) & SD_OBJECT_TRANSFORM_APPLIED)) { - object_dir_transform_auto(kg, sd, &ccl_fetch(sd, dPdu)); - object_dir_transform_auto(kg, sd, &ccl_fetch(sd, dPdv)); + if(!(sd->object_flag & SD_OBJECT_TRANSFORM_APPLIED)) { + object_dir_transform_auto(kg, sd, &sd->dPdu); + object_dir_transform_auto(kg, sd, &sd->dPdv); } # endif #endif } else { #ifdef __DPDU__ - ccl_fetch(sd, dPdu) = make_float3(0.0f, 0.0f, 0.0f); - ccl_fetch(sd, dPdv) = make_float3(0.0f, 0.0f, 0.0f); + sd->dPdu = make_float3(0.0f, 0.0f, 0.0f); + sd->dPdv = make_float3(0.0f, 0.0f, 0.0f); #endif } /* backfacing test */ - if(ccl_fetch(sd, prim) != PRIM_NONE) { - bool backfacing = (dot(ccl_fetch(sd, Ng), ccl_fetch(sd, I)) < 0.0f); + if(sd->prim != PRIM_NONE) { + bool backfacing = (dot(sd->Ng, sd->I) < 0.0f); if(backfacing) { - ccl_fetch(sd, flag) |= SD_BACKFACING; - ccl_fetch(sd, Ng) = -ccl_fetch(sd, Ng); - ccl_fetch(sd, N) = -ccl_fetch(sd, N); + sd->flag |= SD_BACKFACING; + sd->Ng = -sd->Ng; + sd->N = -sd->N; #ifdef __DPDU__ - ccl_fetch(sd, dPdu) = -ccl_fetch(sd, dPdu); - ccl_fetch(sd, dPdv) = -ccl_fetch(sd, dPdv); + sd->dPdu = -sd->dPdu; + sd->dPdv = -sd->dPdv; #endif } } #ifdef __RAY_DIFFERENTIALS__ /* no ray differentials here yet */ - ccl_fetch(sd, dP) = differential3_zero(); - ccl_fetch(sd, dI) = differential3_zero(); - ccl_fetch(sd, du) = differential_zero(); - ccl_fetch(sd, dv) = differential_zero(); + sd->dP = differential3_zero(); + sd->dI = differential3_zero(); + sd->du = differential_zero(); + sd->dv = differential_zero(); #endif } @@ -378,39 +378,39 @@ ccl_device void shader_setup_from_displace(KernelGlobals *kg, ShaderData *sd, ccl_device_inline void shader_setup_from_background(KernelGlobals *kg, ShaderData *sd, const Ray *ray) { /* vectors */ - ccl_fetch(sd, P) = ray->D; - ccl_fetch(sd, N) = -ray->D; - ccl_fetch(sd, Ng) = -ray->D; - ccl_fetch(sd, I) = -ray->D; - ccl_fetch(sd, shader) = kernel_data.background.surface_shader; - ccl_fetch(sd, flag) = kernel_tex_fetch(__shader_flag, (ccl_fetch(sd, shader) & SHADER_MASK)*SHADER_SIZE); - ccl_fetch(sd, object_flag) = 0; + sd->P = ray->D; + sd->N = -ray->D; + sd->Ng = -ray->D; + sd->I = -ray->D; + sd->shader = kernel_data.background.surface_shader; + sd->flag = kernel_tex_fetch(__shader_flag, (sd->shader & SHADER_MASK)*SHADER_SIZE); + sd->object_flag = 0; #ifdef __OBJECT_MOTION__ - ccl_fetch(sd, time) = ray->time; + sd->time = ray->time; #endif - ccl_fetch(sd, ray_length) = 0.0f; + sd->ray_length = 0.0f; #ifdef __INSTANCING__ - ccl_fetch(sd, object) = PRIM_NONE; + sd->object = PRIM_NONE; #endif - ccl_fetch(sd, prim) = PRIM_NONE; + sd->prim = PRIM_NONE; #ifdef __UV__ - ccl_fetch(sd, u) = 0.0f; - ccl_fetch(sd, v) = 0.0f; + sd->u = 0.0f; + sd->v = 0.0f; #endif #ifdef __DPDU__ /* dPdu/dPdv */ - ccl_fetch(sd, dPdu) = make_float3(0.0f, 0.0f, 0.0f); - ccl_fetch(sd, dPdv) = make_float3(0.0f, 0.0f, 0.0f); + sd->dPdu = make_float3(0.0f, 0.0f, 0.0f); + sd->dPdv = make_float3(0.0f, 0.0f, 0.0f); #endif #ifdef __RAY_DIFFERENTIALS__ /* differentials */ - ccl_fetch(sd, dP) = ray->dD; - differential_incoming(&ccl_fetch(sd, dI), ccl_fetch(sd, dP)); - ccl_fetch(sd, du) = differential_zero(); - ccl_fetch(sd, dv) = differential_zero(); + sd->dP = ray->dD; + differential_incoming(&sd->dI, sd->dP); + sd->du = differential_zero(); + sd->dv = differential_zero(); #endif } @@ -505,18 +505,18 @@ ccl_device_inline void _shader_bsdf_multi_eval(KernelGlobals *kg, ShaderData *sd { /* this is the veach one-sample model with balance heuristic, some pdf * factors drop out when using balance heuristic weighting */ - for(int i = 0; i < ccl_fetch(sd, num_closure); i++) { + for(int i = 0; i < sd->num_closure; i++) { if(i == skip_bsdf) continue; - const ShaderClosure *sc = ccl_fetch_array(sd, closure, i); + const ShaderClosure *sc = &sd->closure[i]; if(CLOSURE_IS_BSDF(sc->type)) { float bsdf_pdf = 0.0f; float3 eval = bsdf_eval(kg, sd, sc, omega_in, &bsdf_pdf); if(bsdf_pdf != 0.0f) { - bsdf_eval_accum(result_eval, sc->type, eval*sc->weight); + bsdf_eval_accum(result_eval, sc->type, eval*sc->weight, 1.0f); sum_pdf += bsdf_pdf*sc->sample_weight; } @@ -535,8 +535,8 @@ ccl_device_inline void _shader_bsdf_multi_eval_branched(KernelGlobals *kg, float light_pdf, bool use_mis) { - for(int i = 0; i < ccl_fetch(sd, num_closure); i++) { - const ShaderClosure *sc = ccl_fetch_array(sd, closure, i); + for(int i = 0; i < sd->num_closure; i++) { + const ShaderClosure *sc = &sd->closure[i]; if(CLOSURE_IS_BSDF(sc->type)) { float bsdf_pdf = 0.0f; float3 eval = bsdf_eval(kg, sd, sc, omega_in, &bsdf_pdf); @@ -544,7 +544,8 @@ ccl_device_inline void _shader_bsdf_multi_eval_branched(KernelGlobals *kg, float mis_weight = use_mis? power_heuristic(light_pdf, bsdf_pdf): 1.0f; bsdf_eval_accum(result_eval, sc->type, - eval * sc->weight * mis_weight); + eval * sc->weight, + mis_weight); } } } @@ -576,7 +577,7 @@ void shader_bsdf_eval(KernelGlobals *kg, _shader_bsdf_multi_eval(kg, sd, omega_in, &pdf, -1, eval, 0.0f, 0.0f); if(use_mis) { float weight = power_heuristic(light_pdf, pdf); - bsdf_eval_mul(eval, weight); + bsdf_eval_mis(eval, weight); } } } @@ -591,22 +592,22 @@ ccl_device_inline int shader_bsdf_sample(KernelGlobals *kg, { int sampled = 0; - if(ccl_fetch(sd, num_closure) > 1) { + if(sd->num_closure > 1) { /* pick a BSDF closure based on sample weights */ float sum = 0.0f; - for(sampled = 0; sampled < ccl_fetch(sd, num_closure); sampled++) { - const ShaderClosure *sc = ccl_fetch_array(sd, closure, sampled); + for(sampled = 0; sampled < sd->num_closure; sampled++) { + const ShaderClosure *sc = &sd->closure[sampled]; if(CLOSURE_IS_BSDF(sc->type)) sum += sc->sample_weight; } - float r = ccl_fetch(sd, randb_closure)*sum; + float r = sd->randb_closure*sum; sum = 0.0f; - for(sampled = 0; sampled < ccl_fetch(sd, num_closure); sampled++) { - const ShaderClosure *sc = ccl_fetch_array(sd, closure, sampled); + for(sampled = 0; sampled < sd->num_closure; sampled++) { + const ShaderClosure *sc = &sd->closure[sampled]; if(CLOSURE_IS_BSDF(sc->type)) { sum += sc->sample_weight; @@ -616,13 +617,13 @@ ccl_device_inline int shader_bsdf_sample(KernelGlobals *kg, } } - if(sampled == ccl_fetch(sd, num_closure)) { + if(sampled == sd->num_closure) { *pdf = 0.0f; return LABEL_NONE; } } - const ShaderClosure *sc = ccl_fetch_array(sd, closure, sampled); + const ShaderClosure *sc = &sd->closure[sampled]; int label; float3 eval; @@ -633,7 +634,7 @@ ccl_device_inline int shader_bsdf_sample(KernelGlobals *kg, if(*pdf != 0.0f) { bsdf_eval_init(bsdf_eval, sc->type, eval*sc->weight, kernel_data.film.use_light_pass); - if(ccl_fetch(sd, num_closure) > 1) { + if(sd->num_closure > 1) { float sweight = sc->sample_weight; _shader_bsdf_multi_eval(kg, sd, *omega_in, pdf, sampled, bsdf_eval, *pdf*sweight, sweight); } @@ -660,8 +661,8 @@ ccl_device int shader_bsdf_sample_closure(KernelGlobals *kg, ShaderData *sd, ccl_device void shader_bsdf_blur(KernelGlobals *kg, ShaderData *sd, float roughness) { - for(int i = 0; i < ccl_fetch(sd, num_closure); i++) { - ShaderClosure *sc = ccl_fetch_array(sd, closure, i); + for(int i = 0; i < sd->num_closure; i++) { + ShaderClosure *sc = &sd->closure[i]; if(CLOSURE_IS_BSDF(sc->type)) bsdf_blur(kg, sc, roughness); @@ -670,13 +671,13 @@ ccl_device void shader_bsdf_blur(KernelGlobals *kg, ShaderData *sd, float roughn ccl_device float3 shader_bsdf_transparency(KernelGlobals *kg, ShaderData *sd) { - if(ccl_fetch(sd, flag) & SD_HAS_ONLY_VOLUME) + if(sd->flag & SD_HAS_ONLY_VOLUME) return make_float3(1.0f, 1.0f, 1.0f); float3 eval = make_float3(0.0f, 0.0f, 0.0f); - for(int i = 0; i < ccl_fetch(sd, num_closure); i++) { - ShaderClosure *sc = ccl_fetch_array(sd, closure, i); + for(int i = 0; i < sd->num_closure; i++) { + ShaderClosure *sc = &sd->closure[i]; if(sc->type == CLOSURE_BSDF_TRANSPARENT_ID) // todo: make this work for osl eval += sc->weight; @@ -685,6 +686,18 @@ ccl_device float3 shader_bsdf_transparency(KernelGlobals *kg, ShaderData *sd) return eval; } +ccl_device void shader_bsdf_disable_transparency(KernelGlobals *kg, ShaderData *sd) +{ + for(int i = 0; i < sd->num_closure; i++) { + ShaderClosure *sc = &sd->closure[i]; + + if(sc->type == CLOSURE_BSDF_TRANSPARENT_ID) { + sc->sample_weight = 0.0f; + sc->weight = make_float3(0.0f, 0.0f, 0.0f); + } + } +} + ccl_device float3 shader_bsdf_alpha(KernelGlobals *kg, ShaderData *sd) { float3 alpha = make_float3(1.0f, 1.0f, 1.0f) - shader_bsdf_transparency(kg, sd); @@ -699,8 +712,8 @@ ccl_device float3 shader_bsdf_diffuse(KernelGlobals *kg, ShaderData *sd) { float3 eval = make_float3(0.0f, 0.0f, 0.0f); - for(int i = 0; i < ccl_fetch(sd, num_closure); i++) { - ShaderClosure *sc = ccl_fetch_array(sd, closure, i); + for(int i = 0; i < sd->num_closure; i++) { + ShaderClosure *sc = &sd->closure[i]; if(CLOSURE_IS_BSDF_DIFFUSE(sc->type)) eval += sc->weight; @@ -713,8 +726,8 @@ ccl_device float3 shader_bsdf_glossy(KernelGlobals *kg, ShaderData *sd) { float3 eval = make_float3(0.0f, 0.0f, 0.0f); - for(int i = 0; i < ccl_fetch(sd, num_closure); i++) { - ShaderClosure *sc = ccl_fetch_array(sd, closure, i); + for(int i = 0; i < sd->num_closure; i++) { + ShaderClosure *sc = &sd->closure[i]; if(CLOSURE_IS_BSDF_GLOSSY(sc->type)) eval += sc->weight; @@ -727,8 +740,8 @@ ccl_device float3 shader_bsdf_transmission(KernelGlobals *kg, ShaderData *sd) { float3 eval = make_float3(0.0f, 0.0f, 0.0f); - for(int i = 0; i < ccl_fetch(sd, num_closure); i++) { - ShaderClosure *sc = ccl_fetch_array(sd, closure, i); + for(int i = 0; i < sd->num_closure; i++) { + ShaderClosure *sc = &sd->closure[i]; if(CLOSURE_IS_BSDF_TRANSMISSION(sc->type)) eval += sc->weight; @@ -741,8 +754,8 @@ ccl_device float3 shader_bsdf_subsurface(KernelGlobals *kg, ShaderData *sd) { float3 eval = make_float3(0.0f, 0.0f, 0.0f); - for(int i = 0; i < ccl_fetch(sd, num_closure); i++) { - ShaderClosure *sc = ccl_fetch_array(sd, closure, i); + for(int i = 0; i < sd->num_closure; i++) { + ShaderClosure *sc = &sd->closure[i]; if(CLOSURE_IS_BSSRDF(sc->type) || CLOSURE_IS_BSDF_BSSRDF(sc->type)) eval += sc->weight; @@ -756,8 +769,8 @@ ccl_device float3 shader_bsdf_ao(KernelGlobals *kg, ShaderData *sd, float ao_fac float3 eval = make_float3(0.0f, 0.0f, 0.0f); float3 N = make_float3(0.0f, 0.0f, 0.0f); - for(int i = 0; i < ccl_fetch(sd, num_closure); i++) { - ShaderClosure *sc = ccl_fetch_array(sd, closure, i); + for(int i = 0; i < sd->num_closure; i++) { + ShaderClosure *sc = &sd->closure[i]; if(CLOSURE_IS_BSDF_DIFFUSE(sc->type)) { const DiffuseBsdf *bsdf = (const DiffuseBsdf*)sc; @@ -766,12 +779,12 @@ ccl_device float3 shader_bsdf_ao(KernelGlobals *kg, ShaderData *sd, float ao_fac } else if(CLOSURE_IS_AMBIENT_OCCLUSION(sc->type)) { eval += sc->weight; - N += ccl_fetch(sd, N)*average(sc->weight); + N += sd->N*average(sc->weight); } } if(is_zero(N)) - N = ccl_fetch(sd, N); + N = sd->N; else N = normalize(N); @@ -786,8 +799,8 @@ ccl_device float3 shader_bssrdf_sum(ShaderData *sd, float3 *N_, float *texture_b float3 N = make_float3(0.0f, 0.0f, 0.0f); float texture_blur = 0.0f, weight_sum = 0.0f; - for(int i = 0; i < ccl_fetch(sd, num_closure); i++) { - ShaderClosure *sc = ccl_fetch_array(sd, closure, i); + for(int i = 0; i < sd->num_closure; i++) { + ShaderClosure *sc = &sd->closure[i]; if(CLOSURE_IS_BSSRDF(sc->type)) { const Bssrdf *bssrdf = (const Bssrdf*)sc; @@ -801,10 +814,10 @@ ccl_device float3 shader_bssrdf_sum(ShaderData *sd, float3 *N_, float *texture_b } if(N_) - *N_ = (is_zero(N))? ccl_fetch(sd, N): normalize(N); + *N_ = (is_zero(N))? sd->N: normalize(N); if(texture_blur_) - *texture_blur_ = texture_blur/weight_sum; + *texture_blur_ = safe_divide(texture_blur, weight_sum); return eval; } @@ -814,7 +827,7 @@ ccl_device float3 shader_bssrdf_sum(ShaderData *sd, float3 *N_, float *texture_b ccl_device float3 emissive_eval(KernelGlobals *kg, ShaderData *sd, ShaderClosure *sc) { - return emissive_simple_eval(ccl_fetch(sd, Ng), ccl_fetch(sd, I)); + return emissive_simple_eval(sd->Ng, sd->I); } ccl_device float3 shader_emissive_eval(KernelGlobals *kg, ShaderData *sd) @@ -822,8 +835,8 @@ ccl_device float3 shader_emissive_eval(KernelGlobals *kg, ShaderData *sd) float3 eval; eval = make_float3(0.0f, 0.0f, 0.0f); - for(int i = 0; i < ccl_fetch(sd, num_closure); i++) { - ShaderClosure *sc = ccl_fetch_array(sd, closure, i); + for(int i = 0; i < sd->num_closure; i++) { + ShaderClosure *sc = &sd->closure[i]; if(CLOSURE_IS_EMISSION(sc->type)) eval += emissive_eval(kg, sd, sc)*sc->weight; @@ -838,8 +851,8 @@ ccl_device float3 shader_holdout_eval(KernelGlobals *kg, ShaderData *sd) { float3 weight = make_float3(0.0f, 0.0f, 0.0f); - for(int i = 0; i < ccl_fetch(sd, num_closure); i++) { - ShaderClosure *sc = ccl_fetch_array(sd, closure, i); + for(int i = 0; i < sd->num_closure; i++) { + ShaderClosure *sc = &sd->closure[i]; if(CLOSURE_IS_HOLDOUT(sc->type)) weight += sc->weight; @@ -850,12 +863,12 @@ ccl_device float3 shader_holdout_eval(KernelGlobals *kg, ShaderData *sd) /* Surface Evaluation */ -ccl_device void shader_eval_surface(KernelGlobals *kg, ShaderData *sd, ccl_addr_space RNG *rng, +ccl_device void shader_eval_surface(KernelGlobals *kg, ShaderData *sd, RNG *rng, ccl_addr_space PathState *state, float randb, int path_flag, ShaderContext ctx) { - ccl_fetch(sd, num_closure) = 0; - ccl_fetch(sd, num_closure_extra) = 0; - ccl_fetch(sd, randb_closure) = randb; + sd->num_closure = 0; + sd->num_closure_extra = 0; + sd->randb_closure = randb; #ifdef __OSL__ if(kg->osl) @@ -869,13 +882,13 @@ ccl_device void shader_eval_surface(KernelGlobals *kg, ShaderData *sd, ccl_addr_ DiffuseBsdf *bsdf = (DiffuseBsdf*)bsdf_alloc(sd, sizeof(DiffuseBsdf), make_float3(0.8f, 0.8f, 0.8f)); - bsdf->N = ccl_fetch(sd, N); - ccl_fetch(sd, flag) |= bsdf_diffuse_setup(bsdf); + bsdf->N = sd->N; + sd->flag |= bsdf_diffuse_setup(bsdf); #endif } - if(rng && (ccl_fetch(sd, flag) & SD_BSDF_NEEDS_LCG)) { - ccl_fetch(sd, lcg_state) = lcg_state_init_addrspace(rng, state, 0xb4bc3953); + if(rng && (sd->flag & SD_BSDF_NEEDS_LCG)) { + sd->lcg_state = lcg_state_init(rng, state->rng_offset, state->sample, 0xb4bc3953); } } @@ -884,9 +897,9 @@ ccl_device void shader_eval_surface(KernelGlobals *kg, ShaderData *sd, ccl_addr_ ccl_device float3 shader_eval_background(KernelGlobals *kg, ShaderData *sd, ccl_addr_space PathState *state, int path_flag, ShaderContext ctx) { - ccl_fetch(sd, num_closure) = 0; - ccl_fetch(sd, num_closure_extra) = 0; - ccl_fetch(sd, randb_closure) = 0.0f; + sd->num_closure = 0; + sd->num_closure_extra = 0; + sd->randb_closure = 0.0f; #ifdef __SVM__ #ifdef __OSL__ @@ -901,8 +914,8 @@ ccl_device float3 shader_eval_background(KernelGlobals *kg, ShaderData *sd, float3 eval = make_float3(0.0f, 0.0f, 0.0f); - for(int i = 0; i < ccl_fetch(sd, num_closure); i++) { - const ShaderClosure *sc = ccl_fetch_array(sd, closure, i); + for(int i = 0; i < sd->num_closure; i++) { + const ShaderClosure *sc = &sd->closure[i]; if(CLOSURE_IS_BACKGROUND(sc->type)) eval += sc->weight; @@ -932,7 +945,7 @@ ccl_device_inline void _shader_volume_phase_multi_eval(const ShaderData *sd, con float3 eval = volume_phase_eval(sd, sc, omega_in, &phase_pdf); if(phase_pdf != 0.0f) { - bsdf_eval_accum(result_eval, sc->type, eval); + bsdf_eval_accum(result_eval, sc->type, eval, 1.0f); sum_pdf += phase_pdf*sc->sample_weight; } @@ -1024,8 +1037,8 @@ ccl_device int shader_phase_sample_closure(KernelGlobals *kg, const ShaderData * ccl_device_inline void shader_eval_volume(KernelGlobals *kg, ShaderData *sd, - PathState *state, - VolumeStack *stack, + ccl_addr_space PathState *state, + ccl_addr_space VolumeStack *stack, int path_flag, ShaderContext ctx) { @@ -1081,9 +1094,9 @@ ccl_device_inline void shader_eval_volume(KernelGlobals *kg, ccl_device void shader_eval_displacement(KernelGlobals *kg, ShaderData *sd, ccl_addr_space PathState *state, ShaderContext ctx) { - ccl_fetch(sd, num_closure) = 0; - ccl_fetch(sd, num_closure_extra) = 0; - ccl_fetch(sd, randb_closure) = 0.0f; + sd->num_closure = 0; + sd->num_closure_extra = 0; + sd->randb_closure = 0.0f; /* this will modify sd->P */ #ifdef __SVM__ diff --git a/intern/cycles/kernel/kernel_shadow.h b/intern/cycles/kernel/kernel_shadow.h index 2981f6ac566..0426e0a62c9 100644 --- a/intern/cycles/kernel/kernel_shadow.h +++ b/intern/cycles/kernel/kernel_shadow.h @@ -16,9 +16,84 @@ CCL_NAMESPACE_BEGIN -#ifdef __SHADOW_RECORD_ALL__ +/* Attenuate throughput accordingly to the given intersection event. + * Returns true if the throughput is zero and traversal can be aborted. + */ +ccl_device_forceinline bool shadow_handle_transparent_isect( + KernelGlobals *kg, + ShaderData *shadow_sd, + ccl_addr_space PathState *state, +# ifdef __VOLUME__ + ccl_addr_space struct PathState *volume_state, +# endif + Intersection *isect, + Ray *ray, + float3 *throughput) +{ +#ifdef __VOLUME__ + /* Attenuation between last surface and next surface. */ + if(volume_state->volume_stack[0].shader != SHADER_NONE) { + Ray segment_ray = *ray; + segment_ray.t = isect->t; + kernel_volume_shadow(kg, + shadow_sd, + volume_state, + &segment_ray, + throughput); + } +#endif + /* Setup shader data at surface. */ + shader_setup_from_ray(kg, shadow_sd, isect, ray); + /* Attenuation from transparent surface. */ + if(!(shadow_sd->flag & SD_HAS_ONLY_VOLUME)) { + path_state_modify_bounce(state, true); + shader_eval_surface(kg, + shadow_sd, + NULL, + state, + 0.0f, + PATH_RAY_SHADOW, + SHADER_CONTEXT_SHADOW); + path_state_modify_bounce(state, false); + *throughput *= shader_bsdf_transparency(kg, shadow_sd); + } + /* Stop if all light is blocked. */ + if(is_zero(*throughput)) { + return true; + } +#ifdef __VOLUME__ + /* Exit/enter volume. */ + kernel_volume_stack_enter_exit(kg, shadow_sd, volume_state->volume_stack); +#endif + return false; +} + +/* Special version which only handles opaque shadows. */ +ccl_device bool shadow_blocked_opaque(KernelGlobals *kg, + ShaderData *shadow_sd, + ccl_addr_space PathState *state, + Ray *ray, + Intersection *isect, + float3 *shadow) +{ + const bool blocked = scene_intersect(kg, + *ray, + PATH_RAY_SHADOW_OPAQUE, + isect, + NULL, + 0.0f, 0.0f); +#ifdef __VOLUME__ + if(!blocked && state->volume_stack[0].shader != SHADER_NONE) { + /* Apply attenuation from current volume shader. */ + kernel_volume_shadow(kg, shadow_sd, state, ray, shadow); + } +#endif + return blocked; +} -/* Shadow function to compute how much light is blocked, CPU variation. +#ifdef __TRANSPARENT_SHADOWS__ +# ifdef __SHADOW_RECORD_ALL__ +/* Shadow function to compute how much light is blocked, * * We trace a single ray. If it hits any opaque surface, or more than a given * number of transparent surfaces is hit, then we consider the geometry to be @@ -36,261 +111,403 @@ CCL_NAMESPACE_BEGIN * or there is a performance increase anyway due to avoiding the need to send * two rays with transparent shadows. * - * This is CPU only because of qsort, and malloc or high stack space usage to - * record all these intersections. */ + * On CPU it'll handle all transparent bounces (by allocating storage for + * intersections when they don't fit into the stack storage). + * + * On GPU it'll only handle SHADOW_STACK_MAX_HITS-1 intersections, so this + * is something to be kept an eye on. + */ -#define STACK_MAX_HITS 64 +# define SHADOW_STACK_MAX_HITS 64 -ccl_device_inline bool shadow_blocked(KernelGlobals *kg, ShaderData *shadow_sd, PathState *state, Ray *ray, float3 *shadow) +/* Actual logic with traversal loop implementation which is free from device + * specific tweaks. + * + * Note that hits array should be as big as max_hits+1. + */ +ccl_device bool shadow_blocked_transparent_all_loop(KernelGlobals *kg, + ShaderData *shadow_sd, + ccl_addr_space PathState *state, + const int skip_object, + Ray *ray, + Intersection *hits, + uint max_hits, + float3 *shadow) { - *shadow = make_float3(1.0f, 1.0f, 1.0f); - - if(ray->t == 0.0f) - return false; - - bool blocked; - - if(kernel_data.integrator.transparent_shadows) { - /* check transparent bounces here, for volume scatter which can do - * lighting before surface path termination is checked */ - if(state->transparent_bounce >= kernel_data.integrator.transparent_max_bounce) - return true; - - /* intersect to find an opaque surface, or record all transparent surface hits */ - Intersection hits_stack[STACK_MAX_HITS]; - Intersection *hits = hits_stack; - const int transparent_max_bounce = kernel_data.integrator.transparent_max_bounce; - uint max_hits = transparent_max_bounce - state->transparent_bounce - 1; - - /* prefer to use stack but use dynamic allocation if too deep max hits - * we need max_hits + 1 storage space due to the logic in - * scene_intersect_shadow_all which will first store and then check if - * the limit is exceeded */ - if(max_hits + 1 > STACK_MAX_HITS) { - if(kg->transparent_shadow_intersections == NULL) { - kg->transparent_shadow_intersections = - (Intersection*)malloc(sizeof(Intersection)*(transparent_max_bounce + 1)); + /* Intersect to find an opaque surface, or record all transparent + * surface hits. + */ + uint num_hits; + const bool blocked = scene_intersect_shadow_all(kg, + ray, + hits, + skip_object, + max_hits, + &num_hits); + /* If no opaque surface found but we did find transparent hits, + * shade them. + */ + if(!blocked && num_hits > 0) { + float3 throughput = make_float3(1.0f, 1.0f, 1.0f); + float3 Pend = ray->P + ray->D*ray->t; + float last_t = 0.0f; + int bounce = state->transparent_bounce; + Intersection *isect = hits; +# ifdef __VOLUME__ +# ifdef __SPLIT_KERNEL__ + ccl_addr_space PathState *ps = &kernel_split_state.state_shadow[ccl_global_id(1) * ccl_global_size(0) + ccl_global_id(0)]; +# else + PathState ps_object; + PathState *ps = &ps_object; +# endif + *ps = *state; +# endif + sort_intersections(hits, num_hits); + for(int hit = 0; hit < num_hits; hit++, isect++) { + /* Adjust intersection distance for moving ray forward. */ + float new_t = isect->t; + isect->t -= last_t; + /* Skip hit if we did not move forward, step by step raytracing + * would have skipped it as well then. + */ + if(last_t == new_t) { + continue; } - hits = kg->transparent_shadow_intersections; - } - - uint num_hits; - blocked = scene_intersect_shadow_all(kg, ray, hits, max_hits, &num_hits); - - /* if no opaque surface found but we did find transparent hits, shade them */ - if(!blocked && num_hits > 0) { - float3 throughput = make_float3(1.0f, 1.0f, 1.0f); - float3 Pend = ray->P + ray->D*ray->t; - float last_t = 0.0f; - int bounce = state->transparent_bounce; - Intersection *isect = hits; + last_t = new_t; + /* Attenuate the throughput. */ + if(shadow_handle_transparent_isect(kg, + shadow_sd, + state, #ifdef __VOLUME__ - PathState ps = *state; + ps, #endif - - qsort(hits, num_hits, sizeof(Intersection), intersections_compare); - - for(int hit = 0; hit < num_hits; hit++, isect++) { - /* adjust intersection distance for moving ray forward */ - float new_t = isect->t; - isect->t -= last_t; - - /* skip hit if we did not move forward, step by step raytracing - * would have skipped it as well then */ - if(last_t == new_t) - continue; - - last_t = new_t; - -#ifdef __VOLUME__ - /* attenuation between last surface and next surface */ - if(ps.volume_stack[0].shader != SHADER_NONE) { - Ray segment_ray = *ray; - segment_ray.t = isect->t; - kernel_volume_shadow(kg, shadow_sd, &ps, &segment_ray, &throughput); - } -#endif - - /* setup shader data at surface */ - shader_setup_from_ray(kg, shadow_sd, isect, ray); - - /* attenuation from transparent surface */ - if(!(shadow_sd->flag & SD_HAS_ONLY_VOLUME)) { - path_state_modify_bounce(state, true); - shader_eval_surface(kg, shadow_sd, NULL, state, 0.0f, PATH_RAY_SHADOW, SHADER_CONTEXT_SHADOW); - path_state_modify_bounce(state, false); - - throughput *= shader_bsdf_transparency(kg, shadow_sd); - } - - /* stop if all light is blocked */ - if(is_zero(throughput)) { - return true; - } - - /* move ray forward */ - ray->P = shadow_sd->P; - if(ray->t != FLT_MAX) { - ray->D = normalize_len(Pend - ray->P, &ray->t); - } - -#ifdef __VOLUME__ - /* exit/enter volume */ - kernel_volume_stack_enter_exit(kg, shadow_sd, ps.volume_stack); -#endif - - bounce++; + isect, + ray, + &throughput)) + { + return true; } - -#ifdef __VOLUME__ - /* attenuation for last line segment towards light */ - if(ps.volume_stack[0].shader != SHADER_NONE) - kernel_volume_shadow(kg, shadow_sd, &ps, ray, &throughput); -#endif - - *shadow = throughput; - - return is_zero(throughput); + /* Move ray forward. */ + ray->P = shadow_sd->P; + if(ray->t != FLT_MAX) { + ray->D = normalize_len(Pend - ray->P, &ray->t); + } + bounce++; } +# ifdef __VOLUME__ + /* Attenuation for last line segment towards light. */ + if(ps->volume_stack[0].shader != SHADER_NONE) { + kernel_volume_shadow(kg, shadow_sd, ps, ray, &throughput); + } +# endif + *shadow = throughput; + return is_zero(throughput); } - else { - Intersection isect; - blocked = scene_intersect(kg, *ray, PATH_RAY_SHADOW_OPAQUE, &isect, NULL, 0.0f, 0.0f); - } - -#ifdef __VOLUME__ +# ifdef __VOLUME__ if(!blocked && state->volume_stack[0].shader != SHADER_NONE) { - /* apply attenuation from current volume shader */ + /* Apply attenuation from current volume shader/ */ kernel_volume_shadow(kg, shadow_sd, state, ray, shadow); } -#endif - +# endif return blocked; } -#undef STACK_MAX_HITS - -#else +/* Here we do all device specific trickery before invoking actual traversal + * loop to help readability of the actual logic. + */ +ccl_device bool shadow_blocked_transparent_all(KernelGlobals *kg, + ShaderData *shadow_sd, + ccl_addr_space PathState *state, + const int skip_object, + Ray *ray, + uint max_hits, + float3 *shadow) +{ +# ifdef __SPLIT_KERNEL__ + Intersection hits_[SHADOW_STACK_MAX_HITS]; + Intersection *hits = &hits_[0]; +# elif defined(__KERNEL_CUDA__) + Intersection *hits = kg->hits_stack; +# else + Intersection hits_stack[SHADOW_STACK_MAX_HITS]; + Intersection *hits = hits_stack; +# endif +# ifndef __KERNEL_GPU__ + /* Prefer to use stack but use dynamic allocation if too deep max hits + * we need max_hits + 1 storage space due to the logic in + * scene_intersect_shadow_all which will first store and then check if + * the limit is exceeded. + * + * Ignore this on GPU because of slow/unavailable malloc(). + */ + if(max_hits + 1 > SHADOW_STACK_MAX_HITS) { + if(kg->transparent_shadow_intersections == NULL) { + const int transparent_max_bounce = kernel_data.integrator.transparent_max_bounce; + kg->transparent_shadow_intersections = + (Intersection*)malloc(sizeof(Intersection)*(transparent_max_bounce + 1)); + } + hits = kg->transparent_shadow_intersections; + } +# endif /* __KERNEL_GPU__ */ + /* Invoke actual traversal. */ + return shadow_blocked_transparent_all_loop(kg, + shadow_sd, + state, + skip_object, + ray, + hits, + max_hits, + shadow); +} +# endif /* __SHADOW_RECORD_ALL__ */ -/* Shadow function to compute how much light is blocked, GPU variation. +# if defined(__KERNEL_GPU__) || !defined(__SHADOW_RECORD_ALL__) +/* Shadow function to compute how much light is blocked, * * Here we raytrace from one transparent surface to the next step by step. * To minimize overhead in cases where we don't need transparent shadows, we * first trace a regular shadow ray. We check if the hit primitive was * potentially transparent, and only in that case start marching. this gives - * one extra ray cast for the cases were we do want transparency. */ + * one extra ray cast for the cases were we do want transparency. + */ -ccl_device_noinline bool shadow_blocked(KernelGlobals *kg, - ShaderData *shadow_sd, - ccl_addr_space PathState *state, - ccl_addr_space Ray *ray_input, - float3 *shadow) +/* This function is only implementing device-independent traversal logic + * which requires some precalculation done. + */ +ccl_device bool shadow_blocked_transparent_stepped_loop( + KernelGlobals *kg, + ShaderData *shadow_sd, + ccl_addr_space PathState *state, + const int skip_object, + Ray *ray, + Intersection *isect, + const bool blocked, + const bool is_transparent_isect, + float3 *shadow) { - *shadow = make_float3(1.0f, 1.0f, 1.0f); - - if(ray_input->t == 0.0f) - return false; - -#ifdef __SPLIT_KERNEL__ - Ray private_ray = *ray_input; - Ray *ray = &private_ray; -#else - Ray *ray = ray_input; -#endif - -#ifdef __SPLIT_KERNEL__ - Intersection *isect = &kg->isect_shadow[SD_THREAD]; -#else - Intersection isect_object; - Intersection *isect = &isect_object; -#endif - - bool blocked = scene_intersect(kg, *ray, PATH_RAY_SHADOW_OPAQUE, isect, NULL, 0.0f, 0.0f); - -#ifdef __TRANSPARENT_SHADOWS__ - if(blocked && kernel_data.integrator.transparent_shadows) { - if(shader_transparent_shadow(kg, isect)) { - float3 throughput = make_float3(1.0f, 1.0f, 1.0f); - float3 Pend = ray->P + ray->D*ray->t; - int bounce = state->transparent_bounce; -#ifdef __VOLUME__ - PathState ps = *state; -#endif - - for(;;) { - if(bounce >= kernel_data.integrator.transparent_max_bounce) - return true; - - if(!scene_intersect(kg, *ray, PATH_RAY_SHADOW_TRANSPARENT, isect, NULL, 0.0f, 0.0f)) - { -#ifdef __VOLUME__ - /* attenuation for last line segment towards light */ - if(ps.volume_stack[0].shader != SHADER_NONE) - kernel_volume_shadow(kg, shadow_sd, &ps, ray, &throughput); -#endif - - *shadow *= throughput; - - return false; - } - - if(!shader_transparent_shadow(kg, isect)) { - return true; - } - -#ifdef __VOLUME__ - /* attenuation between last surface and next surface */ - if(ps.volume_stack[0].shader != SHADER_NONE) { - Ray segment_ray = *ray; - segment_ray.t = isect->t; - kernel_volume_shadow(kg, shadow_sd, &ps, &segment_ray, &throughput); + if((blocked && is_transparent_isect) || skip_object != OBJECT_NONE) { + float3 throughput = make_float3(1.0f, 1.0f, 1.0f); + float3 Pend = ray->P + ray->D*ray->t; + int bounce = state->transparent_bounce; +# ifdef __VOLUME__ +# ifdef __SPLIT_KERNEL__ + ccl_addr_space PathState *ps = &kernel_split_state.state_shadow[ccl_global_id(1) * ccl_global_size(0) + ccl_global_id(0)]; +# else + PathState ps_object; + PathState *ps = &ps_object; +# endif + *ps = *state; +# endif + for(;;) { + if(bounce >= kernel_data.integrator.transparent_max_bounce) { + return true; + } + if(!scene_intersect(kg, + *ray, + PATH_RAY_SHADOW_TRANSPARENT, + isect, + NULL, + 0.0f, 0.0f)) + { + break; + } +#ifdef __SHADOW_TRICKS__ + if(skip_object != OBJECT_NONE) { + const int isect_object = (isect->object == PRIM_NONE) + ? kernel_tex_fetch(__prim_object, isect->prim) + : isect->object; + if(isect_object == skip_object) { + shader_setup_from_ray(kg, shadow_sd, isect, ray); + /* Move ray forward. */ + ray->P = ray_offset(shadow_sd->P, -shadow_sd->Ng); + if(ray->t != FLT_MAX) { + ray->D = normalize_len(Pend - ray->P, &ray->t); + } + bounce++; + continue; } + } #endif - - /* setup shader data at surface */ - shader_setup_from_ray(kg, shadow_sd, isect, ray); - - /* attenuation from transparent surface */ - if(!(ccl_fetch(shadow_sd, flag) & SD_HAS_ONLY_VOLUME)) { - path_state_modify_bounce(state, true); - shader_eval_surface(kg, shadow_sd, NULL, state, 0.0f, PATH_RAY_SHADOW, SHADER_CONTEXT_SHADOW); - path_state_modify_bounce(state, false); - - throughput *= shader_bsdf_transparency(kg, shadow_sd); - } - - /* stop if all light is blocked */ - if(is_zero(throughput)) { - return true; - } - - /* move ray forward */ - ray->P = ray_offset(ccl_fetch(shadow_sd, P), -ccl_fetch(shadow_sd, Ng)); - if(ray->t != FLT_MAX) { - ray->D = normalize_len(Pend - ray->P, &ray->t); - } - + if(!shader_transparent_shadow(kg, isect)) { + return true; + } + /* Attenuate the throughput. */ + if(shadow_handle_transparent_isect(kg, + shadow_sd, + state, #ifdef __VOLUME__ - /* exit/enter volume */ - kernel_volume_stack_enter_exit(kg, shadow_sd, ps.volume_stack); + ps, #endif - - bounce++; + isect, + ray, + &throughput)) + { + return true; } + /* Move ray forward. */ + ray->P = ray_offset(shadow_sd->P, -shadow_sd->Ng); + if(ray->t != FLT_MAX) { + ray->D = normalize_len(Pend - ray->P, &ray->t); + } + bounce++; + } +# ifdef __VOLUME__ + /* Attenuation for last line segment towards light. */ + if(ps->volume_stack[0].shader != SHADER_NONE) { + kernel_volume_shadow(kg, shadow_sd, ps, ray, &throughput); } +# endif + *shadow *= throughput; + return is_zero(throughput); } -#ifdef __VOLUME__ - else if(!blocked && state->volume_stack[0].shader != SHADER_NONE) { - /* apply attenuation from current volume shader */ +# ifdef __VOLUME__ + if(!blocked && state->volume_stack[0].shader != SHADER_NONE) { + /* Apply attenuation from current volume shader. */ kernel_volume_shadow(kg, shadow_sd, state, ray, shadow); } -#endif -#endif - +# endif return blocked; } +ccl_device bool shadow_blocked_transparent_stepped( + KernelGlobals *kg, + ShaderData *shadow_sd, + ccl_addr_space PathState *state, + const int skip_object, + Ray *ray, + Intersection *isect, + float3 *shadow) +{ + bool blocked, is_transparent_isect; + if (skip_object == OBJECT_NONE) { + blocked = scene_intersect(kg, + *ray, + PATH_RAY_SHADOW_OPAQUE, + isect, + NULL, + 0.0f, 0.0f); + is_transparent_isect = blocked + ? shader_transparent_shadow(kg, isect) + : false; + } + else { + blocked = false; + is_transparent_isect = false; + } + return shadow_blocked_transparent_stepped_loop(kg, + shadow_sd, + state, + skip_object, + ray, + isect, + blocked, + is_transparent_isect, + shadow); +} + +# endif /* __KERNEL_GPU__ || !__SHADOW_RECORD_ALL__ */ +#endif /* __TRANSPARENT_SHADOWS__ */ + +ccl_device_inline bool shadow_blocked(KernelGlobals *kg, + ShaderData *shadow_sd, + ccl_addr_space PathState *state, + Ray *ray_input, + float3 *shadow) +{ + Ray *ray = ray_input; + Intersection isect; + /* Some common early checks. */ + *shadow = make_float3(1.0f, 1.0f, 1.0f); + if(ray->t == 0.0f) { + return false; + } +#ifdef __SHADOW_TRICKS__ + const int skip_object = state->catcher_object; +#else + const int skip_object = OBJECT_NONE; #endif + /* Do actual shadow shading. */ + /* First of all, we check if integrator requires transparent shadows. + * if not, we use simplest and fastest ever way to calculate occlusion. + * + * NOTE: We can't do quick opaque test here if we are on shadow-catcher + * path because we don't want catcher object to be casting shadow here. + */ +#ifdef __TRANSPARENT_SHADOWS__ + if(!kernel_data.integrator.transparent_shadows && + skip_object == OBJECT_NONE) +#endif + { + return shadow_blocked_opaque(kg, + shadow_sd, + state, + ray, + &isect, + shadow); + } +#ifdef __TRANSPARENT_SHADOWS__ +# ifdef __SHADOW_RECORD_ALL__ + /* For the transparent shadows we try to use record-all logic on the + * devices which supports this. + */ + const int transparent_max_bounce = kernel_data.integrator.transparent_max_bounce; + /* Check transparent bounces here, for volume scatter which can do + * lighting before surface path termination is checked. + */ + if(state->transparent_bounce >= transparent_max_bounce) { + return true; + } + const uint max_hits = transparent_max_bounce - state->transparent_bounce - 1; +# ifdef __KERNEL_GPU__ + /* On GPU we do trickey with tracing opaque ray first, this avoids speed + * regressions in some files. + * + * TODO(sergey): Check why using record-all behavior causes slowdown in such + * cases. Could that be caused by a higher spill pressure? + */ + const bool blocked = scene_intersect(kg, + *ray, + PATH_RAY_SHADOW_OPAQUE, + &isect, + NULL, + 0.0f, 0.0f); + const bool is_transparent_isect = blocked + ? shader_transparent_shadow(kg, &isect) + : false; + if(!blocked || !is_transparent_isect || + max_hits + 1 >= SHADOW_STACK_MAX_HITS) + { + return shadow_blocked_transparent_stepped_loop(kg, + shadow_sd, + state, + skip_object, + ray, + &isect, + blocked, + is_transparent_isect, + shadow); + } +# endif /* __KERNEL_GPU__ */ + return shadow_blocked_transparent_all(kg, + shadow_sd, + state, + skip_object, + ray, + max_hits, + shadow); +# else /* __SHADOW_RECORD_ALL__ */ + /* Fallback to a slowest version which works on all devices. */ + return shadow_blocked_transparent_stepped(kg, + shadow_sd, + state, + skip_object, + ray, + &isect, + shadow); +# endif /* __SHADOW_RECORD_ALL__ */ +#endif /* __TRANSPARENT_SHADOWS__ */ +} -CCL_NAMESPACE_END +#undef SHADOW_STACK_MAX_HITS +CCL_NAMESPACE_END diff --git a/intern/cycles/kernel/kernel_subsurface.h b/intern/cycles/kernel/kernel_subsurface.h index c5652ebf7dc..6c8b7cca4ce 100644 --- a/intern/cycles/kernel/kernel_subsurface.h +++ b/intern/cycles/kernel/kernel_subsurface.h @@ -201,7 +201,7 @@ ccl_device float3 subsurface_color_pow(float3 color, float exponent) ccl_device void subsurface_color_bump_blur(KernelGlobals *kg, ShaderData *sd, - PathState *state, + ccl_addr_space PathState *state, int state_flag, float3 *eval, float3 *N) @@ -239,7 +239,7 @@ ccl_device_inline int subsurface_scatter_multi_intersect( SubsurfaceIntersection *ss_isect, ShaderData *sd, ShaderClosure *sc, - uint *lcg_state, + RNG *lcg_state, float disk_u, float disk_v, bool all) @@ -293,7 +293,12 @@ ccl_device_inline int subsurface_scatter_multi_intersect( float3 disk_P = (disk_r*cosf(phi)) * disk_T + (disk_r*sinf(phi)) * disk_B; /* create ray */ +#ifdef __SPLIT_KERNEL__ + Ray ray_object = ss_isect->ray; + Ray *ray = &ray_object; +#else Ray *ray = &ss_isect->ray; +#endif ray->P = sd->P + disk_N*disk_height + disk_P; ray->D = -disk_N; ray->t = 2.0f*disk_height; @@ -304,7 +309,7 @@ ccl_device_inline int subsurface_scatter_multi_intersect( /* intersect with the same object. if multiple intersections are found it * will use at most BSSRDF_MAX_HITS hits, a random subset of all hits */ scene_intersect_subsurface(kg, - ray, + *ray, ss_isect, sd->object, lcg_state, @@ -314,20 +319,20 @@ ccl_device_inline int subsurface_scatter_multi_intersect( for(int hit = 0; hit < num_eval_hits; hit++) { /* Quickly retrieve P and Ng without setting up ShaderData. */ float3 hit_P; - if(ccl_fetch(sd, type) & PRIMITIVE_TRIANGLE) { + if(sd->type & PRIMITIVE_TRIANGLE) { hit_P = triangle_refine_subsurface(kg, sd, &ss_isect->hits[hit], ray); } #ifdef __OBJECT_MOTION__ - else if(ccl_fetch(sd, type) & PRIMITIVE_MOTION_TRIANGLE) { + else if(sd->type & PRIMITIVE_MOTION_TRIANGLE) { float3 verts[3]; motion_triangle_vertices( kg, - ccl_fetch(sd, object), + sd->object, kernel_tex_fetch(__prim_index, ss_isect->hits[hit].prim), - ccl_fetch(sd, time), + sd->time, verts); hit_P = motion_triangle_refine_subsurface(kg, sd, @@ -367,6 +372,10 @@ ccl_device_inline int subsurface_scatter_multi_intersect( ss_isect->weight[hit] = eval; } +#ifdef __SPLIT_KERNEL__ + ss_isect->ray = *ray; +#endif + return num_eval_hits; } @@ -375,13 +384,19 @@ ccl_device_noinline void subsurface_scatter_multi_setup( SubsurfaceIntersection* ss_isect, int hit, ShaderData *sd, - PathState *state, + ccl_addr_space PathState *state, int state_flag, ShaderClosure *sc, bool all) { +#ifdef __SPLIT_KERNEL__ + Ray ray_object = ss_isect->ray; + Ray *ray = &ray_object; +#else + Ray *ray = &ss_isect->ray; +#endif /* Setup new shading point. */ - shader_setup_from_subsurface(kg, sd, &ss_isect->hits[hit], &ss_isect->ray); + shader_setup_from_subsurface(kg, sd, &ss_isect->hits[hit], ray); /* Optionally blur colors and bump mapping. */ float3 weight = ss_isect->weight[hit]; @@ -392,6 +407,7 @@ ccl_device_noinline void subsurface_scatter_multi_setup( subsurface_scatter_setup_diffuse_bsdf(sd, sc, weight, true, N); } +#ifndef __SPLIT_KERNEL__ /* subsurface scattering step, from a point on the surface to another nearby point on the same object */ ccl_device void subsurface_scatter_step(KernelGlobals *kg, ShaderData *sd, PathState *state, int state_flag, ShaderClosure *sc, uint *lcg_state, float disk_u, float disk_v, bool all) @@ -448,7 +464,7 @@ ccl_device void subsurface_scatter_step(KernelGlobals *kg, ShaderData *sd, PathS /* intersect with the same object. if multiple intersections are * found it will randomly pick one of them */ SubsurfaceIntersection ss_isect; - scene_intersect_subsurface(kg, &ray, &ss_isect, sd->object, lcg_state, 1); + scene_intersect_subsurface(kg, ray, &ss_isect, sd->object, lcg_state, 1); /* evaluate bssrdf */ if(ss_isect.num_hits > 0) { @@ -481,6 +497,7 @@ ccl_device void subsurface_scatter_step(KernelGlobals *kg, ShaderData *sd, PathS /* setup diffuse bsdf */ subsurface_scatter_setup_diffuse_bsdf(sd, sc, eval, (ss_isect.num_hits > 0), N); } +#endif /* ! __SPLIT_KERNEL__ */ CCL_NAMESPACE_END diff --git a/intern/cycles/kernel/kernel_textures.h b/intern/cycles/kernel/kernel_textures.h index 8d5bb75a428..cb1a3f40dee 100644 --- a/intern/cycles/kernel/kernel_textures.h +++ b/intern/cycles/kernel/kernel_textures.h @@ -32,6 +32,7 @@ KERNEL_TEX(uint, texture_uint, __prim_visibility) KERNEL_TEX(uint, texture_uint, __prim_index) KERNEL_TEX(uint, texture_uint, __prim_object) KERNEL_TEX(uint, texture_uint, __object_node) +KERNEL_TEX(float2, texture_float2, __prim_time) /* objects */ KERNEL_TEX(float4, texture_float4, __objects) @@ -177,7 +178,6 @@ KERNEL_IMAGE_TEX(uchar4, texture_image_uchar4, __tex_image_byte4_085) KERNEL_IMAGE_TEX(uchar4, texture_image_uchar4, __tex_image_byte4_086) KERNEL_IMAGE_TEX(uchar4, texture_image_uchar4, __tex_image_byte4_087) KERNEL_IMAGE_TEX(uchar4, texture_image_uchar4, __tex_image_byte4_088) -KERNEL_IMAGE_TEX(uchar4, texture_image_uchar4, __tex_image_byte4_089) # else /* bindless textures */ diff --git a/intern/cycles/kernel/kernel_types.h b/intern/cycles/kernel/kernel_types.h index 8c271c75e44..19c91248922 100644 --- a/intern/cycles/kernel/kernel_types.h +++ b/intern/cycles/kernel/kernel_types.h @@ -17,9 +17,9 @@ #ifndef __KERNEL_TYPES_H__ #define __KERNEL_TYPES_H__ -#include "kernel_math.h" -#include "svm/svm_types.h" -#include "util_static_assert.h" +#include "kernel/kernel_math.h" +#include "kernel/svm/svm_types.h" +#include "util/util_static_assert.h" #ifndef __KERNEL_GPU__ # define __KERNEL_CPU__ @@ -56,6 +56,8 @@ CCL_NAMESPACE_BEGIN #define VOLUME_STACK_SIZE 16 +#define WORK_POOL_SIZE 64 + /* device capabilities */ #ifdef __KERNEL_CPU__ # ifdef __KERNEL_SSE2__ @@ -63,27 +65,34 @@ CCL_NAMESPACE_BEGIN # endif # define __KERNEL_SHADING__ # define __KERNEL_ADV_SHADING__ -# define __BRANCHED_PATH__ +# ifndef __SPLIT_KERNEL__ +# define __BRANCHED_PATH__ +# endif # ifdef WITH_OSL # define __OSL__ # endif # define __SUBSURFACE__ # define __CMJ__ # define __VOLUME__ -# define __VOLUME_DECOUPLED__ # define __VOLUME_SCATTER__ # define __SHADOW_RECORD_ALL__ -# define __VOLUME_RECORD_ALL__ +# ifndef __SPLIT_KERNEL__ +# define __VOLUME_DECOUPLED__ +# define __VOLUME_RECORD_ALL__ +# endif #endif /* __KERNEL_CPU__ */ #ifdef __KERNEL_CUDA__ # define __KERNEL_SHADING__ # define __KERNEL_ADV_SHADING__ -# define __BRANCHED_PATH__ # define __VOLUME__ # define __VOLUME_SCATTER__ # define __SUBSURFACE__ -# define __CMJ__ +# define __SHADOW_RECORD_ALL__ +# ifndef __SPLIT_KERNEL__ +# define __BRANCHED_PATH__ +# define __CMJ__ +# endif #endif /* __KERNEL_CUDA__ */ #ifdef __KERNEL_OPENCL__ @@ -93,6 +102,10 @@ CCL_NAMESPACE_BEGIN # ifdef __KERNEL_OPENCL_NVIDIA__ # define __KERNEL_SHADING__ # define __KERNEL_ADV_SHADING__ +# define __SUBSURFACE__ +# define __VOLUME__ +# define __VOLUME_SCATTER__ +# define __SHADOW_RECORD_ALL__ # ifdef __KERNEL_EXPERIMENTAL__ # define __CMJ__ # endif @@ -114,6 +127,10 @@ CCL_NAMESPACE_BEGIN # define __CL_USE_NATIVE__ # define __KERNEL_SHADING__ # define __KERNEL_ADV_SHADING__ +# define __SUBSURFACE__ +# define __VOLUME__ +# define __VOLUME_SCATTER__ +# define __SHADOW_RECORD_ALL__ # endif /* __KERNEL_OPENCL_AMD__ */ # ifdef __KERNEL_OPENCL_INTEL_CPU__ @@ -140,6 +157,7 @@ CCL_NAMESPACE_BEGIN #define __INTERSECTION_REFINE__ #define __CLAMP_SAMPLE__ #define __PATCH_EVAL__ +#define __SHADOW_TRICKS__ #ifdef __KERNEL_SHADING__ # define __SVM__ @@ -195,6 +213,9 @@ CCL_NAMESPACE_BEGIN #ifdef __NO_TRANSPARENT__ # undef __TRANSPARENT_SHADOWS__ #endif +#ifdef __NO_SHADOW_TRICKS__ +#undef __SHADOW_TRICKS__ +#endif /* Random Numbers */ @@ -299,6 +320,8 @@ enum PathRayFlag { PATH_RAY_MIS_SKIP = 4096, PATH_RAY_DIFFUSE_ANCESTOR = 8192, PATH_RAY_SINGLE_PASS_DONE = 16384, + PATH_RAY_SHADOW_CATCHER = 32768, + PATH_RAY_SHADOW_CATCHER_ONLY = 65536, }; /* Closure Label */ @@ -428,6 +451,20 @@ typedef ccl_addr_space struct PathRadiance { float4 shadow; float mist; #endif + +#ifdef __SHADOW_TRICKS__ + /* Total light reachable across the path, ignoring shadow blocked queries. */ + float3 path_total; + /* Total light reachable across the path with shadow blocked queries + * applied here. + * + * Dividing this figure by path_total will give estimate of shadow pass. + */ + float3 path_total_shaded; + + /* Color of the background on which shadow is alpha-overed. */ + float3 shadow_color; +#endif } PathRadiance; typedef struct BsdfEval { @@ -443,6 +480,9 @@ typedef struct BsdfEval { float3 subsurface; float3 scatter; #endif +#ifdef __SHADOW_TRICKS__ + float3 sum_no_mis; +#endif } BsdfEval; /* Shader Flag */ @@ -536,7 +576,7 @@ typedef struct Ray { /* Intersection */ -typedef ccl_addr_space struct Intersection { +typedef struct Intersection { float t, u, v; int prim; int object; @@ -788,108 +828,89 @@ enum ShaderDataObjectFlag { SD_OBJECT_INTERSECTS_VOLUME = (1 << 5), /* Has position for motion vertices. */ SD_OBJECT_HAS_VERTEX_MOTION = (1 << 6), + /* object is used to catch shadows */ + SD_OBJECT_SHADOW_CATCHER = (1 << 7), SD_OBJECT_FLAGS = (SD_OBJECT_HOLDOUT_MASK | SD_OBJECT_MOTION | SD_OBJECT_TRANSFORM_APPLIED | SD_OBJECT_NEGATIVE_SCALE_APPLIED | SD_OBJECT_HAS_VOLUME | - SD_OBJECT_INTERSECTS_VOLUME) + SD_OBJECT_INTERSECTS_VOLUME | + SD_OBJECT_SHADOW_CATCHER) }; -#ifdef __SPLIT_KERNEL__ -# define SD_THREAD (get_global_id(1) * get_global_size(0) + get_global_id(0)) -# if !defined(__SPLIT_KERNEL_SOA__) - /* ShaderData is stored as an Array-of-Structures */ -# define ccl_soa_member(type, name) type soa_##name -# define ccl_fetch(s, t) (s[SD_THREAD].soa_##t) -# define ccl_fetch_array(s, t, index) (&s[SD_THREAD].soa_##t[index]) -# else - /* ShaderData is stored as an Structure-of-Arrays */ -# define SD_GLOBAL_SIZE (get_global_size(0) * get_global_size(1)) -# define SD_FIELD_SIZE(t) sizeof(((struct ShaderData*)0)->t) -# define SD_OFFSETOF(t) ((char*)(&((struct ShaderData*)0)->t) - (char*)0) -# define ccl_soa_member(type, name) type soa_##name -# define ccl_fetch(s, t) (((ShaderData*)((ccl_addr_space char*)s + SD_GLOBAL_SIZE * SD_OFFSETOF(soa_##t) + SD_FIELD_SIZE(soa_##t) * SD_THREAD - SD_OFFSETOF(soa_##t)))->soa_##t) -# define ccl_fetch_array(s, t, index) (&ccl_fetch(s, t)[index]) -# endif -#else -# define ccl_soa_member(type, name) type name -# define ccl_fetch(s, t) (s->t) -# define ccl_fetch_array(s, t, index) (&s->t[index]) -#endif - typedef ccl_addr_space struct ShaderData { /* position */ - ccl_soa_member(float3, P); + float3 P; /* smooth normal for shading */ - ccl_soa_member(float3, N); + float3 N; /* true geometric normal */ - ccl_soa_member(float3, Ng); + float3 Ng; /* view/incoming direction */ - ccl_soa_member(float3, I); + float3 I; /* shader id */ - ccl_soa_member(int, shader); + int shader; /* booleans describing shader, see ShaderDataFlag */ - ccl_soa_member(int, flag); + int flag; /* booleans describing object of the shader, see ShaderDataObjectFlag */ - ccl_soa_member(int, object_flag); + int object_flag; /* primitive id if there is one, ~0 otherwise */ - ccl_soa_member(int, prim); + int prim; /* combined type and curve segment for hair */ - ccl_soa_member(int, type); + int type; /* parametric coordinates * - barycentric weights for triangles */ - ccl_soa_member(float, u); - ccl_soa_member(float, v); + float u; + float v; /* object id if there is one, ~0 otherwise */ - ccl_soa_member(int, object); + int object; /* motion blur sample time */ - ccl_soa_member(float, time); + float time; /* length of the ray being shaded */ - ccl_soa_member(float, ray_length); + float ray_length; #ifdef __RAY_DIFFERENTIALS__ /* differential of P. these are orthogonal to Ng, not N */ - ccl_soa_member(differential3, dP); + differential3 dP; /* differential of I */ - ccl_soa_member(differential3, dI); + differential3 dI; /* differential of u, v */ - ccl_soa_member(differential, du); - ccl_soa_member(differential, dv); + differential du; + differential dv; #endif #ifdef __DPDU__ /* differential of P w.r.t. parametric coordinates. note that dPdu is * not readily suitable as a tangent for shading on triangles. */ - ccl_soa_member(float3, dPdu); - ccl_soa_member(float3, dPdv); + float3 dPdu; + float3 dPdv; #endif #ifdef __OBJECT_MOTION__ /* object <-> world space transformations, cached to avoid * re-interpolating them constantly for shading */ - ccl_soa_member(Transform, ob_tfm); - ccl_soa_member(Transform, ob_itfm); + Transform ob_tfm; + Transform ob_itfm; #endif /* Closure data, we store a fixed array of closures */ - ccl_soa_member(struct ShaderClosure, closure[MAX_CLOSURE]); - ccl_soa_member(int, num_closure); - ccl_soa_member(int, num_closure_extra); - ccl_soa_member(float, randb_closure); - ccl_soa_member(float3, svm_closure_weight); + struct ShaderClosure closure[MAX_CLOSURE]; + int num_closure; + int num_closure_extra; + float randb_closure; + float3 svm_closure_weight; /* LCG state for closures that require additional random numbers. */ - ccl_soa_member(uint, lcg_state); + uint lcg_state; /* ray start position, only set for backgrounds */ - ccl_soa_member(float3, ray_P); - ccl_soa_member(differential3, ray_dP); + float3 ray_P; + differential3 ray_dP; #ifdef __OSL__ struct KernelGlobals *osl_globals; @@ -935,12 +956,16 @@ typedef struct PathState { RNG rng_congruential; VolumeStack volume_stack[VOLUME_STACK_SIZE]; #endif + +#ifdef __SHADOW_TRICKS__ + int catcher_object; +#endif } PathState; /* Subsurface */ /* Struct to gather multiple SSS hits. */ -struct SubsurfaceIntersection +typedef struct SubsurfaceIntersection { Ray ray; float3 weight[BSSRDF_MAX_HITS]; @@ -948,10 +973,10 @@ struct SubsurfaceIntersection int num_hits; struct Intersection hits[BSSRDF_MAX_HITS]; float3 Ng[BSSRDF_MAX_HITS]; -}; +} SubsurfaceIntersection; /* Struct to gather SSS indirect rays and delay tracing them. */ -struct SubsurfaceIndirectRays +typedef struct SubsurfaceIndirectRays { bool need_update_volume_stack; bool tracing; @@ -962,7 +987,7 @@ struct SubsurfaceIndirectRays struct Ray rays[BSSRDF_MAX_HITS]; float3 throughputs[BSSRDF_MAX_HITS]; struct PathRadiance L[BSSRDF_MAX_HITS]; -}; +} SubsurfaceIndirectRays; /* Constant Kernel Data * @@ -1201,7 +1226,8 @@ typedef struct KernelBVH { int have_curves; int have_instancing; int use_qbvh; - int pad1, pad2; + int use_bvh_steps; + int pad1; } KernelBVH; static_assert_align(KernelBVH, 16); @@ -1296,20 +1322,19 @@ enum QueueNumber { #define RAY_STATE_MASK 0x007 #define RAY_FLAG_MASK 0x0F8 enum RayState { + RAY_INVALID = 0, /* Denotes ray is actively involved in path-iteration. */ - RAY_ACTIVE = 0, + RAY_ACTIVE, /* Denotes ray has completed processing all samples and is inactive. */ - RAY_INACTIVE = 1, + RAY_INACTIVE, /* Denoted ray has exited path-iteration and needs to update output buffer. */ - RAY_UPDATE_BUFFER = 2, + RAY_UPDATE_BUFFER, /* Donotes ray has hit background */ - RAY_HIT_BACKGROUND = 3, + RAY_HIT_BACKGROUND, /* Denotes ray has to be regenerated */ - RAY_TO_REGENERATE = 4, + RAY_TO_REGENERATE, /* Denotes ray has been regenerated */ - RAY_REGENERATED = 5, - /* Denotes ray should skip direct lighting */ - RAY_SKIP_DL = 6, + RAY_REGENERATED, /* Flag's ray has to execute shadow blocked function in AO part */ RAY_SHADOW_RAY_CAST_AO = 16, /* Flag's ray has to execute shadow blocked function in direct lighting part. */ diff --git a/intern/cycles/kernel/kernel_volume.h b/intern/cycles/kernel/kernel_volume.h index c7cb29b5af2..9c0878249d4 100644 --- a/intern/cycles/kernel/kernel_volume.h +++ b/intern/cycles/kernel/kernel_volume.h @@ -38,7 +38,7 @@ typedef struct VolumeShaderCoefficients { /* evaluate shader to get extinction coefficient at P */ ccl_device_inline bool volume_shader_extinction_sample(KernelGlobals *kg, ShaderData *sd, - PathState *state, + ccl_addr_space PathState *state, float3 P, float3 *extinction) { @@ -64,7 +64,7 @@ ccl_device_inline bool volume_shader_extinction_sample(KernelGlobals *kg, /* evaluate shader to get absorption, scattering and emission at P */ ccl_device_inline bool volume_shader_sample(KernelGlobals *kg, ShaderData *sd, - PathState *state, + ccl_addr_space PathState *state, float3 P, VolumeShaderCoefficients *coeff) { @@ -112,7 +112,7 @@ ccl_device float kernel_volume_channel_get(float3 value, int channel) return (channel == 0)? value.x: ((channel == 1)? value.y: value.z); } -ccl_device bool volume_stack_is_heterogeneous(KernelGlobals *kg, VolumeStack *stack) +ccl_device bool volume_stack_is_heterogeneous(KernelGlobals *kg, ccl_addr_space VolumeStack *stack) { for(int i = 0; stack[i].shader != SHADER_NONE; i++) { int shader_flag = kernel_tex_fetch(__shader_flag, (stack[i].shader & SHADER_MASK)*SHADER_SIZE); @@ -161,7 +161,11 @@ ccl_device int volume_stack_sampling_method(KernelGlobals *kg, VolumeStack *stac /* homogeneous volume: assume shader evaluation at the starts gives * the extinction coefficient for the entire line segment */ -ccl_device void kernel_volume_shadow_homogeneous(KernelGlobals *kg, PathState *state, Ray *ray, ShaderData *sd, float3 *throughput) +ccl_device void kernel_volume_shadow_homogeneous(KernelGlobals *kg, + ccl_addr_space PathState *state, + Ray *ray, + ShaderData *sd, + float3 *throughput) { float3 sigma_t; @@ -171,7 +175,11 @@ ccl_device void kernel_volume_shadow_homogeneous(KernelGlobals *kg, PathState *s /* heterogeneous volume: integrate stepping through the volume until we * reach the end, get absorbed entirely, or run out of iterations */ -ccl_device void kernel_volume_shadow_heterogeneous(KernelGlobals *kg, PathState *state, Ray *ray, ShaderData *sd, float3 *throughput) +ccl_device void kernel_volume_shadow_heterogeneous(KernelGlobals *kg, + ccl_addr_space PathState *state, + Ray *ray, + ShaderData *sd, + float3 *throughput) { float3 tp = *throughput; const float tp_eps = 1e-6f; /* todo: this is likely not the right value */ @@ -179,7 +187,7 @@ ccl_device void kernel_volume_shadow_heterogeneous(KernelGlobals *kg, PathState /* prepare for stepping */ int max_steps = kernel_data.integrator.volume_max_steps; float step = kernel_data.integrator.volume_step_size; - float random_jitter_offset = lcg_step_float(&state->rng_congruential) * step; + float random_jitter_offset = lcg_step_float_addrspace(&state->rng_congruential) * step; /* compute extinction at the start */ float t = 0.0f; @@ -193,7 +201,7 @@ ccl_device void kernel_volume_shadow_heterogeneous(KernelGlobals *kg, PathState /* use random position inside this segment to sample shader */ if(new_t == ray->t) - random_jitter_offset = lcg_step_float(&state->rng_congruential) * dt; + random_jitter_offset = lcg_step_float_addrspace(&state->rng_congruential) * dt; float3 new_P = ray->P + ray->D * (t + random_jitter_offset); float3 sigma_t; @@ -227,7 +235,11 @@ ccl_device void kernel_volume_shadow_heterogeneous(KernelGlobals *kg, PathState /* get the volume attenuation over line segment defined by ray, with the * assumption that there are no surfaces blocking light between the endpoints */ -ccl_device_noinline void kernel_volume_shadow(KernelGlobals *kg, ShaderData *shadow_sd, PathState *state, Ray *ray, float3 *throughput) +ccl_device_noinline void kernel_volume_shadow(KernelGlobals *kg, + ShaderData *shadow_sd, + ccl_addr_space PathState *state, + Ray *ray, + float3 *throughput) { shader_setup_from_volume(kg, shadow_sd, ray); @@ -341,9 +353,15 @@ ccl_device float3 kernel_volume_emission_integrate(VolumeShaderCoefficients *coe /* homogeneous volume: assume shader evaluation at the start gives * the volume shading coefficient for the entire line segment */ -ccl_device VolumeIntegrateResult kernel_volume_integrate_homogeneous(KernelGlobals *kg, - PathState *state, Ray *ray, ShaderData *sd, PathRadiance *L, float3 *throughput, - RNG *rng, bool probalistic_scatter) +ccl_device VolumeIntegrateResult kernel_volume_integrate_homogeneous( + KernelGlobals *kg, + ccl_addr_space PathState *state, + Ray *ray, + ShaderData *sd, + PathRadiance *L, + ccl_addr_space float3 *throughput, + RNG *rng, + bool probalistic_scatter) { VolumeShaderCoefficients coeff; @@ -444,8 +462,14 @@ ccl_device VolumeIntegrateResult kernel_volume_integrate_homogeneous(KernelGloba * volume until we reach the end, get absorbed entirely, or run out of * iterations. this does probabilistically scatter or get transmitted through * for path tracing where we don't want to branch. */ -ccl_device VolumeIntegrateResult kernel_volume_integrate_heterogeneous_distance(KernelGlobals *kg, - PathState *state, Ray *ray, ShaderData *sd, PathRadiance *L, float3 *throughput, RNG *rng) +ccl_device VolumeIntegrateResult kernel_volume_integrate_heterogeneous_distance( + KernelGlobals *kg, + ccl_addr_space PathState *state, + Ray *ray, + ShaderData *sd, + PathRadiance *L, + ccl_addr_space float3 *throughput, + RNG *rng) { float3 tp = *throughput; const float tp_eps = 1e-6f; /* todo: this is likely not the right value */ @@ -453,7 +477,7 @@ ccl_device VolumeIntegrateResult kernel_volume_integrate_heterogeneous_distance( /* prepare for stepping */ int max_steps = kernel_data.integrator.volume_max_steps; float step_size = kernel_data.integrator.volume_step_size; - float random_jitter_offset = lcg_step_float(&state->rng_congruential) * step_size; + float random_jitter_offset = lcg_step_float_addrspace(&state->rng_congruential) * step_size; /* compute coefficients at the start */ float t = 0.0f; @@ -474,7 +498,7 @@ ccl_device VolumeIntegrateResult kernel_volume_integrate_heterogeneous_distance( /* use random position inside this segment to sample shader */ if(new_t == ray->t) - random_jitter_offset = lcg_step_float(&state->rng_congruential) * dt; + random_jitter_offset = lcg_step_float_addrspace(&state->rng_congruential) * dt; float3 new_P = ray->P + ray->D * (t + random_jitter_offset); VolumeShaderCoefficients coeff; @@ -579,8 +603,15 @@ ccl_device VolumeIntegrateResult kernel_volume_integrate_heterogeneous_distance( * ray, with the assumption that there are no surfaces blocking light * between the endpoints. distance sampling is used to decide if we will * scatter or not. */ -ccl_device_noinline VolumeIntegrateResult kernel_volume_integrate(KernelGlobals *kg, - PathState *state, ShaderData *sd, Ray *ray, PathRadiance *L, float3 *throughput, RNG *rng, bool heterogeneous) +ccl_device_noinline VolumeIntegrateResult kernel_volume_integrate( + KernelGlobals *kg, + ccl_addr_space PathState *state, + ShaderData *sd, + Ray *ray, + PathRadiance *L, + ccl_addr_space float3 *throughput, + RNG *rng, + bool heterogeneous) { shader_setup_from_volume(kg, sd, ray); @@ -590,6 +621,7 @@ ccl_device_noinline VolumeIntegrateResult kernel_volume_integrate(KernelGlobals return kernel_volume_integrate_homogeneous(kg, state, ray, sd, L, throughput, rng, true); } +#ifndef __SPLIT_KERNEL__ /* Decoupled Volume Sampling * * VolumeSegment is list of coefficients and transmittance stored at all steps @@ -966,7 +998,7 @@ ccl_device VolumeIntegrateResult kernel_volume_decoupled_scatter( mis_weight = 2.0f*power_heuristic(pdf, distance_pdf); } } - if(sample_t < 1e-6f) { + if(sample_t < 1e-6f || pdf == 0.0f) { return VOLUME_PATH_SCATTERED; } @@ -990,6 +1022,7 @@ ccl_device VolumeIntegrateResult kernel_volume_decoupled_scatter( return VOLUME_PATH_SCATTERED; } +#endif /* __SPLIT_KERNEL */ /* decide if we need to use decoupled or not */ ccl_device bool kernel_volume_use_decoupled(KernelGlobals *kg, bool heterogeneous, bool direct, int sampling_method) @@ -1021,9 +1054,9 @@ ccl_device bool kernel_volume_use_decoupled(KernelGlobals *kg, bool heterogeneou ccl_device void kernel_volume_stack_init(KernelGlobals *kg, ShaderData *stack_sd, - const PathState *state, - const Ray *ray, - VolumeStack *stack) + ccl_addr_space const PathState *state, + ccl_addr_space const Ray *ray, + ccl_addr_space VolumeStack *stack) { /* NULL ray happens in the baker, does it need proper initialization of * camera in volume? @@ -1166,7 +1199,7 @@ ccl_device void kernel_volume_stack_init(KernelGlobals *kg, } } -ccl_device void kernel_volume_stack_enter_exit(KernelGlobals *kg, ShaderData *sd, VolumeStack *stack) +ccl_device void kernel_volume_stack_enter_exit(KernelGlobals *kg, ShaderData *sd, ccl_addr_space VolumeStack *stack) { /* todo: we should have some way for objects to indicate if they want the * world shader to work inside them. excluding it by default is problematic @@ -1215,7 +1248,7 @@ ccl_device void kernel_volume_stack_enter_exit(KernelGlobals *kg, ShaderData *sd ccl_device void kernel_volume_stack_update_for_subsurface(KernelGlobals *kg, ShaderData *stack_sd, Ray *ray, - VolumeStack *stack) + ccl_addr_space VolumeStack *stack) { kernel_assert(kernel_data.integrator.use_volumes); @@ -1277,7 +1310,7 @@ ccl_device void kernel_volume_stack_update_for_subsurface(KernelGlobals *kg, * the world's one after the last bounce to avoid render artifacts. */ ccl_device_inline void kernel_volume_clean_stack(KernelGlobals *kg, - VolumeStack *volume_stack) + ccl_addr_space VolumeStack *volume_stack) { if(kernel_data.background.volume_shader != SHADER_NONE) { /* Keep the world's volume in stack. */ diff --git a/intern/cycles/kernel/kernel_work_stealing.h b/intern/cycles/kernel/kernel_work_stealing.h index 7d559b1aa31..28fc5ce1c30 100644 --- a/intern/cycles/kernel/kernel_work_stealing.h +++ b/intern/cycles/kernel/kernel_work_stealing.h @@ -17,177 +17,102 @@ #ifndef __KERNEL_WORK_STEALING_H__ #define __KERNEL_WORK_STEALING_H__ +CCL_NAMESPACE_BEGIN + /* * Utility functions for work stealing */ -#ifdef __WORK_STEALING__ - #ifdef __KERNEL_OPENCL__ # pragma OPENCL EXTENSION cl_khr_global_int32_base_atomics : enable #endif -uint get_group_id_with_ray_index(uint ray_index, - uint tile_dim_x, - uint tile_dim_y, - uint parallel_samples, - int dim) +ccl_device_inline uint kernel_total_work_size(KernelGlobals *kg) +{ + return kernel_split_params.w * kernel_split_params.h * kernel_split_params.num_samples; +} + +ccl_device_inline uint kernel_num_work_pools(KernelGlobals *kg) +{ + return ccl_global_size(0) * ccl_global_size(1) / WORK_POOL_SIZE; +} + +ccl_device_inline uint work_pool_from_ray_index(KernelGlobals *kg, uint ray_index) +{ + return ray_index / WORK_POOL_SIZE; +} + +ccl_device_inline uint work_pool_work_size(KernelGlobals *kg, uint work_pool) { - if(dim == 0) { - uint x_span = ray_index % (tile_dim_x * parallel_samples); - return x_span / get_local_size(0); + uint total_work_size = kernel_total_work_size(kg); + uint num_pools = kernel_num_work_pools(kg); + + if(work_pool >= num_pools || work_pool * WORK_POOL_SIZE >= total_work_size) { + return 0; + } + + uint work_size = (total_work_size / (num_pools * WORK_POOL_SIZE)) * WORK_POOL_SIZE; + + uint remainder = (total_work_size % (num_pools * WORK_POOL_SIZE)); + if(work_pool < remainder / WORK_POOL_SIZE) { + work_size += WORK_POOL_SIZE; } - else /*if(dim == 1)*/ { - kernel_assert(dim == 1); - uint y_span = ray_index / (tile_dim_x * parallel_samples); - return y_span / get_local_size(1); + else if(work_pool == remainder / WORK_POOL_SIZE) { + work_size += remainder % WORK_POOL_SIZE; } + + return work_size; } -uint get_total_work(uint tile_dim_x, - uint tile_dim_y, - uint grp_idx, - uint grp_idy, - uint num_samples) +ccl_device_inline uint get_global_work_index(KernelGlobals *kg, uint work_index, uint ray_index) { - uint threads_within_tile_border_x = - (grp_idx == (get_num_groups(0) - 1)) ? tile_dim_x % get_local_size(0) - : get_local_size(0); - uint threads_within_tile_border_y = - (grp_idy == (get_num_groups(1) - 1)) ? tile_dim_y % get_local_size(1) - : get_local_size(1); - - threads_within_tile_border_x = - (threads_within_tile_border_x == 0) ? get_local_size(0) - : threads_within_tile_border_x; - threads_within_tile_border_y = - (threads_within_tile_border_y == 0) ? get_local_size(1) - : threads_within_tile_border_y; - - return threads_within_tile_border_x * - threads_within_tile_border_y * - num_samples; + uint num_pools = kernel_num_work_pools(kg); + uint pool = work_pool_from_ray_index(kg, ray_index); + + return (work_index / WORK_POOL_SIZE) * (num_pools * WORK_POOL_SIZE) + + (pool * WORK_POOL_SIZE) + + (work_index % WORK_POOL_SIZE); } -/* Returns 0 in case there is no next work available */ -/* Returns 1 in case work assigned is valid */ -int get_next_work(ccl_global uint *work_pool, - ccl_private uint *my_work, - uint tile_dim_x, - uint tile_dim_y, - uint num_samples, - uint parallel_samples, - uint ray_index) +/* Returns true if there is work */ +ccl_device bool get_next_work(KernelGlobals *kg, ccl_private uint *work_index, uint ray_index) { - uint grp_idx = get_group_id_with_ray_index(ray_index, - tile_dim_x, - tile_dim_y, - parallel_samples, - 0); - uint grp_idy = get_group_id_with_ray_index(ray_index, - tile_dim_x, - tile_dim_y, - parallel_samples, - 1); - uint total_work = get_total_work(tile_dim_x, - tile_dim_y, - grp_idx, - grp_idy, - num_samples); - uint group_index = grp_idy * get_num_groups(0) + grp_idx; - *my_work = atomic_inc(&work_pool[group_index]); - return (*my_work < total_work) ? 1 : 0; + uint work_pool = work_pool_from_ray_index(kg, ray_index); + uint pool_size = work_pool_work_size(kg, work_pool); + + if(pool_size == 0) { + return false; + } + + *work_index = atomic_fetch_and_inc_uint32(&kernel_split_params.work_pools[work_pool]); + return (*work_index < pool_size); } -/* This function assumes that the passed my_work is valid. */ -/* Decode sample number w.r.t. assigned my_work. */ -uint get_my_sample(uint my_work, - uint tile_dim_x, - uint tile_dim_y, - uint parallel_samples, - uint ray_index) +/* This function assumes that the passed `work` is valid. */ +/* Decode sample number w.r.t. assigned `work`. */ +ccl_device uint get_work_sample(KernelGlobals *kg, uint work_index, uint ray_index) { - uint grp_idx = get_group_id_with_ray_index(ray_index, - tile_dim_x, - tile_dim_y, - parallel_samples, - 0); - uint grp_idy = get_group_id_with_ray_index(ray_index, - tile_dim_x, - tile_dim_y, - parallel_samples, - 1); - uint threads_within_tile_border_x = - (grp_idx == (get_num_groups(0) - 1)) ? tile_dim_x % get_local_size(0) - : get_local_size(0); - uint threads_within_tile_border_y = - (grp_idy == (get_num_groups(1) - 1)) ? tile_dim_y % get_local_size(1) - : get_local_size(1); - - threads_within_tile_border_x = - (threads_within_tile_border_x == 0) ? get_local_size(0) - : threads_within_tile_border_x; - threads_within_tile_border_y = - (threads_within_tile_border_y == 0) ? get_local_size(1) - : threads_within_tile_border_y; - - return my_work / - (threads_within_tile_border_x * threads_within_tile_border_y); + return get_global_work_index(kg, work_index, ray_index) / (kernel_split_params.w * kernel_split_params.h); } -/* Decode pixel and tile position w.r.t. assigned my_work. */ -void get_pixel_tile_position(ccl_private uint *pixel_x, +/* Decode pixel and tile position w.r.t. assigned `work`. */ +ccl_device void get_work_pixel_tile_position(KernelGlobals *kg, + ccl_private uint *pixel_x, ccl_private uint *pixel_y, ccl_private uint *tile_x, ccl_private uint *tile_y, - uint my_work, - uint tile_dim_x, - uint tile_dim_y, - uint tile_offset_x, - uint tile_offset_y, - uint parallel_samples, + uint work_index, uint ray_index) { - uint grp_idx = get_group_id_with_ray_index(ray_index, - tile_dim_x, - tile_dim_y, - parallel_samples, - 0); - uint grp_idy = get_group_id_with_ray_index(ray_index, - tile_dim_x, - tile_dim_y, - parallel_samples, - 1); - uint threads_within_tile_border_x = - (grp_idx == (get_num_groups(0) - 1)) ? tile_dim_x % get_local_size(0) - : get_local_size(0); - uint threads_within_tile_border_y = - (grp_idy == (get_num_groups(1) - 1)) ? tile_dim_y % get_local_size(1) - : get_local_size(1); - - threads_within_tile_border_x = - (threads_within_tile_border_x == 0) ? get_local_size(0) - : threads_within_tile_border_x; - threads_within_tile_border_y = - (threads_within_tile_border_y == 0) ? get_local_size(1) - : threads_within_tile_border_y; - - uint total_associated_pixels = - threads_within_tile_border_x * threads_within_tile_border_y; - uint work_group_pixel_index = my_work % total_associated_pixels; - uint work_group_pixel_x = - work_group_pixel_index % threads_within_tile_border_x; - uint work_group_pixel_y = - work_group_pixel_index / threads_within_tile_border_x; - - *pixel_x = - tile_offset_x + (grp_idx * get_local_size(0)) + work_group_pixel_x; - *pixel_y = - tile_offset_y + (grp_idy * get_local_size(1)) + work_group_pixel_y; - *tile_x = *pixel_x - tile_offset_x; - *tile_y = *pixel_y - tile_offset_y; + uint pixel_index = get_global_work_index(kg, work_index, ray_index) % (kernel_split_params.w*kernel_split_params.h); + + *tile_x = pixel_index % kernel_split_params.w; + *tile_y = pixel_index / kernel_split_params.w; + + *pixel_x = *tile_x + kernel_split_params.x; + *pixel_y = *tile_y + kernel_split_params.y; } -#endif /* __WORK_STEALING__ */ +CCL_NAMESPACE_END #endif /* __KERNEL_WORK_STEALING_H__ */ diff --git a/intern/cycles/kernel/kernels/cpu/kernel.cpp b/intern/cycles/kernel/kernels/cpu/kernel.cpp index 72dbbd9a416..16992c681e6 100644 --- a/intern/cycles/kernel/kernels/cpu/kernel.cpp +++ b/intern/cycles/kernel/kernels/cpu/kernel.cpp @@ -56,9 +56,9 @@ /* do nothing */ #endif -#include "kernel.h" +#include "kernel/kernel.h" #define KERNEL_ARCH cpu -#include "kernel_cpu_impl.h" +#include "kernel/kernels/cpu/kernel_cpu_impl.h" CCL_NAMESPACE_BEGIN @@ -90,7 +90,7 @@ void kernel_tex_copy(KernelGlobals *kg, kg->tname.width = width; \ } #define KERNEL_IMAGE_TEX(type, ttype, tname) -#include "kernel_textures.h" +#include "kernel/kernel_textures.h" else if(strstr(name, "__tex_image_float4")) { texture_image_float4 *tex = NULL; diff --git a/intern/cycles/kernel/kernels/cpu/kernel_avx.cpp b/intern/cycles/kernel/kernels/cpu/kernel_avx.cpp index 1350d9e5c2e..2600d977972 100644 --- a/intern/cycles/kernel/kernels/cpu/kernel_avx.cpp +++ b/intern/cycles/kernel/kernels/cpu/kernel_avx.cpp @@ -28,10 +28,10 @@ # define __KERNEL_AVX__ #endif -#include "util_optimization.h" +#include "util/util_optimization.h" #ifdef WITH_CYCLES_OPTIMIZED_KERNEL_AVX -# include "kernel.h" +# include "kernel/kernel.h" # define KERNEL_ARCH cpu_avx -# include "kernel_cpu_impl.h" +# include "kernel/kernels/cpu/kernel_cpu_impl.h" #endif /* WITH_CYCLES_OPTIMIZED_KERNEL_AVX */ diff --git a/intern/cycles/kernel/kernels/cpu/kernel_avx2.cpp b/intern/cycles/kernel/kernels/cpu/kernel_avx2.cpp index 1a416e771ee..dba15d037ac 100644 --- a/intern/cycles/kernel/kernels/cpu/kernel_avx2.cpp +++ b/intern/cycles/kernel/kernels/cpu/kernel_avx2.cpp @@ -29,10 +29,10 @@ # define __KERNEL_AVX2__ #endif -#include "util_optimization.h" +#include "util/util_optimization.h" #ifdef WITH_CYCLES_OPTIMIZED_KERNEL_AVX2 -# include "kernel.h" +# include "kernel/kernel.h" # define KERNEL_ARCH cpu_avx2 -# include "kernel_cpu_impl.h" +# include "kernel/kernels/cpu/kernel_cpu_impl.h" #endif /* WITH_CYCLES_OPTIMIZED_KERNEL_AVX2 */ diff --git a/intern/cycles/kernel/kernels/cpu/kernel_cpu.h b/intern/cycles/kernel/kernels/cpu/kernel_cpu.h index 1a07c705f1c..896b80d783e 100644 --- a/intern/cycles/kernel/kernels/cpu/kernel_cpu.h +++ b/intern/cycles/kernel/kernels/cpu/kernel_cpu.h @@ -49,4 +49,44 @@ void KERNEL_FUNCTION_FULL_NAME(shader)(KernelGlobals *kg, int offset, int sample); +/* Split kernels */ + +void KERNEL_FUNCTION_FULL_NAME(data_init)( + KernelGlobals *kg, + ccl_constant KernelData *data, + ccl_global void *split_data_buffer, + int num_elements, + ccl_global char *ray_state, + ccl_global uint *rng_state, + int start_sample, + int end_sample, + int sx, int sy, int sw, int sh, int offset, int stride, + ccl_global int *Queue_index, + int queuesize, + ccl_global char *use_queues_flag, + ccl_global unsigned int *work_pool_wgs, + unsigned int num_samples, + ccl_global float *buffer); + +#define DECLARE_SPLIT_KERNEL_FUNCTION(name) \ + void KERNEL_FUNCTION_FULL_NAME(name)(KernelGlobals *kg, KernelData *data); + +DECLARE_SPLIT_KERNEL_FUNCTION(path_init) +DECLARE_SPLIT_KERNEL_FUNCTION(scene_intersect) +DECLARE_SPLIT_KERNEL_FUNCTION(lamp_emission) +DECLARE_SPLIT_KERNEL_FUNCTION(do_volume) +DECLARE_SPLIT_KERNEL_FUNCTION(queue_enqueue) +DECLARE_SPLIT_KERNEL_FUNCTION(indirect_background) +DECLARE_SPLIT_KERNEL_FUNCTION(shader_eval) +DECLARE_SPLIT_KERNEL_FUNCTION(holdout_emission_blurring_pathtermination_ao) +DECLARE_SPLIT_KERNEL_FUNCTION(subsurface_scatter) +DECLARE_SPLIT_KERNEL_FUNCTION(direct_lighting) +DECLARE_SPLIT_KERNEL_FUNCTION(shadow_blocked_ao) +DECLARE_SPLIT_KERNEL_FUNCTION(shadow_blocked_dl) +DECLARE_SPLIT_KERNEL_FUNCTION(next_iteration_setup) +DECLARE_SPLIT_KERNEL_FUNCTION(indirect_subsurface) +DECLARE_SPLIT_KERNEL_FUNCTION(buffer_update) + +void KERNEL_FUNCTION_FULL_NAME(register_functions)(void(*reg)(const char* name, void* func)); + #undef KERNEL_ARCH diff --git a/intern/cycles/kernel/kernels/cpu/kernel_cpu_impl.h b/intern/cycles/kernel/kernels/cpu/kernel_cpu_impl.h index ec82d4b4c22..148b2eef568 100644 --- a/intern/cycles/kernel/kernels/cpu/kernel_cpu_impl.h +++ b/intern/cycles/kernel/kernels/cpu/kernel_cpu_impl.h @@ -20,18 +20,45 @@ * simply includes this file without worry of copying actual implementation over. */ -#include "kernel_compat_cpu.h" -#include "kernel_math.h" -#include "kernel_types.h" -#include "kernel_globals.h" -#include "kernel_cpu_image.h" -#include "kernel_film.h" -#include "kernel_path.h" -#include "kernel_path_branched.h" -#include "kernel_bake.h" +#include "kernel/kernel_compat_cpu.h" + +#ifndef __SPLIT_KERNEL__ +# include "kernel/kernel_math.h" +# include "kernel/kernel_types.h" + +# include "kernel/split/kernel_split_data.h" +# include "kernel/kernel_globals.h" + +# include "kernel/kernels/cpu/kernel_cpu_image.h" +# include "kernel/kernel_film.h" +# include "kernel/kernel_path.h" +# include "kernel/kernel_path_branched.h" +# include "kernel/kernel_bake.h" +#else +# include "kernel/split/kernel_split_common.h" + +# include "kernel/split/kernel_data_init.h" +# include "kernel/split/kernel_path_init.h" +# include "kernel/split/kernel_scene_intersect.h" +# include "kernel/split/kernel_lamp_emission.h" +# include "kernel/split/kernel_do_volume.h" +# include "kernel/split/kernel_queue_enqueue.h" +# include "kernel/split/kernel_indirect_background.h" +# include "kernel/split/kernel_shader_eval.h" +# include "kernel/split/kernel_holdout_emission_blurring_pathtermination_ao.h" +# include "kernel/split/kernel_subsurface_scatter.h" +# include "kernel/split/kernel_direct_lighting.h" +# include "kernel/split/kernel_shadow_blocked_ao.h" +# include "kernel/split/kernel_shadow_blocked_dl.h" +# include "kernel/split/kernel_next_iteration_setup.h" +# include "kernel/split/kernel_indirect_subsurface.h" +# include "kernel/split/kernel_buffer_update.h" +#endif CCL_NAMESPACE_BEGIN +#ifndef __SPLIT_KERNEL__ + /* Path Tracing */ void KERNEL_FUNCTION_FULL_NAME(path_trace)(KernelGlobals *kg, @@ -131,4 +158,72 @@ void KERNEL_FUNCTION_FULL_NAME(shader)(KernelGlobals *kg, } } +#else /* __SPLIT_KERNEL__ */ + +/* Split Kernel Path Tracing */ + +#define DEFINE_SPLIT_KERNEL_FUNCTION(name) \ + void KERNEL_FUNCTION_FULL_NAME(name)(KernelGlobals *kg, KernelData* /*data*/) \ + { \ + kernel_##name(kg); \ + } + +#define DEFINE_SPLIT_KERNEL_FUNCTION_LOCALS(name, type) \ + void KERNEL_FUNCTION_FULL_NAME(name)(KernelGlobals *kg, KernelData* /*data*/) \ + { \ + ccl_local type locals; \ + kernel_##name(kg, &locals); \ + } + +DEFINE_SPLIT_KERNEL_FUNCTION(path_init) +DEFINE_SPLIT_KERNEL_FUNCTION(scene_intersect) +DEFINE_SPLIT_KERNEL_FUNCTION(lamp_emission) +DEFINE_SPLIT_KERNEL_FUNCTION(do_volume) +DEFINE_SPLIT_KERNEL_FUNCTION_LOCALS(queue_enqueue, QueueEnqueueLocals) +DEFINE_SPLIT_KERNEL_FUNCTION(indirect_background) +DEFINE_SPLIT_KERNEL_FUNCTION_LOCALS(shader_eval, uint) +DEFINE_SPLIT_KERNEL_FUNCTION_LOCALS(holdout_emission_blurring_pathtermination_ao, BackgroundAOLocals) +DEFINE_SPLIT_KERNEL_FUNCTION_LOCALS(subsurface_scatter, uint) +DEFINE_SPLIT_KERNEL_FUNCTION_LOCALS(direct_lighting, uint) +DEFINE_SPLIT_KERNEL_FUNCTION(shadow_blocked_ao) +DEFINE_SPLIT_KERNEL_FUNCTION(shadow_blocked_dl) +DEFINE_SPLIT_KERNEL_FUNCTION_LOCALS(next_iteration_setup, uint) +DEFINE_SPLIT_KERNEL_FUNCTION(indirect_subsurface) +DEFINE_SPLIT_KERNEL_FUNCTION_LOCALS(buffer_update, uint) + +void KERNEL_FUNCTION_FULL_NAME(register_functions)(void(*reg)(const char* name, void* func)) +{ +#define REGISTER_NAME_STRING(name) #name +#define REGISTER_EVAL_NAME(name) REGISTER_NAME_STRING(name) +#define REGISTER(name) reg(REGISTER_EVAL_NAME(KERNEL_FUNCTION_FULL_NAME(name)), (void*)KERNEL_FUNCTION_FULL_NAME(name)); + + REGISTER(path_trace); + REGISTER(convert_to_byte); + REGISTER(convert_to_half_float); + REGISTER(shader); + + REGISTER(data_init); + REGISTER(path_init); + REGISTER(scene_intersect); + REGISTER(lamp_emission); + REGISTER(do_volume); + REGISTER(queue_enqueue); + REGISTER(indirect_background); + REGISTER(shader_eval); + REGISTER(holdout_emission_blurring_pathtermination_ao); + REGISTER(subsurface_scatter); + REGISTER(direct_lighting); + REGISTER(shadow_blocked_ao); + REGISTER(shadow_blocked_dl); + REGISTER(next_iteration_setup); + REGISTER(indirect_subsurface); + REGISTER(buffer_update); + +#undef REGISTER +#undef REGISTER_EVAL_NAME +#undef REGISTER_NAME_STRING +} + +#endif /* __SPLIT_KERNEL__ */ + CCL_NAMESPACE_END diff --git a/intern/cycles/kernel/kernels/cpu/kernel_split.cpp b/intern/cycles/kernel/kernels/cpu/kernel_split.cpp new file mode 100644 index 00000000000..ca750e5a00d --- /dev/null +++ b/intern/cycles/kernel/kernels/cpu/kernel_split.cpp @@ -0,0 +1,63 @@ +/* + * Copyright 2011-2013 Blender Foundation + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +/* CPU kernel entry points */ + +/* On x86-64, we can assume SSE2, so avoid the extra kernel and compile this + * one with SSE2 intrinsics. + */ +#if defined(__x86_64__) || defined(_M_X64) +# define __KERNEL_SSE2__ +#endif + +#define __SPLIT_KERNEL__ + +/* When building kernel for native machine detect kernel features from the flags + * set by compiler. + */ +#ifdef WITH_KERNEL_NATIVE +# ifdef __SSE2__ +# ifndef __KERNEL_SSE2__ +# define __KERNEL_SSE2__ +# endif +# endif +# ifdef __SSE3__ +# define __KERNEL_SSE3__ +# endif +# ifdef __SSSE3__ +# define __KERNEL_SSSE3__ +# endif +# ifdef __SSE4_1__ +# define __KERNEL_SSE41__ +# endif +# ifdef __AVX__ +# define __KERNEL_AVX__ +# endif +# ifdef __AVX2__ +# define __KERNEL_SSE__ +# define __KERNEL_AVX2__ +# endif +#endif + +/* quiet unused define warnings */ +#if defined(__KERNEL_SSE2__) + /* do nothing */ +#endif + +#include "kernel/kernel.h" +#define KERNEL_ARCH cpu +#include "kernel/kernels/cpu/kernel_cpu_impl.h" + diff --git a/intern/cycles/kernel/kernels/cpu/kernel_split_avx.cpp b/intern/cycles/kernel/kernels/cpu/kernel_split_avx.cpp new file mode 100644 index 00000000000..27a746a0799 --- /dev/null +++ b/intern/cycles/kernel/kernels/cpu/kernel_split_avx.cpp @@ -0,0 +1,38 @@ +/* + * Copyright 2011-2013 Blender Foundation + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +/* Optimized CPU kernel entry points. This file is compiled with AVX + * optimization flags and nearly all functions inlined, while kernel.cpp + * is compiled without for other CPU's. */ + +/* SSE optimization disabled for now on 32 bit, see bug #36316 */ +#if !(defined(__GNUC__) && (defined(i386) || defined(_M_IX86))) +# define __KERNEL_SSE2__ +# define __KERNEL_SSE3__ +# define __KERNEL_SSSE3__ +# define __KERNEL_SSE41__ +# define __KERNEL_AVX__ +#endif + +#define __SPLIT_KERNEL__ + +#include "util/util_optimization.h" + +#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_AVX +# include "kernel/kernel.h" +# define KERNEL_ARCH cpu_avx +# include "kernel/kernels/cpu/kernel_cpu_impl.h" +#endif /* WITH_CYCLES_OPTIMIZED_KERNEL_AVX */ diff --git a/intern/cycles/kernel/kernels/cpu/kernel_split_avx2.cpp b/intern/cycles/kernel/kernels/cpu/kernel_split_avx2.cpp new file mode 100644 index 00000000000..364d279a189 --- /dev/null +++ b/intern/cycles/kernel/kernels/cpu/kernel_split_avx2.cpp @@ -0,0 +1,40 @@ +/* + * Copyright 2011-2014 Blender Foundation + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +/* Optimized CPU kernel entry points. This file is compiled with AVX2 + * optimization flags and nearly all functions inlined, while kernel.cpp + * is compiled without for other CPU's. */ + +/* SSE optimization disabled for now on 32 bit, see bug #36316 */ +#if !(defined(__GNUC__) && (defined(i386) || defined(_M_IX86))) +# define __KERNEL_SSE__ +# define __KERNEL_SSE2__ +# define __KERNEL_SSE3__ +# define __KERNEL_SSSE3__ +# define __KERNEL_SSE41__ +# define __KERNEL_AVX__ +# define __KERNEL_AVX2__ +#endif + +#define __SPLIT_KERNEL__ + +#include "util/util_optimization.h" + +#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_AVX2 +# include "kernel/kernel.h" +# define KERNEL_ARCH cpu_avx2 +# include "kernel/kernels/cpu/kernel_cpu_impl.h" +#endif /* WITH_CYCLES_OPTIMIZED_KERNEL_AVX2 */ diff --git a/intern/cycles/kernel/kernels/cpu/kernel_split_sse2.cpp b/intern/cycles/kernel/kernels/cpu/kernel_split_sse2.cpp new file mode 100644 index 00000000000..0afb481296f --- /dev/null +++ b/intern/cycles/kernel/kernels/cpu/kernel_split_sse2.cpp @@ -0,0 +1,34 @@ +/* + * Copyright 2011-2013 Blender Foundation + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +/* Optimized CPU kernel entry points. This file is compiled with SSE2 + * optimization flags and nearly all functions inlined, while kernel.cpp + * is compiled without for other CPU's. */ + +/* SSE optimization disabled for now on 32 bit, see bug #36316 */ +#if !(defined(__GNUC__) && (defined(i386) || defined(_M_IX86))) +# define __KERNEL_SSE2__ +#endif + +#define __SPLIT_KERNEL__ + +#include "util/util_optimization.h" + +#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_SSE2 +# include "kernel/kernel.h" +# define KERNEL_ARCH cpu_sse2 +# include "kernel/kernels/cpu/kernel_cpu_impl.h" +#endif /* WITH_CYCLES_OPTIMIZED_KERNEL_SSE2 */ diff --git a/intern/cycles/kernel/kernels/cpu/kernel_split_sse3.cpp b/intern/cycles/kernel/kernels/cpu/kernel_split_sse3.cpp new file mode 100644 index 00000000000..13d00813591 --- /dev/null +++ b/intern/cycles/kernel/kernels/cpu/kernel_split_sse3.cpp @@ -0,0 +1,36 @@ +/* + * Copyright 2011-2013 Blender Foundation + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +/* Optimized CPU kernel entry points. This file is compiled with SSE3/SSSE3 + * optimization flags and nearly all functions inlined, while kernel.cpp + * is compiled without for other CPU's. */ + +/* SSE optimization disabled for now on 32 bit, see bug #36316 */ +#if !(defined(__GNUC__) && (defined(i386) || defined(_M_IX86))) +# define __KERNEL_SSE2__ +# define __KERNEL_SSE3__ +# define __KERNEL_SSSE3__ +#endif + +#define __SPLIT_KERNEL__ + +#include "util/util_optimization.h" + +#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_SSE3 +# include "kernel/kernel.h" +# define KERNEL_ARCH cpu_sse3 +# include "kernel/kernels/cpu/kernel_cpu_impl.h" +#endif /* WITH_CYCLES_OPTIMIZED_KERNEL_SSE3 */ diff --git a/intern/cycles/kernel/kernels/cpu/kernel_split_sse41.cpp b/intern/cycles/kernel/kernels/cpu/kernel_split_sse41.cpp new file mode 100644 index 00000000000..a4312071edc --- /dev/null +++ b/intern/cycles/kernel/kernels/cpu/kernel_split_sse41.cpp @@ -0,0 +1,37 @@ +/* + * Copyright 2011-2013 Blender Foundation + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +/* Optimized CPU kernel entry points. This file is compiled with SSE3/SSSE3 + * optimization flags and nearly all functions inlined, while kernel.cpp + * is compiled without for other CPU's. */ + +/* SSE optimization disabled for now on 32 bit, see bug #36316 */ +#if !(defined(__GNUC__) && (defined(i386) || defined(_M_IX86))) +# define __KERNEL_SSE2__ +# define __KERNEL_SSE3__ +# define __KERNEL_SSSE3__ +# define __KERNEL_SSE41__ +#endif + +#define __SPLIT_KERNEL__ + +#include "util/util_optimization.h" + +#ifdef WITH_CYCLES_OPTIMIZED_KERNEL_SSE41 +# include "kernel/kernel.h" +# define KERNEL_ARCH cpu_sse41 +# include "kernel/kernels/cpu/kernel_cpu_impl.h" +#endif /* WITH_CYCLES_OPTIMIZED_KERNEL_SSE41 */ diff --git a/intern/cycles/kernel/kernels/cpu/kernel_sse2.cpp b/intern/cycles/kernel/kernels/cpu/kernel_sse2.cpp index a5f2d6e7294..1acfaa91ac9 100644 --- a/intern/cycles/kernel/kernels/cpu/kernel_sse2.cpp +++ b/intern/cycles/kernel/kernels/cpu/kernel_sse2.cpp @@ -23,10 +23,10 @@ # define __KERNEL_SSE2__ #endif -#include "util_optimization.h" +#include "util/util_optimization.h" #ifdef WITH_CYCLES_OPTIMIZED_KERNEL_SSE2 -# include "kernel.h" +# include "kernel/kernel.h" # define KERNEL_ARCH cpu_sse2 -# include "kernel_cpu_impl.h" +# include "kernel/kernels/cpu/kernel_cpu_impl.h" #endif /* WITH_CYCLES_OPTIMIZED_KERNEL_SSE2 */ diff --git a/intern/cycles/kernel/kernels/cpu/kernel_sse3.cpp b/intern/cycles/kernel/kernels/cpu/kernel_sse3.cpp index 86f9ce991f8..f7b6a2e21fe 100644 --- a/intern/cycles/kernel/kernels/cpu/kernel_sse3.cpp +++ b/intern/cycles/kernel/kernels/cpu/kernel_sse3.cpp @@ -25,10 +25,10 @@ # define __KERNEL_SSSE3__ #endif -#include "util_optimization.h" +#include "util/util_optimization.h" #ifdef WITH_CYCLES_OPTIMIZED_KERNEL_SSE3 -# include "kernel.h" +# include "kernel/kernel.h" # define KERNEL_ARCH cpu_sse3 -# include "kernel_cpu_impl.h" +# include "kernel/kernels/cpu/kernel_cpu_impl.h" #endif /* WITH_CYCLES_OPTIMIZED_KERNEL_SSE3 */ diff --git a/intern/cycles/kernel/kernels/cpu/kernel_sse41.cpp b/intern/cycles/kernel/kernels/cpu/kernel_sse41.cpp index c174406047d..1900c6e3012 100644 --- a/intern/cycles/kernel/kernels/cpu/kernel_sse41.cpp +++ b/intern/cycles/kernel/kernels/cpu/kernel_sse41.cpp @@ -26,10 +26,10 @@ # define __KERNEL_SSE41__ #endif -#include "util_optimization.h" +#include "util/util_optimization.h" #ifdef WITH_CYCLES_OPTIMIZED_KERNEL_SSE41 -# include "kernel.h" +# include "kernel/kernel.h" # define KERNEL_ARCH cpu_sse41 -# include "kernel_cpu_impl.h" +# include "kernel/kernels/cpu//kernel_cpu_impl.h" #endif /* WITH_CYCLES_OPTIMIZED_KERNEL_SSE41 */ diff --git a/intern/cycles/kernel/kernels/cuda/kernel.cu b/intern/cycles/kernel/kernels/cuda/kernel.cu index eb2b6ea5414..dc343cb387a 100644 --- a/intern/cycles/kernel/kernels/cuda/kernel.cu +++ b/intern/cycles/kernel/kernels/cuda/kernel.cu @@ -16,113 +16,19 @@ /* CUDA kernel entry points */ -#include "../../kernel_compat_cuda.h" -#include "../../kernel_math.h" -#include "../../kernel_types.h" -#include "../../kernel_globals.h" -#include "../../kernel_film.h" -#include "../../kernel_path.h" -#include "../../kernel_path_branched.h" -#include "../../kernel_bake.h" - -/* device data taken from CUDA occupancy calculator */ - #ifdef __CUDA_ARCH__ -/* 2.0 and 2.1 */ -#if __CUDA_ARCH__ == 200 || __CUDA_ARCH__ == 210 -# define CUDA_MULTIPRESSOR_MAX_REGISTERS 32768 -# define CUDA_MULTIPROCESSOR_MAX_BLOCKS 8 -# define CUDA_BLOCK_MAX_THREADS 1024 -# define CUDA_THREAD_MAX_REGISTERS 63 - -/* tunable parameters */ -# define CUDA_THREADS_BLOCK_WIDTH 16 -# define CUDA_KERNEL_MAX_REGISTERS 32 -# define CUDA_KERNEL_BRANCHED_MAX_REGISTERS 40 - -/* 3.0 and 3.5 */ -#elif __CUDA_ARCH__ == 300 || __CUDA_ARCH__ == 350 -# define CUDA_MULTIPRESSOR_MAX_REGISTERS 65536 -# define CUDA_MULTIPROCESSOR_MAX_BLOCKS 16 -# define CUDA_BLOCK_MAX_THREADS 1024 -# define CUDA_THREAD_MAX_REGISTERS 63 - -/* tunable parameters */ -# define CUDA_THREADS_BLOCK_WIDTH 16 -# define CUDA_KERNEL_MAX_REGISTERS 63 -# define CUDA_KERNEL_BRANCHED_MAX_REGISTERS 63 - -/* 3.2 */ -#elif __CUDA_ARCH__ == 320 -# define CUDA_MULTIPRESSOR_MAX_REGISTERS 32768 -# define CUDA_MULTIPROCESSOR_MAX_BLOCKS 16 -# define CUDA_BLOCK_MAX_THREADS 1024 -# define CUDA_THREAD_MAX_REGISTERS 63 - -/* tunable parameters */ -# define CUDA_THREADS_BLOCK_WIDTH 16 -# define CUDA_KERNEL_MAX_REGISTERS 63 -# define CUDA_KERNEL_BRANCHED_MAX_REGISTERS 63 - -/* 3.7 */ -#elif __CUDA_ARCH__ == 370 -# define CUDA_MULTIPRESSOR_MAX_REGISTERS 65536 -# define CUDA_MULTIPROCESSOR_MAX_BLOCKS 16 -# define CUDA_BLOCK_MAX_THREADS 1024 -# define CUDA_THREAD_MAX_REGISTERS 255 - -/* tunable parameters */ -# define CUDA_THREADS_BLOCK_WIDTH 16 -# define CUDA_KERNEL_MAX_REGISTERS 63 -# define CUDA_KERNEL_BRANCHED_MAX_REGISTERS 63 - -/* 5.0, 5.2, 5.3, 6.0, 6.1 */ -#elif __CUDA_ARCH__ >= 500 -# define CUDA_MULTIPRESSOR_MAX_REGISTERS 65536 -# define CUDA_MULTIPROCESSOR_MAX_BLOCKS 32 -# define CUDA_BLOCK_MAX_THREADS 1024 -# define CUDA_THREAD_MAX_REGISTERS 255 - -/* tunable parameters */ -# define CUDA_THREADS_BLOCK_WIDTH 16 -# define CUDA_KERNEL_MAX_REGISTERS 48 -# define CUDA_KERNEL_BRANCHED_MAX_REGISTERS 63 - -/* unknown architecture */ -#else -# error "Unknown or unsupported CUDA architecture, can't determine launch bounds" -#endif - -/* compute number of threads per block and minimum blocks per multiprocessor - * given the maximum number of registers per thread */ - -#define CUDA_LAUNCH_BOUNDS(threads_block_width, thread_num_registers) \ - __launch_bounds__( \ - threads_block_width*threads_block_width, \ - CUDA_MULTIPRESSOR_MAX_REGISTERS/(threads_block_width*threads_block_width*thread_num_registers) \ - ) - -/* sanity checks */ - -#if CUDA_THREADS_BLOCK_WIDTH*CUDA_THREADS_BLOCK_WIDTH > CUDA_BLOCK_MAX_THREADS -# error "Maximum number of threads per block exceeded" -#endif - -#if CUDA_MULTIPRESSOR_MAX_REGISTERS/(CUDA_THREADS_BLOCK_WIDTH*CUDA_THREADS_BLOCK_WIDTH*CUDA_KERNEL_MAX_REGISTERS) > CUDA_MULTIPROCESSOR_MAX_BLOCKS -# error "Maximum number of blocks per multiprocessor exceeded" -#endif - -#if CUDA_KERNEL_MAX_REGISTERS > CUDA_THREAD_MAX_REGISTERS -# error "Maximum number of registers per thread exceeded" -#endif - -#if CUDA_KERNEL_BRANCHED_MAX_REGISTERS > CUDA_THREAD_MAX_REGISTERS -# error "Maximum number of registers per thread exceeded" -#endif +#include "kernel/kernel_compat_cuda.h" +#include "kernel_config.h" +#include "kernel/kernel_math.h" +#include "kernel/kernel_types.h" +#include "kernel/kernel_globals.h" +#include "kernel/kernel_film.h" +#include "kernel/kernel_path.h" +#include "kernel/kernel_path_branched.h" +#include "kernel/kernel_bake.h" /* kernels */ - extern "C" __global__ void CUDA_LAUNCH_BOUNDS(CUDA_THREADS_BLOCK_WIDTH, CUDA_KERNEL_MAX_REGISTERS) kernel_cuda_path_trace(float *buffer, uint *rng_state, int sample, int sx, int sy, int sw, int sh, int offset, int stride) @@ -130,8 +36,10 @@ kernel_cuda_path_trace(float *buffer, uint *rng_state, int sample, int sx, int s int x = sx + blockDim.x*blockIdx.x + threadIdx.x; int y = sy + blockDim.y*blockIdx.y + threadIdx.y; - if(x < sx + sw && y < sy + sh) - kernel_path_trace(NULL, buffer, rng_state, sample, x, y, offset, stride); + if(x < sx + sw && y < sy + sh) { + KernelGlobals kg; + kernel_path_trace(&kg, buffer, rng_state, sample, x, y, offset, stride); + } } #ifdef __BRANCHED_PATH__ @@ -142,8 +50,10 @@ kernel_cuda_branched_path_trace(float *buffer, uint *rng_state, int sample, int int x = sx + blockDim.x*blockIdx.x + threadIdx.x; int y = sy + blockDim.y*blockIdx.y + threadIdx.y; - if(x < sx + sw && y < sy + sh) - kernel_branched_path_trace(NULL, buffer, rng_state, sample, x, y, offset, stride); + if(x < sx + sw && y < sy + sh) { + KernelGlobals kg; + kernel_branched_path_trace(&kg, buffer, rng_state, sample, x, y, offset, stride); + } } #endif @@ -154,8 +64,9 @@ kernel_cuda_convert_to_byte(uchar4 *rgba, float *buffer, float sample_scale, int int x = sx + blockDim.x*blockIdx.x + threadIdx.x; int y = sy + blockDim.y*blockIdx.y + threadIdx.y; - if(x < sx + sw && y < sy + sh) + if(x < sx + sw && y < sy + sh) { kernel_film_convert_to_byte(NULL, rgba, buffer, sample_scale, x, y, offset, stride); + } } extern "C" __global__ void @@ -165,8 +76,9 @@ kernel_cuda_convert_to_half_float(uchar4 *rgba, float *buffer, float sample_scal int x = sx + blockDim.x*blockIdx.x + threadIdx.x; int y = sy + blockDim.y*blockIdx.y + threadIdx.y; - if(x < sx + sw && y < sy + sh) + if(x < sx + sw && y < sy + sh) { kernel_film_convert_to_half_float(NULL, rgba, buffer, sample_scale, x, y, offset, stride); + } } extern "C" __global__ void @@ -183,7 +95,8 @@ kernel_cuda_shader(uint4 *input, int x = sx + blockDim.x*blockIdx.x + threadIdx.x; if(x < sx + sw) { - kernel_shader_evaluate(NULL, + KernelGlobals kg; + kernel_shader_evaluate(&kg, input, output, output_luma, @@ -200,8 +113,10 @@ kernel_cuda_bake(uint4 *input, float4 *output, int type, int filter, int sx, int { int x = sx + blockDim.x*blockIdx.x + threadIdx.x; - if(x < sx + sw) - kernel_bake_evaluate(NULL, input, output, (ShaderEvalType)type, filter, x, offset, sample); + if(x < sx + sw) { + KernelGlobals kg; + kernel_bake_evaluate(&kg, input, output, (ShaderEvalType)type, filter, x, offset, sample); + } } #endif diff --git a/intern/cycles/kernel/kernels/cuda/kernel_config.h b/intern/cycles/kernel/kernels/cuda/kernel_config.h new file mode 100644 index 00000000000..9fa39dc9ebb --- /dev/null +++ b/intern/cycles/kernel/kernels/cuda/kernel_config.h @@ -0,0 +1,110 @@ +/* + * Copyright 2011-2013 Blender Foundation + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +/* device data taken from CUDA occupancy calculator */ + +/* 2.0 and 2.1 */ +#if __CUDA_ARCH__ == 200 || __CUDA_ARCH__ == 210 +# define CUDA_MULTIPRESSOR_MAX_REGISTERS 32768 +# define CUDA_MULTIPROCESSOR_MAX_BLOCKS 8 +# define CUDA_BLOCK_MAX_THREADS 1024 +# define CUDA_THREAD_MAX_REGISTERS 63 + +/* tunable parameters */ +# define CUDA_THREADS_BLOCK_WIDTH 16 +# define CUDA_KERNEL_MAX_REGISTERS 32 +# define CUDA_KERNEL_BRANCHED_MAX_REGISTERS 40 + +/* 3.0 and 3.5 */ +#elif __CUDA_ARCH__ == 300 || __CUDA_ARCH__ == 350 +# define CUDA_MULTIPRESSOR_MAX_REGISTERS 65536 +# define CUDA_MULTIPROCESSOR_MAX_BLOCKS 16 +# define CUDA_BLOCK_MAX_THREADS 1024 +# define CUDA_THREAD_MAX_REGISTERS 63 + +/* tunable parameters */ +# define CUDA_THREADS_BLOCK_WIDTH 16 +# define CUDA_KERNEL_MAX_REGISTERS 63 +# define CUDA_KERNEL_BRANCHED_MAX_REGISTERS 63 + +/* 3.2 */ +#elif __CUDA_ARCH__ == 320 +# define CUDA_MULTIPRESSOR_MAX_REGISTERS 32768 +# define CUDA_MULTIPROCESSOR_MAX_BLOCKS 16 +# define CUDA_BLOCK_MAX_THREADS 1024 +# define CUDA_THREAD_MAX_REGISTERS 63 + +/* tunable parameters */ +# define CUDA_THREADS_BLOCK_WIDTH 16 +# define CUDA_KERNEL_MAX_REGISTERS 63 +# define CUDA_KERNEL_BRANCHED_MAX_REGISTERS 63 + +/* 3.7 */ +#elif __CUDA_ARCH__ == 370 +# define CUDA_MULTIPRESSOR_MAX_REGISTERS 65536 +# define CUDA_MULTIPROCESSOR_MAX_BLOCKS 16 +# define CUDA_BLOCK_MAX_THREADS 1024 +# define CUDA_THREAD_MAX_REGISTERS 255 + +/* tunable parameters */ +# define CUDA_THREADS_BLOCK_WIDTH 16 +# define CUDA_KERNEL_MAX_REGISTERS 63 +# define CUDA_KERNEL_BRANCHED_MAX_REGISTERS 63 + +/* 5.0, 5.2, 5.3, 6.0, 6.1 */ +#elif __CUDA_ARCH__ >= 500 +# define CUDA_MULTIPRESSOR_MAX_REGISTERS 65536 +# define CUDA_MULTIPROCESSOR_MAX_BLOCKS 32 +# define CUDA_BLOCK_MAX_THREADS 1024 +# define CUDA_THREAD_MAX_REGISTERS 255 + +/* tunable parameters */ +# define CUDA_THREADS_BLOCK_WIDTH 16 +# define CUDA_KERNEL_MAX_REGISTERS 48 +# define CUDA_KERNEL_BRANCHED_MAX_REGISTERS 63 + +/* unknown architecture */ +#else +# error "Unknown or unsupported CUDA architecture, can't determine launch bounds" +#endif + +/* compute number of threads per block and minimum blocks per multiprocessor + * given the maximum number of registers per thread */ + +#define CUDA_LAUNCH_BOUNDS(threads_block_width, thread_num_registers) \ + __launch_bounds__( \ + threads_block_width*threads_block_width, \ + CUDA_MULTIPRESSOR_MAX_REGISTERS/(threads_block_width*threads_block_width*thread_num_registers) \ + ) + +/* sanity checks */ + +#if CUDA_THREADS_BLOCK_WIDTH*CUDA_THREADS_BLOCK_WIDTH > CUDA_BLOCK_MAX_THREADS +# error "Maximum number of threads per block exceeded" +#endif + +#if CUDA_MULTIPRESSOR_MAX_REGISTERS/(CUDA_THREADS_BLOCK_WIDTH*CUDA_THREADS_BLOCK_WIDTH*CUDA_KERNEL_MAX_REGISTERS) > CUDA_MULTIPROCESSOR_MAX_BLOCKS +# error "Maximum number of blocks per multiprocessor exceeded" +#endif + +#if CUDA_KERNEL_MAX_REGISTERS > CUDA_THREAD_MAX_REGISTERS +# error "Maximum number of registers per thread exceeded" +#endif + +#if CUDA_KERNEL_BRANCHED_MAX_REGISTERS > CUDA_THREAD_MAX_REGISTERS +# error "Maximum number of registers per thread exceeded" +#endif + diff --git a/intern/cycles/kernel/kernels/cuda/kernel_split.cu b/intern/cycles/kernel/kernels/cuda/kernel_split.cu new file mode 100644 index 00000000000..a679eff8409 --- /dev/null +++ b/intern/cycles/kernel/kernels/cuda/kernel_split.cu @@ -0,0 +1,144 @@ +/* + * Copyright 2011-2016 Blender Foundation + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +/* CUDA split kernel entry points */ + +#ifdef __CUDA_ARCH__ + +#define __SPLIT_KERNEL__ + +#include "kernel/kernel_compat_cuda.h" +#include "kernel_config.h" + +#include "kernel/split/kernel_split_common.h" +#include "kernel/split/kernel_data_init.h" +#include "kernel/split/kernel_path_init.h" +#include "kernel/split/kernel_scene_intersect.h" +#include "kernel/split/kernel_lamp_emission.h" +#include "kernel/split/kernel_do_volume.h" +#include "kernel/split/kernel_queue_enqueue.h" +#include "kernel/split/kernel_indirect_background.h" +#include "kernel/split/kernel_shader_eval.h" +#include "kernel/split/kernel_holdout_emission_blurring_pathtermination_ao.h" +#include "kernel/split/kernel_subsurface_scatter.h" +#include "kernel/split/kernel_direct_lighting.h" +#include "kernel/split/kernel_shadow_blocked_ao.h" +#include "kernel/split/kernel_shadow_blocked_dl.h" +#include "kernel/split/kernel_next_iteration_setup.h" +#include "kernel/split/kernel_indirect_subsurface.h" +#include "kernel/split/kernel_buffer_update.h" + +#include "kernel/kernel_film.h" + +/* kernels */ +extern "C" __global__ void +CUDA_LAUNCH_BOUNDS(CUDA_THREADS_BLOCK_WIDTH, CUDA_KERNEL_MAX_REGISTERS) +kernel_cuda_state_buffer_size(uint num_threads, uint64_t *size) +{ + *size = split_data_buffer_size(NULL, num_threads); +} + +extern "C" __global__ void +CUDA_LAUNCH_BOUNDS(CUDA_THREADS_BLOCK_WIDTH, CUDA_KERNEL_MAX_REGISTERS) +kernel_cuda_path_trace_data_init( + ccl_global void *split_data_buffer, + int num_elements, + ccl_global char *ray_state, + ccl_global uint *rng_state, + int start_sample, + int end_sample, + int sx, int sy, int sw, int sh, int offset, int stride, + ccl_global int *Queue_index, + int queuesize, + ccl_global char *use_queues_flag, + ccl_global unsigned int *work_pool_wgs, + unsigned int num_samples, + ccl_global float *buffer) +{ + kernel_data_init(NULL, + NULL, + split_data_buffer, + num_elements, + ray_state, + rng_state, + start_sample, + end_sample, + sx, sy, sw, sh, offset, stride, + Queue_index, + queuesize, + use_queues_flag, + work_pool_wgs, + num_samples, + buffer); +} + +#define DEFINE_SPLIT_KERNEL_FUNCTION(name) \ + extern "C" __global__ void \ + CUDA_LAUNCH_BOUNDS(CUDA_THREADS_BLOCK_WIDTH, CUDA_KERNEL_MAX_REGISTERS) \ + kernel_cuda_##name() \ + { \ + kernel_##name(NULL); \ + } + +#define DEFINE_SPLIT_KERNEL_FUNCTION_LOCALS(name, type) \ + extern "C" __global__ void \ + CUDA_LAUNCH_BOUNDS(CUDA_THREADS_BLOCK_WIDTH, CUDA_KERNEL_MAX_REGISTERS) \ + kernel_cuda_##name() \ + { \ + ccl_local type locals; \ + kernel_##name(NULL, &locals); \ + } + +DEFINE_SPLIT_KERNEL_FUNCTION(path_init) +DEFINE_SPLIT_KERNEL_FUNCTION(scene_intersect) +DEFINE_SPLIT_KERNEL_FUNCTION(lamp_emission) +DEFINE_SPLIT_KERNEL_FUNCTION(do_volume) +DEFINE_SPLIT_KERNEL_FUNCTION_LOCALS(queue_enqueue, QueueEnqueueLocals) +DEFINE_SPLIT_KERNEL_FUNCTION(indirect_background) +DEFINE_SPLIT_KERNEL_FUNCTION_LOCALS(shader_eval, uint) +DEFINE_SPLIT_KERNEL_FUNCTION_LOCALS(holdout_emission_blurring_pathtermination_ao, BackgroundAOLocals) +DEFINE_SPLIT_KERNEL_FUNCTION_LOCALS(subsurface_scatter, uint) +DEFINE_SPLIT_KERNEL_FUNCTION_LOCALS(direct_lighting, uint) +DEFINE_SPLIT_KERNEL_FUNCTION(shadow_blocked_ao) +DEFINE_SPLIT_KERNEL_FUNCTION(shadow_blocked_dl) +DEFINE_SPLIT_KERNEL_FUNCTION_LOCALS(next_iteration_setup, uint) +DEFINE_SPLIT_KERNEL_FUNCTION(indirect_subsurface) +DEFINE_SPLIT_KERNEL_FUNCTION_LOCALS(buffer_update, uint) + +extern "C" __global__ void +CUDA_LAUNCH_BOUNDS(CUDA_THREADS_BLOCK_WIDTH, CUDA_KERNEL_MAX_REGISTERS) +kernel_cuda_convert_to_byte(uchar4 *rgba, float *buffer, float sample_scale, int sx, int sy, int sw, int sh, int offset, int stride) +{ + int x = sx + blockDim.x*blockIdx.x + threadIdx.x; + int y = sy + blockDim.y*blockIdx.y + threadIdx.y; + + if(x < sx + sw && y < sy + sh) + kernel_film_convert_to_byte(NULL, rgba, buffer, sample_scale, x, y, offset, stride); +} + +extern "C" __global__ void +CUDA_LAUNCH_BOUNDS(CUDA_THREADS_BLOCK_WIDTH, CUDA_KERNEL_MAX_REGISTERS) +kernel_cuda_convert_to_half_float(uchar4 *rgba, float *buffer, float sample_scale, int sx, int sy, int sw, int sh, int offset, int stride) +{ + int x = sx + blockDim.x*blockIdx.x + threadIdx.x; + int y = sy + blockDim.y*blockIdx.y + threadIdx.y; + + if(x < sx + sw && y < sy + sh) + kernel_film_convert_to_half_float(NULL, rgba, buffer, sample_scale, x, y, offset, stride); +} + +#endif + diff --git a/intern/cycles/kernel/kernels/opencl/kernel.cl b/intern/cycles/kernel/kernels/opencl/kernel.cl index a68f97857b6..078acc1631e 100644 --- a/intern/cycles/kernel/kernels/opencl/kernel.cl +++ b/intern/cycles/kernel/kernels/opencl/kernel.cl @@ -16,34 +16,34 @@ /* OpenCL kernel entry points - unfinished */ -#include "../../kernel_compat_opencl.h" -#include "../../kernel_math.h" -#include "../../kernel_types.h" -#include "../../kernel_globals.h" -#include "../../kernel_image_opencl.h" +#include "kernel/kernel_compat_opencl.h" +#include "kernel/kernel_math.h" +#include "kernel/kernel_types.h" +#include "kernel/kernel_globals.h" +#include "kernel/kernel_image_opencl.h" -#include "../../kernel_film.h" +#include "kernel/kernel_film.h" #if defined(__COMPILE_ONLY_MEGAKERNEL__) || !defined(__NO_BAKING__) -# include "../../kernel_path.h" -# include "../../kernel_path_branched.h" +# include "kernel/kernel_path.h" +# include "kernel/kernel_path_branched.h" #else /* __COMPILE_ONLY_MEGAKERNEL__ */ /* Include only actually used headers for the case * when path tracing kernels are not needed. */ -# include "../../kernel_random.h" -# include "../../kernel_differential.h" -# include "../../kernel_montecarlo.h" -# include "../../kernel_projection.h" -# include "../../geom/geom.h" -# include "../../bvh/bvh.h" - -# include "../../kernel_accumulate.h" -# include "../../kernel_camera.h" -# include "../../kernel_shader.h" +# include "kernel/kernel_random.h" +# include "kernel/kernel_differential.h" +# include "kernel/kernel_montecarlo.h" +# include "kernel/kernel_projection.h" +# include "kernel/geom/geom.h" +# include "kernel/bvh/bvh.h" + +# include "kernel/kernel_accumulate.h" +# include "kernel/kernel_camera.h" +# include "kernel/kernel_shader.h" #endif /* defined(__COMPILE_ONLY_MEGAKERNEL__) || !defined(__NO_BAKING__) */ -#include "../../kernel_bake.h" +#include "kernel/kernel_bake.h" #ifdef __COMPILE_ONLY_MEGAKERNEL__ @@ -54,7 +54,7 @@ __kernel void kernel_ocl_path_trace( #define KERNEL_TEX(type, ttype, name) \ ccl_global type *name, -#include "../../kernel_textures.h" +#include "kernel/kernel_textures.h" int sample, int sx, int sy, int sw, int sh, int offset, int stride) @@ -65,10 +65,10 @@ __kernel void kernel_ocl_path_trace( #define KERNEL_TEX(type, ttype, name) \ kg->name = name; -#include "../../kernel_textures.h" +#include "kernel/kernel_textures.h" - int x = sx + get_global_id(0); - int y = sy + get_global_id(1); + int x = sx + ccl_global_id(0); + int y = sy + ccl_global_id(1); if(x < sx + sw && y < sy + sh) kernel_path_trace(kg, buffer, rng_state, sample, x, y, offset, stride); @@ -84,7 +84,7 @@ __kernel void kernel_ocl_shader( #define KERNEL_TEX(type, ttype, name) \ ccl_global type *name, -#include "../../kernel_textures.h" +#include "kernel/kernel_textures.h" int type, int sx, int sw, int offset, int sample) { @@ -94,9 +94,9 @@ __kernel void kernel_ocl_shader( #define KERNEL_TEX(type, ttype, name) \ kg->name = name; -#include "../../kernel_textures.h" +#include "kernel/kernel_textures.h" - int x = sx + get_global_id(0); + int x = sx + ccl_global_id(0); if(x < sx + sw) { kernel_shader_evaluate(kg, @@ -116,7 +116,7 @@ __kernel void kernel_ocl_bake( #define KERNEL_TEX(type, ttype, name) \ ccl_global type *name, -#include "../../kernel_textures.h" +#include "kernel/kernel_textures.h" int type, int filter, int sx, int sw, int offset, int sample) { @@ -126,9 +126,9 @@ __kernel void kernel_ocl_bake( #define KERNEL_TEX(type, ttype, name) \ kg->name = name; -#include "../../kernel_textures.h" +#include "kernel/kernel_textures.h" - int x = sx + get_global_id(0); + int x = sx + ccl_global_id(0); if(x < sx + sw) { #ifdef __NO_BAKING__ @@ -146,7 +146,7 @@ __kernel void kernel_ocl_convert_to_byte( #define KERNEL_TEX(type, ttype, name) \ ccl_global type *name, -#include "../../kernel_textures.h" +#include "kernel/kernel_textures.h" float sample_scale, int sx, int sy, int sw, int sh, int offset, int stride) @@ -157,10 +157,10 @@ __kernel void kernel_ocl_convert_to_byte( #define KERNEL_TEX(type, ttype, name) \ kg->name = name; -#include "../../kernel_textures.h" +#include "kernel/kernel_textures.h" - int x = sx + get_global_id(0); - int y = sy + get_global_id(1); + int x = sx + ccl_global_id(0); + int y = sy + ccl_global_id(1); if(x < sx + sw && y < sy + sh) kernel_film_convert_to_byte(kg, rgba, buffer, sample_scale, x, y, offset, stride); @@ -173,7 +173,7 @@ __kernel void kernel_ocl_convert_to_half_float( #define KERNEL_TEX(type, ttype, name) \ ccl_global type *name, -#include "../../kernel_textures.h" +#include "kernel/kernel_textures.h" float sample_scale, int sx, int sy, int sw, int sh, int offset, int stride) @@ -184,13 +184,29 @@ __kernel void kernel_ocl_convert_to_half_float( #define KERNEL_TEX(type, ttype, name) \ kg->name = name; -#include "../../kernel_textures.h" +#include "kernel/kernel_textures.h" - int x = sx + get_global_id(0); - int y = sy + get_global_id(1); + int x = sx + ccl_global_id(0); + int y = sy + ccl_global_id(1); if(x < sx + sw && y < sy + sh) kernel_film_convert_to_half_float(kg, rgba, buffer, sample_scale, x, y, offset, stride); } +__kernel void kernel_ocl_zero_buffer(ccl_global float4 *buffer, ulong size, ulong offset) +{ + size_t i = ccl_global_id(0) + ccl_global_id(1) * ccl_global_size(0); + + if(i < size / sizeof(float4)) { + buffer[i+offset/sizeof(float4)] = make_float4(0.0f, 0.0f, 0.0f, 0.0f); + } + else if(i == size / sizeof(float4)) { + ccl_global uchar *b = (ccl_global uchar*)&buffer[i+offset/sizeof(float4)]; + + for(i = 0; i < size % sizeof(float4); i++) { + *(b++) = 0; + } + } +} + #endif /* __COMPILE_ONLY_MEGAKERNEL__ */ diff --git a/intern/cycles/kernel/kernels/opencl/kernel_background_buffer_update.cl b/intern/cycles/kernel/kernels/opencl/kernel_background_buffer_update.cl deleted file mode 100644 index 1914d241eb1..00000000000 --- a/intern/cycles/kernel/kernels/opencl/kernel_background_buffer_update.cl +++ /dev/null @@ -1,125 +0,0 @@ -/* - * Copyright 2011-2015 Blender Foundation - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -#include "split/kernel_background_buffer_update.h" - -__kernel void kernel_ocl_path_trace_background_buffer_update( - ccl_global char *kg, - ccl_constant KernelData *data, - ccl_global float *per_sample_output_buffers, - ccl_global uint *rng_state, - ccl_global uint *rng_coop, /* Required for buffer Update */ - ccl_global float3 *throughput_coop, /* Required for background hit processing */ - PathRadiance *PathRadiance_coop, /* Required for background hit processing and buffer Update */ - ccl_global Ray *Ray_coop, /* Required for background hit processing */ - ccl_global PathState *PathState_coop, /* Required for background hit processing */ - ccl_global float *L_transparent_coop, /* Required for background hit processing and buffer Update */ - ccl_global char *ray_state, /* Stores information on the current state of a ray */ - int sw, int sh, int sx, int sy, int stride, - int rng_state_offset_x, - int rng_state_offset_y, - int rng_state_stride, - ccl_global unsigned int *work_array, /* Denotes work of each ray */ - ccl_global int *Queue_data, /* Queues memory */ - ccl_global int *Queue_index, /* Tracks the number of elements in each queue */ - int queuesize, /* Size (capacity) of each queue */ - int end_sample, - int start_sample, -#ifdef __WORK_STEALING__ - ccl_global unsigned int *work_pool_wgs, - unsigned int num_samples, -#endif -#ifdef __KERNEL_DEBUG__ - DebugData *debugdata_coop, -#endif - int parallel_samples) /* Number of samples to be processed in parallel */ -{ - ccl_local unsigned int local_queue_atomics; - if(get_local_id(0) == 0 && get_local_id(1) == 0) { - local_queue_atomics = 0; - } - barrier(CLK_LOCAL_MEM_FENCE); - - int ray_index = get_global_id(1) * get_global_size(0) + get_global_id(0); - if(ray_index == 0) { - /* We will empty this queue in this kernel. */ - Queue_index[QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS] = 0; - } - char enqueue_flag = 0; - ray_index = get_ray_index(ray_index, - QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS, - Queue_data, - queuesize, - 1); - -#ifdef __COMPUTE_DEVICE_GPU__ - /* If we are executing on a GPU device, we exit all threads that are not - * required. - * - * If we are executing on a CPU device, then we need to keep all threads - * active since we have barrier() calls later in the kernel. CPU devices, - * expect all threads to execute barrier statement. - */ - if(ray_index == QUEUE_EMPTY_SLOT) { - return; - } -#endif - -#ifndef __COMPUTE_DEVICE_GPU__ - if(ray_index != QUEUE_EMPTY_SLOT) { -#endif - enqueue_flag = - kernel_background_buffer_update((KernelGlobals *)kg, - per_sample_output_buffers, - rng_state, - rng_coop, - throughput_coop, - PathRadiance_coop, - Ray_coop, - PathState_coop, - L_transparent_coop, - ray_state, - sw, sh, sx, sy, stride, - rng_state_offset_x, - rng_state_offset_y, - rng_state_stride, - work_array, - end_sample, - start_sample, -#ifdef __WORK_STEALING__ - work_pool_wgs, - num_samples, -#endif -#ifdef __KERNEL_DEBUG__ - debugdata_coop, -#endif - parallel_samples, - ray_index); -#ifndef __COMPUTE_DEVICE_GPU__ - } -#endif - - /* Enqueue RAY_REGENERATED rays into QUEUE_ACTIVE_AND_REGENERATED_RAYS; - * These rays will be made active during next SceneIntersectkernel. - */ - enqueue_ray_index_local(ray_index, - QUEUE_ACTIVE_AND_REGENERATED_RAYS, - enqueue_flag, - queuesize, - &local_queue_atomics, - Queue_data, - Queue_index); -} diff --git a/intern/cycles/kernel/kernels/opencl/kernel_buffer_update.cl b/intern/cycles/kernel/kernels/opencl/kernel_buffer_update.cl new file mode 100644 index 00000000000..db65c91baf7 --- /dev/null +++ b/intern/cycles/kernel/kernels/opencl/kernel_buffer_update.cl @@ -0,0 +1,27 @@ +/* + * Copyright 2011-2015 Blender Foundation + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#include "kernel/kernel_compat_opencl.h" +#include "kernel/split/kernel_split_common.h" +#include "kernel/split/kernel_buffer_update.h" + +__kernel void kernel_ocl_path_trace_buffer_update( + ccl_global char *kg, + ccl_constant KernelData *data) +{ + ccl_local unsigned int local_queue_atomics; + kernel_buffer_update((KernelGlobals*)kg, &local_queue_atomics); +} diff --git a/intern/cycles/kernel/kernels/opencl/kernel_data_init.cl b/intern/cycles/kernel/kernels/opencl/kernel_data_init.cl index 18139687eab..8b85d362f8a 100644 --- a/intern/cycles/kernel/kernels/opencl/kernel_data_init.cl +++ b/intern/cycles/kernel/kernels/opencl/kernel_data_init.cl @@ -14,77 +14,49 @@ * limitations under the License. */ -#include "split/kernel_data_init.h" +#include "kernel/kernel_compat_opencl.h" +#include "kernel/split/kernel_split_common.h" +#include "kernel/split/kernel_data_init.h" __kernel void kernel_ocl_path_trace_data_init( - ccl_global char *globals, - ccl_global char *sd_DL_shadow, + ccl_global char *kg, ccl_constant KernelData *data, - ccl_global float *per_sample_output_buffers, + ccl_global void *split_data_buffer, + int num_elements, + ccl_global char *ray_state, ccl_global uint *rng_state, - ccl_global uint *rng_coop, /* rng array to store rng values for all rays */ - ccl_global float3 *throughput_coop, /* throughput array to store throughput values for all rays */ - ccl_global float *L_transparent_coop, /* L_transparent array to store L_transparent values for all rays */ - PathRadiance *PathRadiance_coop, /* PathRadiance array to store PathRadiance values for all rays */ - ccl_global Ray *Ray_coop, /* Ray array to store Ray information for all rays */ - ccl_global PathState *PathState_coop, /* PathState array to store PathState information for all rays */ - Intersection *Intersection_coop_shadow, - ccl_global char *ray_state, /* Stores information on current state of a ray */ #define KERNEL_TEX(type, ttype, name) \ ccl_global type *name, -#include "../../kernel_textures.h" +#include "kernel/kernel_textures.h" - int start_sample, int sx, int sy, int sw, int sh, int offset, int stride, - int rng_state_offset_x, - int rng_state_offset_y, - int rng_state_stride, - ccl_global int *Queue_data, /* Memory for queues */ + int start_sample, + int end_sample, + int sx, int sy, int sw, int sh, int offset, int stride, ccl_global int *Queue_index, /* Tracks the number of elements in queues */ int queuesize, /* size (capacity) of the queue */ ccl_global char *use_queues_flag, /* flag to decide if scene-intersect kernel should use queues to fetch ray index */ - ccl_global unsigned int *work_array, /* work array to store which work each ray belongs to */ -#ifdef __WORK_STEALING__ ccl_global unsigned int *work_pool_wgs, /* Work pool for each work group */ unsigned int num_samples, /* Total number of samples per pixel */ -#endif -#ifdef __KERNEL_DEBUG__ - DebugData *debugdata_coop, -#endif - int parallel_samples) /* Number of samples to be processed in parallel */ + ccl_global float *buffer) { - kernel_data_init((KernelGlobals *)globals, - (ShaderData *)sd_DL_shadow, + kernel_data_init((KernelGlobals*)kg, data, - per_sample_output_buffers, - rng_state, - rng_coop, - throughput_coop, - L_transparent_coop, - PathRadiance_coop, - Ray_coop, - PathState_coop, - Intersection_coop_shadow, + split_data_buffer, + num_elements, ray_state, + rng_state, #define KERNEL_TEX(type, ttype, name) name, -#include "../../kernel_textures.h" +#include "kernel/kernel_textures.h" - start_sample, sx, sy, sw, sh, offset, stride, - rng_state_offset_x, - rng_state_offset_y, - rng_state_stride, - Queue_data, + start_sample, + end_sample, + sx, sy, sw, sh, offset, stride, Queue_index, queuesize, use_queues_flag, - work_array, -#ifdef __WORK_STEALING__ work_pool_wgs, num_samples, -#endif -#ifdef __KERNEL_DEBUG__ - debugdata_coop, -#endif - parallel_samples); + buffer); } diff --git a/intern/cycles/kernel/kernels/opencl/kernel_direct_lighting.cl b/intern/cycles/kernel/kernels/opencl/kernel_direct_lighting.cl index c6a2c8d050c..eb34f750881 100644 --- a/intern/cycles/kernel/kernels/opencl/kernel_direct_lighting.cl +++ b/intern/cycles/kernel/kernels/opencl/kernel_direct_lighting.cl @@ -14,74 +14,14 @@ * limitations under the License. */ -#include "split/kernel_direct_lighting.h" +#include "kernel/kernel_compat_opencl.h" +#include "kernel/split/kernel_split_common.h" +#include "kernel/split/kernel_direct_lighting.h" __kernel void kernel_ocl_path_trace_direct_lighting( ccl_global char *kg, - ccl_constant KernelData *data, - ccl_global char *sd, /* Required for direct lighting */ - ccl_global uint *rng_coop, /* Required for direct lighting */ - ccl_global PathState *PathState_coop, /* Required for direct lighting */ - ccl_global int *ISLamp_coop, /* Required for direct lighting */ - ccl_global Ray *LightRay_coop, /* Required for direct lighting */ - ccl_global BsdfEval *BSDFEval_coop, /* Required for direct lighting */ - ccl_global char *ray_state, /* Denotes the state of each ray */ - ccl_global int *Queue_data, /* Queue memory */ - ccl_global int *Queue_index, /* Tracks the number of elements in each queue */ - int queuesize) /* Size (capacity) of each queue */ + ccl_constant KernelData *data) { ccl_local unsigned int local_queue_atomics; - if(get_local_id(0) == 0 && get_local_id(1) == 0) { - local_queue_atomics = 0; - } - barrier(CLK_LOCAL_MEM_FENCE); - - char enqueue_flag = 0; - int ray_index = get_global_id(1) * get_global_size(0) + get_global_id(0); - ray_index = get_ray_index(ray_index, - QUEUE_ACTIVE_AND_REGENERATED_RAYS, - Queue_data, - queuesize, - 0); - -#ifdef __COMPUTE_DEVICE_GPU__ - /* If we are executing on a GPU device, we exit all threads that are not - * required. - * - * If we are executing on a CPU device, then we need to keep all threads - * active since we have barrier() calls later in the kernel. CPU devices, - * expect all threads to execute barrier statement. - */ - if(ray_index == QUEUE_EMPTY_SLOT) { - return; - } -#endif - -#ifndef __COMPUTE_DEVICE_GPU__ - if(ray_index != QUEUE_EMPTY_SLOT) { -#endif - enqueue_flag = kernel_direct_lighting((KernelGlobals *)kg, - (ShaderData *)sd, - rng_coop, - PathState_coop, - ISLamp_coop, - LightRay_coop, - BSDFEval_coop, - ray_state, - ray_index); - -#ifndef __COMPUTE_DEVICE_GPU__ - } -#endif - -#ifdef __EMISSION__ - /* Enqueue RAY_SHADOW_RAY_CAST_DL rays. */ - enqueue_ray_index_local(ray_index, - QUEUE_SHADOW_RAY_CAST_DL_RAYS, - enqueue_flag, - queuesize, - &local_queue_atomics, - Queue_data, - Queue_index); -#endif + kernel_direct_lighting((KernelGlobals*)kg, &local_queue_atomics); } diff --git a/intern/cycles/kernel/kernels/opencl/kernel_do_volume.cl b/intern/cycles/kernel/kernels/opencl/kernel_do_volume.cl new file mode 100644 index 00000000000..83ef5f5f3f2 --- /dev/null +++ b/intern/cycles/kernel/kernels/opencl/kernel_do_volume.cl @@ -0,0 +1,26 @@ +/* + * Copyright 2011-2017 Blender Foundation + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#include "kernel/kernel_compat_opencl.h" +#include "kernel/split/kernel_split_common.h" +#include "kernel/split/kernel_do_volume.h" + +__kernel void kernel_ocl_path_trace_do_volume( + ccl_global char *kg, + ccl_constant KernelData *data) +{ + kernel_do_volume((KernelGlobals*)kg); +} diff --git a/intern/cycles/kernel/kernels/opencl/kernel_holdout_emission_blurring_pathtermination_ao.cl b/intern/cycles/kernel/kernels/opencl/kernel_holdout_emission_blurring_pathtermination_ao.cl index e063614da1a..d071b39aa6f 100644 --- a/intern/cycles/kernel/kernels/opencl/kernel_holdout_emission_blurring_pathtermination_ao.cl +++ b/intern/cycles/kernel/kernels/opencl/kernel_holdout_emission_blurring_pathtermination_ao.cl @@ -14,110 +14,16 @@ * limitations under the License. */ -#include "split/kernel_holdout_emission_blurring_pathtermination_ao.h" +#include "kernel/kernel_compat_opencl.h" +#include "kernel/split/kernel_split_common.h" +#include "kernel/split/kernel_holdout_emission_blurring_pathtermination_ao.h" __kernel void kernel_ocl_path_trace_holdout_emission_blurring_pathtermination_ao( ccl_global char *kg, - ccl_constant KernelData *data, - ccl_global char *sd, /* Required throughout the kernel except probabilistic path termination and AO */ - ccl_global float *per_sample_output_buffers, - ccl_global uint *rng_coop, /* Required for "kernel_write_data_passes" and AO */ - ccl_global float3 *throughput_coop, /* Required for handling holdout material and AO */ - ccl_global float *L_transparent_coop, /* Required for handling holdout material */ - PathRadiance *PathRadiance_coop, /* Required for "kernel_write_data_passes" and indirect primitive emission */ - ccl_global PathState *PathState_coop, /* Required throughout the kernel and AO */ - Intersection *Intersection_coop, /* Required for indirect primitive emission */ - ccl_global float3 *AOAlpha_coop, /* Required for AO */ - ccl_global float3 *AOBSDF_coop, /* Required for AO */ - ccl_global Ray *AOLightRay_coop, /* Required for AO */ - int sw, int sh, int sx, int sy, int stride, - ccl_global char *ray_state, /* Denotes the state of each ray */ - ccl_global unsigned int *work_array, /* Denotes the work that each ray belongs to */ - ccl_global int *Queue_data, /* Queue memory */ - ccl_global int *Queue_index, /* Tracks the number of elements in each queue */ - int queuesize, /* Size (capacity) of each queue */ -#ifdef __WORK_STEALING__ - unsigned int start_sample, -#endif - int parallel_samples) /* Number of samples to be processed in parallel */ + ccl_constant KernelData *data) { - ccl_local unsigned int local_queue_atomics_bg; - ccl_local unsigned int local_queue_atomics_ao; - if(get_local_id(0) == 0 && get_local_id(1) == 0) { - local_queue_atomics_bg = 0; - local_queue_atomics_ao = 0; - } - barrier(CLK_LOCAL_MEM_FENCE); - - char enqueue_flag = 0; - char enqueue_flag_AO_SHADOW_RAY_CAST = 0; - int ray_index = get_global_id(1) * get_global_size(0) + get_global_id(0); - ray_index = get_ray_index(ray_index, - QUEUE_ACTIVE_AND_REGENERATED_RAYS, - Queue_data, - queuesize, - 0); - -#ifdef __COMPUTE_DEVICE_GPU__ - /* If we are executing on a GPU device, we exit all threads that are not - * required. - * - * If we are executing on a CPU device, then we need to keep all threads - * active since we have barrier() calls later in the kernel. CPU devices, - * expect all threads to execute barrier statement. - */ - if(ray_index == QUEUE_EMPTY_SLOT) { - return; - } -#endif /* __COMPUTE_DEVICE_GPU__ */ - -#ifndef __COMPUTE_DEVICE_GPU__ - if(ray_index != QUEUE_EMPTY_SLOT) { -#endif - kernel_holdout_emission_blurring_pathtermination_ao( - (KernelGlobals *)kg, - (ShaderData *)sd, - per_sample_output_buffers, - rng_coop, - throughput_coop, - L_transparent_coop, - PathRadiance_coop, - PathState_coop, - Intersection_coop, - AOAlpha_coop, - AOBSDF_coop, - AOLightRay_coop, - sw, sh, sx, sy, stride, - ray_state, - work_array, -#ifdef __WORK_STEALING__ - start_sample, -#endif - parallel_samples, - ray_index, - &enqueue_flag, - &enqueue_flag_AO_SHADOW_RAY_CAST); -#ifndef __COMPUTE_DEVICE_GPU__ - } -#endif - - /* Enqueue RAY_UPDATE_BUFFER rays. */ - enqueue_ray_index_local(ray_index, - QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS, - enqueue_flag, - queuesize, - &local_queue_atomics_bg, - Queue_data, - Queue_index); - -#ifdef __AO__ - /* Enqueue to-shadow-ray-cast rays. */ - enqueue_ray_index_local(ray_index, - QUEUE_SHADOW_RAY_CAST_AO_RAYS, - enqueue_flag_AO_SHADOW_RAY_CAST, - queuesize, - &local_queue_atomics_ao, - Queue_data, - Queue_index); -#endif + ccl_local BackgroundAOLocals locals; + kernel_holdout_emission_blurring_pathtermination_ao( + (KernelGlobals*)kg, + &locals); } diff --git a/intern/cycles/kernel/kernels/opencl/kernel_indirect_background.cl b/intern/cycles/kernel/kernels/opencl/kernel_indirect_background.cl new file mode 100644 index 00000000000..8c213ff5cb2 --- /dev/null +++ b/intern/cycles/kernel/kernels/opencl/kernel_indirect_background.cl @@ -0,0 +1,26 @@ +/* + * Copyright 2011-2017 Blender Foundation + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#include "kernel/kernel_compat_opencl.h" +#include "kernel/split/kernel_split_common.h" +#include "kernel/split/kernel_indirect_background.h" + +__kernel void kernel_ocl_path_trace_indirect_background( + ccl_global char *kg, + ccl_constant KernelData *data) +{ + kernel_indirect_background((KernelGlobals*)kg); +} diff --git a/intern/cycles/kernel/kernels/opencl/kernel_indirect_subsurface.cl b/intern/cycles/kernel/kernels/opencl/kernel_indirect_subsurface.cl new file mode 100644 index 00000000000..998ebc4c0c3 --- /dev/null +++ b/intern/cycles/kernel/kernels/opencl/kernel_indirect_subsurface.cl @@ -0,0 +1,26 @@ +/* + * Copyright 2011-2017 Blender Foundation + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#include "kernel/kernel_compat_opencl.h" +#include "kernel/split/kernel_split_common.h" +#include "kernel/split/kernel_indirect_subsurface.h" + +__kernel void kernel_ocl_path_trace_indirect_subsurface( + ccl_global char *kg, + ccl_constant KernelData *data) +{ + kernel_indirect_subsurface((KernelGlobals*)kg); +} diff --git a/intern/cycles/kernel/kernels/opencl/kernel_lamp_emission.cl b/intern/cycles/kernel/kernels/opencl/kernel_lamp_emission.cl index 267bddc2ffc..822d2287715 100644 --- a/intern/cycles/kernel/kernels/opencl/kernel_lamp_emission.cl +++ b/intern/cycles/kernel/kernels/opencl/kernel_lamp_emission.cl @@ -14,67 +14,13 @@ * limitations under the License. */ -#include "split/kernel_lamp_emission.h" +#include "kernel/kernel_compat_opencl.h" +#include "kernel/split/kernel_split_common.h" +#include "kernel/split/kernel_lamp_emission.h" __kernel void kernel_ocl_path_trace_lamp_emission( ccl_global char *kg, - ccl_constant KernelData *data, - ccl_global float3 *throughput_coop, /* Required for lamp emission */ - PathRadiance *PathRadiance_coop, /* Required for lamp emission */ - ccl_global Ray *Ray_coop, /* Required for lamp emission */ - ccl_global PathState *PathState_coop, /* Required for lamp emission */ - Intersection *Intersection_coop, /* Required for lamp emission */ - ccl_global char *ray_state, /* Denotes the state of each ray */ - int sw, int sh, - ccl_global int *Queue_data, /* Memory for queues */ - ccl_global int *Queue_index, /* Tracks the number of elements in queues */ - int queuesize, /* Size (capacity) of queues */ - ccl_global char *use_queues_flag, /* Used to decide if this kernel should use - * queues to fetch ray index - */ - int parallel_samples) /* Number of samples to be processed in parallel */ + ccl_constant KernelData *data) { - int x = get_global_id(0); - int y = get_global_id(1); - - /* We will empty this queue in this kernel. */ - if(get_global_id(0) == 0 && get_global_id(1) == 0) { - Queue_index[QUEUE_ACTIVE_AND_REGENERATED_RAYS] = 0; - } - /* Fetch use_queues_flag. */ - ccl_local char local_use_queues_flag; - if(get_local_id(0) == 0 && get_local_id(1) == 0) { - local_use_queues_flag = use_queues_flag[0]; - } - barrier(CLK_LOCAL_MEM_FENCE); - - int ray_index; - if(local_use_queues_flag) { - int thread_index = get_global_id(1) * get_global_size(0) + get_global_id(0); - ray_index = get_ray_index(thread_index, - QUEUE_ACTIVE_AND_REGENERATED_RAYS, - Queue_data, - queuesize, - 1); - if(ray_index == QUEUE_EMPTY_SLOT) { - return; - } - } else { - if(x < (sw * parallel_samples) && y < sh) { - ray_index = x + y * (sw * parallel_samples); - } else { - return; - } - } - - kernel_lamp_emission((KernelGlobals *)kg, - throughput_coop, - PathRadiance_coop, - Ray_coop, - PathState_coop, - Intersection_coop, - ray_state, - sw, sh, - use_queues_flag, - ray_index); + kernel_lamp_emission((KernelGlobals*)kg); } diff --git a/intern/cycles/kernel/kernels/opencl/kernel_next_iteration_setup.cl b/intern/cycles/kernel/kernels/opencl/kernel_next_iteration_setup.cl index 6d49b6294a8..6d207253a40 100644 --- a/intern/cycles/kernel/kernels/opencl/kernel_next_iteration_setup.cl +++ b/intern/cycles/kernel/kernels/opencl/kernel_next_iteration_setup.cl @@ -14,101 +14,14 @@ * limitations under the License. */ -#include "split/kernel_next_iteration_setup.h" +#include "kernel/kernel_compat_opencl.h" +#include "kernel/split/kernel_split_common.h" +#include "kernel/split/kernel_next_iteration_setup.h" __kernel void kernel_ocl_path_trace_next_iteration_setup( ccl_global char *kg, - ccl_constant KernelData *data, - ccl_global char *sd, /* Required for setting up ray for next iteration */ - ccl_global uint *rng_coop, /* Required for setting up ray for next iteration */ - ccl_global float3 *throughput_coop, /* Required for setting up ray for next iteration */ - PathRadiance *PathRadiance_coop, /* Required for setting up ray for next iteration */ - ccl_global Ray *Ray_coop, /* Required for setting up ray for next iteration */ - ccl_global PathState *PathState_coop, /* Required for setting up ray for next iteration */ - ccl_global Ray *LightRay_dl_coop, /* Required for radiance update - direct lighting */ - ccl_global int *ISLamp_coop, /* Required for radiance update - direct lighting */ - ccl_global BsdfEval *BSDFEval_coop, /* Required for radiance update - direct lighting */ - ccl_global Ray *LightRay_ao_coop, /* Required for radiance update - AO */ - ccl_global float3 *AOBSDF_coop, /* Required for radiance update - AO */ - ccl_global float3 *AOAlpha_coop, /* Required for radiance update - AO */ - ccl_global char *ray_state, /* Denotes the state of each ray */ - ccl_global int *Queue_data, /* Queue memory */ - ccl_global int *Queue_index, /* Tracks the number of elements in each queue */ - int queuesize, /* Size (capacity) of each queue */ - ccl_global char *use_queues_flag) /* flag to decide if scene_intersect kernel should - * use queues to fetch ray index */ + ccl_constant KernelData *data) { ccl_local unsigned int local_queue_atomics; - if(get_local_id(0) == 0 && get_local_id(1) == 0) { - local_queue_atomics = 0; - } - barrier(CLK_LOCAL_MEM_FENCE); - - if(get_global_id(0) == 0 && get_global_id(1) == 0) { - /* If we are here, then it means that scene-intersect kernel - * has already been executed atleast once. From the next time, - * scene-intersect kernel may operate on queues to fetch ray index - */ - use_queues_flag[0] = 1; - - /* Mark queue indices of QUEUE_SHADOW_RAY_CAST_AO_RAYS and - * QUEUE_SHADOW_RAY_CAST_DL_RAYS queues that were made empty during the - * previous kernel. - */ - Queue_index[QUEUE_SHADOW_RAY_CAST_AO_RAYS] = 0; - Queue_index[QUEUE_SHADOW_RAY_CAST_DL_RAYS] = 0; - } - - char enqueue_flag = 0; - int ray_index = get_global_id(1) * get_global_size(0) + get_global_id(0); - ray_index = get_ray_index(ray_index, - QUEUE_ACTIVE_AND_REGENERATED_RAYS, - Queue_data, - queuesize, - 0); - -#ifdef __COMPUTE_DEVICE_GPU__ - /* If we are executing on a GPU device, we exit all threads that are not - * required. - * - * If we are executing on a CPU device, then we need to keep all threads - * active since we have barrier() calls later in the kernel. CPU devices, - * expect all threads to execute barrier statement. - */ - if(ray_index == QUEUE_EMPTY_SLOT) { - return; - } -#endif - -#ifndef __COMPUTE_DEVICE_GPU__ - if(ray_index != QUEUE_EMPTY_SLOT) { -#endif - enqueue_flag = kernel_next_iteration_setup((KernelGlobals *)kg, - (ShaderData *)sd, - rng_coop, - throughput_coop, - PathRadiance_coop, - Ray_coop, - PathState_coop, - LightRay_dl_coop, - ISLamp_coop, - BSDFEval_coop, - LightRay_ao_coop, - AOBSDF_coop, - AOAlpha_coop, - ray_state, - use_queues_flag, - ray_index); -#ifndef __COMPUTE_DEVICE_GPU__ - } -#endif - - /* Enqueue RAY_UPDATE_BUFFER rays. */ - enqueue_ray_index_local(ray_index, - QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS, - enqueue_flag, - queuesize, - &local_queue_atomics, - Queue_data, - Queue_index); + kernel_next_iteration_setup((KernelGlobals*)kg, &local_queue_atomics); } diff --git a/intern/cycles/kernel/kernels/opencl/kernel_path_init.cl b/intern/cycles/kernel/kernels/opencl/kernel_path_init.cl new file mode 100644 index 00000000000..bd9aa9538c8 --- /dev/null +++ b/intern/cycles/kernel/kernels/opencl/kernel_path_init.cl @@ -0,0 +1,26 @@ +/* + * Copyright 2011-2017 Blender Foundation + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#include "kernel/kernel_compat_opencl.h" +#include "kernel/split/kernel_split_common.h" +#include "kernel/split/kernel_path_init.h" + +__kernel void kernel_ocl_path_trace_path_init( + ccl_global char *kg, + ccl_constant KernelData *data) +{ + kernel_path_init((KernelGlobals*)kg); +} diff --git a/intern/cycles/kernel/kernels/opencl/kernel_queue_enqueue.cl b/intern/cycles/kernel/kernels/opencl/kernel_queue_enqueue.cl index 3156dc255fb..9be154e3d75 100644 --- a/intern/cycles/kernel/kernels/opencl/kernel_queue_enqueue.cl +++ b/intern/cycles/kernel/kernels/opencl/kernel_queue_enqueue.cl @@ -14,93 +14,14 @@ * limitations under the License. */ -#include "../../kernel_compat_opencl.h" -#include "../../kernel_math.h" -#include "../../kernel_types.h" -#include "../../kernel_globals.h" -#include "../../kernel_queues.h" +#include "kernel/kernel_compat_opencl.h" +#include "kernel/split/kernel_split_common.h" +#include "kernel/split/kernel_queue_enqueue.h" -/* - * The kernel "kernel_queue_enqueue" enqueues rays of - * different ray state into their appropriate Queues; - * 1. Rays that have been determined to hit the background from the - * "kernel_scene_intersect" kernel - * are enqueued in QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS; - * 2. Rays that have been determined to be actively participating in path-iteration will be enqueued into QUEUE_ACTIVE_AND_REGENERATED_RAYS. - * - * The input and output of the kernel is as follows, - * - * ray_state -------------------------------------------|--- kernel_queue_enqueue --|--- Queue_data (QUEUE_ACTIVE_AND_REGENERATED_RAYS & QUEUE_HITBF_BUFF_UPDATE_TOREGEN_RAYS) - * Queue_index(QUEUE_ACTIVE_AND_REGENERATED_RAYS) ------| |--- Queue_index (QUEUE_ACTIVE_AND_REGENERATED_RAYS & QUEUE_HITBF_BUFF_UPDATE_TOREGEN_RAYS) - * Queue_index(QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS) ---| | - * queuesize -------------------------------------------| | - * - * Note on Queues : - * State of queues during the first time this kernel is called : - * At entry, - * Both QUEUE_ACTIVE_AND_REGENERATED_RAYS and QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS will be empty. - * At exit, - * QUEUE_ACTIVE_AND_REGENERATED_RAYS will be filled with RAY_ACTIVE rays - * QUEUE_HITBF_BUFF_UPDATE_TOREGEN_RAYS will be filled with RAY_HIT_BACKGROUND rays. - * - * State of queue during other times this kernel is called : - * At entry, - * QUEUE_ACTIVE_AND_REGENERATED_RAYS will be empty. - * QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS will contain RAY_TO_REGENERATE and RAY_UPDATE_BUFFER rays. - * At exit, - * QUEUE_ACTIVE_AND_REGENERATED_RAYS will be filled with RAY_ACTIVE rays. - * QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS will be filled with RAY_TO_REGENERATE, RAY_UPDATE_BUFFER, RAY_HIT_BACKGROUND rays. - */ __kernel void kernel_ocl_path_trace_queue_enqueue( - ccl_global int *Queue_data, /* Queue memory */ - ccl_global int *Queue_index, /* Tracks the number of elements in each queue */ - ccl_global char *ray_state, /* Denotes the state of each ray */ - int queuesize) /* Size (capacity) of each queue */ + ccl_global char *kg, + ccl_constant KernelData *data) { - /* We have only 2 cases (Hit/Not-Hit) */ - ccl_local unsigned int local_queue_atomics[2]; - - int lidx = get_local_id(1) * get_local_size(0) + get_local_id(0); - int ray_index = get_global_id(1) * get_global_size(0) + get_global_id(0); - - if(lidx < 2 ) { - local_queue_atomics[lidx] = 0; - } - barrier(CLK_LOCAL_MEM_FENCE); - - int queue_number = -1; - - if(IS_STATE(ray_state, ray_index, RAY_HIT_BACKGROUND)) { - queue_number = QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS; - } - else if(IS_STATE(ray_state, ray_index, RAY_ACTIVE)) { - queue_number = QUEUE_ACTIVE_AND_REGENERATED_RAYS; - } - - unsigned int my_lqidx; - if(queue_number != -1) { - my_lqidx = get_local_queue_index(queue_number, local_queue_atomics); - } - barrier(CLK_LOCAL_MEM_FENCE); - - if(lidx == 0) { - local_queue_atomics[QUEUE_ACTIVE_AND_REGENERATED_RAYS] = - get_global_per_queue_offset(QUEUE_ACTIVE_AND_REGENERATED_RAYS, - local_queue_atomics, - Queue_index); - local_queue_atomics[QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS] = - get_global_per_queue_offset(QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS, - local_queue_atomics, - Queue_index); - } - barrier(CLK_LOCAL_MEM_FENCE); - - unsigned int my_gqidx; - if(queue_number != -1) { - my_gqidx = get_global_queue_index(queue_number, - queuesize, - my_lqidx, - local_queue_atomics); - Queue_data[my_gqidx] = ray_index; - } + ccl_local QueueEnqueueLocals locals; + kernel_queue_enqueue((KernelGlobals*)kg, &locals); } diff --git a/intern/cycles/kernel/kernels/opencl/kernel_scene_intersect.cl b/intern/cycles/kernel/kernels/opencl/kernel_scene_intersect.cl index 7f3f433c7a6..eb4fb4d153a 100644 --- a/intern/cycles/kernel/kernels/opencl/kernel_scene_intersect.cl +++ b/intern/cycles/kernel/kernels/opencl/kernel_scene_intersect.cl @@ -14,67 +14,13 @@ * limitations under the License. */ -#include "split/kernel_scene_intersect.h" +#include "kernel/kernel_compat_opencl.h" +#include "kernel/split/kernel_split_common.h" +#include "kernel/split/kernel_scene_intersect.h" __kernel void kernel_ocl_path_trace_scene_intersect( ccl_global char *kg, - ccl_constant KernelData *data, - ccl_global uint *rng_coop, - ccl_global Ray *Ray_coop, /* Required for scene_intersect */ - ccl_global PathState *PathState_coop, /* Required for scene_intersect */ - Intersection *Intersection_coop, /* Required for scene_intersect */ - ccl_global char *ray_state, /* Denotes the state of each ray */ - int sw, int sh, - ccl_global int *Queue_data, /* Memory for queues */ - ccl_global int *Queue_index, /* Tracks the number of elements in queues */ - int queuesize, /* Size (capacity) of queues */ - ccl_global char *use_queues_flag, /* used to decide if this kernel should use - * queues to fetch ray index */ -#ifdef __KERNEL_DEBUG__ - DebugData *debugdata_coop, -#endif - int parallel_samples) /* Number of samples to be processed in parallel */ + ccl_constant KernelData *data) { - int x = get_global_id(0); - int y = get_global_id(1); - - /* Fetch use_queues_flag */ - ccl_local char local_use_queues_flag; - if(get_local_id(0) == 0 && get_local_id(1) == 0) { - local_use_queues_flag = use_queues_flag[0]; - } - barrier(CLK_LOCAL_MEM_FENCE); - - int ray_index; - if(local_use_queues_flag) { - int thread_index = get_global_id(1) * get_global_size(0) + get_global_id(0); - ray_index = get_ray_index(thread_index, - QUEUE_ACTIVE_AND_REGENERATED_RAYS, - Queue_data, - queuesize, - 0); - - if(ray_index == QUEUE_EMPTY_SLOT) { - return; - } - } else { - if(x < (sw * parallel_samples) && y < sh) { - ray_index = x + y * (sw * parallel_samples); - } else { - return; - } - } - - kernel_scene_intersect((KernelGlobals *)kg, - rng_coop, - Ray_coop, - PathState_coop, - Intersection_coop, - ray_state, - sw, sh, - use_queues_flag, -#ifdef __KERNEL_DEBUG__ - debugdata_coop, -#endif - ray_index); + kernel_scene_intersect((KernelGlobals*)kg); } diff --git a/intern/cycles/kernel/kernels/opencl/kernel_shader_eval.cl b/intern/cycles/kernel/kernels/opencl/kernel_shader_eval.cl index c37856c8f30..6baee460986 100644 --- a/intern/cycles/kernel/kernels/opencl/kernel_shader_eval.cl +++ b/intern/cycles/kernel/kernels/opencl/kernel_shader_eval.cl @@ -14,55 +14,14 @@ * limitations under the License. */ -#include "split/kernel_shader_eval.h" +#include "kernel/kernel_compat_opencl.h" +#include "kernel/split/kernel_split_common.h" +#include "kernel/split/kernel_shader_eval.h" __kernel void kernel_ocl_path_trace_shader_eval( ccl_global char *kg, - ccl_constant KernelData *data, - ccl_global char *sd, /* Output ShaderData structure to be filled */ - ccl_global uint *rng_coop, /* Required for rbsdf calculation */ - ccl_global Ray *Ray_coop, /* Required for setting up shader from ray */ - ccl_global PathState *PathState_coop, /* Required for all functions in this kernel */ - Intersection *Intersection_coop, /* Required for setting up shader from ray */ - ccl_global char *ray_state, /* Denotes the state of each ray */ - ccl_global int *Queue_data, /* queue memory */ - ccl_global int *Queue_index, /* Tracks the number of elements in each queue */ - int queuesize) /* Size (capacity) of each queue */ + ccl_constant KernelData *data) { - /* Enqeueue RAY_TO_REGENERATE rays into QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS queue. */ ccl_local unsigned int local_queue_atomics; - if(get_local_id(0) == 0 && get_local_id(1) == 0) { - local_queue_atomics = 0; - } - barrier(CLK_LOCAL_MEM_FENCE); - - int ray_index = get_global_id(1) * get_global_size(0) + get_global_id(0); - ray_index = get_ray_index(ray_index, - QUEUE_ACTIVE_AND_REGENERATED_RAYS, - Queue_data, - queuesize, - 0); - - if(ray_index == QUEUE_EMPTY_SLOT) { - return; - } - - char enqueue_flag = (IS_STATE(ray_state, ray_index, RAY_TO_REGENERATE)) ? 1 : 0; - enqueue_ray_index_local(ray_index, - QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS, - enqueue_flag, - queuesize, - &local_queue_atomics, - Queue_data, - Queue_index); - - /* Continue on with shader evaluation. */ - kernel_shader_eval((KernelGlobals *)kg, - (ShaderData *)sd, - rng_coop, - Ray_coop, - PathState_coop, - Intersection_coop, - ray_state, - ray_index); + kernel_shader_eval((KernelGlobals*)kg, &local_queue_atomics); } diff --git a/intern/cycles/kernel/kernels/opencl/kernel_shadow_blocked.cl b/intern/cycles/kernel/kernels/opencl/kernel_shadow_blocked.cl deleted file mode 100644 index edf76fba714..00000000000 --- a/intern/cycles/kernel/kernels/opencl/kernel_shadow_blocked.cl +++ /dev/null @@ -1,65 +0,0 @@ -/* - * Copyright 2011-2015 Blender Foundation - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -#include "split/kernel_shadow_blocked.h" - -__kernel void kernel_ocl_path_trace_shadow_blocked( - ccl_global char *kg, - ccl_constant KernelData *data, - ccl_global PathState *PathState_coop, /* Required for shadow blocked */ - ccl_global Ray *LightRay_dl_coop, /* Required for direct lighting's shadow blocked */ - ccl_global Ray *LightRay_ao_coop, /* Required for AO's shadow blocked */ - ccl_global char *ray_state, - ccl_global int *Queue_data, /* Queue memory */ - ccl_global int *Queue_index, /* Tracks the number of elements in each queue */ - int queuesize) /* Size (capacity) of each queue */ -{ - int lidx = get_local_id(1) * get_local_id(0) + get_local_id(0); - - ccl_local unsigned int ao_queue_length; - ccl_local unsigned int dl_queue_length; - if(lidx == 0) { - ao_queue_length = Queue_index[QUEUE_SHADOW_RAY_CAST_AO_RAYS]; - dl_queue_length = Queue_index[QUEUE_SHADOW_RAY_CAST_DL_RAYS]; - } - barrier(CLK_LOCAL_MEM_FENCE); - - /* flag determining if the current ray is to process shadow ray for AO or DL */ - char shadow_blocked_type = -1; - - int ray_index = QUEUE_EMPTY_SLOT; - int thread_index = get_global_id(1) * get_global_size(0) + get_global_id(0); - if(thread_index < ao_queue_length + dl_queue_length) { - if(thread_index < ao_queue_length) { - ray_index = get_ray_index(thread_index, QUEUE_SHADOW_RAY_CAST_AO_RAYS, Queue_data, queuesize, 1); - shadow_blocked_type = RAY_SHADOW_RAY_CAST_AO; - } else { - ray_index = get_ray_index(thread_index - ao_queue_length, QUEUE_SHADOW_RAY_CAST_DL_RAYS, Queue_data, queuesize, 1); - shadow_blocked_type = RAY_SHADOW_RAY_CAST_DL; - } - } - - if(ray_index == QUEUE_EMPTY_SLOT) - return; - - kernel_shadow_blocked((KernelGlobals *)kg, - PathState_coop, - LightRay_dl_coop, - LightRay_ao_coop, - ray_state, - shadow_blocked_type, - ray_index); -} diff --git a/intern/cycles/kernel/kernels/opencl/kernel_shadow_blocked_ao.cl b/intern/cycles/kernel/kernels/opencl/kernel_shadow_blocked_ao.cl new file mode 100644 index 00000000000..6a8ef81b32a --- /dev/null +++ b/intern/cycles/kernel/kernels/opencl/kernel_shadow_blocked_ao.cl @@ -0,0 +1,26 @@ +/* + * Copyright 2011-2015 Blender Foundation + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#include "kernel/kernel_compat_opencl.h" +#include "kernel/split/kernel_split_common.h" +#include "kernel/split/kernel_shadow_blocked_ao.h" + +__kernel void kernel_ocl_path_trace_shadow_blocked_ao( + ccl_global char *kg, + ccl_constant KernelData *data) +{ + kernel_shadow_blocked_ao((KernelGlobals*)kg); +} diff --git a/intern/cycles/kernel/kernels/opencl/kernel_shadow_blocked_dl.cl b/intern/cycles/kernel/kernels/opencl/kernel_shadow_blocked_dl.cl new file mode 100644 index 00000000000..b255cc5ef8b --- /dev/null +++ b/intern/cycles/kernel/kernels/opencl/kernel_shadow_blocked_dl.cl @@ -0,0 +1,26 @@ +/* + * Copyright 2011-2015 Blender Foundation + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#include "kernel/kernel_compat_opencl.h" +#include "kernel/split/kernel_split_common.h" +#include "kernel/split/kernel_shadow_blocked_dl.h" + +__kernel void kernel_ocl_path_trace_shadow_blocked_dl( + ccl_global char *kg, + ccl_constant KernelData *data) +{ + kernel_shadow_blocked_dl((KernelGlobals*)kg); +} diff --git a/intern/cycles/kernel/kernels/opencl/kernel_split.cl b/intern/cycles/kernel/kernels/opencl/kernel_split.cl new file mode 100644 index 00000000000..732cda30115 --- /dev/null +++ b/intern/cycles/kernel/kernels/opencl/kernel_split.cl @@ -0,0 +1,35 @@ +/* + * Copyright 2011-2017 Blender Foundation + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#include "kernel/kernels/opencl/kernel_state_buffer_size.cl" +#include "kernel/kernels/opencl/kernel_data_init.cl" +#include "kernel/kernels/opencl/kernel_path_init.cl" + +#include "kernel/kernels/opencl/kernel_scene_intersect.cl" +#include "kernel/kernels/opencl/kernel_lamp_emission.cl" +#include "kernel/kernels/opencl/kernel_do_volume.cl" +#include "kernel/kernels/opencl/kernel_indirect_background.cl" +#include "kernel/kernels/opencl/kernel_queue_enqueue.cl" +#include "kernel/kernels/opencl/kernel_shader_eval.cl" +#include "kernel/kernels/opencl/kernel_holdout_emission_blurring_pathtermination_ao.cl" +#include "kernel/kernels/opencl/kernel_subsurface_scatter.cl" +#include "kernel/kernels/opencl/kernel_direct_lighting.cl" +#include "kernel/kernels/opencl/kernel_shadow_blocked_ao.cl" +#include "kernel/kernels/opencl/kernel_shadow_blocked_dl.cl" +#include "kernel/kernels/opencl/kernel_next_iteration_setup.cl" +#include "kernel/kernels/opencl/kernel_indirect_subsurface.cl" +#include "kernel/kernels/opencl/kernel_buffer_update.cl" + diff --git a/intern/cycles/kernel/kernels/opencl/kernel_state_buffer_size.cl b/intern/cycles/kernel/kernels/opencl/kernel_state_buffer_size.cl new file mode 100644 index 00000000000..c10ecc426c6 --- /dev/null +++ b/intern/cycles/kernel/kernels/opencl/kernel_state_buffer_size.cl @@ -0,0 +1,29 @@ +/* + * Copyright 2011-2017 Blender Foundation + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#include "kernel/kernel_compat_opencl.h" +#include "kernel/split/kernel_split_common.h" + +__kernel void kernel_ocl_path_trace_state_buffer_size( + ccl_global char *kg, + ccl_constant KernelData *data, + uint num_threads, + ccl_global uint64_t *size) +{ + ((KernelGlobals*)kg)->data = data; + *size = split_data_buffer_size((KernelGlobals*)kg, num_threads); +} + diff --git a/intern/cycles/kernel/kernels/opencl/kernel_subsurface_scatter.cl b/intern/cycles/kernel/kernels/opencl/kernel_subsurface_scatter.cl new file mode 100644 index 00000000000..7a1838e485f --- /dev/null +++ b/intern/cycles/kernel/kernels/opencl/kernel_subsurface_scatter.cl @@ -0,0 +1,27 @@ +/* + * Copyright 2011-2017 Blender Foundation + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#include "kernel/kernel_compat_opencl.h" +#include "kernel/split/kernel_split_common.h" +#include "kernel/split/kernel_subsurface_scatter.h" + +__kernel void kernel_ocl_path_trace_subsurface_scatter( + ccl_global char *kg, + ccl_constant KernelData *data) +{ + ccl_local unsigned int local_queue_atomics; + kernel_subsurface_scatter((KernelGlobals*)kg, &local_queue_atomics); +} diff --git a/intern/cycles/kernel/kernels/opencl/kernel_sum_all_radiance.cl b/intern/cycles/kernel/kernels/opencl/kernel_sum_all_radiance.cl deleted file mode 100644 index 88a1ed830af..00000000000 --- a/intern/cycles/kernel/kernels/opencl/kernel_sum_all_radiance.cl +++ /dev/null @@ -1,38 +0,0 @@ -/* - * Copyright 2011-2015 Blender Foundation - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -#include "split/kernel_sum_all_radiance.h" - -__kernel void kernel_ocl_path_trace_sum_all_radiance( - ccl_constant KernelData *data, /* To get pass_stride to offet into buffer */ - ccl_global float *buffer, /* Output buffer of RenderTile */ - ccl_global float *per_sample_output_buffer, /* Radiance contributed by all samples */ - int parallel_samples, int sw, int sh, int stride, - int buffer_offset_x, - int buffer_offset_y, - int buffer_stride, - int start_sample) -{ - kernel_sum_all_radiance(data, - buffer, - per_sample_output_buffer, - parallel_samples, - sw, sh, stride, - buffer_offset_x, - buffer_offset_y, - buffer_stride, - start_sample); -} diff --git a/intern/cycles/kernel/osl/CMakeLists.txt b/intern/cycles/kernel/osl/CMakeLists.txt index 98de40e5a8a..d2eb89e0e0a 100644 --- a/intern/cycles/kernel/osl/CMakeLists.txt +++ b/intern/cycles/kernel/osl/CMakeLists.txt @@ -1,12 +1,6 @@ set(INC - . - .. - ../svm - ../../graph - ../../render - ../../util - ../../device + ../.. ) set(INC_SYS diff --git a/intern/cycles/kernel/osl/background.cpp b/intern/cycles/kernel/osl/background.cpp index d835f9be45c..2e73e7a601e 100644 --- a/intern/cycles/kernel/osl/background.cpp +++ b/intern/cycles/kernel/osl/background.cpp @@ -34,10 +34,10 @@ #include <OSL/genclosure.h> -#include "osl_closures.h" +#include "kernel/osl/osl_closures.h" -#include "kernel_compat_cpu.h" -#include "closure/alloc.h" +#include "kernel/kernel_compat_cpu.h" +#include "kernel/closure/alloc.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/kernel/osl/bsdf_diffuse_ramp.cpp b/intern/cycles/kernel/osl/bsdf_diffuse_ramp.cpp index bc26f42b559..ea18f2c8c86 100644 --- a/intern/cycles/kernel/osl/bsdf_diffuse_ramp.cpp +++ b/intern/cycles/kernel/osl/bsdf_diffuse_ramp.cpp @@ -34,13 +34,13 @@ #include <OSL/genclosure.h> -#include "kernel_compat_cpu.h" -#include "osl_closures.h" +#include "kernel/kernel_compat_cpu.h" +#include "kernel/osl/osl_closures.h" -#include "kernel_types.h" -#include "kernel_montecarlo.h" -#include "closure/alloc.h" -#include "closure/bsdf_diffuse_ramp.h" +#include "kernel/kernel_types.h" +#include "kernel/kernel_montecarlo.h" +#include "kernel/closure/alloc.h" +#include "kernel/closure/bsdf_diffuse_ramp.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/kernel/osl/bsdf_phong_ramp.cpp b/intern/cycles/kernel/osl/bsdf_phong_ramp.cpp index 14c7644936e..a26671eb09e 100644 --- a/intern/cycles/kernel/osl/bsdf_phong_ramp.cpp +++ b/intern/cycles/kernel/osl/bsdf_phong_ramp.cpp @@ -34,12 +34,12 @@ #include <OSL/genclosure.h> -#include "kernel_compat_cpu.h" -#include "osl_closures.h" +#include "kernel/kernel_compat_cpu.h" +#include "kernel/osl/osl_closures.h" -#include "kernel_types.h" -#include "closure/alloc.h" -#include "closure/bsdf_phong_ramp.h" +#include "kernel/kernel_types.h" +#include "kernel/closure/alloc.h" +#include "kernel/closure/bsdf_phong_ramp.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/kernel/osl/emissive.cpp b/intern/cycles/kernel/osl/emissive.cpp index 3f13e08b302..8843a196dad 100644 --- a/intern/cycles/kernel/osl/emissive.cpp +++ b/intern/cycles/kernel/osl/emissive.cpp @@ -34,12 +34,12 @@ #include <OSL/genclosure.h> -#include "osl_closures.h" +#include "kernel/osl/osl_closures.h" -#include "kernel_compat_cpu.h" -#include "kernel_types.h" -#include "closure/alloc.h" -#include "closure/emissive.h" +#include "kernel/kernel_compat_cpu.h" +#include "kernel/kernel_types.h" +#include "kernel/closure/alloc.h" +#include "kernel/closure/emissive.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/kernel/osl/osl_bssrdf.cpp b/intern/cycles/kernel/osl/osl_bssrdf.cpp index 44daefee249..188c3960a5f 100644 --- a/intern/cycles/kernel/osl/osl_bssrdf.cpp +++ b/intern/cycles/kernel/osl/osl_bssrdf.cpp @@ -32,17 +32,17 @@ #include <OSL/genclosure.h> -#include "kernel_compat_cpu.h" -#include "osl_closures.h" +#include "kernel/kernel_compat_cpu.h" +#include "kernel/osl/osl_closures.h" -#include "kernel_types.h" -#include "kernel_montecarlo.h" +#include "kernel/kernel_types.h" +#include "kernel/kernel_montecarlo.h" -#include "closure/alloc.h" -#include "closure/bsdf_util.h" -#include "closure/bsdf_diffuse.h" -#include "closure/bsdf_principled_diffuse.h" -#include "closure/bssrdf.h" +#include "kernel/closure/alloc.h" +#include "kernel/closure/bsdf_util.h" +#include "kernel/closure/bsdf_diffuse.h" +#include "kernel/closure/bsdf_principled_diffuse.h" +#include "kernel/closure/bssrdf.h" CCL_NAMESPACE_BEGIN @@ -81,7 +81,7 @@ public: bssrdf->sharpness = sharpness; bssrdf->N = params.N; bssrdf->roughness = params.roughness; - ccl_fetch(sd, flag) |= bssrdf_setup(bssrdf, (ClosureType)type); + sd->flag |= bssrdf_setup(bssrdf, (ClosureType)type); } bssrdf = bssrdf_alloc(sd, make_float3(0.0f, weight.y, 0.0f)); @@ -93,7 +93,7 @@ public: bssrdf->sharpness = sharpness; bssrdf->N = params.N; bssrdf->roughness = params.roughness; - ccl_fetch(sd, flag) |= bssrdf_setup(bssrdf, (ClosureType)type); + sd->flag |= bssrdf_setup(bssrdf, (ClosureType)type); } bssrdf = bssrdf_alloc(sd, make_float3(0.0f, 0.0f, weight.z)); @@ -105,7 +105,7 @@ public: bssrdf->sharpness = sharpness; bssrdf->N = params.N; bssrdf->roughness = params.roughness; - ccl_fetch(sd, flag) |= bssrdf_setup(bssrdf, (ClosureType)type); + sd->flag |= bssrdf_setup(bssrdf, (ClosureType)type); } } } diff --git a/intern/cycles/kernel/osl/osl_closures.cpp b/intern/cycles/kernel/osl/osl_closures.cpp index 5570a22692e..5b66793a05d 100644 --- a/intern/cycles/kernel/osl/osl_closures.cpp +++ b/intern/cycles/kernel/osl/osl_closures.cpp @@ -33,35 +33,36 @@ #include <OSL/genclosure.h> #include <OSL/oslclosure.h> -#include "osl_closures.h" -#include "osl_shader.h" - -#include "util_debug.h" -#include "util_math.h" -#include "util_param.h" - -#include "kernel_types.h" -#include "kernel_compat_cpu.h" -#include "kernel_globals.h" -#include "kernel_montecarlo.h" -#include "kernel_random.h" - -#include "closure/alloc.h" -#include "closure/bsdf_util.h" -#include "closure/bsdf_ashikhmin_velvet.h" -#include "closure/bsdf_diffuse.h" -#include "closure/bsdf_microfacet.h" -#include "closure/bsdf_microfacet_multi.h" -#include "closure/bsdf_oren_nayar.h" -#include "closure/bsdf_reflection.h" -#include "closure/bsdf_refraction.h" -#include "closure/bsdf_transparent.h" -#include "closure/bsdf_ashikhmin_shirley.h" -#include "closure/bsdf_toon.h" -#include "closure/bsdf_hair.h" -#include "closure/bsdf_principled_diffuse.h" -#include "closure/bsdf_principled_sheen.h" -#include "closure/volume.h" +#include "kernel/osl/osl_closures.h" +#include "kernel/osl/osl_shader.h" + +#include "util/util_debug.h" +#include "util/util_math.h" +#include "util/util_param.h" + +#include "kernel/kernel_types.h" +#include "kernel/kernel_compat_cpu.h" +#include "kernel/split/kernel_split_data_types.h" +#include "kernel/kernel_globals.h" +#include "kernel/kernel_montecarlo.h" +#include "kernel/kernel_random.h" + +#include "kernel/closure/alloc.h" +#include "kernel/closure/bsdf_util.h" +#include "kernel/closure/bsdf_ashikhmin_velvet.h" +#include "kernel/closure/bsdf_diffuse.h" +#include "kernel/closure/bsdf_microfacet.h" +#include "kernel/closure/bsdf_microfacet_multi.h" +#include "kernel/closure/bsdf_oren_nayar.h" +#include "kernel/closure/bsdf_reflection.h" +#include "kernel/closure/bsdf_refraction.h" +#include "kernel/closure/bsdf_transparent.h" +#include "kernel/closure/bsdf_ashikhmin_shirley.h" +#include "kernel/closure/bsdf_toon.h" +#include "kernel/closure/bsdf_hair.h" +#include "kernel/closure/bsdf_principled_diffuse.h" +#include "kernel/closure/bsdf_principled_sheen.h" +#include "kernel/closure/volume.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/kernel/osl/osl_closures.h b/intern/cycles/kernel/osl/osl_closures.h index c9740f81c8a..ff5fd9cc905 100644 --- a/intern/cycles/kernel/osl/osl_closures.h +++ b/intern/cycles/kernel/osl/osl_closures.h @@ -33,8 +33,8 @@ #ifndef __OSL_CLOSURES_H__ #define __OSL_CLOSURES_H__ -#include "util_types.h" -#include "kernel_types.h" +#include "util/util_types.h" +#include "kernel/kernel_types.h" #include <OSL/oslclosure.h> #include <OSL/oslexec.h> diff --git a/intern/cycles/kernel/osl/osl_globals.h b/intern/cycles/kernel/osl/osl_globals.h index 65cb7ecc6b4..02c083a83f8 100644 --- a/intern/cycles/kernel/osl/osl_globals.h +++ b/intern/cycles/kernel/osl/osl_globals.h @@ -21,10 +21,10 @@ #include <OSL/oslexec.h> -#include "util_map.h" -#include "util_param.h" -#include "util_thread.h" -#include "util_vector.h" +#include "util/util_map.h" +#include "util/util_param.h" +#include "util/util_thread.h" +#include "util/util_vector.h" #ifndef WIN32 using std::isfinite; diff --git a/intern/cycles/kernel/osl/osl_services.cpp b/intern/cycles/kernel/osl/osl_services.cpp index 58bbdc33920..b767c60c617 100644 --- a/intern/cycles/kernel/osl/osl_services.cpp +++ b/intern/cycles/kernel/osl/osl_services.cpp @@ -25,33 +25,34 @@ #include <string.h> -#include "mesh.h" -#include "object.h" -#include "scene.h" - -#include "osl_closures.h" -#include "osl_globals.h" -#include "osl_services.h" -#include "osl_shader.h" - -#include "util_foreach.h" -#include "util_logging.h" -#include "util_string.h" - -#include "kernel_compat_cpu.h" -#include "kernel_globals.h" -#include "kernel_random.h" -#include "kernel_projection.h" -#include "kernel_differential.h" -#include "kernel_montecarlo.h" -#include "kernel_camera.h" -#include "kernels/cpu/kernel_cpu_image.h" -#include "geom/geom.h" -#include "bvh/bvh.h" - -#include "kernel_projection.h" -#include "kernel_accumulate.h" -#include "kernel_shader.h" +#include "render/mesh.h" +#include "render/object.h" +#include "render/scene.h" + +#include "kernel/osl/osl_closures.h" +#include "kernel/osl/osl_globals.h" +#include "kernel/osl/osl_services.h" +#include "kernel/osl/osl_shader.h" + +#include "util/util_foreach.h" +#include "util/util_logging.h" +#include "util/util_string.h" + +#include "kernel/kernel_compat_cpu.h" +#include "kernel/split/kernel_split_data_types.h" +#include "kernel/kernel_globals.h" +#include "kernel/kernel_random.h" +#include "kernel/kernel_projection.h" +#include "kernel/kernel_differential.h" +#include "kernel/kernel_montecarlo.h" +#include "kernel/kernel_camera.h" +#include "kernel/kernels/cpu/kernel_cpu_image.h" +#include "kernel/geom/geom.h" +#include "kernel/bvh/bvh.h" + +#include "kernel/kernel_projection.h" +#include "kernel/kernel_accumulate.h" +#include "kernel/kernel_shader.h" #ifdef WITH_PTEX # include <Ptexture.h> diff --git a/intern/cycles/kernel/osl/osl_shader.cpp b/intern/cycles/kernel/osl/osl_shader.cpp index 0d762bbdb38..13b19d86eca 100644 --- a/intern/cycles/kernel/osl/osl_shader.cpp +++ b/intern/cycles/kernel/osl/osl_shader.cpp @@ -16,21 +16,22 @@ #include <OSL/oslexec.h> -#include "kernel_compat_cpu.h" -#include "kernel_montecarlo.h" -#include "kernel_types.h" -#include "kernel_globals.h" +#include "kernel/kernel_compat_cpu.h" +#include "kernel/kernel_montecarlo.h" +#include "kernel/kernel_types.h" +#include "kernel/split/kernel_split_data_types.h" +#include "kernel/kernel_globals.h" -#include "geom/geom_object.h" +#include "kernel/geom/geom_object.h" -#include "osl_closures.h" -#include "osl_globals.h" -#include "osl_services.h" -#include "osl_shader.h" +#include "kernel/osl/osl_closures.h" +#include "kernel/osl/osl_globals.h" +#include "kernel/osl/osl_services.h" +#include "kernel/osl/osl_shader.h" -#include "util_foreach.h" +#include "util/util_foreach.h" -#include "attribute.h" +#include "render/attribute.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/kernel/osl/osl_shader.h b/intern/cycles/kernel/osl/osl_shader.h index ad06dd6929d..32121e940b4 100644 --- a/intern/cycles/kernel/osl/osl_shader.h +++ b/intern/cycles/kernel/osl/osl_shader.h @@ -29,7 +29,7 @@ * This means no thread state must be passed along in the kernel itself. */ -#include "kernel_types.h" +#include "kernel/kernel_types.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/kernel/split/kernel_background_buffer_update.h b/intern/cycles/kernel/split/kernel_background_buffer_update.h deleted file mode 100644 index 9bfa71c75ef..00000000000 --- a/intern/cycles/kernel/split/kernel_background_buffer_update.h +++ /dev/null @@ -1,248 +0,0 @@ -/* - * Copyright 2011-2015 Blender Foundation - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -#include "kernel_split_common.h" - -/* Note on kernel_background_buffer_update kernel. - * This is the fourth kernel in the ray tracing logic, and the third - * of the path iteration kernels. This kernel takes care of rays that hit - * the background (sceneintersect kernel), and for the rays of - * state RAY_UPDATE_BUFFER it updates the ray's accumulated radiance in - * the output buffer. This kernel also takes care of rays that have been determined - * to-be-regenerated. - * - * We will empty QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS queue in this kernel - * - * Typically all rays that are in state RAY_HIT_BACKGROUND, RAY_UPDATE_BUFFER - * will be eventually set to RAY_TO_REGENERATE state in this kernel. Finally all rays of ray_state - * RAY_TO_REGENERATE will be regenerated and put in queue QUEUE_ACTIVE_AND_REGENERATED_RAYS. - * - * The input and output are as follows, - * - * rng_coop ---------------------------------------------|--- kernel_background_buffer_update --|--- PathRadiance_coop - * throughput_coop --------------------------------------| |--- L_transparent_coop - * per_sample_output_buffers ----------------------------| |--- per_sample_output_buffers - * Ray_coop ---------------------------------------------| |--- ray_state - * PathState_coop ---------------------------------------| |--- Queue_data (QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS) - * L_transparent_coop -----------------------------------| |--- Queue_data (QUEUE_ACTIVE_AND_REGENERATED_RAYS) - * ray_state --------------------------------------------| |--- Queue_index (QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS) - * Queue_data (QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS) ----| |--- Queue_index (QUEUE_ACTIVE_AND_REGENERATED_RAYS) - * Queue_index (QUEUE_ACTIVE_AND_REGENERATED_RAYS) ------| |--- work_array - * parallel_samples -------------------------------------| |--- PathState_coop - * end_sample -------------------------------------------| |--- throughput_coop - * kg (globals) -----------------------------------------| |--- rng_coop - * rng_state --------------------------------------------| |--- Ray - * PathRadiance_coop ------------------------------------| | - * sw ---------------------------------------------------| | - * sh ---------------------------------------------------| | - * sx ---------------------------------------------------| | - * sy ---------------------------------------------------| | - * stride -----------------------------------------------| | - * work_array -------------------------------------------| |--- work_array - * queuesize --------------------------------------------| | - * start_sample -----------------------------------------| |--- work_pool_wgs - * work_pool_wgs ----------------------------------------| | - * num_samples ------------------------------------------| | - * - * note on sd : sd argument is neither an input nor an output for this kernel. It is just filled and consumed here itself. - * Note on Queues : - * This kernel fetches rays from QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS queue. - * - * State of queues when this kernel is called : - * At entry, - * QUEUE_ACTIVE_AND_REGENERATED_RAYS will be filled with RAY_ACTIVE rays - * QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS will be filled with RAY_UPDATE_BUFFER, RAY_HIT_BACKGROUND, RAY_TO_REGENERATE rays - * At exit, - * QUEUE_ACTIVE_AND_REGENERATED_RAYS will be filled with RAY_ACTIVE and RAY_REGENERATED rays - * QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS will be empty - */ -ccl_device char kernel_background_buffer_update( - KernelGlobals *kg, - ccl_global float *per_sample_output_buffers, - ccl_global uint *rng_state, - ccl_global uint *rng_coop, /* Required for buffer Update */ - ccl_global float3 *throughput_coop, /* Required for background hit processing */ - PathRadiance *PathRadiance_coop, /* Required for background hit processing and buffer Update */ - ccl_global Ray *Ray_coop, /* Required for background hit processing */ - ccl_global PathState *PathState_coop, /* Required for background hit processing */ - ccl_global float *L_transparent_coop, /* Required for background hit processing and buffer Update */ - ccl_global char *ray_state, /* Stores information on the current state of a ray */ - int sw, int sh, int sx, int sy, int stride, - int rng_state_offset_x, - int rng_state_offset_y, - int rng_state_stride, - ccl_global unsigned int *work_array, /* Denotes work of each ray */ - int end_sample, - int start_sample, -#ifdef __WORK_STEALING__ - ccl_global unsigned int *work_pool_wgs, - unsigned int num_samples, -#endif -#ifdef __KERNEL_DEBUG__ - DebugData *debugdata_coop, -#endif - int parallel_samples, /* Number of samples to be processed in parallel */ - int ray_index) -{ - char enqueue_flag = 0; -#ifdef __KERNEL_DEBUG__ - DebugData *debug_data = &debugdata_coop[ray_index]; -#endif - ccl_global PathState *state = &PathState_coop[ray_index]; - PathRadiance *L = L = &PathRadiance_coop[ray_index]; - ccl_global Ray *ray = &Ray_coop[ray_index]; - ccl_global float3 *throughput = &throughput_coop[ray_index]; - ccl_global float *L_transparent = &L_transparent_coop[ray_index]; - ccl_global uint *rng = &rng_coop[ray_index]; - -#ifdef __WORK_STEALING__ - unsigned int my_work; - ccl_global float *initial_per_sample_output_buffers; - ccl_global uint *initial_rng; -#endif - unsigned int sample; - unsigned int tile_x; - unsigned int tile_y; - unsigned int pixel_x; - unsigned int pixel_y; - unsigned int my_sample_tile; - -#ifdef __WORK_STEALING__ - my_work = work_array[ray_index]; - sample = get_my_sample(my_work, sw, sh, parallel_samples, ray_index) + start_sample; - get_pixel_tile_position(&pixel_x, &pixel_y, - &tile_x, &tile_y, - my_work, - sw, sh, sx, sy, - parallel_samples, - ray_index); - my_sample_tile = 0; - initial_per_sample_output_buffers = per_sample_output_buffers; - initial_rng = rng_state; -#else /* __WORK_STEALING__ */ - sample = work_array[ray_index]; - int tile_index = ray_index / parallel_samples; - /* buffer and rng_state's stride is "stride". Find x and y using ray_index */ - tile_x = tile_index % sw; - tile_y = tile_index / sw; - my_sample_tile = ray_index - (tile_index * parallel_samples); -#endif /* __WORK_STEALING__ */ - - rng_state += (rng_state_offset_x + tile_x) + (rng_state_offset_y + tile_y) * rng_state_stride; - per_sample_output_buffers += (((tile_x + (tile_y * stride)) * parallel_samples) + my_sample_tile) * kernel_data.film.pass_stride; - - if(IS_STATE(ray_state, ray_index, RAY_HIT_BACKGROUND)) { - /* eval background shader if nothing hit */ - if(kernel_data.background.transparent && (state->flag & PATH_RAY_CAMERA)) { - *L_transparent = (*L_transparent) + average((*throughput)); -#ifdef __PASSES__ - if(!(kernel_data.film.pass_flag & PASS_BACKGROUND)) -#endif - ASSIGN_RAY_STATE(ray_state, ray_index, RAY_UPDATE_BUFFER); - } - - if(IS_STATE(ray_state, ray_index, RAY_HIT_BACKGROUND)) { -#ifdef __BACKGROUND__ - /* sample background shader */ - float3 L_background = indirect_background(kg, kg->sd_input, state, ray); - path_radiance_accum_background(L, (*throughput), L_background, state->bounce); -#endif - ASSIGN_RAY_STATE(ray_state, ray_index, RAY_UPDATE_BUFFER); - } - } - - if(IS_STATE(ray_state, ray_index, RAY_UPDATE_BUFFER)) { - float3 L_sum = path_radiance_clamp_and_sum(kg, L); - kernel_write_light_passes(kg, per_sample_output_buffers, L, sample); -#ifdef __KERNEL_DEBUG__ - kernel_write_debug_passes(kg, per_sample_output_buffers, state, debug_data, sample); -#endif - float4 L_rad = make_float4(L_sum.x, L_sum.y, L_sum.z, 1.0f - (*L_transparent)); - - /* accumulate result in output buffer */ - kernel_write_pass_float4(per_sample_output_buffers, sample, L_rad); - path_rng_end(kg, rng_state, *rng); - - ASSIGN_RAY_STATE(ray_state, ray_index, RAY_TO_REGENERATE); - } - - if(IS_STATE(ray_state, ray_index, RAY_TO_REGENERATE)) { -#ifdef __WORK_STEALING__ - /* We have completed current work; So get next work */ - int valid_work = get_next_work(work_pool_wgs, &my_work, sw, sh, num_samples, parallel_samples, ray_index); - if(!valid_work) { - /* If work is invalid, this means no more work is available and the thread may exit */ - ASSIGN_RAY_STATE(ray_state, ray_index, RAY_INACTIVE); - } -#else /* __WORK_STEALING__ */ - if((sample + parallel_samples) >= end_sample) { - ASSIGN_RAY_STATE(ray_state, ray_index, RAY_INACTIVE); - } -#endif /* __WORK_STEALING__ */ - - if(IS_STATE(ray_state, ray_index, RAY_TO_REGENERATE)) { -#ifdef __WORK_STEALING__ - work_array[ray_index] = my_work; - /* Get the sample associated with the current work */ - sample = get_my_sample(my_work, sw, sh, parallel_samples, ray_index) + start_sample; - /* Get pixel and tile position associated with current work */ - get_pixel_tile_position(&pixel_x, &pixel_y, &tile_x, &tile_y, my_work, sw, sh, sx, sy, parallel_samples, ray_index); - my_sample_tile = 0; - - /* Remap rng_state according to the current work */ - rng_state = initial_rng + ((rng_state_offset_x + tile_x) + (rng_state_offset_y + tile_y) * rng_state_stride); - /* Remap per_sample_output_buffers according to the current work */ - per_sample_output_buffers = initial_per_sample_output_buffers - + (((tile_x + (tile_y * stride)) * parallel_samples) + my_sample_tile) * kernel_data.film.pass_stride; -#else /* __WORK_STEALING__ */ - work_array[ray_index] = sample + parallel_samples; - sample = work_array[ray_index]; - - /* Get ray position from ray index */ - pixel_x = sx + ((ray_index / parallel_samples) % sw); - pixel_y = sy + ((ray_index / parallel_samples) / sw); -#endif /* __WORK_STEALING__ */ - - /* Initialize random numbers and ray. */ - kernel_path_trace_setup(kg, rng_state, sample, pixel_x, pixel_y, rng, ray); - - if(ray->t != 0.0f) { - /* Initialize throughput, L_transparent, Ray, PathState; - * These rays proceed with path-iteration. - */ - *throughput = make_float3(1.0f, 1.0f, 1.0f); - *L_transparent = 0.0f; - path_radiance_init(L, kernel_data.film.use_light_pass); - path_state_init(kg, kg->sd_input, state, rng, sample, ray); -#ifdef __KERNEL_DEBUG__ - debug_data_init(debug_data); -#endif - ASSIGN_RAY_STATE(ray_state, ray_index, RAY_REGENERATED); - enqueue_flag = 1; - } - else { - /* These rays do not participate in path-iteration. */ - float4 L_rad = make_float4(0.0f, 0.0f, 0.0f, 0.0f); - /* Accumulate result in output buffer. */ - kernel_write_pass_float4(per_sample_output_buffers, sample, L_rad); - path_rng_end(kg, rng_state, *rng); - - ASSIGN_RAY_STATE(ray_state, ray_index, RAY_TO_REGENERATE); - } - } - } - return enqueue_flag; -} diff --git a/intern/cycles/kernel/split/kernel_buffer_update.h b/intern/cycles/kernel/split/kernel_buffer_update.h new file mode 100644 index 00000000000..859c221d976 --- /dev/null +++ b/intern/cycles/kernel/split/kernel_buffer_update.h @@ -0,0 +1,206 @@ +/* + * Copyright 2011-2015 Blender Foundation + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +CCL_NAMESPACE_BEGIN + +/* This kernel takes care of rays that hit the background (sceneintersect + * kernel), and for the rays of state RAY_UPDATE_BUFFER it updates the ray's + * accumulated radiance in the output buffer. This kernel also takes care of + * rays that have been determined to-be-regenerated. + * + * We will empty QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS queue in this kernel. + * + * Typically all rays that are in state RAY_HIT_BACKGROUND, RAY_UPDATE_BUFFER + * will be eventually set to RAY_TO_REGENERATE state in this kernel. + * Finally all rays of ray_state RAY_TO_REGENERATE will be regenerated and put + * in queue QUEUE_ACTIVE_AND_REGENERATED_RAYS. + * + * State of queues when this kernel is called: + * At entry, + * - QUEUE_ACTIVE_AND_REGENERATED_RAYS will be filled with RAY_ACTIVE rays. + * - QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS will be filled with + * RAY_UPDATE_BUFFER, RAY_HIT_BACKGROUND, RAY_TO_REGENERATE rays. + * At exit, + * - QUEUE_ACTIVE_AND_REGENERATED_RAYS will be filled with RAY_ACTIVE and + * RAY_REGENERATED rays. + * - QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS will be empty. + */ +ccl_device void kernel_buffer_update(KernelGlobals *kg, + ccl_local_param unsigned int *local_queue_atomics) +{ + if(ccl_local_id(0) == 0 && ccl_local_id(1) == 0) { + *local_queue_atomics = 0; + } + ccl_barrier(CCL_LOCAL_MEM_FENCE); + + int ray_index = ccl_global_id(1) * ccl_global_size(0) + ccl_global_id(0); + if(ray_index == 0) { + /* We will empty this queue in this kernel. */ + kernel_split_params.queue_index[QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS] = 0; + } + char enqueue_flag = 0; + ray_index = get_ray_index(kg, ray_index, + QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS, + kernel_split_state.queue_data, + kernel_split_params.queue_size, + 1); + +#ifdef __COMPUTE_DEVICE_GPU__ + /* If we are executing on a GPU device, we exit all threads that are not + * required. + * + * If we are executing on a CPU device, then we need to keep all threads + * active since we have barrier() calls later in the kernel. CPU devices, + * expect all threads to execute barrier statement. + */ + if(ray_index == QUEUE_EMPTY_SLOT) { + return; + } +#endif + +#ifndef __COMPUTE_DEVICE_GPU__ + if(ray_index != QUEUE_EMPTY_SLOT) { +#endif + + ccl_global uint *rng_state = kernel_split_params.rng_state; + int stride = kernel_split_params.stride; + + ccl_global char *ray_state = kernel_split_state.ray_state; +#ifdef __KERNEL_DEBUG__ + DebugData *debug_data = &kernel_split_state.debug_data[ray_index]; +#endif + ccl_global PathState *state = &kernel_split_state.path_state[ray_index]; + PathRadiance *L = &kernel_split_state.path_radiance[ray_index]; + ccl_global Ray *ray = &kernel_split_state.ray[ray_index]; + ccl_global float3 *throughput = &kernel_split_state.throughput[ray_index]; + ccl_global float *L_transparent = &kernel_split_state.L_transparent[ray_index]; + RNG rng = kernel_split_state.rng[ray_index]; + ccl_global float *buffer = kernel_split_params.buffer; + + unsigned int work_index; + ccl_global uint *initial_rng; + + unsigned int sample; + unsigned int tile_x; + unsigned int tile_y; + unsigned int pixel_x; + unsigned int pixel_y; + + work_index = kernel_split_state.work_array[ray_index]; + sample = get_work_sample(kg, work_index, ray_index) + kernel_split_params.start_sample; + get_work_pixel_tile_position(kg, &pixel_x, &pixel_y, + &tile_x, &tile_y, + work_index, + ray_index); + initial_rng = rng_state; + + rng_state += kernel_split_params.offset + pixel_x + pixel_y*stride; + buffer += (kernel_split_params.offset + pixel_x + pixel_y*stride) * kernel_data.film.pass_stride; + + if(IS_STATE(ray_state, ray_index, RAY_UPDATE_BUFFER)) { + float3 L_sum; +#ifdef __SHADOW_TRICKS__ + if(state->flag & PATH_RAY_SHADOW_CATCHER) { + L_sum = path_radiance_sum_shadowcatcher(kg, L, L_transparent); + } + else +#endif /* __SHADOW_TRICKS__ */ + { + L_sum = path_radiance_clamp_and_sum(kg, L); + } + kernel_write_light_passes(kg, buffer, L, sample); +#ifdef __KERNEL_DEBUG__ + kernel_write_debug_passes(kg, buffer, state, debug_data, sample); +#endif + float4 L_rad = make_float4(L_sum.x, L_sum.y, L_sum.z, 1.0f - (*L_transparent)); + + /* accumulate result in output buffer */ + kernel_write_pass_float4(buffer, sample, L_rad); + path_rng_end(kg, rng_state, rng); + + ASSIGN_RAY_STATE(ray_state, ray_index, RAY_TO_REGENERATE); + } + + if(IS_STATE(ray_state, ray_index, RAY_TO_REGENERATE)) { + /* We have completed current work; So get next work */ + int valid_work = get_next_work(kg, &work_index, ray_index); + if(!valid_work) { + /* If work is invalid, this means no more work is available and the thread may exit */ + ASSIGN_RAY_STATE(ray_state, ray_index, RAY_INACTIVE); + } + + if(IS_STATE(ray_state, ray_index, RAY_TO_REGENERATE)) { + kernel_split_state.work_array[ray_index] = work_index; + /* Get the sample associated with the current work */ + sample = get_work_sample(kg, work_index, ray_index) + kernel_split_params.start_sample; + /* Get pixel and tile position associated with current work */ + get_work_pixel_tile_position(kg, &pixel_x, &pixel_y, &tile_x, &tile_y, work_index, ray_index); + + /* Remap rng_state according to the current work */ + rng_state = initial_rng + kernel_split_params.offset + pixel_x + pixel_y*stride; + /* Remap buffer according to the current work */ + buffer += (kernel_split_params.offset + pixel_x + pixel_y*stride) * kernel_data.film.pass_stride; + + /* Initialize random numbers and ray. */ + kernel_path_trace_setup(kg, rng_state, sample, pixel_x, pixel_y, &rng, ray); + + if(ray->t != 0.0f) { + /* Initialize throughput, L_transparent, Ray, PathState; + * These rays proceed with path-iteration. + */ + *throughput = make_float3(1.0f, 1.0f, 1.0f); + *L_transparent = 0.0f; + path_radiance_init(L, kernel_data.film.use_light_pass); + path_state_init(kg, &kernel_split_state.sd_DL_shadow[ray_index], state, &rng, sample, ray); +#ifdef __SUBSURFACE__ + kernel_path_subsurface_init_indirect(&kernel_split_state.ss_rays[ray_index]); +#endif +#ifdef __KERNEL_DEBUG__ + debug_data_init(debug_data); +#endif + ASSIGN_RAY_STATE(ray_state, ray_index, RAY_REGENERATED); + enqueue_flag = 1; + } + else { + /* These rays do not participate in path-iteration. */ + float4 L_rad = make_float4(0.0f, 0.0f, 0.0f, 0.0f); + /* Accumulate result in output buffer. */ + kernel_write_pass_float4(buffer, sample, L_rad); + path_rng_end(kg, rng_state, rng); + + ASSIGN_RAY_STATE(ray_state, ray_index, RAY_TO_REGENERATE); + } + } + } + kernel_split_state.rng[ray_index] = rng; + +#ifndef __COMPUTE_DEVICE_GPU__ + } +#endif + + /* Enqueue RAY_REGENERATED rays into QUEUE_ACTIVE_AND_REGENERATED_RAYS; + * These rays will be made active during next SceneIntersectkernel. + */ + enqueue_ray_index_local(ray_index, + QUEUE_ACTIVE_AND_REGENERATED_RAYS, + enqueue_flag, + kernel_split_params.queue_size, + local_queue_atomics, + kernel_split_state.queue_data, + kernel_split_params.queue_index); +} + +CCL_NAMESPACE_END diff --git a/intern/cycles/kernel/split/kernel_data_init.h b/intern/cycles/kernel/split/kernel_data_init.h index 6e158d53d23..9d3d01fff75 100644 --- a/intern/cycles/kernel/split/kernel_data_init.h +++ b/intern/cycles/kernel/split/kernel_data_init.h @@ -14,108 +14,105 @@ * limitations under the License. */ -#include "kernel_split_common.h" +CCL_NAMESPACE_BEGIN -/* Note on kernel_data_initialization kernel - * This kernel Initializes structures needed in path-iteration kernels. - * This is the first kernel in ray-tracing logic. +/* This kernel Initializes structures needed in path-iteration kernels. * - * Ray state of rays outside the tile-boundary will be marked RAY_INACTIVE - * - * Its input and output are as follows, - * - * Un-initialized rng---------------|--- kernel_data_initialization ---|--- Initialized rng - * Un-initialized throughput -------| |--- Initialized throughput - * Un-initialized L_transparent ----| |--- Initialized L_transparent - * Un-initialized PathRadiance -----| |--- Initialized PathRadiance - * Un-initialized Ray --------------| |--- Initialized Ray - * Un-initialized PathState --------| |--- Initialized PathState - * Un-initialized QueueData --------| |--- Initialized QueueData (to QUEUE_EMPTY_SLOT) - * Un-initialized QueueIndex -------| |--- Initialized QueueIndex (to 0) - * Un-initialized use_queues_flag---| |--- Initialized use_queues_flag (to false) - * Un-initialized ray_state --------| |--- Initialized ray_state - * parallel_samples --------------- | |--- Initialized per_sample_output_buffers - * rng_state -----------------------| |--- Initialized work_array - * data ----------------------------| |--- Initialized work_pool_wgs - * start_sample --------------------| | - * sx ------------------------------| | - * sy ------------------------------| | - * sw ------------------------------| | - * sh ------------------------------| | - * stride --------------------------| | - * queuesize -----------------------| | - * num_samples ---------------------| | - * - * Note on Queues : + * Note on Queues: * All slots in queues are initialized to queue empty slot; * The number of elements in the queues is initialized to 0; */ + +/* Distributes an amount of work across all threads + * note: work done inside the loop may not show up to all threads till after + * the current kernel has completed + */ +#define parallel_for(kg, iter_name, work_size) \ + for(size_t _size = (work_size), \ + _global_size = ccl_global_size(0) * ccl_global_size(1), \ + _n = _size / _global_size, \ + _thread = ccl_global_id(0) + ccl_global_id(1) * ccl_global_size(0), \ + iter_name = (_n > 0) ? (_thread * _n) : (_thread) \ + ; \ + (iter_name < (_thread+1) * _n) || (iter_name == _n * _global_size + _thread && _thread < _size % _global_size) \ + ; \ + iter_name = (iter_name != (_thread+1) * _n - 1) ? (iter_name + 1) : (_n * _global_size + _thread) \ + ) + +#ifndef __KERNEL_CPU__ ccl_device void kernel_data_init( +#else +void KERNEL_FUNCTION_FULL_NAME(data_init)( +#endif KernelGlobals *kg, - ShaderData *sd_DL_shadow, ccl_constant KernelData *data, - ccl_global float *per_sample_output_buffers, + ccl_global void *split_data_buffer, + int num_elements, + ccl_global char *ray_state, ccl_global uint *rng_state, - ccl_global uint *rng_coop, /* rng array to store rng values for all rays */ - ccl_global float3 *throughput_coop, /* throughput array to store throughput values for all rays */ - ccl_global float *L_transparent_coop, /* L_transparent array to store L_transparent values for all rays */ - PathRadiance *PathRadiance_coop, /* PathRadiance array to store PathRadiance values for all rays */ - ccl_global Ray *Ray_coop, /* Ray array to store Ray information for all rays */ - ccl_global PathState *PathState_coop, /* PathState array to store PathState information for all rays */ - Intersection *Intersection_coop_shadow, - ccl_global char *ray_state, /* Stores information on current state of a ray */ +#ifdef __KERNEL_OPENCL__ #define KERNEL_TEX(type, ttype, name) \ ccl_global type *name, -#include "../kernel_textures.h" +#include "kernel/kernel_textures.h" +#endif - int start_sample, int sx, int sy, int sw, int sh, int offset, int stride, - int rng_state_offset_x, - int rng_state_offset_y, - int rng_state_stride, - ccl_global int *Queue_data, /* Memory for queues */ + int start_sample, + int end_sample, + int sx, int sy, int sw, int sh, int offset, int stride, ccl_global int *Queue_index, /* Tracks the number of elements in queues */ int queuesize, /* size (capacity) of the queue */ ccl_global char *use_queues_flag, /* flag to decide if scene-intersect kernel should use queues to fetch ray index */ - ccl_global unsigned int *work_array, /* work array to store which work each ray belongs to */ -#ifdef __WORK_STEALING__ - ccl_global unsigned int *work_pool_wgs, /* Work pool for each work group */ - unsigned int num_samples, /* Total number of samples per pixel */ -#endif -#ifdef __KERNEL_DEBUG__ - DebugData *debugdata_coop, -#endif - int parallel_samples) /* Number of samples to be processed in parallel */ + ccl_global unsigned int *work_pools, /* Work pool for each work group */ + unsigned int num_samples, + ccl_global float *buffer) { +#ifdef __KERNEL_OPENCL__ kg->data = data; - kg->sd_input = sd_DL_shadow; - kg->isect_shadow = Intersection_coop_shadow; +#endif + + kernel_split_params.x = sx; + kernel_split_params.y = sy; + kernel_split_params.w = sw; + kernel_split_params.h = sh; + + kernel_split_params.offset = offset; + kernel_split_params.stride = stride; + + kernel_split_params.rng_state = rng_state; + + kernel_split_params.start_sample = start_sample; + kernel_split_params.end_sample = end_sample; + + kernel_split_params.work_pools = work_pools; + kernel_split_params.num_samples = num_samples; + + kernel_split_params.queue_index = Queue_index; + kernel_split_params.queue_size = queuesize; + kernel_split_params.use_queues_flag = use_queues_flag; + + kernel_split_params.buffer = buffer; + + split_data_init(kg, &kernel_split_state, num_elements, split_data_buffer, ray_state); + +#ifdef __KERNEL_OPENCL__ #define KERNEL_TEX(type, ttype, name) \ kg->name = name; -#include "../kernel_textures.h" - - int thread_index = get_global_id(1) * get_global_size(0) + get_global_id(0); +#include "kernel/kernel_textures.h" +#endif -#ifdef __WORK_STEALING__ - int lid = get_local_id(1) * get_local_size(0) + get_local_id(0); - /* Initialize work_pool_wgs */ - if(lid == 0) { - int group_index = get_group_id(1) * get_num_groups(0) + get_group_id(0); - work_pool_wgs[group_index] = 0; - } - barrier(CLK_LOCAL_MEM_FENCE); -#endif /* __WORK_STEALING__ */ + int thread_index = ccl_global_id(1) * ccl_global_size(0) + ccl_global_id(0); /* Initialize queue data and queue index. */ if(thread_index < queuesize) { /* Initialize active ray queue. */ - Queue_data[QUEUE_ACTIVE_AND_REGENERATED_RAYS * queuesize + thread_index] = QUEUE_EMPTY_SLOT; + kernel_split_state.queue_data[QUEUE_ACTIVE_AND_REGENERATED_RAYS * queuesize + thread_index] = QUEUE_EMPTY_SLOT; /* Initialize background and buffer update queue. */ - Queue_data[QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS * queuesize + thread_index] = QUEUE_EMPTY_SLOT; + kernel_split_state.queue_data[QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS * queuesize + thread_index] = QUEUE_EMPTY_SLOT; /* Initialize shadow ray cast of AO queue. */ - Queue_data[QUEUE_SHADOW_RAY_CAST_AO_RAYS * queuesize + thread_index] = QUEUE_EMPTY_SLOT; + kernel_split_state.queue_data[QUEUE_SHADOW_RAY_CAST_AO_RAYS * queuesize + thread_index] = QUEUE_EMPTY_SLOT; /* Initialize shadow ray cast of direct lighting queue. */ - Queue_data[QUEUE_SHADOW_RAY_CAST_DL_RAYS * queuesize + thread_index] = QUEUE_EMPTY_SLOT; + kernel_split_state.queue_data[QUEUE_SHADOW_RAY_CAST_DL_RAYS * queuesize + thread_index] = QUEUE_EMPTY_SLOT; } if(thread_index == 0) { @@ -126,109 +123,31 @@ ccl_device void kernel_data_init( /* The scene-intersect kernel should not use the queues very first time. * since the queue would be empty. */ - use_queues_flag[0] = 0; + *use_queues_flag = 0; } - int x = get_global_id(0); - int y = get_global_id(1); + /* zero the tiles pixels and initialize rng_state if this is the first sample */ + if(start_sample == 0) { + parallel_for(kg, i, sw * sh * kernel_data.film.pass_stride) { + int pixel = i / kernel_data.film.pass_stride; + int pass = i % kernel_data.film.pass_stride; - if(x < (sw * parallel_samples) && y < sh) { - int ray_index = x + y * (sw * parallel_samples); + int x = sx + pixel % sw; + int y = sy + pixel / sw; - /* This is the first assignment to ray_state; - * So we dont use ASSIGN_RAY_STATE macro. - */ - ray_state[ray_index] = RAY_ACTIVE; - - unsigned int my_sample; - unsigned int pixel_x; - unsigned int pixel_y; - unsigned int tile_x; - unsigned int tile_y; - unsigned int my_sample_tile; - -#ifdef __WORK_STEALING__ - unsigned int my_work = 0; - /* Get work. */ - get_next_work(work_pool_wgs, &my_work, sw, sh, num_samples, parallel_samples, ray_index); - /* Get the sample associated with the work. */ - my_sample = get_my_sample(my_work, sw, sh, parallel_samples, ray_index) + start_sample; - - my_sample_tile = 0; - - /* Get pixel and tile position associated with the work. */ - get_pixel_tile_position(&pixel_x, &pixel_y, - &tile_x, &tile_y, - my_work, - sw, sh, sx, sy, - parallel_samples, - ray_index); - work_array[ray_index] = my_work; -#else /* __WORK_STEALING__ */ - unsigned int tile_index = ray_index / parallel_samples; - tile_x = tile_index % sw; - tile_y = tile_index / sw; - my_sample_tile = ray_index - (tile_index * parallel_samples); - my_sample = my_sample_tile + start_sample; - - /* Initialize work array. */ - work_array[ray_index] = my_sample ; - - /* Calculate pixel position of this ray. */ - pixel_x = sx + tile_x; - pixel_y = sy + tile_y; -#endif /* __WORK_STEALING__ */ - - rng_state += (rng_state_offset_x + tile_x) + (rng_state_offset_y + tile_y) * rng_state_stride; - - /* Initialise per_sample_output_buffers to all zeros. */ - per_sample_output_buffers += (((tile_x + (tile_y * stride)) * parallel_samples) + (my_sample_tile)) * kernel_data.film.pass_stride; - int per_sample_output_buffers_iterator = 0; - for(per_sample_output_buffers_iterator = 0; - per_sample_output_buffers_iterator < kernel_data.film.pass_stride; - per_sample_output_buffers_iterator++) - { - per_sample_output_buffers[per_sample_output_buffers_iterator] = 0.0f; - } + int index = (offset + x + y*stride) * kernel_data.film.pass_stride + pass; - /* Initialize random numbers and ray. */ - kernel_path_trace_setup(kg, - rng_state, - my_sample, - pixel_x, pixel_y, - &rng_coop[ray_index], - &Ray_coop[ray_index]); - - if(Ray_coop[ray_index].t != 0.0f) { - /* Initialize throughput, L_transparent, Ray, PathState; - * These rays proceed with path-iteration. - */ - throughput_coop[ray_index] = make_float3(1.0f, 1.0f, 1.0f); - L_transparent_coop[ray_index] = 0.0f; - path_radiance_init(&PathRadiance_coop[ray_index], kernel_data.film.use_light_pass); - path_state_init(kg, - kg->sd_input, - &PathState_coop[ray_index], - &rng_coop[ray_index], - my_sample, - &Ray_coop[ray_index]); -#ifdef __KERNEL_DEBUG__ - debug_data_init(&debugdata_coop[ray_index]); -#endif + *(buffer + index) = 0.0f; } - else { - /* These rays do not participate in path-iteration. */ - float4 L_rad = make_float4(0.0f, 0.0f, 0.0f, 0.0f); - /* Accumulate result in output buffer. */ - kernel_write_pass_float4(per_sample_output_buffers, my_sample, L_rad); - path_rng_end(kg, rng_state, rng_coop[ray_index]); - ASSIGN_RAY_STATE(ray_state, ray_index, RAY_TO_REGENERATE); - } - } - /* Mark rest of the ray-state indices as RAY_INACTIVE. */ - if(thread_index < (get_global_size(0) * get_global_size(1)) - (sh * (sw * parallel_samples))) { - /* First assignment, hence we dont use ASSIGN_RAY_STATE macro */ - ray_state[((sw * parallel_samples) * sh) + thread_index] = RAY_INACTIVE; + parallel_for(kg, i, sw * sh) { + int x = sx + i % sw; + int y = sy + i / sw; + + int index = (offset + x + y*stride); + *(rng_state + index) = hash_int_2d(x, y); + } } } + +CCL_NAMESPACE_END diff --git a/intern/cycles/kernel/split/kernel_direct_lighting.h b/intern/cycles/kernel/split/kernel_direct_lighting.h index 82ca18829d3..bdbf7387b95 100644 --- a/intern/cycles/kernel/split/kernel_direct_lighting.h +++ b/intern/cycles/kernel/split/kernel_direct_lighting.h @@ -14,95 +14,144 @@ * limitations under the License. */ -#include "kernel_split_common.h" +CCL_NAMESPACE_BEGIN -/* Note on kernel_direct_lighting kernel. - * This is the eighth kernel in the ray tracing logic. This is the seventh - * of the path iteration kernels. This kernel takes care of direct lighting - * logic. However, the "shadow ray cast" part of direct lighting is handled +/* This kernel takes care of direct lighting logic. + * However, the "shadow ray cast" part of direct lighting is handled * in the next kernel. * - * This kernels determines the rays for which a shadow_blocked() function associated with direct lighting should be executed. - * Those rays for which a shadow_blocked() function for direct-lighting must be executed, are marked with flag RAY_SHADOW_RAY_CAST_DL and - * enqueued into the queue QUEUE_SHADOW_RAY_CAST_DL_RAYS + * This kernels determines the rays for which a shadow_blocked() function + * associated with direct lighting should be executed. Those rays for which + * a shadow_blocked() function for direct-lighting must be executed, are + * marked with flag RAY_SHADOW_RAY_CAST_DL and enqueued into the queue + * QUEUE_SHADOW_RAY_CAST_DL_RAYS * - * The input and output are as follows, + * Note on Queues: + * This kernel only reads from the QUEUE_ACTIVE_AND_REGENERATED_RAYS queue + * and processes only the rays of state RAY_ACTIVE; If a ray needs to execute + * the corresponding shadow_blocked part, after direct lighting, the ray is + * marked with RAY_SHADOW_RAY_CAST_DL flag. * - * rng_coop -----------------------------------------|--- kernel_direct_lighting --|--- BSDFEval_coop - * PathState_coop -----------------------------------| |--- ISLamp_coop - * sd -----------------------------------------------| |--- LightRay_coop - * ray_state ----------------------------------------| |--- ray_state - * Queue_data (QUEUE_ACTIVE_AND_REGENERATED_RAYS) ---| | - * kg (globals) -------------------------------------| | - * queuesize ----------------------------------------| | - * - * Note on Queues : - * This kernel only reads from the QUEUE_ACTIVE_AND_REGENERATED_RAYS queue and processes - * only the rays of state RAY_ACTIVE; If a ray needs to execute the corresponding shadow_blocked - * part, after direct lighting, the ray is marked with RAY_SHADOW_RAY_CAST_DL flag. - * - * State of queues when this kernel is called : - * state of queues QUEUE_ACTIVE_AND_REGENERATED_RAYS and QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS will be same - * before and after this kernel call. - * QUEUE_SHADOW_RAY_CAST_DL_RAYS queue will be filled with rays for which a shadow_blocked function must be executed, after this - * kernel call. Before this kernel call the QUEUE_SHADOW_RAY_CAST_DL_RAYS will be empty. + * State of queues when this kernel is called: + * - State of queues QUEUE_ACTIVE_AND_REGENERATED_RAYS and + * QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS will be same before and after this + * kernel call. + * - QUEUE_SHADOW_RAY_CAST_DL_RAYS queue will be filled with rays for which a + * shadow_blocked function must be executed, after this kernel call + * Before this kernel call the QUEUE_SHADOW_RAY_CAST_DL_RAYS will be empty. */ -ccl_device char kernel_direct_lighting( - KernelGlobals *kg, - ShaderData *sd, /* Required for direct lighting */ - ccl_global uint *rng_coop, /* Required for direct lighting */ - ccl_global PathState *PathState_coop, /* Required for direct lighting */ - ccl_global int *ISLamp_coop, /* Required for direct lighting */ - ccl_global Ray *LightRay_coop, /* Required for direct lighting */ - ccl_global BsdfEval *BSDFEval_coop, /* Required for direct lighting */ - ccl_global char *ray_state, /* Denotes the state of each ray */ - int ray_index) +ccl_device void kernel_direct_lighting(KernelGlobals *kg, + ccl_local_param unsigned int *local_queue_atomics) { + if(ccl_local_id(0) == 0 && ccl_local_id(1) == 0) { + *local_queue_atomics = 0; + } + ccl_barrier(CCL_LOCAL_MEM_FENCE); + char enqueue_flag = 0; - if(IS_STATE(ray_state, ray_index, RAY_ACTIVE)) { - ccl_global PathState *state = &PathState_coop[ray_index]; + int ray_index = ccl_global_id(1) * ccl_global_size(0) + ccl_global_id(0); + ray_index = get_ray_index(kg, ray_index, + QUEUE_ACTIVE_AND_REGENERATED_RAYS, + kernel_split_state.queue_data, + kernel_split_params.queue_size, + 0); + +#ifdef __COMPUTE_DEVICE_GPU__ + /* If we are executing on a GPU device, we exit all threads that are not + * required. + * + * If we are executing on a CPU device, then we need to keep all threads + * active since we have barrier() calls later in the kernel. CPU devices, + * expect all threads to execute barrier statement. + */ + if(ray_index == QUEUE_EMPTY_SLOT) { + return; + } +#endif + +#ifndef __COMPUTE_DEVICE_GPU__ + if(ray_index != QUEUE_EMPTY_SLOT) { +#endif + + if(IS_STATE(kernel_split_state.ray_state, ray_index, RAY_ACTIVE)) { + ccl_global PathState *state = &kernel_split_state.path_state[ray_index]; + ShaderData *sd = &kernel_split_state.sd[ray_index]; /* direct lighting */ #ifdef __EMISSION__ - if((kernel_data.integrator.use_direct_light && - (ccl_fetch(sd, flag) & SD_BSDF_HAS_EVAL))) - { + RNG rng = kernel_split_state.rng[ray_index]; + bool flag = (kernel_data.integrator.use_direct_light && + (sd->flag & SD_BSDF_HAS_EVAL)); +# ifdef __SHADOW_TRICKS__ + if(flag && state->flag & PATH_RAY_SHADOW_CATCHER) { + flag = false; + ShaderData *emission_sd = &kernel_split_state.sd_DL_shadow[ray_index]; + float3 throughput = kernel_split_state.throughput[ray_index]; + PathRadiance *L = &kernel_split_state.path_radiance[ray_index]; + kernel_branched_path_surface_connect_light(kg, + &rng, + sd, + emission_sd, + state, + throughput, + 1.0f, + L, + 1); + } +# endif /* __SHADOW_TRICKS__ */ + if(flag) { /* Sample illumination from lights to find path contribution. */ - ccl_global RNG* rng = &rng_coop[ray_index]; - float light_t = path_state_rng_1D(kg, rng, state, PRNG_LIGHT); + float light_t = path_state_rng_1D(kg, &rng, state, PRNG_LIGHT); float light_u, light_v; - path_state_rng_2D(kg, rng, state, PRNG_LIGHT_U, &light_u, &light_v); - float terminate = path_state_rng_light_termination(kg, rng, state); + path_state_rng_2D(kg, &rng, state, PRNG_LIGHT_U, &light_u, &light_v); + float terminate = path_state_rng_light_termination(kg, &rng, state); LightSample ls; if(light_sample(kg, light_t, light_u, light_v, - ccl_fetch(sd, time), - ccl_fetch(sd, P), + sd->time, + sd->P, state->bounce, &ls)) { Ray light_ray; -#ifdef __OBJECT_MOTION__ - light_ray.time = ccl_fetch(sd, time); -#endif +# ifdef __OBJECT_MOTION__ + light_ray.time = sd->time; +# endif BsdfEval L_light; bool is_lamp; - if(direct_emission(kg, sd, kg->sd_input, &ls, state, &light_ray, &L_light, &is_lamp, terminate)) { + if(direct_emission(kg, sd, &kernel_split_state.sd_DL_shadow[ray_index], &ls, state, &light_ray, &L_light, &is_lamp, terminate)) { /* Write intermediate data to global memory to access from * the next kernel. */ - LightRay_coop[ray_index] = light_ray; - BSDFEval_coop[ray_index] = L_light; - ISLamp_coop[ray_index] = is_lamp; + kernel_split_state.light_ray[ray_index] = light_ray; + kernel_split_state.bsdf_eval[ray_index] = L_light; + kernel_split_state.is_lamp[ray_index] = is_lamp; /* Mark ray state for next shadow kernel. */ - ADD_RAY_FLAG(ray_state, ray_index, RAY_SHADOW_RAY_CAST_DL); + ADD_RAY_FLAG(kernel_split_state.ray_state, ray_index, RAY_SHADOW_RAY_CAST_DL); enqueue_flag = 1; } } } + kernel_split_state.rng[ray_index] = rng; #endif /* __EMISSION__ */ } - return enqueue_flag; + +#ifndef __COMPUTE_DEVICE_GPU__ + } +#endif + +#ifdef __EMISSION__ + /* Enqueue RAY_SHADOW_RAY_CAST_DL rays. */ + enqueue_ray_index_local(ray_index, + QUEUE_SHADOW_RAY_CAST_DL_RAYS, + enqueue_flag, + kernel_split_params.queue_size, + local_queue_atomics, + kernel_split_state.queue_data, + kernel_split_params.queue_index); +#endif } + +CCL_NAMESPACE_END diff --git a/intern/cycles/kernel/split/kernel_do_volume.h b/intern/cycles/kernel/split/kernel_do_volume.h new file mode 100644 index 00000000000..47d3c280831 --- /dev/null +++ b/intern/cycles/kernel/split/kernel_do_volume.h @@ -0,0 +1,95 @@ +/* + * Copyright 2011-2017 Blender Foundation + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +CCL_NAMESPACE_BEGIN + + +ccl_device void kernel_do_volume(KernelGlobals *kg) +{ +#ifdef __VOLUME__ + /* We will empty this queue in this kernel. */ + if(ccl_global_id(0) == 0 && ccl_global_id(1) == 0) { + kernel_split_params.queue_index[QUEUE_ACTIVE_AND_REGENERATED_RAYS] = 0; + } + /* Fetch use_queues_flag. */ + char local_use_queues_flag = *kernel_split_params.use_queues_flag; + ccl_barrier(CCL_LOCAL_MEM_FENCE); + + int ray_index = ccl_global_id(1) * ccl_global_size(0) + ccl_global_id(0); + if(local_use_queues_flag) { + ray_index = get_ray_index(kg, ray_index, + QUEUE_ACTIVE_AND_REGENERATED_RAYS, + kernel_split_state.queue_data, + kernel_split_params.queue_size, + 1); + if(ray_index == QUEUE_EMPTY_SLOT) { + return; + } + } + + if(IS_STATE(kernel_split_state.ray_state, ray_index, RAY_ACTIVE) || + IS_STATE(kernel_split_state.ray_state, ray_index, RAY_HIT_BACKGROUND)) { + + bool hit = ! IS_STATE(kernel_split_state.ray_state, ray_index, RAY_HIT_BACKGROUND); + + PathRadiance *L = &kernel_split_state.path_radiance[ray_index]; + ccl_global PathState *state = &kernel_split_state.path_state[ray_index]; + + ccl_global float3 *throughput = &kernel_split_state.throughput[ray_index]; + ccl_global Ray *ray = &kernel_split_state.ray[ray_index]; + RNG rng = kernel_split_state.rng[ray_index]; + ccl_global Intersection *isect = &kernel_split_state.isect[ray_index]; + ShaderData *sd = &kernel_split_state.sd[ray_index]; + ShaderData *sd_input = &kernel_split_state.sd_DL_shadow[ray_index]; + + /* Sanitize volume stack. */ + if(!hit) { + kernel_volume_clean_stack(kg, state->volume_stack); + } + /* volume attenuation, emission, scatter */ + if(state->volume_stack[0].shader != SHADER_NONE) { + Ray volume_ray = *ray; + volume_ray.t = (hit)? isect->t: FLT_MAX; + + bool heterogeneous = volume_stack_is_heterogeneous(kg, state->volume_stack); + + { + /* integrate along volume segment with distance sampling */ + VolumeIntegrateResult result = kernel_volume_integrate( + kg, state, sd, &volume_ray, L, throughput, &rng, heterogeneous); + +# ifdef __VOLUME_SCATTER__ + if(result == VOLUME_PATH_SCATTERED) { + /* direct lighting */ + kernel_path_volume_connect_light(kg, &rng, sd, sd_input, *throughput, state, L); + + /* indirect light bounce */ + if(kernel_path_volume_bounce(kg, &rng, sd, throughput, state, L, ray)) + ASSIGN_RAY_STATE(kernel_split_state.ray_state, ray_index, RAY_REGENERATED); + else + ASSIGN_RAY_STATE(kernel_split_state.ray_state, ray_index, RAY_UPDATE_BUFFER); + } +# endif + } + } + kernel_split_state.rng[ray_index] = rng; + } + +#endif +} + + +CCL_NAMESPACE_END diff --git a/intern/cycles/kernel/split/kernel_holdout_emission_blurring_pathtermination_ao.h b/intern/cycles/kernel/split/kernel_holdout_emission_blurring_pathtermination_ao.h index 5d951b972ed..9fc853a84bf 100644 --- a/intern/cycles/kernel/split/kernel_holdout_emission_blurring_pathtermination_ao.h +++ b/intern/cycles/kernel/split/kernel_holdout_emission_blurring_pathtermination_ao.h @@ -14,157 +14,159 @@ * limitations under the License. */ -#include "kernel_split_common.h" +CCL_NAMESPACE_BEGIN -/* Note on kernel_holdout_emission_blurring_pathtermination_ao kernel. - * This is the sixth kernel in the ray tracing logic. This is the fifth - * of the path iteration kernels. This kernel takes care of the logic to process - * "material of type holdout", indirect primitive emission, bsdf blurring, - * probabilistic path termination and AO. +/* This kernel takes care of the logic to process "material of type holdout", + * indirect primitive emission, bsdf blurring, probabilistic path termination + * and AO. * - * This kernels determines the rays for which a shadow_blocked() function associated with AO should be executed. - * Those rays for which a shadow_blocked() function for AO must be executed are marked with flag RAY_SHADOW_RAY_CAST_ao and - * enqueued into the queue QUEUE_SHADOW_RAY_CAST_AO_RAYS + * This kernels determines the rays for which a shadow_blocked() function + * associated with AO should be executed. Those rays for which a + * shadow_blocked() function for AO must be executed are marked with flag + * RAY_SHADOW_RAY_CAST_ao and enqueued into the queue + * QUEUE_SHADOW_RAY_CAST_AO_RAYS * * Ray state of rays that are terminated in this kernel are changed to RAY_UPDATE_BUFFER * - * The input and output are as follows, + * Note on Queues: + * This kernel fetches rays from the queue QUEUE_ACTIVE_AND_REGENERATED_RAYS + * and processes only the rays of state RAY_ACTIVE. + * There are different points in this kernel where a ray may terminate and + * reach RAY_UPDATE_BUFFER state. These rays are enqueued into + * QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS queue. These rays will still be present + * in QUEUE_ACTIVE_AND_REGENERATED_RAYS queue, but since their ray-state has + * been changed to RAY_UPDATE_BUFFER, there is no problem. * - * rng_coop ---------------------------------------------|--- kernel_holdout_emission_blurring_pathtermination_ao ---|--- Queue_index (QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS) - * throughput_coop --------------------------------------| |--- PathState_coop - * PathRadiance_coop ------------------------------------| |--- throughput_coop - * Intersection_coop ------------------------------------| |--- L_transparent_coop - * PathState_coop ---------------------------------------| |--- per_sample_output_buffers - * L_transparent_coop -----------------------------------| |--- PathRadiance_coop - * sd ---------------------------------------------------| |--- ShaderData - * ray_state --------------------------------------------| |--- ray_state - * Queue_data (QUEUE_ACTIVE_AND_REGENERATED_RAYS) -------| |--- Queue_data (QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS) - * Queue_index (QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS) ---| |--- AOAlpha_coop - * kg (globals) -----------------------------------------| |--- AOBSDF_coop - * parallel_samples -------------------------------------| |--- AOLightRay_coop - * per_sample_output_buffers ----------------------------| | - * sw ---------------------------------------------------| | - * sh ---------------------------------------------------| | - * sx ---------------------------------------------------| | - * sy ---------------------------------------------------| | - * stride -----------------------------------------------| | - * work_array -------------------------------------------| | - * queuesize --------------------------------------------| | - * start_sample -----------------------------------------| | - * - * Note on Queues : - * This kernel fetches rays from the queue QUEUE_ACTIVE_AND_REGENERATED_RAYS and processes only - * the rays of state RAY_ACTIVE. - * There are different points in this kernel where a ray may terminate and reach RAY_UPDATE_BUFFER - * state. These rays are enqueued into QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS queue. These rays will - * still be present in QUEUE_ACTIVE_AND_REGENERATED_RAYS queue, but since their ray-state has been - * changed to RAY_UPDATE_BUFFER, there is no problem. - * - * State of queues when this kernel is called : + * State of queues when this kernel is called: * At entry, - * QUEUE_ACTIVE_AND_REGENERATED_RAYS will be filled with RAY_ACTIVE and RAY_REGENERATED rays - * QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS will be filled with RAY_TO_REGENERATE rays. - * QUEUE_SHADOW_RAY_CAST_AO_RAYS will be empty. + * - QUEUE_ACTIVE_AND_REGENERATED_RAYS will be filled with RAY_ACTIVE and + * RAY_REGENERATED rays + * - QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS will be filled with + * RAY_TO_REGENERATE rays. + * - QUEUE_SHADOW_RAY_CAST_AO_RAYS will be empty. * At exit, - * QUEUE_ACTIVE_AND_REGENERATED_RAYS will be filled with RAY_ACTIVE, RAY_REGENERATED and RAY_UPDATE_BUFFER rays - * QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS will be filled with RAY_TO_REGENERATE and RAY_UPDATE_BUFFER rays - * QUEUE_SHADOW_RAY_CAST_AO_RAYS will be filled with rays marked with flag RAY_SHADOW_RAY_CAST_AO + * - QUEUE_ACTIVE_AND_REGENERATED_RAYS will be filled with RAY_ACTIVE, + * RAY_REGENERATED and RAY_UPDATE_BUFFER rays. + * - QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS will be filled with + * RAY_TO_REGENERATE and RAY_UPDATE_BUFFER rays. + * - QUEUE_SHADOW_RAY_CAST_AO_RAYS will be filled with rays marked with + * flag RAY_SHADOW_RAY_CAST_AO */ ccl_device void kernel_holdout_emission_blurring_pathtermination_ao( KernelGlobals *kg, - ShaderData *sd, /* Required throughout the kernel except probabilistic path termination and AO */ - ccl_global float *per_sample_output_buffers, - ccl_global uint *rng_coop, /* Required for "kernel_write_data_passes" and AO */ - ccl_global float3 *throughput_coop, /* Required for handling holdout material and AO */ - ccl_global float *L_transparent_coop, /* Required for handling holdout material */ - PathRadiance *PathRadiance_coop, /* Required for "kernel_write_data_passes" and indirect primitive emission */ - ccl_global PathState *PathState_coop, /* Required throughout the kernel and AO */ - Intersection *Intersection_coop, /* Required for indirect primitive emission */ - ccl_global float3 *AOAlpha_coop, /* Required for AO */ - ccl_global float3 *AOBSDF_coop, /* Required for AO */ - ccl_global Ray *AOLightRay_coop, /* Required for AO */ - int sw, int sh, int sx, int sy, int stride, - ccl_global char *ray_state, /* Denotes the state of each ray */ - ccl_global unsigned int *work_array, /* Denotes the work that each ray belongs to */ -#ifdef __WORK_STEALING__ - unsigned int start_sample, -#endif - int parallel_samples, /* Number of samples to be processed in parallel */ - int ray_index, - char *enqueue_flag, - char *enqueue_flag_AO_SHADOW_RAY_CAST) + ccl_local_param BackgroundAOLocals *locals) { -#ifdef __WORK_STEALING__ - unsigned int my_work; + if(ccl_local_id(0) == 0 && ccl_local_id(1) == 0) { + locals->queue_atomics_bg = 0; + locals->queue_atomics_ao = 0; + } + ccl_barrier(CCL_LOCAL_MEM_FENCE); + + char enqueue_flag = 0; + char enqueue_flag_AO_SHADOW_RAY_CAST = 0; + int ray_index = ccl_global_id(1) * ccl_global_size(0) + ccl_global_id(0); + ray_index = get_ray_index(kg, ray_index, + QUEUE_ACTIVE_AND_REGENERATED_RAYS, + kernel_split_state.queue_data, + kernel_split_params.queue_size, + 0); + +#ifdef __COMPUTE_DEVICE_GPU__ + /* If we are executing on a GPU device, we exit all threads that are not + * required. + * + * If we are executing on a CPU device, then we need to keep all threads + * active since we have barrier() calls later in the kernel. CPU devices, + * expect all threads to execute barrier statement. + */ + if(ray_index == QUEUE_EMPTY_SLOT) { + return; + } +#endif /* __COMPUTE_DEVICE_GPU__ */ + +#ifndef __COMPUTE_DEVICE_GPU__ + if(ray_index != QUEUE_EMPTY_SLOT) { +#endif + + int stride = kernel_split_params.stride; + + unsigned int work_index; unsigned int pixel_x; unsigned int pixel_y; -#endif + unsigned int tile_x; unsigned int tile_y; - int my_sample_tile; unsigned int sample; - ccl_global RNG *rng = 0x0; + RNG rng = kernel_split_state.rng[ray_index]; ccl_global PathState *state = 0x0; float3 throughput; + ccl_global char *ray_state = kernel_split_state.ray_state; + ShaderData *sd = &kernel_split_state.sd[ray_index]; + ccl_global float *buffer = kernel_split_params.buffer; + if(IS_STATE(ray_state, ray_index, RAY_ACTIVE)) { - throughput = throughput_coop[ray_index]; - state = &PathState_coop[ray_index]; - rng = &rng_coop[ray_index]; -#ifdef __WORK_STEALING__ - my_work = work_array[ray_index]; - sample = get_my_sample(my_work, sw, sh, parallel_samples, ray_index) + start_sample; - get_pixel_tile_position(&pixel_x, &pixel_y, + throughput = kernel_split_state.throughput[ray_index]; + state = &kernel_split_state.path_state[ray_index]; + + work_index = kernel_split_state.work_array[ray_index]; + sample = get_work_sample(kg, work_index, ray_index) + kernel_split_params.start_sample; + get_work_pixel_tile_position(kg, &pixel_x, &pixel_y, &tile_x, &tile_y, - my_work, - sw, sh, sx, sy, - parallel_samples, + work_index, ray_index); - my_sample_tile = 0; -#else /* __WORK_STEALING__ */ - sample = work_array[ray_index]; - /* Buffer's stride is "stride"; Find x and y using ray_index. */ - int tile_index = ray_index / parallel_samples; - tile_x = tile_index % sw; - tile_y = tile_index / sw; - my_sample_tile = ray_index - (tile_index * parallel_samples); -#endif /* __WORK_STEALING__ */ - per_sample_output_buffers += - (((tile_x + (tile_y * stride)) * parallel_samples) + my_sample_tile) * - kernel_data.film.pass_stride; + + buffer += (kernel_split_params.offset + pixel_x + pixel_y * stride) * kernel_data.film.pass_stride; + +#ifdef __SHADOW_TRICKS__ + if((sd->object_flag & SD_OBJECT_SHADOW_CATCHER)) { + if (state->flag & PATH_RAY_CAMERA) { + state->flag |= (PATH_RAY_SHADOW_CATCHER | PATH_RAY_SHADOW_CATCHER_ONLY); + state->catcher_object = sd->object; + if(!kernel_data.background.transparent) { + PathRadiance *L = &kernel_split_state.path_radiance[ray_index]; + ccl_global Ray *ray = &kernel_split_state.ray[ray_index]; + L->shadow_color = indirect_background(kg, &kernel_split_state.sd_DL_shadow[ray_index], state, ray); + } + } + } + else { + state->flag &= ~PATH_RAY_SHADOW_CATCHER_ONLY; + } +#endif /* __SHADOW_TRICKS__ */ /* holdout */ #ifdef __HOLDOUT__ - if(((ccl_fetch(sd, flag) & SD_HOLDOUT) || - (ccl_fetch(sd, object_flag) & SD_OBJECT_HOLDOUT_MASK)) && + if(((sd->flag & SD_HOLDOUT) || + (sd->object_flag & SD_OBJECT_HOLDOUT_MASK)) && (state->flag & PATH_RAY_CAMERA)) { if(kernel_data.background.transparent) { float3 holdout_weight; - if(ccl_fetch(sd, object_flag) & SD_OBJECT_HOLDOUT_MASK) { + if(sd->object_flag & SD_OBJECT_HOLDOUT_MASK) { holdout_weight = make_float3(1.0f, 1.0f, 1.0f); } else { holdout_weight = shader_holdout_eval(kg, sd); } /* any throughput is ok, should all be identical here */ - L_transparent_coop[ray_index] += average(holdout_weight*throughput); + kernel_split_state.L_transparent[ray_index] += average(holdout_weight*throughput); } - if(ccl_fetch(sd, object_flag) & SD_OBJECT_HOLDOUT_MASK) { + if(sd->object_flag & SD_OBJECT_HOLDOUT_MASK) { ASSIGN_RAY_STATE(ray_state, ray_index, RAY_UPDATE_BUFFER); - *enqueue_flag = 1; + enqueue_flag = 1; } } #endif /* __HOLDOUT__ */ } if(IS_STATE(ray_state, ray_index, RAY_ACTIVE)) { - PathRadiance *L = &PathRadiance_coop[ray_index]; + PathRadiance *L = &kernel_split_state.path_radiance[ray_index]; /* Holdout mask objects do not write data passes. */ kernel_write_data_passes(kg, - per_sample_output_buffers, + buffer, L, sd, sample, @@ -183,12 +185,12 @@ ccl_device void kernel_holdout_emission_blurring_pathtermination_ao( #ifdef __EMISSION__ /* emission */ - if(ccl_fetch(sd, flag) & SD_EMISSION) { + if(sd->flag & SD_EMISSION) { /* TODO(sergey): is isect.t wrong here for transparent surfaces? */ float3 emission = indirect_primitive_emission( kg, sd, - Intersection_coop[ray_index].t, + kernel_split_state.isect[ray_index].t, state->flag, state->ray_pdf); path_radiance_accum_emission(L, throughput, emission, state->bounce); @@ -203,18 +205,18 @@ ccl_device void kernel_holdout_emission_blurring_pathtermination_ao( if(probability == 0.0f) { ASSIGN_RAY_STATE(ray_state, ray_index, RAY_UPDATE_BUFFER); - *enqueue_flag = 1; + enqueue_flag = 1; } if(IS_STATE(ray_state, ray_index, RAY_ACTIVE)) { if(probability != 1.0f) { - float terminate = path_state_rng_1D_for_decision(kg, rng, state, PRNG_TERMINATE); + float terminate = path_state_rng_1D_for_decision(kg, &rng, state, PRNG_TERMINATE); if(terminate >= probability) { ASSIGN_RAY_STATE(ray_state, ray_index, RAY_UPDATE_BUFFER); - *enqueue_flag = 1; + enqueue_flag = 1; } else { - throughput_coop[ray_index] = throughput/probability; + kernel_split_state.throughput[ray_index] = throughput/probability; } } } @@ -224,37 +226,65 @@ ccl_device void kernel_holdout_emission_blurring_pathtermination_ao( if(IS_STATE(ray_state, ray_index, RAY_ACTIVE)) { /* ambient occlusion */ if(kernel_data.integrator.use_ambient_occlusion || - (ccl_fetch(sd, flag) & SD_AO)) + (sd->flag & SD_AO)) { /* todo: solve correlation */ float bsdf_u, bsdf_v; - path_state_rng_2D(kg, rng, state, PRNG_BSDF_U, &bsdf_u, &bsdf_v); + path_state_rng_2D(kg, &rng, state, PRNG_BSDF_U, &bsdf_u, &bsdf_v); float ao_factor = kernel_data.background.ao_factor; float3 ao_N; - AOBSDF_coop[ray_index] = shader_bsdf_ao(kg, sd, ao_factor, &ao_N); - AOAlpha_coop[ray_index] = shader_bsdf_alpha(kg, sd); + kernel_split_state.ao_bsdf[ray_index] = shader_bsdf_ao(kg, sd, ao_factor, &ao_N); + kernel_split_state.ao_alpha[ray_index] = shader_bsdf_alpha(kg, sd); float3 ao_D; float ao_pdf; sample_cos_hemisphere(ao_N, bsdf_u, bsdf_v, &ao_D, &ao_pdf); - if(dot(ccl_fetch(sd, Ng), ao_D) > 0.0f && ao_pdf != 0.0f) { + if(dot(sd->Ng, ao_D) > 0.0f && ao_pdf != 0.0f) { Ray _ray; - _ray.P = ray_offset(ccl_fetch(sd, P), ccl_fetch(sd, Ng)); + _ray.P = ray_offset(sd->P, sd->Ng); _ray.D = ao_D; _ray.t = kernel_data.background.ao_distance; #ifdef __OBJECT_MOTION__ - _ray.time = ccl_fetch(sd, time); + _ray.time = sd->time; #endif - _ray.dP = ccl_fetch(sd, dP); + _ray.dP = sd->dP; _ray.dD = differential3_zero(); - AOLightRay_coop[ray_index] = _ray; + kernel_split_state.ao_light_ray[ray_index] = _ray; ADD_RAY_FLAG(ray_state, ray_index, RAY_SHADOW_RAY_CAST_AO); - *enqueue_flag_AO_SHADOW_RAY_CAST = 1; + enqueue_flag_AO_SHADOW_RAY_CAST = 1; } } } #endif /* __AO__ */ + kernel_split_state.rng[ray_index] = rng; + + +#ifndef __COMPUTE_DEVICE_GPU__ + } +#endif + + /* Enqueue RAY_UPDATE_BUFFER rays. */ + enqueue_ray_index_local(ray_index, + QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS, + enqueue_flag, + kernel_split_params.queue_size, + &locals->queue_atomics_bg, + kernel_split_state.queue_data, + kernel_split_params.queue_index); + +#ifdef __AO__ + /* Enqueue to-shadow-ray-cast rays. */ + enqueue_ray_index_local(ray_index, + QUEUE_SHADOW_RAY_CAST_AO_RAYS, + enqueue_flag_AO_SHADOW_RAY_CAST, + kernel_split_params.queue_size, + &locals->queue_atomics_ao, + kernel_split_state.queue_data, + kernel_split_params.queue_index); +#endif } + +CCL_NAMESPACE_END diff --git a/intern/cycles/kernel/split/kernel_indirect_background.h b/intern/cycles/kernel/split/kernel_indirect_background.h new file mode 100644 index 00000000000..8192528622e --- /dev/null +++ b/intern/cycles/kernel/split/kernel_indirect_background.h @@ -0,0 +1,82 @@ +/* + * Copyright 2011-2017 Blender Foundation + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +CCL_NAMESPACE_BEGIN + +ccl_device void kernel_indirect_background(KernelGlobals *kg) +{ + ccl_global char *ray_state = kernel_split_state.ray_state; + + int thread_index = ccl_global_id(1) * ccl_global_size(0) + ccl_global_id(0); + int ray_index; + + if(kernel_data.integrator.ao_bounces) { + ray_index = get_ray_index(kg, thread_index, + QUEUE_ACTIVE_AND_REGENERATED_RAYS, + kernel_split_state.queue_data, + kernel_split_params.queue_size, + 0); + + if(ray_index != QUEUE_EMPTY_SLOT) { + if(IS_STATE(ray_state, ray_index, RAY_ACTIVE)) { + ccl_global PathState *state = &kernel_split_state.path_state[ray_index]; + if(state->bounce > kernel_data.integrator.ao_bounces) { + ASSIGN_RAY_STATE(ray_state, ray_index, RAY_UPDATE_BUFFER); + } + } + } + } + + ray_index = get_ray_index(kg, thread_index, + QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS, + kernel_split_state.queue_data, + kernel_split_params.queue_size, + 0); + + if(ray_index == QUEUE_EMPTY_SLOT) { + return; + } + + ccl_global PathState *state = &kernel_split_state.path_state[ray_index]; + PathRadiance *L = &kernel_split_state.path_radiance[ray_index]; + ccl_global Ray *ray = &kernel_split_state.ray[ray_index]; + ccl_global float3 *throughput = &kernel_split_state.throughput[ray_index]; + ccl_global float *L_transparent = &kernel_split_state.L_transparent[ray_index]; + + if(IS_STATE(ray_state, ray_index, RAY_HIT_BACKGROUND)) { + /* eval background shader if nothing hit */ + if(kernel_data.background.transparent && (state->flag & PATH_RAY_CAMERA)) { + *L_transparent = (*L_transparent) + average((*throughput)); +#ifdef __PASSES__ + if(!(kernel_data.film.pass_flag & PASS_BACKGROUND)) +#endif + ASSIGN_RAY_STATE(ray_state, ray_index, RAY_UPDATE_BUFFER); + } + + if(IS_STATE(ray_state, ray_index, RAY_HIT_BACKGROUND)) { +#ifdef __BACKGROUND__ + /* sample background shader */ + float3 L_background = indirect_background(kg, &kernel_split_state.sd_DL_shadow[ray_index], state, ray); + path_radiance_accum_background(L, state, (*throughput), L_background); +#endif + ASSIGN_RAY_STATE(ray_state, ray_index, RAY_UPDATE_BUFFER); + } + } + + +} + +CCL_NAMESPACE_END diff --git a/intern/cycles/kernel/split/kernel_indirect_subsurface.h b/intern/cycles/kernel/split/kernel_indirect_subsurface.h new file mode 100644 index 00000000000..a56e85abeb9 --- /dev/null +++ b/intern/cycles/kernel/split/kernel_indirect_subsurface.h @@ -0,0 +1,77 @@ +/* + * Copyright 2011-2017 Blender Foundation + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +CCL_NAMESPACE_BEGIN + +ccl_device void kernel_indirect_subsurface(KernelGlobals *kg) +{ + int thread_index = ccl_global_id(1) * ccl_global_size(0) + ccl_global_id(0); + if(thread_index == 0) { + /* We will empty both queues in this kernel. */ + kernel_split_params.queue_index[QUEUE_ACTIVE_AND_REGENERATED_RAYS] = 0; + kernel_split_params.queue_index[QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS] = 0; + } + + int ray_index; + get_ray_index(kg, thread_index, + QUEUE_ACTIVE_AND_REGENERATED_RAYS, + kernel_split_state.queue_data, + kernel_split_params.queue_size, + 1); + ray_index = get_ray_index(kg, thread_index, + QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS, + kernel_split_state.queue_data, + kernel_split_params.queue_size, + 1); + +#ifdef __SUBSURFACE__ + + if(ray_index == QUEUE_EMPTY_SLOT) { + return; + } + + ccl_global char *ray_state = kernel_split_state.ray_state; + ccl_global PathState *state = &kernel_split_state.path_state[ray_index]; + PathRadiance *L = &kernel_split_state.path_radiance[ray_index]; + ccl_global Ray *ray = &kernel_split_state.ray[ray_index]; + ccl_global float3 *throughput = &kernel_split_state.throughput[ray_index]; + + if(IS_STATE(ray_state, ray_index, RAY_UPDATE_BUFFER)) { + ccl_addr_space SubsurfaceIndirectRays *ss_indirect = &kernel_split_state.ss_rays[ray_index]; + kernel_path_subsurface_accum_indirect(ss_indirect, L); + + /* Trace indirect subsurface rays by restarting the loop. this uses less + * stack memory than invoking kernel_path_indirect. + */ + if(ss_indirect->num_rays) { + kernel_path_subsurface_setup_indirect(kg, + ss_indirect, + state, + ray, + L, + throughput); + ASSIGN_RAY_STATE(ray_state, ray_index, RAY_REGENERATED); + } + else { + ASSIGN_RAY_STATE(ray_state, ray_index, RAY_UPDATE_BUFFER); + } + } + +#endif /* __SUBSURFACE__ */ + +} + +CCL_NAMESPACE_END diff --git a/intern/cycles/kernel/split/kernel_lamp_emission.h b/intern/cycles/kernel/split/kernel_lamp_emission.h index 3bd0e361078..c669d79ddcd 100644 --- a/intern/cycles/kernel/split/kernel_lamp_emission.h +++ b/intern/cycles/kernel/split/kernel_lamp_emission.h @@ -14,50 +14,49 @@ * limitations under the License. */ -#include "kernel_split_common.h" +CCL_NAMESPACE_BEGIN -/* Note on kernel_lamp_emission - * This is the 3rd kernel in the ray-tracing logic. This is the second of the - * path-iteration kernels. This kernel takes care of the indirect lamp emission logic. - * This kernel operates on QUEUE_ACTIVE_AND_REGENERATED_RAYS. It processes rays of state RAY_ACTIVE - * and RAY_HIT_BACKGROUND. +/* This kernel operates on QUEUE_ACTIVE_AND_REGENERATED_RAYS. + * It processes rays of state RAY_ACTIVE and RAY_HIT_BACKGROUND. * We will empty QUEUE_ACTIVE_AND_REGENERATED_RAYS queue in this kernel. - * The input/output of the kernel is as follows, - * Throughput_coop ------------------------------------|--- kernel_lamp_emission --|--- PathRadiance_coop - * Ray_coop -------------------------------------------| |--- Queue_data(QUEUE_ACTIVE_AND_REGENERATED_RAYS) - * PathState_coop -------------------------------------| |--- Queue_index(QUEUE_ACTIVE_AND_REGENERATED_RAYS) - * kg (globals) ---------------------------------------| | - * Intersection_coop ----------------------------------| | - * ray_state ------------------------------------------| | - * Queue_data (QUEUE_ACTIVE_AND_REGENERATED_RAYS) -----| | - * Queue_index (QUEUE_ACTIVE_AND_REGENERATED_RAYS) ----| | - * queuesize ------------------------------------------| | - * use_queues_flag ------------------------------------| | - * sw -------------------------------------------------| | - * sh -------------------------------------------------| | */ -ccl_device void kernel_lamp_emission( - KernelGlobals *kg, - ccl_global float3 *throughput_coop, /* Required for lamp emission */ - PathRadiance *PathRadiance_coop, /* Required for lamp emission */ - ccl_global Ray *Ray_coop, /* Required for lamp emission */ - ccl_global PathState *PathState_coop, /* Required for lamp emission */ - Intersection *Intersection_coop, /* Required for lamp emission */ - ccl_global char *ray_state, /* Denotes the state of each ray */ - int sw, int sh, - ccl_global char *use_queues_flag, /* Used to decide if this kernel should use - * queues to fetch ray index - */ - int ray_index) +ccl_device void kernel_lamp_emission(KernelGlobals *kg) { - if(IS_STATE(ray_state, ray_index, RAY_ACTIVE) || - IS_STATE(ray_state, ray_index, RAY_HIT_BACKGROUND)) +#ifndef __VOLUME__ + /* We will empty this queue in this kernel. */ + if(ccl_global_id(0) == 0 && ccl_global_id(1) == 0) { + kernel_split_params.queue_index[QUEUE_ACTIVE_AND_REGENERATED_RAYS] = 0; + } +#endif + /* Fetch use_queues_flag. */ + char local_use_queues_flag = *kernel_split_params.use_queues_flag; + ccl_barrier(CCL_LOCAL_MEM_FENCE); + + int ray_index = ccl_global_id(1) * ccl_global_size(0) + ccl_global_id(0); + if(local_use_queues_flag) { + ray_index = get_ray_index(kg, ray_index, + QUEUE_ACTIVE_AND_REGENERATED_RAYS, + kernel_split_state.queue_data, + kernel_split_params.queue_size, +#ifndef __VOLUME__ + 1 +#else + 0 +#endif + ); + if(ray_index == QUEUE_EMPTY_SLOT) { + return; + } + } + + if(IS_STATE(kernel_split_state.ray_state, ray_index, RAY_ACTIVE) || + IS_STATE(kernel_split_state.ray_state, ray_index, RAY_HIT_BACKGROUND)) { - PathRadiance *L = &PathRadiance_coop[ray_index]; - ccl_global PathState *state = &PathState_coop[ray_index]; + PathRadiance *L = &kernel_split_state.path_radiance[ray_index]; + ccl_global PathState *state = &kernel_split_state.path_state[ray_index]; - float3 throughput = throughput_coop[ray_index]; - Ray ray = Ray_coop[ray_index]; + float3 throughput = kernel_split_state.throughput[ray_index]; + Ray ray = kernel_split_state.ray[ray_index]; #ifdef __LAMP_MIS__ if(kernel_data.integrator.use_lamp_mis && !(state->flag & PATH_RAY_CAMERA)) { @@ -65,7 +64,7 @@ ccl_device void kernel_lamp_emission( Ray light_ray; light_ray.P = ray.P - state->ray_t*ray.D; - state->ray_t += Intersection_coop[ray_index].t; + state->ray_t += kernel_split_state.isect[ray_index].t; light_ray.D = ray.D; light_ray.t = state->ray_t; light_ray.time = ray.time; @@ -74,10 +73,13 @@ ccl_device void kernel_lamp_emission( /* intersect with lamp */ float3 emission; - if(indirect_lamp_emission(kg, kg->sd_input, state, &light_ray, &emission)) { + if(indirect_lamp_emission(kg, &kernel_split_state.sd_DL_shadow[ray_index], state, &light_ray, &emission)) { path_radiance_accum_emission(L, throughput, emission, state->bounce); } } #endif /* __LAMP_MIS__ */ } } + +CCL_NAMESPACE_END + diff --git a/intern/cycles/kernel/split/kernel_next_iteration_setup.h b/intern/cycles/kernel/split/kernel_next_iteration_setup.h index 816f3a6fbff..1bebc16e25b 100644 --- a/intern/cycles/kernel/split/kernel_next_iteration_setup.h +++ b/intern/cycles/kernel/split/kernel_next_iteration_setup.h @@ -14,128 +14,163 @@ * limitations under the License. */ -#include "kernel_split_common.h" +CCL_NAMESPACE_BEGIN -/* Note on kernel_setup_next_iteration kernel. - * This is the tenth kernel in the ray tracing logic. This is the ninth - * of the path iteration kernels. This kernel takes care of setting up - * Ray for the next iteration of path-iteration and accumulating radiance - * corresponding to AO and direct-lighting +/*This kernel takes care of setting up ray for the next iteration of + * path-iteration and accumulating radiance corresponding to AO and + * direct-lighting * - * Ray state of rays that are terminated in this kernel are changed to RAY_UPDATE_BUFFER + * Ray state of rays that are terminated in this kernel are changed + * to RAY_UPDATE_BUFFER. * - * The input and output are as follows, + * Note on queues: + * This kernel fetches rays from the queue QUEUE_ACTIVE_AND_REGENERATED_RAYS + * and processes only the rays of state RAY_ACTIVE. + * There are different points in this kernel where a ray may terminate and + * reach RAY_UPDATE_BUFF state. These rays are enqueued into + * QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS queue. These rays will still be present + * in QUEUE_ACTIVE_AND_REGENERATED_RAYS queue, but since their ray-state has + * been changed to RAY_UPDATE_BUFF, there is no problem. * - * rng_coop ---------------------------------------------|--- kernel_next_iteration_setup -|--- Queue_index (QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS) - * throughput_coop --------------------------------------| |--- Queue_data (QUEUE_HITBF_BUFF_UPDATE_TOREGEN_RAYS) - * PathRadiance_coop ------------------------------------| |--- throughput_coop - * PathState_coop ---------------------------------------| |--- PathRadiance_coop - * sd ---------------------------------------------------| |--- PathState_coop - * ray_state --------------------------------------------| |--- ray_state - * Queue_data (QUEUE_ACTIVE_AND_REGENERATD_RAYS) --------| |--- Ray_coop - * Queue_index (QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS) ---| |--- use_queues_flag - * Ray_coop ---------------------------------------------| | - * kg (globals) -----------------------------------------| | - * LightRay_dl_coop -------------------------------------| - * ISLamp_coop ------------------------------------------| - * BSDFEval_coop ----------------------------------------| - * LightRay_ao_coop -------------------------------------| - * AOBSDF_coop ------------------------------------------| - * AOAlpha_coop -----------------------------------------| - * - * Note on queues, - * This kernel fetches rays from the queue QUEUE_ACTIVE_AND_REGENERATED_RAYS and processes only - * the rays of state RAY_ACTIVE. - * There are different points in this kernel where a ray may terminate and reach RAY_UPDATE_BUFF - * state. These rays are enqueued into QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS queue. These rays will - * still be present in QUEUE_ACTIVE_AND_REGENERATED_RAYS queue, but since their ray-state has been - * changed to RAY_UPDATE_BUFF, there is no problem. - * - * State of queues when this kernel is called : + * State of queues when this kernel is called: * At entry, - * QUEUE_ACTIVE_AND_REGENERATED_RAYS will be filled with RAY_ACTIVE, RAY_REGENERATED, RAY_UPDATE_BUFFER rays. - * QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS will be filled with RAY_TO_REGENERATE and RAY_UPDATE_BUFFER rays + * - QUEUE_ACTIVE_AND_REGENERATED_RAYS will be filled with RAY_ACTIVE, + * RAY_REGENERATED, RAY_UPDATE_BUFFER rays. + * - QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS will be filled with + * RAY_TO_REGENERATE and RAY_UPDATE_BUFFER rays. * At exit, - * QUEUE_ACTIVE_AND_REGENERATED_RAYS will be filled with RAY_ACTIVE, RAY_REGENERATED and more RAY_UPDATE_BUFFER rays. - * QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS will be filled with RAY_TO_REGENERATE and more RAY_UPDATE_BUFFER rays + * - QUEUE_ACTIVE_AND_REGENERATED_RAYS will be filled with RAY_ACTIVE, + * RAY_REGENERATED and more RAY_UPDATE_BUFFER rays. + * - QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS will be filled with + * RAY_TO_REGENERATE and more RAY_UPDATE_BUFFER rays. */ -ccl_device char kernel_next_iteration_setup( - KernelGlobals *kg, - ShaderData *sd, /* Required for setting up ray for next iteration */ - ccl_global uint *rng_coop, /* Required for setting up ray for next iteration */ - ccl_global float3 *throughput_coop, /* Required for setting up ray for next iteration */ - PathRadiance *PathRadiance_coop, /* Required for setting up ray for next iteration */ - ccl_global Ray *Ray_coop, /* Required for setting up ray for next iteration */ - ccl_global PathState *PathState_coop, /* Required for setting up ray for next iteration */ - ccl_global Ray *LightRay_dl_coop, /* Required for radiance update - direct lighting */ - ccl_global int *ISLamp_coop, /* Required for radiance update - direct lighting */ - ccl_global BsdfEval *BSDFEval_coop, /* Required for radiance update - direct lighting */ - ccl_global Ray *LightRay_ao_coop, /* Required for radiance update - AO */ - ccl_global float3 *AOBSDF_coop, /* Required for radiance update - AO */ - ccl_global float3 *AOAlpha_coop, /* Required for radiance update - AO */ - ccl_global char *ray_state, /* Denotes the state of each ray */ - ccl_global char *use_queues_flag, /* flag to decide if scene_intersect kernel should - * use queues to fetch ray index */ - int ray_index) +ccl_device void kernel_next_iteration_setup(KernelGlobals *kg, + ccl_local_param unsigned int *local_queue_atomics) { + if(ccl_local_id(0) == 0 && ccl_local_id(1) == 0) { + *local_queue_atomics = 0; + } + ccl_barrier(CCL_LOCAL_MEM_FENCE); + + if(ccl_global_id(0) == 0 && ccl_global_id(1) == 0) { + /* If we are here, then it means that scene-intersect kernel + * has already been executed atleast once. From the next time, + * scene-intersect kernel may operate on queues to fetch ray index + */ + *kernel_split_params.use_queues_flag = 1; + + /* Mark queue indices of QUEUE_SHADOW_RAY_CAST_AO_RAYS and + * QUEUE_SHADOW_RAY_CAST_DL_RAYS queues that were made empty during the + * previous kernel. + */ + kernel_split_params.queue_index[QUEUE_SHADOW_RAY_CAST_AO_RAYS] = 0; + kernel_split_params.queue_index[QUEUE_SHADOW_RAY_CAST_DL_RAYS] = 0; + } + char enqueue_flag = 0; + int ray_index = ccl_global_id(1) * ccl_global_size(0) + ccl_global_id(0); + ray_index = get_ray_index(kg, ray_index, + QUEUE_ACTIVE_AND_REGENERATED_RAYS, + kernel_split_state.queue_data, + kernel_split_params.queue_size, + 0); + +#ifdef __COMPUTE_DEVICE_GPU__ + /* If we are executing on a GPU device, we exit all threads that are not + * required. + * + * If we are executing on a CPU device, then we need to keep all threads + * active since we have barrier() calls later in the kernel. CPU devices, + * expect all threads to execute barrier statement. + */ + if(ray_index == QUEUE_EMPTY_SLOT) { + return; + } +#endif + +#ifndef __COMPUTE_DEVICE_GPU__ + if(ray_index != QUEUE_EMPTY_SLOT) { +#endif /* Load ShaderData structure. */ PathRadiance *L = NULL; ccl_global PathState *state = NULL; + ccl_global char *ray_state = kernel_split_state.ray_state; /* Path radiance update for AO/Direct_lighting's shadow blocked. */ if(IS_FLAG(ray_state, ray_index, RAY_SHADOW_RAY_CAST_DL) || IS_FLAG(ray_state, ray_index, RAY_SHADOW_RAY_CAST_AO)) { - state = &PathState_coop[ray_index]; - L = &PathRadiance_coop[ray_index]; - float3 _throughput = throughput_coop[ray_index]; + state = &kernel_split_state.path_state[ray_index]; + L = &kernel_split_state.path_radiance[ray_index]; + float3 _throughput = kernel_split_state.throughput[ray_index]; if(IS_FLAG(ray_state, ray_index, RAY_SHADOW_RAY_CAST_AO)) { - float3 shadow = LightRay_ao_coop[ray_index].P; - char update_path_radiance = LightRay_ao_coop[ray_index].t; + float3 shadow = kernel_split_state.ao_light_ray[ray_index].P; + // TODO(mai): investigate correctness here + char update_path_radiance = (char)kernel_split_state.ao_light_ray[ray_index].t; if(update_path_radiance) { path_radiance_accum_ao(L, _throughput, - AOAlpha_coop[ray_index], - AOBSDF_coop[ray_index], + kernel_split_state.ao_alpha[ray_index], + kernel_split_state.ao_bsdf[ray_index], shadow, state->bounce); } + else { + path_radiance_accum_total_ao(L, _throughput, kernel_split_state.ao_bsdf[ray_index]); + } REMOVE_RAY_FLAG(ray_state, ray_index, RAY_SHADOW_RAY_CAST_AO); } if(IS_FLAG(ray_state, ray_index, RAY_SHADOW_RAY_CAST_DL)) { - float3 shadow = LightRay_dl_coop[ray_index].P; - char update_path_radiance = LightRay_dl_coop[ray_index].t; + float3 shadow = kernel_split_state.light_ray[ray_index].P; + // TODO(mai): investigate correctness here + char update_path_radiance = (char)kernel_split_state.light_ray[ray_index].t; + BsdfEval L_light = kernel_split_state.bsdf_eval[ray_index]; if(update_path_radiance) { - BsdfEval L_light = BSDFEval_coop[ray_index]; path_radiance_accum_light(L, _throughput, &L_light, shadow, 1.0f, state->bounce, - ISLamp_coop[ray_index]); + kernel_split_state.is_lamp[ray_index]); + } + else { + path_radiance_accum_total_light(L, _throughput, &L_light); } REMOVE_RAY_FLAG(ray_state, ray_index, RAY_SHADOW_RAY_CAST_DL); } } if(IS_STATE(ray_state, ray_index, RAY_ACTIVE)) { - ccl_global float3 *throughput = &throughput_coop[ray_index]; - ccl_global Ray *ray = &Ray_coop[ray_index]; - ccl_global RNG *rng = &rng_coop[ray_index]; - state = &PathState_coop[ray_index]; - L = &PathRadiance_coop[ray_index]; + ccl_global float3 *throughput = &kernel_split_state.throughput[ray_index]; + ccl_global Ray *ray = &kernel_split_state.ray[ray_index]; + RNG rng = kernel_split_state.rng[ray_index]; + state = &kernel_split_state.path_state[ray_index]; + L = &kernel_split_state.path_radiance[ray_index]; /* Compute direct lighting and next bounce. */ - if(!kernel_path_surface_bounce(kg, rng, sd, throughput, state, L, ray)) { + if(!kernel_path_surface_bounce(kg, &rng, &kernel_split_state.sd[ray_index], throughput, state, L, ray)) { ASSIGN_RAY_STATE(ray_state, ray_index, RAY_UPDATE_BUFFER); enqueue_flag = 1; } + kernel_split_state.rng[ray_index] = rng; } - return enqueue_flag; +#ifndef __COMPUTE_DEVICE_GPU__ + } +#endif + + /* Enqueue RAY_UPDATE_BUFFER rays. */ + enqueue_ray_index_local(ray_index, + QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS, + enqueue_flag, + kernel_split_params.queue_size, + local_queue_atomics, + kernel_split_state.queue_data, + kernel_split_params.queue_index); } + +CCL_NAMESPACE_END diff --git a/intern/cycles/kernel/split/kernel_path_init.h b/intern/cycles/kernel/split/kernel_path_init.h new file mode 100644 index 00000000000..a7ecde7c80d --- /dev/null +++ b/intern/cycles/kernel/split/kernel_path_init.h @@ -0,0 +1,105 @@ +/* + * Copyright 2011-2017 Blender Foundation + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +CCL_NAMESPACE_BEGIN + +/* This kernel initializes structures needed in path-iteration kernels. + * This is the first kernel in ray-tracing logic. + * + * Ray state of rays outside the tile-boundary will be marked RAY_INACTIVE + */ +ccl_device void kernel_path_init(KernelGlobals *kg) { + int ray_index = ccl_global_id(0) + ccl_global_id(1) * ccl_global_size(0); + + /* This is the first assignment to ray_state; + * So we dont use ASSIGN_RAY_STATE macro. + */ + kernel_split_state.ray_state[ray_index] = RAY_ACTIVE; + + unsigned int my_sample; + unsigned int pixel_x; + unsigned int pixel_y; + unsigned int tile_x; + unsigned int tile_y; + + unsigned int work_index = 0; + /* Get work. */ + if(!get_next_work(kg, &work_index, ray_index)) { + /* No more work, mark ray as inactive */ + kernel_split_state.ray_state[ray_index] = RAY_INACTIVE; + + return; + } + + /* Get the sample associated with the work. */ + my_sample = get_work_sample(kg, work_index, ray_index) + kernel_split_params.start_sample; + + /* Get pixel and tile position associated with the work. */ + get_work_pixel_tile_position(kg, &pixel_x, &pixel_y, + &tile_x, &tile_y, + work_index, + ray_index); + kernel_split_state.work_array[ray_index] = work_index; + + ccl_global uint *rng_state = kernel_split_params.rng_state; + rng_state += kernel_split_params.offset + pixel_x + pixel_y*kernel_split_params.stride; + + ccl_global float *buffer = kernel_split_params.buffer; + buffer += (kernel_split_params.offset + pixel_x + pixel_y * kernel_split_params.stride) * kernel_data.film.pass_stride; + + RNG rng = kernel_split_state.rng[ray_index]; + + /* Initialize random numbers and ray. */ + kernel_path_trace_setup(kg, + rng_state, + my_sample, + pixel_x, pixel_y, + &rng, + &kernel_split_state.ray[ray_index]); + + if(kernel_split_state.ray[ray_index].t != 0.0f) { + /* Initialize throughput, L_transparent, Ray, PathState; + * These rays proceed with path-iteration. + */ + kernel_split_state.throughput[ray_index] = make_float3(1.0f, 1.0f, 1.0f); + kernel_split_state.L_transparent[ray_index] = 0.0f; + path_radiance_init(&kernel_split_state.path_radiance[ray_index], kernel_data.film.use_light_pass); + path_state_init(kg, + &kernel_split_state.sd_DL_shadow[ray_index], + &kernel_split_state.path_state[ray_index], + &rng, + my_sample, + &kernel_split_state.ray[ray_index]); +#ifdef __SUBSURFACE__ + kernel_path_subsurface_init_indirect(&kernel_split_state.ss_rays[ray_index]); +#endif + +#ifdef __KERNEL_DEBUG__ + debug_data_init(&kernel_split_state.debug_data[ray_index]); +#endif + } + else { + /* These rays do not participate in path-iteration. */ + float4 L_rad = make_float4(0.0f, 0.0f, 0.0f, 0.0f); + /* Accumulate result in output buffer. */ + kernel_write_pass_float4(buffer, my_sample, L_rad); + path_rng_end(kg, rng_state, kernel_split_state.rng[ray_index]); + ASSIGN_RAY_STATE(kernel_split_state.ray_state, ray_index, RAY_TO_REGENERATE); + } + kernel_split_state.rng[ray_index] = rng; +} + +CCL_NAMESPACE_END diff --git a/intern/cycles/kernel/split/kernel_queue_enqueue.h b/intern/cycles/kernel/split/kernel_queue_enqueue.h new file mode 100644 index 00000000000..e2e841f36d3 --- /dev/null +++ b/intern/cycles/kernel/split/kernel_queue_enqueue.h @@ -0,0 +1,90 @@ +/* + * Copyright 2011-2016 Blender Foundation + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +CCL_NAMESPACE_BEGIN + +/* This kernel enqueues rays of different ray state into their + * appropriate queues: + * + * 1. Rays that have been determined to hit the background from the + * "kernel_scene_intersect" kernel are enqueued in + * QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS; + * 2. Rays that have been determined to be actively participating in pat + * -iteration will be enqueued into QUEUE_ACTIVE_AND_REGENERATED_RAYS. + * + * State of queue during other times this kernel is called: + * At entry, + * - QUEUE_ACTIVE_AND_REGENERATED_RAYS will be empty. + * - QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS will contain RAY_TO_REGENERATE + * and RAY_UPDATE_BUFFER rays. + * At exit, + * - QUEUE_ACTIVE_AND_REGENERATED_RAYS will be filled with RAY_ACTIVE rays. + * - QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS will be filled with + * RAY_TO_REGENERATE, RAY_UPDATE_BUFFER, RAY_HIT_BACKGROUND rays. + */ +ccl_device void kernel_queue_enqueue(KernelGlobals *kg, + ccl_local_param QueueEnqueueLocals *locals) +{ + /* We have only 2 cases (Hit/Not-Hit) */ + int lidx = ccl_local_id(1) * ccl_local_size(0) + ccl_local_id(0); + int ray_index = ccl_global_id(1) * ccl_global_size(0) + ccl_global_id(0); + + if(lidx == 0) { + locals->queue_atomics[0] = 0; + locals->queue_atomics[1] = 0; + } + ccl_barrier(CCL_LOCAL_MEM_FENCE); + + int queue_number = -1; + + if(IS_STATE(kernel_split_state.ray_state, ray_index, RAY_HIT_BACKGROUND) || + IS_STATE(kernel_split_state.ray_state, ray_index, RAY_UPDATE_BUFFER)) { + queue_number = QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS; + } + else if(IS_STATE(kernel_split_state.ray_state, ray_index, RAY_ACTIVE) || + IS_STATE(kernel_split_state.ray_state, ray_index, RAY_REGENERATED)) { + queue_number = QUEUE_ACTIVE_AND_REGENERATED_RAYS; + } + + unsigned int my_lqidx; + if(queue_number != -1) { + my_lqidx = get_local_queue_index(queue_number, locals->queue_atomics); + } + ccl_barrier(CCL_LOCAL_MEM_FENCE); + + if(lidx == 0) { + locals->queue_atomics[QUEUE_ACTIVE_AND_REGENERATED_RAYS] = + get_global_per_queue_offset(QUEUE_ACTIVE_AND_REGENERATED_RAYS, + locals->queue_atomics, + kernel_split_params.queue_index); + locals->queue_atomics[QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS] = + get_global_per_queue_offset(QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS, + locals->queue_atomics, + kernel_split_params.queue_index); + } + ccl_barrier(CCL_LOCAL_MEM_FENCE); + + unsigned int my_gqidx; + if(queue_number != -1) { + my_gqidx = get_global_queue_index(queue_number, + kernel_split_params.queue_size, + my_lqidx, + locals->queue_atomics); + kernel_split_state.queue_data[my_gqidx] = ray_index; + } +} + +CCL_NAMESPACE_END diff --git a/intern/cycles/kernel/split/kernel_scene_intersect.h b/intern/cycles/kernel/split/kernel_scene_intersect.h index 2388580051f..684760eedee 100644 --- a/intern/cycles/kernel/split/kernel_scene_intersect.h +++ b/intern/cycles/kernel/split/kernel_scene_intersect.h @@ -14,81 +14,47 @@ * limitations under the License. */ -#include "kernel_split_common.h" +CCL_NAMESPACE_BEGIN -/* Note on kernel_scene_intersect kernel. - * This is the second kernel in the ray tracing logic. This is the first - * of the path iteration kernels. This kernel takes care of scene_intersect function. +/* This kernel takes care of scene_intersect function. * * This kernel changes the ray_state of RAY_REGENERATED rays to RAY_ACTIVE. * This kernel processes rays of ray state RAY_ACTIVE - * This kernel determines the rays that have hit the background and changes their ray state to RAY_HIT_BACKGROUND. - * - * The input and output are as follows, - * - * Ray_coop ---------------------------------------|--------- kernel_scene_intersect----------|--- PathState - * PathState_coop ---------------------------------| |--- Intersection - * ray_state --------------------------------------| |--- ray_state - * use_queues_flag --------------------------------| | - * QueueData(QUEUE_ACTIVE_AND_REGENERATED_RAYS) ---| | - * kg (globals) -----------------------------------| | - * rng_coop ---------------------------------------| | - * sw ---------------------------------------------| | - * sh ---------------------------------------------| | - * queuesize --------------------------------------| | - * - * Note on Queues : - * Ideally we would want kernel_scene_intersect to work on queues. - * But during the very first time, the queues will be empty and hence we perform a direct mapping - * between ray-index and thread-index; From the next time onward, the queue will be filled and - * we may start operating on queues. - * - * State of queue during the first time this kernel is called : - * QUEUE_ACTIVE_AND_REGENERATED_RAYS and QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS will be empty.before and after this kernel - * - * State of queues during other times this kernel is called : - * At entry, - * QUEUE_ACTIVE_AND_REGENERATED_RAYS will have a mix of RAY_ACTIVE, RAY_UPDATE_BUFFER and RAY_REGENERATED rays; - * QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS will be filled with RAY_TO_REGENERATE and RAY_UPDATE_BUFFER rays ; - * (The rays that are in the state RAY_UPDATE_BUFFER in both the queues are actually the same rays; These - * are the rays that were in RAY_ACTIVE state during the initial enqueue but on further processing - * , by different kernels, have turned into RAY_UPDATE_BUFFER rays. Since all kernel, even after fetching from - * QUEUE_ACTIVE_AND_REGENERATED_RAYS, proceed further based on ray state information, RAY_UPDATE_BUFFER rays - * being present in QUEUE_ACTIVE_AND_REGENERATED_RAYS does not cause any logical issues) - * At exit, - * QUEUE_ACTIVE_AND_REGENERATED_RAYS - All RAY_REGENERATED rays will have been converted to RAY_ACTIVE and - * Some rays in QUEUE_ACTIVE_AND_REGENERATED_RAYS queue will move to state RAY_HIT_BACKGROUND - * QUEUE_HITBF_BUFF_UPDATE_TOREGEN_RAYS - no change + * This kernel determines the rays that have hit the background and changes + * their ray state to RAY_HIT_BACKGROUND. */ - -ccl_device void kernel_scene_intersect( - KernelGlobals *kg, - ccl_global uint *rng_coop, - ccl_global Ray *Ray_coop, /* Required for scene_intersect */ - ccl_global PathState *PathState_coop, /* Required for scene_intersect */ - Intersection *Intersection_coop, /* Required for scene_intersect */ - ccl_global char *ray_state, /* Denotes the state of each ray */ - int sw, int sh, - ccl_global char *use_queues_flag, /* used to decide if this kernel should use - * queues to fetch ray index */ -#ifdef __KERNEL_DEBUG__ - DebugData *debugdata_coop, -#endif - int ray_index) +ccl_device void kernel_scene_intersect(KernelGlobals *kg) { + /* Fetch use_queues_flag */ + char local_use_queues_flag = *kernel_split_params.use_queues_flag; + ccl_barrier(CCL_LOCAL_MEM_FENCE); + + int ray_index = ccl_global_id(1) * ccl_global_size(0) + ccl_global_id(0); + if(local_use_queues_flag) { + ray_index = get_ray_index(kg, ray_index, + QUEUE_ACTIVE_AND_REGENERATED_RAYS, + kernel_split_state.queue_data, + kernel_split_params.queue_size, + 0); + + if(ray_index == QUEUE_EMPTY_SLOT) { + return; + } + } + /* All regenerated rays become active here */ - if(IS_STATE(ray_state, ray_index, RAY_REGENERATED)) - ASSIGN_RAY_STATE(ray_state, ray_index, RAY_ACTIVE); + if(IS_STATE(kernel_split_state.ray_state, ray_index, RAY_REGENERATED)) + ASSIGN_RAY_STATE(kernel_split_state.ray_state, ray_index, RAY_ACTIVE); - if(!IS_STATE(ray_state, ray_index, RAY_ACTIVE)) + if(!IS_STATE(kernel_split_state.ray_state, ray_index, RAY_ACTIVE)) return; #ifdef __KERNEL_DEBUG__ - DebugData *debug_data = &debugdata_coop[ray_index]; + DebugData *debug_data = &kernel_split_state.debug_data[ray_index]; #endif - Intersection *isect = &Intersection_coop[ray_index]; - PathState state = PathState_coop[ray_index]; - Ray ray = Ray_coop[ray_index]; + Intersection isect; + PathState state = kernel_split_state.path_state[ray_index]; + Ray ray = kernel_split_state.ray[ray_index]; /* intersect scene */ uint visibility = path_state_ray_visibility(kg, &state); @@ -96,7 +62,7 @@ ccl_device void kernel_scene_intersect( #ifdef __HAIR__ float difl = 0.0f, extmax = 0.0f; uint lcg_state = 0; - RNG rng = rng_coop[ray_index]; + RNG rng = kernel_split_state.rng[ray_index]; if(kernel_data.bvh.have_curves) { if((kernel_data.cam.resolution == 1) && (state.flag & PATH_RAY_CAMERA)) { @@ -106,19 +72,25 @@ ccl_device void kernel_scene_intersect( } extmax = kernel_data.curve.maximum_width; - lcg_state = lcg_state_init(&rng, &state, 0x51633e2d); + lcg_state = lcg_state_init(&rng, state.rng_offset, state.sample, 0x51633e2d); + } + + if(state.bounce > kernel_data.integrator.ao_bounces) { + visibility = PATH_RAY_SHADOW; + ray.t = kernel_data.background.ao_distance; } - bool hit = scene_intersect(kg, ray, visibility, isect, &lcg_state, difl, extmax); + bool hit = scene_intersect(kg, ray, visibility, &isect, &lcg_state, difl, extmax); #else - bool hit = scene_intersect(kg, ray, visibility, isect, NULL, 0.0f, 0.0f); + bool hit = scene_intersect(kg, ray, visibility, &isect, NULL, 0.0f, 0.0f); #endif + kernel_split_state.isect[ray_index] = isect; #ifdef __KERNEL_DEBUG__ if(state.flag & PATH_RAY_CAMERA) { - debug_data->num_bvh_traversed_nodes += isect->num_traversed_nodes; - debug_data->num_bvh_traversed_instances += isect->num_traversed_instances; - debug_data->num_bvh_intersections += isect->num_intersections; + debug_data->num_bvh_traversed_nodes += isect.num_traversed_nodes; + debug_data->num_bvh_traversed_instances += isect.num_traversed_instances; + debug_data->num_bvh_intersections += isect.num_intersections; } debug_data->num_ray_bounces++; #endif @@ -128,6 +100,8 @@ ccl_device void kernel_scene_intersect( * These rays undergo special processing in the * background_bufferUpdate kernel. */ - ASSIGN_RAY_STATE(ray_state, ray_index, RAY_HIT_BACKGROUND); + ASSIGN_RAY_STATE(kernel_split_state.ray_state, ray_index, RAY_HIT_BACKGROUND); } } + +CCL_NAMESPACE_END diff --git a/intern/cycles/kernel/split/kernel_shader_eval.h b/intern/cycles/kernel/split/kernel_shader_eval.h index cef64bf5f36..0f1696e34a0 100644 --- a/intern/cycles/kernel/split/kernel_shader_eval.h +++ b/intern/cycles/kernel/split/kernel_shader_eval.h @@ -14,57 +14,58 @@ * limitations under the License. */ -#include "kernel_split_common.h" +CCL_NAMESPACE_BEGIN -/* Note on kernel_shader_eval kernel - * This kernel is the 5th kernel in the ray tracing logic. This is - * the 4rd kernel in path iteration. This kernel sets up the ShaderData - * structure from the values computed by the previous kernels. It also identifies - * the rays of state RAY_TO_REGENERATE and enqueues them in QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS queue. +/* This kernel sets up the ShaderData structure from the values computed + * by the previous kernels. * - * The input and output of the kernel is as follows, - * rng_coop -------------------------------------------|--- kernel_shader_eval --|--- sd - * Ray_coop -------------------------------------------| |--- Queue_data (QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS) - * PathState_coop -------------------------------------| |--- Queue_index (QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS) - * Intersection_coop ----------------------------------| | - * Queue_data (QUEUE_ACTIVE_AND_REGENERATD_RAYS)-------| | - * Queue_index(QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS)---| | - * ray_state ------------------------------------------| | - * kg (globals) ---------------------------------------| | - * queuesize ------------------------------------------| | - * - * Note on Queues : - * This kernel reads from the QUEUE_ACTIVE_AND_REGENERATED_RAYS queue and processes - * only the rays of state RAY_ACTIVE; - * State of queues when this kernel is called, - * at entry, - * QUEUE_ACTIVE_AND_REGENERATED_RAYS will be filled with RAY_ACTIVE and RAY_REGENERATED rays - * QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS will be empty. - * at exit, - * QUEUE_ACTIVE_AND_REGENERATED_RAYS will be filled with RAY_ACTIVE and RAY_REGENERATED rays - * QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS will be filled with RAY_TO_REGENERATE rays + * It also identifies the rays of state RAY_TO_REGENERATE and enqueues them + * in QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS queue. */ -ccl_device void kernel_shader_eval( - KernelGlobals *kg, - ShaderData *sd, /* Output ShaderData structure to be filled */ - ccl_global uint *rng_coop, /* Required for rbsdf calculation */ - ccl_global Ray *Ray_coop, /* Required for setting up shader from ray */ - ccl_global PathState *PathState_coop, /* Required for all functions in this kernel */ - Intersection *Intersection_coop, /* Required for setting up shader from ray */ - ccl_global char *ray_state, /* Denotes the state of each ray */ - int ray_index) +ccl_device void kernel_shader_eval(KernelGlobals *kg, + ccl_local_param unsigned int *local_queue_atomics) { - if(IS_STATE(ray_state, ray_index, RAY_ACTIVE)) { - Intersection *isect = &Intersection_coop[ray_index]; - ccl_global uint *rng = &rng_coop[ray_index]; - ccl_global PathState *state = &PathState_coop[ray_index]; - Ray ray = Ray_coop[ray_index]; + /* Enqeueue RAY_TO_REGENERATE rays into QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS queue. */ + if(ccl_local_id(0) == 0 && ccl_local_id(1) == 0) { + *local_queue_atomics = 0; + } + ccl_barrier(CCL_LOCAL_MEM_FENCE); + + int ray_index = ccl_global_id(1) * ccl_global_size(0) + ccl_global_id(0); + ray_index = get_ray_index(kg, ray_index, + QUEUE_ACTIVE_AND_REGENERATED_RAYS, + kernel_split_state.queue_data, + kernel_split_params.queue_size, + 0); + + char enqueue_flag = 0; + if((ray_index != QUEUE_EMPTY_SLOT) && IS_STATE(kernel_split_state.ray_state, ray_index, RAY_TO_REGENERATE)) { + enqueue_flag = 1; + } + + enqueue_ray_index_local(ray_index, + QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS, + enqueue_flag, + kernel_split_params.queue_size, + local_queue_atomics, + kernel_split_state.queue_data, + kernel_split_params.queue_index); + + /* Continue on with shader evaluation. */ + if((ray_index != QUEUE_EMPTY_SLOT) && IS_STATE(kernel_split_state.ray_state, ray_index, RAY_ACTIVE)) { + Intersection isect = kernel_split_state.isect[ray_index]; + RNG rng = kernel_split_state.rng[ray_index]; + ccl_global PathState *state = &kernel_split_state.path_state[ray_index]; + Ray ray = kernel_split_state.ray[ray_index]; shader_setup_from_ray(kg, - sd, - isect, + &kernel_split_state.sd[ray_index], + &isect, &ray); - float rbsdf = path_state_rng_1D_for_decision(kg, rng, state, PRNG_BSDF); - shader_eval_surface(kg, sd, rng, state, rbsdf, state->flag, SHADER_CONTEXT_MAIN); + float rbsdf = path_state_rng_1D_for_decision(kg, &rng, state, PRNG_BSDF); + shader_eval_surface(kg, &kernel_split_state.sd[ray_index], &rng, state, rbsdf, state->flag, SHADER_CONTEXT_MAIN); + kernel_split_state.rng[ray_index] = rng; } } + +CCL_NAMESPACE_END diff --git a/intern/cycles/kernel/split/kernel_shadow_blocked.h b/intern/cycles/kernel/split/kernel_shadow_blocked.h deleted file mode 100644 index 6153af47f96..00000000000 --- a/intern/cycles/kernel/split/kernel_shadow_blocked.h +++ /dev/null @@ -1,85 +0,0 @@ -/* - * Copyright 2011-2015 Blender Foundation - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -#include "kernel_split_common.h" - -/* Note on kernel_shadow_blocked kernel. - * This is the ninth kernel in the ray tracing logic. This is the eighth - * of the path iteration kernels. This kernel takes care of "shadow ray cast" - * logic of the direct lighting and AO part of ray tracing. - * - * The input and output are as follows, - * - * PathState_coop ----------------------------------|--- kernel_shadow_blocked --| - * LightRay_dl_coop --------------------------------| |--- LightRay_dl_coop - * LightRay_ao_coop --------------------------------| |--- LightRay_ao_coop - * ray_state ---------------------------------------| |--- ray_state - * Queue_data(QUEUE_SHADOW_RAY_CAST_AO_RAYS & | |--- Queue_data (QUEUE_SHADOW_RAY_CAST_AO_RAYS & QUEUE_SHADOW_RAY_CAST_AO_RAYS) - QUEUE_SHADOW_RAY_CAST_DL_RAYS) -------| | - * Queue_index(QUEUE_SHADOW_RAY_CAST_AO_RAYS& - QUEUE_SHADOW_RAY_CAST_DL_RAYS) -------| | - * kg (globals) ------------------------------------| | - * queuesize ---------------------------------------| | - * - * Note on sd_shadow : sd_shadow is neither input nor output to this kernel. sd_shadow is filled and consumed in this kernel itself. - * Note on queues : - * The kernel fetches from QUEUE_SHADOW_RAY_CAST_AO_RAYS and QUEUE_SHADOW_RAY_CAST_DL_RAYS queues. We will empty - * these queues this kernel. - * State of queues when this kernel is called : - * state of queues QUEUE_ACTIVE_AND_REGENERATED_RAYS and QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS will be same - * before and after this kernel call. - * QUEUE_SHADOW_RAY_CAST_AO_RAYS & QUEUE_SHADOW_RAY_CAST_DL_RAYS will be filled with rays marked with flags RAY_SHADOW_RAY_CAST_AO - * and RAY_SHADOW_RAY_CAST_DL respectively, during kernel entry. - * QUEUE_SHADOW_RAY_CAST_AO_RAYS and QUEUE_SHADOW_RAY_CAST_DL_RAYS will be empty at kernel exit. - */ -ccl_device void kernel_shadow_blocked( - KernelGlobals *kg, - ccl_global PathState *PathState_coop, /* Required for shadow blocked */ - ccl_global Ray *LightRay_dl_coop, /* Required for direct lighting's shadow blocked */ - ccl_global Ray *LightRay_ao_coop, /* Required for AO's shadow blocked */ - ccl_global char *ray_state, - char shadow_blocked_type, - int ray_index) -{ - /* Flag determining if we need to update L. */ - char update_path_radiance = 0; - - if(IS_FLAG(ray_state, ray_index, RAY_SHADOW_RAY_CAST_DL) || - IS_FLAG(ray_state, ray_index, RAY_SHADOW_RAY_CAST_AO)) - { - ccl_global PathState *state = &PathState_coop[ray_index]; - ccl_global Ray *light_ray_dl_global = &LightRay_dl_coop[ray_index]; - ccl_global Ray *light_ray_ao_global = &LightRay_ao_coop[ray_index]; - - ccl_global Ray *light_ray_global = - shadow_blocked_type == RAY_SHADOW_RAY_CAST_AO - ? light_ray_ao_global - : light_ray_dl_global; - - float3 shadow; - update_path_radiance = !(shadow_blocked(kg, - kg->sd_input, - state, - light_ray_global, - &shadow)); - - /* We use light_ray_global's P and t to store shadow and - * update_path_radiance. - */ - light_ray_global->P = shadow; - light_ray_global->t = update_path_radiance; - } -} diff --git a/intern/cycles/kernel/split/kernel_shadow_blocked_ao.h b/intern/cycles/kernel/split/kernel_shadow_blocked_ao.h new file mode 100644 index 00000000000..4243e18de72 --- /dev/null +++ b/intern/cycles/kernel/split/kernel_shadow_blocked_ao.h @@ -0,0 +1,60 @@ +/* + * Copyright 2011-2015 Blender Foundation + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +CCL_NAMESPACE_BEGIN + +/* Shadow ray cast for AO. */ +ccl_device void kernel_shadow_blocked_ao(KernelGlobals *kg) +{ + unsigned int ao_queue_length = kernel_split_params.queue_index[QUEUE_SHADOW_RAY_CAST_AO_RAYS]; + ccl_barrier(CCL_LOCAL_MEM_FENCE); + + int ray_index = QUEUE_EMPTY_SLOT; + int thread_index = ccl_global_id(1) * ccl_global_size(0) + ccl_global_id(0); + if(thread_index < ao_queue_length) { + ray_index = get_ray_index(kg, thread_index, QUEUE_SHADOW_RAY_CAST_AO_RAYS, + kernel_split_state.queue_data, kernel_split_params.queue_size, 1); + } + + if(ray_index == QUEUE_EMPTY_SLOT) + return; + + /* Flag determining if we need to update L. */ + char update_path_radiance = 0; + + if(IS_FLAG(kernel_split_state.ray_state, ray_index, RAY_SHADOW_RAY_CAST_AO)) { + ccl_global PathState *state = &kernel_split_state.path_state[ray_index]; + ccl_global Ray *light_ray_global = &kernel_split_state.ao_light_ray[ray_index]; + + float3 shadow; + Ray ray = *light_ray_global; + update_path_radiance = !(shadow_blocked(kg, + &kernel_split_state.sd_DL_shadow[ray_index], + state, + &ray, + &shadow)); + + *light_ray_global = ray; + /* We use light_ray_global's P and t to store shadow and + * update_path_radiance. + */ + light_ray_global->P = shadow; + light_ray_global->t = update_path_radiance; + } +} + +CCL_NAMESPACE_END + diff --git a/intern/cycles/kernel/split/kernel_shadow_blocked_dl.h b/intern/cycles/kernel/split/kernel_shadow_blocked_dl.h new file mode 100644 index 00000000000..bb8f0157965 --- /dev/null +++ b/intern/cycles/kernel/split/kernel_shadow_blocked_dl.h @@ -0,0 +1,59 @@ +/* + * Copyright 2011-2015 Blender Foundation + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +CCL_NAMESPACE_BEGIN + +/* Shadow ray cast for direct visible light. */ +ccl_device void kernel_shadow_blocked_dl(KernelGlobals *kg) +{ + unsigned int dl_queue_length = kernel_split_params.queue_index[QUEUE_SHADOW_RAY_CAST_DL_RAYS]; + ccl_barrier(CCL_LOCAL_MEM_FENCE); + + int ray_index = QUEUE_EMPTY_SLOT; + int thread_index = ccl_global_id(1) * ccl_global_size(0) + ccl_global_id(0); + if(thread_index < dl_queue_length) { + ray_index = get_ray_index(kg, thread_index, QUEUE_SHADOW_RAY_CAST_DL_RAYS, + kernel_split_state.queue_data, kernel_split_params.queue_size, 1); + } + + if(ray_index == QUEUE_EMPTY_SLOT) + return; + + /* Flag determining if we need to update L. */ + char update_path_radiance = 0; + + if(IS_FLAG(kernel_split_state.ray_state, ray_index, RAY_SHADOW_RAY_CAST_DL)) { + ccl_global PathState *state = &kernel_split_state.path_state[ray_index]; + ccl_global Ray *light_ray_global = &kernel_split_state.light_ray[ray_index]; + + float3 shadow; + Ray ray = *light_ray_global; + update_path_radiance = !(shadow_blocked(kg, + &kernel_split_state.sd_DL_shadow[ray_index], + state, + &ray, + &shadow)); + + *light_ray_global = ray; + /* We use light_ray_global's P and t to store shadow and + * update_path_radiance. + */ + light_ray_global->P = shadow; + light_ray_global->t = update_path_radiance; + } +} + +CCL_NAMESPACE_END diff --git a/intern/cycles/kernel/split/kernel_split_common.h b/intern/cycles/kernel/split/kernel_split_common.h index 2135ee22b2e..4303ba0a905 100644 --- a/intern/cycles/kernel/split/kernel_split_common.h +++ b/intern/cycles/kernel/split/kernel_split_common.h @@ -17,48 +17,61 @@ #ifndef __KERNEL_SPLIT_H__ #define __KERNEL_SPLIT_H__ -#include "kernel_compat_opencl.h" -#include "kernel_math.h" -#include "kernel_types.h" -#include "kernel_globals.h" -#include "kernel_image_opencl.h" +#include "kernel/kernel_math.h" +#include "kernel/kernel_types.h" -#include "util_atomic.h" +#include "kernel/split/kernel_split_data.h" -#include "kernel_random.h" -#include "kernel_projection.h" -#include "kernel_montecarlo.h" -#include "kernel_differential.h" -#include "kernel_camera.h" +#include "kernel/kernel_globals.h" -#include "geom/geom.h" -#include "bvh/bvh.h" +#ifdef __OSL__ +# include "kernel/osl/osl_shader.h" +#endif + +#ifdef __KERNEL_OPENCL__ +# include "kernel/kernel_image_opencl.h" +#endif +#ifdef __KERNEL_CPU__ +# include "kernel/kernels/cpu/kernel_cpu_image.h" +#endif + +#include "util/util_atomic.h" + +#include "kernel/kernel_random.h" +#include "kernel/kernel_projection.h" +#include "kernel/kernel_montecarlo.h" +#include "kernel/kernel_differential.h" +#include "kernel/kernel_camera.h" + +#include "kernel/geom/geom.h" +#include "kernel/bvh/bvh.h" -#include "kernel_accumulate.h" -#include "kernel_shader.h" -#include "kernel_light.h" -#include "kernel_passes.h" +#include "kernel/kernel_accumulate.h" +#include "kernel/kernel_shader.h" +#include "kernel/kernel_light.h" +#include "kernel/kernel_passes.h" #ifdef __SUBSURFACE__ -#include "kernel_subsurface.h" +# include "kernel/kernel_subsurface.h" #endif #ifdef __VOLUME__ -#include "kernel_volume.h" +# include "kernel/kernel_volume.h" #endif -#include "kernel_path_state.h" -#include "kernel_shadow.h" -#include "kernel_emission.h" -#include "kernel_path_common.h" -#include "kernel_path_surface.h" -#include "kernel_path_volume.h" +#include "kernel/kernel_path_state.h" +#include "kernel/kernel_shadow.h" +#include "kernel/kernel_emission.h" +#include "kernel/kernel_path_common.h" +#include "kernel/kernel_path_surface.h" +#include "kernel/kernel_path_volume.h" +#include "kernel/kernel_path_subsurface.h" #ifdef __KERNEL_DEBUG__ -#include "kernel_debug.h" +# include "kernel/kernel_debug.h" #endif -#include "kernel_queues.h" -#include "kernel_work_stealing.h" +#include "kernel/kernel_queues.h" +#include "kernel/kernel_work_stealing.h" #endif /* __KERNEL_SPLIT_H__ */ diff --git a/intern/cycles/kernel/split/kernel_split_data.h b/intern/cycles/kernel/split/kernel_split_data.h new file mode 100644 index 00000000000..17e6587883a --- /dev/null +++ b/intern/cycles/kernel/split/kernel_split_data.h @@ -0,0 +1,75 @@ +/* + * Copyright 2011-2016 Blender Foundation + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#ifndef __KERNEL_SPLIT_DATA_H__ +#define __KERNEL_SPLIT_DATA_H__ + +#include "kernel/split/kernel_split_data_types.h" +#include "kernel/kernel_globals.h" + +CCL_NAMESPACE_BEGIN + +ccl_device_inline uint64_t split_data_buffer_size(KernelGlobals *kg, size_t num_elements) +{ + (void)kg; /* Unused on CPU. */ + + uint64_t size = 0; +#define SPLIT_DATA_ENTRY(type, name, num) + align_up(num_elements * num * sizeof(type), 16) + size = size SPLIT_DATA_ENTRIES; +#undef SPLIT_DATA_ENTRY + +#ifdef __SUBSURFACE__ + size += align_up(num_elements * sizeof(SubsurfaceIndirectRays), 16); /* ss_rays */ +#endif + +#ifdef __VOLUME__ + size += align_up(2 * num_elements * sizeof(PathState), 16); /* state_shadow */ +#endif + + return size; +} + +ccl_device_inline void split_data_init(KernelGlobals *kg, + ccl_global SplitData *split_data, + size_t num_elements, + ccl_global void *data, + ccl_global char *ray_state) +{ + (void)kg; /* Unused on CPU. */ + + ccl_global char *p = (ccl_global char*)data; + +#define SPLIT_DATA_ENTRY(type, name, num) \ + split_data->name = (type*)p; p += align_up(num_elements * num * sizeof(type), 16); + SPLIT_DATA_ENTRIES; +#undef SPLIT_DATA_ENTRY + +#ifdef __SUBSURFACE__ + split_data->ss_rays = (ccl_global SubsurfaceIndirectRays*)p; + p += align_up(num_elements * sizeof(SubsurfaceIndirectRays), 16); +#endif + +#ifdef __VOLUME__ + split_data->state_shadow = (ccl_global PathState*)p; + p += align_up(2 * num_elements * sizeof(PathState), 16); +#endif + + split_data->ray_state = ray_state; +} + +CCL_NAMESPACE_END + +#endif /* __KERNEL_SPLIT_DATA_H__ */ diff --git a/intern/cycles/kernel/split/kernel_split_data_types.h b/intern/cycles/kernel/split/kernel_split_data_types.h new file mode 100644 index 00000000000..748197b7183 --- /dev/null +++ b/intern/cycles/kernel/split/kernel_split_data_types.h @@ -0,0 +1,127 @@ +/* + * Copyright 2011-2016 Blender Foundation + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#ifndef __KERNEL_SPLIT_DATA_TYPES_H__ +#define __KERNEL_SPLIT_DATA_TYPES_H__ + +CCL_NAMESPACE_BEGIN + +/* parameters used by the split kernels, we use a single struct to avoid passing these to each kernel */ + +typedef struct SplitParams { + int x; + int y; + int w; + int h; + + int offset; + int stride; + + ccl_global uint *rng_state; + + int start_sample; + int end_sample; + + ccl_global unsigned int *work_pools; + unsigned int num_samples; + + ccl_global int *queue_index; + int queue_size; + ccl_global char *use_queues_flag; + + ccl_global float *buffer; +} SplitParams; + +/* Global memory variables [porting]; These memory is used for + * co-operation between different kernels; Data written by one + * kernel will be available to another kernel via this global + * memory. + */ + +/* SPLIT_DATA_ENTRY(type, name, num) */ + +#if defined(WITH_CYCLES_DEBUG) || defined(__KERNEL_DEBUG__) +/* DebugData memory */ +# define SPLIT_DATA_DEBUG_ENTRIES \ + SPLIT_DATA_ENTRY(DebugData, debug_data, 1) +#else +# define SPLIT_DATA_DEBUG_ENTRIES +#endif + +#define SPLIT_DATA_ENTRIES \ + SPLIT_DATA_ENTRY(ccl_global RNG, rng, 1) \ + SPLIT_DATA_ENTRY(ccl_global float3, throughput, 1) \ + SPLIT_DATA_ENTRY(ccl_global float, L_transparent, 1) \ + SPLIT_DATA_ENTRY(PathRadiance, path_radiance, 1) \ + SPLIT_DATA_ENTRY(ccl_global Ray, ray, 1) \ + SPLIT_DATA_ENTRY(ccl_global PathState, path_state, 1) \ + SPLIT_DATA_ENTRY(ccl_global Intersection, isect, 1) \ + SPLIT_DATA_ENTRY(ccl_global float3, ao_alpha, 1) \ + SPLIT_DATA_ENTRY(ccl_global float3, ao_bsdf, 1) \ + SPLIT_DATA_ENTRY(ccl_global Ray, ao_light_ray, 1) \ + SPLIT_DATA_ENTRY(ccl_global BsdfEval, bsdf_eval, 1) \ + SPLIT_DATA_ENTRY(ccl_global int, is_lamp, 1) \ + SPLIT_DATA_ENTRY(ccl_global Ray, light_ray, 1) \ + SPLIT_DATA_ENTRY(ccl_global int, queue_data, (NUM_QUEUES*2)) /* TODO(mai): this is too large? */ \ + SPLIT_DATA_ENTRY(ccl_global uint, work_array, 1) \ + SPLIT_DATA_ENTRY(ShaderData, sd, 1) \ + SPLIT_DATA_ENTRY(ShaderData, sd_DL_shadow, 1) \ + SPLIT_DATA_DEBUG_ENTRIES \ + +/* struct that holds pointers to data in the shared state buffer */ +typedef struct SplitData { +#define SPLIT_DATA_ENTRY(type, name, num) type *name; + SPLIT_DATA_ENTRIES +#undef SPLIT_DATA_ENTRY + +#ifdef __SUBSURFACE__ + ccl_global SubsurfaceIndirectRays *ss_rays; +#endif + +#ifdef __VOLUME__ + ccl_global PathState *state_shadow; +#endif + + /* this is actually in a separate buffer from the rest of the split state data (so it can be read back from + * the host easily) but is still used the same as the other data so we have it here in this struct as well + */ + ccl_global char *ray_state; +} SplitData; + +#ifndef __KERNEL_CUDA__ +# define kernel_split_state (kg->split_data) +# define kernel_split_params (kg->split_param_data) +#else +__device__ SplitData __split_data; +# define kernel_split_state (__split_data) +__device__ SplitParams __split_param_data; +# define kernel_split_params (__split_param_data) +#endif /* __KERNEL_CUDA__ */ + +/* Local storage for queue_enqueue kernel. */ +typedef struct QueueEnqueueLocals { + uint queue_atomics[2]; +} QueueEnqueueLocals; + +/* Local storage for holdout_emission_blurring_pathtermination_ao kernel. */ +typedef struct BackgroundAOLocals { + uint queue_atomics_bg; + uint queue_atomics_ao; +} BackgroundAOLocals; + +CCL_NAMESPACE_END + +#endif /* __KERNEL_SPLIT_DATA_TYPES_H__ */ diff --git a/intern/cycles/kernel/split/kernel_subsurface_scatter.h b/intern/cycles/kernel/split/kernel_subsurface_scatter.h new file mode 100644 index 00000000000..0b4d50c70ee --- /dev/null +++ b/intern/cycles/kernel/split/kernel_subsurface_scatter.h @@ -0,0 +1,100 @@ +/* + * Copyright 2011-2017 Blender Foundation + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +CCL_NAMESPACE_BEGIN + + +ccl_device void kernel_subsurface_scatter(KernelGlobals *kg, + ccl_local_param unsigned int* local_queue_atomics) +{ +#ifdef __SUBSURFACE__ + if(ccl_local_id(0) == 0 && ccl_local_id(1) == 0) { + *local_queue_atomics = 0; + } + ccl_barrier(CCL_LOCAL_MEM_FENCE); + + int ray_index = ccl_global_id(1) * ccl_global_size(0) + ccl_global_id(0); + ray_index = get_ray_index(kg, ray_index, + QUEUE_ACTIVE_AND_REGENERATED_RAYS, + kernel_split_state.queue_data, + kernel_split_params.queue_size, + 0); + +#ifdef __COMPUTE_DEVICE_GPU__ + /* If we are executing on a GPU device, we exit all threads that are not + * required. + * + * If we are executing on a CPU device, then we need to keep all threads + * active since we have barrier() calls later in the kernel. CPU devices, + * expect all threads to execute barrier statement. + */ + if(ray_index == QUEUE_EMPTY_SLOT) { + return; + } +#endif + + char enqueue_flag = 0; + +#ifndef __COMPUTE_DEVICE_GPU__ + if(ray_index != QUEUE_EMPTY_SLOT) { +#endif + + ccl_global char *ray_state = kernel_split_state.ray_state; + ccl_global PathState *state = &kernel_split_state.path_state[ray_index]; + PathRadiance *L = &kernel_split_state.path_radiance[ray_index]; + RNG rng = kernel_split_state.rng[ray_index]; + ccl_global Ray *ray = &kernel_split_state.ray[ray_index]; + ccl_global float3 *throughput = &kernel_split_state.throughput[ray_index]; + ccl_global SubsurfaceIndirectRays *ss_indirect = &kernel_split_state.ss_rays[ray_index]; + ShaderData *sd = &kernel_split_state.sd[ray_index]; + ShaderData *emission_sd = &kernel_split_state.sd_DL_shadow[ray_index]; + + if(IS_STATE(ray_state, ray_index, RAY_ACTIVE)) { + if(sd->flag & SD_BSSRDF) { + if(kernel_path_subsurface_scatter(kg, + sd, + emission_sd, + L, + state, + &rng, + ray, + throughput, + ss_indirect)) { + ASSIGN_RAY_STATE(ray_state, ray_index, RAY_UPDATE_BUFFER); + enqueue_flag = 1; + } + } + kernel_split_state.rng[ray_index] = rng; + } + +#ifndef __COMPUTE_DEVICE_GPU__ + } +#endif + + /* Enqueue RAY_UPDATE_BUFFER rays. */ + enqueue_ray_index_local(ray_index, + QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS, + enqueue_flag, + kernel_split_params.queue_size, + local_queue_atomics, + kernel_split_state.queue_data, + kernel_split_params.queue_index); + +#endif /* __SUBSURFACE__ */ + +} + +CCL_NAMESPACE_END diff --git a/intern/cycles/kernel/split/kernel_sum_all_radiance.h b/intern/cycles/kernel/split/kernel_sum_all_radiance.h deleted file mode 100644 index a21e9b6a0b1..00000000000 --- a/intern/cycles/kernel/split/kernel_sum_all_radiance.h +++ /dev/null @@ -1,59 +0,0 @@ -/* - * Copyright 2011-2015 Blender Foundation - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -#include "../kernel_compat_opencl.h" -#include "../kernel_math.h" -#include "../kernel_types.h" -#include "../kernel_globals.h" - -/* Since we process various samples in parallel; The output radiance of different samples - * are stored in different locations; This kernel combines the output radiance contributed - * by all different samples and stores them in the RenderTile's output buffer. - */ -ccl_device void kernel_sum_all_radiance( - ccl_constant KernelData *data, /* To get pass_stride to offet into buffer */ - ccl_global float *buffer, /* Output buffer of RenderTile */ - ccl_global float *per_sample_output_buffer, /* Radiance contributed by all samples */ - int parallel_samples, int sw, int sh, int stride, - int buffer_offset_x, - int buffer_offset_y, - int buffer_stride, - int start_sample) -{ - int x = get_global_id(0); - int y = get_global_id(1); - - if(x < sw && y < sh) { - buffer += ((buffer_offset_x + x) + (buffer_offset_y + y) * buffer_stride) * (data->film.pass_stride); - per_sample_output_buffer += ((x + y * stride) * parallel_samples) * (data->film.pass_stride); - - int sample_stride = (data->film.pass_stride); - - int sample_iterator = 0; - int pass_stride_iterator = 0; - int num_floats = data->film.pass_stride; - - for(sample_iterator = 0; sample_iterator < parallel_samples; sample_iterator++) { - for(pass_stride_iterator = 0; pass_stride_iterator < num_floats; pass_stride_iterator++) { - *(buffer + pass_stride_iterator) = - (start_sample == 0 && sample_iterator == 0) - ? *(per_sample_output_buffer + pass_stride_iterator) - : *(buffer + pass_stride_iterator) + *(per_sample_output_buffer + pass_stride_iterator); - } - per_sample_output_buffer += sample_stride; - } - } -} diff --git a/intern/cycles/kernel/svm/svm.h b/intern/cycles/kernel/svm/svm.h index 88ec7fe6fcc..d748e76fa80 100644 --- a/intern/cycles/kernel/svm/svm.h +++ b/intern/cycles/kernel/svm/svm.h @@ -39,7 +39,7 @@ * mostly taken care of in the SVM compiler. */ -#include "svm_types.h" +#include "kernel/svm/svm_types.h" CCL_NAMESPACE_BEGIN @@ -139,49 +139,49 @@ CCL_NAMESPACE_END /* Nodes */ -#include "svm_noise.h" +#include "kernel/svm/svm_noise.h" #include "svm_texture.h" -#include "svm_color_util.h" -#include "svm_math_util.h" - -#include "svm_attribute.h" -#include "svm_gradient.h" -#include "svm_blackbody.h" -#include "svm_closure.h" -#include "svm_noisetex.h" -#include "svm_convert.h" -#include "svm_displace.h" -#include "svm_fresnel.h" -#include "svm_wireframe.h" -#include "svm_wavelength.h" -#include "svm_camera.h" -#include "svm_geometry.h" -#include "svm_hsv.h" -#include "svm_image.h" -#include "svm_gamma.h" -#include "svm_brightness.h" -#include "svm_invert.h" -#include "svm_light_path.h" -#include "svm_magic.h" -#include "svm_mapping.h" -#include "svm_normal.h" -#include "svm_wave.h" -#include "svm_math.h" -#include "svm_mix.h" -#include "svm_ramp.h" -#include "svm_sepcomb_hsv.h" -#include "svm_sepcomb_vector.h" -#include "svm_musgrave.h" -#include "svm_sky.h" -#include "svm_tex_coord.h" -#include "svm_value.h" -#include "svm_voronoi.h" -#include "svm_checker.h" -#include "svm_brick.h" -#include "svm_vector_transform.h" -#include "svm_voxel.h" -#include "svm_bump.h" +#include "kernel/svm/svm_color_util.h" +#include "kernel/svm/svm_math_util.h" + +#include "kernel/svm/svm_attribute.h" +#include "kernel/svm/svm_gradient.h" +#include "kernel/svm/svm_blackbody.h" +#include "kernel/svm/svm_closure.h" +#include "kernel/svm/svm_noisetex.h" +#include "kernel/svm/svm_convert.h" +#include "kernel/svm/svm_displace.h" +#include "kernel/svm/svm_fresnel.h" +#include "kernel/svm/svm_wireframe.h" +#include "kernel/svm/svm_wavelength.h" +#include "kernel/svm/svm_camera.h" +#include "kernel/svm/svm_geometry.h" +#include "kernel/svm/svm_hsv.h" +#include "kernel/svm/svm_image.h" +#include "kernel/svm/svm_gamma.h" +#include "kernel/svm/svm_brightness.h" +#include "kernel/svm/svm_invert.h" +#include "kernel/svm/svm_light_path.h" +#include "kernel/svm/svm_magic.h" +#include "kernel/svm/svm_mapping.h" +#include "kernel/svm/svm_normal.h" +#include "kernel/svm/svm_wave.h" +#include "kernel/svm/svm_math.h" +#include "kernel/svm/svm_mix.h" +#include "kernel/svm/svm_ramp.h" +#include "kernel/svm/svm_sepcomb_hsv.h" +#include "kernel/svm/svm_sepcomb_vector.h" +#include "kernel/svm/svm_musgrave.h" +#include "kernel/svm/svm_sky.h" +#include "kernel/svm/svm_tex_coord.h" +#include "kernel/svm/svm_value.h" +#include "kernel/svm/svm_voronoi.h" +#include "kernel/svm/svm_checker.h" +#include "kernel/svm/svm_brick.h" +#include "kernel/svm/svm_vector_transform.h" +#include "kernel/svm/svm_voxel.h" +#include "kernel/svm/svm_bump.h" CCL_NAMESPACE_BEGIN @@ -192,7 +192,7 @@ CCL_NAMESPACE_BEGIN ccl_device_noinline void svm_eval_nodes(KernelGlobals *kg, ShaderData *sd, ccl_addr_space PathState *state, ShaderType type, int path_flag) { float stack[SVM_STACK_SIZE]; - int offset = ccl_fetch(sd, shader) & SHADER_MASK; + int offset = sd->shader & SHADER_MASK; while(1) { uint4 node = read_node(kg, &offset); diff --git a/intern/cycles/kernel/svm/svm_attribute.h b/intern/cycles/kernel/svm/svm_attribute.h index 0e55c99ae97..229a3f20421 100644 --- a/intern/cycles/kernel/svm/svm_attribute.h +++ b/intern/cycles/kernel/svm/svm_attribute.h @@ -27,7 +27,7 @@ ccl_device AttributeDescriptor svm_node_attr_init(KernelGlobals *kg, ShaderData AttributeDescriptor desc; - if(ccl_fetch(sd, object) != OBJECT_NONE) { + if(sd->object != OBJECT_NONE) { desc = find_attribute(kg, sd, node.y); if(desc.offset == ATTR_STD_NOT_FOUND) { desc = attribute_not_found(); diff --git a/intern/cycles/kernel/svm/svm_bump.h b/intern/cycles/kernel/svm/svm_bump.h index 04a8c7b64e5..610d9af9e1f 100644 --- a/intern/cycles/kernel/svm/svm_bump.h +++ b/intern/cycles/kernel/svm/svm_bump.h @@ -21,9 +21,9 @@ CCL_NAMESPACE_BEGIN ccl_device void svm_node_enter_bump_eval(KernelGlobals *kg, ShaderData *sd, float *stack, uint offset) { /* save state */ - stack_store_float3(stack, offset+0, ccl_fetch(sd, P)); - stack_store_float3(stack, offset+3, ccl_fetch(sd, dP).dx); - stack_store_float3(stack, offset+6, ccl_fetch(sd, dP).dy); + stack_store_float3(stack, offset+0, sd->P); + stack_store_float3(stack, offset+3, sd->dP.dx); + stack_store_float3(stack, offset+6, sd->dP.dy); /* set state as if undisplaced */ const AttributeDescriptor desc = find_attribute(kg, sd, ATTR_STD_POSITION_UNDISPLACED); @@ -36,18 +36,18 @@ ccl_device void svm_node_enter_bump_eval(KernelGlobals *kg, ShaderData *sd, floa object_dir_transform(kg, sd, &dPdx); object_dir_transform(kg, sd, &dPdy); - ccl_fetch(sd, P) = P; - ccl_fetch(sd, dP).dx = dPdx; - ccl_fetch(sd, dP).dy = dPdy; + sd->P = P; + sd->dP.dx = dPdx; + sd->dP.dy = dPdy; } } ccl_device void svm_node_leave_bump_eval(KernelGlobals *kg, ShaderData *sd, float *stack, uint offset) { /* restore state */ - ccl_fetch(sd, P) = stack_load_float3(stack, offset+0); - ccl_fetch(sd, dP).dx = stack_load_float3(stack, offset+3); - ccl_fetch(sd, dP).dy = stack_load_float3(stack, offset+6); + sd->P = stack_load_float3(stack, offset+0); + sd->dP.dx = stack_load_float3(stack, offset+3); + sd->dP.dy = stack_load_float3(stack, offset+6); } CCL_NAMESPACE_END diff --git a/intern/cycles/kernel/svm/svm_camera.h b/intern/cycles/kernel/svm/svm_camera.h index 00678a49d70..90249dfd978 100644 --- a/intern/cycles/kernel/svm/svm_camera.h +++ b/intern/cycles/kernel/svm/svm_camera.h @@ -23,7 +23,7 @@ ccl_device void svm_node_camera(KernelGlobals *kg, ShaderData *sd, float *stack, float3 vector; Transform tfm = kernel_data.cam.worldtocamera; - vector = transform_point(&tfm, ccl_fetch(sd, P)); + vector = transform_point(&tfm, sd->P); zdepth = vector.z; distance = len(vector); diff --git a/intern/cycles/kernel/svm/svm_closure.h b/intern/cycles/kernel/svm/svm_closure.h index 024d7d6447a..9a3689a94f4 100644 --- a/intern/cycles/kernel/svm/svm_closure.h +++ b/intern/cycles/kernel/svm/svm_closure.h @@ -25,13 +25,13 @@ ccl_device void svm_node_glass_setup(ShaderData *sd, MicrofacetBsdf *bsdf, int t bsdf->alpha_y = 0.0f; bsdf->alpha_x = 0.0f; bsdf->ior = eta; - ccl_fetch(sd, flag) |= bsdf_refraction_setup(bsdf); + sd->flag |= bsdf_refraction_setup(bsdf); } else { bsdf->alpha_y = 0.0f; bsdf->alpha_x = 0.0f; bsdf->ior = 0.0f; - ccl_fetch(sd, flag) |= bsdf_reflection_setup(bsdf); + sd->flag |= bsdf_reflection_setup(bsdf); } } else if(type == CLOSURE_BSDF_MICROFACET_BECKMANN_GLASS_ID) { @@ -40,9 +40,9 @@ ccl_device void svm_node_glass_setup(ShaderData *sd, MicrofacetBsdf *bsdf, int t bsdf->ior = eta; if(refract) - ccl_fetch(sd, flag) |= bsdf_microfacet_beckmann_refraction_setup(bsdf); + sd->flag |= bsdf_microfacet_beckmann_refraction_setup(bsdf); else - ccl_fetch(sd, flag) |= bsdf_microfacet_beckmann_setup(bsdf); + sd->flag |= bsdf_microfacet_beckmann_setup(bsdf); } else { bsdf->alpha_x = roughness; @@ -50,9 +50,9 @@ ccl_device void svm_node_glass_setup(ShaderData *sd, MicrofacetBsdf *bsdf, int t bsdf->ior = eta; if(refract) - ccl_fetch(sd, flag) |= bsdf_microfacet_ggx_refraction_setup(bsdf); + sd->flag |= bsdf_microfacet_ggx_refraction_setup(bsdf); else - ccl_fetch(sd, flag) |= bsdf_microfacet_ggx_setup(bsdf); + sd->flag |= bsdf_microfacet_ggx_setup(bsdf); } } @@ -70,7 +70,7 @@ ccl_device void svm_node_closure_bsdf(KernelGlobals *kg, ShaderData *sd, float * if(mix_weight == 0.0f) return; - float3 N = stack_valid(data_node.x)? stack_load_float3(stack, data_node.x): ccl_fetch(sd, N); + float3 N = stack_valid(data_node.x)? stack_load_float3(stack, data_node.x): sd->N; float param1 = (stack_valid(param1_offset))? stack_load_float(stack, param1_offset): __uint_as_float(node.z); float param2 = (stack_valid(param2_offset))? stack_load_float(stack, param2_offset): __uint_as_float(node.w); @@ -110,10 +110,10 @@ ccl_device void svm_node_closure_bsdf(KernelGlobals *kg, ShaderData *sd, float * T = rotate_around_axis(T, N, anisotropic_rotation * M_2PI_F); /* calculate ior */ - float ior = (ccl_fetch(sd, flag) & SD_BACKFACING) ? 1.0f / eta : eta; + float ior = (sd->flag & SD_BACKFACING) ? 1.0f / eta : eta; // calculate fresnel for refraction - float cosNO = dot(N, ccl_fetch(sd, I)); + float cosNO = dot(N, sd->I); float fresnel = fresnel_dielectric_cos(cosNO, ior); // calculate weights of the diffuse and specular part @@ -129,7 +129,7 @@ ccl_device void svm_node_closure_bsdf(KernelGlobals *kg, ShaderData *sd, float * // get the additional clearcoat normal and subsurface scattering radius uint4 data_cn_ssr = read_node(kg, offset); - float3 clearcoat_normal = stack_valid(data_cn_ssr.x) ? stack_load_float3(stack, data_cn_ssr.x) : ccl_fetch(sd, N); + float3 clearcoat_normal = stack_valid(data_cn_ssr.x) ? stack_load_float3(stack, data_cn_ssr.x) : sd->N; float3 subsurface_radius = stack_valid(data_cn_ssr.y) ? stack_load_float3(stack, data_cn_ssr.y) : make_float3(1.0f, 1.0f, 1.0f); // get the subsurface color @@ -137,7 +137,7 @@ ccl_device void svm_node_closure_bsdf(KernelGlobals *kg, ShaderData *sd, float * float3 subsurface_color = stack_valid(data_subsurface_color.x) ? stack_load_float3(stack, data_subsurface_color.x) : make_float3(__uint_as_float(data_subsurface_color.y), __uint_as_float(data_subsurface_color.z), __uint_as_float(data_subsurface_color.w)); - float3 weight = ccl_fetch(sd, svm_closure_weight) * mix_weight; + float3 weight = sd->svm_closure_weight * mix_weight; #ifdef __SUBSURFACE__ float3 albedo = subsurface_color * subsurface + base_color * (1.0f - subsurface); @@ -163,7 +163,7 @@ ccl_device void svm_node_closure_bsdf(KernelGlobals *kg, ShaderData *sd, float * bsdf->roughness = roughness; /* setup bsdf */ - ccl_fetch(sd, flag) |= bsdf_principled_diffuse_setup(bsdf); + sd->flag |= bsdf_principled_diffuse_setup(bsdf); } } else if(subsurface > CLOSURE_WEIGHT_CUTOFF && subsurf_sample_weight > CLOSURE_WEIGHT_CUTOFF) { @@ -186,7 +186,7 @@ ccl_device void svm_node_closure_bsdf(KernelGlobals *kg, ShaderData *sd, float * bssrdf->roughness = roughness; /* setup bsdf */ - ccl_fetch(sd, flag) |= bssrdf_setup(bssrdf, (ClosureType)CLOSURE_BSSRDF_PRINCIPLED_ID); + sd->flag |= bssrdf_setup(bssrdf, (ClosureType)CLOSURE_BSSRDF_PRINCIPLED_ID); } bssrdf = bssrdf_alloc(sd, make_float3(0.0f, subsurf_weight.y, 0.0f)); @@ -200,7 +200,7 @@ ccl_device void svm_node_closure_bsdf(KernelGlobals *kg, ShaderData *sd, float * bssrdf->roughness = roughness; /* setup bsdf */ - ccl_fetch(sd, flag) |= bssrdf_setup(bssrdf, (ClosureType)CLOSURE_BSSRDF_PRINCIPLED_ID); + sd->flag |= bssrdf_setup(bssrdf, (ClosureType)CLOSURE_BSSRDF_PRINCIPLED_ID); } bssrdf = bssrdf_alloc(sd, make_float3(0.0f, 0.0f, subsurf_weight.z)); @@ -214,7 +214,7 @@ ccl_device void svm_node_closure_bsdf(KernelGlobals *kg, ShaderData *sd, float * bssrdf->roughness = roughness; /* setup bsdf */ - ccl_fetch(sd, flag) |= bssrdf_setup(bssrdf, (ClosureType)CLOSURE_BSSRDF_PRINCIPLED_ID); + sd->flag |= bssrdf_setup(bssrdf, (ClosureType)CLOSURE_BSSRDF_PRINCIPLED_ID); } } } @@ -230,7 +230,7 @@ ccl_device void svm_node_closure_bsdf(KernelGlobals *kg, ShaderData *sd, float * bsdf->roughness = roughness; /* setup bsdf */ - ccl_fetch(sd, flag) |= bsdf_principled_diffuse_setup(bsdf); + sd->flag |= bsdf_principled_diffuse_setup(bsdf); } } #endif @@ -251,7 +251,7 @@ ccl_device void svm_node_closure_bsdf(KernelGlobals *kg, ShaderData *sd, float * bsdf->N = N; /* setup bsdf */ - ccl_fetch(sd, flag) |= bsdf_principled_sheen_setup(bsdf); + sd->flag |= bsdf_principled_sheen_setup(bsdf); } } @@ -286,9 +286,9 @@ ccl_device void svm_node_closure_bsdf(KernelGlobals *kg, ShaderData *sd, float * /* setup bsdf */ if(distribution == CLOSURE_BSDF_MICROFACET_GGX_GLASS_ID || roughness <= 0.075f) /* use single-scatter GGX */ - ccl_fetch(sd, flag) |= bsdf_microfacet_ggx_aniso_fresnel_setup(bsdf); + sd->flag |= bsdf_microfacet_ggx_aniso_fresnel_setup(bsdf); else /* use multi-scatter GGX */ - ccl_fetch(sd, flag) |= bsdf_microfacet_multi_ggx_aniso_fresnel_setup(bsdf); + sd->flag |= bsdf_microfacet_multi_ggx_aniso_fresnel_setup(bsdf); } } #ifdef __CAUSTICS_TRICKS__ @@ -326,7 +326,7 @@ ccl_device void svm_node_closure_bsdf(KernelGlobals *kg, ShaderData *sd, float * bsdf->extra->cspec0 = cspec0; /* setup bsdf */ - ccl_fetch(sd, flag) |= bsdf_microfacet_ggx_fresnel_setup(bsdf); + sd->flag |= bsdf_microfacet_ggx_fresnel_setup(bsdf); } } @@ -350,7 +350,7 @@ ccl_device void svm_node_closure_bsdf(KernelGlobals *kg, ShaderData *sd, float * bsdf->ior = ior; /* setup bsdf */ - ccl_fetch(sd, flag) |= bsdf_microfacet_ggx_refraction_setup(bsdf); + sd->flag |= bsdf_microfacet_ggx_refraction_setup(bsdf); } } } @@ -371,7 +371,7 @@ ccl_device void svm_node_closure_bsdf(KernelGlobals *kg, ShaderData *sd, float * bsdf->extra->cspec0 = cspec0; /* setup bsdf */ - ccl_fetch(sd, flag) |= bsdf_microfacet_multi_ggx_glass_fresnel_setup(bsdf); + sd->flag |= bsdf_microfacet_multi_ggx_glass_fresnel_setup(bsdf); } } } @@ -399,7 +399,7 @@ ccl_device void svm_node_closure_bsdf(KernelGlobals *kg, ShaderData *sd, float * bsdf->extra->clearcoat = clearcoat; /* setup bsdf */ - ccl_fetch(sd, flag) |= bsdf_microfacet_ggx_clearcoat_setup(bsdf); + sd->flag |= bsdf_microfacet_ggx_clearcoat_setup(bsdf); } } #ifdef __CAUSTICS_TRICKS__ @@ -409,7 +409,7 @@ ccl_device void svm_node_closure_bsdf(KernelGlobals *kg, ShaderData *sd, float * break; } case CLOSURE_BSDF_DIFFUSE_ID: { - float3 weight = ccl_fetch(sd, svm_closure_weight) * mix_weight; + float3 weight = sd->svm_closure_weight * mix_weight; OrenNayarBsdf *bsdf = (OrenNayarBsdf*)bsdf_alloc(sd, sizeof(OrenNayarBsdf), weight); if(bsdf) { @@ -418,31 +418,31 @@ ccl_device void svm_node_closure_bsdf(KernelGlobals *kg, ShaderData *sd, float * float roughness = param1; if(roughness == 0.0f) { - ccl_fetch(sd, flag) |= bsdf_diffuse_setup((DiffuseBsdf*)bsdf); + sd->flag |= bsdf_diffuse_setup((DiffuseBsdf*)bsdf); } else { bsdf->roughness = roughness; - ccl_fetch(sd, flag) |= bsdf_oren_nayar_setup(bsdf); + sd->flag |= bsdf_oren_nayar_setup(bsdf); } } break; } case CLOSURE_BSDF_TRANSLUCENT_ID: { - float3 weight = ccl_fetch(sd, svm_closure_weight) * mix_weight; + float3 weight = sd->svm_closure_weight * mix_weight; DiffuseBsdf *bsdf = (DiffuseBsdf*)bsdf_alloc(sd, sizeof(DiffuseBsdf), weight); if(bsdf) { bsdf->N = N; - ccl_fetch(sd, flag) |= bsdf_translucent_setup(bsdf); + sd->flag |= bsdf_translucent_setup(bsdf); } break; } case CLOSURE_BSDF_TRANSPARENT_ID: { - float3 weight = ccl_fetch(sd, svm_closure_weight) * mix_weight; + float3 weight = sd->svm_closure_weight * mix_weight; ShaderClosure *bsdf = bsdf_alloc(sd, sizeof(ShaderClosure), weight); if(bsdf) { - ccl_fetch(sd, flag) |= bsdf_transparent_setup(bsdf); + sd->flag |= bsdf_transparent_setup(bsdf); } break; } @@ -455,7 +455,7 @@ ccl_device void svm_node_closure_bsdf(KernelGlobals *kg, ShaderData *sd, float * if(!kernel_data.integrator.caustics_reflective && (path_flag & PATH_RAY_DIFFUSE)) break; #endif - float3 weight = ccl_fetch(sd, svm_closure_weight) * mix_weight; + float3 weight = sd->svm_closure_weight * mix_weight; MicrofacetBsdf *bsdf = (MicrofacetBsdf*)bsdf_alloc(sd, sizeof(MicrofacetBsdf), weight); if(bsdf) { @@ -467,21 +467,21 @@ ccl_device void svm_node_closure_bsdf(KernelGlobals *kg, ShaderData *sd, float * /* setup bsdf */ if(type == CLOSURE_BSDF_REFLECTION_ID) - ccl_fetch(sd, flag) |= bsdf_reflection_setup(bsdf); + sd->flag |= bsdf_reflection_setup(bsdf); else if(type == CLOSURE_BSDF_MICROFACET_BECKMANN_ID) - ccl_fetch(sd, flag) |= bsdf_microfacet_beckmann_setup(bsdf); + sd->flag |= bsdf_microfacet_beckmann_setup(bsdf); else if(type == CLOSURE_BSDF_MICROFACET_GGX_ID) - ccl_fetch(sd, flag) |= bsdf_microfacet_ggx_setup(bsdf); + sd->flag |= bsdf_microfacet_ggx_setup(bsdf); else if(type == CLOSURE_BSDF_MICROFACET_MULTI_GGX_ID) { kernel_assert(stack_valid(data_node.z)); bsdf->extra = (MicrofacetExtra*)closure_alloc_extra(sd, sizeof(MicrofacetExtra)); if(bsdf->extra) { bsdf->extra->color = stack_load_float3(stack, data_node.z); - ccl_fetch(sd, flag) |= bsdf_microfacet_multi_ggx_setup(bsdf); + sd->flag |= bsdf_microfacet_multi_ggx_setup(bsdf); } } else - ccl_fetch(sd, flag) |= bsdf_ashikhmin_shirley_setup(bsdf); + sd->flag |= bsdf_ashikhmin_shirley_setup(bsdf); } break; @@ -493,7 +493,7 @@ ccl_device void svm_node_closure_bsdf(KernelGlobals *kg, ShaderData *sd, float * if(!kernel_data.integrator.caustics_refractive && (path_flag & PATH_RAY_DIFFUSE)) break; #endif - float3 weight = ccl_fetch(sd, svm_closure_weight) * mix_weight; + float3 weight = sd->svm_closure_weight * mix_weight; MicrofacetBsdf *bsdf = (MicrofacetBsdf*)bsdf_alloc(sd, sizeof(MicrofacetBsdf), weight); if(bsdf) { @@ -501,7 +501,7 @@ ccl_device void svm_node_closure_bsdf(KernelGlobals *kg, ShaderData *sd, float * bsdf->extra = NULL; float eta = fmaxf(param2, 1e-5f); - eta = (ccl_fetch(sd, flag) & SD_BACKFACING)? 1.0f/eta: eta; + eta = (sd->flag & SD_BACKFACING)? 1.0f/eta: eta; /* setup bsdf */ if(type == CLOSURE_BSDF_REFRACTION_ID) { @@ -509,7 +509,7 @@ ccl_device void svm_node_closure_bsdf(KernelGlobals *kg, ShaderData *sd, float * bsdf->alpha_y = 0.0f; bsdf->ior = eta; - ccl_fetch(sd, flag) |= bsdf_refraction_setup(bsdf); + sd->flag |= bsdf_refraction_setup(bsdf); } else { bsdf->alpha_x = param1; @@ -517,9 +517,9 @@ ccl_device void svm_node_closure_bsdf(KernelGlobals *kg, ShaderData *sd, float * bsdf->ior = eta; if(type == CLOSURE_BSDF_MICROFACET_BECKMANN_REFRACTION_ID) - ccl_fetch(sd, flag) |= bsdf_microfacet_beckmann_refraction_setup(bsdf); + sd->flag |= bsdf_microfacet_beckmann_refraction_setup(bsdf); else - ccl_fetch(sd, flag) |= bsdf_microfacet_ggx_refraction_setup(bsdf); + sd->flag |= bsdf_microfacet_ggx_refraction_setup(bsdf); } } @@ -535,14 +535,14 @@ ccl_device void svm_node_closure_bsdf(KernelGlobals *kg, ShaderData *sd, float * break; } #endif - float3 weight = ccl_fetch(sd, svm_closure_weight) * mix_weight; + float3 weight = sd->svm_closure_weight * mix_weight; /* index of refraction */ float eta = fmaxf(param2, 1e-5f); - eta = (ccl_fetch(sd, flag) & SD_BACKFACING)? 1.0f/eta: eta; + eta = (sd->flag & SD_BACKFACING)? 1.0f/eta: eta; /* fresnel */ - float cosNO = dot(N, ccl_fetch(sd, I)); + float cosNO = dot(N, sd->I); float fresnel = fresnel_dielectric_cos(cosNO, eta); float roughness = param1; @@ -581,7 +581,7 @@ ccl_device void svm_node_closure_bsdf(KernelGlobals *kg, ShaderData *sd, float * if(!kernel_data.integrator.caustics_reflective && !kernel_data.integrator.caustics_refractive && (path_flag & PATH_RAY_DIFFUSE)) break; #endif - float3 weight = ccl_fetch(sd, svm_closure_weight) * mix_weight; + float3 weight = sd->svm_closure_weight * mix_weight; MicrofacetBsdf *bsdf = (MicrofacetBsdf*)bsdf_alloc(sd, sizeof(MicrofacetBsdf), weight); MicrofacetExtra *extra = (MicrofacetExtra*)closure_alloc_extra(sd, sizeof(MicrofacetExtra)); @@ -593,13 +593,13 @@ ccl_device void svm_node_closure_bsdf(KernelGlobals *kg, ShaderData *sd, float * bsdf->alpha_x = param1; bsdf->alpha_y = param1; float eta = fmaxf(param2, 1e-5f); - bsdf->ior = (ccl_fetch(sd, flag) & SD_BACKFACING)? 1.0f/eta: eta; + bsdf->ior = (sd->flag & SD_BACKFACING)? 1.0f/eta: eta; kernel_assert(stack_valid(data_node.z)); bsdf->extra->color = stack_load_float3(stack, data_node.z); /* setup bsdf */ - ccl_fetch(sd, flag) |= bsdf_microfacet_multi_ggx_glass_setup(bsdf); + sd->flag |= bsdf_microfacet_multi_ggx_glass_setup(bsdf); } break; @@ -612,7 +612,7 @@ ccl_device void svm_node_closure_bsdf(KernelGlobals *kg, ShaderData *sd, float * if(!kernel_data.integrator.caustics_reflective && (path_flag & PATH_RAY_DIFFUSE)) break; #endif - float3 weight = ccl_fetch(sd, svm_closure_weight) * mix_weight; + float3 weight = sd->svm_closure_weight * mix_weight; MicrofacetBsdf *bsdf = (MicrofacetBsdf*)bsdf_alloc(sd, sizeof(MicrofacetBsdf), weight); if(bsdf) { @@ -642,33 +642,33 @@ ccl_device void svm_node_closure_bsdf(KernelGlobals *kg, ShaderData *sd, float * bsdf->ior = 0.0f; if(type == CLOSURE_BSDF_MICROFACET_BECKMANN_ANISO_ID) { - ccl_fetch(sd, flag) |= bsdf_microfacet_beckmann_aniso_setup(bsdf); + sd->flag |= bsdf_microfacet_beckmann_aniso_setup(bsdf); } else if(type == CLOSURE_BSDF_MICROFACET_GGX_ANISO_ID) { - ccl_fetch(sd, flag) |= bsdf_microfacet_ggx_aniso_setup(bsdf); + sd->flag |= bsdf_microfacet_ggx_aniso_setup(bsdf); } else if(type == CLOSURE_BSDF_MICROFACET_MULTI_GGX_ANISO_ID) { kernel_assert(stack_valid(data_node.w)); bsdf->extra = (MicrofacetExtra*)closure_alloc_extra(sd, sizeof(MicrofacetExtra)); if(bsdf->extra) { bsdf->extra->color = stack_load_float3(stack, data_node.w); - ccl_fetch(sd, flag) |= bsdf_microfacet_multi_ggx_aniso_setup(bsdf); + sd->flag |= bsdf_microfacet_multi_ggx_aniso_setup(bsdf); } } else - ccl_fetch(sd, flag) |= bsdf_ashikhmin_shirley_aniso_setup(bsdf); + sd->flag |= bsdf_ashikhmin_shirley_aniso_setup(bsdf); } break; } case CLOSURE_BSDF_ASHIKHMIN_VELVET_ID: { - float3 weight = ccl_fetch(sd, svm_closure_weight) * mix_weight; + float3 weight = sd->svm_closure_weight * mix_weight; VelvetBsdf *bsdf = (VelvetBsdf*)bsdf_alloc(sd, sizeof(VelvetBsdf), weight); if(bsdf) { bsdf->N = N; bsdf->sigma = saturate(param1); - ccl_fetch(sd, flag) |= bsdf_ashikhmin_velvet_setup(bsdf); + sd->flag |= bsdf_ashikhmin_velvet_setup(bsdf); } break; } @@ -678,7 +678,7 @@ ccl_device void svm_node_closure_bsdf(KernelGlobals *kg, ShaderData *sd, float * break; #endif case CLOSURE_BSDF_DIFFUSE_TOON_ID: { - float3 weight = ccl_fetch(sd, svm_closure_weight) * mix_weight; + float3 weight = sd->svm_closure_weight * mix_weight; ToonBsdf *bsdf = (ToonBsdf*)bsdf_alloc(sd, sizeof(ToonBsdf), weight); if(bsdf) { @@ -687,18 +687,18 @@ ccl_device void svm_node_closure_bsdf(KernelGlobals *kg, ShaderData *sd, float * bsdf->smooth = param2; if(type == CLOSURE_BSDF_DIFFUSE_TOON_ID) - ccl_fetch(sd, flag) |= bsdf_diffuse_toon_setup(bsdf); + sd->flag |= bsdf_diffuse_toon_setup(bsdf); else - ccl_fetch(sd, flag) |= bsdf_glossy_toon_setup(bsdf); + sd->flag |= bsdf_glossy_toon_setup(bsdf); } break; } #ifdef __HAIR__ case CLOSURE_BSDF_HAIR_REFLECTION_ID: case CLOSURE_BSDF_HAIR_TRANSMISSION_ID: { - float3 weight = ccl_fetch(sd, svm_closure_weight) * mix_weight; + float3 weight = sd->svm_closure_weight * mix_weight; - if(ccl_fetch(sd, flag) & SD_BACKFACING && ccl_fetch(sd, type) & PRIMITIVE_ALL_CURVE) { + if(sd->flag & SD_BACKFACING && sd->type & PRIMITIVE_ALL_CURVE) { ShaderClosure *bsdf = bsdf_alloc(sd, sizeof(ShaderClosure), weight); if(bsdf) { @@ -708,7 +708,7 @@ ccl_device void svm_node_closure_bsdf(KernelGlobals *kg, ShaderData *sd, float * * better figure out a way to skip backfaces from rays * spawned by transmission from the front */ bsdf->weight = make_float3(1.0f, 1.0f, 1.0f); - ccl_fetch(sd, flag) |= bsdf_transparent_setup(bsdf); + sd->flag |= bsdf_transparent_setup(bsdf); } } else { @@ -722,18 +722,18 @@ ccl_device void svm_node_closure_bsdf(KernelGlobals *kg, ShaderData *sd, float * if(stack_valid(data_node.y)) { bsdf->T = normalize(stack_load_float3(stack, data_node.y)); } - else if(!(ccl_fetch(sd, type) & PRIMITIVE_ALL_CURVE)) { - bsdf->T = normalize(ccl_fetch(sd, dPdv)); + else if(!(sd->type & PRIMITIVE_ALL_CURVE)) { + bsdf->T = normalize(sd->dPdv); bsdf->offset = 0.0f; } else - bsdf->T = normalize(ccl_fetch(sd, dPdu)); + bsdf->T = normalize(sd->dPdu); if(type == CLOSURE_BSDF_HAIR_REFLECTION_ID) { - ccl_fetch(sd, flag) |= bsdf_hair_reflection_setup(bsdf); + sd->flag |= bsdf_hair_reflection_setup(bsdf); } else { - ccl_fetch(sd, flag) |= bsdf_hair_transmission_setup(bsdf); + sd->flag |= bsdf_hair_transmission_setup(bsdf); } } } @@ -746,8 +746,8 @@ ccl_device void svm_node_closure_bsdf(KernelGlobals *kg, ShaderData *sd, float * case CLOSURE_BSSRDF_CUBIC_ID: case CLOSURE_BSSRDF_GAUSSIAN_ID: case CLOSURE_BSSRDF_BURLEY_ID: { - float3 albedo = ccl_fetch(sd, svm_closure_weight); - float3 weight = ccl_fetch(sd, svm_closure_weight) * mix_weight; + float3 albedo = sd->svm_closure_weight; + float3 weight = sd->svm_closure_weight * mix_weight; float sample_weight = fabsf(average(weight)); /* disable in case of diffuse ancestor, can't see it well then and @@ -773,7 +773,7 @@ ccl_device void svm_node_closure_bsdf(KernelGlobals *kg, ShaderData *sd, float * bssrdf->albedo = albedo.x; bssrdf->sharpness = sharpness; bssrdf->N = N; - ccl_fetch(sd, flag) |= bssrdf_setup(bssrdf, (ClosureType)type); + sd->flag |= bssrdf_setup(bssrdf, (ClosureType)type); } bssrdf = bssrdf_alloc(sd, make_float3(0.0f, weight.y, 0.0f)); @@ -784,7 +784,7 @@ ccl_device void svm_node_closure_bsdf(KernelGlobals *kg, ShaderData *sd, float * bssrdf->albedo = albedo.y; bssrdf->sharpness = sharpness; bssrdf->N = N; - ccl_fetch(sd, flag) |= bssrdf_setup(bssrdf, (ClosureType)type); + sd->flag |= bssrdf_setup(bssrdf, (ClosureType)type); } bssrdf = bssrdf_alloc(sd, make_float3(0.0f, 0.0f, weight.z)); @@ -795,7 +795,7 @@ ccl_device void svm_node_closure_bsdf(KernelGlobals *kg, ShaderData *sd, float * bssrdf->albedo = albedo.z; bssrdf->sharpness = sharpness; bssrdf->N = N; - ccl_fetch(sd, flag) |= bssrdf_setup(bssrdf, (ClosureType)type); + sd->flag |= bssrdf_setup(bssrdf, (ClosureType)type); } } @@ -825,21 +825,21 @@ ccl_device void svm_node_closure_volume(KernelGlobals *kg, ShaderData *sd, float switch(type) { case CLOSURE_VOLUME_ABSORPTION_ID: { - float3 weight = (make_float3(1.0f, 1.0f, 1.0f) - ccl_fetch(sd, svm_closure_weight)) * mix_weight * density; + float3 weight = (make_float3(1.0f, 1.0f, 1.0f) - sd->svm_closure_weight) * mix_weight * density; ShaderClosure *sc = closure_alloc(sd, sizeof(ShaderClosure), CLOSURE_NONE_ID, weight); if(sc) { - ccl_fetch(sd, flag) |= volume_absorption_setup(sc); + sd->flag |= volume_absorption_setup(sc); } break; } case CLOSURE_VOLUME_HENYEY_GREENSTEIN_ID: { - float3 weight = ccl_fetch(sd, svm_closure_weight) * mix_weight * density; + float3 weight = sd->svm_closure_weight * mix_weight * density; HenyeyGreensteinVolume *volume = (HenyeyGreensteinVolume*)bsdf_alloc(sd, sizeof(HenyeyGreensteinVolume), weight); if(volume) { volume->g = param2; /* g */ - ccl_fetch(sd, flag) |= volume_henyey_greenstein_setup(volume); + sd->flag |= volume_henyey_greenstein_setup(volume); } break; } @@ -859,12 +859,12 @@ ccl_device void svm_node_closure_emission(ShaderData *sd, float *stack, uint4 no if(mix_weight == 0.0f) return; - closure_alloc(sd, sizeof(ShaderClosure), CLOSURE_EMISSION_ID, ccl_fetch(sd, svm_closure_weight) * mix_weight); + closure_alloc(sd, sizeof(ShaderClosure), CLOSURE_EMISSION_ID, sd->svm_closure_weight * mix_weight); } else - closure_alloc(sd, sizeof(ShaderClosure), CLOSURE_EMISSION_ID, ccl_fetch(sd, svm_closure_weight)); + closure_alloc(sd, sizeof(ShaderClosure), CLOSURE_EMISSION_ID, sd->svm_closure_weight); - ccl_fetch(sd, flag) |= SD_EMISSION; + sd->flag |= SD_EMISSION; } ccl_device void svm_node_closure_background(ShaderData *sd, float *stack, uint4 node) @@ -877,10 +877,10 @@ ccl_device void svm_node_closure_background(ShaderData *sd, float *stack, uint4 if(mix_weight == 0.0f) return; - closure_alloc(sd, sizeof(ShaderClosure), CLOSURE_BACKGROUND_ID, ccl_fetch(sd, svm_closure_weight) * mix_weight); + closure_alloc(sd, sizeof(ShaderClosure), CLOSURE_BACKGROUND_ID, sd->svm_closure_weight * mix_weight); } else - closure_alloc(sd, sizeof(ShaderClosure), CLOSURE_BACKGROUND_ID, ccl_fetch(sd, svm_closure_weight)); + closure_alloc(sd, sizeof(ShaderClosure), CLOSURE_BACKGROUND_ID, sd->svm_closure_weight); } ccl_device void svm_node_closure_holdout(ShaderData *sd, float *stack, uint4 node) @@ -893,12 +893,12 @@ ccl_device void svm_node_closure_holdout(ShaderData *sd, float *stack, uint4 nod if(mix_weight == 0.0f) return; - closure_alloc(sd, sizeof(ShaderClosure), CLOSURE_HOLDOUT_ID, ccl_fetch(sd, svm_closure_weight) * mix_weight); + closure_alloc(sd, sizeof(ShaderClosure), CLOSURE_HOLDOUT_ID, sd->svm_closure_weight * mix_weight); } else - closure_alloc(sd, sizeof(ShaderClosure), CLOSURE_HOLDOUT_ID, ccl_fetch(sd, svm_closure_weight)); + closure_alloc(sd, sizeof(ShaderClosure), CLOSURE_HOLDOUT_ID, sd->svm_closure_weight); - ccl_fetch(sd, flag) |= SD_HOLDOUT; + sd->flag |= SD_HOLDOUT; } ccl_device void svm_node_closure_ambient_occlusion(ShaderData *sd, float *stack, uint4 node) @@ -911,19 +911,19 @@ ccl_device void svm_node_closure_ambient_occlusion(ShaderData *sd, float *stack, if(mix_weight == 0.0f) return; - closure_alloc(sd, sizeof(ShaderClosure), CLOSURE_AMBIENT_OCCLUSION_ID, ccl_fetch(sd, svm_closure_weight) * mix_weight); + closure_alloc(sd, sizeof(ShaderClosure), CLOSURE_AMBIENT_OCCLUSION_ID, sd->svm_closure_weight * mix_weight); } else - closure_alloc(sd, sizeof(ShaderClosure), CLOSURE_AMBIENT_OCCLUSION_ID, ccl_fetch(sd, svm_closure_weight)); + closure_alloc(sd, sizeof(ShaderClosure), CLOSURE_AMBIENT_OCCLUSION_ID, sd->svm_closure_weight); - ccl_fetch(sd, flag) |= SD_AO; + sd->flag |= SD_AO; } /* Closure Nodes */ ccl_device_inline void svm_node_closure_store_weight(ShaderData *sd, float3 weight) { - ccl_fetch(sd, svm_closure_weight) = weight; + sd->svm_closure_weight = weight; } ccl_device void svm_node_closure_set_weight(ShaderData *sd, uint r, uint g, uint b) @@ -973,7 +973,7 @@ ccl_device void svm_node_mix_closure(ShaderData *sd, float *stack, uint4 node) ccl_device void svm_node_set_normal(KernelGlobals *kg, ShaderData *sd, float *stack, uint in_direction, uint out_normal) { float3 normal = stack_load_float3(stack, in_direction); - ccl_fetch(sd, N) = normal; + sd->N = normal; stack_store_float3(stack, out_normal, normal); } diff --git a/intern/cycles/kernel/svm/svm_displace.h b/intern/cycles/kernel/svm/svm_displace.h index 890ab41aaaa..c94fa130af7 100644 --- a/intern/cycles/kernel/svm/svm_displace.h +++ b/intern/cycles/kernel/svm/svm_displace.h @@ -25,10 +25,10 @@ ccl_device void svm_node_set_bump(KernelGlobals *kg, ShaderData *sd, float *stac uint normal_offset, distance_offset, invert, use_object_space; decode_node_uchar4(node.y, &normal_offset, &distance_offset, &invert, &use_object_space); - float3 normal_in = stack_valid(normal_offset)? stack_load_float3(stack, normal_offset): ccl_fetch(sd, N); + float3 normal_in = stack_valid(normal_offset)? stack_load_float3(stack, normal_offset): sd->N; - float3 dPdx = ccl_fetch(sd, dP).dx; - float3 dPdy = ccl_fetch(sd, dP).dy; + float3 dPdx = sd->dP.dx; + float3 dPdy = sd->dP.dy; if(use_object_space) { object_inverse_normal_transform(kg, sd, &normal_in); @@ -80,14 +80,14 @@ ccl_device void svm_node_set_displacement(KernelGlobals *kg, ShaderData *sd, flo { float d = stack_load_float(stack, fac_offset); - float3 dP = ccl_fetch(sd, N); + float3 dP = sd->N; object_inverse_normal_transform(kg, sd, &dP); dP *= d*0.1f; /* todo: get rid of this factor */ object_dir_transform(kg, sd, &dP); - ccl_fetch(sd, P) += dP; + sd->P += dP; } CCL_NAMESPACE_END diff --git a/intern/cycles/kernel/svm/svm_fresnel.h b/intern/cycles/kernel/svm/svm_fresnel.h index 23c97d80cb0..3703ec55015 100644 --- a/intern/cycles/kernel/svm/svm_fresnel.h +++ b/intern/cycles/kernel/svm/svm_fresnel.h @@ -23,12 +23,12 @@ ccl_device void svm_node_fresnel(ShaderData *sd, float *stack, uint ior_offset, uint normal_offset, out_offset; decode_node_uchar4(node, &normal_offset, &out_offset, NULL, NULL); float eta = (stack_valid(ior_offset))? stack_load_float(stack, ior_offset): __uint_as_float(ior_value); - float3 normal_in = stack_valid(normal_offset)? stack_load_float3(stack, normal_offset): ccl_fetch(sd, N); + float3 normal_in = stack_valid(normal_offset)? stack_load_float3(stack, normal_offset): sd->N; eta = fmaxf(eta, 1e-5f); - eta = (ccl_fetch(sd, flag) & SD_BACKFACING)? 1.0f/eta: eta; + eta = (sd->flag & SD_BACKFACING)? 1.0f/eta: eta; - float f = fresnel_dielectric_cos(dot(ccl_fetch(sd, I), normal_in), eta); + float f = fresnel_dielectric_cos(dot(sd->I, normal_in), eta); stack_store_float(stack, out_offset, f); } @@ -44,18 +44,18 @@ ccl_device void svm_node_layer_weight(ShaderData *sd, float *stack, uint4 node) decode_node_uchar4(node.w, &type, &normal_offset, &out_offset, NULL); float blend = (stack_valid(blend_offset))? stack_load_float(stack, blend_offset): __uint_as_float(blend_value); - float3 normal_in = (stack_valid(normal_offset))? stack_load_float3(stack, normal_offset): ccl_fetch(sd, N); + float3 normal_in = (stack_valid(normal_offset))? stack_load_float3(stack, normal_offset): sd->N; float f; if(type == NODE_LAYER_WEIGHT_FRESNEL) { float eta = fmaxf(1.0f - blend, 1e-5f); - eta = (ccl_fetch(sd, flag) & SD_BACKFACING)? eta: 1.0f/eta; + eta = (sd->flag & SD_BACKFACING)? eta: 1.0f/eta; - f = fresnel_dielectric_cos(dot(ccl_fetch(sd, I), normal_in), eta); + f = fresnel_dielectric_cos(dot(sd->I, normal_in), eta); } else { - f = fabsf(dot(ccl_fetch(sd, I), normal_in)); + f = fabsf(dot(sd->I, normal_in)); if(blend != 0.5f) { blend = clamp(blend, 0.0f, 1.0f-1e-5f); diff --git a/intern/cycles/kernel/svm/svm_geometry.h b/intern/cycles/kernel/svm/svm_geometry.h index 7d512f7ff4d..4a09d9f6653 100644 --- a/intern/cycles/kernel/svm/svm_geometry.h +++ b/intern/cycles/kernel/svm/svm_geometry.h @@ -27,15 +27,15 @@ ccl_device_inline void svm_node_geometry(KernelGlobals *kg, float3 data; switch(type) { - case NODE_GEOM_P: data = ccl_fetch(sd, P); break; - case NODE_GEOM_N: data = ccl_fetch(sd, N); break; + case NODE_GEOM_P: data = sd->P; break; + case NODE_GEOM_N: data = sd->N; break; #ifdef __DPDU__ case NODE_GEOM_T: data = primitive_tangent(kg, sd); break; #endif - case NODE_GEOM_I: data = ccl_fetch(sd, I); break; - case NODE_GEOM_Ng: data = ccl_fetch(sd, Ng); break; + case NODE_GEOM_I: data = sd->I; break; + case NODE_GEOM_Ng: data = sd->Ng; break; #ifdef __UV__ - case NODE_GEOM_uv: data = make_float3(ccl_fetch(sd, u), ccl_fetch(sd, v), 0.0f); break; + case NODE_GEOM_uv: data = make_float3(sd->u, sd->v, 0.0f); break; #endif } @@ -48,8 +48,8 @@ ccl_device void svm_node_geometry_bump_dx(KernelGlobals *kg, ShaderData *sd, flo float3 data; switch(type) { - case NODE_GEOM_P: data = ccl_fetch(sd, P) + ccl_fetch(sd, dP).dx; break; - case NODE_GEOM_uv: data = make_float3(ccl_fetch(sd, u) + ccl_fetch(sd, du).dx, ccl_fetch(sd, v) + ccl_fetch(sd, dv).dx, 0.0f); break; + case NODE_GEOM_P: data = sd->P + sd->dP.dx; break; + case NODE_GEOM_uv: data = make_float3(sd->u + sd->du.dx, sd->v + sd->dv.dx, 0.0f); break; default: svm_node_geometry(kg, sd, stack, type, out_offset); return; } @@ -65,8 +65,8 @@ ccl_device void svm_node_geometry_bump_dy(KernelGlobals *kg, ShaderData *sd, flo float3 data; switch(type) { - case NODE_GEOM_P: data = ccl_fetch(sd, P) + ccl_fetch(sd, dP).dy; break; - case NODE_GEOM_uv: data = make_float3(ccl_fetch(sd, u) + ccl_fetch(sd, du).dy, ccl_fetch(sd, v) + ccl_fetch(sd, dv).dy, 0.0f); break; + case NODE_GEOM_P: data = sd->P + sd->dP.dy; break; + case NODE_GEOM_uv: data = make_float3(sd->u + sd->du.dy, sd->v + sd->dv.dy, 0.0f); break; default: svm_node_geometry(kg, sd, stack, type, out_offset); return; } @@ -87,9 +87,9 @@ ccl_device void svm_node_object_info(KernelGlobals *kg, ShaderData *sd, float *s stack_store_float3(stack, out_offset, object_location(kg, sd)); return; } - case NODE_INFO_OB_INDEX: data = object_pass_id(kg, ccl_fetch(sd, object)); break; + case NODE_INFO_OB_INDEX: data = object_pass_id(kg, sd->object); break; case NODE_INFO_MAT_INDEX: data = shader_pass_id(kg, sd); break; - case NODE_INFO_OB_RANDOM: data = object_random_number(kg, ccl_fetch(sd, object)); break; + case NODE_INFO_OB_RANDOM: data = object_random_number(kg, sd->object); break; default: data = 0.0f; break; } @@ -106,44 +106,44 @@ ccl_device void svm_node_particle_info(KernelGlobals *kg, { switch(type) { case NODE_INFO_PAR_INDEX: { - int particle_id = object_particle_id(kg, ccl_fetch(sd, object)); + int particle_id = object_particle_id(kg, sd->object); stack_store_float(stack, out_offset, particle_index(kg, particle_id)); break; } case NODE_INFO_PAR_AGE: { - int particle_id = object_particle_id(kg, ccl_fetch(sd, object)); + int particle_id = object_particle_id(kg, sd->object); stack_store_float(stack, out_offset, particle_age(kg, particle_id)); break; } case NODE_INFO_PAR_LIFETIME: { - int particle_id = object_particle_id(kg, ccl_fetch(sd, object)); + int particle_id = object_particle_id(kg, sd->object); stack_store_float(stack, out_offset, particle_lifetime(kg, particle_id)); break; } case NODE_INFO_PAR_LOCATION: { - int particle_id = object_particle_id(kg, ccl_fetch(sd, object)); + int particle_id = object_particle_id(kg, sd->object); stack_store_float3(stack, out_offset, particle_location(kg, particle_id)); break; } #if 0 /* XXX float4 currently not supported in SVM stack */ case NODE_INFO_PAR_ROTATION: { - int particle_id = object_particle_id(kg, ccl_fetch(sd, object)); + int particle_id = object_particle_id(kg, sd->object); stack_store_float4(stack, out_offset, particle_rotation(kg, particle_id)); break; } #endif case NODE_INFO_PAR_SIZE: { - int particle_id = object_particle_id(kg, ccl_fetch(sd, object)); + int particle_id = object_particle_id(kg, sd->object); stack_store_float(stack, out_offset, particle_size(kg, particle_id)); break; } case NODE_INFO_PAR_VELOCITY: { - int particle_id = object_particle_id(kg, ccl_fetch(sd, object)); + int particle_id = object_particle_id(kg, sd->object); stack_store_float3(stack, out_offset, particle_velocity(kg, particle_id)); break; } case NODE_INFO_PAR_ANGULAR_VELOCITY: { - int particle_id = object_particle_id(kg, ccl_fetch(sd, object)); + int particle_id = object_particle_id(kg, sd->object); stack_store_float3(stack, out_offset, particle_angular_velocity(kg, particle_id)); break; } @@ -165,7 +165,7 @@ ccl_device void svm_node_hair_info(KernelGlobals *kg, switch(type) { case NODE_INFO_CURVE_IS_STRAND: { - data = (ccl_fetch(sd, type) & PRIMITIVE_ALL_CURVE) != 0; + data = (sd->type & PRIMITIVE_ALL_CURVE) != 0; stack_store_float(stack, out_offset, data); break; } @@ -177,7 +177,7 @@ ccl_device void svm_node_hair_info(KernelGlobals *kg, break; } /*case NODE_INFO_CURVE_FADE: { - data = ccl_fetch(sd, curve_transparency); + data = sd->curve_transparency; stack_store_float(stack, out_offset, data); break; }*/ diff --git a/intern/cycles/kernel/svm/svm_image.h b/intern/cycles/kernel/svm/svm_image.h index 2afdf61b476..76acc9253a1 100644 --- a/intern/cycles/kernel/svm/svm_image.h +++ b/intern/cycles/kernel/svm/svm_image.h @@ -144,7 +144,6 @@ ccl_device float4 svm_image_texture(KernelGlobals *kg, int id, float x, float y, case 86: r = kernel_tex_image_interp(__tex_image_byte4_086, x, y); break; case 87: r = kernel_tex_image_interp(__tex_image_byte4_087, x, y); break; case 88: r = kernel_tex_image_interp(__tex_image_byte4_088, x, y); break; - case 89: r = kernel_tex_image_interp(__tex_image_byte4_089, x, y); break; default: kernel_assert(0); return make_float4(0.0f, 0.0f, 0.0f, 0.0f); @@ -238,9 +237,9 @@ ccl_device void svm_node_tex_image(KernelGlobals *kg, ShaderData *sd, float *sta ccl_device void svm_node_tex_image_box(KernelGlobals *kg, ShaderData *sd, float *stack, uint4 node) { /* get object space normal */ - float3 N = ccl_fetch(sd, N); + float3 N = sd->N; - N = ccl_fetch(sd, N); + N = sd->N; object_inverse_normal_transform(kg, sd, &N); /* project from direction vector to barycentric coordinates in triangles */ diff --git a/intern/cycles/kernel/svm/svm_light_path.h b/intern/cycles/kernel/svm/svm_light_path.h index 04f6f623f18..1492e358608 100644 --- a/intern/cycles/kernel/svm/svm_light_path.h +++ b/intern/cycles/kernel/svm/svm_light_path.h @@ -31,8 +31,8 @@ ccl_device void svm_node_light_path(ShaderData *sd, ccl_addr_space PathState *st case NODE_LP_reflection: info = (path_flag & PATH_RAY_REFLECT)? 1.0f: 0.0f; break; case NODE_LP_transmission: info = (path_flag & PATH_RAY_TRANSMIT)? 1.0f: 0.0f; break; case NODE_LP_volume_scatter: info = (path_flag & PATH_RAY_VOLUME_SCATTER)? 1.0f: 0.0f; break; - case NODE_LP_backfacing: info = (ccl_fetch(sd, flag) & SD_BACKFACING)? 1.0f: 0.0f; break; - case NODE_LP_ray_length: info = ccl_fetch(sd, ray_length); break; + case NODE_LP_backfacing: info = (sd->flag & SD_BACKFACING)? 1.0f: 0.0f; break; + case NODE_LP_ray_length: info = sd->ray_length; break; case NODE_LP_ray_depth: info = (float)state->bounce; break; case NODE_LP_ray_diffuse: info = (float)state->diffuse_bounce; break; case NODE_LP_ray_glossy: info = (float)state->glossy_bounce; break; @@ -56,14 +56,14 @@ ccl_device void svm_node_light_falloff(ShaderData *sd, float *stack, uint4 node) switch(type) { case NODE_LIGHT_FALLOFF_QUADRATIC: break; - case NODE_LIGHT_FALLOFF_LINEAR: strength *= ccl_fetch(sd, ray_length); break; - case NODE_LIGHT_FALLOFF_CONSTANT: strength *= ccl_fetch(sd, ray_length)*ccl_fetch(sd, ray_length); break; + case NODE_LIGHT_FALLOFF_LINEAR: strength *= sd->ray_length; break; + case NODE_LIGHT_FALLOFF_CONSTANT: strength *= sd->ray_length*sd->ray_length; break; } float smooth = stack_load_float(stack, smooth_offset); if(smooth > 0.0f) { - float squared = ccl_fetch(sd, ray_length)*ccl_fetch(sd, ray_length); + float squared = sd->ray_length*sd->ray_length; /* Distant lamps set the ray length to FLT_MAX, which causes squared to overflow. */ if(isfinite(squared)) { strength *= squared/(smooth + squared); diff --git a/intern/cycles/kernel/svm/svm_math_util.h b/intern/cycles/kernel/svm/svm_math_util.h index 01547b60014..a7f15de7325 100644 --- a/intern/cycles/kernel/svm/svm_math_util.h +++ b/intern/cycles/kernel/svm/svm_math_util.h @@ -134,32 +134,37 @@ ccl_device float3 svm_math_blackbody_color(float t) { { 6.72595954e-13f, -2.73059993e-08f, 4.24068546e-04f, -7.52204323e-01f }, }; - if(t >= 12000.0f) + int i; + if(t >= 12000.0f) { return make_float3(0.826270103f, 0.994478524f, 1.56626022f); + } + else if(t >= 6365.0f) { + i = 5; + } + else if(t >= 3315.0f) { + i = 4; + } + else if(t >= 1902.0f) { + i = 3; + } + else if(t >= 1449.0f) { + i = 2; + } + else if(t >= 1167.0f) { + i = 1; + } + else if(t >= 965.0f) { + i = 0; + } + else { + /* For 800 <= t < 965 color does not change in OSL implementation, so keep color the same */ + return make_float3(4.70366907f, 0.0f, 0.0f); + } - /* Define a macro to reduce stack usage for nvcc */ -#define MAKE_BB_RGB(i) make_float3(\ - rc[i][0] / t + rc[i][1] * t + rc[i][2],\ - gc[i][0] / t + gc[i][1] * t + gc[i][2],\ - ((bc[i][0] * t + bc[i][1]) * t + bc[i][2]) * t + bc[i][3]) - - if(t >= 6365.0f) - return MAKE_BB_RGB(5); - if(t >= 3315.0f) - return MAKE_BB_RGB(4); - if(t >= 1902.0f) - return MAKE_BB_RGB(3); - if(t >= 1449.0f) - return MAKE_BB_RGB(2); - if(t >= 1167.0f) - return MAKE_BB_RGB(1); - if(t >= 965.0f) - return MAKE_BB_RGB(0); - -#undef MAKE_BB_RGB - - /* For 800 <= t < 965 color does not change in OSL implementation, so keep color the same */ - return make_float3(4.70366907f, 0.0f, 0.0f); + const float t_inv = 1.0f / t; + return make_float3(rc[i][0] * t_inv + rc[i][1] * t + rc[i][2], + gc[i][0] * t_inv + gc[i][1] * t + gc[i][2], + ((bc[i][0] * t + bc[i][1]) * t + bc[i][2]) * t + bc[i][3]); } ccl_device_inline float3 svm_math_gamma_color(float3 color, float gamma) diff --git a/intern/cycles/kernel/svm/svm_noisetex.h b/intern/cycles/kernel/svm/svm_noisetex.h index 62ff38cf1c5..0347ab7b193 100644 --- a/intern/cycles/kernel/svm/svm_noisetex.h +++ b/intern/cycles/kernel/svm/svm_noisetex.h @@ -18,50 +18,42 @@ CCL_NAMESPACE_BEGIN /* Noise */ -ccl_device_inline void svm_noise(float3 p, float detail, float distortion, float *fac, float3 *color) -{ - int hard = 0; - - if(distortion != 0.0f) { - float3 r, offset = make_float3(13.5f, 13.5f, 13.5f); - - r.x = noise(p + offset) * distortion; - r.y = noise(p) * distortion; - r.z = noise(p - offset) * distortion; - - p += r; - } - - *fac = noise_turbulence(p, detail, hard); - *color = make_float3(*fac, - noise_turbulence(make_float3(p.y, p.x, p.z), detail, hard), - noise_turbulence(make_float3(p.y, p.z, p.x), detail, hard)); -} - ccl_device void svm_node_tex_noise(KernelGlobals *kg, ShaderData *sd, float *stack, uint4 node, int *offset) { uint co_offset, scale_offset, detail_offset, distortion_offset, fac_offset, color_offset; decode_node_uchar4(node.y, &co_offset, &scale_offset, &detail_offset, &distortion_offset); + decode_node_uchar4(node.z, &color_offset, &fac_offset, NULL, NULL); uint4 node2 = read_node(kg, offset); float scale = stack_load_float_default(stack, scale_offset, node2.x); float detail = stack_load_float_default(stack, detail_offset, node2.y); float distortion = stack_load_float_default(stack, distortion_offset, node2.z); - float3 co = stack_load_float3(stack, co_offset); + float3 p = stack_load_float3(stack, co_offset) * scale; + int hard = 0; - float3 color; - float f; + if(distortion != 0.0f) { + float3 r, offset = make_float3(13.5f, 13.5f, 13.5f); + + r.x = noise(p + offset) * distortion; + r.y = noise(p) * distortion; + r.z = noise(p - offset) * distortion; - svm_noise(co*scale, detail, distortion, &f, &color); + p += r; + } - decode_node_uchar4(node.z, &color_offset, &fac_offset, NULL, NULL); + float f = noise_turbulence(p, detail, hard); - if(stack_valid(fac_offset)) + if(stack_valid(fac_offset)) { stack_store_float(stack, fac_offset, f); - if(stack_valid(color_offset)) + } + if(stack_valid(color_offset)) { + float3 color = make_float3(f, + noise_turbulence(make_float3(p.y, p.x, p.z), detail, hard), + noise_turbulence(make_float3(p.y, p.z, p.x), detail, hard)); stack_store_float3(stack, color_offset, color); + } } CCL_NAMESPACE_END diff --git a/intern/cycles/kernel/svm/svm_tex_coord.h b/intern/cycles/kernel/svm/svm_tex_coord.h index c0b01262212..c94327401f5 100644 --- a/intern/cycles/kernel/svm/svm_tex_coord.h +++ b/intern/cycles/kernel/svm/svm_tex_coord.h @@ -31,9 +31,9 @@ ccl_device void svm_node_tex_coord(KernelGlobals *kg, switch(type) { case NODE_TEXCO_OBJECT: { - data = ccl_fetch(sd, P); + data = sd->P; if(node.w == 0) { - if(ccl_fetch(sd, object) != OBJECT_NONE) { + if(sd->object != OBJECT_NONE) { object_inverse_position_transform(kg, sd, &data); } } @@ -48,47 +48,47 @@ ccl_device void svm_node_tex_coord(KernelGlobals *kg, break; } case NODE_TEXCO_NORMAL: { - data = ccl_fetch(sd, N); + data = sd->N; object_inverse_normal_transform(kg, sd, &data); break; } case NODE_TEXCO_CAMERA: { Transform tfm = kernel_data.cam.worldtocamera; - if(ccl_fetch(sd, object) != OBJECT_NONE) - data = transform_point(&tfm, ccl_fetch(sd, P)); + if(sd->object != OBJECT_NONE) + data = transform_point(&tfm, sd->P); else - data = transform_point(&tfm, ccl_fetch(sd, P) + camera_position(kg)); + data = transform_point(&tfm, sd->P + camera_position(kg)); break; } case NODE_TEXCO_WINDOW: { - if((path_flag & PATH_RAY_CAMERA) && ccl_fetch(sd, object) == OBJECT_NONE && kernel_data.cam.type == CAMERA_ORTHOGRAPHIC) - data = camera_world_to_ndc(kg, sd, ccl_fetch(sd, ray_P)); + if((path_flag & PATH_RAY_CAMERA) && sd->object == OBJECT_NONE && kernel_data.cam.type == CAMERA_ORTHOGRAPHIC) + data = camera_world_to_ndc(kg, sd, sd->ray_P); else - data = camera_world_to_ndc(kg, sd, ccl_fetch(sd, P)); + data = camera_world_to_ndc(kg, sd, sd->P); data.z = 0.0f; break; } case NODE_TEXCO_REFLECTION: { - if(ccl_fetch(sd, object) != OBJECT_NONE) - data = 2.0f*dot(ccl_fetch(sd, N), ccl_fetch(sd, I))*ccl_fetch(sd, N) - ccl_fetch(sd, I); + if(sd->object != OBJECT_NONE) + data = 2.0f*dot(sd->N, sd->I)*sd->N - sd->I; else - data = ccl_fetch(sd, I); + data = sd->I; break; } case NODE_TEXCO_DUPLI_GENERATED: { - data = object_dupli_generated(kg, ccl_fetch(sd, object)); + data = object_dupli_generated(kg, sd->object); break; } case NODE_TEXCO_DUPLI_UV: { - data = object_dupli_uv(kg, ccl_fetch(sd, object)); + data = object_dupli_uv(kg, sd->object); break; } case NODE_TEXCO_VOLUME_GENERATED: { - data = ccl_fetch(sd, P); + data = sd->P; #ifdef __VOLUME__ - if(ccl_fetch(sd, object) != OBJECT_NONE) + if(sd->object != OBJECT_NONE) data = volume_normalized_position(kg, sd, data); #endif break; @@ -112,9 +112,9 @@ ccl_device void svm_node_tex_coord_bump_dx(KernelGlobals *kg, switch(type) { case NODE_TEXCO_OBJECT: { - data = ccl_fetch(sd, P) + ccl_fetch(sd, dP).dx; + data = sd->P + sd->dP.dx; if(node.w == 0) { - if(ccl_fetch(sd, object) != OBJECT_NONE) { + if(sd->object != OBJECT_NONE) { object_inverse_position_transform(kg, sd, &data); } } @@ -129,47 +129,47 @@ ccl_device void svm_node_tex_coord_bump_dx(KernelGlobals *kg, break; } case NODE_TEXCO_NORMAL: { - data = ccl_fetch(sd, N); + data = sd->N; object_inverse_normal_transform(kg, sd, &data); break; } case NODE_TEXCO_CAMERA: { Transform tfm = kernel_data.cam.worldtocamera; - if(ccl_fetch(sd, object) != OBJECT_NONE) - data = transform_point(&tfm, ccl_fetch(sd, P) + ccl_fetch(sd, dP).dx); + if(sd->object != OBJECT_NONE) + data = transform_point(&tfm, sd->P + sd->dP.dx); else - data = transform_point(&tfm, ccl_fetch(sd, P) + ccl_fetch(sd, dP).dx + camera_position(kg)); + data = transform_point(&tfm, sd->P + sd->dP.dx + camera_position(kg)); break; } case NODE_TEXCO_WINDOW: { - if((path_flag & PATH_RAY_CAMERA) && ccl_fetch(sd, object) == OBJECT_NONE && kernel_data.cam.type == CAMERA_ORTHOGRAPHIC) - data = camera_world_to_ndc(kg, sd, ccl_fetch(sd, ray_P) + ccl_fetch(sd, ray_dP).dx); + if((path_flag & PATH_RAY_CAMERA) && sd->object == OBJECT_NONE && kernel_data.cam.type == CAMERA_ORTHOGRAPHIC) + data = camera_world_to_ndc(kg, sd, sd->ray_P + sd->ray_dP.dx); else - data = camera_world_to_ndc(kg, sd, ccl_fetch(sd, P) + ccl_fetch(sd, dP).dx); + data = camera_world_to_ndc(kg, sd, sd->P + sd->dP.dx); data.z = 0.0f; break; } case NODE_TEXCO_REFLECTION: { - if(ccl_fetch(sd, object) != OBJECT_NONE) - data = 2.0f*dot(ccl_fetch(sd, N), ccl_fetch(sd, I))*ccl_fetch(sd, N) - ccl_fetch(sd, I); + if(sd->object != OBJECT_NONE) + data = 2.0f*dot(sd->N, sd->I)*sd->N - sd->I; else - data = ccl_fetch(sd, I); + data = sd->I; break; } case NODE_TEXCO_DUPLI_GENERATED: { - data = object_dupli_generated(kg, ccl_fetch(sd, object)); + data = object_dupli_generated(kg, sd->object); break; } case NODE_TEXCO_DUPLI_UV: { - data = object_dupli_uv(kg, ccl_fetch(sd, object)); + data = object_dupli_uv(kg, sd->object); break; } case NODE_TEXCO_VOLUME_GENERATED: { - data = ccl_fetch(sd, P) + ccl_fetch(sd, dP).dx; + data = sd->P + sd->dP.dx; #ifdef __VOLUME__ - if(ccl_fetch(sd, object) != OBJECT_NONE) + if(sd->object != OBJECT_NONE) data = volume_normalized_position(kg, sd, data); #endif break; @@ -196,9 +196,9 @@ ccl_device void svm_node_tex_coord_bump_dy(KernelGlobals *kg, switch(type) { case NODE_TEXCO_OBJECT: { - data = ccl_fetch(sd, P) + ccl_fetch(sd, dP).dy; + data = sd->P + sd->dP.dy; if(node.w == 0) { - if(ccl_fetch(sd, object) != OBJECT_NONE) { + if(sd->object != OBJECT_NONE) { object_inverse_position_transform(kg, sd, &data); } } @@ -213,47 +213,47 @@ ccl_device void svm_node_tex_coord_bump_dy(KernelGlobals *kg, break; } case NODE_TEXCO_NORMAL: { - data = ccl_fetch(sd, N); + data = sd->N; object_inverse_normal_transform(kg, sd, &data); break; } case NODE_TEXCO_CAMERA: { Transform tfm = kernel_data.cam.worldtocamera; - if(ccl_fetch(sd, object) != OBJECT_NONE) - data = transform_point(&tfm, ccl_fetch(sd, P) + ccl_fetch(sd, dP).dy); + if(sd->object != OBJECT_NONE) + data = transform_point(&tfm, sd->P + sd->dP.dy); else - data = transform_point(&tfm, ccl_fetch(sd, P) + ccl_fetch(sd, dP).dy + camera_position(kg)); + data = transform_point(&tfm, sd->P + sd->dP.dy + camera_position(kg)); break; } case NODE_TEXCO_WINDOW: { - if((path_flag & PATH_RAY_CAMERA) && ccl_fetch(sd, object) == OBJECT_NONE && kernel_data.cam.type == CAMERA_ORTHOGRAPHIC) - data = camera_world_to_ndc(kg, sd, ccl_fetch(sd, ray_P) + ccl_fetch(sd, ray_dP).dy); + if((path_flag & PATH_RAY_CAMERA) && sd->object == OBJECT_NONE && kernel_data.cam.type == CAMERA_ORTHOGRAPHIC) + data = camera_world_to_ndc(kg, sd, sd->ray_P + sd->ray_dP.dy); else - data = camera_world_to_ndc(kg, sd, ccl_fetch(sd, P) + ccl_fetch(sd, dP).dy); + data = camera_world_to_ndc(kg, sd, sd->P + sd->dP.dy); data.z = 0.0f; break; } case NODE_TEXCO_REFLECTION: { - if(ccl_fetch(sd, object) != OBJECT_NONE) - data = 2.0f*dot(ccl_fetch(sd, N), ccl_fetch(sd, I))*ccl_fetch(sd, N) - ccl_fetch(sd, I); + if(sd->object != OBJECT_NONE) + data = 2.0f*dot(sd->N, sd->I)*sd->N - sd->I; else - data = ccl_fetch(sd, I); + data = sd->I; break; } case NODE_TEXCO_DUPLI_GENERATED: { - data = object_dupli_generated(kg, ccl_fetch(sd, object)); + data = object_dupli_generated(kg, sd->object); break; } case NODE_TEXCO_DUPLI_UV: { - data = object_dupli_uv(kg, ccl_fetch(sd, object)); + data = object_dupli_uv(kg, sd->object); break; } case NODE_TEXCO_VOLUME_GENERATED: { - data = ccl_fetch(sd, P) + ccl_fetch(sd, dP).dy; + data = sd->P + sd->dP.dy; #ifdef __VOLUME__ - if(ccl_fetch(sd, object) != OBJECT_NONE) + if(sd->object != OBJECT_NONE) data = volume_normalized_position(kg, sd, data); #endif break; @@ -274,12 +274,12 @@ ccl_device void svm_node_normal_map(KernelGlobals *kg, ShaderData *sd, float *st float3 color = stack_load_float3(stack, color_offset); color = 2.0f*make_float3(color.x - 0.5f, color.y - 0.5f, color.z - 0.5f); - bool is_backfacing = (ccl_fetch(sd, flag) & SD_BACKFACING) != 0; + bool is_backfacing = (sd->flag & SD_BACKFACING) != 0; float3 N; if(space == NODE_NORMAL_MAP_TANGENT) { /* tangent space */ - if(ccl_fetch(sd, object) == OBJECT_NONE) { + if(sd->object == OBJECT_NONE) { stack_store_float3(stack, normal_offset, make_float3(0.0f, 0.0f, 0.0f)); return; } @@ -299,11 +299,11 @@ ccl_device void svm_node_normal_map(KernelGlobals *kg, ShaderData *sd, float *st float sign = primitive_attribute_float(kg, sd, attr_sign, NULL, NULL); float3 normal; - if(ccl_fetch(sd, shader) & SHADER_SMOOTH_NORMAL) { + if(sd->shader & SHADER_SMOOTH_NORMAL) { normal = primitive_attribute_float3(kg, sd, attr_normal, NULL, NULL); } else { - normal = ccl_fetch(sd, Ng); + normal = sd->Ng; /* the normal is already inverted, which is too soon for the math here */ if(is_backfacing) { @@ -345,11 +345,11 @@ ccl_device void svm_node_normal_map(KernelGlobals *kg, ShaderData *sd, float *st if(strength != 1.0f) { strength = max(strength, 0.0f); - N = safe_normalize(ccl_fetch(sd, N) + (N - ccl_fetch(sd, N))*strength); + N = safe_normalize(sd->N + (N - sd->N)*strength); } if(is_zero(N)) { - N = ccl_fetch(sd, N); + N = sd->N; } stack_store_float3(stack, normal_offset, N); @@ -377,7 +377,7 @@ ccl_device void svm_node_tangent(KernelGlobals *kg, ShaderData *sd, float *stack float3 generated; if(desc.offset == ATTR_STD_NOT_FOUND) - generated = ccl_fetch(sd, P); + generated = sd->P; else generated = primitive_attribute_float3(kg, sd, desc, NULL, NULL); @@ -390,7 +390,7 @@ ccl_device void svm_node_tangent(KernelGlobals *kg, ShaderData *sd, float *stack } object_normal_transform(kg, sd, &tangent); - tangent = cross(ccl_fetch(sd, N), normalize(cross(tangent, ccl_fetch(sd, N)))); + tangent = cross(sd->N, normalize(cross(tangent, sd->N))); stack_store_float3(stack, tangent_offset, tangent); } diff --git a/intern/cycles/kernel/svm/svm_vector_transform.h b/intern/cycles/kernel/svm/svm_vector_transform.h index 4c32130d06d..4e92f27acdb 100644 --- a/intern/cycles/kernel/svm/svm_vector_transform.h +++ b/intern/cycles/kernel/svm/svm_vector_transform.h @@ -33,7 +33,7 @@ ccl_device void svm_node_vector_transform(KernelGlobals *kg, ShaderData *sd, flo NodeVectorTransformConvertSpace to = (NodeVectorTransformConvertSpace)ito; Transform tfm; - bool is_object = (ccl_fetch(sd, object) != OBJECT_NONE); + bool is_object = (sd->object != OBJECT_NONE); bool is_direction = (type == NODE_VECTOR_TRANSFORM_TYPE_VECTOR || type == NODE_VECTOR_TRANSFORM_TYPE_NORMAL); /* From world */ diff --git a/intern/cycles/kernel/svm/svm_voxel.h b/intern/cycles/kernel/svm/svm_voxel.h index a8b3604a8a7..9e826c8c23f 100644 --- a/intern/cycles/kernel/svm/svm_voxel.h +++ b/intern/cycles/kernel/svm/svm_voxel.h @@ -46,7 +46,7 @@ ccl_device void svm_node_tex_voxel(KernelGlobals *kg, # if defined(__KERNEL_CUDA__) # if __CUDA_ARCH__ >= 300 CUtexObject tex = kernel_tex_fetch(__bindless_mapping, id); - if(id < 2048) /* TODO(dingto): Make this a variable */ + if(id < TEX_START_HALF4_CUDA_KEPLER) r = kernel_tex_image_interp_3d_float4(tex, co.x, co.y, co.z); else { float f = kernel_tex_image_interp_3d_float(tex, co.x, co.y, co.z); diff --git a/intern/cycles/kernel/svm/svm_wireframe.h b/intern/cycles/kernel/svm/svm_wireframe.h index 87e40791333..3c6353c8001 100644 --- a/intern/cycles/kernel/svm/svm_wireframe.h +++ b/intern/cycles/kernel/svm/svm_wireframe.h @@ -41,9 +41,9 @@ ccl_device_inline float wireframe(KernelGlobals *kg, float3 *P) { #ifdef __HAIR__ - if(ccl_fetch(sd, prim) != PRIM_NONE && ccl_fetch(sd, type) & PRIMITIVE_ALL_TRIANGLE) + if(sd->prim != PRIM_NONE && sd->type & PRIMITIVE_ALL_TRIANGLE) #else - if(ccl_fetch(sd, prim) != PRIM_NONE) + if(sd->prim != PRIM_NONE) #endif { float3 Co[3]; @@ -52,12 +52,12 @@ ccl_device_inline float wireframe(KernelGlobals *kg, /* Triangles */ int np = 3; - if(ccl_fetch(sd, type) & PRIMITIVE_TRIANGLE) - triangle_vertices(kg, ccl_fetch(sd, prim), Co); + if(sd->type & PRIMITIVE_TRIANGLE) + triangle_vertices(kg, sd->prim, Co); else - motion_triangle_vertices(kg, ccl_fetch(sd, object), ccl_fetch(sd, prim), ccl_fetch(sd, time), Co); + motion_triangle_vertices(kg, sd->object, sd->prim, sd->time, Co); - if(!(ccl_fetch(sd, object_flag) & SD_OBJECT_TRANSFORM_APPLIED)) { + if(!(sd->object_flag & SD_OBJECT_TRANSFORM_APPLIED)) { object_position_transform(kg, sd, &Co[0]); object_position_transform(kg, sd, &Co[1]); object_position_transform(kg, sd, &Co[2]); @@ -66,8 +66,8 @@ ccl_device_inline float wireframe(KernelGlobals *kg, if(pixel_size) { // Project the derivatives of P to the viewing plane defined // by I so we have a measure of how big is a pixel at this point - float pixelwidth_x = len(ccl_fetch(sd, dP).dx - dot(ccl_fetch(sd, dP).dx, ccl_fetch(sd, I)) * ccl_fetch(sd, I)); - float pixelwidth_y = len(ccl_fetch(sd, dP).dy - dot(ccl_fetch(sd, dP).dy, ccl_fetch(sd, I)) * ccl_fetch(sd, I)); + float pixelwidth_x = len(sd->dP.dx - dot(sd->dP.dx, sd->I) * sd->I); + float pixelwidth_y = len(sd->dP.dy - dot(sd->dP.dy, sd->I) * sd->I); // Take the average of both axis' length pixelwidth = (pixelwidth_x + pixelwidth_y) * 0.5f; } @@ -113,20 +113,20 @@ ccl_device void svm_node_wireframe(KernelGlobals *kg, * With OpenCL 2.0 it's possible to avoid this change, but for until * then we'll be living with such an exception. */ - float3 P = ccl_fetch(sd, P); + float3 P = sd->P; float f = wireframe(kg, sd, size, pixel_size, &P); #else - float f = wireframe(kg, sd, size, pixel_size, &ccl_fetch(sd, P)); + float f = wireframe(kg, sd, size, pixel_size, &sd->P); #endif /* TODO(sergey): Think of faster way to calculate derivatives. */ if(bump_offset == NODE_BUMP_OFFSET_DX) { - float3 Px = ccl_fetch(sd, P) - ccl_fetch(sd, dP).dx; - f += (f - wireframe(kg, sd, size, pixel_size, &Px)) / len(ccl_fetch(sd, dP).dx); + float3 Px = sd->P - sd->dP.dx; + f += (f - wireframe(kg, sd, size, pixel_size, &Px)) / len(sd->dP.dx); } else if(bump_offset == NODE_BUMP_OFFSET_DY) { - float3 Py = ccl_fetch(sd, P) - ccl_fetch(sd, dP).dy; - f += (f - wireframe(kg, sd, size, pixel_size, &Py)) / len(ccl_fetch(sd, dP).dy); + float3 Py = sd->P - sd->dP.dy; + f += (f - wireframe(kg, sd, size, pixel_size, &Py)) / len(sd->dP.dy); } if(stack_valid(out_fac)) diff --git a/intern/cycles/render/CMakeLists.txt b/intern/cycles/render/CMakeLists.txt index 8eaa9de3874..17ac66644e2 100644 --- a/intern/cycles/render/CMakeLists.txt +++ b/intern/cycles/render/CMakeLists.txt @@ -1,14 +1,6 @@ set(INC - . - ../device - ../graph - ../kernel - ../kernel/svm - ../kernel/osl - ../bvh - ../subd - ../util + .. ../../glew-mx ) diff --git a/intern/cycles/render/attribute.cpp b/intern/cycles/render/attribute.cpp index c0d429a583c..e157a385904 100644 --- a/intern/cycles/render/attribute.cpp +++ b/intern/cycles/render/attribute.cpp @@ -14,13 +14,13 @@ * limitations under the License. */ -#include "image.h" -#include "mesh.h" -#include "attribute.h" +#include "render/image.h" +#include "render/mesh.h" +#include "render/attribute.h" -#include "util_debug.h" -#include "util_foreach.h" -#include "util_transform.h" +#include "util/util_debug.h" +#include "util/util_foreach.h" +#include "util/util_transform.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/render/attribute.h b/intern/cycles/render/attribute.h index f4538c76369..a64eb6542d5 100644 --- a/intern/cycles/render/attribute.h +++ b/intern/cycles/render/attribute.h @@ -17,12 +17,12 @@ #ifndef __ATTRIBUTE_H__ #define __ATTRIBUTE_H__ -#include "kernel_types.h" +#include "kernel/kernel_types.h" -#include "util_list.h" -#include "util_param.h" -#include "util_types.h" -#include "util_vector.h" +#include "util/util_list.h" +#include "util/util_param.h" +#include "util/util_types.h" +#include "util/util_vector.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/render/background.cpp b/intern/cycles/render/background.cpp index 8d7d7b847fd..930debe1e33 100644 --- a/intern/cycles/render/background.cpp +++ b/intern/cycles/render/background.cpp @@ -14,17 +14,17 @@ * limitations under the License. */ -#include "background.h" -#include "device.h" -#include "integrator.h" -#include "graph.h" -#include "nodes.h" -#include "scene.h" -#include "shader.h" - -#include "util_foreach.h" -#include "util_math.h" -#include "util_types.h" +#include "render/background.h" +#include "device/device.h" +#include "render/integrator.h" +#include "render/graph.h" +#include "render/nodes.h" +#include "render/scene.h" +#include "render/shader.h" + +#include "util/util_foreach.h" +#include "util/util_math.h" +#include "util/util_types.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/render/background.h b/intern/cycles/render/background.h index 8029c6a9e80..db20b6ebf87 100644 --- a/intern/cycles/render/background.h +++ b/intern/cycles/render/background.h @@ -17,9 +17,9 @@ #ifndef __BACKGROUND_H__ #define __BACKGROUND_H__ -#include "node.h" +#include "graph/node.h" -#include "util_types.h" +#include "util/util_types.h" CCL_NAMESPACE_BEGIN @@ -30,7 +30,7 @@ class Shader; class Background : public Node { public: - NODE_DECLARE; + NODE_DECLARE float ao_factor; float ao_distance; diff --git a/intern/cycles/render/bake.cpp b/intern/cycles/render/bake.cpp index d9a297002c6..c0fcd517390 100644 --- a/intern/cycles/render/bake.cpp +++ b/intern/cycles/render/bake.cpp @@ -14,8 +14,8 @@ * limitations under the License. */ -#include "bake.h" -#include "integrator.h" +#include "render/bake.h" +#include "render/integrator.h" CCL_NAMESPACE_BEGIN @@ -171,9 +171,9 @@ bool BakeManager::bake(Device *device, DeviceScene *dscene, Scene *scene, Progre /* needs to be up to data for attribute access */ device->const_copy_to("__data", &dscene->data, sizeof(dscene->data)); - device->mem_alloc(d_input, MEM_READ_ONLY); + device->mem_alloc("bake_input", d_input, MEM_READ_ONLY); device->mem_copy_to(d_input); - device->mem_alloc(d_output, MEM_READ_WRITE); + device->mem_alloc("bake_output", d_output, MEM_READ_WRITE); DeviceTask task(DeviceTask::SHADER); task.shader_input = d_input.device_pointer; diff --git a/intern/cycles/render/bake.h b/intern/cycles/render/bake.h index 25f5eb3c897..ceb94cfb682 100644 --- a/intern/cycles/render/bake.h +++ b/intern/cycles/render/bake.h @@ -17,11 +17,11 @@ #ifndef __BAKE_H__ #define __BAKE_H__ -#include "device.h" -#include "scene.h" +#include "device/device.h" +#include "render/scene.h" -#include "util_progress.h" -#include "util_vector.h" +#include "util/util_progress.h" +#include "util/util_vector.h" CCL_NAMESPACE_BEGIN @@ -73,7 +73,7 @@ public: bool need_update; - int total_pixel_samples; + size_t total_pixel_samples; private: BakeData *m_bake_data; diff --git a/intern/cycles/render/buffers.cpp b/intern/cycles/render/buffers.cpp index f1692712d61..fe2c2e78926 100644 --- a/intern/cycles/render/buffers.cpp +++ b/intern/cycles/render/buffers.cpp @@ -16,17 +16,17 @@ #include <stdlib.h> -#include "buffers.h" -#include "device.h" - -#include "util_debug.h" -#include "util_foreach.h" -#include "util_hash.h" -#include "util_image.h" -#include "util_math.h" -#include "util_opengl.h" -#include "util_time.h" -#include "util_types.h" +#include "render/buffers.h" +#include "device/device.h" + +#include "util/util_debug.h" +#include "util/util_foreach.h" +#include "util/util_hash.h" +#include "util/util_image.h" +#include "util/util_math.h" +#include "util/util_opengl.h" +#include "util/util_time.h" +#include "util/util_types.h" CCL_NAMESPACE_BEGIN @@ -129,13 +129,13 @@ void RenderBuffers::reset(Device *device, BufferParams& params_) /* allocate buffer */ buffer.resize(params.width*params.height*params.get_passes_size()); - device->mem_alloc(buffer, MEM_READ_WRITE); + device->mem_alloc("render_buffer", buffer, MEM_READ_WRITE); device->mem_zero(buffer); /* allocate rng state */ rng_state.resize(params.width, params.height); - device->mem_alloc(rng_state, MEM_READ_WRITE); + device->mem_alloc("rng_state", rng_state, MEM_READ_WRITE); } bool RenderBuffers::copy_from_device() diff --git a/intern/cycles/render/buffers.h b/intern/cycles/render/buffers.h index c9c2a21079a..5c78971678a 100644 --- a/intern/cycles/render/buffers.h +++ b/intern/cycles/render/buffers.h @@ -17,16 +17,16 @@ #ifndef __BUFFERS_H__ #define __BUFFERS_H__ -#include "device_memory.h" +#include "device/device_memory.h" -#include "film.h" +#include "render/film.h" -#include "kernel_types.h" +#include "kernel/kernel_types.h" -#include "util_half.h" -#include "util_string.h" -#include "util_thread.h" -#include "util_types.h" +#include "util/util_half.h" +#include "util/util_string.h" +#include "util/util_thread.h" +#include "util/util_types.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/render/camera.cpp b/intern/cycles/render/camera.cpp index c8c51ec96d2..83ff8a10618 100644 --- a/intern/cycles/render/camera.cpp +++ b/intern/cycles/render/camera.cpp @@ -14,18 +14,18 @@ * limitations under the License. */ -#include "camera.h" -#include "mesh.h" -#include "object.h" -#include "scene.h" -#include "tables.h" - -#include "device.h" - -#include "util_foreach.h" -#include "util_function.h" -#include "util_math_cdf.h" -#include "util_vector.h" +#include "render/camera.h" +#include "render/mesh.h" +#include "render/object.h" +#include "render/scene.h" +#include "render/tables.h" + +#include "device/device.h" + +#include "util/util_foreach.h" +#include "util/util_function.h" +#include "util/util_math_cdf.h" +#include "util/util_vector.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/render/camera.h b/intern/cycles/render/camera.h index 141ef9cccef..dd6b831b347 100644 --- a/intern/cycles/render/camera.h +++ b/intern/cycles/render/camera.h @@ -17,13 +17,13 @@ #ifndef __CAMERA_H__ #define __CAMERA_H__ -#include "kernel_types.h" +#include "kernel/kernel_types.h" -#include "node.h" +#include "graph/node.h" -#include "util_boundbox.h" -#include "util_transform.h" -#include "util_types.h" +#include "util/util_boundbox.h" +#include "util/util_transform.h" +#include "util/util_types.h" CCL_NAMESPACE_BEGIN @@ -39,7 +39,7 @@ class Scene; class Camera : public Node { public: - NODE_DECLARE; + NODE_DECLARE /* Specifies an offset for the shutter's time interval. */ enum MotionPosition { diff --git a/intern/cycles/render/constant_fold.cpp b/intern/cycles/render/constant_fold.cpp index b7f25663bc3..2569d9eec27 100644 --- a/intern/cycles/render/constant_fold.cpp +++ b/intern/cycles/render/constant_fold.cpp @@ -14,11 +14,11 @@ * limitations under the License. */ -#include "constant_fold.h" -#include "graph.h" +#include "render/constant_fold.h" +#include "render/graph.h" -#include "util_foreach.h" -#include "util_logging.h" +#include "util/util_foreach.h" +#include "util/util_logging.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/render/constant_fold.h b/intern/cycles/render/constant_fold.h index 7962698319f..33f93b8c0ab 100644 --- a/intern/cycles/render/constant_fold.h +++ b/intern/cycles/render/constant_fold.h @@ -17,8 +17,8 @@ #ifndef __CONSTANT_FOLD_H__ #define __CONSTANT_FOLD_H__ -#include "util_types.h" -#include "svm_types.h" +#include "util/util_types.h" +#include "kernel/svm/svm_types.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/render/curves.cpp b/intern/cycles/render/curves.cpp index f671eb19cae..4c085b928fb 100644 --- a/intern/cycles/render/curves.cpp +++ b/intern/cycles/render/curves.cpp @@ -14,16 +14,16 @@ * limitations under the License. */ -#include "device.h" -#include "curves.h" -#include "mesh.h" -#include "object.h" -#include "scene.h" - -#include "util_foreach.h" -#include "util_map.h" -#include "util_progress.h" -#include "util_vector.h" +#include "device/device.h" +#include "render/curves.h" +#include "render/mesh.h" +#include "render/object.h" +#include "render/scene.h" + +#include "util/util_foreach.h" +#include "util/util_map.h" +#include "util/util_progress.h" +#include "util/util_vector.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/render/curves.h b/intern/cycles/render/curves.h index e41967eebf5..8834764bd63 100644 --- a/intern/cycles/render/curves.h +++ b/intern/cycles/render/curves.h @@ -17,8 +17,8 @@ #ifndef __CURVES_H__ #define __CURVES_H__ -#include "util_types.h" -#include "util_vector.h" +#include "util/util_types.h" +#include "util/util_vector.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/render/film.cpp b/intern/cycles/render/film.cpp index 923252bb375..7809f4345f1 100644 --- a/intern/cycles/render/film.cpp +++ b/intern/cycles/render/film.cpp @@ -14,19 +14,19 @@ * limitations under the License. */ -#include "camera.h" -#include "device.h" -#include "film.h" -#include "integrator.h" -#include "mesh.h" -#include "scene.h" -#include "tables.h" - -#include "util_algorithm.h" -#include "util_debug.h" -#include "util_foreach.h" -#include "util_math.h" -#include "util_math_cdf.h" +#include "render/camera.h" +#include "device/device.h" +#include "render/film.h" +#include "render/integrator.h" +#include "render/mesh.h" +#include "render/scene.h" +#include "render/tables.h" + +#include "util/util_algorithm.h" +#include "util/util_debug.h" +#include "util/util_foreach.h" +#include "util/util_math.h" +#include "util/util_math_cdf.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/render/film.h b/intern/cycles/render/film.h index 9fa51c51f52..83c941d5c57 100644 --- a/intern/cycles/render/film.h +++ b/intern/cycles/render/film.h @@ -17,12 +17,12 @@ #ifndef __FILM_H__ #define __FILM_H__ -#include "util_string.h" -#include "util_vector.h" +#include "util/util_string.h" +#include "util/util_vector.h" -#include "kernel_types.h" +#include "kernel/kernel_types.h" -#include "node.h" +#include "graph/node.h" CCL_NAMESPACE_BEGIN @@ -53,7 +53,7 @@ public: class Film : public Node { public: - NODE_DECLARE; + NODE_DECLARE float exposure; array<Pass> passes; diff --git a/intern/cycles/render/graph.cpp b/intern/cycles/render/graph.cpp index 52c94ec2716..8e61daab49b 100644 --- a/intern/cycles/render/graph.cpp +++ b/intern/cycles/render/graph.cpp @@ -14,17 +14,18 @@ * limitations under the License. */ -#include "attribute.h" -#include "graph.h" -#include "nodes.h" -#include "shader.h" -#include "constant_fold.h" - -#include "util_algorithm.h" -#include "util_debug.h" -#include "util_foreach.h" -#include "util_queue.h" -#include "util_logging.h" +#include "render/attribute.h" +#include "render/graph.h" +#include "render/nodes.h" +#include "render/scene.h" +#include "render/shader.h" +#include "render/constant_fold.h" + +#include "util/util_algorithm.h" +#include "util/util_debug.h" +#include "util/util_foreach.h" +#include "util/util_queue.h" +#include "util/util_logging.h" CCL_NAMESPACE_BEGIN @@ -195,6 +196,7 @@ bool ShaderNode::equals(const ShaderNode& other) ShaderGraph::ShaderGraph() { finalized = false; + simplified = false; num_node_ids = 0; add(new OutputNode()); } @@ -207,6 +209,8 @@ ShaderGraph::~ShaderGraph() ShaderNode *ShaderGraph::add(ShaderNode *node) { assert(!finalized); + simplified = false; + node->id = num_node_ids++; nodes.push_back(node); return node; @@ -241,6 +245,7 @@ void ShaderGraph::connect(ShaderOutput *from, ShaderInput *to) { assert(!finalized); assert(from && to); + simplified = false; if(to->link) { fprintf(stderr, "Cycles shader graph connect: input already connected.\n"); @@ -273,6 +278,7 @@ void ShaderGraph::connect(ShaderOutput *from, ShaderInput *to) void ShaderGraph::disconnect(ShaderOutput *from) { assert(!finalized); + simplified = false; foreach(ShaderInput *sock, from->links) { sock->link = NULL; @@ -285,6 +291,7 @@ void ShaderGraph::disconnect(ShaderInput *to) { assert(!finalized); assert(to->link); + simplified = false; ShaderOutput *from = to->link; @@ -294,6 +301,8 @@ void ShaderGraph::disconnect(ShaderInput *to) void ShaderGraph::relink(ShaderNode *node, ShaderOutput *from, ShaderOutput *to) { + simplified = false; + /* Copy because disconnect modifies this list */ vector<ShaderInput*> outputs = from->links; @@ -310,9 +319,19 @@ void ShaderGraph::relink(ShaderNode *node, ShaderOutput *from, ShaderOutput *to) } } +void ShaderGraph::simplify(Scene *scene) +{ + if(!simplified) { + default_inputs(scene->shader_manager->use_osl()); + clean(scene); + refine_bump_nodes(); + + simplified = true; + } +} + void ShaderGraph::finalize(Scene *scene, bool do_bump, - bool do_osl, bool do_simplify, bool bump_in_object_space) { @@ -322,9 +341,7 @@ void ShaderGraph::finalize(Scene *scene, * modified afterwards. */ if(!finalized) { - default_inputs(do_osl); - clean(scene); - refine_bump_nodes(); + simplify(scene); if(do_bump) bump_from_displacement(bump_in_object_space); diff --git a/intern/cycles/render/graph.h b/intern/cycles/render/graph.h index 780fdf49ca4..09932695d1f 100644 --- a/intern/cycles/render/graph.h +++ b/intern/cycles/render/graph.h @@ -17,17 +17,17 @@ #ifndef __GRAPH_H__ #define __GRAPH_H__ -#include "node.h" -#include "node_type.h" +#include "graph/node.h" +#include "graph/node_type.h" -#include "kernel_types.h" +#include "kernel/kernel_types.h" -#include "util_list.h" -#include "util_map.h" -#include "util_param.h" -#include "util_set.h" -#include "util_types.h" -#include "util_vector.h" +#include "util/util_list.h" +#include "util/util_map.h" +#include "util/util_param.h" +#include "util/util_set.h" +#include "util/util_types.h" +#include "util/util_vector.h" CCL_NAMESPACE_BEGIN @@ -201,14 +201,14 @@ public: /* Node definition utility macros */ #define SHADER_NODE_CLASS(type) \ - NODE_DECLARE; \ + NODE_DECLARE \ type(); \ virtual ShaderNode *clone() const { return new type(*this); } \ virtual void compile(SVMCompiler& compiler); \ virtual void compile(OSLCompiler& compiler); \ #define SHADER_NODE_NO_CLONE_CLASS(type) \ - NODE_DECLARE; \ + NODE_DECLARE \ type(); \ virtual void compile(SVMCompiler& compiler); \ virtual void compile(OSLCompiler& compiler); \ @@ -240,6 +240,7 @@ public: list<ShaderNode*> nodes; size_t num_node_ids; bool finalized; + bool simplified; ShaderGraph(); ~ShaderGraph(); @@ -255,9 +256,9 @@ public: void relink(ShaderNode *node, ShaderOutput *from, ShaderOutput *to); void remove_proxy_nodes(); + void simplify(Scene *scene); void finalize(Scene *scene, bool do_bump = false, - bool do_osl = false, bool do_simplify = false, bool bump_in_object_space = false); diff --git a/intern/cycles/render/image.cpp b/intern/cycles/render/image.cpp index fd8a1262208..a8c4f446bea 100644 --- a/intern/cycles/render/image.cpp +++ b/intern/cycles/render/image.cpp @@ -14,15 +14,15 @@ * limitations under the License. */ -#include "device.h" -#include "image.h" -#include "scene.h" +#include "device/device.h" +#include "render/image.h" +#include "render/scene.h" -#include "util_foreach.h" -#include "util_logging.h" -#include "util_path.h" -#include "util_progress.h" -#include "util_texture.h" +#include "util/util_foreach.h" +#include "util/util_logging.h" +#include "util/util_path.h" +#include "util/util_progress.h" +#include "util/util_texture.h" #ifdef WITH_OSL #include <OSL/oslexec.h> @@ -156,6 +156,16 @@ ImageManager::ImageDataType ImageManager::get_image_metadata(const string& filen } } + /* Perform preliminary checks, with meaningful logging. */ + if(!path_exists(filename)) { + VLOG(1) << "File '" << filename << "' does not exist."; + return IMAGE_DATA_TYPE_BYTE4; + } + if(path_is_directory(filename)) { + VLOG(1) << "File '" << filename << "' is a directory, can't use as image."; + return IMAGE_DATA_TYPE_BYTE4; + } + ImageInput *in = ImageInput::create(filename); if(in) { @@ -285,9 +295,8 @@ int ImageManager::add_image(const string& filename, thread_scoped_lock device_lock(device_mutex); - /* Do we have a float? */ - if(type == IMAGE_DATA_TYPE_FLOAT || type == IMAGE_DATA_TYPE_FLOAT4) - is_float = true; + /* Check whether it's a float texture. */ + is_float = (type == IMAGE_DATA_TYPE_FLOAT || type == IMAGE_DATA_TYPE_FLOAT4); /* No single channel and half textures on CUDA (Fermi) and no half on OpenCL, use available slots */ if((type == IMAGE_DATA_TYPE_FLOAT || @@ -433,6 +442,11 @@ bool ImageManager::file_load_image_generic(Image *img, ImageInput **in, int &wid return false; if(!img->builtin_data) { + /* NOTE: Error logging is done in meta data acquisition. */ + if(!path_exists(img->filename) || path_is_directory(img->filename)) { + return false; + } + /* load image from file through OIIO */ *in = ImageInput::create(img->filename); diff --git a/intern/cycles/render/image.h b/intern/cycles/render/image.h index 494c74f0cdd..996b5a5b65f 100644 --- a/intern/cycles/render/image.h +++ b/intern/cycles/render/image.h @@ -17,13 +17,13 @@ #ifndef __IMAGE_H__ #define __IMAGE_H__ -#include "device.h" -#include "device_memory.h" +#include "device/device.h" +#include "device/device_memory.h" -#include "util_image.h" -#include "util_string.h" -#include "util_thread.h" -#include "util_vector.h" +#include "util/util_image.h" +#include "util/util_string.h" +#include "util/util_thread.h" +#include "util/util_vector.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/render/integrator.cpp b/intern/cycles/render/integrator.cpp index 1ab0f9874f2..a004bb5b856 100644 --- a/intern/cycles/render/integrator.cpp +++ b/intern/cycles/render/integrator.cpp @@ -14,16 +14,16 @@ * limitations under the License. */ -#include "device.h" -#include "integrator.h" -#include "film.h" -#include "light.h" -#include "scene.h" -#include "shader.h" -#include "sobol.h" - -#include "util_foreach.h" -#include "util_hash.h" +#include "device/device.h" +#include "render/integrator.h" +#include "render/film.h" +#include "render/light.h" +#include "render/scene.h" +#include "render/shader.h" +#include "render/sobol.h" + +#include "util/util_foreach.h" +#include "util/util_hash.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/render/integrator.h b/intern/cycles/render/integrator.h index 27fff4831e5..9501d7f8416 100644 --- a/intern/cycles/render/integrator.h +++ b/intern/cycles/render/integrator.h @@ -17,9 +17,9 @@ #ifndef __INTEGRATOR_H__ #define __INTEGRATOR_H__ -#include "kernel_types.h" +#include "kernel/kernel_types.h" -#include "node.h" +#include "graph/node.h" CCL_NAMESPACE_BEGIN @@ -29,7 +29,7 @@ class Scene; class Integrator : public Node { public: - NODE_DECLARE; + NODE_DECLARE int min_bounce; int max_bounce; diff --git a/intern/cycles/render/light.cpp b/intern/cycles/render/light.cpp index 2245c861d5a..4886dcd563f 100644 --- a/intern/cycles/render/light.cpp +++ b/intern/cycles/render/light.cpp @@ -14,19 +14,19 @@ * limitations under the License. */ -#include "background.h" -#include "device.h" -#include "integrator.h" -#include "film.h" -#include "light.h" -#include "mesh.h" -#include "object.h" -#include "scene.h" -#include "shader.h" - -#include "util_foreach.h" -#include "util_progress.h" -#include "util_logging.h" +#include "render/background.h" +#include "device/device.h" +#include "render/integrator.h" +#include "render/film.h" +#include "render/light.h" +#include "render/mesh.h" +#include "render/object.h" +#include "render/scene.h" +#include "render/shader.h" + +#include "util/util_foreach.h" +#include "util/util_progress.h" +#include "util/util_logging.h" CCL_NAMESPACE_BEGIN @@ -57,9 +57,9 @@ static void shade_background_pixels(Device *device, DeviceScene *dscene, int res device->const_copy_to("__data", &dscene->data, sizeof(dscene->data)); - device->mem_alloc(d_input, MEM_READ_ONLY); + device->mem_alloc("shade_background_pixels_input", d_input, MEM_READ_ONLY); device->mem_copy_to(d_input); - device->mem_alloc(d_output, MEM_WRITE_ONLY); + device->mem_alloc("shade_background_pixels_output", d_output, MEM_WRITE_ONLY); DeviceTask main_task(DeviceTask::SHADER); main_task.shader_input = d_input.device_pointer; @@ -486,10 +486,18 @@ static void background_cdf(int start, float2 *cond_cdf) { /* Conditional CDFs (rows, U direction). */ + /* NOTE: It is possible to have some NaN pixels on background + * which will ruin CDF causing wrong shading. We replace such + * pixels with black. + */ for(int i = start; i < end; i++) { float sin_theta = sinf(M_PI_F * (i + 0.5f) / res); float3 env_color = (*pixels)[i * res]; float ave_luminance = average(env_color); + /* TODO(sergey): Consider adding average_safe(). */ + if(!isfinite(ave_luminance)) { + ave_luminance = 0.0f; + } cond_cdf[i * cdf_count].x = ave_luminance * sin_theta; cond_cdf[i * cdf_count].y = 0.0f; @@ -497,6 +505,9 @@ static void background_cdf(int start, for(int j = 1; j < res; j++) { env_color = (*pixels)[i * res + j]; ave_luminance = average(env_color); + if(!isfinite(ave_luminance)) { + ave_luminance = 0.0f; + } cond_cdf[i * cdf_count + j].x = ave_luminance * sin_theta; cond_cdf[i * cdf_count + j].y = cond_cdf[i * cdf_count + j - 1].y + cond_cdf[i * cdf_count + j - 1].x / res; diff --git a/intern/cycles/render/light.h b/intern/cycles/render/light.h index f56530b6490..7e9014eb823 100644 --- a/intern/cycles/render/light.h +++ b/intern/cycles/render/light.h @@ -17,12 +17,12 @@ #ifndef __LIGHT_H__ #define __LIGHT_H__ -#include "kernel_types.h" +#include "kernel/kernel_types.h" -#include "node.h" +#include "graph/node.h" -#include "util_types.h" -#include "util_vector.h" +#include "util/util_types.h" +#include "util/util_vector.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/render/mesh.cpp b/intern/cycles/render/mesh.cpp index c42b32919d4..a4dc06c4345 100644 --- a/intern/cycles/render/mesh.cpp +++ b/intern/cycles/render/mesh.cpp @@ -14,29 +14,29 @@ * limitations under the License. */ -#include "bvh.h" -#include "bvh_build.h" - -#include "camera.h" -#include "curves.h" -#include "device.h" -#include "graph.h" -#include "shader.h" -#include "light.h" -#include "mesh.h" -#include "nodes.h" -#include "object.h" -#include "scene.h" - -#include "osl_globals.h" - -#include "subd_split.h" -#include "subd_patch_table.h" - -#include "util_foreach.h" -#include "util_logging.h" -#include "util_progress.h" -#include "util_set.h" +#include "bvh/bvh.h" +#include "bvh/bvh_build.h" + +#include "render/camera.h" +#include "render/curves.h" +#include "device/device.h" +#include "render/graph.h" +#include "render/shader.h" +#include "render/light.h" +#include "render/mesh.h" +#include "render/nodes.h" +#include "render/object.h" +#include "render/scene.h" + +#include "kernel/osl/osl_globals.h" + +#include "subd/subd_split.h" +#include "subd/subd_patch_table.h" + +#include "util/util_foreach.h" +#include "util/util_logging.h" +#include "util/util_progress.h" +#include "util/util_set.h" CCL_NAMESPACE_BEGIN @@ -1873,9 +1873,14 @@ void MeshManager::device_update_bvh(Device *device, DeviceScene *dscene, Scene * dscene->prim_object.reference((uint*)&pack.prim_object[0], pack.prim_object.size()); device->tex_alloc("__prim_object", dscene->prim_object); } + if(pack.prim_time.size()) { + dscene->prim_time.reference((float2*)&pack.prim_time[0], pack.prim_time.size()); + device->tex_alloc("__prim_time", dscene->prim_time); + } dscene->data.bvh.root = pack.root_index; dscene->data.bvh.use_qbvh = scene->params.use_qbvh; + dscene->data.bvh.use_bvh_steps = (scene->params.num_bvh_time_steps != 0); } void MeshManager::device_update_flags(Device * /*device*/, @@ -2152,6 +2157,7 @@ void MeshManager::device_free(Device *device, DeviceScene *dscene) device->tex_free(dscene->prim_visibility); device->tex_free(dscene->prim_index); device->tex_free(dscene->prim_object); + device->tex_free(dscene->prim_time); device->tex_free(dscene->tri_shader); device->tex_free(dscene->tri_vnormal); device->tex_free(dscene->tri_vindex); @@ -2173,6 +2179,7 @@ void MeshManager::device_free(Device *device, DeviceScene *dscene) dscene->prim_visibility.clear(); dscene->prim_index.clear(); dscene->prim_object.clear(); + dscene->prim_time.clear(); dscene->tri_shader.clear(); dscene->tri_vnormal.clear(); dscene->tri_vindex.clear(); diff --git a/intern/cycles/render/mesh.h b/intern/cycles/render/mesh.h index 5f33e30eac2..043ce9d0ffc 100644 --- a/intern/cycles/render/mesh.h +++ b/intern/cycles/render/mesh.h @@ -17,17 +17,18 @@ #ifndef __MESH_H__ #define __MESH_H__ -#include "attribute.h" -#include "node.h" -#include "shader.h" - -#include "util_boundbox.h" -#include "util_list.h" -#include "util_map.h" -#include "util_param.h" -#include "util_transform.h" -#include "util_types.h" -#include "util_vector.h" +#include "graph/node.h" + +#include "render/attribute.h" +#include "render/shader.h" + +#include "util/util_boundbox.h" +#include "util/util_list.h" +#include "util/util_map.h" +#include "util/util_param.h" +#include "util/util_transform.h" +#include "util/util_types.h" +#include "util/util_vector.h" CCL_NAMESPACE_BEGIN @@ -48,7 +49,7 @@ struct PackedPatchTable; class Mesh : public Node { public: - NODE_DECLARE; + NODE_DECLARE /* Mesh Triangle */ struct Triangle { diff --git a/intern/cycles/render/mesh_displace.cpp b/intern/cycles/render/mesh_displace.cpp index adc5b820298..cf28bb16bb7 100644 --- a/intern/cycles/render/mesh_displace.cpp +++ b/intern/cycles/render/mesh_displace.cpp @@ -14,15 +14,15 @@ * limitations under the License. */ -#include "device.h" +#include "device/device.h" -#include "mesh.h" -#include "object.h" -#include "scene.h" -#include "shader.h" +#include "render/mesh.h" +#include "render/object.h" +#include "render/scene.h" +#include "render/shader.h" -#include "util_foreach.h" -#include "util_progress.h" +#include "util/util_foreach.h" +#include "util/util_progress.h" CCL_NAMESPACE_BEGIN @@ -121,9 +121,9 @@ bool MeshManager::displace(Device *device, DeviceScene *dscene, Scene *scene, Me /* needs to be up to data for attribute access */ device->const_copy_to("__data", &dscene->data, sizeof(dscene->data)); - device->mem_alloc(d_input, MEM_READ_ONLY); + device->mem_alloc("displace_input", d_input, MEM_READ_ONLY); device->mem_copy_to(d_input); - device->mem_alloc(d_output, MEM_WRITE_ONLY); + device->mem_alloc("displace_output", d_output, MEM_WRITE_ONLY); DeviceTask task(DeviceTask::SHADER); task.shader_input = d_input.device_pointer; diff --git a/intern/cycles/render/mesh_subdivision.cpp b/intern/cycles/render/mesh_subdivision.cpp index 57c76a9f1c8..585ed77b026 100644 --- a/intern/cycles/render/mesh_subdivision.cpp +++ b/intern/cycles/render/mesh_subdivision.cpp @@ -14,16 +14,16 @@ * limitations under the License. */ -#include "mesh.h" -#include "attribute.h" -#include "camera.h" +#include "render/mesh.h" +#include "render/attribute.h" +#include "render/camera.h" -#include "subd_split.h" -#include "subd_patch.h" -#include "subd_patch_table.h" +#include "subd/subd_split.h" +#include "subd/subd_patch.h" +#include "subd/subd_patch_table.h" -#include "util_foreach.h" -#include "util_algorithm.h" +#include "util/util_foreach.h" +#include "util/util_algorithm.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/render/nodes.cpp b/intern/cycles/render/nodes.cpp index c02c1adb989..3f56690d0c1 100644 --- a/intern/cycles/render/nodes.cpp +++ b/intern/cycles/render/nodes.cpp @@ -14,20 +14,21 @@ * limitations under the License. */ -#include "image.h" -#include "integrator.h" -#include "nodes.h" -#include "scene.h" -#include "svm.h" -#include "svm_color_util.h" -#include "svm_ramp_util.h" -#include "svm_math_util.h" -#include "osl.h" -#include "constant_fold.h" - -#include "util_sky_model.h" -#include "util_foreach.h" -#include "util_transform.h" +#include "render/image.h" +#include "render/integrator.h" +#include "render/nodes.h" +#include "render/scene.h" +#include "render/svm.h" +#include "kernel/svm/svm_color_util.h" +#include "kernel/svm/svm_ramp_util.h" +#include "kernel/svm/svm_math_util.h" +#include "render/osl.h" +#include "render/constant_fold.h" + +#include "util/util_sky_model.h" +#include "util/util_foreach.h" +#include "util/util_logging.h" +#include "util/util_transform.h" CCL_NAMESPACE_BEGIN @@ -1931,21 +1932,38 @@ GlossyBsdfNode::GlossyBsdfNode() void GlossyBsdfNode::simplify_settings(Scene *scene) { if(distribution_orig == NBUILTIN_CLOSURES) { + roughness_orig = roughness; distribution_orig = distribution; } + else { + /* By default we use original values, so we don't worry about restoring + * defaults later one and can only do override when needed. + */ + roughness = roughness_orig; + distribution = distribution_orig; + } Integrator *integrator = scene->integrator; + ShaderInput *roughness_input = input("Roughness"); if(integrator->filter_glossy == 0.0f) { /* Fallback to Sharp closure for Roughness close to 0. * Note: Keep the epsilon in sync with kernel! */ - ShaderInput *roughness_input = input("Roughness"); if(!roughness_input->link && roughness <= 1e-4f) { + VLOG(1) << "Using sharp glossy BSDF."; distribution = CLOSURE_BSDF_REFLECTION_ID; } } else { - /* Rollback to original distribution when filter glossy is used. */ - distribution = distribution_orig; + /* If filter glossy is used we replace Sharp glossy with GGX so we can + * benefit from closure blur to remove unwanted noise. + */ + if(roughness_input->link == NULL && + distribution == CLOSURE_BSDF_REFLECTION_ID) + { + VLOG(1) << "Using GGX glossy with filter glossy."; + distribution = CLOSURE_BSDF_MICROFACET_GGX_ID; + roughness = 0.0f; + } } closure = distribution; } @@ -1953,7 +1971,8 @@ void GlossyBsdfNode::simplify_settings(Scene *scene) bool GlossyBsdfNode::has_integrator_dependency() { ShaderInput *roughness_input = input("Roughness"); - return !roughness_input->link && roughness <= 1e-4f; + return !roughness_input->link && + (distribution == CLOSURE_BSDF_REFLECTION_ID || roughness <= 1e-4f); } void GlossyBsdfNode::compile(SVMCompiler& compiler) @@ -2008,21 +2027,38 @@ GlassBsdfNode::GlassBsdfNode() void GlassBsdfNode::simplify_settings(Scene *scene) { if(distribution_orig == NBUILTIN_CLOSURES) { + roughness_orig = roughness; distribution_orig = distribution; } + else { + /* By default we use original values, so we don't worry about restoring + * defaults later one and can only do override when needed. + */ + roughness = roughness_orig; + distribution = distribution_orig; + } Integrator *integrator = scene->integrator; + ShaderInput *roughness_input = input("Roughness"); if(integrator->filter_glossy == 0.0f) { /* Fallback to Sharp closure for Roughness close to 0. * Note: Keep the epsilon in sync with kernel! */ - ShaderInput *roughness_input = input("Roughness"); if(!roughness_input->link && roughness <= 1e-4f) { + VLOG(1) << "Using sharp glass BSDF."; distribution = CLOSURE_BSDF_SHARP_GLASS_ID; } } else { - /* Rollback to original distribution when filter glossy is used. */ - distribution = distribution_orig; + /* If filter glossy is used we replace Sharp glossy with GGX so we can + * benefit from closure blur to remove unwanted noise. + */ + if(roughness_input->link == NULL && + distribution == CLOSURE_BSDF_SHARP_GLASS_ID) + { + VLOG(1) << "Using GGX glass with filter glossy."; + distribution = CLOSURE_BSDF_MICROFACET_GGX_GLASS_ID; + roughness = 0.0f; + } } closure = distribution; } @@ -2030,7 +2066,8 @@ void GlassBsdfNode::simplify_settings(Scene *scene) bool GlassBsdfNode::has_integrator_dependency() { ShaderInput *roughness_input = input("Roughness"); - return !roughness_input->link && roughness <= 1e-4f; + return !roughness_input->link && + (distribution == CLOSURE_BSDF_SHARP_GLASS_ID || roughness <= 1e-4f); } void GlassBsdfNode::compile(SVMCompiler& compiler) @@ -2085,21 +2122,38 @@ RefractionBsdfNode::RefractionBsdfNode() void RefractionBsdfNode::simplify_settings(Scene *scene) { if(distribution_orig == NBUILTIN_CLOSURES) { + roughness_orig = roughness; distribution_orig = distribution; } + else { + /* By default we use original values, so we don't worry about restoring + * defaults later one and can only do override when needed. + */ + roughness = roughness_orig; + distribution = distribution_orig; + } Integrator *integrator = scene->integrator; + ShaderInput *roughness_input = input("Roughness"); if(integrator->filter_glossy == 0.0f) { /* Fallback to Sharp closure for Roughness close to 0. * Note: Keep the epsilon in sync with kernel! */ - ShaderInput *roughness_input = input("Roughness"); if(!roughness_input->link && roughness <= 1e-4f) { + VLOG(1) << "Using sharp refraction BSDF."; distribution = CLOSURE_BSDF_REFRACTION_ID; } } else { - /* Rollback to original distribution when filter glossy is used. */ - distribution = distribution_orig; + /* If filter glossy is used we replace Sharp glossy with GGX so we can + * benefit from closure blur to remove unwanted noise. + */ + if(roughness_input->link == NULL && + distribution == CLOSURE_BSDF_REFRACTION_ID) + { + VLOG(1) << "Using GGX refraction with filter glossy."; + distribution = CLOSURE_BSDF_MICROFACET_GGX_REFRACTION_ID; + roughness = 0.0f; + } } closure = distribution; } @@ -2107,7 +2161,8 @@ void RefractionBsdfNode::simplify_settings(Scene *scene) bool RefractionBsdfNode::has_integrator_dependency() { ShaderInput *roughness_input = input("Roughness"); - return !roughness_input->link && roughness <= 1e-4f; + return !roughness_input->link && + (distribution == CLOSURE_BSDF_REFRACTION_ID || roughness <= 1e-4f); } void RefractionBsdfNode::compile(SVMCompiler& compiler) diff --git a/intern/cycles/render/nodes.h b/intern/cycles/render/nodes.h index 8d2df673688..d8023747860 100644 --- a/intern/cycles/render/nodes.h +++ b/intern/cycles/render/nodes.h @@ -17,10 +17,10 @@ #ifndef __NODES_H__ #define __NODES_H__ -#include "graph.h" -#include "node.h" +#include "render/graph.h" +#include "graph/node.h" -#include "util_string.h" +#include "util/util_string.h" CCL_NAMESPACE_BEGIN @@ -324,7 +324,7 @@ private: class BsdfNode : public ShaderNode { public: explicit BsdfNode(const NodeType *node_type); - SHADER_NODE_BASE_CLASS(BsdfNode); + SHADER_NODE_BASE_CLASS(BsdfNode) bool has_spatial_varying() { return true; } void compile(SVMCompiler& compiler, ShaderInput *param1, ShaderInput *param2, ShaderInput *param3 = NULL, ShaderInput *param4 = NULL); @@ -421,7 +421,7 @@ public: bool has_integrator_dependency(); ClosureType get_closure_type() { return distribution; } - float roughness; + float roughness, roughness_orig; ClosureType distribution, distribution_orig; }; @@ -433,7 +433,7 @@ public: bool has_integrator_dependency(); ClosureType get_closure_type() { return distribution; } - float roughness, IOR; + float roughness, roughness_orig, IOR; ClosureType distribution, distribution_orig; }; @@ -445,7 +445,7 @@ public: bool has_integrator_dependency(); ClosureType get_closure_type() { return distribution; } - float roughness, IOR; + float roughness, roughness_orig, IOR; ClosureType distribution, distribution_orig; }; @@ -674,7 +674,7 @@ public: class MixClosureWeightNode : public ShaderNode { public: - SHADER_NODE_CLASS(MixClosureWeightNode); + SHADER_NODE_CLASS(MixClosureWeightNode) float weight; float fac; @@ -920,7 +920,7 @@ public: class CurvesNode : public ShaderNode { public: explicit CurvesNode(const NodeType *node_type); - SHADER_NODE_BASE_CLASS(CurvesNode); + SHADER_NODE_BASE_CLASS(CurvesNode) virtual int get_group() { return NODE_GROUP_LEVEL_3; } diff --git a/intern/cycles/render/object.cpp b/intern/cycles/render/object.cpp index 8342f376836..375abfeb27a 100644 --- a/intern/cycles/render/object.cpp +++ b/intern/cycles/render/object.cpp @@ -14,22 +14,22 @@ * limitations under the License. */ -#include "camera.h" -#include "device.h" -#include "light.h" -#include "mesh.h" -#include "curves.h" -#include "object.h" -#include "particles.h" -#include "scene.h" - -#include "util_foreach.h" -#include "util_logging.h" -#include "util_map.h" -#include "util_progress.h" -#include "util_vector.h" - -#include "subd_patch_table.h" +#include "render/camera.h" +#include "device/device.h" +#include "render/light.h" +#include "render/mesh.h" +#include "render/curves.h" +#include "render/object.h" +#include "render/particles.h" +#include "render/scene.h" + +#include "util/util_foreach.h" +#include "util/util_logging.h" +#include "util/util_map.h" +#include "util/util_progress.h" +#include "util/util_vector.h" + +#include "subd/subd_patch_table.h" CCL_NAMESPACE_BEGIN @@ -49,6 +49,8 @@ NODE_DEFINE(Object) SOCKET_POINT(dupli_generated, "Dupli Generated", make_float3(0.0f, 0.0f, 0.0f)); SOCKET_POINT2(dupli_uv, "Dupli UV", make_float2(0.0f, 0.0f)); + SOCKET_BOOLEAN(is_shadow_catcher, "Shadow Catcher", false); + return type; } @@ -597,6 +599,12 @@ void ObjectManager::device_update_flags(Device *device, else { object_flag[object_index] &= ~SD_OBJECT_HAS_VOLUME; } + if(object->is_shadow_catcher) { + object_flag[object_index] |= SD_OBJECT_SHADOW_CATCHER; + } + else { + object_flag[object_index] &= ~SD_OBJECT_SHADOW_CATCHER; + } if(bounds_valid) { foreach(Object *volume_object, volume_objects) { diff --git a/intern/cycles/render/object.h b/intern/cycles/render/object.h index 7e306fab2a8..12d7b2c81cf 100644 --- a/intern/cycles/render/object.h +++ b/intern/cycles/render/object.h @@ -17,14 +17,14 @@ #ifndef __OBJECT_H__ #define __OBJECT_H__ -#include "node.h" -#include "scene.h" +#include "graph/node.h" +#include "render/scene.h" -#include "util_boundbox.h" -#include "util_param.h" -#include "util_transform.h" -#include "util_thread.h" -#include "util_types.h" +#include "util/util_boundbox.h" +#include "util/util_param.h" +#include "util/util_transform.h" +#include "util/util_thread.h" +#include "util/util_types.h" CCL_NAMESPACE_BEGIN @@ -40,7 +40,7 @@ struct Transform; class Object : public Node { public: - NODE_DECLARE; + NODE_DECLARE Mesh *mesh; Transform tfm; @@ -53,6 +53,7 @@ public: bool use_motion; bool hide_on_missing_motion; bool use_holdout; + bool is_shadow_catcher; float3 dupli_generated; float2 dupli_uv; diff --git a/intern/cycles/render/osl.cpp b/intern/cycles/render/osl.cpp index 67b68e63cb2..6bff29d1c76 100644 --- a/intern/cycles/render/osl.cpp +++ b/intern/cycles/render/osl.cpp @@ -14,26 +14,26 @@ * limitations under the License. */ -#include "device.h" +#include "device/device.h" -#include "graph.h" -#include "light.h" -#include "osl.h" -#include "scene.h" -#include "shader.h" -#include "nodes.h" +#include "render/graph.h" +#include "render/light.h" +#include "render/osl.h" +#include "render/scene.h" +#include "render/shader.h" +#include "render/nodes.h" #ifdef WITH_OSL -#include "osl_globals.h" -#include "osl_services.h" -#include "osl_shader.h" +#include "kernel/osl/osl_globals.h" +#include "kernel/osl/osl_services.h" +#include "kernel/osl/osl_shader.h" -#include "util_foreach.h" -#include "util_logging.h" -#include "util_md5.h" -#include "util_path.h" -#include "util_progress.h" +#include "util/util_foreach.h" +#include "util/util_logging.h" +#include "util/util_md5.h" +#include "util/util_path.h" +#include "util/util_progress.h" #endif @@ -1096,12 +1096,10 @@ void OSLCompiler::compile(Scene *scene, OSLGlobals *og, Shader *shader) /* finalize */ shader->graph->finalize(scene, false, - true, shader->has_integrator_dependency); if(shader->graph_bump) { shader->graph_bump->finalize(scene, true, - true, shader->has_integrator_dependency, shader->displacement_method == DISPLACE_BOTH); } diff --git a/intern/cycles/render/osl.h b/intern/cycles/render/osl.h index b131b672b8c..2be1126fdd3 100644 --- a/intern/cycles/render/osl.h +++ b/intern/cycles/render/osl.h @@ -17,13 +17,13 @@ #ifndef __OSL_H__ #define __OSL_H__ -#include "util_set.h" -#include "util_string.h" -#include "util_thread.h" +#include "util/util_set.h" +#include "util/util_string.h" +#include "util/util_thread.h" -#include "graph.h" -#include "nodes.h" -#include "shader.h" +#include "render/graph.h" +#include "render/nodes.h" +#include "render/shader.h" #ifdef WITH_OSL #include <OSL/oslcomp.h> diff --git a/intern/cycles/render/particles.cpp b/intern/cycles/render/particles.cpp index 1a35d60fb4b..a51822a08be 100644 --- a/intern/cycles/render/particles.cpp +++ b/intern/cycles/render/particles.cpp @@ -14,15 +14,15 @@ * limitations under the License. */ -#include "device.h" -#include "particles.h" -#include "scene.h" - -#include "util_foreach.h" -#include "util_logging.h" -#include "util_map.h" -#include "util_progress.h" -#include "util_vector.h" +#include "device/device.h" +#include "render/particles.h" +#include "render/scene.h" + +#include "util/util_foreach.h" +#include "util/util_logging.h" +#include "util/util_map.h" +#include "util/util_progress.h" +#include "util/util_vector.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/render/particles.h b/intern/cycles/render/particles.h index 2509e27b44b..66d46114b3e 100644 --- a/intern/cycles/render/particles.h +++ b/intern/cycles/render/particles.h @@ -17,8 +17,8 @@ #ifndef __PARTICLES_H__ #define __PARTICLES_H__ -#include "util_types.h" -#include "util_vector.h" +#include "util/util_types.h" +#include "util/util_vector.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/render/scene.cpp b/intern/cycles/render/scene.cpp index 68124e78cb5..4db20338744 100644 --- a/intern/cycles/render/scene.cpp +++ b/intern/cycles/render/scene.cpp @@ -16,27 +16,27 @@ #include <stdlib.h> -#include "background.h" -#include "bake.h" -#include "camera.h" -#include "curves.h" -#include "device.h" -#include "film.h" -#include "integrator.h" -#include "light.h" -#include "mesh.h" -#include "object.h" -#include "osl.h" -#include "particles.h" -#include "scene.h" -#include "shader.h" -#include "svm.h" -#include "tables.h" - -#include "util_foreach.h" -#include "util_guarded_allocator.h" -#include "util_logging.h" -#include "util_progress.h" +#include "render/background.h" +#include "render/bake.h" +#include "render/camera.h" +#include "render/curves.h" +#include "device/device.h" +#include "render/film.h" +#include "render/integrator.h" +#include "render/light.h" +#include "render/mesh.h" +#include "render/object.h" +#include "render/osl.h" +#include "render/particles.h" +#include "render/scene.h" +#include "render/shader.h" +#include "render/svm.h" +#include "render/tables.h" + +#include "util/util_foreach.h" +#include "util/util_guarded_allocator.h" +#include "util/util_logging.h" +#include "util/util_progress.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/render/scene.h b/intern/cycles/render/scene.h index 8768682043f..2b5267642a2 100644 --- a/intern/cycles/render/scene.h +++ b/intern/cycles/render/scene.h @@ -17,18 +17,18 @@ #ifndef __SCENE_H__ #define __SCENE_H__ -#include "image.h" -#include "shader.h" +#include "render/image.h" +#include "render/shader.h" -#include "device_memory.h" +#include "device/device_memory.h" -#include "util_param.h" -#include "util_string.h" -#include "util_system.h" -#include "util_texture.h" -#include "util_thread.h" -#include "util_types.h" -#include "util_vector.h" +#include "util/util_param.h" +#include "util/util_string.h" +#include "util/util_system.h" +#include "util/util_texture.h" +#include "util/util_thread.h" +#include "util/util_types.h" +#include "util/util_vector.h" CCL_NAMESPACE_BEGIN @@ -69,6 +69,7 @@ public: device_vector<uint> prim_visibility; device_vector<uint> prim_index; device_vector<uint> prim_object; + device_vector<float2> prim_time; /* mesh */ device_vector<uint> tri_shader; diff --git a/intern/cycles/render/session.cpp b/intern/cycles/render/session.cpp index 7c01934cfd8..c9b5547b407 100644 --- a/intern/cycles/render/session.cpp +++ b/intern/cycles/render/session.cpp @@ -17,24 +17,24 @@ #include <string.h> #include <limits.h> -#include "buffers.h" -#include "camera.h" -#include "device.h" -#include "graph.h" -#include "integrator.h" -#include "mesh.h" -#include "object.h" -#include "scene.h" -#include "session.h" -#include "bake.h" - -#include "util_foreach.h" -#include "util_function.h" -#include "util_logging.h" -#include "util_math.h" -#include "util_opengl.h" -#include "util_task.h" -#include "util_time.h" +#include "render/buffers.h" +#include "render/camera.h" +#include "device/device.h" +#include "render/graph.h" +#include "render/integrator.h" +#include "render/mesh.h" +#include "render/object.h" +#include "render/scene.h" +#include "render/session.h" +#include "render/bake.h" + +#include "util/util_foreach.h" +#include "util/util_function.h" +#include "util/util_logging.h" +#include "util/util_math.h" +#include "util/util_opengl.h" +#include "util/util_task.h" +#include "util/util_time.h" CCL_NAMESPACE_BEGIN @@ -230,7 +230,9 @@ void Session::run_gpu() while(1) { scoped_timer pause_timer; pause_cond.wait(pause_lock); - progress.add_skip_time(pause_timer, params.background); + if(pause) { + progress.add_skip_time(pause_timer, params.background); + } update_status_time(pause, no_tiles); progress.set_update(); @@ -520,7 +522,9 @@ void Session::run_cpu() while(1) { scoped_timer pause_timer; pause_cond.wait(pause_lock); - progress.add_skip_time(pause_timer, params.background); + if(pause) { + progress.add_skip_time(pause_timer, params.background); + } update_status_time(pause, no_tiles); progress.set_update(); @@ -633,6 +637,9 @@ DeviceRequestedFeatures Session::get_requested_device_features() requested_features.use_patch_evaluation = true; } #endif + if(object->is_shadow_catcher) { + requested_features.use_shadow_tricks = true; + } } BakeManager *bake_manager = scene->bake_manager; @@ -650,6 +657,8 @@ void Session::load_kernels() if(!kernels_loaded) { progress.set_status("Loading render kernels (may take a few minutes the first time)"); + scoped_timer timer; + DeviceRequestedFeatures requested_features = get_requested_device_features(); VLOG(2) << "Requested features:\n" << requested_features; if(!device->load_kernels(requested_features)) { @@ -663,6 +672,9 @@ void Session::load_kernels() return; } + progress.add_skip_time(timer, false); + VLOG(1) << "Total time spent loading kernels: " << time_dt() - timer.get_start(); + kernels_loaded = true; } } @@ -824,7 +836,7 @@ void Session::update_status_time(bool show_pause, bool show_done) int progressive_sample = tile_manager.state.sample; int num_samples = tile_manager.get_num_effective_samples(); - int tile = tile_manager.state.num_rendered_tiles; + int tile = progress.get_finished_tiles(); int num_tiles = tile_manager.state.num_tiles; /* update status */ @@ -832,7 +844,7 @@ void Session::update_status_time(bool show_pause, bool show_done) if(!params.progressive) { const bool is_cpu = params.device.type == DEVICE_CPU; - const bool is_last_tile = (progress.get_finished_tiles() + 1) == num_tiles; + const bool is_last_tile = (tile + 1) == num_tiles; substatus = string_printf("Path Tracing Tile %d/%d", tile, num_tiles); @@ -883,6 +895,7 @@ void Session::path_trace() task.need_finish_queue = params.progressive_refine; task.integrator_branched = scene->integrator->method == Integrator::BRANCHED_PATH; task.requested_tile_size = params.tile_size; + task.passes_size = tile_manager.params.get_passes_size(); device->task_add(task); } diff --git a/intern/cycles/render/session.h b/intern/cycles/render/session.h index c7ff1446171..a7e5f78a64d 100644 --- a/intern/cycles/render/session.h +++ b/intern/cycles/render/session.h @@ -17,15 +17,15 @@ #ifndef __SESSION_H__ #define __SESSION_H__ -#include "buffers.h" -#include "device.h" -#include "shader.h" -#include "tile.h" - -#include "util_progress.h" -#include "util_stats.h" -#include "util_thread.h" -#include "util_vector.h" +#include "render/buffers.h" +#include "device/device.h" +#include "render/shader.h" +#include "render/tile.h" + +#include "util/util_progress.h" +#include "util/util_stats.h" +#include "util/util_thread.h" +#include "util/util_vector.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/render/shader.cpp b/intern/cycles/render/shader.cpp index 335edcbe609..12d3c6cf832 100644 --- a/intern/cycles/render/shader.cpp +++ b/intern/cycles/render/shader.cpp @@ -14,22 +14,22 @@ * limitations under the License. */ -#include "background.h" -#include "camera.h" -#include "device.h" -#include "graph.h" -#include "integrator.h" -#include "light.h" -#include "mesh.h" -#include "nodes.h" -#include "object.h" -#include "osl.h" -#include "scene.h" -#include "shader.h" -#include "svm.h" -#include "tables.h" - -#include "util_foreach.h" +#include "render/background.h" +#include "render/camera.h" +#include "device/device.h" +#include "render/graph.h" +#include "render/integrator.h" +#include "render/light.h" +#include "render/mesh.h" +#include "render/nodes.h" +#include "render/object.h" +#include "render/osl.h" +#include "render/scene.h" +#include "render/shader.h" +#include "render/svm.h" +#include "render/tables.h" + +#include "util/util_foreach.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/render/shader.h b/intern/cycles/render/shader.h index 7d896652196..87fef19c592 100644 --- a/intern/cycles/render/shader.h +++ b/intern/cycles/render/shader.h @@ -19,20 +19,20 @@ #ifdef WITH_OSL /* So no context pollution happens from indirectly included windows.h */ -# include "util_windows.h" +# include "util/util_windows.h" # include <OSL/oslexec.h> #endif -#include "attribute.h" -#include "kernel_types.h" +#include "render/attribute.h" +#include "kernel/kernel_types.h" -#include "node.h" +#include "graph/node.h" -#include "util_map.h" -#include "util_param.h" -#include "util_string.h" -#include "util_thread.h" -#include "util_types.h" +#include "util/util_map.h" +#include "util/util_param.h" +#include "util/util_string.h" +#include "util/util_thread.h" +#include "util/util_types.h" CCL_NAMESPACE_BEGIN @@ -82,7 +82,7 @@ enum DisplacementMethod { class Shader : public Node { public: - NODE_DECLARE; + NODE_DECLARE int pass_id; diff --git a/intern/cycles/render/sobol.cpp b/intern/cycles/render/sobol.cpp index e3c2e802067..ce93dc8c5d5 100644 --- a/intern/cycles/render/sobol.cpp +++ b/intern/cycles/render/sobol.cpp @@ -46,10 +46,10 @@ * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ -#include "util_debug.h" -#include "util_types.h" +#include "util/util_debug.h" +#include "util/util_types.h" -#include "sobol.h" +#include "render/sobol.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/render/sobol.h b/intern/cycles/render/sobol.h index 574f148b9a2..9fbce4e14a5 100644 --- a/intern/cycles/render/sobol.h +++ b/intern/cycles/render/sobol.h @@ -17,7 +17,7 @@ #ifndef __SOBOL_H__ #define __SOBOL_H__ -#include "util_types.h" +#include "util/util_types.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/render/svm.cpp b/intern/cycles/render/svm.cpp index 955b892f4c3..4cb4018e2b4 100644 --- a/intern/cycles/render/svm.cpp +++ b/intern/cycles/render/svm.cpp @@ -14,20 +14,20 @@ * limitations under the License. */ -#include "device.h" -#include "graph.h" -#include "light.h" -#include "mesh.h" -#include "nodes.h" -#include "scene.h" -#include "shader.h" -#include "svm.h" - -#include "util_debug.h" -#include "util_logging.h" -#include "util_foreach.h" -#include "util_progress.h" -#include "util_task.h" +#include "device/device.h" +#include "render/graph.h" +#include "render/light.h" +#include "render/mesh.h" +#include "render/nodes.h" +#include "render/scene.h" +#include "render/shader.h" +#include "render/svm.h" + +#include "util/util_debug.h" +#include "util/util_logging.h" +#include "util/util_foreach.h" +#include "util/util_progress.h" +#include "util/util_task.h" CCL_NAMESPACE_BEGIN @@ -813,7 +813,6 @@ void SVMCompiler::compile(Scene *scene, scoped_timer timer((summary != NULL)? &summary->time_finalize: NULL); shader->graph->finalize(scene, false, - false, shader->has_integrator_dependency); } @@ -821,7 +820,6 @@ void SVMCompiler::compile(Scene *scene, scoped_timer timer((summary != NULL)? &summary->time_finalize_bump: NULL); shader->graph_bump->finalize(scene, true, - false, shader->has_integrator_dependency, shader->displacement_method == DISPLACE_BOTH); } diff --git a/intern/cycles/render/svm.h b/intern/cycles/render/svm.h index a501b6bc8b1..abbd9e50610 100644 --- a/intern/cycles/render/svm.h +++ b/intern/cycles/render/svm.h @@ -17,13 +17,13 @@ #ifndef __SVM_H__ #define __SVM_H__ -#include "attribute.h" -#include "graph.h" -#include "shader.h" +#include "render/attribute.h" +#include "render/graph.h" +#include "render/shader.h" -#include "util_set.h" -#include "util_string.h" -#include "util_thread.h" +#include "util/util_set.h" +#include "util/util_string.h" +#include "util/util_thread.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/render/tables.cpp b/intern/cycles/render/tables.cpp index dfafd99961b..bf1ef12d602 100644 --- a/intern/cycles/render/tables.cpp +++ b/intern/cycles/render/tables.cpp @@ -14,12 +14,12 @@ * limitations under the License. */ -#include "device.h" -#include "scene.h" -#include "tables.h" +#include "device/device.h" +#include "render/scene.h" +#include "render/tables.h" -#include "util_debug.h" -#include "util_logging.h" +#include "util/util_debug.h" +#include "util/util_logging.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/render/tables.h b/intern/cycles/render/tables.h index 1bb70b22762..bc261c2a74d 100644 --- a/intern/cycles/render/tables.h +++ b/intern/cycles/render/tables.h @@ -17,7 +17,7 @@ #ifndef __TABLES_H__ #define __TABLES_H__ -#include <util_list.h> +#include "util/util_list.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/render/tile.cpp b/intern/cycles/render/tile.cpp index a493c3fa1cd..944e746ca2d 100644 --- a/intern/cycles/render/tile.cpp +++ b/intern/cycles/render/tile.cpp @@ -14,10 +14,10 @@ * limitations under the License. */ -#include "tile.h" +#include "render/tile.h" -#include "util_algorithm.h" -#include "util_types.h" +#include "util/util_algorithm.h" +#include "util/util_types.h" CCL_NAMESPACE_BEGIN @@ -131,7 +131,6 @@ void TileManager::reset(BufferParams& params_, int num_samples_) state.buffer = BufferParams(); state.sample = range_start_sample - 1; state.num_tiles = 0; - state.num_rendered_tiles = 0; state.num_samples = 0; state.resolution_divider = get_divider(params.width, params.height, start_resolution); state.tiles.clear(); @@ -343,7 +342,6 @@ bool TileManager::next_tile(Tile& tile, int device) tile = Tile(state.tiles[logical_device].front()); state.tiles[logical_device].pop_front(); - state.num_rendered_tiles++; return true; } diff --git a/intern/cycles/render/tile.h b/intern/cycles/render/tile.h index 5d92ebac355..622b89f7670 100644 --- a/intern/cycles/render/tile.h +++ b/intern/cycles/render/tile.h @@ -19,8 +19,8 @@ #include <limits.h> -#include "buffers.h" -#include "util_list.h" +#include "render/buffers.h" +#include "util/util_list.h" CCL_NAMESPACE_BEGIN @@ -63,7 +63,6 @@ public: int num_samples; int resolution_divider; int num_tiles; - int num_rendered_tiles; /* Total samples over all pixels: Generally num_samples*num_pixels, * but can be higher due to the initial resolution division for previews. */ diff --git a/intern/cycles/subd/CMakeLists.txt b/intern/cycles/subd/CMakeLists.txt index dafb807bdf3..fe0c221ab0d 100644 --- a/intern/cycles/subd/CMakeLists.txt +++ b/intern/cycles/subd/CMakeLists.txt @@ -1,11 +1,6 @@ set(INC - . - ../graph - ../kernel - ../kernel/svm - ../render - ../util + .. ) set(INC_SYS diff --git a/intern/cycles/subd/subd_dice.cpp b/intern/cycles/subd/subd_dice.cpp index a1bd349b167..fae815901ee 100644 --- a/intern/cycles/subd/subd_dice.cpp +++ b/intern/cycles/subd/subd_dice.cpp @@ -14,13 +14,13 @@ * limitations under the License. */ -#include "camera.h" -#include "mesh.h" +#include "render/camera.h" +#include "render/mesh.h" -#include "subd_dice.h" -#include "subd_patch.h" +#include "subd/subd_dice.h" +#include "subd/subd_patch.h" -#include "util_debug.h" +#include "util/util_debug.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/subd/subd_dice.h b/intern/cycles/subd/subd_dice.h index 33d13a4ab3a..c0e32be18c4 100644 --- a/intern/cycles/subd/subd_dice.h +++ b/intern/cycles/subd/subd_dice.h @@ -22,8 +22,8 @@ * DiagSplit. For more algorithm details, see the DiagSplit paper or the * ARB_tessellation_shader OpenGL extension, Section 2.X.2. */ -#include "util_types.h" -#include "util_vector.h" +#include "util/util_types.h" +#include "util/util_vector.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/subd/subd_patch.cpp b/intern/cycles/subd/subd_patch.cpp index d3319c5ccf5..fa2fe2bf113 100644 --- a/intern/cycles/subd/subd_patch.cpp +++ b/intern/cycles/subd/subd_patch.cpp @@ -16,12 +16,12 @@ /* Parts adapted from code in the public domain in NVidia Mesh Tools. */ -#include "mesh.h" +#include "render/mesh.h" -#include "subd_patch.h" +#include "subd/subd_patch.h" -#include "util_math.h" -#include "util_types.h" +#include "util/util_math.h" +#include "util/util_types.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/subd/subd_patch.h b/intern/cycles/subd/subd_patch.h index 360c1abf27b..1bb81588835 100644 --- a/intern/cycles/subd/subd_patch.h +++ b/intern/cycles/subd/subd_patch.h @@ -17,8 +17,8 @@ #ifndef __SUBD_PATCH_H__ #define __SUBD_PATCH_H__ -#include "util_boundbox.h" -#include "util_types.h" +#include "util/util_boundbox.h" +#include "util/util_types.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/subd/subd_patch_table.cpp b/intern/cycles/subd/subd_patch_table.cpp index d437b045c07..63bf673a90b 100644 --- a/intern/cycles/subd/subd_patch_table.cpp +++ b/intern/cycles/subd/subd_patch_table.cpp @@ -25,10 +25,10 @@ * */ -#include "subd_patch_table.h" -#include "kernel_types.h" +#include "subd/subd_patch_table.h" +#include "kernel/kernel_types.h" -#include "util_math.h" +#include "util/util_math.h" #ifdef WITH_OPENSUBDIV #include <opensubdiv/far/patchTable.h> diff --git a/intern/cycles/subd/subd_patch_table.h b/intern/cycles/subd/subd_patch_table.h index 3166a1691d8..907f2dd6c28 100644 --- a/intern/cycles/subd/subd_patch_table.h +++ b/intern/cycles/subd/subd_patch_table.h @@ -17,8 +17,8 @@ #ifndef __SUBD_PATCH_TABLE_H__ #define __SUBD_PATCH_TABLE_H__ -#include "util_types.h" -#include "util_vector.h" +#include "util/util_types.h" +#include "util/util_vector.h" #ifdef WITH_OPENSUBDIV #ifdef _MSC_VER diff --git a/intern/cycles/subd/subd_split.cpp b/intern/cycles/subd/subd_split.cpp index 3c91ad8ab0d..9dbfc1c4e2f 100644 --- a/intern/cycles/subd/subd_split.cpp +++ b/intern/cycles/subd/subd_split.cpp @@ -14,16 +14,16 @@ * limitations under the License. */ -#include "camera.h" -#include "mesh.h" +#include "render/camera.h" +#include "render/mesh.h" -#include "subd_dice.h" -#include "subd_patch.h" -#include "subd_split.h" +#include "subd/subd_dice.h" +#include "subd/subd_patch.h" +#include "subd/subd_split.h" -#include "util_debug.h" -#include "util_math.h" -#include "util_types.h" +#include "util/util_debug.h" +#include "util/util_math.h" +#include "util/util_types.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/subd/subd_split.h b/intern/cycles/subd/subd_split.h index a2f76dd2e03..f869cc6a48e 100644 --- a/intern/cycles/subd/subd_split.h +++ b/intern/cycles/subd/subd_split.h @@ -22,10 +22,10 @@ * evaluation at arbitrary points is required for this to work. See the paper * for more details. */ -#include "subd_dice.h" +#include "subd/subd_dice.h" -#include "util_types.h" -#include "util_vector.h" +#include "util/util_types.h" +#include "util/util_vector.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/util/CMakeLists.txt b/intern/cycles/util/CMakeLists.txt index d8abf671bd6..a015fef8284 100644 --- a/intern/cycles/util/CMakeLists.txt +++ b/intern/cycles/util/CMakeLists.txt @@ -1,6 +1,6 @@ set(INC - . + .. ../../glew-mx ) @@ -52,6 +52,7 @@ set(SRC_HEADERS util_math.h util_math_cdf.h util_math_fast.h + util_math_intersect.h util_md5.h util_opengl.h util_optimization.h diff --git a/intern/cycles/util/util_aligned_malloc.cpp b/intern/cycles/util/util_aligned_malloc.cpp index 15d2eb3271b..cc7252dcc58 100644 --- a/intern/cycles/util/util_aligned_malloc.cpp +++ b/intern/cycles/util/util_aligned_malloc.cpp @@ -14,8 +14,8 @@ * limitations under the License. */ -#include "util_aligned_malloc.h" -#include "util_guarded_allocator.h" +#include "util/util_aligned_malloc.h" +#include "util/util_guarded_allocator.h" #include <cassert> diff --git a/intern/cycles/util/util_aligned_malloc.h b/intern/cycles/util/util_aligned_malloc.h index ecc0f28c376..cf1e86ca916 100644 --- a/intern/cycles/util/util_aligned_malloc.h +++ b/intern/cycles/util/util_aligned_malloc.h @@ -17,7 +17,7 @@ #ifndef __UTIL_ALIGNED_MALLOC_H__ #define __UTIL_ALIGNED_MALLOC_H__ -#include "util_types.h" +#include "util/util_types.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/util/util_atomic.h b/intern/cycles/util/util_atomic.h index 433e41fbbb6..6c52117ef9a 100644 --- a/intern/cycles/util/util_atomic.h +++ b/intern/cycles/util/util_atomic.h @@ -32,6 +32,13 @@ ATOMIC_INLINE void atomic_update_max_z(size_t *maximum_value, size_t value) } } +#define atomic_add_and_fetch_float(p, x) atomic_add_and_fetch_fl((p), (x)) + +#define atomic_fetch_and_inc_uint32(p) atomic_fetch_and_add_uint32((p), 1) + +#define CCL_LOCAL_MEM_FENCE 0 +#define ccl_barrier(flags) (void)0 + #else /* __KERNEL_GPU__ */ #ifdef __KERNEL_OPENCL__ @@ -39,7 +46,7 @@ ATOMIC_INLINE void atomic_update_max_z(size_t *maximum_value, size_t value) /* Float atomics implementation credits: * http://suhorukov.blogspot.in/2011/12/opencl-11-atomic-operations-on-floating.html */ -ccl_device_inline void atomic_add_and_fetch_float(volatile ccl_global float *source, +ccl_device_inline float atomic_add_and_fetch_float(volatile ccl_global float *source, const float operand) { union { @@ -56,10 +63,29 @@ ccl_device_inline void atomic_add_and_fetch_float(volatile ccl_global float *sou } while(atomic_cmpxchg((volatile ccl_global unsigned int *)source, prev_value.int_value, new_value.int_value) != prev_value.int_value); + return new_value.float_value; } +#define atomic_fetch_and_add_uint32(p, x) atomic_add((p), (x)) +#define atomic_fetch_and_inc_uint32(p) atomic_inc((p)) + +#define CCL_LOCAL_MEM_FENCE CLK_LOCAL_MEM_FENCE +#define ccl_barrier(flags) barrier(flags) + #endif /* __KERNEL_OPENCL__ */ +#ifdef __KERNEL_CUDA__ + +#define atomic_add_and_fetch_float(p, x) (atomicAdd((float*)(p), (float)(x)) + (float)(x)) + +#define atomic_fetch_and_add_uint32(p, x) atomicAdd((unsigned int*)(p), (unsigned int)(x)) +#define atomic_fetch_and_inc_uint32(p) atomic_fetch_and_add_uint32((p), 1) + +#define CCL_LOCAL_MEM_FENCE +#define ccl_barrier(flags) __syncthreads() + +#endif /* __KERNEL_CUDA__ */ + #endif /* __KERNEL_GPU__ */ #endif /* __UTIL_ATOMIC_H__ */ diff --git a/intern/cycles/util/util_boundbox.h b/intern/cycles/util/util_boundbox.h index dfe4977aef3..ed94ca20211 100644 --- a/intern/cycles/util/util_boundbox.h +++ b/intern/cycles/util/util_boundbox.h @@ -20,10 +20,10 @@ #include <math.h> #include <float.h> -#include "util_math.h" -#include "util_string.h" -#include "util_transform.h" -#include "util_types.h" +#include "util/util_math.h" +#include "util/util_string.h" +#include "util/util_transform.h" +#include "util/util_types.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/util/util_color.h b/intern/cycles/util/util_color.h index d3598f84b94..4d673dc34d8 100644 --- a/intern/cycles/util/util_color.h +++ b/intern/cycles/util/util_color.h @@ -17,11 +17,11 @@ #ifndef __UTIL_COLOR_H__ #define __UTIL_COLOR_H__ -#include "util_math.h" -#include "util_types.h" +#include "util/util_math.h" +#include "util/util_types.h" #ifdef __KERNEL_SSE2__ -#include "util_simd.h" +#include "util/util_simd.h" #endif CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/util/util_debug.cpp b/intern/cycles/util/util_debug.cpp index 80d177d2cae..9cfa57dd741 100644 --- a/intern/cycles/util/util_debug.cpp +++ b/intern/cycles/util/util_debug.cpp @@ -14,12 +14,12 @@ * limitations under the License. */ -#include "util_debug.h" +#include "util/util_debug.h" #include <stdlib.h> -#include "util_logging.h" -#include "util_string.h" +#include "util/util_logging.h" +#include "util/util_string.h" CCL_NAMESPACE_BEGIN @@ -29,7 +29,8 @@ DebugFlags::CPU::CPU() sse41(true), sse3(true), sse2(true), - qbvh(true) + qbvh(true), + split_kernel(false) { reset(); } @@ -55,10 +56,12 @@ void DebugFlags::CPU::reset() #undef CHECK_CPU_FLAGS qbvh = true; + split_kernel = false; } DebugFlags::CUDA::CUDA() - : adaptive_compile(false) + : adaptive_compile(false), + split_kernel(false) { reset(); } @@ -67,12 +70,15 @@ void DebugFlags::CUDA::reset() { if(getenv("CYCLES_CUDA_ADAPTIVE_COMPILE") != NULL) adaptive_compile = true; + + split_kernel = false; } DebugFlags::OpenCL::OpenCL() : device_type(DebugFlags::OpenCL::DEVICE_ALL), kernel_type(DebugFlags::OpenCL::KERNEL_DEFAULT), - debug(false) + debug(false), + single_program(false) { reset(); } @@ -112,6 +118,7 @@ void DebugFlags::OpenCL::reset() } /* Initialize other flags from environment variables. */ debug = (getenv("CYCLES_OPENCL_DEBUG") != NULL); + single_program = (getenv("CYCLES_OPENCL_SINGLE_PROGRAM") != NULL); } DebugFlags::DebugFlags() @@ -133,7 +140,9 @@ std::ostream& operator <<(std::ostream &os, << " AVX : " << string_from_bool(debug_flags.cpu.avx) << "\n" << " SSE4.1 : " << string_from_bool(debug_flags.cpu.sse41) << "\n" << " SSE3 : " << string_from_bool(debug_flags.cpu.sse3) << "\n" - << " SSE2 : " << string_from_bool(debug_flags.cpu.sse2) << "\n"; + << " SSE2 : " << string_from_bool(debug_flags.cpu.sse2) << "\n" + << " QBVH : " << string_from_bool(debug_flags.cpu.qbvh) << "\n" + << " Split : " << string_from_bool(debug_flags.cpu.split_kernel) << "\n"; os << "CUDA flags:\n" << " Adaptive Compile: " << string_from_bool(debug_flags.cuda.adaptive_compile) << "\n"; @@ -172,9 +181,10 @@ std::ostream& operator <<(std::ostream &os, break; } os << "OpenCL flags:\n" - << " Device type : " << opencl_device_type << "\n" - << " Kernel type : " << opencl_kernel_type << "\n" - << " Debug : " << string_from_bool(debug_flags.opencl.debug) + << " Device type : " << opencl_device_type << "\n" + << " Kernel type : " << opencl_kernel_type << "\n" + << " Debug : " << string_from_bool(debug_flags.opencl.debug) << "\n" + << " Signle program : " << string_from_bool(debug_flags.opencl.single_program) << "\n"; return os; } diff --git a/intern/cycles/util/util_debug.h b/intern/cycles/util/util_debug.h index 73fd228b5d9..4505d584490 100644 --- a/intern/cycles/util/util_debug.h +++ b/intern/cycles/util/util_debug.h @@ -20,7 +20,7 @@ #include <cassert> #include <iostream> -#include "util_static_assert.h" +#include "util/util_static_assert.h" CCL_NAMESPACE_BEGIN @@ -46,6 +46,9 @@ public: /* Whether QBVH usage is allowed or not. */ bool qbvh; + + /* Whether split kernel is used */ + bool split_kernel; }; /* Descriptor of CUDA feature-set to be used. */ @@ -58,6 +61,9 @@ public: /* Whether adaptive feature based runtime compile is enabled or not. * Requires the CUDA Toolkit and only works on Linux atm. */ bool adaptive_compile; + + /* Whether split kernel is used */ + bool split_kernel; }; /* Descriptor of OpenCL feature-set to be used. */ @@ -106,6 +112,9 @@ public: /* Use debug version of the kernel. */ bool debug; + + /* Use single program */ + bool single_program; }; /* Get instance of debug flags registry. */ diff --git a/intern/cycles/util/util_guarded_allocator.cpp b/intern/cycles/util/util_guarded_allocator.cpp index 615ac95f324..54fa6a80df5 100644 --- a/intern/cycles/util/util_guarded_allocator.cpp +++ b/intern/cycles/util/util_guarded_allocator.cpp @@ -14,8 +14,8 @@ * limitations under the License. */ -#include "util_guarded_allocator.h" -#include "util_stats.h" +#include "util/util_guarded_allocator.h" +#include "util/util_stats.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/util/util_guarded_allocator.h b/intern/cycles/util/util_guarded_allocator.h index 78453d214be..5f9dcfb2481 100644 --- a/intern/cycles/util/util_guarded_allocator.h +++ b/intern/cycles/util/util_guarded_allocator.h @@ -20,8 +20,8 @@ #include <cstddef> #include <memory> -#include "util_debug.h" -#include "util_types.h" +#include "util/util_debug.h" +#include "util/util_types.h" #ifdef WITH_BLENDER_GUARDEDALLOC # include "../../guardedalloc/MEM_guardedalloc.h" diff --git a/intern/cycles/util/util_half.h b/intern/cycles/util/util_half.h index 5db3384cda4..612228dd1c1 100644 --- a/intern/cycles/util/util_half.h +++ b/intern/cycles/util/util_half.h @@ -17,10 +17,11 @@ #ifndef __UTIL_HALF_H__ #define __UTIL_HALF_H__ -#include "util_types.h" +#include "util/util_types.h" +#include "util/util_math.h" #ifdef __KERNEL_SSE2__ -#include "util_simd.h" +#include "util/util_simd.h" #endif CCL_NAMESPACE_BEGIN @@ -110,6 +111,28 @@ ccl_device_inline float4 half4_to_float4(half4 h) return f; } +ccl_device_inline half float_to_half(float f) +{ + const uint u = __float_as_uint(f); + /* Sign bit, shifted to it's position. */ + uint sign_bit = u & 0x80000000; + sign_bit >>= 16; + /* Exponent. */ + uint exponent_bits = u & 0x7f800000; + /* Non-sign bits. */ + uint value_bits = u & 0x7fffffff; + value_bits >>= 13; /* Align mantissa on MSB. */ + value_bits -= 0x1c000; /* Adjust bias. */ + /* Flush-to-zero. */ + value_bits = (exponent_bits < 0x38800000) ? 0 : value_bits; + /* Clamp-to-max. */ + value_bits = (exponent_bits > 0x47000000) ? 0x7bff : value_bits; + /* Denormals-as-zero. */ + value_bits = (exponent_bits == 0 ? 0 : value_bits); + /* Re-insert sign bit and return. */ + return (value_bits | sign_bit); +} + #endif #endif diff --git a/intern/cycles/util/util_hash.h b/intern/cycles/util/util_hash.h index 98c3a681ff2..a30b7fe288e 100644 --- a/intern/cycles/util/util_hash.h +++ b/intern/cycles/util/util_hash.h @@ -17,7 +17,7 @@ #ifndef __UTIL_HASH_H__ #define __UTIL_HASH_H__ -#include "util_types.h" +#include "util/util_types.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/util/util_image.h b/intern/cycles/util/util_image.h index c8efc551d97..18876841b5b 100644 --- a/intern/cycles/util/util_image.h +++ b/intern/cycles/util/util_image.h @@ -21,7 +21,7 @@ #include <OpenImageIO/imageio.h> -#include "util_vector.h" +#include "util/util_vector.h" CCL_NAMESPACE_BEGIN @@ -42,4 +42,4 @@ CCL_NAMESPACE_END #endif /* __UTIL_IMAGE_H__ */ -#include "util_image_impl.h" +#include "util/util_image_impl.h" diff --git a/intern/cycles/util/util_image_impl.h b/intern/cycles/util/util_image_impl.h index 73ecfda0855..a0f9c66f979 100644 --- a/intern/cycles/util/util_image_impl.h +++ b/intern/cycles/util/util_image_impl.h @@ -17,9 +17,10 @@ #ifndef __UTIL_IMAGE_IMPL_H__ #define __UTIL_IMAGE_IMPL_H__ -#include "util_algorithm.h" -#include "util_debug.h" -#include "util_image.h" +#include "util/util_algorithm.h" +#include "util/util_debug.h" +#include "util/util_half.h" +#include "util/util_image.h" CCL_NAMESPACE_BEGIN @@ -38,6 +39,52 @@ const T *util_image_read(const vector<T>& pixels, return &pixels[index]; } +/* Cast input pixel from unknown storage to float. */ +template<typename T> +inline float cast_to_float(T value); + +template<> +inline float cast_to_float(float value) +{ + return value; +} +template<> +inline float cast_to_float(uchar value) +{ + return (float)value / 255.0f; +} +template<> +inline float cast_to_float(half value) +{ + return half_to_float(value); +} + +/* Cast float value to output pixel type. */ +template<typename T> +inline T cast_from_float(float value); + +template<> +inline float cast_from_float(float value) +{ + return value; +} +template<> +inline uchar cast_from_float(float value) +{ + if(value < 0.0f) { + return 0; + } + else if(value > (1.0f - 0.5f / 255.0f)) { + return 255; + } + return (uchar)((255.0f * value) + 0.5f); +} +template<> +inline half cast_from_float(float value) +{ + return float_to_half(value); +} + template<typename T> void util_image_downscale_sample(const vector<T>& pixels, const size_t width, @@ -71,15 +118,22 @@ void util_image_downscale_sample(const vector<T>& pixels, components, nx, ny, nz); for(size_t k = 0; k < components; ++k) { - accum[k] += pixel[k]; + accum[k] += cast_to_float(pixel[k]); } ++count; } } } - const float inv_count = 1.0f / (float)count; - for(size_t k = 0; k < components; ++k) { - result[k] = T(accum[k] * inv_count); + if(count != 0) { + const float inv_count = 1.0f / (float)count; + for(size_t k = 0; k < components; ++k) { + result[k] = cast_from_float<T>(accum[k] * inv_count); + } + } + else { + for(size_t k = 0; k < components; ++k) { + result[k] = T(0.0f); + } } } diff --git a/intern/cycles/util/util_logging.cpp b/intern/cycles/util/util_logging.cpp index 03041723e15..a5a3bd34fff 100644 --- a/intern/cycles/util/util_logging.cpp +++ b/intern/cycles/util/util_logging.cpp @@ -14,9 +14,9 @@ * limitations under the License. */ -#include <util_logging.h> +#include "util/util_logging.h" -#include "util_math.h" +#include "util/util_math.h" #include <stdio.h> #ifdef _MSC_VER @@ -69,6 +69,15 @@ void util_logging_verbosity_set(int verbosity) } std::ostream& operator <<(std::ostream &os, + const int2 &value) +{ + os << "(" << value.x + << ", " << value.y + << ")"; + return os; +} + +std::ostream& operator <<(std::ostream &os, const float3 &value) { os << "(" << value.x diff --git a/intern/cycles/util/util_logging.h b/intern/cycles/util/util_logging.h index 2aa9c25b1a0..ecf9c9cfee0 100644 --- a/intern/cycles/util/util_logging.h +++ b/intern/cycles/util/util_logging.h @@ -45,6 +45,7 @@ public: #define VLOG_ONCE(level, flag) if(!flag) flag = true, VLOG(level) +struct int2; struct float3; void util_logging_init(const char *argv0); @@ -52,6 +53,8 @@ void util_logging_start(void); void util_logging_verbosity_set(int verbosity); std::ostream& operator <<(std::ostream &os, + const int2 &value); +std::ostream& operator <<(std::ostream &os, const float3 &value); CCL_NAMESPACE_END diff --git a/intern/cycles/util/util_math.h b/intern/cycles/util/util_math.h index 2b81c8c498a..e0305b978b9 100644 --- a/intern/cycles/util/util_math.h +++ b/intern/cycles/util/util_math.h @@ -35,7 +35,7 @@ #endif -#include "util_types.h" +#include "util/util_types.h" CCL_NAMESPACE_BEGIN @@ -43,41 +43,41 @@ CCL_NAMESPACE_BEGIN /* Division */ #ifndef M_PI_F -#define M_PI_F ((float)3.14159265358979323846264338327950288) /* pi */ +#define M_PI_F (3.1415926535897932f) /* pi */ #endif #ifndef M_PI_2_F -#define M_PI_2_F ((float)1.57079632679489661923132169163975144) /* pi/2 */ +#define M_PI_2_F (1.5707963267948966f) /* pi/2 */ #endif #ifndef M_PI_4_F -#define M_PI_4_F ((float)0.785398163397448309615660845819875721) /* pi/4 */ +#define M_PI_4_F (0.7853981633974830f) /* pi/4 */ #endif #ifndef M_1_PI_F -#define M_1_PI_F ((float)0.318309886183790671537767526745028724) /* 1/pi */ +#define M_1_PI_F (0.3183098861837067f) /* 1/pi */ #endif #ifndef M_2_PI_F -#define M_2_PI_F ((float)0.636619772367581343075535053490057448) /* 2/pi */ +#define M_2_PI_F (0.6366197723675813f) /* 2/pi */ #endif /* Multiplication */ #ifndef M_2PI_F -#define M_2PI_F ((float)6.283185307179586476925286766559005768) /* 2*pi */ +#define M_2PI_F (6.2831853071795864f) /* 2*pi */ #endif #ifndef M_4PI_F -#define M_4PI_F ((float)12.56637061435917295385057353311801153) /* 4*pi */ +#define M_4PI_F (12.566370614359172f) /* 4*pi */ #endif /* Float sqrt variations */ #ifndef M_SQRT2_F -#define M_SQRT2_F ((float)1.41421356237309504880) /* sqrt(2) */ +#define M_SQRT2_F (1.4142135623730950f) /* sqrt(2) */ #endif #ifndef M_LN2_F -#define M_LN2_F ((float)0.6931471805599453) /* ln(2) */ +#define M_LN2_F (0.6931471805599453f) /* ln(2) */ #endif #ifndef M_LN10_F -#define M_LN10_F ((float)2.3025850929940457) /* ln(10) */ +#define M_LN10_F (2.3025850929940457f) /* ln(10) */ #endif /* Scalar */ @@ -774,6 +774,7 @@ template<size_t index_0, size_t index_1, size_t index_2, size_t index_3> __force return _mm_castsi128_ps(_mm_shuffle_epi32(_mm_castps_si128(b), _MM_SHUFFLE(index_3, index_2, index_1, index_0))); } +#if defined(__KERNEL_SSE3__) template<> __forceinline const float4 shuffle<0, 0, 2, 2>(const float4& b) { return _mm_moveldup_ps(b); @@ -783,6 +784,7 @@ template<> __forceinline const float4 shuffle<1, 1, 3, 3>(const float4& b) { return _mm_movehdup_ps(b); } +#endif template<> __forceinline const float4 shuffle<0, 1, 0, 1>(const float4& b) { @@ -1241,19 +1243,6 @@ ccl_device_inline float __uint_as_float(uint i) return u.f; } -/* Versions of functions which are safe for fast math. */ -ccl_device_inline bool isnan_safe(float f) -{ - unsigned int x = __float_as_uint(f); - return (x << 1) > 0xff000000u; -} - -ccl_device_inline bool isfinite_safe(float f) -{ - /* By IEEE 754 rule, 2*Inf equals Inf */ - unsigned int x = __float_as_uint(f); - return (f == f) && (x == 0 || (f != 2.0f*f)); -} /* Interpolation */ @@ -1271,6 +1260,20 @@ ccl_device_inline float triangle_area(const float3& v1, const float3& v2, const #endif +/* Versions of functions which are safe for fast math. */ +ccl_device_inline bool isnan_safe(float f) +{ + unsigned int x = __float_as_uint(f); + return (x << 1) > 0xff000000u; +} + +ccl_device_inline bool isfinite_safe(float f) +{ + /* By IEEE 754 rule, 2*Inf equals Inf */ + unsigned int x = __float_as_uint(f); + return (f == f) && (x == 0 || (f != 2.0f*f)) && !((x << 1) > 0xff000000u); +} + /* Orthonormal vectors */ ccl_device_inline void make_orthonormals(const float3 N, float3 *a, float3 *b) @@ -1329,7 +1332,7 @@ ccl_device_inline float3 safe_divide_even_color(float3 a, float3 b) y = (b.y != 0.0f)? a.y/b.y: 0.0f; z = (b.z != 0.0f)? a.z/b.z: 0.0f; - /* try to get grey even if b is zero */ + /* try to get gray even if b is zero */ if(b.x == 0.0f) { if(b.y == 0.0f) { x = z; @@ -1451,181 +1454,9 @@ ccl_device_inline float beta(float x, float y) #endif } -/* Ray Intersection */ - -ccl_device bool ray_sphere_intersect( - float3 ray_P, float3 ray_D, float ray_t, - float3 sphere_P, float sphere_radius, - float3 *isect_P, float *isect_t) -{ - float3 d = sphere_P - ray_P; - float radiussq = sphere_radius*sphere_radius; - float tsq = dot(d, d); - - if(tsq > radiussq) { /* ray origin outside sphere */ - float tp = dot(d, ray_D); - - if(tp < 0.0f) /* dir points away from sphere */ - return false; - - float dsq = tsq - tp*tp; /* pythagoras */ - - if(dsq > radiussq) /* closest point on ray outside sphere */ - return false; - - float t = tp - sqrtf(radiussq - dsq); /* pythagoras */ - - if(t < ray_t) { - *isect_t = t; - *isect_P = ray_P + ray_D*t; - return true; - } - } - - return false; -} - -ccl_device bool ray_aligned_disk_intersect( - float3 ray_P, float3 ray_D, float ray_t, - float3 disk_P, float disk_radius, - float3 *isect_P, float *isect_t) +ccl_device_inline float xor_signmask(float x, int y) { - /* aligned disk normal */ - float disk_t; - float3 disk_N = normalize_len(ray_P - disk_P, &disk_t); - float div = dot(ray_D, disk_N); - - if(UNLIKELY(div == 0.0f)) - return false; - - /* compute t to intersection point */ - float t = -disk_t/div; - if(t < 0.0f || t > ray_t) - return false; - - /* test if within radius */ - float3 P = ray_P + ray_D*t; - if(len_squared(P - disk_P) > disk_radius*disk_radius) - return false; - - *isect_P = P; - *isect_t = t; - - return true; -} - -ccl_device bool ray_triangle_intersect( - float3 ray_P, float3 ray_D, float ray_t, - float3 v0, float3 v1, float3 v2, - float3 *isect_P, float *isect_t) -{ - /* Calculate intersection */ - float3 e1 = v1 - v0; - float3 e2 = v2 - v0; - float3 s1 = cross(ray_D, e2); - - const float divisor = dot(s1, e1); - if(UNLIKELY(divisor == 0.0f)) - return false; - - const float invdivisor = 1.0f/divisor; - - /* compute first barycentric coordinate */ - const float3 d = ray_P - v0; - const float u = dot(d, s1)*invdivisor; - if(u < 0.0f) - return false; - - /* Compute second barycentric coordinate */ - const float3 s2 = cross(d, e1); - const float v = dot(ray_D, s2)*invdivisor; - if(v < 0.0f) - return false; - - const float b0 = 1.0f - u - v; - if(b0 < 0.0f) - return false; - - /* compute t to intersection point */ - const float t = dot(e2, s2)*invdivisor; - if(t < 0.0f || t > ray_t) - return false; - - *isect_t = t; - *isect_P = ray_P + ray_D*t; - - return true; -} - -ccl_device_inline bool ray_triangle_intersect_uv( - float3 ray_P, float3 ray_D, float ray_t, - float3 v0, float3 v1, float3 v2, - float *isect_u, float *isect_v, float *isect_t) -{ - /* Calculate intersection */ - float3 e1 = v1 - v0; - float3 e2 = v2 - v0; - float3 s1 = cross(ray_D, e2); - - const float divisor = dot(s1, e1); - if(UNLIKELY(divisor == 0.0f)) - return false; - - const float invdivisor = 1.0f/divisor; - - /* compute first barycentric coordinate */ - const float3 d = ray_P - v0; - const float u = dot(d, s1)*invdivisor; - if(u < 0.0f) - return false; - - /* Compute second barycentric coordinate */ - const float3 s2 = cross(d, e1); - const float v = dot(ray_D, s2)*invdivisor; - if(v < 0.0f) - return false; - - const float b0 = 1.0f - u - v; - if(b0 < 0.0f) - return false; - - /* compute t to intersection point */ - const float t = dot(e2, s2)*invdivisor; - if(t < 0.0f || t > ray_t) - return false; - - *isect_u = u; - *isect_v = v; - *isect_t = t; - - return true; -} - -ccl_device bool ray_quad_intersect(float3 ray_P, float3 ray_D, float ray_mint, float ray_maxt, - float3 quad_P, float3 quad_u, float3 quad_v, float3 quad_n, - float3 *isect_P, float *isect_t, float *isect_u, float *isect_v) -{ - float t = -(dot(ray_P, quad_n) - dot(quad_P, quad_n)) / dot(ray_D, quad_n); - if(t < ray_mint || t > ray_maxt) - return false; - - float3 hit = ray_P + t*ray_D; - float3 inplane = hit - quad_P; - - float u = dot(inplane, quad_u) / dot(quad_u, quad_u) + 0.5f; - if(u < 0.0f || u > 1.0f) - return false; - - float v = dot(inplane, quad_v) / dot(quad_v, quad_v) + 0.5f; - if(v < 0.0f || v > 1.0f) - return false; - - if(isect_P) *isect_P = hit; - if(isect_t) *isect_t = t; - if(isect_u) *isect_u = u; - if(isect_v) *isect_v = v; - - return true; + return __int_as_float(__float_as_int(x) ^ y); } /* projections */ @@ -1690,4 +1521,3 @@ ccl_device_inline int util_max_axis(float3 vec) CCL_NAMESPACE_END #endif /* __UTIL_MATH_H__ */ - diff --git a/intern/cycles/util/util_math_cdf.cpp b/intern/cycles/util/util_math_cdf.cpp index ec78ca15d88..c14d4793ea1 100644 --- a/intern/cycles/util/util_math_cdf.cpp +++ b/intern/cycles/util/util_math_cdf.cpp @@ -14,10 +14,10 @@ * limitations under the License. */ -#include "util_math_cdf.h" +#include "util/util_math_cdf.h" -#include "util_algorithm.h" -#include "util_math.h" +#include "util/util_algorithm.h" +#include "util/util_math.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/util/util_math_cdf.h b/intern/cycles/util/util_math_cdf.h index 47dfb68ba44..79643fe26e3 100644 --- a/intern/cycles/util/util_math_cdf.h +++ b/intern/cycles/util/util_math_cdf.h @@ -17,9 +17,9 @@ #ifndef __UTIL_MATH_CDF_H__ #define __UTIL_MATH_CDF_H__ -#include "util_algorithm.h" -#include "util_math.h" -#include "util_vector.h" +#include "util/util_algorithm.h" +#include "util/util_math.h" +#include "util/util_vector.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/util/util_math_intersect.h b/intern/cycles/util/util_math_intersect.h new file mode 100644 index 00000000000..2b65a0dfa48 --- /dev/null +++ b/intern/cycles/util/util_math_intersect.h @@ -0,0 +1,221 @@ +/* + * Copyright 2011-2017 Blender Foundation + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#ifndef __UTIL_MATH_INTERSECT_H__ +#define __UTIL_MATH_INTERSECT_H__ + +CCL_NAMESPACE_BEGIN + +/* Ray Intersection */ + +ccl_device bool ray_sphere_intersect( + float3 ray_P, float3 ray_D, float ray_t, + float3 sphere_P, float sphere_radius, + float3 *isect_P, float *isect_t) +{ + const float3 d = sphere_P - ray_P; + const float radiussq = sphere_radius*sphere_radius; + const float tsq = dot(d, d); + + if(tsq > radiussq) { + /* Ray origin outside sphere. */ + const float tp = dot(d, ray_D); + if(tp < 0.0f) { + /* Ray points away from sphere. */ + return false; + } + const float dsq = tsq - tp*tp; /* pythagoras */ + if(dsq > radiussq) { + /* Closest point on ray outside sphere. */ + return false; + } + const float t = tp - sqrtf(radiussq - dsq); /* pythagoras */ + if(t < ray_t) { + *isect_t = t; + *isect_P = ray_P + ray_D*t; + return true; + } + } + return false; +} + +ccl_device bool ray_aligned_disk_intersect( + float3 ray_P, float3 ray_D, float ray_t, + float3 disk_P, float disk_radius, + float3 *isect_P, float *isect_t) +{ + /* Aligned disk normal. */ + float disk_t; + const float3 disk_N = normalize_len(ray_P - disk_P, &disk_t); + const float div = dot(ray_D, disk_N); + if(UNLIKELY(div == 0.0f)) { + return false; + } + /* Compute t to intersection point. */ + const float t = -disk_t/div; + if(t < 0.0f || t > ray_t) { + return false; + } + /* Test if within radius. */ + float3 P = ray_P + ray_D*t; + if(len_squared(P - disk_P) > disk_radius*disk_radius) { + return false; + } + *isect_P = P; + *isect_t = t; + return true; +} + +#if defined(__KERNEL_CUDA__) && __CUDA_ARCH__ < 300 +ccl_device_inline +#else +ccl_device_forceinline +#endif +bool ray_triangle_intersect( + float3 ray_P, float3 ray_dir, float ray_t, +#if defined(__KERNEL_SSE2__) && defined(__KERNEL_SSE__) + const ssef *ssef_verts, +#else + const float3 tri_a, const float3 tri_b, const float3 tri_c, +#endif + float *isect_u, float *isect_v, float *isect_t) +{ +#if defined(__KERNEL_SSE2__) && defined(__KERNEL_SSE__) + typedef ssef float3; + const float3 tri_a(ssef_verts[0]); + const float3 tri_b(ssef_verts[1]); + const float3 tri_c(ssef_verts[2]); + const float3 P(ray_P); + const float3 dir(ray_dir); +#else +# define dot3(a, b) dot(a, b) + const float3 P = ray_P; + const float3 dir = ray_dir; +#endif + + /* Calculate vertices relative to ray origin. */ + const float3 v0 = tri_c - P; + const float3 v1 = tri_a - P; + const float3 v2 = tri_b - P; + + /* Calculate triangle edges. */ + const float3 e0 = v2 - v0; + const float3 e1 = v0 - v1; + const float3 e2 = v1 - v2; + + /* Perform edge tests. */ +#ifdef __KERNEL_SSE2__ + const float3 crossU = cross(v2 + v0, e0); + const float3 crossV = cross(v0 + v1, e1); + const float3 crossW = cross(v1 + v2, e2); +# ifndef __KERNEL_SSE__ + const ssef crossX(crossU.x, crossV.x, crossW.x, crossW.x); + const ssef crossY(crossU.y, crossV.y, crossW.y, crossW.y); + const ssef crossZ(crossU.z, crossV.z, crossW.z, crossW.z); +# else + ssef crossX(crossU); + ssef crossY(crossV); + ssef crossZ(crossW); + ssef zero = _mm_setzero_ps(); + _MM_TRANSPOSE4_PS(crossX, crossY, crossZ, zero); +# endif + const ssef dirX(ray_dir.x); + const ssef dirY(ray_dir.y); + const ssef dirZ(ray_dir.z); + /*const*/ ssef UVWW = crossX*dirX + crossY*dirY + crossZ*dirZ; + const float minUVW = reduce_min(UVWW); + const float maxUVW = reduce_max(UVWW); +#else /* __KERNEL_SSE2__ */ + const float U = dot(cross(v2 + v0, e0), ray_dir); + const float V = dot(cross(v0 + v1, e1), ray_dir); + const float W = dot(cross(v1 + v2, e2), ray_dir); + const float minUVW = min(U, min(V, W)); + const float maxUVW = max(U, max(V, W)); +#endif /* __KERNEL_SSE2__ */ + + if(minUVW < 0.0f && maxUVW > 0.0f) { + return false; + } + + /* Calculate geometry normal and denominator. */ + const float3 Ng1 = cross(e1, e0); + //const Vec3vfM Ng1 = stable_triangle_normal(e2,e1,e0); + const float3 Ng = Ng1 + Ng1; + const float den = dot3(Ng, dir); + /* Avoid division by 0. */ + if(UNLIKELY(den == 0.0f)) { + return false; + } + + /* Perform depth test. */ + const float T = dot3(v0, Ng); + const int sign_den = (__float_as_int(den) & 0x80000000); + const float sign_T = xor_signmask(T, sign_den); + if((sign_T < 0.0f) || + (sign_T > ray_t * xor_signmask(den, sign_den))) + { + return false; + } + + const float inv_den = 1.0f / den; +#ifdef __KERNEL_SSE2__ + UVWW *= inv_den; + _mm_store_ss(isect_u, UVWW); + _mm_store_ss(isect_v, shuffle<1,1,3,3>(UVWW)); +#else + *isect_u = U * inv_den; + *isect_v = V * inv_den; +#endif + *isect_t = T * inv_den; + return true; + +#undef dot3 +} + +ccl_device bool ray_quad_intersect(float3 ray_P, float3 ray_D, + float ray_mint, float ray_maxt, + float3 quad_P, + float3 quad_u, float3 quad_v, float3 quad_n, + float3 *isect_P, float *isect_t, + float *isect_u, float *isect_v) +{ + /* Perform intersection test. */ + float t = -(dot(ray_P, quad_n) - dot(quad_P, quad_n)) / dot(ray_D, quad_n); + if(t < ray_mint || t > ray_maxt) { + return false; + } + const float3 hit = ray_P + t*ray_D; + const float3 inplane = hit - quad_P; + const float u = dot(inplane, quad_u) / dot(quad_u, quad_u) + 0.5f; + if(u < 0.0f || u > 1.0f) { + return false; + } + const float v = dot(inplane, quad_v) / dot(quad_v, quad_v) + 0.5f; + if(v < 0.0f || v > 1.0f) { + return false; + } + /* Store the result. */ + /* TODO(sergey): Check whether we can avoid some checks here. */ + if(isect_P != NULL) *isect_P = hit; + if(isect_t != NULL) *isect_t = t; + if(isect_u != NULL) *isect_u = u; + if(isect_v != NULL) *isect_v = v; + return true; +} + +CCL_NAMESPACE_END + +#endif /* __UTIL_MATH_INTERSECT_H__ */ diff --git a/intern/cycles/util/util_md5.h b/intern/cycles/util/util_md5.h index d0af9fdb004..e4cd66c85b0 100644 --- a/intern/cycles/util/util_md5.h +++ b/intern/cycles/util/util_md5.h @@ -30,8 +30,8 @@ #ifndef __UTIL_MD5_H__ #define __UTIL_MD5_H__ -#include "util_string.h" -#include "util_types.h" +#include "util/util_string.h" +#include "util/util_types.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/util/util_optimization.h b/intern/cycles/util/util_optimization.h index adc141a7b28..6f70a474fe7 100644 --- a/intern/cycles/util/util_optimization.h +++ b/intern/cycles/util/util_optimization.h @@ -111,7 +111,7 @@ /* MinGW64 has conflicting declarations for these SSE headers in <windows.h>. * Since we can't avoid including <windows.h>, better only include that */ -#include "util_windows.h" +#include "util/util_windows.h" #endif diff --git a/intern/cycles/util/util_path.cpp b/intern/cycles/util/util_path.cpp index 5df262fcbbb..cd3067f7650 100644 --- a/intern/cycles/util/util_path.cpp +++ b/intern/cycles/util/util_path.cpp @@ -14,10 +14,10 @@ * limitations under the License. */ -#include "util_debug.h" -#include "util_md5.h" -#include "util_path.h" -#include "util_string.h" +#include "util/util_debug.h" +#include "util/util_md5.h" +#include "util/util_path.h" +#include "util/util_string.h" #include <OpenImageIO/filesystem.h> #include <OpenImageIO/strutil.h> @@ -45,7 +45,7 @@ OIIO_NAMESPACE_USING # include <shlwapi.h> #endif -#include "util_windows.h" +#include "util/util_windows.h" CCL_NAMESPACE_BEGIN @@ -320,17 +320,18 @@ static char *path_specials(const string& sub) { static bool env_init = false; static char *env_shader_path; - static char *env_kernel_path; + static char *env_source_path; if(!env_init) { env_shader_path = getenv("CYCLES_SHADER_PATH"); - env_kernel_path = getenv("CYCLES_KERNEL_PATH"); + /* NOTE: It is KERNEL in env variable for compatibility reasons. */ + env_source_path = getenv("CYCLES_KERNEL_PATH"); env_init = true; } if(env_shader_path != NULL && sub == "shader") { return env_shader_path; } - else if(env_shader_path != NULL && sub == "kernel") { - return env_kernel_path; + else if(env_shader_path != NULL && sub == "source") { + return env_source_path; } return NULL; } @@ -814,7 +815,7 @@ string path_source_replace_includes(const string& source, /* Use line directives for better error messages. */ line = line_directive(filepath, 1) + token.replace(0, n_end + 1, "\n" + text + "\n") - + line_directive(path_join(path, source_filename), i); + + line_directive(path_join(path, source_filename), i + 1); } } } diff --git a/intern/cycles/util/util_path.h b/intern/cycles/util/util_path.h index 70dbb5ae403..0e5e2d2c837 100644 --- a/intern/cycles/util/util_path.h +++ b/intern/cycles/util/util_path.h @@ -24,10 +24,10 @@ #include <stdio.h> -#include "util_set.h" -#include "util_string.h" -#include "util_types.h" -#include "util_vector.h" +#include "util/util_set.h" +#include "util/util_string.h" +#include "util/util_types.h" +#include "util/util_vector.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/util/util_progress.h b/intern/cycles/util/util_progress.h index 14215056840..39c1eed04e7 100644 --- a/intern/cycles/util/util_progress.h +++ b/intern/cycles/util/util_progress.h @@ -23,10 +23,10 @@ * update notifications from a job running in another thread. All methods * except for the constructor/destructor are thread safe. */ -#include "util_function.h" -#include "util_string.h" -#include "util_time.h" -#include "util_thread.h" +#include "util/util_function.h" +#include "util/util_string.h" +#include "util/util_time.h" +#include "util/util_thread.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/util/util_simd.cpp b/intern/cycles/util/util_simd.cpp index de2df612578..f90439c188b 100644 --- a/intern/cycles/util/util_simd.cpp +++ b/intern/cycles/util/util_simd.cpp @@ -19,7 +19,7 @@ (defined(WITH_KERNEL_NATIVE) && defined(__SSE2__)) #define __KERNEL_SSE2__ -#include "util_simd.h" +#include "util/util_simd.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/util/util_simd.h b/intern/cycles/util/util_simd.h index 756bd15ed25..557809a5719 100644 --- a/intern/cycles/util/util_simd.h +++ b/intern/cycles/util/util_simd.h @@ -20,8 +20,8 @@ #include <limits> -#include "util_debug.h" -#include "util_types.h" +#include "util/util_debug.h" +#include "util/util_types.h" CCL_NAMESPACE_BEGIN @@ -451,11 +451,11 @@ ccl_device_inline int bitscan(int value) CCL_NAMESPACE_END -#include "util_math.h" -#include "util_sseb.h" -#include "util_ssei.h" -#include "util_ssef.h" -#include "util_avxf.h" +#include "util/util_math.h" +#include "util/util_sseb.h" +#include "util/util_ssei.h" +#include "util/util_ssef.h" +#include "util/util_avxf.h" #endif /* __UTIL_SIMD_TYPES_H__ */ diff --git a/intern/cycles/util/util_sky_model.cpp b/intern/cycles/util/util_sky_model.cpp index 5730986cc4f..6dda8469907 100644 --- a/intern/cycles/util/util_sky_model.cpp +++ b/intern/cycles/util/util_sky_model.cpp @@ -97,8 +97,8 @@ All instructions on how to use this code are in the accompanying header file. */ -#include "util_sky_model.h" -#include "util_sky_model_data.h" +#include "util/util_sky_model.h" +#include "util/util_sky_model_data.h" #include <assert.h> #include <stdio.h> diff --git a/intern/cycles/util/util_ssef.h b/intern/cycles/util/util_ssef.h index 2f5295b5463..cf99a08efae 100644 --- a/intern/cycles/util/util_ssef.h +++ b/intern/cycles/util/util_ssef.h @@ -514,12 +514,12 @@ ccl_device_inline float len3(const ssef& a) /* faster version for SSSE3 */ typedef ssei shuffle_swap_t; -ccl_device_inline const shuffle_swap_t shuffle_swap_identity(void) +ccl_device_inline shuffle_swap_t shuffle_swap_identity(void) { return _mm_set_epi8(15, 14, 13, 12, 11, 10, 9, 8, 7, 6, 5, 4, 3, 2, 1, 0); } -ccl_device_inline const shuffle_swap_t shuffle_swap_swap(void) +ccl_device_inline shuffle_swap_t shuffle_swap_swap(void) { return _mm_set_epi8(7, 6, 5, 4, 3, 2, 1, 0, 15, 14, 13, 12, 11, 10, 9, 8); } @@ -534,12 +534,12 @@ ccl_device_inline const ssef shuffle_swap(const ssef& a, const shuffle_swap_t& s /* somewhat slower version for SSE2 */ typedef int shuffle_swap_t; -ccl_device_inline const shuffle_swap_t shuffle_swap_identity(void) +ccl_device_inline shuffle_swap_t shuffle_swap_identity(void) { return 0; } -ccl_device_inline const shuffle_swap_t shuffle_swap_swap(void) +ccl_device_inline shuffle_swap_t shuffle_swap_swap(void) { return 1; } diff --git a/intern/cycles/util/util_stack_allocator.h b/intern/cycles/util/util_stack_allocator.h index d7aab5b250c..2f1799a739e 100644 --- a/intern/cycles/util/util_stack_allocator.h +++ b/intern/cycles/util/util_stack_allocator.h @@ -20,8 +20,8 @@ #include <cstddef> #include <memory> -#include "util_debug.h" -#include "util_types.h" +#include "util/util_debug.h" +#include "util/util_types.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/util/util_static_assert.h b/intern/cycles/util/util_static_assert.h index 033d85e8ec6..e90049254de 100644 --- a/intern/cycles/util/util_static_assert.h +++ b/intern/cycles/util/util_static_assert.h @@ -43,7 +43,9 @@ template <> class StaticAssertFailure<true> {}; # endif /* __COUNTER__ */ # endif /* C++11 or MSVC2015 */ #else /* __KERNEL_GPU__ */ -# define static_assert(statement, message) +# ifndef static_assert +# define static_assert(statement, message) +# endif #endif /* __KERNEL_GPU__ */ /* TODO(sergey): For until C++11 is a bare minimum for us, diff --git a/intern/cycles/util/util_stats.h b/intern/cycles/util/util_stats.h index c21a8488c81..baba549753d 100644 --- a/intern/cycles/util/util_stats.h +++ b/intern/cycles/util/util_stats.h @@ -17,7 +17,7 @@ #ifndef __UTIL_STATS_H__ #define __UTIL_STATS_H__ -#include "util_atomic.h" +#include "util/util_atomic.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/util/util_string.cpp b/intern/cycles/util/util_string.cpp index 5594aa8edb6..a1008d510d1 100644 --- a/intern/cycles/util/util_string.cpp +++ b/intern/cycles/util/util_string.cpp @@ -17,9 +17,9 @@ #include <stdarg.h> #include <stdio.h> -#include "util_foreach.h" -#include "util_string.h" -#include "util_windows.h" +#include "util/util_foreach.h" +#include "util/util_string.h" +#include "util/util_windows.h" #ifdef _WIN32 # ifndef vsnprintf diff --git a/intern/cycles/util/util_string.h b/intern/cycles/util/util_string.h index 7aeed96f00b..e2c105db9c1 100644 --- a/intern/cycles/util/util_string.h +++ b/intern/cycles/util/util_string.h @@ -21,7 +21,7 @@ #include <string> #include <sstream> -#include "util_vector.h" +#include "util/util_vector.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/util/util_system.cpp b/intern/cycles/util/util_system.cpp index 87d885c44cf..a942d738b8a 100644 --- a/intern/cycles/util/util_system.cpp +++ b/intern/cycles/util/util_system.cpp @@ -14,12 +14,12 @@ * limitations under the License. */ -#include "util_system.h" +#include "util/util_system.h" -#include "util_debug.h" -#include "util_logging.h" -#include "util_types.h" -#include "util_string.h" +#include "util/util_debug.h" +#include "util/util_logging.h" +#include "util/util_types.h" +#include "util/util_string.h" #ifdef _WIN32 # if(!defined(FREE_WINDOWS)) diff --git a/intern/cycles/util/util_system.h b/intern/cycles/util/util_system.h index ff61b260bed..db7a45b2d59 100644 --- a/intern/cycles/util/util_system.h +++ b/intern/cycles/util/util_system.h @@ -17,7 +17,7 @@ #ifndef __UTIL_SYSTEM_H__ #define __UTIL_SYSTEM_H__ -#include "util_string.h" +#include "util/util_string.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/util/util_task.cpp b/intern/cycles/util/util_task.cpp index 0d1fed3ebbf..fb0c34e1dc4 100644 --- a/intern/cycles/util/util_task.cpp +++ b/intern/cycles/util/util_task.cpp @@ -14,12 +14,12 @@ * limitations under the License. */ -#include "util_debug.h" -#include "util_foreach.h" -#include "util_logging.h" -#include "util_system.h" -#include "util_task.h" -#include "util_time.h" +#include "util/util_debug.h" +#include "util/util_foreach.h" +#include "util/util_logging.h" +#include "util/util_system.h" +#include "util/util_task.h" +#include "util/util_time.h" //#define THREADING_DEBUG_ENABLED diff --git a/intern/cycles/util/util_task.h b/intern/cycles/util/util_task.h index 0b82f14f66f..3ebfb007e40 100644 --- a/intern/cycles/util/util_task.h +++ b/intern/cycles/util/util_task.h @@ -17,10 +17,10 @@ #ifndef __UTIL_TASK_H__ #define __UTIL_TASK_H__ -#include "util_list.h" -#include "util_string.h" -#include "util_thread.h" -#include "util_vector.h" +#include "util/util_list.h" +#include "util/util_string.h" +#include "util/util_thread.h" +#include "util/util_vector.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/util/util_thread.cpp b/intern/cycles/util/util_thread.cpp index 3db8b4bd197..3dcb09804b0 100644 --- a/intern/cycles/util/util_thread.cpp +++ b/intern/cycles/util/util_thread.cpp @@ -14,10 +14,10 @@ * limitations under the License. */ -#include "util_thread.h" +#include "util/util_thread.h" -#include "util_system.h" -#include "util_windows.h" +#include "util/util_system.h" +#include "util/util_windows.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/util/util_thread.h b/intern/cycles/util/util_thread.h index 427c633d2ce..1b4e87ebf03 100644 --- a/intern/cycles/util/util_thread.h +++ b/intern/cycles/util/util_thread.h @@ -32,7 +32,7 @@ # include <libkern/OSAtomic.h> #endif -#include "util_function.h" +#include "util/util_function.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/util/util_time.cpp b/intern/cycles/util/util_time.cpp index 59c963cfafb..7c39aa294bf 100644 --- a/intern/cycles/util/util_time.cpp +++ b/intern/cycles/util/util_time.cpp @@ -16,8 +16,8 @@ #include <stdlib.h> -#include "util_time.h" -#include "util_windows.h" +#include "util/util_time.h" +#include "util/util_windows.h" #ifdef _WIN32 diff --git a/intern/cycles/util/util_transform.cpp b/intern/cycles/util/util_transform.cpp index 2f10540c94e..b8f182ae962 100644 --- a/intern/cycles/util/util_transform.cpp +++ b/intern/cycles/util/util_transform.cpp @@ -46,10 +46,10 @@ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ -#include "util_transform.h" +#include "util/util_transform.h" -#include "util_boundbox.h" -#include "util_math.h" +#include "util/util_boundbox.h" +#include "util/util_math.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/util/util_transform.h b/intern/cycles/util/util_transform.h index a0695f20488..aef168ca64d 100644 --- a/intern/cycles/util/util_transform.h +++ b/intern/cycles/util/util_transform.h @@ -21,8 +21,8 @@ #include <string.h> #endif -#include "util_math.h" -#include "util_types.h" +#include "util/util_math.h" +#include "util/util_types.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/util/util_types.h b/intern/cycles/util/util_types.h index a000fae4bd6..bf4a134b998 100644 --- a/intern/cycles/util/util_types.h +++ b/intern/cycles/util/util_types.h @@ -37,6 +37,9 @@ #define ccl_device_noinline static #define ccl_global #define ccl_constant +#define ccl_local +#define ccl_local_param +#define ccl_private #define ccl_restrict __restrict #define __KERNEL_WITH_SSE_ALIGN__ @@ -82,7 +85,7 @@ /* SIMD Types */ -#include "util_optimization.h" +#include "util/util_optimization.h" #endif @@ -103,10 +106,16 @@ typedef unsigned int uint; #endif -#ifndef __KERNEL_GPU__ - /* Fixed Bits Types */ +#ifdef __KERNEL_OPENCL__ + +typedef ulong uint64_t; + +#endif + +#ifndef __KERNEL_GPU__ + #ifdef _WIN32 typedef signed char int8_t; @@ -171,7 +180,7 @@ struct ccl_try_align(16) int3 { }; __forceinline int3() {} - __forceinline int3(const __m128i a) : m128(a) {} + __forceinline int3(const __m128i& a) : m128(a) {} __forceinline operator const __m128i&(void) const { return m128; } __forceinline operator __m128i&(void) { return m128; } @@ -193,7 +202,7 @@ struct ccl_try_align(16) int4 { }; __forceinline int4() {} - __forceinline int4(const __m128i a) : m128(a) {} + __forceinline int4(const __m128i& a) : m128(a) {} __forceinline operator const __m128i&(void) const { return m128; } __forceinline operator __m128i&(void) { return m128; } @@ -265,7 +274,7 @@ struct ccl_try_align(16) float4 { }; __forceinline float4() {} - __forceinline float4(const __m128 a) : m128(a) {} + __forceinline float4(const __m128& a) : m128(a) {} __forceinline operator const __m128&(void) const { return m128; } __forceinline operator __m128&(void) { return m128; } @@ -397,11 +406,6 @@ ccl_device_inline float4 make_float4(float x, float y, float z, float w) return a; } -ccl_device_inline int align_up(int offset, int alignment) -{ - return (offset + alignment - 1) & ~(alignment - 1); -} - ccl_device_inline int3 make_int3(int i) { #ifdef __KERNEL_SSE__ @@ -476,6 +480,21 @@ ccl_device_inline int4 make_int4(const float3& f) #endif +ccl_device_inline size_t align_up(size_t offset, size_t alignment) +{ + return (offset + alignment - 1) & ~(alignment - 1); +} + +ccl_device_inline size_t round_up(size_t x, size_t multiple) +{ + return ((x + multiple - 1) / multiple) * multiple; +} + +ccl_device_inline size_t round_down(size_t x, size_t multiple) +{ + return (x / multiple) * multiple; +} + /* Interpolation types for textures * cuda also use texture space to store other objects */ enum InterpolationType { diff --git a/intern/cycles/util/util_vector.h b/intern/cycles/util/util_vector.h index 546b17570bb..4add91a3368 100644 --- a/intern/cycles/util/util_vector.h +++ b/intern/cycles/util/util_vector.h @@ -23,9 +23,9 @@ #include <cstring> #include <vector> -#include "util_aligned_malloc.h" -#include "util_guarded_allocator.h" -#include "util_types.h" +#include "util/util_aligned_malloc.h" +#include "util/util_guarded_allocator.h" +#include "util/util_types.h" CCL_NAMESPACE_BEGIN diff --git a/intern/cycles/util/util_view.cpp b/intern/cycles/util/util_view.cpp index 9796a5f896d..10d86167921 100644 --- a/intern/cycles/util/util_view.cpp +++ b/intern/cycles/util/util_view.cpp @@ -17,11 +17,11 @@ #include <stdio.h> #include <stdlib.h> -#include "util_opengl.h" -#include "util_string.h" -#include "util_time.h" -#include "util_version.h" -#include "util_view.h" +#include "util/util_opengl.h" +#include "util/util_string.h" +#include "util/util_time.h" +#include "util/util_version.h" +#include "util/util_view.h" #ifdef __APPLE__ #include <GLUT/glut.h> diff --git a/intern/cycles/util/util_windows.cpp b/intern/cycles/util/util_windows.cpp index 4de8483564b..073db2a27db 100644 --- a/intern/cycles/util/util_windows.cpp +++ b/intern/cycles/util/util_windows.cpp @@ -14,7 +14,7 @@ * limitations under the License. */ -#include "util_windows.h" +#include "util/util_windows.h" #ifdef _WIN32 diff --git a/intern/decklink/CMakeLists.txt b/intern/decklink/CMakeLists.txt index fbef65cdba4..b7727cbc32a 100644 --- a/intern/decklink/CMakeLists.txt +++ b/intern/decklink/CMakeLists.txt @@ -34,14 +34,14 @@ set(SRC DeckLinkAPI.h ) -if (WIN32) +if(WIN32) list(APPEND SRC win/DeckLinkAPI_h.h win/DeckLinkAPI_i.c ) endif() -if (UNIX AND NOT APPLE) +if(UNIX AND NOT APPLE) list(APPEND SRC linux/DeckLinkAPI.h linux/DeckLinkAPIConfiguration.h diff --git a/intern/ffmpeg/ffmpeg_compat.h b/intern/ffmpeg/ffmpeg_compat.h index bcfa24b06a8..d6220ebf562 100644 --- a/intern/ffmpeg/ffmpeg_compat.h +++ b/intern/ffmpeg/ffmpeg_compat.h @@ -350,7 +350,12 @@ int avcodec_decode_video2(AVCodecContext *avctx, AVFrame *picture, FFMPEG_INLINE int64_t av_get_pts_from_frame(AVFormatContext *avctx, AVFrame * picture) { - int64_t pts = picture->pkt_pts; + int64_t pts; +#if LIBAVUTIL_VERSION_INT >= AV_VERSION_INT(55, 34, 100) + pts = picture->pts; +#else + pts = picture->pkt_pts; +#endif if (pts == AV_NOPTS_VALUE) { pts = picture->pkt_dts; diff --git a/intern/ghost/GHOST_C-api.h b/intern/ghost/GHOST_C-api.h index ff1922af4f3..6887063eae9 100644 --- a/intern/ghost/GHOST_C-api.h +++ b/intern/ghost/GHOST_C-api.h @@ -908,6 +908,11 @@ extern int GHOST_UseNativePixels(void); extern float GHOST_GetNativePixelSize(GHOST_WindowHandle windowhandle); /** + * Returns the suggested DPI for this window. + */ +extern GHOST_TUns16 GHOST_GetDPIHint(GHOST_WindowHandle windowhandle); + +/** * Enable IME attached to the given window, i.e. allows user-input * events to be dispatched to the IME. * \param windowhandle Window handle of the caller diff --git a/intern/ghost/GHOST_IWindow.h b/intern/ghost/GHOST_IWindow.h index 688ebecf557..4a4d6be5ced 100644 --- a/intern/ghost/GHOST_IWindow.h +++ b/intern/ghost/GHOST_IWindow.h @@ -332,6 +332,12 @@ public: virtual float getNativePixelSize(void) = 0; + /** + * Returns the recommended DPI for this window. + * \return The recommended DPI for this window. + */ + virtual GHOST_TUns16 getDPIHint() = 0; + #ifdef WITH_INPUT_IME /** * Enable IME attached to the given window, i.e. allows user-input diff --git a/intern/ghost/GHOST_Types.h b/intern/ghost/GHOST_Types.h index 9ee4599a4a6..02b5063e515 100644 --- a/intern/ghost/GHOST_Types.h +++ b/intern/ghost/GHOST_Types.h @@ -189,6 +189,7 @@ typedef enum { GHOST_kEventWindowUpdate, GHOST_kEventWindowSize, GHOST_kEventWindowMove, + GHOST_kEventWindowDPIHintChanged, GHOST_kEventDraggingEntered, GHOST_kEventDraggingUpdated, diff --git a/intern/ghost/intern/GHOST_C-api.cpp b/intern/ghost/intern/GHOST_C-api.cpp index 41bc735e1e2..ce653188760 100644 --- a/intern/ghost/intern/GHOST_C-api.cpp +++ b/intern/ghost/intern/GHOST_C-api.cpp @@ -914,6 +914,12 @@ float GHOST_GetNativePixelSize(GHOST_WindowHandle windowhandle) return 1.0f; } +GHOST_TUns16 GHOST_GetDPIHint(GHOST_WindowHandle windowhandle) +{ + GHOST_IWindow *window = (GHOST_IWindow *)windowhandle; + return window->getDPIHint(); +} + #ifdef WITH_INPUT_IME void GHOST_BeginIME(GHOST_WindowHandle windowhandle, diff --git a/intern/ghost/intern/GHOST_SystemWin32.cpp b/intern/ghost/intern/GHOST_SystemWin32.cpp index 7d55a973f91..240d7ccd2fe 100644 --- a/intern/ghost/intern/GHOST_SystemWin32.cpp +++ b/intern/ghost/intern/GHOST_SystemWin32.cpp @@ -111,6 +111,11 @@ #define VK_MEDIA_PLAY_PAUSE 0xB3 #endif // VK_MEDIA_PLAY_PAUSE +// Window message newer than Windows 7 +#ifndef WM_DPICHANGED +#define WM_DPICHANGED 0x02E0 +#endif // WM_DPICHANGED + /* Workaround for some laptop touchpads, some of which seems to * have driver issues which makes it so window function receives * the message, but PeekMessage doesn't pick those messages for @@ -152,6 +157,27 @@ static void initRawInput() #undef DEVICE_COUNT } +#ifndef DPI_ENUMS_DECLARED +typedef enum PROCESS_DPI_AWARENESS { + PROCESS_DPI_UNAWARE = 0, + PROCESS_SYSTEM_DPI_AWARE = 1, + PROCESS_PER_MONITOR_DPI_AWARE = 2 +} PROCESS_DPI_AWARENESS; + +typedef enum MONITOR_DPI_TYPE { + MDT_EFFECTIVE_DPI = 0, + MDT_ANGULAR_DPI = 1, + MDT_RAW_DPI = 2, + MDT_DEFAULT = MDT_EFFECTIVE_DPI +} MONITOR_DPI_TYPE; + +#define USER_DEFAULT_SCREEN_DPI 96 + +#define DPI_ENUMS_DECLARED +#endif +typedef HRESULT(API * GHOST_WIN32_SetProcessDpiAwareness)(PROCESS_DPI_AWARENESS); +typedef BOOL(API * GHOST_WIN32_EnableNonClientDpiScaling)(HWND); + GHOST_SystemWin32::GHOST_SystemWin32() : m_hasPerformanceCounter(false), m_freq(0), m_start(0) { @@ -161,6 +187,18 @@ GHOST_SystemWin32::GHOST_SystemWin32() m_consoleStatus = 1; + // Tell Windows we are per monitor DPI aware. This disables the default + // blurry scaling and enables WM_DPICHANGED to allow us to draw at proper DPI. + HMODULE m_shcore = ::LoadLibrary("Shcore.dll"); + if (m_shcore) { + GHOST_WIN32_SetProcessDpiAwareness fpSetProcessDpiAwareness = + (GHOST_WIN32_SetProcessDpiAwareness) ::GetProcAddress(m_shcore, "SetProcessDpiAwareness"); + + if (fpSetProcessDpiAwareness) { + fpSetProcessDpiAwareness(PROCESS_PER_MONITOR_DPI_AWARE); + } + } + // Check if current keyboard layout uses AltGr and save keylayout ID for // specialized handling if keys like VK_OEM_*. I.e. french keylayout // generates VK_OEM_8 for their exclamation key (key left of right shift) @@ -922,6 +960,20 @@ LRESULT WINAPI GHOST_SystemWin32::s_wndProc(HWND hwnd, UINT msg, WPARAM wParam, GHOST_ASSERT(system, "GHOST_SystemWin32::s_wndProc(): system not initialized"); if (hwnd) { + if(msg == WM_NCCREATE) { + // Tell Windows to automatically handle scaling of non-client areas + // such as the caption bar. EnableNonClientDpiScaling was introduced in Windows 10 + HMODULE m_user32 = ::LoadLibrary("User32.dll"); + if (m_user32) { + GHOST_WIN32_EnableNonClientDpiScaling fpEnableNonClientDpiScaling = + (GHOST_WIN32_EnableNonClientDpiScaling) ::GetProcAddress(m_user32, "EnableNonClientDpiScaling"); + + if (fpEnableNonClientDpiScaling) { + fpEnableNonClientDpiScaling(hwnd); + } + } + } + GHOST_WindowWin32 *window = (GHOST_WindowWin32 *)::GetWindowLongPtr(hwnd, GWLP_USERDATA); if (window) { switch (msg) { @@ -1294,6 +1346,32 @@ LRESULT WINAPI GHOST_SystemWin32::s_wndProc(HWND hwnd, UINT msg, WPARAM wParam, } break; + case WM_DPICHANGED: + /* The WM_DPICHANGED message is sent when the effective dots per inch (dpi) for a window has changed. + * The DPI is the scale factor for a window. There are multiple events that can cause the DPI to + * change such as when the window is moved to a monitor with a different DPI. + */ + { + WORD newYAxisDPI = HIWORD(wParam); + WORD newXAxisDPI = LOWORD(wParam); + // The suggested new size and position of the window. + RECT* const suggestedWindowRect = (RECT*)lParam; + + // Push DPI change event first + system->pushEvent(processWindowEvent(GHOST_kEventWindowDPIHintChanged, window)); + system->dispatchEvents(); + eventHandled = true; + + // Then move and resize window + SetWindowPos(hwnd, + NULL, + suggestedWindowRect->left, + suggestedWindowRect->top, + suggestedWindowRect->right - suggestedWindowRect->left, + suggestedWindowRect->bottom - suggestedWindowRect->top, + SWP_NOZORDER | SWP_NOACTIVATE); + } + break; //////////////////////////////////////////////////////////////////////// // Window events, ignored //////////////////////////////////////////////////////////////////////// diff --git a/intern/ghost/intern/GHOST_Window.h b/intern/ghost/intern/GHOST_Window.h index d778628ea37..2798bdf72f3 100644 --- a/intern/ghost/intern/GHOST_Window.h +++ b/intern/ghost/intern/GHOST_Window.h @@ -295,6 +295,15 @@ public: return 1.0f; } + /** + * Returns the recommended DPI for this window. + * \return The recommended DPI for this window. + */ + virtual inline GHOST_TUns16 getDPIHint() + { + return 96; + } + #ifdef WITH_INPUT_IME virtual void beginIME(GHOST_TInt32 x, GHOST_TInt32 y, diff --git a/intern/ghost/intern/GHOST_WindowSDL.cpp b/intern/ghost/intern/GHOST_WindowSDL.cpp index 1335c38d977..aeb6188daef 100644 --- a/intern/ghost/intern/GHOST_WindowSDL.cpp +++ b/intern/ghost/intern/GHOST_WindowSDL.cpp @@ -563,3 +563,19 @@ GHOST_WindowSDL::setWindowCursorVisibility(bool visible) SDL_ShowCursor(visible); return GHOST_kSuccess; } + +GHOST_TUns16 +GHOST_WindowSDL::getDPIHint() +{ + int displayIndex = SDL_GetWindowDisplayIndex(m_sdl_win); + if (displayIndex < 0) { + return 96; + } + + float ddpi; + if (SDL_GetDisplayDPI(displayIndex, &ddpi, NULL, NULL) != 0) { + return 96; + } + + return (int)ddpi; +} diff --git a/intern/ghost/intern/GHOST_WindowSDL.h b/intern/ghost/intern/GHOST_WindowSDL.h index 5f658e8ad01..96104ec28b4 100644 --- a/intern/ghost/intern/GHOST_WindowSDL.h +++ b/intern/ghost/intern/GHOST_WindowSDL.h @@ -168,6 +168,8 @@ protected: GHOST_TSuccess beginFullScreen() const { return GHOST_kFailure; } GHOST_TSuccess endFullScreen() const { return GHOST_kFailure; } + + GHOST_TUns16 getDPIHint(); }; diff --git a/intern/ghost/intern/GHOST_WindowWin32.cpp b/intern/ghost/intern/GHOST_WindowWin32.cpp index 7d80aa43a40..fc46164c135 100644 --- a/intern/ghost/intern/GHOST_WindowWin32.cpp +++ b/intern/ghost/intern/GHOST_WindowWin32.cpp @@ -92,6 +92,7 @@ GHOST_WindowWin32::GHOST_WindowWin32(GHOST_SystemWin32 *system, m_tablet(0), m_maxPressure(0), m_normal_state(GHOST_kWindowStateNormal), + m_user32(NULL), m_parentWindowHwnd(parentwindowhwnd), m_debug_context(is_debug) { @@ -965,6 +966,23 @@ void GHOST_WindowWin32::bringTabletContextToFront() } } +GHOST_TUns16 GHOST_WindowWin32::getDPIHint() +{ + if (!m_user32) { + m_user32 = ::LoadLibrary("user32.dll"); + } + + if (m_user32) { + GHOST_WIN32_GetDpiForWindow fpGetDpiForWindow = (GHOST_WIN32_GetDpiForWindow) ::GetProcAddress(m_user32, "GetDpiForWindow"); + + if (fpGetDpiForWindow) { + return fpGetDpiForWindow(this->m_hWnd); + } + } + + return USER_DEFAULT_SCREEN_DPI; +} + /** Reverse the bits in a GHOST_TUns8 */ static GHOST_TUns8 uns8ReverseBits(GHOST_TUns8 ch) { diff --git a/intern/ghost/intern/GHOST_WindowWin32.h b/intern/ghost/intern/GHOST_WindowWin32.h index a1cf58c9ceb..75a33951ff4 100644 --- a/intern/ghost/intern/GHOST_WindowWin32.h +++ b/intern/ghost/intern/GHOST_WindowWin32.h @@ -58,6 +58,12 @@ typedef BOOL (API * GHOST_WIN32_WTClose)(HCTX); typedef BOOL (API * GHOST_WIN32_WTPacket)(HCTX, UINT, LPVOID); typedef BOOL (API * GHOST_WIN32_WTOverlap)(HCTX, BOOL); +// typedefs for user32 functions to allow dynamic loading of Windows 10 DPI scaling functions +typedef UINT(API * GHOST_WIN32_GetDpiForWindow)(HWND); +#ifndef USER_DEFAULT_SCREEN_DPI +#define USER_DEFAULT_SCREEN_DPI 96 +#endif // USER_DEFAULT_SCREEN_DPI + /** * GHOST window on M$ Windows OSs. * \author Maarten Gribnau @@ -251,6 +257,8 @@ public: GHOST_TSuccess endFullScreen() const {return GHOST_kFailure;} + GHOST_TUns16 getDPIHint() override; + /** if the window currently resizing */ bool m_inLiveResize; @@ -351,6 +359,9 @@ private: GHOST_TWindowState m_normal_state; + /** user32 dll handle*/ + HMODULE m_user32; + /** Hwnd to parent window */ GHOST_TEmbedderWindowID m_parentWindowHwnd; diff --git a/intern/ghost/intern/GHOST_WindowX11.cpp b/intern/ghost/intern/GHOST_WindowX11.cpp index 47fbe1256b1..2019f58251f 100644 --- a/intern/ghost/intern/GHOST_WindowX11.cpp +++ b/intern/ghost/intern/GHOST_WindowX11.cpp @@ -56,6 +56,9 @@ # include <X11/extensions/XInput2.h> #endif +//For DPI value +#include <X11/Xresource.h> + #if defined(__sun__) || defined(__sun) || defined(__sparc) || defined(__sparc__) || defined(_AIX) # include <strings.h> #endif @@ -68,6 +71,7 @@ #include <algorithm> #include <string> +#include <math.h> /* For obscure full screen mode stuff * lifted verbatim from blut. */ @@ -1672,3 +1676,44 @@ endFullScreen() const return GHOST_kSuccess; } + +GHOST_TUns16 +GHOST_WindowX11:: +getDPIHint() +{ + /* Try to read DPI setting set using xrdb */ + char* resMan = XResourceManagerString(m_display); + if (resMan) { + XrmDatabase xrdb = XrmGetStringDatabase(resMan); + if (xrdb) { + char* type = NULL; + XrmValue val; + + int success = XrmGetResource(xrdb, "Xft.dpi", "Xft.Dpi", &type, &val); + if (success && type) { + if (strcmp(type, "String") == 0) { + return atoi((char*)val.addr); + } + } + } + } + + /* Fallback to calculating DPI using X reported DPI, set using xrandr --dpi */ + XWindowAttributes attr; + if (!XGetWindowAttributes(m_display, m_window, &attr)) { + /* Failed to get window attributes, return X11 default DPI */ + return 96; + } + + Screen* screen = attr.screen; + int pixelWidth = WidthOfScreen(screen); + int pixelHeight = HeightOfScreen(screen); + int mmWidth = WidthMMOfScreen(screen); + int mmHeight = HeightMMOfScreen(screen); + + double pixelDiagonal = sqrt((pixelWidth * pixelWidth) + (pixelHeight * pixelHeight)); + double mmDiagonal = sqrt((mmWidth * mmWidth) + (mmHeight * mmHeight)); + float inchDiagonal = mmDiagonal * 0.039f; + int dpi = pixelDiagonal / inchDiagonal; + return dpi; +} diff --git a/intern/ghost/intern/GHOST_WindowX11.h b/intern/ghost/intern/GHOST_WindowX11.h index 9380aa9d631..5c54c1e8162 100644 --- a/intern/ghost/intern/GHOST_WindowX11.h +++ b/intern/ghost/intern/GHOST_WindowX11.h @@ -235,6 +235,8 @@ public: GHOST_TSuccess endFullScreen() const; + GHOST_TUns16 getDPIHint(); + protected: /** * \param type The type of rendering context create. diff --git a/intern/libmv/libmv/numeric/numeric.h b/intern/libmv/libmv/numeric/numeric.h index a42dab8c7a2..1a23d653676 100644 --- a/intern/libmv/libmv/numeric/numeric.h +++ b/intern/libmv/libmv/numeric/numeric.h @@ -36,7 +36,7 @@ #if !defined(__MINGW64__) # if defined(_WIN32) || defined(__APPLE__) || \ defined(__FreeBSD__) || defined(__NetBSD__) -static void sincos(double x, double *sinx, double *cosx) { +inline void sincos(double x, double *sinx, double *cosx) { *sinx = sin(x); *cosx = cos(x); } @@ -5,8 +5,8 @@ REM This is for users who like to configure & build Blender with a single comman setlocal ENABLEEXTENSIONS set BLENDER_DIR=%~dp0 set BLENDER_DIR_NOSPACES=%BLENDER_DIR: =% -if not "%BLENDER_DIR%"=="%BLENDER_DIR_NOSPACES%" ( - echo There are spaces detected in the build path "%BLENDER_DIR%", this is currently not supported, exiting.... +if not "%BLENDER_DIR%"=="%BLENDER_DIR_NOSPACES%" ( + echo There are spaces detected in the build path "%BLENDER_DIR%", this is currently not supported, exiting.... goto EOF ) set BUILD_DIR=%BLENDER_DIR%..\build_windows @@ -79,7 +79,7 @@ if NOT "%1" == "" ( set NOBUILD=1 ) else if "%1" == "showhash" ( for /f "delims=" %%i in ('git rev-parse HEAD') do echo Branch_hash=%%i - cd release/datafiles/locale + cd release/datafiles/locale for /f "delims=" %%i in ('git rev-parse HEAD') do echo Locale_hash=%%i cd %~dp0 cd release/scripts/addons @@ -132,13 +132,13 @@ if "%BUILD_ARCH%"=="x64" ( if "%target%"=="Release" ( - rem for vc12 check for both cuda 7.5 and 8 + rem for vc12 check for both cuda 7.5 and 8 if "%CUDA_PATH%"=="" ( echo Cuda Not found, aborting! goto EOF ) set BUILD_CMAKE_ARGS=%BUILD_CMAKE_ARGS% ^ - -C"%BLENDER_DIR%\build_files\cmake\config\blender_release.cmake" + -C"%BLENDER_DIR%\build_files\cmake\config\blender_release.cmake" ) :DetectMSVC @@ -157,7 +157,7 @@ if DEFINED MSVC_VC_DIR goto msvc_detect_finally if DEFINED MSVC_VC_DIR call "%MSVC_VC_DIR%\vcvarsall.bat" if DEFINED MSVC_VC_DIR goto sanity_checks -rem MSVC Build environment 2017 and up. +rem MSVC Build environment 2017 and up. for /F "usebackq skip=2 tokens=1-2*" %%A IN (`REG QUERY "HKEY_LOCAL_MACHINE\SOFTWARE\Wow6432Node\Microsoft\VisualStudio\SXS\VS7" /v %BUILD_VS_VER%.0 2^>nul`) DO set MSVC_VS_DIR=%%C if DEFINED MSVC_VS_DIR goto msvc_detect_finally_2017 REM Check 32 bits @@ -202,7 +202,7 @@ if NOT EXIST %BLENDER_DIR%..\lib\nul ( if "%TARGET%"=="" ( echo Error: Convenience target not set echo This is required for building, aborting! - echo . + echo . goto HELP ) @@ -266,15 +266,15 @@ echo. echo At any point you can optionally modify your build configuration by editing: echo "%BUILD_DIR%\CMakeCache.txt", then run "make" again to build with the changes applied. echo. -echo Blender successfully built, run from: "%BUILD_DIR%\bin\%BUILD_TYPE%" +echo Blender successfully built, run from: "%BUILD_DIR%\bin\%BUILD_TYPE%\blender.exe" echo. goto EOF :HELP echo. echo Convenience targets - echo - release ^(identical to the offical blender.org builds^) + echo - release ^(identical to the official blender.org builds^) echo - full ^(same as release minus the cuda kernels^) - echo - lite + echo - lite echo - headless echo - cycles echo - bpy @@ -289,11 +289,10 @@ goto EOF echo - with_tests ^(enable building unit tests^) echo - debug ^(Build an unoptimized debuggable build^) echo - packagename [newname] ^(override default cpack package name^) - echo - x86 ^(override host autodetect and build 32 bit code^) - echo - x64 ^(override host autodetect and build 64 bit code^) + echo - x86 ^(override host auto-detect and build 32 bit code^) + echo - x64 ^(override host auto-detect and build 64 bit code^) echo - 2013 ^(build with visual studio 2013^) echo - 2015 ^(build with visual studio 2015^) [EXPERIMENTAL] echo. :EOF - diff --git a/release/datafiles/blender_icons.svg b/release/datafiles/blender_icons.svg index e9c114ba1bd..d88788fa904 100644 --- a/release/datafiles/blender_icons.svg +++ b/release/datafiles/blender_icons.svg @@ -14,7 +14,7 @@ height="640" id="svg2" sodipodi:version="0.32" - inkscape:version="0.91 r" + inkscape:version="0.91 r13725" version="1.0" sodipodi:docname="blender_icons.svg" inkscape:output_extension="org.inkscape.output.svg.inkscape" @@ -31338,6 +31338,26 @@ d="m 125.5,433.5 23,0 0,41 -33,0 0,-31 10,-10 z" style="display:inline;fill:url(#linearGradient13110);fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:1;marker:none" /> </clipPath> + <linearGradient + inkscape:collect="always" + xlink:href="#linearGradient1610-6" + id="linearGradient18199" + gradientUnits="userSpaceOnUse" + x1="189.76083" + y1="248.13905" + x2="116.05637" + y2="183.6826" /> + <radialGradient + inkscape:collect="always" + xlink:href="#linearGradient22562" + id="radialGradient23167-6" + gradientUnits="userSpaceOnUse" + gradientTransform="matrix(0.99220964,-0.12457927,0.11585516,0.92272644,-34.13325,22.766225)" + cx="-0.78262758" + cy="294.63174" + fx="-0.78262758" + fy="294.63174" + r="6.6750002" /> </defs> <sodipodi:namedview id="base" @@ -31349,16 +31369,16 @@ objecttolerance="10000" inkscape:pageopacity="0.0" inkscape:pageshadow="2" - inkscape:zoom="14.413868" - inkscape:cx="480.24726" - inkscape:cy="269.95478" + inkscape:zoom="19.997864" + inkscape:cx="462.52244" + inkscape:cy="435.14241" inkscape:document-units="px" - inkscape:current-layer="layer1" + inkscape:current-layer="g23149-4" showgrid="true" - inkscape:window-width="1680" - inkscape:window-height="1020" - inkscape:window-x="0" - inkscape:window-y="30" + inkscape:window-width="1920" + inkscape:window-height="1025" + inkscape:window-x="-8" + inkscape:window-y="-8" inkscape:snap-nodes="true" inkscape:snap-bbox="true" showguides="true" @@ -92660,6 +92680,56 @@ style="opacity:0.51999996;fill:url(#radialGradient21448-8-143);fill-opacity:1;fill-rule:evenodd;stroke:none" /> </g> </g> + <g + transform="translate(335.99871,21.048284)" + style="display:inline;enable-background:new" + id="ICON_ROTATE-7"> + <rect + y="178" + x="110" + height="16" + width="16" + id="rect37989-8" + style="display:inline;overflow:visible;visibility:visible;opacity:0;fill:#b3b3b3;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:2.79999995;marker:none;enable-background:accumulate" /> + <path + style="display:inline;overflow:visible;visibility:visible;fill:none;stroke:#000000;stroke-width:2.4000001;stroke-linecap:square;stroke-linejoin:round;stroke-miterlimit:4;stroke-dasharray:none;stroke-dashoffset:0;stroke-opacity:1;marker:none;enable-background:new" + d="m 114.5,192.5 -3,0 0,-3 m 13,0 0,3 -3,0 m -0.25,-13 3.25,0 0,3 m -13,0 0,-3 3,0" + id="path37498-7" + sodipodi:nodetypes="cccccccccccc" + inkscape:connector-curvature="0" /> + <path + sodipodi:nodetypes="cccccccccccc" + id="rect38140-7" + d="m 114.5,192.5 -3,0 0,-3 m 13,0 0,3 -3,0 m -0.25,-13 3.25,0 0,3 m -13,0 0,-3 3,0" + style="display:inline;overflow:visible;visibility:visible;fill:none;stroke:url(#linearGradient18199);stroke-width:1;stroke-linecap:square;stroke-linejoin:miter;stroke-miterlimit:4;stroke-dashoffset:0;stroke-opacity:1;marker:none;enable-background:new" + inkscape:connector-curvature="0" /> + <g + transform="matrix(0.59971056,0,0,0.59971056,116.78278,9.7425599)" + style="display:inline;enable-background:new" + id="g23145-9"> + <g + id="g23149-4"> + <path + id="path39832-9" + style="display:inline;overflow:visible;visibility:visible;fill:none;stroke:#1a1a1a;stroke-width:4.66725159;stroke-linecap:square;stroke-linejoin:round;stroke-miterlimit:4;stroke-dasharray:none;stroke-dashoffset:0;stroke-opacity:1;marker:none;enable-background:accumulate" + d="m -4.3682386,287.81345 1.5,0 c 0.999089,0 2.07885534,1.30514 2.50490386,2.78207 1.06592652,3.69512 2.80867074,9.82446 5.88525404,9.96406 2.6782554,0 1.6181317,-5.11535 3.1736046,-5.26275 l 0.25,0" + sodipodi:nodetypes="cssccc" + inkscape:connector-curvature="0" /> + <path + sodipodi:nodetypes="ccscc" + d="m 9.3647983,295.22328 -0.4793018,0 c -2.2335161,0 0.1796731,4.94901 -3.4398065,5.09984 -4.44796752,0.18536 -5.37272213,-12.59185 -8.0767581,-12.56237 l -2,0" + style="display:inline;overflow:visible;visibility:visible;fill:none;stroke:#a8df84;stroke-width:2.93474906;stroke-linecap:square;stroke-linejoin:round;stroke-miterlimit:4;stroke-dasharray:none;stroke-dashoffset:0;stroke-opacity:1;marker:none;enable-background:accumulate" + id="path39834-2" + inkscape:connector-curvature="0" /> + </g> + <path + id="path39836-9" + style="display:inline;overflow:visible;visibility:visible;opacity:0.35;fill:none;stroke:url(#radialGradient23167-6);stroke-width:3.53503864;stroke-linecap:round;stroke-linejoin:round;stroke-miterlimit:4;stroke-dasharray:none;stroke-dashoffset:0;stroke-opacity:1;marker:none;enable-background:accumulate" + d="M 5.6770841,300.48165 C 0.7393262,300.21066 0.54777814,287.99792 -2.9522219,287.99792" + sodipodi:nodetypes="cc" + inkscape:connector-curvature="0" /> + </g> + </g> </g> <g inkscape:groupmode="layer" diff --git a/release/datafiles/blender_icons16/icon16_normalize_fcurves.dat b/release/datafiles/blender_icons16/icon16_normalize_fcurves.dat Binary files differnew file mode 100644 index 00000000000..fe118a23d79 --- /dev/null +++ b/release/datafiles/blender_icons16/icon16_normalize_fcurves.dat diff --git a/release/datafiles/blender_icons32/icon32_normalize_fcurves.dat b/release/datafiles/blender_icons32/icon32_normalize_fcurves.dat Binary files differnew file mode 100644 index 00000000000..5b1f546a563 --- /dev/null +++ b/release/datafiles/blender_icons32/icon32_normalize_fcurves.dat diff --git a/release/scripts/freestyle/modules/freestyle/shaders.py b/release/scripts/freestyle/modules/freestyle/shaders.py index 633def38b5b..bce6642220b 100644 --- a/release/scripts/freestyle/modules/freestyle/shaders.py +++ b/release/scripts/freestyle/modules/freestyle/shaders.py @@ -568,7 +568,7 @@ class pyRandomColorShader(StrokeShader): class py2DCurvatureColorShader(StrokeShader): """ - Assigns a color (greyscale) to the stroke based on the curvature. + Assigns a color (grayscale) to the stroke based on the curvature. A higher curvature will yield a brighter color. """ def shade(self, stroke): @@ -584,7 +584,7 @@ class py2DCurvatureColorShader(StrokeShader): class pyTimeColorShader(StrokeShader): """ - Assigns a greyscale value that increases for every vertex. + Assigns a grayscale value that increases for every vertex. The brightness will increase along the stroke. """ def __init__(self, step=0.01): diff --git a/release/scripts/freestyle/modules/parameter_editor.py b/release/scripts/freestyle/modules/parameter_editor.py index 93305cb7c5a..b093920a4cb 100644 --- a/release/scripts/freestyle/modules/parameter_editor.py +++ b/release/scripts/freestyle/modules/parameter_editor.py @@ -1170,6 +1170,7 @@ class Seed: _seed = Seed() + def get_dashed_pattern(linestyle): """Extracts the dashed pattern from the various UI options """ pattern = [] @@ -1185,6 +1186,15 @@ def get_dashed_pattern(linestyle): return pattern +def get_grouped_objects(group): + for ob in group.objects: + if ob.dupli_type == 'GROUP' and ob.dupli_group is not None: + for dupli in get_grouped_objects(ob.dupli_group): + yield dupli + else: + yield ob + + integration_types = { 'MEAN': IntegrationType.MEAN, 'MIN': IntegrationType.MIN, @@ -1267,7 +1277,7 @@ def process(layer_name, lineset_name): # prepare selection criteria by group of objects if lineset.select_by_group: if lineset.group is not None: - names = {getQualifiedObjectName(ob): True for ob in lineset.group.objects} + names = {getQualifiedObjectName(ob): True for ob in get_grouped_objects(lineset.group)} upred = ObjectNamesUP1D(names, lineset.group_negation == 'EXCLUSIVE') selection_criteria.append(upred) # prepare selection criteria by image border diff --git a/release/scripts/modules/addon_utils.py b/release/scripts/modules/addon_utils.py index 0f096f5812c..51e3e65b78c 100644 --- a/release/scripts/modules/addon_utils.py +++ b/release/scripts/modules/addon_utils.py @@ -24,6 +24,7 @@ __all__ = ( "check", "enable", "disable", + "disable_all", "reset_all", "module_bl_info", ) @@ -31,8 +32,9 @@ __all__ = ( import bpy as _bpy _user_preferences = _bpy.context.user_preferences -error_duplicates = False error_encoding = False +# (name, file, path) +error_duplicates = [] addons_fake_modules = {} @@ -57,12 +59,11 @@ def paths(): def modules_refresh(module_cache=addons_fake_modules): - global error_duplicates global error_encoding import os - error_duplicates = False error_encoding = False + error_duplicates.clear() path_list = paths() @@ -168,7 +169,7 @@ def modules_refresh(module_cache=addons_fake_modules): if mod.__file__ != mod_path: print("multiple addons with the same name:\n %r\n %r" % (mod.__file__, mod_path)) - error_duplicates = True + error_duplicates.append((mod.bl_info["name"], mod.__file__, mod_path)) elif mod.__time__ != os.path.getmtime(mod_path): print("reloading addon:", @@ -444,6 +445,13 @@ def reset_all(*, reload_scripts=False): disable(mod_name) +def disable_all(): + import sys + for mod_name, mod in sys.modules.items(): + if getattr(mod, "__addon_enabled__", False): + disable(mod_name) + + def module_bl_info(mod, info_basis=None): if info_basis is None: info_basis = { diff --git a/release/scripts/modules/bl_app_override/__init__.py b/release/scripts/modules/bl_app_override/__init__.py new file mode 100644 index 00000000000..89cc8a0eb28 --- /dev/null +++ b/release/scripts/modules/bl_app_override/__init__.py @@ -0,0 +1,202 @@ +# ##### BEGIN GPL LICENSE BLOCK ##### +# +# This program is free software; you can redistribute it and/or +# modify it under the terms of the GNU General Public License +# as published by the Free Software Foundation; either version 2 +# of the License, or (at your option) any later version. +# +# This program is distributed in the hope that it will be useful, +# but WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +# GNU General Public License for more details. +# +# You should have received a copy of the GNU General Public License +# along with this program; if not, write to the Free Software Foundation, +# Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. +# +# ##### END GPL LICENSE BLOCK ##### + +# <pep8-80 compliant> + +""" +Module to manage overriding various parts of Blender. + +Intended for use with 'app_templates', though it can be used from anywhere. +""" + + +# TODO, how to check these aren't from add-ons. +# templates might need to un-register while filtering. +def class_filter(cls_parent, **kw): + whitelist = kw.pop("whitelist", None) + blacklist = kw.pop("blacklist", None) + kw_items = tuple(kw.items()) + for cls in cls_parent.__subclasses__(): + # same as is_registered() + if "bl_rna" in cls.__dict__: + if blacklist is not None and cls.__name__ in blacklist: + continue + if ((whitelist is not None and cls.__name__ is whitelist) or + all((getattr(cls, attr) in expect) for attr, expect in kw_items)): + yield cls + + +def ui_draw_filter_register( + *, + ui_ignore_classes=None, + ui_ignore_operator=None, + ui_ignore_property=None, + ui_ignore_menu=None, + ui_ignore_label=None +): + import bpy + + UILayout = bpy.types.UILayout + + if ui_ignore_classes is None: + ui_ignore_classes = ( + bpy.types.Panel, + bpy.types.Menu, + bpy.types.Header, + ) + + class OperatorProperties_Fake: + pass + + class UILayout_Fake(bpy.types.UILayout): + __slots__ = () + + def __getattribute__(self, attr): + # ensure we always pass down UILayout_Fake instances + if attr in {"row", "split", "column", "box", "column_flow"}: + real_func = UILayout.__getattribute__(self, attr) + + def dummy_func(*args, **kw): + # print("wrapped", attr) + ret = real_func(*args, **kw) + return UILayout_Fake(ret) + return dummy_func + + elif attr in {"operator", "operator_menu_enum", "operator_enum"}: + if ui_ignore_operator is None: + return UILayout.__getattribute__(self, attr) + + real_func = UILayout.__getattribute__(self, attr) + + def dummy_func(*args, **kw): + # print("wrapped", attr) + if not ui_ignore_operator(args[0]): + ret = real_func(*args, **kw) + else: + # UILayout.__getattribute__(self, "label")() + # may need to be set + ret = OperatorProperties_Fake() + return ret + return dummy_func + + elif attr in {"prop", "prop_enum"}: + if ui_ignore_property is None: + return UILayout.__getattribute__(self, attr) + + real_func = UILayout.__getattribute__(self, attr) + + def dummy_func(*args, **kw): + # print("wrapped", attr) + if not ui_ignore_property(args[0].__class__.__name__, args[1]): + ret = real_func(*args, **kw) + else: + ret = None + return ret + return dummy_func + + elif attr == "menu": + if ui_ignore_menu is None: + return UILayout.__getattribute__(self, attr) + + real_func = UILayout.__getattribute__(self, attr) + + def dummy_func(*args, **kw): + # print("wrapped", attr) + if not ui_ignore_menu(args[0]): + ret = real_func(*args, **kw) + else: + ret = None + return ret + return dummy_func + + elif attr == "label": + if ui_ignore_label is None: + return UILayout.__getattribute__(self, attr) + + real_func = UILayout.__getattribute__(self, attr) + + def dummy_func(*args, **kw): + # print("wrapped", attr) + if not ui_ignore_label(args[0] if args else kw.get("text", "")): + ret = real_func(*args, **kw) + else: + # ret = real_func() + ret = None + return ret + return dummy_func + else: + return UILayout.__getattribute__(self, attr) + # print(self, attr) + + def operator(*args, **kw): + return super().operator(*args, **kw) + + def draw_override(func_orig, self_real, context): + cls_real = self_real.__class__ + if cls_real is super: + # simple, no wrapping + return func_orig(self_real, context) + + class Wrapper(cls_real): + __slots__ = () + def __getattribute__(self, attr): + if attr == "layout": + return UILayout_Fake(self_real.layout) + else: + cls = super() + try: + return cls.__getattr__(self, attr) + except AttributeError: + # class variable + try: + return getattr(cls, attr) + except AttributeError: + # for preset bl_idname access + return getattr(UILayout(self), attr) + + @property + def layout(self): + # print("wrapped") + return self_real.layout + + return func_orig(Wrapper(self_real), context) + + ui_ignore_store = [] + + for cls in ui_ignore_classes: + for subcls in list(cls.__subclasses__()): + if "draw" in subcls.__dict__: # don't want to get parents draw() + + def replace_draw(): + # function also serves to hold draw_old in a local name-space + draw_orig = subcls.draw + + def draw(self, context): + return draw_override(draw_orig, self, context) + subcls.draw = draw + + ui_ignore_store.append((subcls, "draw", subcls.draw)) + + replace_draw() + + return ui_ignore_store + + +def ui_draw_filter_unregister(ui_ignore_store): + for (obj, attr, value) in ui_ignore_store: + setattr(obj, attr, value) diff --git a/release/scripts/modules/bl_app_override/helpers.py b/release/scripts/modules/bl_app_override/helpers.py new file mode 100644 index 00000000000..981039e8ddc --- /dev/null +++ b/release/scripts/modules/bl_app_override/helpers.py @@ -0,0 +1,167 @@ +# ##### BEGIN GPL LICENSE BLOCK ##### +# +# This program is free software; you can redistribute it and/or +# modify it under the terms of the GNU General Public License +# as published by the Free Software Foundation; either version 2 +# of the License, or (at your option) any later version. +# +# This program is distributed in the hope that it will be useful, +# but WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +# GNU General Public License for more details. +# +# You should have received a copy of the GNU General Public License +# along with this program; if not, write to the Free Software Foundation, +# Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. +# +# ##### END GPL LICENSE BLOCK ##### + +# <pep8-80 compliant> + +# ----------------------------------------------------------------------------- +# AppOverrideState + + +class AppOverrideState: + """ + Utility class to encapsulate overriding the application state + so that settings can be restored afterwards. + """ + __slots__ = ( + # setup_classes + "_class_store", + # setup_ui_ignore + "_ui_ignore_store", + # setup_addons + "_addon_store", + ) + + # --------- + # Callbacks + # + # Set as None, to make it simple to check if they're being overridden. + + # setup/teardown classes + class_ignore = None + + # setup/teardown ui_ignore + ui_ignore_classes = None + ui_ignore_operator = None + ui_ignore_property = None + ui_ignore_menu = None + ui_ignore_label = None + + addon_paths = None + addons = None + + # End callbacks + + def __init__(self): + self._class_store = None + self._addon_store = None + self._ui_ignore_store = None + + def _setup_classes(self): + import bpy + assert(self._class_store is None) + self._class_store = self.class_ignore() + from bpy.utils import unregister_class + for cls in self._class_store: + unregister_class(cls) + + def _teardown_classes(self): + assert(self._class_store is not None) + + from bpy.utils import register_class + for cls in self._class_store: + register_class(cls) + self._class_store = None + + def _setup_ui_ignore(self): + import bl_app_override + + self._ui_ignore_store = bl_app_override.ui_draw_filter_register( + ui_ignore_classes=( + None if self.ui_ignore_classes is None + else self.ui_ignore_classes() + ), + ui_ignore_operator=self.ui_ignore_operator, + ui_ignore_property=self.ui_ignore_property, + ui_ignore_menu=self.ui_ignore_menu, + ui_ignore_label=self.ui_ignore_label, + ) + + def _teardown_ui_ignore(self): + import bl_app_override + bl_app_override.ui_draw_filter_unregister( + self._ui_ignore_store + ) + self._ui_ignore_store = None + + def _setup_addons(self): + import sys + import os + + sys_path = [] + if self.addon_paths is not None: + for path in self.addon_paths(): + if path not in sys.path: + sys.path.append(path) + + import addon_utils + addons = [] + if self.addons is not None: + addons.extend(self.addons()) + for addon in addons: + addon_utils.enable(addon) + + self._addon_store = { + "sys_path": sys_path, + "addons": addons, + } + + def _teardown_addons(self): + import sys + + sys_path = self._addon_store["sys_path"] + for path in sys_path: + # should always succeed, but if not it doesn't matter + # (someone else was changing the sys.path), ignore! + try: + sys.path.remove(path) + except: + pass + + addons = self._addon_store["addons"] + import addon_utils + for addon in addons: + addon_utils.disable(addon) + + self._addon_store.clear() + self._addon_store = None + + def setup(self): + if self.class_ignore is not None: + self._setup_classes() + + if any((self.addon_paths, + self.addons, + )): + self._setup_addons() + + if any((self.ui_ignore_operator, + self.ui_ignore_property, + self.ui_ignore_menu, + self.ui_ignore_label, + )): + self._setup_ui_ignore() + + def teardown(self): + if self._class_store is not None: + self._teardown_classes() + + if self._addon_store is not None: + self._teardown_addons() + + if self._ui_ignore_store is not None: + self._teardown_ui_ignore() diff --git a/release/scripts/modules/bl_app_template_utils.py b/release/scripts/modules/bl_app_template_utils.py new file mode 100644 index 00000000000..b3a4824aa7b --- /dev/null +++ b/release/scripts/modules/bl_app_template_utils.py @@ -0,0 +1,198 @@ +# ##### BEGIN GPL LICENSE BLOCK ##### +# +# This program is free software; you can redistribute it and/or +# modify it under the terms of the GNU General Public License +# as published by the Free Software Foundation; either version 2 +# of the License, or (at your option) any later version. +# +# This program is distributed in the hope that it will be useful, +# but WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +# GNU General Public License for more details. +# +# You should have received a copy of the GNU General Public License +# along with this program; if not, write to the Free Software Foundation, +# Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. +# +# ##### END GPL LICENSE BLOCK ##### + +# <pep8-80 compliant> + +""" +Similar to ``addon_utils``, except we can only have one active at a time. + +In most cases users of this module will simply call 'activate'. +""" + +__all__ = ( + "activate", + "import_from_path", + "import_from_id", + "reset", +) + +import bpy as _bpy + +# Normally matches 'user_preferences.app_template_id', +# but loading new preferences will get us out of sync. +_app_template = { + "id": "", +} + +# instead of sys.modules +# note that we only ever have one template enabled at a time +# so it may not seem necessary to use this. +# +# However, templates may want to share between each-other, +# so any loaded modules are stored here? +# +# Note that the ID here is the app_template_id , not the modules __name__. +_modules = {} + + +def _enable(template_id, *, handle_error=None, ignore_not_found=False): + import os + import sys + from bpy_restrict_state import RestrictBlend + + if handle_error is None: + def handle_error(ex): + import traceback + traceback.print_exc() + + # Split registering up into 2 steps so we can undo + # if it fails par way through. + + # disable the context, using the context at all is + # really bad while loading an template, don't do it! + with RestrictBlend(): + + # 1) try import + try: + mod = import_from_id(template_id, ignore_not_found=ignore_not_found) + if mod is None: + return None + mod.__template_enabled__ = False + _modules[template_id] = mod + except Exception as ex: + handle_error(ex) + return None + + # 2) try run the modules register function + try: + mod.register() + except Exception as ex: + print("Exception in module register(): %r" % + getattr(mod, "__file__", template_id)) + handle_error(ex) + del _modules[template_id] + return None + + # * OK loaded successfully! * + mod.__template_enabled__ = True + + if _bpy.app.debug_python: + print("\tapp_template_utils.enable", mod.__name__) + + return mod + + +def _disable(template_id, *, handle_error=None): + """ + Disables a template by name. + + :arg template_id: The name of the template and module. + :type template_id: string + :arg handle_error: Called in the case of an error, + taking an exception argument. + :type handle_error: function + """ + import sys + + if handle_error is None: + def handle_error(ex): + import traceback + traceback.print_exc() + + mod = _modules.get(template_id) + + if mod and getattr(mod, "__template_enabled__", False) is not False: + mod.__template_enabled__ = False + + try: + mod.unregister() + except Exception as ex: + print("Exception in module unregister(): %r" % + getattr(mod, "__file__", template_id)) + handle_error(ex) + else: + print("\tapp_template_utils.disable: %s not %s." % + (template_id, "disabled" if mod is None else "loaded")) + + if _bpy.app.debug_python: + print("\tapp_template_utils.disable", template_id) + + +def import_from_path(path, ignore_not_found=False): + import os + from importlib import import_module + base_module, template_id = path.rsplit(os.sep, 2)[-2:] + module_name = base_module + "." + template_id + + try: + return import_module(module_name) + except ModuleNotFoundError as ex: + if ignore_not_found and ex.name == module_name: + return None + raise ex + + +def import_from_id(template_id, ignore_not_found=False): + import os + path = next(iter(_bpy.utils.app_template_paths(template_id)), None) + if path is None: + if ignore_not_found: + return None + else: + raise Exception("%r template not found!" % template_id) + else: + if ignore_not_found: + if not os.path.exists(os.path.join(path, "__init__.py")): + return None + return import_from_path(path, ignore_not_found=ignore_not_found) + + +def activate(template_id=None): + template_id_prev = _app_template["id"] + + # not needed but may as well avoid activating same template + # ... in fact keep this, it will show errors early on! + """ + if template_id_prev == template_id: + return + """ + + if template_id_prev: + _disable(template_id_prev) + + # Disable all addons, afterwards caller must reset. + import addon_utils + addon_utils.disable_all() + + # ignore_not_found so modules that don't contain scripts don't raise errors + mod = _enable(template_id, ignore_not_found=True) if template_id else None + + _app_template["id"] = template_id + + +def reset(*, reload_scripts=False): + """ + Sets default state. + """ + template_id = _bpy.context.user_preferences.app_template + if _bpy.app.debug_python: + print("bl_app_template_utils.reset('%s')" % template_id) + + # TODO reload_scripts + + activate(template_id) diff --git a/release/scripts/modules/bpy/__init__.py b/release/scripts/modules/bpy/__init__.py index 26fdbc8cc56..545b891505f 100644 --- a/release/scripts/modules/bpy/__init__.py +++ b/release/scripts/modules/bpy/__init__.py @@ -48,11 +48,11 @@ def main(): import sys # Possibly temp. addons path - from os.path import join, dirname, normpath - sys.path.append(normpath(join(dirname(__file__), - "..", "..", "addons", "modules"))) - sys.path.append(join(utils.user_resource('SCRIPTS'), - "addons", "modules")) + from os.path import join, dirname + sys.path.extend([ + join(dirname(dirname(dirname(__file__))), "addons", "modules"), + join(utils.user_resource('SCRIPTS'), "addons", "modules"), + ]) # fake module to allow: # from bpy.types import Panel diff --git a/release/scripts/modules/bpy/path.py b/release/scripts/modules/bpy/path.py index a864a86eba7..e17d710068c 100644 --- a/release/scripts/modules/bpy/path.py +++ b/release/scripts/modules/bpy/path.py @@ -204,7 +204,9 @@ def display_name(name): name = name.replace("_colon_", ":") name = name.replace("_plus_", "+") - name = name.replace("_", " ") + # strip to allow underscore prefix + # (when paths can't start with numbers for eg). + name = name.replace("_", " ").lstrip(" ") if name.islower(): name = name.lower().title() diff --git a/release/scripts/modules/bpy/utils/__init__.py b/release/scripts/modules/bpy/utils/__init__.py index 31dd836e034..1d555ae7123 100644 --- a/release/scripts/modules/bpy/utils/__init__.py +++ b/release/scripts/modules/bpy/utils/__init__.py @@ -32,6 +32,7 @@ __all__ = ( "preset_find", "preset_paths", "refresh_script_paths", + "app_template_paths", "register_class", "register_module", "register_manual_map", @@ -49,18 +50,18 @@ __all__ = ( "unregister_class", "unregister_module", "user_resource", - ) +) from _bpy import ( - _utils_units as units, - blend_paths, - escape_identifier, - register_class, - resource_path, - script_paths as _bpy_script_paths, - unregister_class, - user_resource as _user_resource, - ) + _utils_units as units, + blend_paths, + escape_identifier, + register_class, + resource_path, + script_paths as _bpy_script_paths, + unregister_class, + user_resource as _user_resource, +) import bpy as _bpy import os as _os @@ -142,7 +143,7 @@ def load_scripts(reload_scripts=False, refresh_scripts=False): as modules. :type refresh_scripts: bool """ - use_time = _bpy.app.debug_python + use_time = use_class_register_check = _bpy.app.debug_python if use_time: import time @@ -161,7 +162,8 @@ def load_scripts(reload_scripts=False, refresh_scripts=False): for module_name in [ext.module for ext in _user_preferences.addons]: _addon_utils.disable(module_name) - # *AFTER* unregistering all add-ons, otherwise all calls to unregister_module() will silently fail (do nothing). + # *AFTER* unregistering all add-ons, otherwise all calls to + # unregister_module() will silently fail (do nothing). _bpy_types.TypeMap.clear() def register_module_call(mod): @@ -245,6 +247,12 @@ def load_scripts(reload_scripts=False, refresh_scripts=False): for mod in modules_from_path(path, loaded_modules): test_register(mod) + # load template (if set) + if any(_bpy.utils.app_template_paths()): + import bl_app_template_utils + bl_app_template_utils.reset(reload_scripts=reload_scripts) + del bl_app_template_utils + # deal with addons separately _initialize = getattr(_addon_utils, "_initialize", None) if _initialize is not None: @@ -269,13 +277,21 @@ def load_scripts(reload_scripts=False, refresh_scripts=False): if use_time: print("Python Script Load Time %.4f" % (time.time() - t_main)) + if use_class_register_check: + for cls in _bpy.types.bpy_struct.__subclasses__(): + if getattr(cls, "is_registered", False): + for subcls in cls.__subclasses__(): + if not subcls.is_registered: + print( + "Warning, unregistered class: %s(%s)" % + (subcls.__name__, cls.__name__) + ) + # base scripts -_scripts = _os.path.join(_os.path.dirname(__file__), - _os.path.pardir, - _os.path.pardir, - ) -_scripts = (_os.path.normpath(_scripts), ) +_scripts = ( + _os.path.dirname(_os.path.dirname(_os.path.dirname(__file__))), +) def script_path_user(): @@ -356,6 +372,38 @@ def refresh_script_paths(): _sys_path_ensure(path) +def app_template_paths(subdir=None): + """ + Returns valid application template paths. + + :arg subdir: Optional subdir. + :type subdir: string + :return: app template paths. + :rtype: generator + """ + + # note: LOCAL, USER, SYSTEM order matches script resolution order. + subdir_tuple = (subdir,) if subdir is not None else () + + path = _os.path.join(*( + resource_path('LOCAL'), "scripts", "startup", + "bl_app_templates_user", *subdir_tuple)) + if _os.path.isdir(path): + yield path + else: + path = _os.path.join(*( + resource_path('USER'), "scripts", "startup", + "bl_app_templates_user", *subdir_tuple)) + if _os.path.isdir(path): + yield path + + path = _os.path.join(*( + resource_path('SYSTEM'), "scripts", "startup", + "bl_app_templates_system", *subdir_tuple)) + if _os.path.isdir(path): + yield path + + def preset_paths(subdir): """ Returns a list of paths for a specific preset. diff --git a/release/scripts/modules/bpy_types.py b/release/scripts/modules/bpy_types.py index d64acd2ce3b..600b29a6b2b 100644 --- a/release/scripts/modules/bpy_types.py +++ b/release/scripts/modules/bpy_types.py @@ -683,6 +683,10 @@ class _GenericUI: return draw_funcs @classmethod + def is_extended(cls): + return bool(getattr(cls.draw, "_draw_funcs", None)) + + @classmethod def append(cls, draw_func): """ Append a draw function to this menu, @@ -725,11 +729,30 @@ class Header(StructRNA, _GenericUI, metaclass=RNAMeta): class Menu(StructRNA, _GenericUI, metaclass=RNAMeta): __slots__ = () - def path_menu(self, searchpaths, operator, - props_default=None, filter_ext=None): + def path_menu(self, searchpaths, operator, *, + props_default=None, prop_filepath="filepath", + filter_ext=None, filter_path=None, display_name=None): + """ + Populate a menu from a list of paths. + + :arg searchpaths: Paths to scan. + :type searchpaths: sequence of strings. + :arg operator: The operator id to use with each file. + :type operator: string + :arg prop_filepath: Optional operator filepath property (defaults to "filepath"). + :type prop_filepath: string + :arg props_default: Properties to assign to each operator. + :type props_default: dict + :arg filter_ext: Optional callback that takes the file extensions. + + Returning false excludes the file from the list. + + :type filter_ext: Callable that takes a string and returns a bool. + :arg display_name: Optional callback that takes the full path, returns the name to display. + :type display_name: Callable that takes a string and returns a string. + """ layout = self.layout - # hard coded to set the operators 'filepath' to the filename. import os import bpy.utils @@ -742,25 +765,32 @@ class Menu(StructRNA, _GenericUI, metaclass=RNAMeta): # collect paths files = [] for directory in searchpaths: - files.extend([(f, os.path.join(directory, f)) - for f in os.listdir(directory) - if (not f.startswith(".")) - if ((filter_ext is None) or - (filter_ext(os.path.splitext(f)[1]))) - ]) + files.extend( + [(f, os.path.join(directory, f)) + for f in os.listdir(directory) + if (not f.startswith(".")) + if ((filter_ext is None) or + (filter_ext(os.path.splitext(f)[1]))) + if ((filter_path is None) or + (filter_path(f))) + ]) files.sort() for f, filepath in files: - props = layout.operator(operator, - text=bpy.path.display_name(f), - translate=False) + # Intentionally pass the full path to 'display_name' callback, + # since the callback may want to use part a directory in the name. + props = layout.operator( + operator, + text=display_name(filepath) if display_name else bpy.path.display_name(f), + translate=False, + ) if props_default is not None: for attr, value in props_default.items(): setattr(props, attr, value) - props.filepath = filepath + setattr(props, prop_filepath, filepath) if operator == "script.execute_preset": props.menu_idname = self.bl_idname diff --git a/release/scripts/modules/rna_keymap_ui.py b/release/scripts/modules/rna_keymap_ui.py index 2ca7a7997a5..a1a4e5b8763 100644 --- a/release/scripts/modules/rna_keymap_ui.py +++ b/release/scripts/modules/rna_keymap_ui.py @@ -128,16 +128,15 @@ def draw_kmi(display_keymaps, kc, km, kmi, layout, level): else: box = col.column() - split = box.split(percentage=0.01) + split = box.split() # header bar - row = split.row() + row = split.row(align=True) row.prop(kmi, "show_expanded", text="", emboss=False) - - row = split.row() row.prop(kmi, "active", text="", emboss=False) if km.is_modal: + row.separator() row.prop(kmi, "propvalue", text="") else: row.label(text=kmi.name) diff --git a/release/scripts/presets/interface_theme/back_to_black.xml b/release/scripts/presets/interface_theme/back_to_black.xml index 915e9cb64f1..1636f5b5cf6 100644 --- a/release/scripts/presets/interface_theme/back_to_black.xml +++ b/release/scripts/presets/interface_theme/back_to_black.xml @@ -18,7 +18,7 @@ text_sel="#ffffff" show_shaded="TRUE" shadetop="-100" - shadedown="0"> + shadedown="5"> </ThemeWidgetColors> </wcol_regular> <wcol_tool> @@ -30,19 +30,19 @@ text_sel="#ffffff" show_shaded="TRUE" shadetop="-100" - shadedown="0"> + shadedown="5"> </ThemeWidgetColors> </wcol_tool> <wcol_radio> <ThemeWidgetColors outline="#2a2a2a" inner="#111111ff" inner_sel="#33406bff" - item="#191919ff" + item="#444444ff" text="#929292" text_sel="#ffffff" show_shaded="TRUE" shadetop="-100" - shadedown="0"> + shadedown="5"> </ThemeWidgetColors> </wcol_radio> <wcol_text> @@ -50,23 +50,23 @@ inner="#111111ff" inner_sel="#33406bff" item="#191919ff" - text="#e4e4e4" + text="#929292" text_sel="#ffffff" show_shaded="TRUE" shadetop="-100" - shadedown="0"> + shadedown="5"> </ThemeWidgetColors> </wcol_text> <wcol_option> - <ThemeWidgetColors outline="#2a2a2a" + <ThemeWidgetColors outline="#535353" inner="#111111ff" inner_sel="#33406bff" - item="#000000ff" - text="#c7c7c7" + item="#a3a3a3ff" + text="#929292" text_sel="#ffffff" show_shaded="TRUE" shadetop="-100" - shadedown="0"> + shadedown="5"> </ThemeWidgetColors> </wcol_option> <wcol_toggle> @@ -78,7 +78,7 @@ text_sel="#ffffff" show_shaded="TRUE" shadetop="-100" - shadedown="0"> + shadedown="5"> </ThemeWidgetColors> </wcol_toggle> <wcol_num> @@ -90,7 +90,7 @@ text_sel="#ffffff" show_shaded="TRUE" shadetop="-100" - shadedown="0"> + shadedown="5"> </ThemeWidgetColors> </wcol_num> <wcol_numslider> @@ -102,7 +102,7 @@ text_sel="#ffffff" show_shaded="TRUE" shadetop="-100" - shadedown="0"> + shadedown="5"> </ThemeWidgetColors> </wcol_numslider> <wcol_box> @@ -114,7 +114,7 @@ text_sel="#ffffff" show_shaded="TRUE" shadetop="-100" - shadedown="0"> + shadedown="5"> </ThemeWidgetColors> </wcol_box> <wcol_menu> @@ -126,7 +126,7 @@ text_sel="#ffffff" show_shaded="TRUE" shadetop="-100" - shadedown="0"> + shadedown="5"> </ThemeWidgetColors> </wcol_menu> <wcol_pulldown> @@ -138,7 +138,7 @@ text_sel="#ffffff" show_shaded="TRUE" shadetop="-100" - shadedown="0"> + shadedown="5"> </ThemeWidgetColors> </wcol_pulldown> <wcol_menu_back> @@ -150,7 +150,7 @@ text_sel="#ffffff" show_shaded="TRUE" shadetop="-100" - shadedown="0"> + shadedown="5"> </ThemeWidgetColors> </wcol_menu_back> <wcol_pie_menu> @@ -170,7 +170,7 @@ inner="#191919e6" inner_sel="#2d2d2de6" item="#646464ff" - text="#ffffff" + text="#929292" text_sel="#ffffff" show_shaded="FALSE" shadetop="25" @@ -186,7 +186,7 @@ text_sel="#ffffff" show_shaded="TRUE" shadetop="-100" - shadedown="0"> + shadedown="5"> </ThemeWidgetColors> </wcol_menu_item> <wcol_scroll> @@ -198,7 +198,7 @@ text_sel="#ffffff" show_shaded="TRUE" shadetop="-100" - shadedown="0"> + shadedown="5"> </ThemeWidgetColors> </wcol_scroll> <wcol_progress> @@ -210,7 +210,7 @@ text_sel="#ffffff" show_shaded="TRUE" shadetop="-100" - shadedown="0"> + shadedown="5"> </ThemeWidgetColors> </wcol_progress> <wcol_list_item> @@ -222,7 +222,7 @@ text_sel="#ffffff" show_shaded="TRUE" shadetop="-100" - shadedown="0"> + shadedown="5"> </ThemeWidgetColors> </wcol_list_item> <wcol_state> @@ -239,32 +239,35 @@ </user_interface> <view_3d> <ThemeView3D grid="#222222" + clipping_border_3d="#313131ff" wire="#888888" - wire_edit="#000000" + wire_edit="#6c75ff" gp_vertex="#000000" gp_vertex_select="#ff8500" gp_vertex_size="3" - lamp="#c1d40028" - speaker="#535353" - camera="#000000" - view_overlay="#000000" - empty="#000000" + text_grease_pencil="#b5e61d" object_selected="#f15800" object_active="#ff8c19" object_grouped="#083008" object_grouped_active="#55bb55" - transform="#ffffff" + text_keyframe="#ddd700" + camera="#535353" + empty="#535353" + lamp="#fff0d328" + speaker="#535353" vertex="#72cfdd" vertex_select="#ff8500" vertex_size="3" + vertex_bevel="#00a5ff" vertex_unreferenced="#000000" edge_select="#ffa000" edge_seam="#db2512" edge_sharp="#ff2020" edge_crease="#cc0099" + edge_bevel="#00a5ff" edge_facesel="#6b6b6b" freestyle_edge_mark="#7fff7f" - face="#73828f12" + face="#73828f41" face_select="#ffa4003c" face_dot="#ffa900" facedot_size="4" @@ -291,19 +294,18 @@ normal="#22dddd" vertex_normal="#2361dd" split_normal="#dd23dd" - bone_solid="#c8c8c8" bone_pose="#50c8ff" bone_pose_active="#8cffff" - frame_current="#60c040" - outline_width="1" + bone_solid="#c8c8c8" bundle_solid="#c8c8c8" camera_path="#5a5a5a" skin_root="#000000" - clipping_border_3d="#313131ff" - text_keyframe="#ddd700" - text_grease_pencil="#b5e61d" + view_overlay="#000000" + transform="#ffffff" + frame_current="#60c040" paint_curve_handle="#7fff7f7f" - paint_curve_pivot="#ff7f7f7f"> + paint_curve_pivot="#ff7f7f7f" + outline_width="1"> <space> <ThemeSpaceGradient title="#5d5d5d" text="#7d7d7d" @@ -312,23 +314,23 @@ header_text="#979797" header_text_hi="#ffffff" button="#00000057" - button_title="#c5c5c5" + button_title="#929292" button_text="#c3c3c3" - button_text_hi="#ffffff" + button_text_hi="#e5e5e5" tab_active="#212947" tab_inactive="#000000" tab_back="#060606ff" tab_outline="#000000"> <gradients> <ThemeGradientColors show_grad="TRUE" - gradient="#0a0a0a" + gradient="#1d1d1d" high_gradient="#000000"> </ThemeGradientColors> </gradients> <panelcolors> <ThemePanelColors header="#00000019" back="#72727280" - show_header="FALSE" + show_header="TRUE" show_back="FALSE"> </ThemePanelColors> </panelcolors> @@ -348,6 +350,7 @@ vertex="#ffffff" vertex_select="#ff8500" vertex_size="3" + vertex_bevel="#000000" vertex_unreferenced="#000000" handle_free="#808080" handle_auto="#909000" @@ -382,7 +385,7 @@ <panelcolors> <ThemePanelColors header="#00000019" back="#72727280" - show_header="FALSE" + show_header="TRUE" show_back="FALSE"> </ThemePanelColors> </panelcolors> @@ -418,7 +421,7 @@ <panelcolors> <ThemePanelColors header="#00000019" back="#72727280" - show_header="FALSE" + show_header="TRUE" show_back="FALSE"> </ThemePanelColors> </panelcolors> @@ -463,7 +466,7 @@ <panelcolors> <ThemePanelColors header="#00000019" back="#72727280" - show_header="FALSE" + show_header="TRUE" show_back="FALSE"> </ThemePanelColors> </panelcolors> @@ -501,6 +504,7 @@ keyframe_jitter_selected="#61c042" keyframe_border="#000000ff" keyframe_border_selected="#000000ff" + keyframe_scale_factor="1" summary="#00000000"> <space> <ThemeSpaceGeneric back="#080808" @@ -521,7 +525,7 @@ <panelcolors> <ThemePanelColors header="#00000019" back="#72727280" - show_header="FALSE" + show_header="TRUE" show_back="FALSE"> </ThemePanelColors> </panelcolors> @@ -543,6 +547,7 @@ vertex="#0f13bb" vertex_select="#ff8500" vertex_size="3" + vertex_bevel="#000000" vertex_unreferenced="#000000" face="#ffffff0a" face_select="#ff85003c" @@ -596,7 +601,7 @@ <panelcolors> <ThemePanelColors header="#00000019" back="#72727280" - show_header="FALSE" + show_header="TRUE" show_back="FALSE"> </ThemePanelColors> </panelcolors> @@ -644,7 +649,7 @@ <panelcolors> <ThemePanelColors header="#00000019" back="#72727280" - show_header="FALSE" + show_header="TRUE" show_back="FALSE"> </ThemePanelColors> </panelcolors> @@ -673,7 +678,7 @@ <panelcolors> <ThemePanelColors header="#00000019" back="#72727280" - show_header="FALSE" + show_header="TRUE" show_back="FALSE"> </ThemePanelColors> </panelcolors> @@ -712,7 +717,7 @@ <panelcolors> <ThemePanelColors header="#00000019" back="#72727280" - show_header="FALSE" + show_header="TRUE" show_back="FALSE"> </ThemePanelColors> </panelcolors> @@ -744,7 +749,7 @@ <panelcolors> <ThemePanelColors header="#00000019" back="#72727280" - show_header="FALSE" + show_header="TRUE" show_back="FALSE"> </ThemePanelColors> </panelcolors> @@ -799,7 +804,7 @@ <panelcolors> <ThemePanelColors header="#00000019" back="#72727280" - show_header="FALSE" + show_header="TRUE" show_back="FALSE"> </ThemePanelColors> </panelcolors> @@ -835,7 +840,7 @@ <panelcolors> <ThemePanelColors header="#00000019" back="#72727280" - show_header="FALSE" + show_header="TRUE" show_back="FALSE"> </ThemePanelColors> </panelcolors> @@ -865,7 +870,7 @@ <panelcolors> <ThemePanelColors header="#00000019" back="#72727280" - show_header="FALSE" + show_header="TRUE" show_back="FALSE"> </ThemePanelColors> </panelcolors> @@ -903,7 +908,7 @@ <panelcolors> <ThemePanelColors header="#00000019" back="#72727280" - show_header="FALSE" + show_header="TRUE" show_back="FALSE"> </ThemePanelColors> </panelcolors> @@ -932,7 +937,7 @@ <panelcolors> <ThemePanelColors header="#00000019" back="#72727280" - show_header="FALSE" + show_header="TRUE" show_back="FALSE"> </ThemePanelColors> </panelcolors> @@ -966,7 +971,7 @@ <panelcolors> <ThemePanelColors header="#00000019" back="#72727280" - show_header="FALSE" + show_header="TRUE" show_back="FALSE"> </ThemePanelColors> </panelcolors> @@ -1019,7 +1024,7 @@ <panelcolors> <ThemePanelColors header="#00000019" back="#72727280" - show_header="FALSE" + show_header="TRUE" show_back="FALSE"> </ThemePanelColors> </panelcolors> diff --git a/release/scripts/presets/keyconfig/3dsmax.py b/release/scripts/presets/keyconfig/3dsmax.py index 7694e338d68..6d05ff6982c 100644 --- a/release/scripts/presets/keyconfig/3dsmax.py +++ b/release/scripts/presets/keyconfig/3dsmax.py @@ -401,6 +401,12 @@ kmi = km.keymap_items.new('particle.hide', 'H', 'PRESS') kmi.properties.unselected = False kmi = km.keymap_items.new('particle.hide', 'H', 'PRESS', shift=True) kmi.properties.unselected = True +kmi = km.keymap_items.new('view3d.manipulator', 'LEFTMOUSE', 'PRESS', shift=True) +kmi.properties.release_confirm = True +kmi.properties.use_planar_constraint= True +kmi = km.keymap_items.new('view3d.manipulator', 'LEFTMOUSE', 'PRESS', shift=True) +kmi.properties.release_confirm = True +kmi.properties.use_accurate= True kmi = km.keymap_items.new('view3d.manipulator', 'LEFTMOUSE', 'PRESS', any=True) kmi.properties.release_confirm = True kmi = km.keymap_items.new('particle.brush_edit', 'LEFTMOUSE', 'PRESS') @@ -422,6 +428,12 @@ kmi.properties.value_2 = 'ENABLED' # Map 3D View km = kc.keymaps.new('3D View', space_type='VIEW_3D', region_type='WINDOW', modal=False) +kmi = km.keymap_items.new('view3d.manipulator', 'LEFTMOUSE', 'PRESS', shift=True) +kmi.properties.release_confirm = True +kmi.properties.use_planar_constraint= True +kmi = km.keymap_items.new('view3d.manipulator', 'LEFTMOUSE', 'PRESS', shift=True) +kmi.properties.release_confirm = True +kmi.properties.use_accurate= True kmi = km.keymap_items.new('view3d.manipulator', 'LEFTMOUSE', 'PRESS', any=True) kmi.properties.release_confirm = True kmi = km.keymap_items.new('view3d.cursor3d', 'ACTIONMOUSE', 'PRESS') diff --git a/release/scripts/presets/keyconfig/maya.py b/release/scripts/presets/keyconfig/maya.py index 67fd1fddcac..53129593a59 100644 --- a/release/scripts/presets/keyconfig/maya.py +++ b/release/scripts/presets/keyconfig/maya.py @@ -933,6 +933,12 @@ km = kc.keymaps.new('3D View', space_type='VIEW_3D', region_type='WINDOW', modal kmi = km.keymap_items.new('view3d.cursor3d', 'ACTIONMOUSE', 'PRESS') kmi = km.keymap_items.new('view3d.rotate', 'LEFTMOUSE', 'PRESS', alt=True) +kmi = km.keymap_items.new('view3d.manipulator', 'LEFTMOUSE', 'PRESS', shift=True) +kmi.properties.release_confirm = True +kmi.properties.use_planar_constraint = True +kmi = km.keymap_items.new('view3d.manipulator', 'LEFTMOUSE', 'PRESS', shift=True) +kmi.properties.release_confirm = True +kmi.properties.use_accurate = True kmi = km.keymap_items.new('view3d.manipulator', 'LEFTMOUSE', 'PRESS', any=True) kmi.properties.release_confirm = True kmi = km.keymap_items.new('view3d.move', 'MIDDLEMOUSE', 'PRESS', alt=True) diff --git a/release/scripts/startup/bl_operators/__init__.py b/release/scripts/startup/bl_operators/__init__.py index a696410ca1c..1e0dbe6925e 100644 --- a/release/scripts/startup/bl_operators/__init__.py +++ b/release/scripts/startup/bl_operators/__init__.py @@ -21,8 +21,7 @@ # support reloading sub-modules if "bpy" in locals(): from importlib import reload - for val in _modules_loaded: - reload(val) + _modules_loaded[:] = [reload(val) for val in _modules_loaded] del reload _modules = [ @@ -63,8 +62,15 @@ del _namespace def register(): - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for mod in _modules_loaded: + for cls in mod.classes: + register_class(cls) def unregister(): - bpy.utils.unregister_module(__name__) + from bpy.utils import unregister_class + for mod in reversed(_modules_loaded): + for cls in reversed(mod.classes): + if cls.is_registered: + unregister_class(cls) diff --git a/release/scripts/startup/bl_operators/add_mesh_torus.py b/release/scripts/startup/bl_operators/add_mesh_torus.py index 247b91e147f..0e5acea94f9 100644 --- a/release/scripts/startup/bl_operators/add_mesh_torus.py +++ b/release/scripts/startup/bl_operators/add_mesh_torus.py @@ -282,3 +282,8 @@ class AddTorus(Operator, object_utils.AddObjectHelper): object_utils.object_data_add(context, mesh, operator=self) return {'FINISHED'} + + +classes = ( + AddTorus, +)
\ No newline at end of file diff --git a/release/scripts/startup/bl_operators/anim.py b/release/scripts/startup/bl_operators/anim.py index c20d591241c..78fcf0dd124 100644 --- a/release/scripts/startup/bl_operators/anim.py +++ b/release/scripts/startup/bl_operators/anim.py @@ -28,11 +28,11 @@ if "bpy" in locals(): import bpy from bpy.types import Operator from bpy.props import ( - IntProperty, - BoolProperty, - EnumProperty, - StringProperty, - ) + IntProperty, + BoolProperty, + EnumProperty, + StringProperty, +) class ANIM_OT_keying_set_export(Operator): @@ -115,30 +115,30 @@ class ANIM_OT_keying_set_export(Operator): # Find material or lamp using this node tree... id_bpy_path = "bpy.data.nodes[\"%s\"]" found = False - + for mat in bpy.data.materials: if mat.node_tree == ksp.id: id_bpy_path = "bpy.data.materials[\"%s\"].node_tree" % (mat.name) found = True - break; - + break + if not found: for lamp in bpy.data.lamps: if lamp.node_tree == ksp.id: id_bpy_path = "bpy.data.lamps[\"%s\"].node_tree" % (lamp.name) found = True - break; - + break + if not found: - self.report({'WARN'}, "Could not find material or lamp using Shader Node Tree - %s" % (ksp.id)) + self.report({'WARN'}, "Could not find material or lamp using Shader Node Tree - %s" % (ksp.id)) elif ksp.id.bl_rna.identifier.startswith("CompositorNodeTree"): # Find compositor nodetree using this node tree... for scene in bpy.data.scenes: if scene.node_tree == ksp.id: id_bpy_path = "bpy.data.scenes[\"%s\"].node_tree" % (scene.name) - break; + break else: - self.report({'WARN'}, "Could not find scene using Compositor Node Tree - %s" % (ksp.id)) + self.report({'WARN'}, "Could not find scene using Compositor Node Tree - %s" % (ksp.id)) else: idtype_list = ksp.id.bl_rna.name.lower() + "s" id_bpy_path = "bpy.data.%s[\"%s\"]" % (idtype_list, ksp.id.name) @@ -302,9 +302,11 @@ class ClearUselessActions(Operator): bl_label = "Clear Useless Actions" bl_options = {'REGISTER', 'UNDO'} - only_unused = BoolProperty(name="Only Unused", + only_unused = BoolProperty( + name="Only Unused", description="Only unused (Fake User only) actions get considered", - default=True) + default=True, + ) @classmethod def poll(cls, context): @@ -393,7 +395,7 @@ class UpdateAnimatedTransformConstraint(Operator): except: pass ret = (data, new_path) - #print(ret) + # print(ret) return ret @@ -412,3 +414,11 @@ class UpdateAnimatedTransformConstraint(Operator): text.from_string(log) self.report({'INFO'}, "Complete report available on '%s' text datablock" % text.name) return {'FINISHED'} + + +classes = ( + ANIM_OT_keying_set_export, + BakeAction, + ClearUselessActions, + UpdateAnimatedTransformConstraint, +) diff --git a/release/scripts/startup/bl_operators/clip.py b/release/scripts/startup/bl_operators/clip.py index 0c77ea2ab7e..e52d577b900 100644 --- a/release/scripts/startup/bl_operators/clip.py +++ b/release/scripts/startup/bl_operators/clip.py @@ -21,7 +21,10 @@ import bpy import os from bpy.types import Operator from bpy.props import FloatProperty -from mathutils import Vector, Matrix +from mathutils import ( + Vector, + Matrix, +) def CLIP_spaces_walk(context, all_screens, tarea, tspace, callback, *args): @@ -1071,3 +1074,17 @@ class CLIP_OT_track_settings_to_track(bpy.types.Operator): setattr(marker_selected, attr, getattr(marker, attr)) return {'FINISHED'} + + +classes = ( + CLIP_OT_bundles_to_mesh, + CLIP_OT_constraint_to_fcurve, + CLIP_OT_delete_proxy, + CLIP_OT_filter_tracks, + CLIP_OT_set_active_clip, + CLIP_OT_set_viewport_background, + CLIP_OT_setup_tracking_scene, + CLIP_OT_track_settings_as_default, + CLIP_OT_track_settings_to_track, + CLIP_OT_track_to_empty, +) diff --git a/release/scripts/startup/bl_operators/console.py b/release/scripts/startup/bl_operators/console.py index 8cfc977294a..fb36f80239e 100644 --- a/release/scripts/startup/bl_operators/console.py +++ b/release/scripts/startup/bl_operators/console.py @@ -159,3 +159,12 @@ class ConsoleLanguage(Operator): remove_duplicates=True) return {'FINISHED'} + + +classes = ( + ConsoleAutocomplete, + ConsoleBanner, + ConsoleCopyAsScript, + ConsoleExec, + ConsoleLanguage, +)
\ No newline at end of file diff --git a/release/scripts/startup/bl_operators/file.py b/release/scripts/startup/bl_operators/file.py index 51e079164b6..d710b9af715 100644 --- a/release/scripts/startup/bl_operators/file.py +++ b/release/scripts/startup/bl_operators/file.py @@ -248,3 +248,8 @@ class WM_OT_previews_batch_clear(Operator): return {'FINISHED'} + +classes = ( + WM_OT_previews_batch_clear, + WM_OT_previews_batch_generate, +)
\ No newline at end of file diff --git a/release/scripts/startup/bl_operators/freestyle.py b/release/scripts/startup/bl_operators/freestyle.py index fe3b00d54cd..2e46160aeeb 100644 --- a/release/scripts/startup/bl_operators/freestyle.py +++ b/release/scripts/startup/bl_operators/freestyle.py @@ -16,13 +16,15 @@ # # ##### END GPL LICENSE BLOCK ##### +# <pep8 compliant> + import bpy from bpy.props import ( - BoolProperty, - EnumProperty, - StringProperty, - ) + BoolProperty, + EnumProperty, + StringProperty, +) class SCENE_OT_freestyle_fill_range_by_selection(bpy.types.Operator): @@ -218,3 +220,11 @@ class SCENE_OT_freestyle_module_open(bpy.types.Operator): text = bpy.data.texts.load(self.filepath, self.make_internal) self.freestyle_module.script = text return {'FINISHED'} + + +classes = ( + SCENE_OT_freestyle_add_edge_marks_to_keying_set, + SCENE_OT_freestyle_add_face_marks_to_keying_set, + SCENE_OT_freestyle_fill_range_by_selection, + SCENE_OT_freestyle_module_open, +) diff --git a/release/scripts/startup/bl_operators/image.py b/release/scripts/startup/bl_operators/image.py index d3460383fe7..6a538f0ae33 100644 --- a/release/scripts/startup/bl_operators/image.py +++ b/release/scripts/startup/bl_operators/image.py @@ -242,3 +242,11 @@ class ProjectApply(Operator): bpy.ops.paint.project_image(image=image_name) return {'FINISHED'} + + +classes = ( + EditExternally, + ProjectApply, + ProjectEdit, + SaveDirty, +)
\ No newline at end of file diff --git a/release/scripts/startup/bl_operators/mask.py b/release/scripts/startup/bl_operators/mask.py index aa984659430..78a4bd9af27 100644 --- a/release/scripts/startup/bl_operators/mask.py +++ b/release/scripts/startup/bl_operators/mask.py @@ -31,3 +31,8 @@ class MASK_MT_add(Menu): layout.operator_context = 'INVOKE_REGION_WIN' layout.operator("mask.primitive_circle_add", text="Circle", icon='MESH_CIRCLE') layout.operator("mask.primitive_square_add", text="Square", icon='MESH_PLANE') + + +classes = ( + MASK_MT_add, +)
\ No newline at end of file diff --git a/release/scripts/startup/bl_operators/mesh.py b/release/scripts/startup/bl_operators/mesh.py index 58eab5436e6..4edefd7bf9b 100644 --- a/release/scripts/startup/bl_operators/mesh.py +++ b/release/scripts/startup/bl_operators/mesh.py @@ -21,7 +21,10 @@ import bpy from bpy.types import Operator -from bpy.props import EnumProperty, IntProperty +from bpy.props import ( + EnumProperty, + IntProperty, +) class MeshMirrorUV(Operator): @@ -248,3 +251,10 @@ class MehsSetNormalsFromFaces(Operator): return {'FINISHED'} + +classes = ( + MehsSetNormalsFromFaces, + MeshMirrorUV, + MeshSelectNext, + MeshSelectPrev, +) diff --git a/release/scripts/startup/bl_operators/node.py b/release/scripts/startup/bl_operators/node.py index acff259e503..40876e2b069 100644 --- a/release/scripts/startup/bl_operators/node.py +++ b/release/scripts/startup/bl_operators/node.py @@ -21,16 +21,16 @@ import bpy import nodeitems_utils from bpy.types import ( - Operator, - PropertyGroup, - ) + Operator, + PropertyGroup, +) from bpy.props import ( - BoolProperty, - CollectionProperty, - EnumProperty, - IntProperty, - StringProperty, - ) + BoolProperty, + CollectionProperty, + EnumProperty, + IntProperty, + StringProperty, +) class NodeSetting(PropertyGroup): @@ -295,3 +295,14 @@ class NODE_OT_tree_path_parent(Operator): space.path.pop() return {'FINISHED'} + + +classes = ( + NodeSetting, + + NODE_OT_add_and_link_node, + NODE_OT_add_node, + NODE_OT_add_search, + NODE_OT_collapse_hide_unused_toggle, + NODE_OT_tree_path_parent, +) diff --git a/release/scripts/startup/bl_operators/object.py b/release/scripts/startup/bl_operators/object.py index 6356da216b1..3a42d8d2e78 100644 --- a/release/scripts/startup/bl_operators/object.py +++ b/release/scripts/startup/bl_operators/object.py @@ -21,12 +21,12 @@ import bpy from bpy.types import Operator from bpy.props import ( - StringProperty, - BoolProperty, - EnumProperty, - IntProperty, - FloatProperty, - ) + BoolProperty, + EnumProperty, + FloatProperty, + IntProperty, + StringProperty, +) class SelectPattern(Operator): @@ -1034,3 +1034,22 @@ class LodGenerate(Operator): scene.objects.active = ob return {'FINISHED'} + + +classes = ( + ClearAllRestrictRender, + DupliOffsetFromCursor, + IsolateTypeRender, + JoinUVs, + LodByName, + LodClearAll, + LodGenerate, + MakeDupliFace, + SelectCamera, + SelectHierarchy, + SelectPattern, + ShapeTransfer, + SubdivisionSet, + TransformsToDeltas, + TransformsToDeltasAnim, +) diff --git a/release/scripts/startup/bl_operators/object_align.py b/release/scripts/startup/bl_operators/object_align.py index a6ee16e6b71..1539ffb3545 100644 --- a/release/scripts/startup/bl_operators/object_align.py +++ b/release/scripts/startup/bl_operators/object_align.py @@ -26,13 +26,14 @@ from mathutils import Vector def GlobalBB_LQ(bb_world): # Initialize the variables with the 8th vertex - left, right, front, back, down, up = (bb_world[7][0], - bb_world[7][0], - bb_world[7][1], - bb_world[7][1], - bb_world[7][2], - bb_world[7][2], - ) + left, right, front, back, down, up = ( + bb_world[7][0], + bb_world[7][0], + bb_world[7][1], + bb_world[7][1], + bb_world[7][2], + bb_world[7][2], + ) # Test against the other 7 verts for i in range(7): @@ -398,16 +399,23 @@ class AlignObjects(Operator): def execute(self, context): align_axis = self.align_axis - ret = align_objects(context, - 'X' in align_axis, - 'Y' in align_axis, - 'Z' in align_axis, - self.align_mode, - self.relative_to, - self.bb_quality) + ret = align_objects( + context, + 'X' in align_axis, + 'Y' in align_axis, + 'Z' in align_axis, + self.align_mode, + self.relative_to, + self.bb_quality, + ) if not ret: self.report({'WARNING'}, "No objects with bound-box selected") return {'CANCELLED'} else: return {'FINISHED'} + + +classes = ( + AlignObjects, +) diff --git a/release/scripts/startup/bl_operators/object_quick_effects.py b/release/scripts/startup/bl_operators/object_quick_effects.py index ef10e279bb4..16f29c77bb9 100644 --- a/release/scripts/startup/bl_operators/object_quick_effects.py +++ b/release/scripts/startup/bl_operators/object_quick_effects.py @@ -22,12 +22,12 @@ from mathutils import Vector import bpy from bpy.types import Operator from bpy.props import ( - BoolProperty, - EnumProperty, - IntProperty, - FloatProperty, - FloatVectorProperty, - ) + BoolProperty, + EnumProperty, + IntProperty, + FloatProperty, + FloatVectorProperty, +) def object_ensure_material(obj, mat_name): @@ -319,7 +319,7 @@ class QuickSmoke(Operator): def execute(self, context): if not bpy.app.build_options.mod_smoke: - self.report({'ERROR'}, "Build without Smoke modifier support") + self.report({'ERROR'}, "Built without Smoke modifier support") return {'CANCELLED'} fake_context = context.copy() @@ -568,7 +568,7 @@ class QuickFluid(Operator): def execute(self, context): if not bpy.app.build_options.mod_fluid: - self.report({'ERROR'}, "Build without Fluid modifier support") + self.report({'ERROR'}, "Built without Fluid modifier support") return {'CANCELLED'} fake_context = context.copy() @@ -645,3 +645,11 @@ class QuickFluid(Operator): bpy.ops.fluid.bake('INVOKE_DEFAULT') return {'FINISHED'} + + +classes = ( + QuickExplode, + QuickFluid, + QuickFur, + QuickSmoke, +) diff --git a/release/scripts/startup/bl_operators/object_randomize_transform.py b/release/scripts/startup/bl_operators/object_randomize_transform.py index 38110328603..f856b85844e 100644 --- a/release/scripts/startup/bl_operators/object_randomize_transform.py +++ b/release/scripts/startup/bl_operators/object_randomize_transform.py @@ -185,3 +185,8 @@ class RandomizeLocRotSize(Operator): loc, rot, scale, scale_even, scale_min) return {'FINISHED'} + + +classes = ( + RandomizeLocRotSize, +)
\ No newline at end of file diff --git a/release/scripts/startup/bl_operators/presets.py b/release/scripts/startup/bl_operators/presets.py index e01e509b292..fdacf24f6e0 100644 --- a/release/scripts/startup/bl_operators/presets.py +++ b/release/scripts/startup/bl_operators/presets.py @@ -135,7 +135,7 @@ class AddPresetBase: file_preset.write("%s = %r\n" % (rna_path_step, value)) - file_preset = open(filepath, 'w') + file_preset = open(filepath, 'w', encoding="utf-8") file_preset.write("import bpy\n") if hasattr(self, "preset_defines"): @@ -680,3 +680,26 @@ class AddPresetUnitsLength(AddPresetBase, Operator): ] preset_subdir = "units_length" + + +classes = ( + AddPresetCamera, + AddPresetCloth, + AddPresetFluid, + AddPresetHairDynamics, + AddPresetInteraction, + AddPresetInterfaceTheme, + AddPresetKeyconfig, + AddPresetNodeColor, + AddPresetOperator, + AddPresetRender, + AddPresetSSS, + AddPresetSafeAreas, + AddPresetSunSky, + AddPresetTrackingCamera, + AddPresetTrackingSettings, + AddPresetTrackingTrackColor, + AddPresetUnitsLength, + ExecutePreset, + WM_MT_operator_presets, +)
\ No newline at end of file diff --git a/release/scripts/startup/bl_operators/rigidbody.py b/release/scripts/startup/bl_operators/rigidbody.py index 750a5b0bf0f..6792d525683 100644 --- a/release/scripts/startup/bl_operators/rigidbody.py +++ b/release/scripts/startup/bl_operators/rigidbody.py @@ -20,8 +20,10 @@ import bpy from bpy.types import Operator -from bpy.props import IntProperty -from bpy.props import EnumProperty +from bpy.props import ( + EnumProperty, + IntProperty, +) class CopyRigidbodySettings(Operator): @@ -309,3 +311,10 @@ class ConnectRigidBodies(Operator): else: self.report({'WARNING'}, "No other objects selected") return {'CANCELLED'} + + +classes = ( + BakeToKeyframes, + ConnectRigidBodies, + CopyRigidbodySettings, +) diff --git a/release/scripts/startup/bl_operators/screen_play_rendered_anim.py b/release/scripts/startup/bl_operators/screen_play_rendered_anim.py index a5565699364..f4d6c7065a9 100644 --- a/release/scripts/startup/bl_operators/screen_play_rendered_anim.py +++ b/release/scripts/startup/bl_operators/screen_play_rendered_anim.py @@ -180,3 +180,8 @@ class PlayRenderedAnim(Operator): return {'CANCELLED'} return {'FINISHED'} + + +classes = ( + PlayRenderedAnim, +)
\ No newline at end of file diff --git a/release/scripts/startup/bl_operators/sequencer.py b/release/scripts/startup/bl_operators/sequencer.py index 31ca4249a9d..7209b6b478f 100644 --- a/release/scripts/startup/bl_operators/sequencer.py +++ b/release/scripts/startup/bl_operators/sequencer.py @@ -134,3 +134,10 @@ class SequencerDeinterlaceSelectedMovies(Operator): s.use_deinterlace = True return {'FINISHED'} + + +classes = ( + SequencerCrossfadeSounds, + SequencerCutMulticam, + SequencerDeinterlaceSelectedMovies, +)
\ No newline at end of file diff --git a/release/scripts/startup/bl_operators/uvcalc_follow_active.py b/release/scripts/startup/bl_operators/uvcalc_follow_active.py index d1ac9e0b586..25ee5cafe81 100644 --- a/release/scripts/startup/bl_operators/uvcalc_follow_active.py +++ b/release/scripts/startup/bl_operators/uvcalc_follow_active.py @@ -248,3 +248,8 @@ class FollowActiveQuads(Operator): def invoke(self, context, event): wm = context.window_manager return wm.invoke_props_dialog(self) + + +classes = ( + FollowActiveQuads, +)
\ No newline at end of file diff --git a/release/scripts/startup/bl_operators/uvcalc_lightmap.py b/release/scripts/startup/bl_operators/uvcalc_lightmap.py index 3b095c883a3..8ee29d15d1b 100644 --- a/release/scripts/startup/bl_operators/uvcalc_lightmap.py +++ b/release/scripts/startup/bl_operators/uvcalc_lightmap.py @@ -668,3 +668,8 @@ class LightMapPack(Operator): def invoke(self, context, event): wm = context.window_manager return wm.invoke_props_dialog(self) + + +classes = ( + LightMapPack, +)
\ No newline at end of file diff --git a/release/scripts/startup/bl_operators/uvcalc_smart_project.py b/release/scripts/startup/bl_operators/uvcalc_smart_project.py index 52e7b0e0ae4..5581415c083 100644 --- a/release/scripts/startup/bl_operators/uvcalc_smart_project.py +++ b/release/scripts/startup/bl_operators/uvcalc_smart_project.py @@ -18,7 +18,11 @@ # TODO <pep8 compliant> -from mathutils import Matrix, Vector, geometry +from mathutils import ( + Matrix, + Vector, + geometry, +) import bpy from bpy.types import Operator @@ -1100,3 +1104,8 @@ class SmartProject(Operator): def invoke(self, context, event): wm = context.window_manager return wm.invoke_props_dialog(self) + + +classes = ( + SmartProject, +) diff --git a/release/scripts/startup/bl_operators/vertexpaint_dirt.py b/release/scripts/startup/bl_operators/vertexpaint_dirt.py index 892e1822d68..c006e8e6e92 100644 --- a/release/scripts/startup/bl_operators/vertexpaint_dirt.py +++ b/release/scripts/startup/bl_operators/vertexpaint_dirt.py @@ -182,3 +182,8 @@ class VertexPaintDirt(Operator): ret = applyVertexDirt(mesh, self.blur_iterations, self.blur_strength, self.dirt_angle, self.clean_angle, self.dirt_only) return ret + + +classes = ( + VertexPaintDirt, +)
\ No newline at end of file diff --git a/release/scripts/startup/bl_operators/view3d.py b/release/scripts/startup/bl_operators/view3d.py index df4a93bb87f..acec2d8fe91 100644 --- a/release/scripts/startup/bl_operators/view3d.py +++ b/release/scripts/startup/bl_operators/view3d.py @@ -213,3 +213,11 @@ class VIEW3D_OT_select_or_deselect_all(Operator): enumerate=self.enumerate, object=self.object, location=(x, y)) + + +classes = ( + VIEW3D_OT_edit_mesh_extrude_individual_move, + VIEW3D_OT_edit_mesh_extrude_move, + VIEW3D_OT_edit_mesh_extrude_shrink_fatten, + VIEW3D_OT_select_or_deselect_all, +)
\ No newline at end of file diff --git a/release/scripts/startup/bl_operators/wm.py b/release/scripts/startup/bl_operators/wm.py index f5460d58d44..d94951341c1 100644 --- a/release/scripts/startup/bl_operators/wm.py +++ b/release/scripts/startup/bl_operators/wm.py @@ -21,12 +21,12 @@ import bpy from bpy.types import Operator from bpy.props import ( - StringProperty, - BoolProperty, - IntProperty, - FloatProperty, - EnumProperty, - ) + StringProperty, + BoolProperty, + IntProperty, + FloatProperty, + EnumProperty, +) from bpy.app.translations import pgettext_tip as tip_ @@ -130,6 +130,20 @@ def execute_context_assign(self, context): return operator_path_undo_return(context, data_path) +def module_filesystem_remove(path_base, module_name): + import os + module_name = os.path.splitext(module_name)[0] + for f in os.listdir(path_base): + f_base = os.path.splitext(f)[0] + if f_base == module_name: + f_full = os.path.join(path_base, f) + + if os.path.isdir(f_full): + os.rmdir(f_full) + else: + os.remove(f_full) + + class BRUSH_OT_active_index_set(Operator): """Set active sculpt/paint brush from it's number""" bl_idname = "brush.active_index_set" @@ -907,7 +921,10 @@ def _wm_doc_get_id(doc_id, do_url=True, url_prefix=""): # an operator (common case - just button referencing an op) if hasattr(bpy.types, class_name.upper() + "_OT_" + class_prop): if do_url: - url = ("%s/bpy.ops.%s.html#bpy.ops.%s.%s" % (url_prefix, class_name, class_name, class_prop)) + url = ( + "%s/bpy.ops.%s.html#bpy.ops.%s.%s" % + (url_prefix, class_name, class_name, class_prop) + ) else: rna = "bpy.ops.%s.%s" % (class_name, class_prop) else: @@ -922,7 +939,10 @@ def _wm_doc_get_id(doc_id, do_url=True, url_prefix=""): class_name, class_prop = class_name.split("_OT_", 1) class_name = class_name.lower() if do_url: - url = ("%s/bpy.ops.%s.html#bpy.ops.%s.%s" % (url_prefix, class_name, class_name, class_prop)) + url = ( + "%s/bpy.ops.%s.html#bpy.ops.%s.%s" % + (url_prefix, class_name, class_name, class_prop) + ) else: rna = "bpy.ops.%s.%s" % (class_name, class_prop) else: @@ -938,9 +958,12 @@ def _wm_doc_get_id(doc_id, do_url=True, url_prefix=""): rna_parent = rna_parent.base if do_url: - url = ("%s/bpy.types.%s.html#bpy.types.%s.%s" % (url_prefix, class_name, class_name, class_prop)) + url = ( + "%s/bpy.types.%s.html#bpy.types.%s.%s" % + (url_prefix, class_name, class_name, class_prop) + ) else: - rna = ("bpy.types.%s.%s" % (class_name, class_prop)) + rna = "bpy.types.%s.%s" % (class_name, class_prop) else: # We assume this is custom property, only try to generate generic url/rna_id... if do_url: @@ -1008,11 +1031,9 @@ class WM_OT_doc_view(Operator): doc_id = doc_id if bpy.app.version_cycle == "release": - _prefix = ("https://www.blender.org/api/blender_python_api_%s%s_release" % - ("_".join(str(v) for v in bpy.app.version[:2]), bpy.app.version_char)) + _prefix = ("https://docs.blender.org/api/blender_python_api_current") else: - _prefix = ("https://www.blender.org/api/blender_python_api_%s" % - "_".join(str(v) for v in bpy.app.version)) + _prefix = ("https://docs.blender.org/api/blender_python_api_master") def execute(self, context): url = _wm_doc_get_id(self.doc_id, do_url=True, url_prefix=self._prefix) @@ -1089,10 +1110,10 @@ class WM_OT_properties_edit(Operator): def execute(self, context): from rna_prop_ui import ( - rna_idprop_ui_prop_get, - rna_idprop_ui_prop_clear, - rna_idprop_ui_prop_update, - ) + rna_idprop_ui_prop_get, + rna_idprop_ui_prop_clear, + rna_idprop_ui_prop_update, + ) data_path = self.data_path value = self.value @@ -1269,9 +1290,9 @@ class WM_OT_properties_add(Operator): def execute(self, context): from rna_prop_ui import ( - rna_idprop_ui_prop_get, - rna_idprop_ui_prop_update, - ) + rna_idprop_ui_prop_get, + rna_idprop_ui_prop_update, + ) data_path = self.data_path item = eval("context.%s" % data_path) @@ -1286,10 +1307,10 @@ class WM_OT_properties_add(Operator): return prop_new - prop = unique_name( - {*item.keys(), - *type(item).bl_rna.properties.keys(), - }) + prop = unique_name({ + *item.keys(), + *type(item).bl_rna.properties.keys(), + }) item[prop] = 1.0 rna_idprop_ui_prop_update(item, prop) @@ -1329,9 +1350,9 @@ class WM_OT_properties_remove(Operator): def execute(self, context): from rna_prop_ui import ( - rna_idprop_ui_prop_clear, - rna_idprop_ui_prop_update, - ) + rna_idprop_ui_prop_clear, + rna_idprop_ui_prop_update, + ) data_path = self.data_path item = eval("context.%s" % data_path) prop = self.property @@ -1369,7 +1390,10 @@ class WM_OT_appconfig_default(Operator): filepath = os.path.join(bpy.utils.preset_paths("interaction")[0], "blender.py") if os.path.exists(filepath): - bpy.ops.script.execute_preset(filepath=filepath, menu_idname="USERPREF_MT_interaction_presets") + bpy.ops.script.execute_preset( + filepath=filepath, + menu_idname="USERPREF_MT_interaction_presets", + ) return {'FINISHED'} @@ -1389,7 +1413,10 @@ class WM_OT_appconfig_activate(Operator): filepath = self.filepath.replace("keyconfig", "interaction") if os.path.exists(filepath): - bpy.ops.script.execute_preset(filepath=filepath, menu_idname="USERPREF_MT_interaction_presets") + bpy.ops.script.execute_preset( + filepath=filepath, + menu_idname="USERPREF_MT_interaction_presets", + ) return {'FINISHED'} @@ -1494,7 +1521,7 @@ class WM_OT_blenderplayer_start(Operator): "-g", "show_profile", "=", "%d" % gs.show_framerate_profile, "-g", "show_properties", "=", "%d" % gs.show_debug_properties, "-g", "ignore_deprecation_warnings", "=", "%d" % (not gs.use_deprecation_warnings), - ]) + ]) # finish the call with the path to the blend file args.append(filepath) @@ -1624,10 +1651,11 @@ class WM_OT_keyconfig_export(Operator): wm = context.window_manager - keyconfig_utils.keyconfig_export(wm, - wm.keyconfigs.active, - self.filepath, - ) + keyconfig_utils.keyconfig_export( + wm, + wm.keyconfigs.active, + self.filepath, + ) return {'FINISHED'} @@ -1892,7 +1920,10 @@ class WM_OT_theme_install(Operator): try: shutil.copyfile(xmlfile, path_dest) - bpy.ops.script.execute_preset(filepath=path_dest, menu_idname="USERPREF_MT_interface_theme_presets") + bpy.ops.script.execute_preset( + filepath=path_dest, + menu_idname="USERPREF_MT_interface_theme_presets", + ) except: traceback.print_exc() @@ -1919,10 +1950,12 @@ class WM_OT_addon_refresh(Operator): return {'FINISHED'} +# Note: shares some logic with WM_OT_app_template_install +# but not enough to de-duplicate. Fixed here may apply to both. class WM_OT_addon_install(Operator): "Install an add-on" bl_idname = "wm.addon_install" - bl_label = "Install from File..." + bl_label = "Install Add-on from File..." overwrite = BoolProperty( name="Overwrite", @@ -1953,20 +1986,6 @@ class WM_OT_addon_install(Operator): options={'HIDDEN'}, ) - @staticmethod - def _module_remove(path_addons, module): - import os - module = os.path.splitext(module)[0] - for f in os.listdir(path_addons): - f_base = os.path.splitext(f)[0] - if f_base == module: - f_full = os.path.join(path_addons, f) - - if os.path.isdir(f_full): - os.rmdir(f_full) - else: - os.remove(f_full) - def execute(self, context): import addon_utils import traceback @@ -2019,7 +2038,7 @@ class WM_OT_addon_install(Operator): if self.overwrite: for f in file_to_extract.namelist(): - WM_OT_addon_install._module_remove(path_addons, f) + module_filesystem_remove(path_addons, f) else: for f in file_to_extract.namelist(): path_dest = os.path.join(path_addons, os.path.basename(f)) @@ -2037,7 +2056,7 @@ class WM_OT_addon_install(Operator): path_dest = os.path.join(path_addons, os.path.basename(pyfile)) if self.overwrite: - WM_OT_addon_install._module_remove(path_addons, os.path.basename(pyfile)) + module_filesystem_remove(path_addons, os.path.basename(pyfile)) elif os.path.exists(path_dest): self.report({'WARNING'}, "File already installed to %r\n" % path_dest) return {'CANCELLED'} @@ -2072,7 +2091,10 @@ class WM_OT_addon_install(Operator): bpy.utils.refresh_script_paths() # print message - msg = tip_("Modules Installed from %r into %r (%s)") % (pyfile, path_addons, ", ".join(sorted(addons_new))) + msg = ( + tip_("Modules Installed (%s) from %r into %r") % + (", ".join(sorted(addons_new)), pyfile, path_addons) + ) print(msg) self.report({'INFO'}, msg) @@ -2166,6 +2188,7 @@ class WM_OT_addon_expand(Operator): return {'FINISHED'} + class WM_OT_addon_userpref_show(Operator): "Show add-on user preferences" bl_idname = "wm.addon_userpref_show" @@ -2194,3 +2217,160 @@ class WM_OT_addon_userpref_show(Operator): bpy.ops.screen.userpref_show('INVOKE_DEFAULT') return {'FINISHED'} + + +# Note: shares some logic with WM_OT_addon_install +# but not enough to de-duplicate. Fixes here may apply to both. +class WM_OT_app_template_install(Operator): + "Install an application-template" + bl_idname = "wm.app_template_install" + bl_label = "Install Template from File..." + + overwrite = BoolProperty( + name="Overwrite", + description="Remove existing template with the same ID", + default=True, + ) + + filepath = StringProperty( + subtype='FILE_PATH', + ) + filter_folder = BoolProperty( + name="Filter folders", + default=True, + options={'HIDDEN'}, + ) + filter_glob = StringProperty( + default="*.zip", + options={'HIDDEN'}, + ) + + def execute(self, context): + import traceback + import zipfile + import shutil + import os + + filepath = self.filepath + + path_app_templates = bpy.utils.user_resource( + 'SCRIPTS', os.path.join("startup", "bl_app_templates_user"), + create=True, + ) + + if not path_app_templates: + self.report({'ERROR'}, "Failed to get add-ons path") + return {'CANCELLED'} + + if not os.path.isdir(path_app_templates): + try: + os.makedirs(path_app_templates, exist_ok=True) + except: + traceback.print_exc() + + app_templates_old = set(os.listdir(path_app_templates)) + + # check to see if the file is in compressed format (.zip) + if zipfile.is_zipfile(filepath): + try: + file_to_extract = zipfile.ZipFile(filepath, 'r') + except: + traceback.print_exc() + return {'CANCELLED'} + + if self.overwrite: + for f in file_to_extract.namelist(): + module_filesystem_remove(path_app_templates, f) + else: + for f in file_to_extract.namelist(): + path_dest = os.path.join(path_app_templates, os.path.basename(f)) + if os.path.exists(path_dest): + self.report({'WARNING'}, "File already installed to %r\n" % path_dest) + return {'CANCELLED'} + + try: # extract the file to "bl_app_templates_user" + file_to_extract.extractall(path_app_templates) + except: + traceback.print_exc() + return {'CANCELLED'} + + else: + # Only support installing zipfiles + self.report({'WARNING'}, "Expected a zip-file %r\n" % filepath) + return {'CANCELLED'} + + app_templates_new = set(os.listdir(path_app_templates)) - app_templates_old + + # in case a new module path was created to install this addon. + bpy.utils.refresh_script_paths() + + # print message + msg = ( + tip_("Template Installed (%s) from %r into %r") % + (", ".join(sorted(app_templates_new)), filepath, path_app_templates) + ) + print(msg) + self.report({'INFO'}, msg) + + return {'FINISHED'} + + def invoke(self, context, event): + wm = context.window_manager + wm.fileselect_add(self) + return {'RUNNING_MODAL'} + + +classes = ( + BRUSH_OT_active_index_set, + WM_OT_addon_disable, + WM_OT_addon_enable, + WM_OT_addon_expand, + WM_OT_addon_install, + WM_OT_addon_refresh, + WM_OT_addon_remove, + WM_OT_addon_userpref_show, + WM_OT_app_template_install, + WM_OT_appconfig_activate, + WM_OT_appconfig_default, + WM_OT_blenderplayer_start, + WM_OT_context_collection_boolean_set, + WM_OT_context_cycle_array, + WM_OT_context_cycle_enum, + WM_OT_context_cycle_int, + WM_OT_context_menu_enum, + WM_OT_context_modal_mouse, + WM_OT_context_pie_enum, + WM_OT_context_scale_float, + WM_OT_context_scale_int, + WM_OT_context_set_boolean, + WM_OT_context_set_enum, + WM_OT_context_set_float, + WM_OT_context_set_id, + WM_OT_context_set_int, + WM_OT_context_set_string, + WM_OT_context_set_value, + WM_OT_context_toggle, + WM_OT_context_toggle_enum, + WM_OT_copy_prev_settings, + WM_OT_doc_view, + WM_OT_doc_view_manual, + WM_OT_keyconfig_activate, + WM_OT_keyconfig_export, + WM_OT_keyconfig_import, + WM_OT_keyconfig_remove, + WM_OT_keyconfig_test, + WM_OT_keyitem_add, + WM_OT_keyitem_remove, + WM_OT_keyitem_restore, + WM_OT_keymap_restore, + WM_OT_operator_cheat_sheet, + WM_OT_operator_pie_enum, + WM_OT_path_open, + WM_OT_properties_add, + WM_OT_properties_context_change, + WM_OT_properties_edit, + WM_OT_properties_remove, + WM_OT_sysinfo, + WM_OT_theme_install, + WM_OT_url_open, +) diff --git a/release/scripts/startup/bl_ui/__init__.py b/release/scripts/startup/bl_ui/__init__.py index 2389be6787d..5b609605cee 100644 --- a/release/scripts/startup/bl_ui/__init__.py +++ b/release/scripts/startup/bl_ui/__init__.py @@ -23,8 +23,7 @@ # support reloading sub-modules if "bpy" in locals(): from importlib import reload - for val in _modules_loaded: - reload(val) + _modules_loaded[:] = [reload(val) for val in _modules_loaded] del reload _modules = [ @@ -94,7 +93,10 @@ del _namespace def register(): - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for mod in _modules_loaded: + for cls in mod.classes: + register_class(cls) # space_userprefs.py from bpy.props import StringProperty, EnumProperty @@ -143,8 +145,11 @@ def register(): def unregister(): - bpy.utils.unregister_module(__name__) - + from bpy.utils import unregister_class + for mod in reversed(_modules_loaded): + for cls in reversed(mod.classes): + if cls.is_registered: + unregister_class(cls) # Define a default UIList, when a list does not need any custom drawing... # Keep in sync with its #defined name in UI_interface.h diff --git a/release/scripts/startup/bl_ui/properties_animviz.py b/release/scripts/startup/bl_ui/properties_animviz.py index 046b5eb2aa5..84bae18dd6f 100644 --- a/release/scripts/startup/bl_ui/properties_animviz.py +++ b/release/scripts/startup/bl_ui/properties_animviz.py @@ -139,6 +139,11 @@ class OnionSkinButtonsPanel: col.label(text="Display:") col.prop(arm, "show_only_ghost_selected", text="Selected Only") + +classes = ( +) + if __name__ == "__main__": # only for live edit. - import bpy - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/properties_constraint.py b/release/scripts/startup/bl_ui/properties_constraint.py index cb5f1595ff3..00892d5f85b 100644 --- a/release/scripts/startup/bl_ui/properties_constraint.py +++ b/release/scripts/startup/bl_ui/properties_constraint.py @@ -941,5 +941,12 @@ class BONE_PT_constraints(ConstraintButtonsPanel, Panel): for con in context.pose_bone.constraints: self.draw_constraint(context, con) +classes = ( + OBJECT_PT_constraints, + BONE_PT_constraints, +) + if __name__ == "__main__": # only for live edit. - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/properties_data_armature.py b/release/scripts/startup/bl_ui/properties_data_armature.py index 8261f0c7770..a2ecf984eb5 100644 --- a/release/scripts/startup/bl_ui/properties_data_armature.py +++ b/release/scripts/startup/bl_ui/properties_data_armature.py @@ -328,5 +328,21 @@ class DATA_PT_custom_props_arm(ArmatureButtonsPanel, PropertyPanel, Panel): _context_path = "object.data" _property_type = bpy.types.Armature + +classes = ( + DATA_PT_context_arm, + DATA_PT_skeleton, + DATA_PT_display, + DATA_PT_bone_group_specials, + DATA_PT_bone_groups, + DATA_PT_pose_library, + DATA_PT_ghost, + DATA_PT_iksolver_itasc, + DATA_PT_motion_paths, + DATA_PT_custom_props_arm, +) + if __name__ == "__main__": # only for live edit. - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/properties_data_bone.py b/release/scripts/startup/bl_ui/properties_data_bone.py index 3a5475514c9..132c355ed99 100644 --- a/release/scripts/startup/bl_ui/properties_data_bone.py +++ b/release/scripts/startup/bl_ui/properties_data_bone.py @@ -449,5 +449,20 @@ class BONE_PT_custom_props(BoneButtonsPanel, PropertyPanel, Panel): else: return "active_bone" + +classes = ( + BONE_PT_context_bone, + BONE_PT_transform, + BONE_PT_transform_locks, + BONE_PT_curved, + BONE_PT_relations, + BONE_PT_display, + BONE_PT_inverse_kinematics, + BONE_PT_deform, + BONE_PT_custom_props, +) + if __name__ == "__main__": # only for live edit. - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/properties_data_camera.py b/release/scripts/startup/bl_ui/properties_data_camera.py index c6fa77aec00..101062095c4 100644 --- a/release/scripts/startup/bl_ui/properties_data_camera.py +++ b/release/scripts/startup/bl_ui/properties_data_camera.py @@ -325,5 +325,20 @@ def draw_display_safe_settings(layout, safe_data, settings): col.prop(safe_data, "action_center", slider=True) +classes = ( + CAMERA_MT_presets, + SAFE_AREAS_MT_presets, + DATA_PT_context_camera, + DATA_PT_lens, + DATA_PT_camera, + DATA_PT_camera_stereoscopy, + DATA_PT_camera_dof, + DATA_PT_camera_display, + DATA_PT_camera_safe_areas, + DATA_PT_custom_props_camera, +) + if __name__ == "__main__": # only for live edit. - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/properties_data_curve.py b/release/scripts/startup/bl_ui/properties_data_curve.py index 33af812b6c4..ac1eb9505a6 100644 --- a/release/scripts/startup/bl_ui/properties_data_curve.py +++ b/release/scripts/startup/bl_ui/properties_data_curve.py @@ -431,5 +431,21 @@ class DATA_PT_custom_props_curve(CurveButtonsPanel, PropertyPanel, Panel): _context_path = "object.data" _property_type = bpy.types.Curve + +classes = ( + DATA_PT_context_curve, + DATA_PT_shape_curve, + DATA_PT_curve_texture_space, + DATA_PT_geometry_curve, + DATA_PT_pathanim, + DATA_PT_active_spline, + DATA_PT_font, + DATA_PT_paragraph, + DATA_PT_text_boxes, + DATA_PT_custom_props_curve, +) + if __name__ == "__main__": # only for live edit. - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/properties_data_empty.py b/release/scripts/startup/bl_ui/properties_data_empty.py index a752d3517de..c66ff87ecbd 100644 --- a/release/scripts/startup/bl_ui/properties_data_empty.py +++ b/release/scripts/startup/bl_ui/properties_data_empty.py @@ -55,5 +55,12 @@ class DATA_PT_empty(DataButtonsPanel, Panel): layout.prop(ob, "empty_draw_size", text="Size") + +classes = ( + DATA_PT_empty, +) + if __name__ == "__main__": # only for live edit. - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/properties_data_lamp.py b/release/scripts/startup/bl_ui/properties_data_lamp.py index 30cd39d2e2f..f913ef51381 100644 --- a/release/scripts/startup/bl_ui/properties_data_lamp.py +++ b/release/scripts/startup/bl_ui/properties_data_lamp.py @@ -394,5 +394,21 @@ class DATA_PT_custom_props_lamp(DataButtonsPanel, PropertyPanel, Panel): _context_path = "object.data" _property_type = bpy.types.Lamp + +classes = ( + LAMP_MT_sunsky_presets, + DATA_PT_context_lamp, + DATA_PT_preview, + DATA_PT_lamp, + DATA_PT_sunsky, + DATA_PT_shadow, + DATA_PT_area, + DATA_PT_spot, + DATA_PT_falloff_curve, + DATA_PT_custom_props_lamp, +) + if __name__ == "__main__": # only for live edit. - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/properties_data_lattice.py b/release/scripts/startup/bl_ui/properties_data_lattice.py index cbfa6961e8c..4b3fd48c195 100644 --- a/release/scripts/startup/bl_ui/properties_data_lattice.py +++ b/release/scripts/startup/bl_ui/properties_data_lattice.py @@ -82,5 +82,14 @@ class DATA_PT_custom_props_lattice(DataButtonsPanel, PropertyPanel, Panel): _context_path = "object.data" _property_type = bpy.types.Lattice + +classes = ( + DATA_PT_context_lattice, + DATA_PT_lattice, + DATA_PT_custom_props_lattice, +) + if __name__ == "__main__": # only for live edit. - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/properties_data_mesh.py b/release/scripts/startup/bl_ui/properties_data_mesh.py index 59907692fe0..de55b4152ba 100644 --- a/release/scripts/startup/bl_ui/properties_data_mesh.py +++ b/release/scripts/startup/bl_ui/properties_data_mesh.py @@ -392,5 +392,24 @@ class DATA_PT_custom_props_mesh(MeshButtonsPanel, PropertyPanel, Panel): _property_type = bpy.types.Mesh +classes = ( + MESH_MT_vertex_group_specials, + MESH_MT_shape_key_specials, + MESH_UL_vgroups, + MESH_UL_shape_keys, + MESH_UL_uvmaps_vcols, + DATA_PT_context_mesh, + DATA_PT_normals, + DATA_PT_texture_space, + DATA_PT_vertex_groups, + DATA_PT_shape_keys, + DATA_PT_uv_texture, + DATA_PT_vertex_colors, + DATA_PT_customdata, + DATA_PT_custom_props_mesh, +) + if __name__ == "__main__": # only for live edit. - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/properties_data_metaball.py b/release/scripts/startup/bl_ui/properties_data_metaball.py index b42258c386a..a621dc7210f 100644 --- a/release/scripts/startup/bl_ui/properties_data_metaball.py +++ b/release/scripts/startup/bl_ui/properties_data_metaball.py @@ -135,5 +135,16 @@ class DATA_PT_custom_props_metaball(DataButtonsPanel, PropertyPanel, Panel): _context_path = "object.data" _property_type = bpy.types.MetaBall + +classes = ( + DATA_PT_context_metaball, + DATA_PT_metaball, + DATA_PT_mball_texture_space, + DATA_PT_metaball_element, + DATA_PT_custom_props_metaball, +) + if __name__ == "__main__": # only for live edit. - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/properties_data_modifier.py b/release/scripts/startup/bl_ui/properties_data_modifier.py index d66fb08bcd6..a37d61dd0af 100644 --- a/release/scripts/startup/bl_ui/properties_data_modifier.py +++ b/release/scripts/startup/bl_ui/properties_data_modifier.py @@ -569,6 +569,14 @@ class DATA_PT_modifiers(ModifierButtonsPanel, Panel): col.prop(md, "use_mirror_u", text="U") col.prop(md, "use_mirror_v", text="V") + col = layout.column(align=True) + + if md.use_mirror_u: + col.prop(md, "mirror_offset_u") + + if md.use_mirror_v: + col.prop(md, "mirror_offset_v") + col = layout.column() if md.use_mirror_merge is True: @@ -951,6 +959,23 @@ class DATA_PT_modifiers(ModifierButtonsPanel, Panel): def SURFACE(self, layout, ob, md): layout.label(text="Settings are inside the Physics tab") + def SURFACE_DEFORM(self, layout, ob, md): + col = layout.column() + col.active = not md.is_bound + + col.prop(md, "target") + col.prop(md, "falloff") + + layout.separator() + + col = layout.column() + col.active = md.target is not None + + if md.is_bound: + col.operator("object.surfacedeform_bind", text="Unbind") + else: + col.operator("object.surfacedeform_bind", text="Bind") + def UV_PROJECT(self, layout, ob, md): split = layout.split() @@ -1320,7 +1345,9 @@ class DATA_PT_modifiers(ModifierButtonsPanel, Panel): row.prop(md, "thickness_vertex_group", text="Factor") col.prop(md, "use_crease", text="Crease Edges") - col.prop(md, "crease_weight", text="Crease Weight") + row = col.row() + row.active = md.use_crease + row.prop(md, "crease_weight", text="Crease Weight") col = split.column() @@ -1503,5 +1530,11 @@ class DATA_PT_modifiers(ModifierButtonsPanel, Panel): layout.operator("object.correctivesmooth_bind", text="Unbind" if is_bind else "Bind") +classes = ( + DATA_PT_modifiers, +) + if __name__ == "__main__": # only for live edit. - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/properties_data_speaker.py b/release/scripts/startup/bl_ui/properties_data_speaker.py index 5ed0f8c3265..eecb2690302 100644 --- a/release/scripts/startup/bl_ui/properties_data_speaker.py +++ b/release/scripts/startup/bl_ui/properties_data_speaker.py @@ -121,5 +121,16 @@ class DATA_PT_custom_props_speaker(DataButtonsPanel, PropertyPanel, Panel): _context_path = "object.data" _property_type = bpy.types.Speaker + +classes = ( + DATA_PT_context_speaker, + DATA_PT_speaker, + DATA_PT_distance, + DATA_PT_cone, + DATA_PT_custom_props_speaker, +) + if __name__ == "__main__": # only for live edit. - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/properties_freestyle.py b/release/scripts/startup/bl_ui/properties_freestyle.py index e0620447185..9c5be7624e0 100644 --- a/release/scripts/startup/bl_ui/properties_freestyle.py +++ b/release/scripts/startup/bl_ui/properties_freestyle.py @@ -828,5 +828,17 @@ class MATERIAL_PT_freestyle_line(MaterialFreestyleButtonsPanel, Panel): row.prop(mat, "line_priority", text="Priority") +classes = ( + RENDER_PT_freestyle, + RENDERLAYER_UL_linesets, + RENDER_MT_lineset_specials, + RENDERLAYER_PT_freestyle, + RENDERLAYER_PT_freestyle_lineset, + RENDERLAYER_PT_freestyle_linestyle, + MATERIAL_PT_freestyle_line, +) + if __name__ == "__main__": # only for live edit. - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/properties_game.py b/release/scripts/startup/bl_ui/properties_game.py index ec061370fe5..46b34373aa5 100644 --- a/release/scripts/startup/bl_ui/properties_game.py +++ b/release/scripts/startup/bl_ui/properties_game.py @@ -852,5 +852,30 @@ class OBJECT_PT_levels_of_detail(ObjectButtonsPanel, Panel): row.menu("OBJECT_MT_lod_tools", text="", icon='TRIA_DOWN') +classes = ( + PHYSICS_PT_game_physics, + PHYSICS_PT_game_collision_bounds, + PHYSICS_PT_game_obstacles, + RENDER_PT_embedded, + RENDER_PT_game_player, + RENDER_PT_game_stereo, + RENDER_PT_game_shading, + RENDER_PT_game_system, + RENDER_PT_game_display, + SCENE_PT_game_physics, + SCENE_PT_game_physics_obstacles, + SCENE_PT_game_navmesh, + SCENE_PT_game_hysteresis, + WORLD_PT_game_context_world, + WORLD_PT_game_world, + WORLD_PT_game_environment_lighting, + WORLD_PT_game_mist, + DATA_PT_shadow_game, + OBJECT_MT_lod_tools, + OBJECT_PT_levels_of_detail, +) + if __name__ == "__main__": # only for live edit. - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/properties_grease_pencil_common.py b/release/scripts/startup/bl_ui/properties_grease_pencil_common.py index 4529c127839..42e8d5272b3 100644 --- a/release/scripts/startup/bl_ui/properties_grease_pencil_common.py +++ b/release/scripts/startup/bl_ui/properties_grease_pencil_common.py @@ -1155,3 +1155,24 @@ class GreasePencilToolsPanel: layout.separator() gpencil_stroke_placement_settings(context, layout) + + +classes = ( + GPENCIL_PIE_tool_palette, + GPENCIL_PIE_settings_palette, + GPENCIL_PIE_tools_more, + GPENCIL_PIE_sculpt, + GPENCIL_MT_snap, + GPENCIL_MT_gpencil_edit_specials, + GPENCIL_UL_brush, + GPENCIL_UL_palettecolor, + GPENCIL_UL_layer, + GPENCIL_MT_layer_specials, + GPENCIL_MT_brush_specials, + GPENCIL_MT_palettecolor_specials, +) + +if __name__ == "__main__": # only for live edit. + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/properties_mask_common.py b/release/scripts/startup/bl_ui/properties_mask_common.py index f4836da50bc..a7a67130f2f 100644 --- a/release/scripts/startup/bl_ui/properties_mask_common.py +++ b/release/scripts/startup/bl_ui/properties_mask_common.py @@ -392,5 +392,17 @@ class MASK_MT_select(Menu): layout.operator("mask.select_all", text="Inverse").action = 'INVERT' layout.operator("mask.select_linked", text="Select Linked") + +classes = ( + MASK_UL_layers, + MASK_MT_mask, + MASK_MT_visibility, + MASK_MT_transform, + MASK_MT_animation, + MASK_MT_select, +) + if __name__ == "__main__": # only for live edit. - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/properties_material.py b/release/scripts/startup/bl_ui/properties_material.py index d916007ea41..06ae1847d06 100644 --- a/release/scripts/startup/bl_ui/properties_material.py +++ b/release/scripts/startup/bl_ui/properties_material.py @@ -71,6 +71,7 @@ class MATERIAL_MT_specials(Menu): class MATERIAL_UL_matslots(UIList): + def draw_item(self, context, layout, data, item, icon, active_data, active_propname, index): # assert(isinstance(item, bpy.types.MaterialSlot) # ob = data @@ -1052,5 +1053,38 @@ class MATERIAL_PT_custom_props(MaterialButtonsPanel, PropertyPanel, Panel): _context_path = "material" _property_type = bpy.types.Material + +classes = ( + MATERIAL_MT_sss_presets, + MATERIAL_MT_specials, + MATERIAL_UL_matslots, + MATERIAL_PT_context_material, + MATERIAL_PT_preview, + MATERIAL_PT_pipeline, + MATERIAL_PT_diffuse, + MATERIAL_PT_specular, + MATERIAL_PT_shading, + MATERIAL_PT_transp, + MATERIAL_PT_mirror, + MATERIAL_PT_sss, + MATERIAL_PT_halo, + MATERIAL_PT_flare, + MATERIAL_PT_game_settings, + MATERIAL_PT_physics, + MATERIAL_PT_strand, + MATERIAL_PT_options, + MATERIAL_PT_shadow, + MATERIAL_PT_transp_game, + MATERIAL_PT_volume_density, + MATERIAL_PT_volume_shading, + MATERIAL_PT_volume_lighting, + MATERIAL_PT_volume_transp, + MATERIAL_PT_volume_integration, + MATERIAL_PT_volume_options, + MATERIAL_PT_custom_props, +) + if __name__ == "__main__": # only for live edit. - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/properties_object.py b/release/scripts/startup/bl_ui/properties_object.py index 3ff7a248c60..d7e18f81232 100644 --- a/release/scripts/startup/bl_ui/properties_object.py +++ b/release/scripts/startup/bl_ui/properties_object.py @@ -152,6 +152,33 @@ class OBJECT_PT_relations(ObjectButtonsPanel, Panel): sub.active = (parent is not None) +class OBJECT_PT_relations_extras(ObjectButtonsPanel, Panel): + bl_label = "Relations Extras" + bl_options = {'DEFAULT_CLOSED'} + + def draw(self, context): + layout = self.layout + + ob = context.object + + split = layout.split() + + if context.scene.render.engine != 'BLENDER_GAME': + col = split.column() + col.label(text="Tracking Axes:") + col.prop(ob, "track_axis", text="Axis") + col.prop(ob, "up_axis", text="Up Axis") + + col = split.column() + col.prop(ob, "use_slow_parent") + row = col.row() + row.active = ((ob.parent is not None) and (ob.use_slow_parent)) + row.prop(ob, "slow_parent_offset", text="Offset") + + layout.prop(ob, "use_extra_recalc_object") + layout.prop(ob, "use_extra_recalc_data") + + class GROUP_MT_specials(Menu): bl_label = "Group Specials" @@ -296,33 +323,6 @@ class OBJECT_PT_duplication(ObjectButtonsPanel, Panel): layout.prop(ob, "dupli_group", text="Group") -class OBJECT_PT_relations_extras(ObjectButtonsPanel, Panel): - bl_label = "Relations Extras" - bl_options = {'DEFAULT_CLOSED'} - - def draw(self, context): - layout = self.layout - - ob = context.object - - split = layout.split() - - if context.scene.render.engine != 'BLENDER_GAME': - col = split.column() - col.label(text="Tracking Axes:") - col.prop(ob, "track_axis", text="Axis") - col.prop(ob, "up_axis", text="Up Axis") - - col = split.column() - col.prop(ob, "use_slow_parent") - row = col.row() - row.active = ((ob.parent is not None) and (ob.use_slow_parent)) - row.prop(ob, "slow_parent_offset", text="Offset") - - layout.prop(ob, "use_extra_recalc_object") - layout.prop(ob, "use_extra_recalc_data") - - from bl_ui.properties_animviz import ( MotionPathButtonsPanel, OnionSkinButtonsPanel, @@ -366,5 +366,23 @@ class OBJECT_PT_custom_props(ObjectButtonsPanel, PropertyPanel, Panel): _context_path = "object" _property_type = bpy.types.Object + +classes = ( + OBJECT_PT_context_object, + OBJECT_PT_transform, + OBJECT_PT_delta_transform, + OBJECT_PT_transform_locks, + OBJECT_PT_relations, + OBJECT_PT_relations_extras, + GROUP_MT_specials, + OBJECT_PT_groups, + OBJECT_PT_display, + OBJECT_PT_duplication, + OBJECT_PT_motion_paths, + OBJECT_PT_custom_props, +) + if __name__ == "__main__": # only for live edit. - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/properties_paint_common.py b/release/scripts/startup/bl_ui/properties_paint_common.py index 09a3a19cbce..dfd20d90ec4 100644 --- a/release/scripts/startup/bl_ui/properties_paint_common.py +++ b/release/scripts/startup/bl_ui/properties_paint_common.py @@ -309,3 +309,13 @@ def brush_mask_texture_settings(layout, brush): split = layout.split() split.prop(mask_tex_slot, "offset") split.prop(mask_tex_slot, "scale") + + +classes = ( + VIEW3D_MT_tools_projectpaint_clone, +) + +if __name__ == "__main__": # only for live edit. + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/properties_particle.py b/release/scripts/startup/bl_ui/properties_particle.py index 29fc56c3fad..2c27c68ed9d 100644 --- a/release/scripts/startup/bl_ui/properties_particle.py +++ b/release/scripts/startup/bl_ui/properties_particle.py @@ -23,11 +23,11 @@ from rna_prop_ui import PropertyPanel from bpy.app.translations import pgettext_iface as iface_ from bl_ui.properties_physics_common import ( - point_cache_ui, - effector_weights_ui, - basic_force_field_settings_ui, - basic_force_field_falloff_ui, - ) + point_cache_ui, + effector_weights_ui, + basic_force_field_settings_ui, + basic_force_field_falloff_ui, +) def particle_panel_enabled(context, psys): @@ -108,6 +108,7 @@ def find_modifier(ob, psys): class PARTICLE_UL_particle_systems(bpy.types.UIList): + def draw_item(self, context, layout, data, item, icon, active_data, active_propname, index, flt_flag): ob = data psys = item @@ -117,8 +118,10 @@ class PARTICLE_UL_particle_systems(bpy.types.UIList): layout.prop(psys, "name", text="", emboss=False, icon_value=icon) if md: - layout.prop(md, "show_render", emboss=False, icon_only=True, icon='RESTRICT_RENDER_OFF' if md.show_render else 'RESTRICT_RENDER_ON') - layout.prop(md, "show_viewport", emboss=False, icon_only=True, icon='RESTRICT_VIEW_OFF' if md.show_viewport else 'RESTRICT_VIEW_ON') + layout.prop(md, "show_render", emboss=False, icon_only=True, + icon='RESTRICT_RENDER_OFF' if md.show_render else 'RESTRICT_RENDER_ON') + layout.prop(md, "show_viewport", emboss=False, icon_only=True, + icon='RESTRICT_VIEW_OFF' if md.show_viewport else 'RESTRICT_VIEW_ON') elif self.layout_type == 'GRID': layout.alignment = 'CENTER' @@ -401,7 +404,8 @@ class PARTICLE_PT_hair_dynamics(ParticleButtonsPanel, Panel): label = "ERROR" icon = 'ERROR' box.label(label, icon=icon) - box.label("Iterations: %d .. %d (avg. %d)" % (result.min_iterations, result.max_iterations, result.avg_iterations)) + box.label("Iterations: %d .. %d (avg. %d)" % + (result.min_iterations, result.max_iterations, result.avg_iterations)) box.label("Error: %.5f .. %.5f (avg. %.5f)" % (result.min_error, result.max_error, result.avg_error)) @@ -423,7 +427,12 @@ class PARTICLE_PT_cache(ParticleButtonsPanel, Panel): phystype = psys.settings.physics_type if phystype == 'NO' or phystype == 'KEYED': return False - return (psys.settings.type in {'EMITTER', 'REACTOR'} or (psys.settings.type == 'HAIR' and (psys.use_hair_dynamics or psys.point_cache.is_baked))) and engine in cls.COMPAT_ENGINES + return ( + (psys.settings.type in {'EMITTER', 'REACTOR'} or + (psys.settings.type == 'HAIR' and + (psys.use_hair_dynamics or psys.point_cache.is_baked))) and + engine in cls.COMPAT_ENGINES + ) def draw(self, context): psys = context.particle_system @@ -743,7 +752,8 @@ class PARTICLE_PT_physics(ParticleButtonsPanel, Panel): layout.label(text="Fluid interaction:") row = layout.row() - row.template_list("UI_UL_list", "particle_targets", psys, "targets", psys, "active_particle_target_index", rows=4) + row.template_list("UI_UL_list", "particle_targets", psys, "targets", + psys, "active_particle_target_index", rows=4) col = row.column() sub = col.row() @@ -760,7 +770,7 @@ class PARTICLE_PT_physics(ParticleButtonsPanel, Panel): row = layout.row() if part.physics_type == 'KEYED': col = row.column() - #doesn't work yet + # doesn't work yet #col.alert = key.valid col.prop(key, "object", text="") col.prop(key, "system", text="System") @@ -770,7 +780,7 @@ class PARTICLE_PT_physics(ParticleButtonsPanel, Panel): col.prop(key, "duration") elif part.physics_type == 'BOIDS': sub = row.row() - #doesn't work yet + # doesn't work yet #sub.alert = key.valid sub.prop(key, "object", text="") sub.prop(key, "system", text="System") @@ -778,7 +788,7 @@ class PARTICLE_PT_physics(ParticleButtonsPanel, Panel): layout.prop(key, "alliance", expand=True) elif part.physics_type == 'FLUID': sub = row.row() - #doesn't work yet + # doesn't work yet #sub.alert = key.valid sub.prop(key, "object", text="") sub.prop(key, "system", text="System") @@ -831,7 +841,8 @@ class PARTICLE_PT_boidbrain(ParticleButtonsPanel, Panel): row.label(text="") row = layout.row() - row.template_list("UI_UL_list", "particle_boids_rules", state, "rules", state, "active_boid_rule_index", rows=4) + row.template_list("UI_UL_list", "particle_boids_rules", state, + "rules", state, "active_boid_rule_index", rows=4) col = row.column() sub = col.row() @@ -848,7 +859,7 @@ class PARTICLE_PT_boidbrain(ParticleButtonsPanel, Panel): if rule: row = layout.row() row.prop(rule, "name", text="") - #somebody make nice icons for boids here please! -jahka + # somebody make nice icons for boids here please! -jahka row.prop(rule, "use_in_air", icon='TRIA_UP', text="") row.prop(rule, "use_on_land", icon='TRIA_DOWN', text="") @@ -1404,5 +1415,29 @@ class PARTICLE_PT_custom_props(ParticleButtonsPanel, PropertyPanel, Panel): _context_path = "particle_system.settings" _property_type = bpy.types.ParticleSettings + +classes = ( + PARTICLE_MT_specials, + PARTICLE_MT_hair_dynamics_presets, + PARTICLE_UL_particle_systems, + PARTICLE_PT_context_particles, + PARTICLE_PT_emission, + PARTICLE_PT_hair_dynamics, + PARTICLE_PT_cache, + PARTICLE_PT_velocity, + PARTICLE_PT_rotation, + PARTICLE_PT_physics, + PARTICLE_PT_boidbrain, + PARTICLE_PT_render, + PARTICLE_PT_draw, + PARTICLE_PT_children, + PARTICLE_PT_field_weights, + PARTICLE_PT_force_fields, + PARTICLE_PT_vertexgroups, + PARTICLE_PT_custom_props, +) + if __name__ == "__main__": # only for live edit. - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/properties_physics_cloth.py b/release/scripts/startup/bl_ui/properties_physics_cloth.py index 3ebf2691b4c..0a7890670b2 100644 --- a/release/scripts/startup/bl_ui/properties_physics_cloth.py +++ b/release/scripts/startup/bl_ui/properties_physics_cloth.py @@ -256,5 +256,17 @@ class PHYSICS_PT_cloth_field_weights(PhysicButtonsPanel, Panel): cloth = context.cloth.settings effector_weights_ui(self, context, cloth.effector_weights, 'CLOTH') +classes = ( + CLOTH_MT_presets, + PHYSICS_PT_cloth, + PHYSICS_PT_cloth_cache, + PHYSICS_PT_cloth_collision, + PHYSICS_PT_cloth_stiffness, + PHYSICS_PT_cloth_sewing, + PHYSICS_PT_cloth_field_weights, +) + if __name__ == "__main__": # only for live edit. - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/properties_physics_common.py b/release/scripts/startup/bl_ui/properties_physics_common.py index 277b59d187d..0b98d8738dc 100644 --- a/release/scripts/startup/bl_ui/properties_physics_common.py +++ b/release/scripts/startup/bl_ui/properties_physics_common.py @@ -274,6 +274,8 @@ def basic_force_field_settings_ui(self, context, field): col.prop(field, "use_global_coords", text="Global") elif field.type == 'HARMONIC': col.prop(field, "use_multiple_springs") + if field.type == 'FORCE': + col.prop(field, "use_gravity_falloff", text="Gravitation") split = layout.split() @@ -316,5 +318,12 @@ def basic_force_field_falloff_ui(self, context, field): sub.active = field.use_max_distance sub.prop(field, "distance_max", text="Maximum") + +classes = ( + PHYSICS_PT_add, +) + if __name__ == "__main__": # only for live edit. - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/properties_physics_dynamicpaint.py b/release/scripts/startup/bl_ui/properties_physics_dynamicpaint.py index 6c3a3246cf6..c813350be08 100644 --- a/release/scripts/startup/bl_ui/properties_physics_dynamicpaint.py +++ b/release/scripts/startup/bl_ui/properties_physics_dynamicpaint.py @@ -529,12 +529,20 @@ class PHYSICS_PT_dp_brush_wave(PhysicButtonsPanel, Panel): row.prop(brush, "wave_clamp") -def register(): - bpy.utils.register_module(__name__) - - -def unregister(): - bpy.utils.register_module(__name__) - -if __name__ == "__main__": - register() +classes = ( + PHYSICS_UL_dynapaint_surfaces, + PHYSICS_PT_dynamic_paint, + PHYSICS_PT_dp_advanced_canvas, + PHYSICS_PT_dp_canvas_output, + PHYSICS_PT_dp_canvas_initial_color, + PHYSICS_PT_dp_effects, + PHYSICS_PT_dp_cache, + PHYSICS_PT_dp_brush_source, + PHYSICS_PT_dp_brush_velocity, + PHYSICS_PT_dp_brush_wave, +) + +if __name__ == "__main__": # only for live edit. + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/properties_physics_field.py b/release/scripts/startup/bl_ui/properties_physics_field.py index 1cebc0496b0..2b12fcf982d 100644 --- a/release/scripts/startup/bl_ui/properties_physics_field.py +++ b/release/scripts/startup/bl_ui/properties_physics_field.py @@ -228,5 +228,13 @@ class PHYSICS_PT_collision(PhysicButtonsPanel, Panel): col.label(text="Force Fields:") col.prop(settings, "absorption", text="Absorption") + +classes = ( + PHYSICS_PT_field, + PHYSICS_PT_collision, +) + if __name__ == "__main__": # only for live edit. - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/properties_physics_fluid.py b/release/scripts/startup/bl_ui/properties_physics_fluid.py index 5d7034c2e68..ab92370f9ae 100644 --- a/release/scripts/startup/bl_ui/properties_physics_fluid.py +++ b/release/scripts/startup/bl_ui/properties_physics_fluid.py @@ -308,5 +308,16 @@ class PHYSICS_PT_domain_particles(PhysicButtonsPanel, Panel): row.prop(fluid, "tracer_particles", text="Tracer") row.prop(fluid, "generate_particles", text="Generate") + +classes = ( + FLUID_MT_presets, + PHYSICS_PT_fluid, + PHYSICS_PT_domain_gravity, + PHYSICS_PT_domain_boundary, + PHYSICS_PT_domain_particles, +) + if __name__ == "__main__": # only for live edit. - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/properties_physics_rigidbody.py b/release/scripts/startup/bl_ui/properties_physics_rigidbody.py index eeb21c046bf..6afdd800b88 100644 --- a/release/scripts/startup/bl_ui/properties_physics_rigidbody.py +++ b/release/scripts/startup/bl_ui/properties_physics_rigidbody.py @@ -137,5 +137,14 @@ class PHYSICS_PT_rigid_body_dynamics(PHYSICS_PT_rigidbody_panel, Panel): col.prop(rbo, "linear_damping", text="Translation") col.prop(rbo, "angular_damping", text="Rotation") + +classes = ( + PHYSICS_PT_rigid_body, + PHYSICS_PT_rigid_body_collisions, + PHYSICS_PT_rigid_body_dynamics, +) + if __name__ == "__main__": # only for live edit. - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/properties_physics_rigidbody_constraint.py b/release/scripts/startup/bl_ui/properties_physics_rigidbody_constraint.py index 9d4f51b256b..84a4cbb4b68 100644 --- a/release/scripts/startup/bl_ui/properties_physics_rigidbody_constraint.py +++ b/release/scripts/startup/bl_ui/properties_physics_rigidbody_constraint.py @@ -260,5 +260,11 @@ class PHYSICS_PT_rigid_body_constraint(PHYSICS_PT_rigidbody_constraint_panel, Pa sub.prop(rbc, "spring_damping_ang_z", text="Damping") +classes = ( + PHYSICS_PT_rigid_body_constraint, +) + if __name__ == "__main__": # only for live edit. - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/properties_physics_smoke.py b/release/scripts/startup/bl_ui/properties_physics_smoke.py index ee9135b9dbf..f2e6c1e22e3 100644 --- a/release/scripts/startup/bl_ui/properties_physics_smoke.py +++ b/release/scripts/startup/bl_ui/properties_physics_smoke.py @@ -21,9 +21,9 @@ import bpy from bpy.types import Panel from bl_ui.properties_physics_common import ( - point_cache_ui, - effector_weights_ui, - ) + point_cache_ui, + effector_weights_ui, +) class PhysicButtonsPanel: @@ -380,23 +380,23 @@ class PHYSICS_PT_smoke_display_settings(PhysicButtonsPanel, Panel): do_axis_slicing = (slice_method == 'AXIS_ALIGNED') do_full_slicing = (axis_slice_method == 'FULL') - row = layout.row(); + row = layout.row() row.enabled = do_axis_slicing row.prop(domain, "axis_slice_method") - col = layout.column(); + col = layout.column() col.enabled = not do_full_slicing and do_axis_slicing col.prop(domain, "slice_axis") col.prop(domain, "slice_depth") - row = layout.row(); + row = layout.row() row.enabled = do_full_slicing or not do_axis_slicing row.prop(domain, "slice_per_voxel") layout.separator() layout.label(text="Debug:") layout.prop(domain, "draw_velocity") - col = layout.column(); + col = layout.column() col.enabled = domain.draw_velocity col.prop(domain, "vector_draw_type") col.prop(domain, "vector_scale") @@ -404,11 +404,25 @@ class PHYSICS_PT_smoke_display_settings(PhysicButtonsPanel, Panel): layout.separator() layout.label(text="Color Mapping:") layout.prop(domain, "use_color_ramp") - col = layout.column(); + col = layout.column() col.enabled = domain.use_color_ramp col.prop(domain, "coba_field") col.template_color_ramp(domain, "color_ramp", expand=True) +classes = ( + PHYSICS_PT_smoke, + PHYSICS_PT_smoke_flow_advanced, + PHYSICS_PT_smoke_fire, + PHYSICS_PT_smoke_adaptive_domain, + PHYSICS_PT_smoke_highres, + PHYSICS_PT_smoke_groups, + PHYSICS_PT_smoke_cache, + PHYSICS_PT_smoke_field_weights, + PHYSICS_PT_smoke_display_settings, +) + if __name__ == "__main__": # only for live edit. - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/properties_physics_softbody.py b/release/scripts/startup/bl_ui/properties_physics_softbody.py index a458af739f2..5960428e4ae 100644 --- a/release/scripts/startup/bl_ui/properties_physics_softbody.py +++ b/release/scripts/startup/bl_ui/properties_physics_softbody.py @@ -244,5 +244,18 @@ class PHYSICS_PT_softbody_field_weights(PhysicButtonsPanel, Panel): effector_weights_ui(self, context, softbody.effector_weights, 'SOFTBODY') + +classes = ( + PHYSICS_PT_softbody, + PHYSICS_PT_softbody_cache, + PHYSICS_PT_softbody_goal, + PHYSICS_PT_softbody_edge, + PHYSICS_PT_softbody_collision, + PHYSICS_PT_softbody_solver, + PHYSICS_PT_softbody_field_weights, +) + if __name__ == "__main__": # only for live edit. - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/properties_render.py b/release/scripts/startup/bl_ui/properties_render.py index 6d23c07152e..a7e8d9273ad 100644 --- a/release/scripts/startup/bl_ui/properties_render.py +++ b/release/scripts/startup/bl_ui/properties_render.py @@ -584,5 +584,24 @@ class RENDER_PT_bake(RenderButtonsPanel, Panel): sub.prop(rd, "bake_user_scale", text="User Scale") +classes = ( + RENDER_MT_presets, + RENDER_MT_ffmpeg_presets, + RENDER_MT_framerate_presets, + RENDER_PT_render, + RENDER_PT_dimensions, + RENDER_PT_antialiasing, + RENDER_PT_motion_blur, + RENDER_PT_shading, + RENDER_PT_performance, + RENDER_PT_post_processing, + RENDER_PT_stamp, + RENDER_PT_output, + RENDER_PT_encoding, + RENDER_PT_bake, +) + if __name__ == "__main__": # only for live edit. - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/properties_render_layer.py b/release/scripts/startup/bl_ui/properties_render_layer.py index 9b8bc237db9..084bc387822 100644 --- a/release/scripts/startup/bl_ui/properties_render_layer.py +++ b/release/scripts/startup/bl_ui/properties_render_layer.py @@ -227,5 +227,16 @@ class RENDERLAYER_PT_views(RenderLayerButtonsPanel, Panel): row.prop(rv, "camera_suffix", text="") +classes = ( + RENDERLAYER_UL_renderlayers, + RENDERLAYER_PT_layers, + RENDERLAYER_PT_layer_options, + RENDERLAYER_PT_layer_passes, + RENDERLAYER_UL_renderviews, + RENDERLAYER_PT_views, +) + if __name__ == "__main__": # only for live edit. - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/properties_scene.py b/release/scripts/startup/bl_ui/properties_scene.py index d6253ec7fbc..fb93c0b56f2 100644 --- a/release/scripts/startup/bl_ui/properties_scene.py +++ b/release/scripts/startup/bl_ui/properties_scene.py @@ -443,5 +443,25 @@ class SCENE_PT_custom_props(SceneButtonsPanel, PropertyPanel, Panel): _context_path = "scene" _property_type = bpy.types.Scene + +classes = ( + SCENE_MT_units_length_presets, + SCENE_UL_keying_set_paths, + SCENE_PT_scene, + SCENE_PT_unit, + SCENE_PT_keying_sets, + SCENE_PT_keying_set_paths, + SCENE_PT_color_management, + SCENE_PT_audio, + SCENE_PT_physics, + SCENE_PT_rigid_body_world, + SCENE_PT_rigid_body_cache, + SCENE_PT_rigid_body_field_weights, + SCENE_PT_simplify, + SCENE_PT_custom_props, +) + if __name__ == "__main__": # only for live edit. - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/properties_texture.py b/release/scripts/startup/bl_ui/properties_texture.py index caf19a9e469..d05527b7ef6 100644 --- a/release/scripts/startup/bl_ui/properties_texture.py +++ b/release/scripts/startup/bl_ui/properties_texture.py @@ -21,15 +21,15 @@ import bpy from bpy.types import Menu, Panel, UIList from bpy.types import ( - Brush, - FreestyleLineStyle, - Lamp, - Material, - Object, - ParticleSettings, - Texture, - World, - ) + Brush, + FreestyleLineStyle, + Lamp, + Material, + Object, + ParticleSettings, + Texture, + World, +) from rna_prop_ui import PropertyPanel @@ -60,6 +60,7 @@ class TEXTURE_MT_envmap_specials(Menu): class TEXTURE_UL_texslots(UIList): + def draw_item(self, context, layout, data, item, icon, active_data, active_propname, index): # assert(isinstance(item, bpy.types.MaterialTextureSlot) ma = data @@ -198,7 +199,8 @@ class TEXTURE_PT_context_texture(TextureButtonsPanel, Panel): if tex_collection: row = layout.row() - row.template_list("TEXTURE_UL_texslots", "", idblock, "texture_slots", idblock, "active_texture_index", rows=2) + row.template_list("TEXTURE_UL_texslots", "", idblock, "texture_slots", + idblock, "active_texture_index", rows=2) col = row.column(align=True) col.operator("texture.slot_move", text="", icon='TRIA_UP').type = 'UP' @@ -1210,12 +1212,22 @@ class TEXTURE_PT_influence(TextureSlotPanel, Panel): row = layout.row() sub = row.row() - sub.active = (tex.use_map_normal or tex.use_map_warp) and not (tex.texture.type == 'IMAGE' and (tex.texture.use_normal_map or tex.texture.use_derivative_map)) + sub.active = ( + (tex.use_map_normal or tex.use_map_warp) and + not (tex.texture.type == 'IMAGE' and + (tex.texture.use_normal_map or tex.texture.use_derivative_map)) + ) sub.prop(tex, "bump_method", text="Method") - # the space setting is supported for: derivative-maps + bump-maps (DEFAULT,BEST_QUALITY), not for normal-maps + # the space setting is supported for: derivative-maps + bump-maps + # (DEFAULT,BEST_QUALITY), not for normal-maps sub = row.row() - sub.active = (tex.use_map_normal or tex.use_map_warp) and not (tex.texture.type == 'IMAGE' and tex.texture.use_normal_map) and ((tex.bump_method in {'BUMP_LOW_QUALITY', 'BUMP_MEDIUM_QUALITY', 'BUMP_BEST_QUALITY'}) or (tex.texture.type == 'IMAGE' and tex.texture.use_derivative_map)) + sub.active = ( + (tex.use_map_normal or tex.use_map_warp) and + not (tex.texture.type == 'IMAGE' and tex.texture.use_normal_map) and + ((tex.bump_method in {'BUMP_LOW_QUALITY', 'BUMP_MEDIUM_QUALITY', 'BUMP_BEST_QUALITY'}) or + (tex.texture.type == 'IMAGE' and tex.texture.use_derivative_map)) + ) sub.prop(tex, "bump_objectspace", text="Space") @@ -1224,5 +1236,38 @@ class TEXTURE_PT_custom_props(TextureButtonsPanel, PropertyPanel, Panel): _context_path = "texture" _property_type = Texture + +classes = ( + TEXTURE_MT_specials, + TEXTURE_MT_envmap_specials, + TEXTURE_UL_texslots, + TEXTURE_PT_context_texture, + TEXTURE_PT_preview, + TEXTURE_PT_colors, + TEXTURE_PT_clouds, + TEXTURE_PT_wood, + TEXTURE_PT_marble, + TEXTURE_PT_magic, + TEXTURE_PT_blend, + TEXTURE_PT_stucci, + TEXTURE_PT_image, + TEXTURE_PT_image_sampling, + TEXTURE_PT_image_mapping, + TEXTURE_PT_envmap, + TEXTURE_PT_envmap_sampling, + TEXTURE_PT_musgrave, + TEXTURE_PT_voronoi, + TEXTURE_PT_distortednoise, + TEXTURE_PT_voxeldata, + TEXTURE_PT_pointdensity, + TEXTURE_PT_pointdensity_turbulence, + TEXTURE_PT_ocean, + TEXTURE_PT_mapping, + TEXTURE_PT_influence, + TEXTURE_PT_custom_props, +) + if __name__ == "__main__": # only for live edit. - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/properties_world.py b/release/scripts/startup/bl_ui/properties_world.py index e07349a99ee..6aa39580d34 100644 --- a/release/scripts/startup/bl_ui/properties_world.py +++ b/release/scripts/startup/bl_ui/properties_world.py @@ -248,5 +248,20 @@ class WORLD_PT_custom_props(WorldButtonsPanel, PropertyPanel, Panel): _context_path = "world" _property_type = bpy.types.World + +classes = ( + WORLD_PT_context_world, + WORLD_PT_preview, + WORLD_PT_world, + WORLD_PT_ambient_occlusion, + WORLD_PT_environment_lighting, + WORLD_PT_indirect_lighting, + WORLD_PT_gather, + WORLD_PT_mist, + WORLD_PT_custom_props, +) + if __name__ == "__main__": # only for live edit. - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/space_clip.py b/release/scripts/startup/bl_ui/space_clip.py index 5cc135392c1..3f05620fcf6 100644 --- a/release/scripts/startup/bl_ui/space_clip.py +++ b/release/scripts/startup/bl_ui/space_clip.py @@ -1486,5 +1486,69 @@ class CLIP_MT_stabilize_2d_rotation_specials(Menu): layout.operator("clip.stabilize_2d_rotation_select") +classes = ( + CLIP_UL_tracking_objects, + CLIP_HT_header, + CLIP_MT_track, + CLIP_MT_tracking_editor_menus, + CLIP_MT_masking_editor_menus, + CLIP_PT_track, + CLIP_PT_tools_clip, + CLIP_PT_tools_marker, + CLIP_PT_tracking_settings, + CLIP_PT_tools_tracking, + CLIP_PT_tools_plane_tracking, + CLIP_PT_tools_solve, + CLIP_PT_tools_cleanup, + CLIP_PT_tools_geometry, + CLIP_PT_tools_orientation, + CLIP_PT_tools_object, + CLIP_PT_objects, + CLIP_PT_plane_track, + CLIP_PT_track_settings, + CLIP_PT_tracking_camera, + CLIP_PT_tracking_lens, + CLIP_PT_display, + CLIP_PT_marker, + CLIP_PT_marker_display, + CLIP_PT_stabilization, + CLIP_PT_proxy, + CLIP_PT_mask, + CLIP_PT_mask_layers, + CLIP_PT_mask_display, + CLIP_PT_active_mask_spline, + CLIP_PT_active_mask_point, + CLIP_PT_tools_mask, + CLIP_PT_tools_mask_add, + CLIP_PT_tools_mask_transforms, + CLIP_PT_footage, + CLIP_PT_footage_info, + CLIP_PT_tools_scenesetup, + CLIP_PT_grease_pencil, + CLIP_PT_grease_pencil_palettecolor, + CLIP_PT_tools_grease_pencil_draw, + CLIP_PT_tools_grease_pencil_edit, + CLIP_PT_tools_grease_pencil_sculpt, + CLIP_PT_tools_grease_pencil_brush, + CLIP_PT_tools_grease_pencil_brushcurves, + CLIP_MT_view, + CLIP_MT_clip, + CLIP_MT_proxy, + CLIP_MT_reconstruction, + CLIP_MT_track_visibility, + CLIP_MT_track_transform, + CLIP_MT_select, + CLIP_MT_select_grouped, + CLIP_MT_tracking_specials, + CLIP_MT_camera_presets, + CLIP_MT_track_color_presets, + CLIP_MT_tracking_settings_presets, + CLIP_MT_track_color_specials, + CLIP_MT_stabilize_2d_specials, + CLIP_MT_stabilize_2d_rotation_specials, +) + if __name__ == "__main__": # only for live edit. - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/space_console.py b/release/scripts/startup/bl_ui/space_console.py index 4266abc5ad7..e6f96d989c7 100644 --- a/release/scripts/startup/bl_ui/space_console.py +++ b/release/scripts/startup/bl_ui/space_console.py @@ -102,5 +102,15 @@ def add_scrollback(text, text_type): bpy.ops.console.scrollback_append(text=l.expandtabs(4), type=text_type) + +classes = ( + CONSOLE_HT_header, + CONSOLE_MT_editor_menus, + CONSOLE_MT_console, + CONSOLE_MT_language, +) + if __name__ == "__main__": # only for live edit. - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/space_dopesheet.py b/release/scripts/startup/bl_ui/space_dopesheet.py index 4d365c8dc08..e02649219dd 100644 --- a/release/scripts/startup/bl_ui/space_dopesheet.py +++ b/release/scripts/startup/bl_ui/space_dopesheet.py @@ -449,5 +449,21 @@ class DOPESHEET_MT_delete(Menu): layout.operator("action.clean", text="Clean Channels").channels = True +classes = ( + DOPESHEET_HT_header, + DOPESHEET_MT_editor_menus, + DOPESHEET_MT_view, + DOPESHEET_MT_select, + DOPESHEET_MT_marker, + DOPESHEET_MT_channel, + DOPESHEET_MT_key, + DOPESHEET_MT_key_transform, + DOPESHEET_MT_gpencil_channel, + DOPESHEET_MT_gpencil_frame, + DOPESHEET_MT_delete, +) + if __name__ == "__main__": # only for live edit. - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/space_filebrowser.py b/release/scripts/startup/bl_ui/space_filebrowser.py index d7b93dc36f0..bd72dfa9370 100644 --- a/release/scripts/startup/bl_ui/space_filebrowser.py +++ b/release/scripts/startup/bl_ui/space_filebrowser.py @@ -247,5 +247,18 @@ class FILEBROWSER_PT_advanced_filter(Panel): col.prop(params, "filter_id") +classes = ( + FILEBROWSER_HT_header, + FILEBROWSER_UL_dir, + FILEBROWSER_PT_system_folders, + FILEBROWSER_PT_system_bookmarks, + FILEBROWSER_MT_bookmarks_specials, + FILEBROWSER_PT_bookmarks, + FILEBROWSER_PT_recent_folders, + FILEBROWSER_PT_advanced_filter, +) + if __name__ == "__main__": # only for live edit. - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/space_graph.py b/release/scripts/startup/bl_ui/space_graph.py index 446df9e6e79..21b49df61ae 100644 --- a/release/scripts/startup/bl_ui/space_graph.py +++ b/release/scripts/startup/bl_ui/space_graph.py @@ -42,10 +42,11 @@ class GRAPH_HT_header(Header): dopesheet_filter(layout, context) - layout.prop(st, "use_normalization", text="Normalize") - row = layout.row() - row.active = st.use_normalization - row.prop(st, "use_auto_normalization", text="Auto") + row = layout.row(align=True) + row.prop(st, "use_normalization", icon='NORMALIZE_FCURVES', text="Normalize", toggle=True) + sub = row.row(align=True) + sub.active = st.use_normalization + sub.prop(st, "use_auto_normalization", icon='FILE_REFRESH', text="", toggle=True) row = layout.row(align=True) @@ -298,6 +299,19 @@ class GRAPH_MT_delete(Menu): layout.operator("graph.clean").channels = False layout.operator("graph.clean", text="Clean Channels").channels = True +classes = ( + GRAPH_HT_header, + GRAPH_MT_editor_menus, + GRAPH_MT_view, + GRAPH_MT_select, + GRAPH_MT_marker, + GRAPH_MT_channel, + GRAPH_MT_key, + GRAPH_MT_key_transform, + GRAPH_MT_delete, +) if __name__ == "__main__": # only for live edit. - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/space_image.py b/release/scripts/startup/bl_ui/space_image.py index 04b4cef9512..b274657b4f4 100644 --- a/release/scripts/startup/bl_ui/space_image.py +++ b/release/scripts/startup/bl_ui/space_image.py @@ -1242,5 +1242,59 @@ class IMAGE_PT_tools_grease_pencil_brushcurves(GreasePencilBrushCurvesPanel, Pan bl_space_type = 'IMAGE_EDITOR' +classes = ( + IMAGE_MT_view, + IMAGE_MT_select, + IMAGE_MT_brush, + IMAGE_MT_image, + IMAGE_MT_image_invert, + IMAGE_MT_uvs, + IMAGE_MT_uvs_showhide, + IMAGE_MT_uvs_proportional, + IMAGE_MT_uvs_transform, + IMAGE_MT_uvs_snap, + IMAGE_MT_uvs_mirror, + IMAGE_MT_uvs_weldalign, + IMAGE_MT_uvs_select_mode, + IMAGE_HT_header, + MASK_MT_editor_menus, + IMAGE_PT_mask, + IMAGE_PT_mask_layers, + IMAGE_PT_mask_display, + IMAGE_PT_active_mask_spline, + IMAGE_PT_active_mask_point, + IMAGE_PT_image_properties, + IMAGE_PT_game_properties, + IMAGE_PT_view_properties, + IMAGE_PT_tools_transform_uvs, + IMAGE_PT_paint, + IMAGE_PT_tools_brush_overlay, + IMAGE_PT_tools_brush_texture, + IMAGE_PT_tools_mask, + IMAGE_PT_tools_mask_texture, + IMAGE_PT_tools_brush_tool, + IMAGE_PT_paint_stroke, + IMAGE_PT_paint_curve, + IMAGE_PT_tools_imagepaint_symmetry, + IMAGE_PT_tools_brush_appearance, + IMAGE_PT_tools_paint_options, + IMAGE_UV_sculpt, + IMAGE_UV_sculpt_curve, + IMAGE_PT_view_histogram, + IMAGE_PT_view_waveform, + IMAGE_PT_view_vectorscope, + IMAGE_PT_sample_line, + IMAGE_PT_scope_sample, + IMAGE_PT_grease_pencil, + IMAGE_PT_grease_pencil_palettecolor, + IMAGE_PT_tools_grease_pencil_draw, + IMAGE_PT_tools_grease_pencil_edit, + IMAGE_PT_tools_grease_pencil_sculpt, + IMAGE_PT_tools_grease_pencil_brush, + IMAGE_PT_tools_grease_pencil_brushcurves, +) + if __name__ == "__main__": # only for live edit. - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/space_info.py b/release/scripts/startup/bl_ui/space_info.py index 2a085b8aada..a7b518dfd2e 100644 --- a/release/scripts/startup/bl_ui/space_info.py +++ b/release/scripts/startup/bl_ui/space_info.py @@ -127,6 +127,18 @@ class INFO_MT_file(Menu): layout.operator("wm.save_homefile", icon='SAVE_PREFS') layout.operator("wm.read_factory_settings", icon='LOAD_FACTORY') + if any(bpy.utils.app_template_paths()): + app_template = context.user_preferences.app_template + if app_template: + layout.operator( + "wm.read_factory_settings", + text="Load Factory Template Settings", + icon='LOAD_FACTORY', + ).app_template = app_template + del app_template + + layout.menu("USERPREF_MT_app_templates", icon='FILE_BLEND') + layout.separator() layout.operator_context = 'INVOKE_AREA' @@ -340,5 +352,23 @@ class INFO_MT_help(Menu): layout.operator("wm.splash", icon='BLENDER') + +classes = ( + INFO_HT_header, + INFO_MT_editor_menus, + INFO_MT_file, + INFO_MT_file_import, + INFO_MT_file_export, + INFO_MT_file_external_data, + INFO_MT_file_previews, + INFO_MT_game, + INFO_MT_render, + INFO_MT_opengl_render, + INFO_MT_window, + INFO_MT_help, +) + if __name__ == "__main__": # only for live edit. - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/space_logic.py b/release/scripts/startup/bl_ui/space_logic.py index 1b316a3eede..b552181f491 100644 --- a/release/scripts/startup/bl_ui/space_logic.py +++ b/release/scripts/startup/bl_ui/space_logic.py @@ -130,5 +130,16 @@ class LOGIC_MT_view(Menu): layout.operator("screen.screen_full_area") layout.operator("screen.screen_full_area", text="Toggle Fullscreen Area").use_hide_panels = True + +classes = ( + LOGIC_PT_properties, + LOGIC_MT_logicbricks_add, + LOGIC_HT_header, + LOGIC_MT_editor_menus, + LOGIC_MT_view, +) + if __name__ == "__main__": # only for live edit. - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/space_nla.py b/release/scripts/startup/bl_ui/space_nla.py index 8fbf9bfc6ac..65d020477dc 100644 --- a/release/scripts/startup/bl_ui/space_nla.py +++ b/release/scripts/startup/bl_ui/space_nla.py @@ -209,5 +209,19 @@ class NLA_MT_edit_transform(Menu): layout.operator("transform.transform", text="Extend").mode = 'TIME_EXTEND' layout.operator("transform.transform", text="Scale").mode = 'TIME_SCALE' + +classes = ( + NLA_HT_header, + NLA_MT_edit, + NLA_MT_editor_menus, + NLA_MT_view, + NLA_MT_select, + NLA_MT_marker, + NLA_MT_add, + NLA_MT_edit_transform, +) + if __name__ == "__main__": # only for live edit. - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/space_node.py b/release/scripts/startup/bl_ui/space_node.py index 8821fa0ca58..b939d34e154 100644 --- a/release/scripts/startup/bl_ui/space_node.py +++ b/release/scripts/startup/bl_ui/space_node.py @@ -526,6 +526,33 @@ class NODE_PT_tools_grease_pencil_brushcurves(GreasePencilBrushCurvesPanel, Pane def node_draw_tree_view(layout, context): pass +classes = ( + NODE_HT_header, + NODE_MT_editor_menus, + NODE_MT_add, + NODE_MT_view, + NODE_MT_select, + NODE_MT_node, + NODE_MT_node_color_presets, + NODE_MT_node_color_specials, + NODE_PT_active_node_generic, + NODE_PT_active_node_color, + NODE_PT_active_node_properties, + NODE_PT_backdrop, + NODE_PT_quality, + NODE_UL_interface_sockets, + NODE_PT_grease_pencil, + NODE_PT_grease_pencil_palettecolor, + NODE_PT_grease_pencil_tools, + NODE_PT_tools_grease_pencil_draw, + NODE_PT_tools_grease_pencil_edit, + NODE_PT_tools_grease_pencil_sculpt, + NODE_PT_tools_grease_pencil_brush, + NODE_PT_tools_grease_pencil_brushcurves, +) + if __name__ == "__main__": # only for live edit. - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/space_outliner.py b/release/scripts/startup/bl_ui/space_outliner.py index 6f7c535fe21..4eb550bbb35 100644 --- a/release/scripts/startup/bl_ui/space_outliner.py +++ b/release/scripts/startup/bl_ui/space_outliner.py @@ -130,5 +130,16 @@ class OUTLINER_MT_edit_datablocks(Menu): layout.operator("outliner.drivers_add_selected") layout.operator("outliner.drivers_delete_selected") + +classes = ( + OUTLINER_HT_header, + OUTLINER_MT_editor_menus, + OUTLINER_MT_view, + OUTLINER_MT_search, + OUTLINER_MT_edit_datablocks, +) + if __name__ == "__main__": # only for live edit. - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/space_properties.py b/release/scripts/startup/bl_ui/space_properties.py index f0ca01ec3ba..20cb5719e20 100644 --- a/release/scripts/startup/bl_ui/space_properties.py +++ b/release/scripts/startup/bl_ui/space_properties.py @@ -34,13 +34,11 @@ class PROPERTIES_HT_header(Header): row.prop(view, "context", expand=True, icon_only=True) -def register(): - bpy.utils.register_module(__name__) - - -def unregister(): - bpy.utils.unregister_module(__name__) - - -if __name__ == "__main__": - register() +classes = ( + PROPERTIES_HT_header, +) + +if __name__ == "__main__": # only for live edit. + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/space_sequencer.py b/release/scripts/startup/bl_ui/space_sequencer.py index 6bb516cf929..01da3b48f7d 100644 --- a/release/scripts/startup/bl_ui/space_sequencer.py +++ b/release/scripts/startup/bl_ui/space_sequencer.py @@ -652,17 +652,39 @@ class SEQUENCER_PT_effect(SequencerButtonsPanel, Panel): col.prop(strip, "rotation_start", text="Rotation") elif strip.type == 'MULTICAM': - layout.prop(strip, "multicam_source") - - row = layout.row(align=True) - sub = row.row(align=True) - sub.scale_x = 2.0 - - sub.operator("screen.animation_play", text="", icon='PAUSE' if context.screen.is_animation_playing else 'PLAY') + col = layout.column(align=True) + strip_channel = strip.channel + + col.prop(strip, "multicam_source", text="Source Channel") + + # The multicam strip needs at least 2 strips to be useful + if strip_channel > 2: + BT_ROW = 4 + + col.label("Cut To:") + row = col.row() + + for i in range(1, strip_channel): + if (i % BT_ROW) == 1: + row = col.row(align=True) + + # Workaround - .enabled has to have a separate UI block to work + if i == strip.multicam_source: + sub = row.row(align=True) + sub.enabled = False + sub.operator("sequencer.cut_multicam", text="%d" % i).camera = i + else: + sub_1 = row.row(align=True) + sub_1.enabled = True + sub_1.operator("sequencer.cut_multicam", text="%d" % i).camera = i + + if strip.channel > BT_ROW and (strip_channel - 1) % BT_ROW: + for i in range(strip.channel, strip_channel + ((BT_ROW + 1 - strip_channel) % BT_ROW)): + row.label("") + else: + col.separator() + col.label(text="Two or more channels are needed below this strip", icon="INFO") - row.label("Cut To") - for i in range(1, strip.channel): - row.operator("sequencer.cut_multicam", text="%d" % i).camera = i elif strip.type == 'TEXT': col = layout.column() @@ -1217,5 +1239,37 @@ class SEQUENCER_PT_custom_props(SequencerButtonsPanel, PropertyPanel, Panel): bl_category = "Strip" +classes = ( + SEQUENCER_HT_header, + SEQUENCER_MT_editor_menus, + SEQUENCER_MT_view, + SEQUENCER_MT_view_toggle, + SEQUENCER_MT_select, + SEQUENCER_MT_marker, + SEQUENCER_MT_change, + SEQUENCER_MT_frame, + SEQUENCER_MT_add, + SEQUENCER_MT_add_effect, + SEQUENCER_MT_strip, + SEQUENCER_PT_edit, + SEQUENCER_PT_effect, + SEQUENCER_PT_input, + SEQUENCER_PT_sound, + SEQUENCER_PT_scene, + SEQUENCER_PT_mask, + SEQUENCER_PT_filter, + SEQUENCER_PT_proxy, + SEQUENCER_PT_preview, + SEQUENCER_PT_view, + SEQUENCER_PT_view_safe_areas, + SEQUENCER_PT_modifiers, + SEQUENCER_PT_grease_pencil, + SEQUENCER_PT_grease_pencil_palettecolor, + SEQUENCER_PT_grease_pencil_tools, + SEQUENCER_PT_custom_props, +) + if __name__ == "__main__": # only for live edit. - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/space_text.py b/release/scripts/startup/bl_ui/space_text.py index 1fd10575e07..98a7da855ca 100644 --- a/release/scripts/startup/bl_ui/space_text.py +++ b/release/scripts/startup/bl_ui/space_text.py @@ -215,20 +215,22 @@ class TEXT_MT_templates_py(Menu): bl_label = "Python" def draw(self, context): - self.path_menu(bpy.utils.script_paths("templates_py"), - "text.open", - {"internal": True}, - ) + self.path_menu( + bpy.utils.script_paths("templates_py"), + "text.open", + props_default={"internal": True}, + ) class TEXT_MT_templates_osl(Menu): bl_label = "Open Shading Language" def draw(self, context): - self.path_menu(bpy.utils.script_paths("templates_osl"), - "text.open", - {"internal": True}, - ) + self.path_menu( + bpy.utils.script_paths("templates_osl"), + "text.open", + props_default={"internal": True}, + ) class TEXT_MT_templates(Menu): @@ -341,5 +343,25 @@ class TEXT_MT_toolbox(Menu): layout.operator("text.run_script") + +classes = ( + TEXT_HT_header, + TEXT_MT_edit, + TEXT_MT_editor_menus, + TEXT_PT_properties, + TEXT_PT_find, + TEXT_MT_view, + TEXT_MT_text, + TEXT_MT_templates, + TEXT_MT_templates_py, + TEXT_MT_templates_osl, + TEXT_MT_edit_select, + TEXT_MT_format, + TEXT_MT_edit_to3d, + TEXT_MT_toolbox, +) + if __name__ == "__main__": # only for live edit. - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/space_time.py b/release/scripts/startup/bl_ui/space_time.py index 508e62e4f56..b9a25cd72a0 100644 --- a/release/scripts/startup/bl_ui/space_time.py +++ b/release/scripts/startup/bl_ui/space_time.py @@ -49,7 +49,10 @@ class TIME_HT_header(Header): row.prop(scene, "frame_preview_start", text="Start") row.prop(scene, "frame_preview_end", text="End") - layout.prop(scene, "frame_current", text="") + if scene.show_subframe: + layout.prop(scene, "frame_float", text="") + else: + layout.prop(scene, "frame_current", text="") layout.separator() @@ -135,6 +138,7 @@ class TIME_MT_view(Menu): layout.prop(st, "show_frame_indicator") layout.prop(scene, "show_keys_from_selected_only") + layout.prop(scene, "show_subframe") layout.separator() @@ -265,5 +269,18 @@ def marker_menu_generic(layout): layout.prop(ts, "lock_markers") +classes = ( + TIME_HT_header, + TIME_MT_editor_menus, + TIME_MT_marker, + TIME_MT_view, + TIME_MT_cache, + TIME_MT_frame, + TIME_MT_playback, + TIME_MT_autokey, +) + if __name__ == "__main__": # only for live edit. - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/space_userpref.py b/release/scripts/startup/bl_ui/space_userpref.py index 075a6f870fa..e50beba50d8 100644 --- a/release/scripts/startup/bl_ui/space_userpref.py +++ b/release/scripts/startup/bl_ui/space_userpref.py @@ -90,6 +90,63 @@ class USERPREF_MT_interaction_presets(Menu): draw = Menu.draw_preset +class USERPREF_MT_app_templates(Menu): + bl_label = "Application Templates" + preset_subdir = "app_templates" + + def draw_ex(self, context, *, use_splash=False, use_default=False, use_install=False): + import os + + layout = self.layout + + # now draw the presets + layout.operator_context = 'EXEC_DEFAULT' + + if use_default: + props = layout.operator("wm.read_homefile", text="Default") + props.use_splash = True + props.app_template = "" + layout.separator() + + template_paths = bpy.utils.app_template_paths() + + # expand template paths + app_templates = [] + for path in template_paths: + for d in os.listdir(path): + if d.startswith(("__", ".")): + continue + template = os.path.join(path, d) + if os.path.isdir(template): + # template_paths_expand.append(template) + app_templates.append(d) + + for d in sorted(app_templates): + props = layout.operator( + "wm.read_homefile", + text=bpy.path.display_name(d), + ) + props.use_splash = True + props.app_template = d; + + if use_install: + layout.separator() + layout.operator_context = 'INVOKE_DEFAULT' + props = layout.operator("wm.app_template_install") + + + def draw(self, context): + self.draw_ex(context, use_splash=False, use_default=True, use_install=True) + + +class USERPREF_MT_templates_splash(Menu): + bl_label = "Startup Templates" + preset_subdir = "templates" + + def draw(self, context): + USERPREF_MT_app_templates.draw_ex(self, context, use_splash=True, use_default=True) + + class USERPREF_MT_appconfigs(Menu): bl_label = "AppPresets" preset_subdir = "keyconfig" @@ -110,7 +167,17 @@ class USERPREF_MT_splash(Menu): split = layout.split() row = split.row() - row.label("") + + if any(bpy.utils.app_template_paths()): + row.label("Template:") + template = context.user_preferences.app_template + row.menu( + "USERPREF_MT_templates_splash", + text=bpy.path.display_name(template) if template else "Default", + ) + else: + row.label("") + row = split.row() row.label("Interaction:") @@ -150,6 +217,7 @@ class USERPREF_PT_interface(Panel): col = row.column() col.label(text="Display:") + col.prop(view, "ui_scale", text="Scale") col.prop(view, "show_tooltips") col.prop(view, "show_tooltips_python") col.prop(view, "show_object_info", text="Object Info") @@ -400,11 +468,6 @@ class USERPREF_PT_system(Panel): col = colsplit.column() col.label(text="General:") - col.prop(system, "dpi") - col.label("Virtual Pixel Mode:") - col.prop(system, "virtual_pixel_mode", text="") - - col.separator() col.prop(system, "frame_server_port") col.prop(system, "scrollback", text="Console Scrollback") @@ -453,6 +516,7 @@ class USERPREF_PT_system(Panel): col.separator() col.label(text="Selection") col.prop(system, "select_method", text="") + col.prop(system, "use_select_pick_depth") col.separator() @@ -1243,7 +1307,7 @@ class USERPREF_MT_addons_online_resources(Menu): "wm.url_open", text="API Concepts", icon='URL', ).url = bpy.types.WM_OT_doc_view._prefix + "/info_quickstart.html" layout.operator("wm.url_open", text="Add-on Tutorial", icon='URL', - ).url = "http://www.blender.org/api/blender_python_api_current/info_tutorial_addon.html" + ).url = bpy.types.WM_OT_doc_view._prefix + "/info_tutorial_addon.html" class USERPREF_PT_addons(Panel): @@ -1317,11 +1381,18 @@ class USERPREF_PT_addons(Panel): # set in addon_utils.modules_refresh() if addon_utils.error_duplicates: - self.draw_error(col, - "Multiple addons using the same name found!\n" - "likely a problem with the script search path.\n" - "(see console for details)", - ) + box = col.box() + row = box.row() + row.label("Multiple add-ons with the same name found!") + row.label(icon='ERROR') + box.label("Please delete one of each pair:") + for (addon_name, addon_file, addon_path) in addon_utils.error_duplicates: + box.separator() + sub_col = box.column(align=True) + sub_col.label(addon_name + ":") + sub_col.label(" " + addon_file) + sub_col.label(" " + addon_path) + if addon_utils.error_encoding: self.draw_error(col, @@ -1473,5 +1544,29 @@ class USERPREF_PT_addons(Panel): row.label(text=module_name, translate=False) +classes = ( + USERPREF_HT_header, + USERPREF_PT_tabs, + USERPREF_MT_interaction_presets, + USERPREF_MT_templates_splash, + USERPREF_MT_app_templates, + USERPREF_MT_appconfigs, + USERPREF_MT_splash, + USERPREF_MT_splash_footer, + USERPREF_PT_interface, + USERPREF_PT_edit, + USERPREF_PT_system, + USERPREF_MT_interface_theme_presets, + USERPREF_PT_theme, + USERPREF_PT_file, + USERPREF_MT_ndof_settings, + USERPREF_MT_keyconfigs, + USERPREF_PT_input, + USERPREF_MT_addons_online_resources, + USERPREF_PT_addons, +) + if __name__ == "__main__": # only for live edit. - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/space_view3d.py b/release/scripts/startup/bl_ui/space_view3d.py index 1183d078a8c..f13c7095f67 100644 --- a/release/scripts/startup/bl_ui/space_view3d.py +++ b/release/scripts/startup/bl_ui/space_view3d.py @@ -284,7 +284,8 @@ class VIEW3D_MT_transform_object(VIEW3D_MT_transform_base): layout.separator() layout.operator_context = 'EXEC_REGION_WIN' - layout.operator("transform.transform", text="Align to Transform Orientation").mode = 'ALIGN' # XXX see alignmenu() in edit.c of b2.4x to get this working + # XXX see alignmenu() in edit.c of b2.4x to get this working + layout.operator("transform.transform", text="Align to Transform Orientation").mode = 'ALIGN' layout.separator() @@ -1205,6 +1206,16 @@ class INFO_MT_lamp_add(Menu): layout.operator_enum("object.lamp_add", "type") +class INFO_MT_camera_add(Menu): + bl_idname = "INFO_MT_camera_add" + bl_label = "Camera" + + def draw(self, context): + layout = self.layout + layout.operator_context = 'EXEC_REGION_WIN' + layout.operator("object.camera_add", text="Camera", icon='OUTLINER_OB_CAMERA') + + class INFO_MT_add(Menu): bl_label = "Add" @@ -1236,7 +1247,11 @@ class INFO_MT_add(Menu): layout.operator("object.speaker_add", text="Speaker", icon='OUTLINER_OB_SPEAKER') layout.separator() - layout.operator("object.camera_add", text="Camera", icon='OUTLINER_OB_CAMERA') + if INFO_MT_camera_add.is_extended(): + layout.menu("INFO_MT_camera_add", icon='OUTLINER_OB_CAMERA') + else: + INFO_MT_camera_add.draw(self, context) + layout.menu("INFO_MT_lamp_add", icon='OUTLINER_OB_LAMP') layout.separator() @@ -1533,6 +1548,7 @@ class VIEW3D_MT_object_apply(Menu): layout.separator() layout.operator("object.visual_transform_apply", text="Visual Transform", text_ctxt=i18n_contexts.default) + layout.operator("object.convert", text="Visual Geometry to Mesh", text_ctxt=i18n_contexts.default).target = 'MESH' layout.operator("object.duplicates_make_real") @@ -1744,6 +1760,7 @@ class VIEW3D_MT_brush_paint_modes(Menu): layout.prop(brush, "use_paint_weight", text="Weight Paint") layout.prop(brush, "use_paint_image", text="Texture Paint") + # ********** Vertex paint menu ********** @@ -1813,6 +1830,7 @@ class VIEW3D_MT_vertex_group(Menu): layout.operator("object.vertex_group_remove", text="Remove Active Group").all = False layout.operator("object.vertex_group_remove", text="Remove All Groups").all = True + # ********** Weight paint menu ********** @@ -1851,6 +1869,7 @@ class VIEW3D_MT_paint_weight(Menu): layout.operator("paint.weight_set") + # ********** Sculpt menu ********** @@ -2004,6 +2023,7 @@ class VIEW3D_MT_particle_specials(Menu): class VIEW3D_MT_particle_showhide(ShowHideMenu, Menu): _operator_name = "particle" + # ********** Pose Menu ********** @@ -2277,6 +2297,7 @@ class VIEW3D_MT_bone_options_disable(Menu, BoneOptions): bl_label = "Disable Bone Options" type = 'DISABLE' + # ********** Edit Menus, suffix from ob.type ********** @@ -2444,6 +2465,7 @@ class VIEW3D_MT_edit_mesh_vertices(Menu): with_bullet = bpy.app.build_options.bullet layout.operator("mesh.merge") + layout.operator("mesh.remove_doubles") layout.operator("mesh.rip_move") layout.operator("mesh.rip_move_fill") layout.operator("mesh.rip_edge_move") @@ -2466,7 +2488,6 @@ class VIEW3D_MT_edit_mesh_vertices(Menu): if with_bullet: layout.operator("mesh.convex_hull") layout.operator("mesh.vertices_smooth") - layout.operator("mesh.remove_doubles") layout.operator("mesh.blend_from_shape") @@ -2623,6 +2644,7 @@ class VIEW3D_MT_edit_mesh_clean(Menu): layout.operator("mesh.face_make_planar") layout.operator("mesh.vert_connect_nonplanar") layout.operator("mesh.vert_connect_concave") + layout.operator("mesh.remove_doubles") layout.operator("mesh.fill_holes") @@ -3780,13 +3802,143 @@ class VIEW3D_PT_context_properties(Panel): # Draw with no edit button rna_prop_ui.draw(self.layout, context, member, object, False) - -def register(): - bpy.utils.register_module(__name__) - - -def unregister(): - bpy.utils.unregister_module(__name__) - -if __name__ == "__main__": - register() +classes = ( + VIEW3D_HT_header, + VIEW3D_MT_editor_menus, + VIEW3D_MT_transform, + VIEW3D_MT_transform_base, + VIEW3D_MT_transform_object, + VIEW3D_MT_transform_armature, + VIEW3D_MT_mirror, + VIEW3D_MT_snap, + VIEW3D_MT_uv_map, + VIEW3D_MT_edit_proportional, + VIEW3D_MT_view, + VIEW3D_MT_view_navigation, + VIEW3D_MT_view_align, + VIEW3D_MT_view_align_selected, + VIEW3D_MT_view_cameras, + VIEW3D_MT_select_object, + VIEW3D_MT_select_object_more_less, + VIEW3D_MT_select_pose, + VIEW3D_MT_select_pose_more_less, + VIEW3D_MT_select_particle, + VIEW3D_MT_edit_mesh, + VIEW3D_MT_edit_mesh_select_similar, + VIEW3D_MT_edit_mesh_select_by_trait, + VIEW3D_MT_edit_mesh_select_more_less, + VIEW3D_MT_select_edit_mesh, + VIEW3D_MT_select_edit_curve, + VIEW3D_MT_select_edit_surface, + VIEW3D_MT_select_edit_text, + VIEW3D_MT_select_edit_metaball, + VIEW3D_MT_select_edit_lattice, + VIEW3D_MT_select_edit_armature, + VIEW3D_MT_select_gpencil, + VIEW3D_MT_select_paint_mask, + VIEW3D_MT_select_paint_mask_vertex, + VIEW3D_MT_angle_control, + INFO_MT_mesh_add, + INFO_MT_curve_add, + INFO_MT_surface_add, + INFO_MT_metaball_add, + INFO_MT_edit_curve_add, + INFO_MT_edit_armature_add, + INFO_MT_armature_add, + INFO_MT_lamp_add, + INFO_MT_camera_add, + INFO_MT_add, + VIEW3D_MT_object, + VIEW3D_MT_object_animation, + VIEW3D_MT_object_clear, + VIEW3D_MT_object_specials, + VIEW3D_MT_object_apply, + VIEW3D_MT_object_parent, + VIEW3D_MT_object_track, + VIEW3D_MT_object_group, + VIEW3D_MT_object_constraints, + VIEW3D_MT_object_quick_effects, + VIEW3D_MT_object_showhide, + VIEW3D_MT_make_single_user, + VIEW3D_MT_make_links, + VIEW3D_MT_object_game, + VIEW3D_MT_brush, + VIEW3D_MT_brush_paint_modes, + VIEW3D_MT_paint_vertex, + VIEW3D_MT_hook, + VIEW3D_MT_vertex_group, + VIEW3D_MT_paint_weight, + VIEW3D_MT_sculpt, + VIEW3D_MT_hide_mask, + VIEW3D_MT_particle, + VIEW3D_MT_particle_specials, + VIEW3D_MT_particle_showhide, + VIEW3D_MT_pose, + VIEW3D_MT_pose_transform, + VIEW3D_MT_pose_slide, + VIEW3D_MT_pose_propagate, + VIEW3D_MT_pose_library, + VIEW3D_MT_pose_motion, + VIEW3D_MT_pose_group, + VIEW3D_MT_pose_ik, + VIEW3D_MT_pose_constraints, + VIEW3D_MT_pose_showhide, + VIEW3D_MT_pose_apply, + VIEW3D_MT_pose_specials, + VIEW3D_MT_bone_options_toggle, + VIEW3D_MT_bone_options_enable, + VIEW3D_MT_bone_options_disable, + VIEW3D_MT_edit_mesh_specials, + VIEW3D_MT_edit_mesh_select_mode, + VIEW3D_MT_edit_mesh_extrude, + VIEW3D_MT_edit_mesh_vertices, + VIEW3D_MT_edit_mesh_edges, + VIEW3D_MT_edit_mesh_faces, + VIEW3D_MT_edit_mesh_normals, + VIEW3D_MT_edit_mesh_clean, + VIEW3D_MT_edit_mesh_delete, + VIEW3D_MT_edit_mesh_showhide, + VIEW3D_MT_edit_gpencil, + VIEW3D_MT_edit_gpencil_delete, + VIEW3D_MT_edit_curve, + VIEW3D_MT_edit_curve_ctrlpoints, + VIEW3D_MT_edit_curve_segments, + VIEW3D_MT_edit_curve_specials, + VIEW3D_MT_edit_curve_delete, + VIEW3D_MT_edit_curve_showhide, + VIEW3D_MT_edit_surface, + VIEW3D_MT_edit_font, + VIEW3D_MT_edit_text_chars, + VIEW3D_MT_edit_meta, + VIEW3D_MT_edit_meta_showhide, + VIEW3D_MT_edit_lattice, + VIEW3D_MT_edit_armature, + VIEW3D_MT_armature_specials, + VIEW3D_MT_edit_armature_parent, + VIEW3D_MT_edit_armature_roll, + VIEW3D_MT_edit_armature_delete, + VIEW3D_MT_edit_gpencil_transform, + VIEW3D_MT_edit_gpencil_interpolate, + VIEW3D_PT_grease_pencil, + VIEW3D_PT_grease_pencil_palettecolor, + VIEW3D_PT_view3d_properties, + VIEW3D_PT_view3d_cursor, + VIEW3D_PT_view3d_name, + VIEW3D_PT_view3d_display, + VIEW3D_PT_view3d_stereo, + VIEW3D_PT_view3d_shading, + VIEW3D_PT_view3d_motion_tracking, + VIEW3D_PT_view3d_meshdisplay, + VIEW3D_PT_view3d_meshstatvis, + VIEW3D_PT_view3d_curvedisplay, + VIEW3D_PT_background_image, + VIEW3D_PT_transform_orientations, + VIEW3D_PT_etch_a_ton, + VIEW3D_PT_context_properties, +) + + +if __name__ == "__main__": # only for live edit. + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/bl_ui/space_view3d_toolbar.py b/release/scripts/startup/bl_ui/space_view3d_toolbar.py index 60e86d7544d..d58453deaef 100644 --- a/release/scripts/startup/bl_ui/space_view3d_toolbar.py +++ b/release/scripts/startup/bl_ui/space_view3d_toolbar.py @@ -2008,5 +2008,69 @@ class VIEW3D_PT_tools_history(View3DPanel, Panel): col.operator("screen.repeat_history", text="History...") +classes = ( + VIEW3D_PT_tools_transform, + VIEW3D_PT_tools_object, + VIEW3D_PT_tools_add_object, + VIEW3D_PT_tools_relations, + VIEW3D_PT_tools_animation, + VIEW3D_PT_tools_rigid_body, + VIEW3D_PT_tools_transform_mesh, + VIEW3D_PT_tools_meshedit, + VIEW3D_PT_tools_meshweight, + VIEW3D_PT_tools_add_mesh_edit, + VIEW3D_PT_tools_shading, + VIEW3D_PT_tools_uvs, + VIEW3D_PT_tools_meshedit_options, + VIEW3D_PT_tools_transform_curve, + VIEW3D_PT_tools_curveedit, + VIEW3D_PT_tools_add_curve_edit, + VIEW3D_PT_tools_curveedit_options_stroke, + VIEW3D_PT_tools_transform_surface, + VIEW3D_PT_tools_surfaceedit, + VIEW3D_PT_tools_add_surface_edit, + VIEW3D_PT_tools_textedit, + VIEW3D_PT_tools_armatureedit, + VIEW3D_PT_tools_armatureedit_transform, + VIEW3D_PT_tools_armatureedit_options, + VIEW3D_PT_tools_mballedit, + VIEW3D_PT_tools_add_mball_edit, + VIEW3D_PT_tools_latticeedit, + VIEW3D_PT_tools_posemode, + VIEW3D_PT_tools_posemode_options, + VIEW3D_PT_imapaint_tools_missing, + VIEW3D_PT_tools_brush, + TEXTURE_UL_texpaintslots, + VIEW3D_MT_tools_projectpaint_uvlayer, + VIEW3D_PT_slots_projectpaint, + VIEW3D_PT_stencil_projectpaint, + VIEW3D_PT_tools_brush_overlay, + VIEW3D_PT_tools_brush_texture, + VIEW3D_PT_tools_mask_texture, + VIEW3D_PT_tools_brush_stroke, + VIEW3D_PT_tools_brush_curve, + VIEW3D_PT_sculpt_dyntopo, + VIEW3D_PT_sculpt_options, + VIEW3D_PT_sculpt_symmetry, + VIEW3D_PT_tools_brush_appearance, + VIEW3D_PT_tools_weightpaint, + VIEW3D_PT_tools_weightpaint_options, + VIEW3D_PT_tools_vertexpaint, + VIEW3D_PT_tools_imagepaint_external, + VIEW3D_PT_tools_imagepaint_symmetry, + VIEW3D_PT_tools_projectpaint, + VIEW3D_MT_tools_projectpaint_stencil, + VIEW3D_PT_tools_particlemode, + VIEW3D_PT_tools_grease_pencil_draw, + VIEW3D_PT_tools_grease_pencil_edit, + VIEW3D_PT_tools_grease_pencil_interpolate, + VIEW3D_PT_tools_grease_pencil_sculpt, + VIEW3D_PT_tools_grease_pencil_brush, + VIEW3D_PT_tools_grease_pencil_brushcurves, + VIEW3D_PT_tools_history, +) + if __name__ == "__main__": # only for live edit. - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) diff --git a/release/scripts/startup/keyingsets_builtins.py b/release/scripts/startup/keyingsets_builtins.py index 97320e56378..ce0b1b62d44 100644 --- a/release/scripts/startup/keyingsets_builtins.py +++ b/release/scripts/startup/keyingsets_builtins.py @@ -643,13 +643,41 @@ class BUILTIN_KSI_DeltaScale(KeyingSetInfo): ############################### +classes = ( + BUILTIN_KSI_Available, + BUILTIN_KSI_BendyBones, + BUILTIN_KSI_DeltaLocation, + BUILTIN_KSI_DeltaRotation, + BUILTIN_KSI_DeltaScale, + BUILTIN_KSI_LocRot, + BUILTIN_KSI_LocRotScale, + BUILTIN_KSI_LocScale, + BUILTIN_KSI_Location, + BUILTIN_KSI_RotScale, + BUILTIN_KSI_Rotation, + BUILTIN_KSI_Scaling, + BUILTIN_KSI_VisualLoc, + BUILTIN_KSI_VisualLocRot, + BUILTIN_KSI_VisualLocRotScale, + BUILTIN_KSI_VisualLocScale, + BUILTIN_KSI_VisualRot, + BUILTIN_KSI_VisualRotScale, + BUILTIN_KSI_VisualScaling, + BUILTIN_KSI_WholeCharacter, + BUILTIN_KSI_WholeCharacterSelected, +) + def register(): - bpy.utils.register_module(__name__) + from bpy.utils import register_class + for cls in classes: + register_class(cls) def unregister(): - bpy.utils.unregister_module(__name__) + from bpy.utils import unregister_class + for cls in classes: + unregister_class(cls) if __name__ == "__main__": diff --git a/release/scripts/startup/nodeitems_builtins.py b/release/scripts/startup/nodeitems_builtins.py index 7c0adb3077a..8fd039f2bf7 100644 --- a/release/scripts/startup/nodeitems_builtins.py +++ b/release/scripts/startup/nodeitems_builtins.py @@ -19,7 +19,11 @@ # <pep8 compliant> import bpy import nodeitems_utils -from nodeitems_utils import NodeCategory, NodeItem, NodeItemCustom +from nodeitems_utils import ( + NodeCategory, + NodeItem, + NodeItemCustom, +) # Subclasses for standard node types diff --git a/source/blender/alembic/intern/abc_archive.cc b/source/blender/alembic/intern/abc_archive.cc index 0985a06d732..5f8fc1a3739 100644 --- a/source/blender/alembic/intern/abc_archive.cc +++ b/source/blender/alembic/intern/abc_archive.cc @@ -113,25 +113,25 @@ static OArchive create_archive(std::ostream *ostream, Alembic::Abc::MetaData &md, bool ogawa) { - md.set(Alembic::Abc::kApplicationNameKey, "Blender"); + md.set(Alembic::Abc::kApplicationNameKey, "Blender"); md.set(Alembic::Abc::kUserDescriptionKey, scene_name); - time_t raw_time; - time(&raw_time); - char buffer[128]; + time_t raw_time; + time(&raw_time); + char buffer[128]; #if defined _WIN32 || defined _WIN64 - ctime_s(buffer, 128, &raw_time); + ctime_s(buffer, 128, &raw_time); #else - ctime_r(&raw_time, buffer); + ctime_r(&raw_time, buffer); #endif - const std::size_t buffer_len = strlen(buffer); - if (buffer_len > 0 && buffer[buffer_len - 1] == '\n') { - buffer[buffer_len - 1] = '\0'; - } + const std::size_t buffer_len = strlen(buffer); + if (buffer_len > 0 && buffer[buffer_len - 1] == '\n') { + buffer[buffer_len - 1] = '\0'; + } - md.set(Alembic::Abc::kDateWrittenKey, buffer); + md.set(Alembic::Abc::kDateWrittenKey, buffer); ErrorHandler::Policy policy = ErrorHandler::kThrowPolicy; diff --git a/source/blender/alembic/intern/abc_curves.cc b/source/blender/alembic/intern/abc_curves.cc index 4ecb9d944f2..0542255d84b 100644 --- a/source/blender/alembic/intern/abc_curves.cc +++ b/source/blender/alembic/intern/abc_curves.cc @@ -102,7 +102,7 @@ void AbcCurveWriter::do_write() const BPoint *point = nurbs->bp; for (int i = 0; i < totpoint; ++i, ++point) { - copy_zup_yup(temp_vert.getValue(), point->vec); + copy_yup_from_zup(temp_vert.getValue(), point->vec); verts.push_back(temp_vert); weights.push_back(point->vec[3]); widths.push_back(point->radius); @@ -118,7 +118,7 @@ void AbcCurveWriter::do_write() /* TODO(kevin): store info about handles, Alembic doesn't have this. */ for (int i = 0; i < totpoint; ++i, ++bezier) { - copy_zup_yup(temp_vert.getValue(), bezier->vec[1]); + copy_yup_from_zup(temp_vert.getValue(), bezier->vec[1]); verts.push_back(temp_vert); widths.push_back(bezier->radius); } @@ -322,7 +322,7 @@ void read_curve_sample(Curve *cu, const ICurvesSchema &schema, const float time) weight = (*weights)[idx]; } - copy_yup_zup(bp->vec, pos.getValue()); + copy_zup_from_yup(bp->vec, pos.getValue()); bp->vec[3] = weight; bp->f1 = SELECT; bp->radius = radius; @@ -361,7 +361,7 @@ void read_curve_sample(Curve *cu, const ICurvesSchema &schema, const float time) * object directly and create a new DerivedMesh from that. Also we might need to * create new or delete existing NURBS in the curve. */ -DerivedMesh *AbcCurveReader::read_derivedmesh(DerivedMesh */*dm*/, const float time, int /*read_flag*/, const char **/*err_str*/) +DerivedMesh *AbcCurveReader::read_derivedmesh(DerivedMesh * /*dm*/, const float time, int /*read_flag*/, const char ** /*err_str*/) { ISampleSelector sample_sel(time); const ICurvesSchema::Sample sample = m_curves_schema.getValue(sample_sel); @@ -389,7 +389,7 @@ DerivedMesh *AbcCurveReader::read_derivedmesh(DerivedMesh */*dm*/, const float t for (int i = 0; i < totpoint; ++i, ++point, ++vertex_idx) { const Imath::V3f &pos = (*positions)[vertex_idx]; - copy_yup_zup(point->vec, pos.getValue()); + copy_zup_from_yup(point->vec, pos.getValue()); } } else if (nurbs->bezt) { @@ -397,7 +397,7 @@ DerivedMesh *AbcCurveReader::read_derivedmesh(DerivedMesh */*dm*/, const float t for (int i = 0; i < totpoint; ++i, ++bezier, ++vertex_idx) { const Imath::V3f &pos = (*positions)[vertex_idx]; - copy_yup_zup(bezier->vec[1], pos.getValue()); + copy_zup_from_yup(bezier->vec[1], pos.getValue()); } } } diff --git a/source/blender/alembic/intern/abc_customdata.cc b/source/blender/alembic/intern/abc_customdata.cc index ebf1b2ba96e..0d11ab79ddd 100644 --- a/source/blender/alembic/intern/abc_customdata.cc +++ b/source/blender/alembic/intern/abc_customdata.cc @@ -327,6 +327,11 @@ static void read_custom_data_ex(const ICompoundProperty &prop, } else if (data_type == CD_MLOOPUV) { IV2fGeomParam uv_param(prop, prop_header.getName()); + + if (!uv_param.isIndexed()) { + return; + } + IV2fGeomParam::Sample sample; uv_param.getIndexed(sample, iss); diff --git a/source/blender/alembic/intern/abc_exporter.cc b/source/blender/alembic/intern/abc_exporter.cc index ff8b0442ab6..564157fbc97 100644 --- a/source/blender/alembic/intern/abc_exporter.cc +++ b/source/blender/alembic/intern/abc_exporter.cc @@ -47,7 +47,7 @@ extern "C" { #ifdef WIN32 /* needed for MSCV because of snprintf from BLI_string */ -# include "BLI_winstuff.h" +# include "BLI_winstuff.h" #endif #include "BKE_anim.h" @@ -347,7 +347,7 @@ void AbcExporter::createTransformWritersHierarchy(EvaluationContext *eval_ctx) Object *ob = base->object; if (export_object(&m_settings, ob)) { - switch(ob->type) { + switch (ob->type) { case OB_LAMP: case OB_LATTICE: case OB_MBALL: @@ -382,7 +382,10 @@ void AbcExporter::createTransformWritersFlat() void AbcExporter::exploreTransform(EvaluationContext *eval_ctx, Object *ob, Object *parent, Object *dupliObParent) { - createTransformWriter(ob, parent, dupliObParent); + + if (export_object(&m_settings, ob) && object_is_shape(ob)) { + createTransformWriter(ob, parent, dupliObParent); + } ListBase *lb = object_duplilist(eval_ctx, m_scene, ob); @@ -410,8 +413,12 @@ void AbcExporter::createTransformWriter(Object *ob, Object *parent, Object *dupl { const std::string name = get_object_dag_path_name(ob, dupliObParent); + /* An object should not be its own parent, or we'll get infinite loops. */ + BLI_assert(ob != parent); + BLI_assert(ob != dupliObParent); + /* check if we have already created a transform writer for this object */ - if (m_xforms.find(name) != m_xforms.end()){ + if (getXForm(name) != NULL) { std::cerr << "xform " << name << " already exists\n"; return; } @@ -426,6 +433,14 @@ void AbcExporter::createTransformWriter(Object *ob, Object *parent, Object *dupl if (parent->parent) { createTransformWriter(parent, parent->parent, dupliObParent); } + else if (parent == dupliObParent) { + if (dupliObParent->parent == NULL) { + createTransformWriter(parent, NULL, NULL); + } + else { + createTransformWriter(parent, dupliObParent->parent, dupliObParent->parent); + } + } else { createTransformWriter(parent, dupliObParent, dupliObParent); } @@ -518,7 +533,7 @@ void AbcExporter::createShapeWriter(Object *ob, Object *dupliObParent) } } - switch(ob->type) { + switch (ob->type) { case OB_MESH: { Mesh *me = static_cast<Mesh *>(ob->data); diff --git a/source/blender/alembic/intern/abc_hair.cc b/source/blender/alembic/intern/abc_hair.cc index 14bcf6731ea..e328df24e53 100644 --- a/source/blender/alembic/intern/abc_hair.cc +++ b/source/blender/alembic/intern/abc_hair.cc @@ -77,7 +77,6 @@ void AbcHairWriter::do_write() DerivedMesh *dm = mesh_create_derived_view(m_scene, m_object, CD_MASK_MESH); DM_ensure_tessface(dm); - DM_update_tessface_data(dm); std::vector<Imath::V3f> verts; std::vector<int32_t> hvertices; @@ -164,7 +163,7 @@ void AbcHairWriter::write_hair_sample(DerivedMesh *dm, psys_interpolate_face(mverts, face, tface, NULL, mapfw, vec, normal, NULL, NULL, NULL, NULL); - copy_zup_yup(tmp_nor.getValue(), normal); + copy_yup_from_zup(tmp_nor.getValue(), normal); norm_values.push_back(tmp_nor); } } @@ -198,7 +197,7 @@ void AbcHairWriter::write_hair_sample(DerivedMesh *dm, MVert *mv = mverts + vtx[o]; normal_short_to_float_v3(normal, mv->no); - copy_zup_yup(tmp_nor.getValue(), normal); + copy_yup_from_zup(tmp_nor.getValue(), normal); norm_values.push_back(tmp_nor); found = true; break; diff --git a/source/blender/alembic/intern/abc_mesh.cc b/source/blender/alembic/intern/abc_mesh.cc index bdd75f93189..5a57e43326a 100644 --- a/source/blender/alembic/intern/abc_mesh.cc +++ b/source/blender/alembic/intern/abc_mesh.cc @@ -112,7 +112,7 @@ static void get_vertices(DerivedMesh *dm, std::vector<Imath::V3f> &points) MVert *verts = dm->getVertArray(dm); for (int i = 0, e = dm->getNumVerts(dm); i < e; ++i) { - copy_zup_yup(points[i].getValue(), verts[i].co); + copy_yup_from_zup(points[i].getValue(), verts[i].co); } } @@ -182,7 +182,7 @@ static void get_vertex_normals(DerivedMesh *dm, std::vector<Imath::V3f> &normals for (int i = 0, e = dm->getNumVerts(dm); i < e; ++i) { normal_short_to_float_v3(no, verts[i].no); - copy_zup_yup(normals[i].getValue(), no); + copy_yup_from_zup(normals[i].getValue(), no); } } @@ -211,7 +211,7 @@ static void get_loop_normals(DerivedMesh *dm, std::vector<Imath::V3f> &normals) for (int j = 0; j < mp->totloop; --ml, ++j, ++loop_index) { const int index = ml->v; - copy_zup_yup(normals[loop_index].getValue(), lnors[index]); + copy_yup_from_zup(normals[loop_index].getValue(), lnors[index]); } } } @@ -226,14 +226,14 @@ static void get_loop_normals(DerivedMesh *dm, std::vector<Imath::V3f> &normals) BKE_mesh_calc_poly_normal(mp, ml - (mp->totloop - 1), verts, no); for (int j = 0; j < mp->totloop; --ml, ++j, ++loop_index) { - copy_zup_yup(normals[loop_index].getValue(), no); + copy_yup_from_zup(normals[loop_index].getValue(), no); } } else { /* Smooth shaded, use individual vert normals. */ for (int j = 0; j < mp->totloop; --ml, ++j, ++loop_index) { normal_short_to_float_v3(no, verts[ml->v].no); - copy_zup_yup(normals[loop_index].getValue(), no); + copy_yup_from_zup(normals[loop_index].getValue(), no); } } } @@ -590,7 +590,7 @@ void AbcMeshWriter::getVelocities(DerivedMesh *dm, std::vector<Imath::V3f> &vels float *mesh_vels = reinterpret_cast<float *>(fss->meshVelocities); for (int i = 0; i < totverts; ++i) { - copy_zup_yup(vels[i].getValue(), mesh_vels); + copy_yup_from_zup(vels[i].getValue(), mesh_vels); mesh_vels += 3; } } @@ -691,7 +691,7 @@ static void assign_materials(Main *bmain, Object *ob, const std::map<std::string assigned_name = mat_iter->second; } - assign_material(ob, assigned_name, it->second, BKE_MAT_ASSIGN_OBJECT); + assign_material(ob, assigned_name, it->second, BKE_MAT_ASSIGN_OBDATA); } } } @@ -726,7 +726,7 @@ static void read_mverts_interp(MVert *mverts, const P3fArraySamplePtr &positions const Imath::V3f &ceil_pos = (*ceil_positions)[i]; interp_v3_v3v3(tmp, floor_pos.getValue(), ceil_pos.getValue(), weight); - copy_yup_zup(mvert.co, tmp); + copy_zup_from_yup(mvert.co, tmp); mvert.bweight = 0; } @@ -755,7 +755,7 @@ void read_mverts(MVert *mverts, const P3fArraySamplePtr &positions, const N3fArr MVert &mvert = mverts[i]; Imath::V3f pos_in = (*positions)[i]; - copy_yup_zup(mvert.co, pos_in.getValue()); + copy_zup_from_yup(mvert.co, pos_in.getValue()); mvert.bweight = 0; @@ -765,7 +765,7 @@ void read_mverts(MVert *mverts, const P3fArraySamplePtr &positions, const N3fArr short no[3]; normal_float_to_short_v3(no, nor_in.getValue()); - copy_yup_zup(mvert.no, no); + copy_zup_from_yup(mvert.no, no); } } } diff --git a/source/blender/alembic/intern/abc_nurbs.cc b/source/blender/alembic/intern/abc_nurbs.cc index 4f57dfdae9e..d0b9561f679 100644 --- a/source/blender/alembic/intern/abc_nurbs.cc +++ b/source/blender/alembic/intern/abc_nurbs.cc @@ -153,7 +153,7 @@ void AbcNurbsWriter::do_write() const BPoint *bp = nu->bp; for (int i = 0; i < size; ++i, ++bp) { - copy_zup_yup(positions[i].getValue(), bp->vec); + copy_yup_from_zup(positions[i].getValue(), bp->vec); weights[i] = bp->vec[3]; } @@ -281,7 +281,7 @@ void AbcNurbsReader::readObjectData(Main *bmain, float time) posw_in = (*weights)[i]; } - copy_yup_zup(bp->vec, pos_in.getValue()); + copy_zup_from_yup(bp->vec, pos_in.getValue()); bp->vec[3] = posw_in; bp->f1 = SELECT; bp->radius = 1.0f; diff --git a/source/blender/alembic/intern/abc_object.cc b/source/blender/alembic/intern/abc_object.cc index 9dfccdb8c7f..a5b8af542fc 100644 --- a/source/blender/alembic/intern/abc_object.cc +++ b/source/blender/alembic/intern/abc_object.cc @@ -97,14 +97,14 @@ Imath::Box3d AbcObjectWriter::bounds() return Imath::Box3d(); } - /* Convert Z-up to Y-up. */ + /* Convert Z-up to Y-up. This also changes which vector goes into which min/max property. */ this->m_bounds.min.x = bb->vec[0][0]; this->m_bounds.min.y = bb->vec[0][2]; - this->m_bounds.min.z = -bb->vec[0][1]; + this->m_bounds.min.z = -bb->vec[6][1]; this->m_bounds.max.x = bb->vec[6][0]; this->m_bounds.max.y = bb->vec[6][2]; - this->m_bounds.max.z = -bb->vec[6][1]; + this->m_bounds.max.z = -bb->vec[0][1]; return this->m_bounds; } diff --git a/source/blender/alembic/intern/abc_points.cc b/source/blender/alembic/intern/abc_points.cc index 4c78f3e83c7..fc84759b1d9 100644 --- a/source/blender/alembic/intern/abc_points.cc +++ b/source/blender/alembic/intern/abc_points.cc @@ -200,7 +200,7 @@ void read_points_sample(const IPointsSchema &schema, read_mverts(config.mvert, positions, vnormals); } -DerivedMesh *AbcPointsReader::read_derivedmesh(DerivedMesh *dm, const float time, int /*read_flag*/, const char **/*err_str*/) +DerivedMesh *AbcPointsReader::read_derivedmesh(DerivedMesh *dm, const float time, int /*read_flag*/, const char ** /*err_str*/) { ISampleSelector sample_sel(time); const IPointsSchema::Sample sample = m_schema.getValue(sample_sel); diff --git a/source/blender/alembic/intern/abc_transform.cc b/source/blender/alembic/intern/abc_transform.cc index e2fc7674c4e..2c6ef09326c 100644 --- a/source/blender/alembic/intern/abc_transform.cc +++ b/source/blender/alembic/intern/abc_transform.cc @@ -122,7 +122,7 @@ Imath::Box3d AbcTransformWriter::bounds() return Imath::transform(bounds, m_matrix); } -bool AbcTransformWriter::hasAnimation(Object */*ob*/) const +bool AbcTransformWriter::hasAnimation(Object * /*ob*/) const { /* TODO(kevin): implement this. */ return true; @@ -146,6 +146,6 @@ bool AbcEmptyReader::valid() const void AbcEmptyReader::readObjectData(Main *bmain, float /*time*/) { - m_object = BKE_object_add_only_object(bmain, OB_EMPTY, m_object_name.c_str()); + m_object = BKE_object_add_only_object(bmain, OB_EMPTY, m_data_name.c_str()); m_object->data = NULL; } diff --git a/source/blender/alembic/intern/abc_util.cc b/source/blender/alembic/intern/abc_util.cc index f8ce72d845d..50fa43a3491 100644 --- a/source/blender/alembic/intern/abc_util.cc +++ b/source/blender/alembic/intern/abc_util.cc @@ -37,6 +37,8 @@ extern "C" { #include "DNA_object_types.h" #include "BLI_math.h" + +#include "PIL_time.h" } std::string get_id_name(Object *ob) @@ -197,7 +199,7 @@ void create_transform_matrix(float r_mat[4][4]) copy_m4_m3(transform_mat, rot_mat); /* Add translation to transformation matrix. */ - copy_yup_zup(transform_mat[3], loc); + copy_zup_from_yup(transform_mat[3], loc); /* Create scale matrix. */ scale_mat[0][0] = scale[0]; @@ -417,7 +419,7 @@ void create_transform_matrix(Object *obj, float transform_mat[4][4]) copy_m4_m3(transform_mat, rot_mat); /* Add translation to transformation matrix. */ - copy_zup_yup(transform_mat[3], loc); + copy_yup_from_zup(transform_mat[3], loc); /* Create scale matrix. */ scale_mat[0][0] = scale[0]; @@ -523,3 +525,15 @@ AbcObjectReader *create_reader(const Alembic::AbcGeom::IObject &object, ImportSe return reader; } + +/* ********************** */ + +ScopeTimer::ScopeTimer(const char *message) + : m_message(message) + , m_start(PIL_check_seconds_timer()) +{} + +ScopeTimer::~ScopeTimer() +{ + fprintf(stderr, "%s: %fs\n", m_message, PIL_check_seconds_timer() - m_start); +} diff --git a/source/blender/alembic/intern/abc_util.h b/source/blender/alembic/intern/abc_util.h index 60a96855d14..85ba4d5c9c7 100644 --- a/source/blender/alembic/intern/abc_util.h +++ b/source/blender/alembic/intern/abc_util.h @@ -116,14 +116,14 @@ AbcObjectReader *create_reader(const Alembic::AbcGeom::IObject &object, ImportSe /* Copy from Y-up to Z-up. */ -ABC_INLINE void copy_yup_zup(float zup[3], const float yup[3]) +ABC_INLINE void copy_zup_from_yup(float zup[3], const float yup[3]) { zup[0] = yup[0]; zup[1] = -yup[2]; zup[2] = yup[1]; } -ABC_INLINE void copy_yup_zup(short zup[3], const short yup[3]) +ABC_INLINE void copy_zup_from_yup(short zup[3], const short yup[3]) { zup[0] = yup[0]; zup[1] = -yup[2]; @@ -132,18 +132,37 @@ ABC_INLINE void copy_yup_zup(short zup[3], const short yup[3]) /* Copy from Z-up to Y-up. */ -ABC_INLINE void copy_zup_yup(float yup[3], const float zup[3]) +ABC_INLINE void copy_yup_from_zup(float yup[3], const float zup[3]) { yup[0] = zup[0]; yup[1] = zup[2]; yup[2] = -zup[1]; } -ABC_INLINE void copy_zup_yup(short yup[3], const short zup[3]) +ABC_INLINE void copy_yup_from_zup(short yup[3], const short zup[3]) { yup[0] = zup[0]; yup[1] = zup[2]; yup[2] = -zup[1]; } +/* *************************** */ + +#undef ABC_DEBUG_TIME + +class ScopeTimer { + const char *m_message; + double m_start; + +public: + ScopeTimer(const char *message); + ~ScopeTimer(); +}; + +#ifdef ABC_DEBUG_TIME +# define SCOPE_TIMER(message) ScopeTimer prof(message) +#else +# define SCOPE_TIMER(message) +#endif + #endif /* __ABC_UTIL_H__ */ diff --git a/source/blender/alembic/intern/alembic_capi.cc b/source/blender/alembic/intern/alembic_capi.cc index d8d017119b1..dc5146a26e0 100644 --- a/source/blender/alembic/intern/alembic_capi.cc +++ b/source/blender/alembic/intern/alembic_capi.cc @@ -542,6 +542,8 @@ ABC_INLINE bool is_mesh_and_strands(const IObject &object) static void import_startjob(void *user_data, short *stop, short *do_update, float *progress) { + SCOPE_TIMER("Alembic import, objects reading and creation"); + ImportJobData *data = static_cast<ImportJobData *>(user_data); data->stop = stop; @@ -677,6 +679,8 @@ static void import_startjob(void *user_data, short *stop, short *do_update, floa static void import_endjob(void *user_data) { + SCOPE_TIMER("Alembic import, cleanup"); + ImportJobData *data = static_cast<ImportJobData *>(user_data); std::vector<AbcObjectReader *>::iterator iter; diff --git a/source/blender/blenkernel/BKE_appdir.h b/source/blender/blenkernel/BKE_appdir.h index 077fe2a629c..ac8f861fa56 100644 --- a/source/blender/blenkernel/BKE_appdir.h +++ b/source/blender/blenkernel/BKE_appdir.h @@ -27,11 +27,15 @@ /* note on naming: typical _get() suffix is omitted here, * since its the main purpose of the API. */ const char *BKE_appdir_folder_default(void); +const char *BKE_appdir_folder_id_ex(const int folder_id, const char *subfolder, char *path, size_t path_len); const char *BKE_appdir_folder_id(const int folder_id, const char *subfolder); const char *BKE_appdir_folder_id_create(const int folder_id, const char *subfolder); const char *BKE_appdir_folder_id_user_notest(const int folder_id, const char *subfolder); const char *BKE_appdir_folder_id_version(const int folder_id, const int ver, const bool do_check); +bool BKE_appdir_app_template_any(void); +bool BKE_appdir_app_template_id_search(const char *app_template, char *path, size_t path_len); + /* Initialize path to program executable */ void BKE_appdir_program_path_init(const char *argv0); diff --git a/source/blender/blenkernel/BKE_blender.h b/source/blender/blenkernel/BKE_blender.h index d2d9c763031..d55926ffb1e 100644 --- a/source/blender/blenkernel/BKE_blender.h +++ b/source/blender/blenkernel/BKE_blender.h @@ -38,14 +38,22 @@ extern "C" { #endif +struct UserDef; + void BKE_blender_free(void); void BKE_blender_globals_init(void); void BKE_blender_globals_clear(void); +void BKE_blender_version_string( + char *version_str, size_t maxncpy, + short version, short subversion, bool v_prefix, bool include_subversion); -void BKE_blender_userdef_free(void); +void BKE_blender_userdef_set_data(struct UserDef *userdef); +void BKE_blender_userdef_free_data(struct UserDef *userdef); void BKE_blender_userdef_refresh(void); - + +void BKE_blender_userdef_set_app_template(struct UserDef *userdef); + /* set this callback when a UI is running */ void BKE_blender_callback_test_break_set(void (*func)(void)); int BKE_blender_test_break(void); diff --git a/source/blender/blenkernel/BKE_blendfile.h b/source/blender/blenkernel/BKE_blendfile.h index 6767fce3abd..ac58451e412 100644 --- a/source/blender/blenkernel/BKE_blendfile.h +++ b/source/blender/blenkernel/BKE_blendfile.h @@ -33,8 +33,7 @@ struct ID; struct Main; struct MemFile; struct ReportList; - -int BKE_blendfile_read(struct bContext *C, const char *filepath, struct ReportList *reports); +struct UserDef; enum { BKE_BLENDFILE_READ_FAIL = 0, /* no load */ @@ -42,15 +41,24 @@ enum { BKE_BLENDFILE_READ_OK_USERPREFS = 2, /* OK, and with new user settings */ }; +int BKE_blendfile_read( + struct bContext *C, const char *filepath, + struct ReportList *reports, int skip_flag); bool BKE_blendfile_read_from_memory( - struct bContext *C, const void *filebuf, - int filelength, struct ReportList *reports, bool update_defaults); + struct bContext *C, const void *filebuf, int filelength, + struct ReportList *reports, int skip_flag, bool update_defaults); bool BKE_blendfile_read_from_memfile( struct bContext *C, struct MemFile *memfile, + struct ReportList *reports, int skip_flag); +void BKE_blendfile_read_make_empty(struct bContext *C); + +struct UserDef *BKE_blendfile_userdef_read( + const char *filepath, struct ReportList *reports); +struct UserDef *BKE_blendfile_userdef_read_from_memory( + const void *filebuf, int filelength, struct ReportList *reports); -int BKE_blendfile_read_userdef(const char *filepath, struct ReportList *reports); -int BKE_blendfile_write_userdef(const char *filepath, struct ReportList *reports); +int BKE_blendfile_userdef_write(const char *filepath, struct ReportList *reports); /* partial blend file writing */ diff --git a/source/blender/blenkernel/BKE_bvhutils.h b/source/blender/blenkernel/BKE_bvhutils.h index 07db2217bac..cb72f0859d5 100644 --- a/source/blender/blenkernel/BKE_bvhutils.h +++ b/source/blender/blenkernel/BKE_bvhutils.h @@ -54,7 +54,6 @@ typedef struct BVHTreeFromEditMesh { /* default callbacks to bvh nearest and raycast */ BVHTree_NearestPointCallback nearest_callback; BVHTree_RayCastCallback raycast_callback; - BVHTree_NearestToRayCallback nearest_to_ray_callback; struct BMEditMesh *em; @@ -75,7 +74,6 @@ typedef struct BVHTreeFromMesh { /* default callbacks to bvh nearest and raycast */ BVHTree_NearestPointCallback nearest_callback; BVHTree_RayCastCallback raycast_callback; - BVHTree_NearestToRayCallback nearest_to_ray_callback; /* Vertex array, so that callbacks have instante access to data */ const struct MVert *vert; @@ -104,7 +102,7 @@ typedef struct BVHTreeFromMesh { * The tree is build in mesh space coordinates, this means special care must be made on queries * so that the coordinates and rays are first translated on the mesh local coordinates. * Reason for this is that bvh_from_mesh_* can use a cache in some cases and so it becomes possible to reuse a BVHTree. - * + * * free_bvhtree_from_mesh should be called when the tree is no longer needed. */ BVHTree *bvhtree_from_editmesh_verts( @@ -118,7 +116,7 @@ BVHTree *bvhtree_from_editmesh_verts_ex( BVHTree *bvhtree_from_mesh_verts( struct BVHTreeFromMesh *data, struct DerivedMesh *mesh, float epsilon, int tree_type, int axis); BVHTree *bvhtree_from_mesh_verts_ex( - struct BVHTreeFromMesh *data, struct MVert *vert, const int numVerts, + struct BVHTreeFromMesh *data, const struct MVert *vert, const int numVerts, const bool vert_allocated, const BLI_bitmap *mask, int verts_num_active, float epsilon, int tree_type, int axis); @@ -133,14 +131,20 @@ BVHTree *bvhtree_from_editmesh_edges_ex( BVHTree *bvhtree_from_mesh_edges( struct BVHTreeFromMesh *data, struct DerivedMesh *mesh, float epsilon, int tree_type, int axis); +BVHTree *bvhtree_from_mesh_edges_ex( + struct BVHTreeFromMesh *data, + const struct MVert *vert, const bool vert_allocated, + const struct MEdge *edge, const int edges_num, const bool edge_allocated, + const BLI_bitmap *edges_mask, int edges_num_active, + float epsilon, int tree_type, int axis); BVHTree *bvhtree_from_mesh_faces( struct BVHTreeFromMesh *data, struct DerivedMesh *mesh, float epsilon, int tree_type, int axis); BVHTree *bvhtree_from_mesh_faces_ex( struct BVHTreeFromMesh *data, - struct MVert *vert, const bool vert_allocated, - struct MFace *face, const int numFaces, const bool face_allocated, + const struct MVert *vert, const bool vert_allocated, + const struct MFace *face, const int numFaces, const bool face_allocated, const BLI_bitmap *mask, int numFaces_active, float epsilon, int tree_type, int axis); diff --git a/source/blender/blenkernel/BKE_cachefile.h b/source/blender/blenkernel/BKE_cachefile.h index 7e1c069df9a..0d4ed2083b1 100644 --- a/source/blender/blenkernel/BKE_cachefile.h +++ b/source/blender/blenkernel/BKE_cachefile.h @@ -55,7 +55,7 @@ void BKE_cachefile_reload(const struct Main *bmain, struct CacheFile *cache_file void BKE_cachefile_ensure_handle(const struct Main *bmain, struct CacheFile *cache_file); -void BKE_cachefile_update_frame(struct Main *bmain, struct Scene *scene,const float ctime, const float fps); +void BKE_cachefile_update_frame(struct Main *bmain, struct Scene *scene, const float ctime, const float fps); bool BKE_cachefile_filepath_get( const struct Main *bmain, const struct CacheFile *cache_file, float frame, diff --git a/source/blender/blenkernel/BKE_cdderivedmesh.h b/source/blender/blenkernel/BKE_cdderivedmesh.h index 9948f21ba90..4876461bfe0 100644 --- a/source/blender/blenkernel/BKE_cdderivedmesh.h +++ b/source/blender/blenkernel/BKE_cdderivedmesh.h @@ -79,6 +79,7 @@ DerivedMesh *CDDM_from_curve_displist(struct Object *ob, struct ListBase *dispba */ struct DerivedMesh *CDDM_copy(struct DerivedMesh *dm); struct DerivedMesh *CDDM_copy_from_tessface(struct DerivedMesh *dm); +struct DerivedMesh *CDDM_copy_with_tessface(struct DerivedMesh *dm); /* creates a CDDerivedMesh with the same layer stack configuration as the * given DerivedMesh and containing the requested numbers of elements. diff --git a/source/blender/blenkernel/BKE_cloth.h b/source/blender/blenkernel/BKE_cloth.h index 36330242f18..6c517bd02df 100644 --- a/source/blender/blenkernel/BKE_cloth.h +++ b/source/blender/blenkernel/BKE_cloth.h @@ -136,9 +136,6 @@ typedef struct ClothSpring { float restlen; /* The original length of the spring. */ int type; /* types defined in BKE_cloth.h ("springType") */ int flags; /* defined in BKE_cloth.h, e.g. deactivated due to tearing */ - float dfdx[3][3]; - float dfdv[3][3]; - float f[3]; float stiffness; /* stiffness factor from the vertex groups */ float editrestlen; @@ -240,9 +237,6 @@ void bvhselftree_update_from_cloth(struct ClothModifierData *clmd, bool moving); // needed for button_object.c void cloth_clear_cache (struct Object *ob, struct ClothModifierData *clmd, float framenr ); -// needed for cloth.c -int cloth_add_spring (struct ClothModifierData *clmd, unsigned int indexA, unsigned int indexB, float restlength, int spring_type); - void cloth_parallel_transport_hair_frame(float mat[3][3], const float dir_old[3], const float dir_new[3]); //////////////////////////////////////////////// diff --git a/source/blender/blenkernel/BKE_curve.h b/source/blender/blenkernel/BKE_curve.h index 5558786d254..e111bd0e16b 100644 --- a/source/blender/blenkernel/BKE_curve.h +++ b/source/blender/blenkernel/BKE_curve.h @@ -36,6 +36,7 @@ struct BezTriple; struct Curve; struct EditNurb; +struct GHash; struct ListBase; struct Main; struct Nurb; @@ -52,6 +53,13 @@ typedef struct CurveCache { struct Path *path; } CurveCache; +/* Definitions needed for shape keys */ +typedef struct CVKeyIndex { + void *orig_cv; + int key_index, nu_index, pt_index, vertex_index; + bool switched; +} CVKeyIndex; + #define KNOTSU(nu) ( (nu)->orderu + (nu)->pntsu + (((nu)->flagu & CU_NURB_CYCLIC) ? ((nu)->orderu - 1) : 0) ) #define KNOTSV(nu) ( (nu)->orderv + (nu)->pntsv + (((nu)->flagv & CU_NURB_CYCLIC) ? ((nu)->orderv - 1) : 0) ) @@ -108,7 +116,8 @@ void BK_curve_nurbs_vertexCos_apply(struct ListBase *lb, float (*vertexCos)[3]); float (*BKE_curve_nurbs_keyVertexCos_get(struct ListBase *lb, float *key))[3]; void BKE_curve_nurbs_keyVertexTilts_apply(struct ListBase *lb, float *key); -void BKE_curve_editNurb_keyIndex_free(struct EditNurb *editnurb); +void BKE_curve_editNurb_keyIndex_delCV(struct GHash *keyindex, const void *cv); +void BKE_curve_editNurb_keyIndex_free(struct GHash **keyindex); void BKE_curve_editNurb_free(struct Curve *cu); struct ListBase *BKE_curve_editNurbs_get(struct Curve *cu); diff --git a/source/blender/blenkernel/BKE_mesh.h b/source/blender/blenkernel/BKE_mesh.h index d41878825bb..b83bec5a302 100644 --- a/source/blender/blenkernel/BKE_mesh.h +++ b/source/blender/blenkernel/BKE_mesh.h @@ -131,8 +131,7 @@ bool BKE_mesh_uv_cdlayer_rename(struct Mesh *me, const char *old_name, const cha float (*BKE_mesh_vertexCos_get(const struct Mesh *me, int *r_numVerts))[3]; -void BKE_mesh_calc_normals_split(struct Mesh *mesh); -void BKE_mesh_split_faces(struct Mesh *mesh); +void BKE_mesh_split_faces(struct Mesh *mesh, bool free_loop_normals); struct Mesh *BKE_mesh_new_from_object(struct Main *bmain, struct Scene *sce, struct Object *ob, int apply_modifiers, int settings, int calc_tessface, int calc_undeformed); @@ -228,6 +227,9 @@ void BKE_lnor_space_custom_normal_to_data(MLoopNorSpace *lnor_space, const float bool BKE_mesh_has_custom_loop_normals(struct Mesh *me); +void BKE_mesh_calc_normals_split(struct Mesh *mesh); +void BKE_mesh_calc_normals_split_ex(struct Mesh *mesh, struct MLoopNorSpaceArray *r_lnors_spacearr); + void BKE_mesh_normals_loop_split( const struct MVert *mverts, const int numVerts, struct MEdge *medges, const int numEdges, struct MLoop *mloops, float (*r_loopnors)[3], const int numLoops, diff --git a/source/blender/blenkernel/BKE_node.h b/source/blender/blenkernel/BKE_node.h index e3f28b332cf..7544e8f5b3e 100644 --- a/source/blender/blenkernel/BKE_node.h +++ b/source/blender/blenkernel/BKE_node.h @@ -49,7 +49,7 @@ #include "RNA_types.h" /* not very important, but the stack solver likes to know a maximum */ -#define MAX_SOCKET 64 +#define MAX_SOCKET 512 struct bContext; struct bNode; diff --git a/source/blender/blenkernel/BKE_object.h b/source/blender/blenkernel/BKE_object.h index d812ab832a1..b6633ea6198 100644 --- a/source/blender/blenkernel/BKE_object.h +++ b/source/blender/blenkernel/BKE_object.h @@ -139,8 +139,6 @@ void BKE_boundbox_init_from_minmax(struct BoundBox *bb, const float min[3], cons void BKE_boundbox_calc_center_aabb(const struct BoundBox *bb, float r_cent[3]); void BKE_boundbox_calc_size_aabb(const struct BoundBox *bb, float r_size[3]); void BKE_boundbox_minmax(const struct BoundBox *bb, float obmat[4][4], float r_min[3], float r_max[3]); -struct BoundBox *BKE_boundbox_ensure_minimum_dimensions( - struct BoundBox *bb, struct BoundBox *bb_temp, const float epsilon); struct BoundBox *BKE_object_boundbox_get(struct Object *ob); void BKE_object_dimensions_get(struct Object *ob, float vec[3]); @@ -202,6 +200,11 @@ void BKE_object_eval_uber_data(struct EvaluationContext *eval_ctx, struct Scene *scene, struct Object *ob); +void BKE_object_eval_cloth(struct EvaluationContext *eval_ctx, + struct Scene *scene, + struct Object *object); + + void BKE_object_handle_data_update(struct EvaluationContext *eval_ctx, struct Scene *scene, struct Object *ob); diff --git a/source/blender/blenkernel/BKE_particle.h b/source/blender/blenkernel/BKE_particle.h index e5967be0bc7..2b6a84a2f87 100644 --- a/source/blender/blenkernel/BKE_particle.h +++ b/source/blender/blenkernel/BKE_particle.h @@ -207,7 +207,7 @@ typedef struct ParticleCollisionElement { typedef struct ParticleCollision { struct Object *current; struct Object *hit; - struct Object *skip[PARTICLE_COLLISION_MAX_COLLISIONS+1]; + struct Object *skip[PARTICLE_COLLISION_MAX_COLLISIONS + 1]; struct Object *emitter; struct CollisionModifierData *md; // collision modifier for current object; diff --git a/source/blender/blenkernel/BKE_texture.h b/source/blender/blenkernel/BKE_texture.h index 1c5ea946f59..9a60eb29957 100644 --- a/source/blender/blenkernel/BKE_texture.h +++ b/source/blender/blenkernel/BKE_texture.h @@ -42,6 +42,7 @@ struct Brush; struct ColorBand; struct EnvMap; struct FreestyleLineStyle; +struct ImagePool; struct Lamp; struct Main; struct Material; @@ -133,6 +134,12 @@ struct OceanTex *BKE_texture_ocean_copy(struct OceanTex *ot); bool BKE_texture_dependsOnTime(const struct Tex *texture); bool BKE_texture_is_image_user(const struct Tex *tex); +void BKE_texture_get_value_ex( + const struct Scene *scene, struct Tex *texture, + float *tex_co, struct TexResult *texres, + struct ImagePool *pool, + bool use_color_management); + void BKE_texture_get_value( const struct Scene *scene, struct Tex *texture, float *tex_co, struct TexResult *texres, bool use_color_management); diff --git a/source/blender/blenkernel/intern/DerivedMesh.c b/source/blender/blenkernel/intern/DerivedMesh.c index 160adcc3988..8b7329cf940 100644 --- a/source/blender/blenkernel/intern/DerivedMesh.c +++ b/source/blender/blenkernel/intern/DerivedMesh.c @@ -2912,9 +2912,6 @@ DerivedMesh *editbmesh_get_derived_base(Object *obedit, BMEditMesh *em, CustomDa /* get derived mesh from an object, using editbmesh if available. */ DerivedMesh *object_get_derived_final(Object *ob, const bool for_render) { - Mesh *me = ob->data; - BMEditMesh *em = me->edit_btmesh; - if (for_render) { /* TODO(sergey): use proper derived render here in the future. */ return ob->derivedFinal; @@ -2922,9 +2919,13 @@ DerivedMesh *object_get_derived_final(Object *ob, const bool for_render) /* only return the editmesh if its from this object because * we don't a mesh from another object's modifier stack: T43122 */ - if (em && (em->ob == ob)) { - DerivedMesh *dm = em->derivedFinal; - return dm; + if (ob->type == OB_MESH) { + Mesh *me = ob->data; + BMEditMesh *em = me->edit_btmesh; + if (em && (em->ob == ob)) { + DerivedMesh *dm = em->derivedFinal; + return dm; + } } return ob->derivedFinal; diff --git a/source/blender/blenkernel/intern/appdir.c b/source/blender/blenkernel/intern/appdir.c index b1dcc40279f..43fd47981b1 100644 --- a/source/blender/blenkernel/intern/appdir.c +++ b/source/blender/blenkernel/intern/appdir.c @@ -114,18 +114,26 @@ static char *blender_version_decimal(const int ver) * Concatenates path_base, (optional) path_sep and (optional) folder_name into targetpath, * returning true if result points to a directory. */ -static bool test_path(char *targetpath, const char *path_base, const char *path_sep, const char *folder_name) +static bool test_path( + char *targetpath, size_t targetpath_len, + const char *path_base, const char *path_sep, const char *folder_name) { char tmppath[FILE_MAX]; - if (path_sep) BLI_join_dirfile(tmppath, sizeof(tmppath), path_base, path_sep); - else BLI_strncpy(tmppath, path_base, sizeof(tmppath)); - - /* rare cases folder_name is omitted (when looking for ~/.blender/2.xx dir only) */ - if (folder_name) - BLI_make_file_string("/", targetpath, tmppath, folder_name); - else - BLI_strncpy(targetpath, tmppath, sizeof(tmppath)); + if (path_sep) { + BLI_join_dirfile(tmppath, sizeof(tmppath), path_base, path_sep); + } + else { + BLI_strncpy(tmppath, path_base, sizeof(tmppath)); + } + + /* rare cases folder_name is omitted (when looking for ~/.config/blender/2.xx dir only) */ + if (folder_name) { + BLI_join_dirfile(targetpath, targetpath_len, tmppath, folder_name); + } + else { + BLI_strncpy(targetpath, tmppath, targetpath_len); + } /* FIXME: why is "//" on front of tmppath expanded to "/" (by BLI_join_dirfile) * if folder_name is specified but not otherwise? */ @@ -179,7 +187,9 @@ static bool test_env_path(char *path, const char *envvar) * \param ver To construct name of version-specific directory within bprogdir * \return true if such a directory exists. */ -static bool get_path_local(char *targetpath, const char *folder_name, const char *subfolder_name, const int ver) +static bool get_path_local( + char *targetpath, size_t targetpath_len, + const char *folder_name, const char *subfolder_name, const int ver) { char relfolder[FILE_MAX]; @@ -201,11 +211,12 @@ static bool get_path_local(char *targetpath, const char *folder_name, const char /* try EXECUTABLE_DIR/2.5x/folder_name - new default directory for local blender installed files */ #ifdef __APPLE__ - static char osx_resourses[FILE_MAX]; /* due new codesign situation in OSX > 10.9.5 we must move the blender_version dir with contents to Resources */ + /* due new codesign situation in OSX > 10.9.5 we must move the blender_version dir with contents to Resources */ + static char osx_resourses[FILE_MAX]; sprintf(osx_resourses, "%s../Resources", bprogdir); - return test_path(targetpath, osx_resourses, blender_version_decimal(ver), relfolder); + return test_path(targetpath, targetpath_len, osx_resourses, blender_version_decimal(ver), relfolder); #else - return test_path(targetpath, bprogdir, blender_version_decimal(ver), relfolder); + return test_path(targetpath, targetpath_len, bprogdir, blender_version_decimal(ver), relfolder); #endif } @@ -219,7 +230,7 @@ static bool is_portable_install(void) const int ver = BLENDER_VERSION; char path[FILE_MAX]; - return get_path_local(path, "config", NULL, ver); + return get_path_local(path, sizeof(path), "config", NULL, ver); } /** @@ -233,20 +244,22 @@ static bool is_portable_install(void) * \param ver Blender version, used to construct a subdirectory name * \return true if it was able to construct such a path. */ -static bool get_path_user(char *targetpath, const char *folder_name, const char *subfolder_name, const char *envvar, const int ver) +static bool get_path_user( + char *targetpath, size_t targetpath_len, const char *folder_name, const char *subfolder_name, + const char *envvar, const int ver) { char user_path[FILE_MAX]; const char *user_base_path; /* for portable install, user path is always local */ - if (is_portable_install()) - return get_path_local(targetpath, folder_name, subfolder_name, ver); - + if (is_portable_install()) { + return get_path_local(targetpath, targetpath_len, folder_name, subfolder_name, ver); + } user_path[0] = '\0'; if (test_env_path(user_path, envvar)) { if (subfolder_name) { - return test_path(targetpath, user_path, NULL, subfolder_name); + return test_path(targetpath, targetpath_len, user_path, NULL, subfolder_name); } else { BLI_strncpy(targetpath, user_path, FILE_MAX); @@ -266,10 +279,10 @@ static bool get_path_user(char *targetpath, const char *folder_name, const char #endif if (subfolder_name) { - return test_path(targetpath, user_path, folder_name, subfolder_name); + return test_path(targetpath, targetpath_len, user_path, folder_name, subfolder_name); } else { - return test_path(targetpath, user_path, NULL, folder_name); + return test_path(targetpath, targetpath_len, user_path, NULL, folder_name); } } @@ -283,7 +296,9 @@ static bool get_path_user(char *targetpath, const char *folder_name, const char * \param ver Blender version, used to construct a subdirectory name * \return true if it was able to construct such a path. */ -static bool get_path_system(char *targetpath, const char *folder_name, const char *subfolder_name, const char *envvar, const int ver) +static bool get_path_system( + char *targetpath, size_t targetpath_len, const char *folder_name, const char *subfolder_name, + const char *envvar, const int ver) { char system_path[FILE_MAX]; const char *system_base_path; @@ -307,13 +322,13 @@ static bool get_path_system(char *targetpath, const char *folder_name, const cha /* try CWD/release/folder_name */ if (BLI_current_working_dir(cwd, sizeof(cwd))) { - if (test_path(targetpath, cwd, "release", relfolder)) { + if (test_path(targetpath, targetpath_len, cwd, "release", relfolder)) { return true; } } /* try EXECUTABLE_DIR/release/folder_name */ - if (test_path(targetpath, bprogdir, "release", relfolder)) + if (test_path(targetpath, targetpath_len, bprogdir, "release", relfolder)) return true; /* end developer overrides */ @@ -324,7 +339,7 @@ static bool get_path_system(char *targetpath, const char *folder_name, const cha if (test_env_path(system_path, envvar)) { if (subfolder_name) { - return test_path(targetpath, system_path, NULL, subfolder_name); + return test_path(targetpath, targetpath_len, system_path, NULL, subfolder_name); } else { BLI_strncpy(targetpath, system_path, FILE_MAX); @@ -345,57 +360,63 @@ static bool get_path_system(char *targetpath, const char *folder_name, const cha if (subfolder_name) { /* try $BLENDERPATH/folder_name/subfolder_name */ - return test_path(targetpath, system_path, folder_name, subfolder_name); + return test_path(targetpath, targetpath_len, system_path, folder_name, subfolder_name); } else { /* try $BLENDERPATH/folder_name */ - return test_path(targetpath, system_path, NULL, folder_name); + return test_path(targetpath, targetpath_len, system_path, NULL, folder_name); } } -/* get a folder out of the 'folder_id' presets for paths */ -/* returns the path if found, NULL string if not */ -const char *BKE_appdir_folder_id(const int folder_id, const char *subfolder) +/** + * Get a folder out of the 'folder_id' presets for paths. + * returns the path if found, NULL string if not + * + * \param subfolder: The name of a directory to check for, + * this may contain path separators but must resolve to a directory, checked with #BLI_is_dir. + */ +const char *BKE_appdir_folder_id_ex( + const int folder_id, const char *subfolder, + char *path, size_t path_len) { const int ver = BLENDER_VERSION; - static char path[FILE_MAX] = ""; - + switch (folder_id) { case BLENDER_DATAFILES: /* general case */ - if (get_path_user(path, "datafiles", subfolder, "BLENDER_USER_DATAFILES", ver)) break; - if (get_path_local(path, "datafiles", subfolder, ver)) break; - if (get_path_system(path, "datafiles", subfolder, "BLENDER_SYSTEM_DATAFILES", ver)) break; + if (get_path_user(path, path_len, "datafiles", subfolder, "BLENDER_USER_DATAFILES", ver)) break; + if (get_path_local(path, path_len, "datafiles", subfolder, ver)) break; + if (get_path_system(path, path_len, "datafiles", subfolder, "BLENDER_SYSTEM_DATAFILES", ver)) break; return NULL; case BLENDER_USER_DATAFILES: - if (get_path_user(path, "datafiles", subfolder, "BLENDER_USER_DATAFILES", ver)) break; + if (get_path_user(path, path_len, "datafiles", subfolder, "BLENDER_USER_DATAFILES", ver)) break; return NULL; case BLENDER_SYSTEM_DATAFILES: - if (get_path_local(path, "datafiles", subfolder, ver)) break; - if (get_path_system(path, "datafiles", subfolder, "BLENDER_SYSTEM_DATAFILES", ver)) break; + if (get_path_local(path, path_len, "datafiles", subfolder, ver)) break; + if (get_path_system(path, path_len, "datafiles", subfolder, "BLENDER_SYSTEM_DATAFILES", ver)) break; return NULL; case BLENDER_USER_AUTOSAVE: - if (get_path_user(path, "autosave", subfolder, "BLENDER_USER_DATAFILES", ver)) break; + if (get_path_user(path, path_len, "autosave", subfolder, "BLENDER_USER_DATAFILES", ver)) break; return NULL; case BLENDER_USER_CONFIG: - if (get_path_user(path, "config", subfolder, "BLENDER_USER_CONFIG", ver)) break; + if (get_path_user(path, path_len, "config", subfolder, "BLENDER_USER_CONFIG", ver)) break; return NULL; case BLENDER_USER_SCRIPTS: - if (get_path_user(path, "scripts", subfolder, "BLENDER_USER_SCRIPTS", ver)) break; + if (get_path_user(path, path_len, "scripts", subfolder, "BLENDER_USER_SCRIPTS", ver)) break; return NULL; case BLENDER_SYSTEM_SCRIPTS: - if (get_path_local(path, "scripts", subfolder, ver)) break; - if (get_path_system(path, "scripts", subfolder, "BLENDER_SYSTEM_SCRIPTS", ver)) break; + if (get_path_local(path, path_len, "scripts", subfolder, ver)) break; + if (get_path_system(path, path_len, "scripts", subfolder, "BLENDER_SYSTEM_SCRIPTS", ver)) break; return NULL; case BLENDER_SYSTEM_PYTHON: - if (get_path_local(path, "python", subfolder, ver)) break; - if (get_path_system(path, "python", subfolder, "BLENDER_SYSTEM_PYTHON", ver)) break; + if (get_path_local(path, path_len, "python", subfolder, ver)) break; + if (get_path_system(path, path_len, "python", subfolder, "BLENDER_SYSTEM_PYTHON", ver)) break; return NULL; default: @@ -406,6 +427,13 @@ const char *BKE_appdir_folder_id(const int folder_id, const char *subfolder) return path; } +const char *BKE_appdir_folder_id( + const int folder_id, const char *subfolder) +{ + static char path[FILE_MAX] = ""; + return BKE_appdir_folder_id_ex(folder_id, subfolder, path, sizeof(path)); +} + /** * Returns the path to a folder in the user area without checking that it actually exists first. */ @@ -416,16 +444,16 @@ const char *BKE_appdir_folder_id_user_notest(const int folder_id, const char *su switch (folder_id) { case BLENDER_USER_DATAFILES: - get_path_user(path, "datafiles", subfolder, "BLENDER_USER_DATAFILES", ver); + get_path_user(path, sizeof(path), "datafiles", subfolder, "BLENDER_USER_DATAFILES", ver); break; case BLENDER_USER_CONFIG: - get_path_user(path, "config", subfolder, "BLENDER_USER_CONFIG", ver); + get_path_user(path, sizeof(path), "config", subfolder, "BLENDER_USER_CONFIG", ver); break; case BLENDER_USER_AUTOSAVE: - get_path_user(path, "autosave", subfolder, "BLENDER_USER_AUTOSAVE", ver); + get_path_user(path, sizeof(path), "autosave", subfolder, "BLENDER_USER_AUTOSAVE", ver); break; case BLENDER_USER_SCRIPTS: - get_path_user(path, "scripts", subfolder, "BLENDER_USER_SCRIPTS", ver); + get_path_user(path, sizeof(path), "scripts", subfolder, "BLENDER_USER_SCRIPTS", ver); break; default: BLI_assert(0); @@ -469,13 +497,13 @@ const char *BKE_appdir_folder_id_version(const int folder_id, const int ver, con bool ok; switch (folder_id) { case BLENDER_RESOURCE_PATH_USER: - ok = get_path_user(path, NULL, NULL, NULL, ver); + ok = get_path_user(path, sizeof(path), NULL, NULL, NULL, ver); break; case BLENDER_RESOURCE_PATH_LOCAL: - ok = get_path_local(path, NULL, NULL, ver); + ok = get_path_local(path, sizeof(path), NULL, NULL, ver); break; case BLENDER_RESOURCE_PATH_SYSTEM: - ok = get_path_system(path, NULL, NULL, NULL, ver); + ok = get_path_system(path, sizeof(path), NULL, NULL, NULL, ver); break; default: path[0] = '\0'; /* in case do_check is false */ @@ -655,6 +683,48 @@ bool BKE_appdir_program_python_search( return is_found; } +static const char *app_template_directory_search[2] = { + "startup" SEP_STR "bl_app_templates_user", + "startup" SEP_STR "bl_app_templates_system", +}; + +static const int app_template_directory_id[2] = { + BLENDER_USER_SCRIPTS, + BLENDER_SYSTEM_SCRIPTS, +}; + +/** + * Return true if templates exist + */ +bool BKE_appdir_app_template_any(void) +{ + char temp_dir[FILE_MAX]; + for (int i = 0; i < 2; i++) { + if (BKE_appdir_folder_id_ex( + app_template_directory_id[i], app_template_directory_search[i], + temp_dir, sizeof(temp_dir))) + { + return true; + } + } + return false; +} + +bool BKE_appdir_app_template_id_search(const char *app_template, char *path, size_t path_len) +{ + for (int i = 0; i < 2; i++) { + char subdir[FILE_MAX]; + BLI_join_dirfile(subdir, sizeof(subdir), app_template_directory_search[i], app_template); + if (BKE_appdir_folder_id_ex( + app_template_directory_id[i], subdir, + path, path_len)) + { + return true; + } + } + return false; +} + /** * Gets the temp directory when blender first runs. * If the default path is not found, use try $TEMP @@ -713,7 +783,8 @@ static void where_is_temp(char *fullname, char *basename, const size_t maxlen, c BLI_add_slash(fullname); #ifdef WIN32 if (userdir && userdir != fullname) { - BLI_strncpy(userdir, fullname, maxlen); /* also set user pref to show %TEMP%. /tmp/ is just plain confusing for Windows users. */ + /* also set user pref to show %TEMP%. /tmp/ is just plain confusing for Windows users. */ + BLI_strncpy(userdir, fullname, maxlen); } #endif } @@ -755,7 +826,6 @@ static void where_is_temp(char *fullname, char *basename, const size_t maxlen, c void BKE_tempdir_init(char *userdir) { where_is_temp(btempdir_session, btempdir_base, FILE_MAX, userdir); -; } /** diff --git a/source/blender/blenkernel/intern/armature.c b/source/blender/blenkernel/intern/armature.c index 0287d6ae9ca..02cc31e5977 100644 --- a/source/blender/blenkernel/intern/armature.c +++ b/source/blender/blenkernel/intern/armature.c @@ -1037,6 +1037,17 @@ void armature_deform_verts(Object *armOb, Object *target, DerivedMesh *dm, float if (use_dverts) { defnrToPC = MEM_callocN(sizeof(*defnrToPC) * defbase_tot, "defnrToBone"); defnrToPCIndex = MEM_callocN(sizeof(*defnrToPCIndex) * defbase_tot, "defnrToIndex"); + /* TODO(sergey): Some considerations here: + * + * - Make it more generic function, maybe even keep together with chanhash. + * - Check whether keeping this consistent across frames gives speedup. + * - Don't use hash for small armatures. + */ + GHash *idx_hash = BLI_ghash_ptr_new("pose channel index by name"); + int pchan_index = 0; + for (pchan = armOb->pose->chanbase.first; pchan != NULL; pchan = pchan->next, ++pchan_index) { + BLI_ghash_insert(idx_hash, pchan, SET_INT_IN_POINTER(pchan_index)); + } for (i = 0, dg = target->defbase.first; dg; i++, dg = dg->next) { defnrToPC[i] = BKE_pose_channel_find_name(armOb->pose, dg->name); /* exclude non-deforming bones */ @@ -1045,10 +1056,11 @@ void armature_deform_verts(Object *armOb, Object *target, DerivedMesh *dm, float defnrToPC[i] = NULL; } else { - defnrToPCIndex[i] = BLI_findindex(&armOb->pose->chanbase, defnrToPC[i]); + defnrToPCIndex[i] = GET_INT_FROM_POINTER(BLI_ghash_lookup(idx_hash, defnrToPC[i])); } } } + BLI_ghash_free(idx_hash, NULL, NULL); } } } diff --git a/source/blender/blenkernel/intern/blender.c b/source/blender/blenkernel/intern/blender.c index a8670395fc4..ceb641073e0 100644 --- a/source/blender/blenkernel/intern/blender.c +++ b/source/blender/blenkernel/intern/blender.c @@ -100,6 +100,18 @@ void BKE_blender_free(void) free_nodesystem(); } +void BKE_blender_version_string(char *version_str, size_t maxncpy, short version, short subversion, bool v_prefix, bool include_subversion) +{ + const char *prefix = v_prefix ? "v" : ""; + + if (include_subversion && subversion > 0) { + BLI_snprintf(version_str, maxncpy, "%s%d.%02d.%d", prefix, version / 100, version % 100, subversion); + } + else { + BLI_snprintf(version_str, maxncpy, "%s%d.%02d", prefix, version / 100, version % 100); + } +} + void BKE_blender_globals_init(void) { memset(&G, 0, sizeof(Global)); @@ -110,10 +122,7 @@ void BKE_blender_globals_init(void) strcpy(G.ima, "//"); - if (BLENDER_SUBVERSION) - BLI_snprintf(versionstr, sizeof(versionstr), "v%d.%02d.%d", BLENDER_VERSION / 100, BLENDER_VERSION % 100, BLENDER_SUBVERSION); - else - BLI_snprintf(versionstr, sizeof(versionstr), "v%d.%02d", BLENDER_VERSION / 100, BLENDER_VERSION % 100); + BKE_blender_version_string(versionstr, sizeof(versionstr), BLENDER_VERSION, BLENDER_SUBVERSION, true, true); #ifndef WITH_PYTHON_SECURITY /* default */ G.f |= G_SCRIPT_AUTOEXEC; @@ -141,20 +150,18 @@ static void keymap_item_free(wmKeyMapItem *kmi) MEM_freeN(kmi->ptr); } -/** - * When loading a new userdef from file, - * or when exiting Blender. - */ -void BKE_blender_userdef_free(void) +void BKE_blender_userdef_set_data(UserDef *userdef) { - wmKeyMap *km; - wmKeyMapItem *kmi; - wmKeyMapDiffItem *kmdi; - bAddon *addon, *addon_next; - uiFont *font; - - for (km = U.user_keymaps.first; km; km = km->next) { - for (kmdi = km->diff_items.first; kmdi; kmdi = kmdi->next) { + /* only here free userdef themes... */ + BKE_blender_userdef_free_data(&U); + U = *userdef; +} + +static void userdef_free_keymaps(UserDef *userdef) +{ + for (wmKeyMap *km = userdef->user_keymaps.first, *km_next; km; km = km_next) { + km_next = km->next; + for (wmKeyMapDiffItem *kmdi = km->diff_items.first; kmdi; kmdi = kmdi->next) { if (kmdi->add_item) { keymap_item_free(kmdi->add_item); MEM_freeN(kmdi->add_item); @@ -165,14 +172,21 @@ void BKE_blender_userdef_free(void) } } - for (kmi = km->items.first; kmi; kmi = kmi->next) + for (wmKeyMapItem *kmi = km->items.first; kmi; kmi = kmi->next) { keymap_item_free(kmi); + } BLI_freelistN(&km->diff_items); BLI_freelistN(&km->items); + + MEM_freeN(km); } - - for (addon = U.addons.first; addon; addon = addon_next) { + BLI_listbase_clear(&userdef->user_keymaps); +} + +static void userdef_free_addons(UserDef *userdef) +{ + for (bAddon *addon = userdef->addons.first, *addon_next; addon; addon = addon_next) { addon_next = addon->next; if (addon->prop) { IDP_FreeProperty(addon->prop); @@ -180,19 +194,35 @@ void BKE_blender_userdef_free(void) } MEM_freeN(addon); } + BLI_listbase_clear(&userdef->addons); +} - for (font = U.uifonts.first; font; font = font->next) { +/** + * When loading a new userdef from file, + * or when exiting Blender. + */ +void BKE_blender_userdef_free_data(UserDef *userdef) +{ +#define U _invalid_access_ /* ensure no accidental global access */ +#ifdef U /* quiet warning */ +#endif + + userdef_free_keymaps(userdef); + userdef_free_addons(userdef); + + for (uiFont *font = userdef->uifonts.first; font; font = font->next) { BLF_unload_id(font->blf_id); } BLF_default_set(-1); - BLI_freelistN(&U.autoexec_paths); + BLI_freelistN(&userdef->autoexec_paths); + + BLI_freelistN(&userdef->uistyles); + BLI_freelistN(&userdef->uifonts); + BLI_freelistN(&userdef->themes); - BLI_freelistN(&U.uistyles); - BLI_freelistN(&U.uifonts); - BLI_freelistN(&U.themes); - BLI_freelistN(&U.user_keymaps); +#undef U } /** @@ -208,6 +238,44 @@ void BKE_blender_userdef_refresh(void) } +/** + * Write U from userdef. + * This function defines which settings a template will override for the user preferences. + */ +void BKE_blender_userdef_set_app_template(UserDef *userdef) +{ + /* TODO: + * - keymaps + * - various minor settings (add as needed). + */ + +#define LIST_OVERRIDE(id) { \ + BLI_freelistN(&U.id); \ + BLI_movelisttolist(&U.id, &userdef->id); \ +} ((void)0) + +#define MEMCPY_OVERRIDE(id) \ + memcpy(U.id, userdef->id, sizeof(U.id)); + + /* for some types we need custom free functions */ + userdef_free_addons(&U); + userdef_free_keymaps(&U); + + LIST_OVERRIDE(uistyles); + LIST_OVERRIDE(uifonts); + LIST_OVERRIDE(themes); + LIST_OVERRIDE(addons); + LIST_OVERRIDE(user_keymaps); + + MEMCPY_OVERRIDE(light); + + MEMCPY_OVERRIDE(font_path_ui); + MEMCPY_OVERRIDE(font_path_ui_mono); + +#undef LIST_OVERRIDE +#undef MEMCPY_OVERRIDE +} + /* ***************** testing for break ************* */ static void (*blender_test_break_cb)(void) = NULL; @@ -234,7 +302,7 @@ int BKE_blender_test_break(void) * \note Don't use MEM_mallocN so functions can be registered at any time. * \{ */ -struct AtExitData { +static struct AtExitData { struct AtExitData *next; void (*func)(void *user_data); diff --git a/source/blender/blenkernel/intern/blender_undo.c b/source/blender/blenkernel/intern/blender_undo.c index bc98d6f6805..e3a5edb2049 100644 --- a/source/blender/blenkernel/intern/blender_undo.c +++ b/source/blender/blenkernel/intern/blender_undo.c @@ -108,9 +108,9 @@ static int read_undosave(bContext *C, UndoElem *uel) G.fileflags |= G_FILE_NO_UI; if (UNDO_DISK) - success = (BKE_blendfile_read(C, uel->str, NULL) != BKE_BLENDFILE_READ_FAIL); + success = (BKE_blendfile_read(C, uel->str, NULL, 0) != BKE_BLENDFILE_READ_FAIL); else - success = BKE_blendfile_read_from_memfile(C, &uel->memfile, NULL); + success = BKE_blendfile_read_from_memfile(C, &uel->memfile, NULL, 0); /* restore */ BLI_strncpy(G.main->name, mainstr, sizeof(G.main->name)); /* restore */ @@ -389,7 +389,7 @@ bool BKE_undo_save_file(const char *filename) Main *BKE_undo_get_main(Scene **r_scene) { Main *mainp = NULL; - BlendFileData *bfd = BLO_read_from_memfile(G.main, G.main->name, &curundo->memfile, NULL); + BlendFileData *bfd = BLO_read_from_memfile(G.main, G.main->name, &curundo->memfile, NULL, BLO_READ_SKIP_NONE); if (bfd) { mainp = bfd->main; diff --git a/source/blender/blenkernel/intern/blendfile.c b/source/blender/blenkernel/intern/blendfile.c index 54f709a1e5b..a521d671ea4 100644 --- a/source/blender/blenkernel/intern/blendfile.c +++ b/source/blender/blenkernel/intern/blendfile.c @@ -227,7 +227,7 @@ static void setup_app_data( if (bfd->user) { /* only here free userdef themes... */ - BKE_blender_userdef_free(); + BKE_blender_userdef_free_data(&U); U = *bfd->user; @@ -349,17 +349,23 @@ static int handle_subversion_warning(Main *main, ReportList *reports) return 1; } -int BKE_blendfile_read(bContext *C, const char *filepath, ReportList *reports) +int BKE_blendfile_read( + bContext *C, const char *filepath, + ReportList *reports, int skip_flags) { BlendFileData *bfd; int retval = BKE_BLENDFILE_READ_OK; - if (strstr(filepath, BLENDER_STARTUP_FILE) == NULL) /* don't print user-pref loading */ - printf("read blend: %s\n", filepath); + /* don't print user-pref loading */ + if (strstr(filepath, BLENDER_STARTUP_FILE) == NULL) { + printf("Read blend: %s\n", filepath); + } - bfd = BLO_read_from_file(filepath, reports); + bfd = BLO_read_from_file(filepath, reports, skip_flags); if (bfd) { - if (bfd->user) retval = BKE_BLENDFILE_READ_OK_USERPREFS; + if (bfd->user) { + retval = BKE_BLENDFILE_READ_OK_USERPREFS; + } if (0 == handle_subversion_warning(bfd->main, reports)) { BKE_main_free(bfd->main); @@ -379,11 +385,11 @@ int BKE_blendfile_read(bContext *C, const char *filepath, ReportList *reports) bool BKE_blendfile_read_from_memory( bContext *C, const void *filebuf, int filelength, - ReportList *reports, bool update_defaults) + ReportList *reports, int skip_flags, bool update_defaults) { BlendFileData *bfd; - bfd = BLO_read_from_memory(filebuf, filelength, reports); + bfd = BLO_read_from_memory(filebuf, filelength, reports, skip_flags); if (bfd) { if (update_defaults) BLO_update_defaults_startup_blend(bfd->main); @@ -399,11 +405,11 @@ bool BKE_blendfile_read_from_memory( /* memfile is the undo buffer */ bool BKE_blendfile_read_from_memfile( bContext *C, struct MemFile *memfile, - ReportList *reports) + ReportList *reports, int skip_flags) { BlendFileData *bfd; - bfd = BLO_read_from_memfile(CTX_data_main(C), G.main->name, memfile, reports); + bfd = BLO_read_from_memfile(CTX_data_main(C), G.main->name, memfile, reports, skip_flags); if (bfd) { /* remove the unused screens and wm */ while (bfd->main->wm.first) @@ -420,32 +426,76 @@ bool BKE_blendfile_read_from_memfile( return (bfd != NULL); } +/** + * Utility to make a file 'empty' used for startup to optionally give an empty file. + * Handy for tests. + */ +void BKE_blendfile_read_make_empty(bContext *C) +{ + Main *bmain = CTX_data_main(C); + + ListBase *lbarray[MAX_LIBARRAY]; + ID *id; + int a; + + a = set_listbasepointers(bmain, lbarray); + while (a--) { + id = lbarray[a]->first; + if (id != NULL) { + if (ELEM(GS(id->name), ID_SCE, ID_SCR, ID_WM)) { + continue; + } + while ((id = lbarray[a]->first)) { + BKE_libblock_delete(bmain, id); + } + } + } +} + /* only read the userdef from a .blend */ -int BKE_blendfile_read_userdef(const char *filepath, ReportList *reports) +UserDef *BKE_blendfile_userdef_read(const char *filepath, ReportList *reports) { BlendFileData *bfd; - int retval = BKE_BLENDFILE_READ_FAIL; + UserDef *userdef = NULL; - bfd = BLO_read_from_file(filepath, reports); + bfd = BLO_read_from_file(filepath, reports, BLO_READ_SKIP_ALL & ~BLO_READ_SKIP_USERDEF); if (bfd) { if (bfd->user) { - retval = BKE_BLENDFILE_READ_OK_USERPREFS; + userdef = bfd->user; + } + BKE_main_free(bfd->main); + MEM_freeN(bfd); + } - /* only here free userdef themes... */ - BKE_blender_userdef_free(); + return userdef; +} - U = *bfd->user; - MEM_freeN(bfd->user); + +UserDef *BKE_blendfile_userdef_read_from_memory( + const void *filebuf, int filelength, + ReportList *reports) +{ + BlendFileData *bfd; + UserDef *userdef = NULL; + + bfd = BLO_read_from_memory(filebuf, filelength, reports, BLO_READ_SKIP_ALL & ~BLO_READ_SKIP_USERDEF); + if (bfd) { + if (bfd->user) { + userdef = bfd->user; } BKE_main_free(bfd->main); MEM_freeN(bfd); } + else { + BKE_reports_prepend(reports, "Loading failed: "); + } - return retval; + return userdef; } + /* only write the userdef in a .blend */ -int BKE_blendfile_write_userdef(const char *filepath, ReportList *reports) +int BKE_blendfile_userdef_write(const char *filepath, ReportList *reports) { Main *mainb = MEM_callocN(sizeof(Main), "empty main"); int retval = 0; diff --git a/source/blender/blenkernel/intern/bvhutils.c b/source/blender/blenkernel/intern/bvhutils.c index c1dd3c0a5d8..c0e4ef37a93 100644 --- a/source/blender/blenkernel/intern/bvhutils.c +++ b/source/blender/blenkernel/intern/bvhutils.c @@ -393,7 +393,6 @@ static BVHTree *bvhtree_from_editmesh_verts_create_tree( BMEditMesh *em, const int verts_num, const BLI_bitmap *verts_mask, int verts_num_active) { - BVHTree *tree = NULL; BM_mesh_elem_table_ensure(em->bm, BM_VERT); if (verts_mask) { BLI_assert(IN_RANGE_INCL(verts_num_active, 0, verts_num)); @@ -402,14 +401,15 @@ static BVHTree *bvhtree_from_editmesh_verts_create_tree( verts_num_active = verts_num; } - tree = BLI_bvhtree_new(verts_num_active, epsilon, tree_type, axis); + BVHTree *tree = BLI_bvhtree_new(verts_num_active, epsilon, tree_type, axis); if (tree) { for (int i = 0; i < verts_num; i++) { - if (!verts_mask || BLI_BITMAP_TEST_BOOL(verts_mask, i)) { - BMVert *eve = BM_vert_at_index(em->bm, i); - BLI_bvhtree_insert(tree, i, eve->co, 1); + if (verts_mask && !BLI_BITMAP_TEST_BOOL(verts_mask, i)) { + continue; } + BMVert *eve = BM_vert_at_index(em->bm, i); + BLI_bvhtree_insert(tree, i, eve->co, 1); } BLI_assert(BLI_bvhtree_get_size(tree) == verts_num_active); BLI_bvhtree_balance(tree); @@ -420,31 +420,28 @@ static BVHTree *bvhtree_from_editmesh_verts_create_tree( static BVHTree *bvhtree_from_mesh_verts_create_tree( float epsilon, int tree_type, int axis, - MVert *vert, const int verts_num, + const MVert *vert, const int verts_num, const BLI_bitmap *verts_mask, int verts_num_active) { - BVHTree *tree = NULL; - int i; - if (vert) { - if (verts_mask) { - BLI_assert(IN_RANGE_INCL(verts_num_active, 0, verts_num)); - } - else { - verts_num_active = verts_num; - } + BLI_assert(vert != NULL); + if (verts_mask) { + BLI_assert(IN_RANGE_INCL(verts_num_active, 0, verts_num)); + } + else { + verts_num_active = verts_num; + } - tree = BLI_bvhtree_new(verts_num_active, epsilon, tree_type, axis); + BVHTree *tree = BLI_bvhtree_new(verts_num_active, epsilon, tree_type, axis); - if (tree) { - for (i = 0; i < verts_num; i++) { - if (verts_mask && !BLI_BITMAP_TEST_BOOL(verts_mask, i)) { - continue; - } - BLI_bvhtree_insert(tree, i, vert[i].co, 1); + if (tree) { + for (int i = 0; i < verts_num; i++) { + if (verts_mask && !BLI_BITMAP_TEST_BOOL(verts_mask, i)) { + continue; } - BLI_assert(BLI_bvhtree_get_size(tree) == verts_num_active); - BLI_bvhtree_balance(tree); + BLI_bvhtree_insert(tree, i, vert[i].co, 1); } + BLI_assert(BLI_bvhtree_get_size(tree) == verts_num_active); + BLI_bvhtree_balance(tree); } return tree; @@ -452,31 +449,23 @@ static BVHTree *bvhtree_from_mesh_verts_create_tree( static void bvhtree_from_mesh_verts_setup_data( BVHTreeFromMesh *data, BVHTree *tree, const bool is_cached, float epsilon, - MVert *vert, const bool vert_allocated) + const MVert *vert, const bool vert_allocated) { memset(data, 0, sizeof(*data)); - if (tree) { - data->tree = tree; - data->cached = is_cached; + data->tree = tree; + data->cached = is_cached; - /* a NULL nearest callback works fine - * remember the min distance to point is the same as the min distance to BV of point */ - data->nearest_callback = NULL; - data->raycast_callback = mesh_verts_spherecast; - data->nearest_to_ray_callback = NULL; + /* a NULL nearest callback works fine + * remember the min distance to point is the same as the min distance to BV of point */ + data->nearest_callback = NULL; + data->raycast_callback = mesh_verts_spherecast; - data->vert = vert; - data->vert_allocated = vert_allocated; - //data->face = DM_get_tessface_array(dm, &data->face_allocated); /* XXX WHY???? */ + data->vert = vert; + data->vert_allocated = vert_allocated; + //data->face = DM_get_tessface_array(dm, &data->face_allocated); /* XXX WHY???? */ - data->sphere_radius = epsilon; - } - else { - if (vert_allocated) { - MEM_freeN(vert); - } - } + data->sphere_radius = epsilon; } /* Builds a bvh tree where nodes are the vertices of the given em */ @@ -485,11 +474,9 @@ BVHTree *bvhtree_from_editmesh_verts_ex( const BLI_bitmap *verts_mask, int verts_num_active, float epsilon, int tree_type, int axis) { - int vert_num = em->bm->totvert; - BVHTree *tree = bvhtree_from_editmesh_verts_create_tree( epsilon, tree_type, axis, - em, vert_num, verts_mask, verts_num_active); + em, em->bm->totvert, verts_mask, verts_num_active); if (tree) { memset(data, 0, sizeof(*data)); @@ -497,11 +484,11 @@ BVHTree *bvhtree_from_editmesh_verts_ex( data->em = em; data->nearest_callback = NULL; data->raycast_callback = editmesh_verts_spherecast; - data->nearest_to_ray_callback = NULL; } return tree; } + BVHTree *bvhtree_from_editmesh_verts( BVHTreeFromEditMesh *data, BMEditMesh *em, float epsilon, int tree_type, int axis) @@ -512,8 +499,8 @@ BVHTree *bvhtree_from_editmesh_verts( epsilon, tree_type, axis); } - -/* Builds a bvh tree where nodes are the vertices of the given dm */ +/* Builds a bvh tree where nodes are the vertices of the given dm + * and stores the BVHTree in dm->bvhCache */ BVHTree *bvhtree_from_mesh_verts( BVHTreeFromMesh *data, DerivedMesh *dm, float epsilon, int tree_type, int axis) @@ -553,10 +540,18 @@ BVHTree *bvhtree_from_mesh_verts( /* printf("BVHTree is already build, using cached tree\n"); */ } - /* Setup BVHTreeFromMesh */ - bvhtree_from_mesh_verts_setup_data(data, tree, true, epsilon, vert, vert_allocated); - - return data->tree; + if (tree) { + /* Setup BVHTreeFromMesh */ + bvhtree_from_mesh_verts_setup_data( + data, tree, true, epsilon, vert, vert_allocated); + } + else { + if (vert_allocated) { + MEM_freeN(vert); + } + memset(data, 0, sizeof(*data)); + } + return tree; } /** @@ -566,7 +561,7 @@ BVHTree *bvhtree_from_mesh_verts( * \param verts_num_active if >= 0, number of active verts to add to BVH tree (else will be computed from mask). */ BVHTree *bvhtree_from_mesh_verts_ex( - BVHTreeFromMesh *data, MVert *vert, const int verts_num, const bool vert_allocated, + BVHTreeFromMesh *data, const MVert *vert, const int verts_num, const bool vert_allocated, const BLI_bitmap *verts_mask, int verts_num_active, float epsilon, int tree_type, int axis) { @@ -574,9 +569,10 @@ BVHTree *bvhtree_from_mesh_verts_ex( epsilon, tree_type, axis, vert, verts_num, verts_mask, verts_num_active); /* Setup BVHTreeFromMesh */ - bvhtree_from_mesh_verts_setup_data(data, tree, false, epsilon, vert, vert_allocated); + bvhtree_from_mesh_verts_setup_data( + data, tree, false, epsilon, vert, vert_allocated); - return data->tree; + return tree; } /** \} */ @@ -592,8 +588,6 @@ static BVHTree *bvhtree_from_editmesh_edges_create_tree( BMEditMesh *em, const int edges_num, const BLI_bitmap *edges_mask, int edges_num_active) { - BVHTree *tree = NULL; - int i; BM_mesh_elem_table_ensure(em->bm, BM_EDGE); if (edges_mask) { BLI_assert(IN_RANGE_INCL(edges_num_active, 0, edges_num)); @@ -602,9 +596,10 @@ static BVHTree *bvhtree_from_editmesh_edges_create_tree( edges_num_active = edges_num; } - tree = BLI_bvhtree_new(edges_num_active, epsilon, tree_type, axis); + BVHTree *tree = BLI_bvhtree_new(edges_num_active, epsilon, tree_type, axis); if (tree) { + int i; BMIter iter; BMEdge *eed; BM_ITER_MESH_INDEX (eed, &iter, em->bm, BM_EDGES_OF_MESH, i) { @@ -624,6 +619,62 @@ static BVHTree *bvhtree_from_editmesh_edges_create_tree( return tree; } +static BVHTree *bvhtree_from_mesh_edges_create_tree( + const MVert *vert, const MEdge *edge, const int edge_num, + const BLI_bitmap *edges_mask, int edges_num_active, + float epsilon, int tree_type, int axis) +{ + if (edges_mask) { + BLI_assert(IN_RANGE_INCL(edges_num_active, 0, edge_num)); + } + else { + edges_num_active = edge_num; + } + BLI_assert(vert != NULL); + BLI_assert(edge != NULL); + + /* Create a bvh-tree of the given target */ + BVHTree *tree = BLI_bvhtree_new(edges_num_active, epsilon, tree_type, axis); + if (tree) { + for (int i = 0; i < edge_num; i++) { + if (edges_mask && !BLI_BITMAP_TEST_BOOL(edges_mask, i)) { + continue; + } + float co[2][3]; + copy_v3_v3(co[0], vert[edge[i].v1].co); + copy_v3_v3(co[1], vert[edge[i].v2].co); + + BLI_bvhtree_insert(tree, i, co[0], 2); + } + BLI_bvhtree_balance(tree); + } + + return tree; +} + +static void bvhtree_from_mesh_edges_setup_data( + BVHTreeFromMesh *data, BVHTree *tree, + const bool is_cached, float epsilon, + const MVert *vert, const bool vert_allocated, + const MEdge *edge, const bool edge_allocated) +{ + memset(data, 0, sizeof(*data)); + + data->tree = tree; + + data->cached = is_cached; + + data->nearest_callback = mesh_edges_nearest_point; + data->raycast_callback = mesh_edges_spherecast; + + data->vert = vert; + data->vert_allocated = vert_allocated; + data->edge = edge; + data->edge_allocated = edge_allocated; + + data->sphere_radius = epsilon; +} + /* Builds a bvh tree where nodes are the edges of the given em */ BVHTree *bvhtree_from_editmesh_edges_ex( BVHTreeFromEditMesh *data, BMEditMesh *em, @@ -642,12 +693,11 @@ BVHTree *bvhtree_from_editmesh_edges_ex( data->em = em; data->nearest_callback = NULL; /* TODO */ data->raycast_callback = NULL; /* TODO */ - /* TODO: not urgent however since users currently define own callbacks */ - data->nearest_to_ray_callback = NULL; } return tree; } + BVHTree *bvhtree_from_editmesh_edges( BVHTreeFromEditMesh *data, BMEditMesh *em, float epsilon, int tree_type, int axis) @@ -680,27 +730,13 @@ BVHTree *bvhtree_from_mesh_edges( BLI_rw_mutex_lock(&cache_rwlock, THREAD_LOCK_WRITE); tree = bvhcache_find(dm->bvhCache, BVHTREE_FROM_EDGES); if (tree == NULL) { - int i; - int numEdges = dm->getNumEdges(dm); - - if (vert != NULL && edge != NULL) { - /* Create a bvh-tree of the given target */ - tree = BLI_bvhtree_new(numEdges, epsilon, tree_type, axis); - if (tree != NULL) { - for (i = 0; i < numEdges; i++) { - float co[2][3]; - copy_v3_v3(co[0], vert[edge[i].v1].co); - copy_v3_v3(co[1], vert[edge[i].v2].co); - - BLI_bvhtree_insert(tree, i, co[0], 2); - } - BLI_bvhtree_balance(tree); + tree = bvhtree_from_mesh_edges_create_tree( + vert, edge, dm->getNumEdges(dm), + NULL, -1, epsilon, tree_type, axis); - /* Save on cache for later use */ - /* printf("BVHTree built and saved on cache\n"); */ - bvhcache_insert(&dm->bvhCache, tree, BVHTREE_FROM_EDGES); - } - } + /* Save on cache for later use */ + /* printf("BVHTree built and saved on cache\n"); */ + bvhcache_insert(&dm->bvhCache, tree, BVHTREE_FROM_EDGES); } BLI_rw_mutex_unlock(&cache_rwlock); } @@ -708,24 +744,10 @@ BVHTree *bvhtree_from_mesh_edges( /* printf("BVHTree is already build, using cached tree\n"); */ } - - /* Setup BVHTreeFromMesh */ - memset(data, 0, sizeof(*data)); - data->tree = tree; - - if (data->tree) { - data->cached = true; - - data->nearest_callback = mesh_edges_nearest_point; - data->raycast_callback = mesh_edges_spherecast; - data->nearest_to_ray_callback = NULL; - - data->vert = vert; - data->vert_allocated = vert_allocated; - data->edge = edge; - data->edge_allocated = edge_allocated; - - data->sphere_radius = epsilon; + if (tree) { + /* Setup BVHTreeFromMesh */ + bvhtree_from_mesh_edges_setup_data( + data, tree, true, epsilon, vert, vert_allocated, edge, edge_allocated); } else { if (vert_allocated) { @@ -734,8 +756,33 @@ BVHTree *bvhtree_from_mesh_edges( if (edge_allocated) { MEM_freeN(edge); } + memset(data, 0, sizeof(*data)); } - return data->tree; + return tree; +} + +/** + * Builds a bvh tree where nodes are the given edges . + * \param vert/edge_allocated if true, elem freeing will be done when freeing data. + * \param edges_mask if not null, true elements give which vert to add to BVH tree. + * \param edges_num_active if >= 0, number of active edges to add to BVH tree (else will be computed from mask). + */ +BVHTree *bvhtree_from_mesh_edges_ex( + BVHTreeFromMesh *data, + const MVert *vert, const bool vert_allocated, + const MEdge *edge, const int edges_num, const bool edge_allocated, + const BLI_bitmap *edges_mask, int edges_num_active, + float epsilon, int tree_type, int axis) +{ + BVHTree *tree = bvhtree_from_mesh_edges_create_tree( + vert, edge, edges_num, edges_mask, edges_num_active, + epsilon, tree_type, axis); + + /* Setup BVHTreeFromMesh */ + bvhtree_from_mesh_edges_setup_data( + data, tree, false, epsilon, vert, vert_allocated, edge, edge_allocated); + + return tree; } /** \} */ @@ -748,7 +795,7 @@ BVHTree *bvhtree_from_mesh_edges( static BVHTree *bvhtree_from_mesh_faces_create_tree( float epsilon, int tree_type, int axis, - MVert *vert, MFace *face, const int faces_num, + const MVert *vert, const MFace *face, const int faces_num, const BLI_bitmap *faces_mask, int faces_num_active) { BVHTree *tree = NULL; @@ -792,34 +839,23 @@ static BVHTree *bvhtree_from_mesh_faces_create_tree( static void bvhtree_from_mesh_faces_setup_data( BVHTreeFromMesh *data, BVHTree *tree, const bool is_cached, float epsilon, - MVert *vert, const bool vert_allocated, - MFace *face, const bool face_allocated) + const MVert *vert, const bool vert_allocated, + const MFace *face, const bool face_allocated) { memset(data, 0, sizeof(*data)); - if (tree) { - data->tree = tree; - data->cached = is_cached; + data->tree = tree; + data->cached = is_cached; - data->nearest_callback = mesh_faces_nearest_point; - data->raycast_callback = mesh_faces_spherecast; - data->nearest_to_ray_callback = NULL; + data->nearest_callback = mesh_faces_nearest_point; + data->raycast_callback = mesh_faces_spherecast; - data->vert = vert; - data->vert_allocated = vert_allocated; - data->face = face; - data->face_allocated = face_allocated; + data->vert = vert; + data->vert_allocated = vert_allocated; + data->face = face; + data->face_allocated = face_allocated; - data->sphere_radius = epsilon; - } - else { - if (vert_allocated) { - MEM_freeN(vert); - } - if (face_allocated) { - MEM_freeN(face); - } - } + data->sphere_radius = epsilon; } /* Builds a bvh tree where nodes are the tesselated faces of the given dm */ @@ -862,10 +898,21 @@ BVHTree *bvhtree_from_mesh_faces( /* printf("BVHTree is already build, using cached tree\n"); */ } - /* Setup BVHTreeFromMesh */ - bvhtree_from_mesh_faces_setup_data(data, tree, true, epsilon, vert, vert_allocated, face, face_allocated); - - return data->tree; + if (tree) { + /* Setup BVHTreeFromMesh */ + bvhtree_from_mesh_faces_setup_data( + data, tree, true, epsilon, vert, vert_allocated, face, face_allocated); + } + else { + if (vert_allocated) { + MEM_freeN(vert); + } + if (face_allocated) { + MEM_freeN(face); + } + memset(data, 0, sizeof(*data)); + } + return tree; } /** @@ -876,8 +923,8 @@ BVHTree *bvhtree_from_mesh_faces( * \param numFaces_active if >= 0, number of active faces to add to BVH tree (else will be computed from mask). */ BVHTree *bvhtree_from_mesh_faces_ex( - BVHTreeFromMesh *data, MVert *vert, const bool vert_allocated, - MFace *face, const int numFaces, const bool face_allocated, + BVHTreeFromMesh *data, const MVert *vert, const bool vert_allocated, + const MFace *face, const int numFaces, const bool face_allocated, const BLI_bitmap *faces_mask, int faces_num_active, float epsilon, int tree_type, int axis) { @@ -887,9 +934,10 @@ BVHTree *bvhtree_from_mesh_faces_ex( faces_mask, faces_num_active); /* Setup BVHTreeFromMesh */ - bvhtree_from_mesh_faces_setup_data(data, tree, false, epsilon, vert, vert_allocated, face, face_allocated); + bvhtree_from_mesh_faces_setup_data( + data, tree, false, epsilon, vert, vert_allocated, face, face_allocated); - return data->tree; + return tree; } /** \} */ @@ -1000,34 +1048,20 @@ static void bvhtree_from_mesh_looptri_setup_data( { memset(data, 0, sizeof(*data)); - if (tree) { - data->tree = tree; - data->cached = is_cached; + data->tree = tree; + data->cached = is_cached; - data->nearest_callback = mesh_looptri_nearest_point; - data->raycast_callback = mesh_looptri_spherecast; - data->nearest_to_ray_callback = NULL; + data->nearest_callback = mesh_looptri_nearest_point; + data->raycast_callback = mesh_looptri_spherecast; - data->vert = vert; - data->vert_allocated = vert_allocated; - data->loop = mloop; - data->loop_allocated = loop_allocated; - data->looptri = looptri; - data->looptri_allocated = looptri_allocated; + data->vert = vert; + data->vert_allocated = vert_allocated; + data->loop = mloop; + data->loop_allocated = loop_allocated; + data->looptri = looptri; + data->looptri_allocated = looptri_allocated; - data->sphere_radius = epsilon; - } - else { - if (vert_allocated) { - MEM_freeN((void *)vert); - } - if (loop_allocated) { - MEM_freeN((void *)mloop); - } - if (looptri_allocated) { - MEM_freeN((void *)looptri); - } - } + data->sphere_radius = epsilon; } /** @@ -1072,7 +1106,6 @@ BVHTree *bvhtree_from_editmesh_looptri_ex( data->tree = tree; data->nearest_callback = editmesh_looptri_nearest_point; data->raycast_callback = editmesh_looptri_spherecast; - data->nearest_to_ray_callback = NULL; data->sphere_radius = 0.0f; data->em = em; data->cached = bvhCache != NULL; @@ -1154,14 +1187,28 @@ BVHTree *bvhtree_from_mesh_looptri( /* printf("BVHTree is already build, using cached tree\n"); */ } - /* Setup BVHTreeFromMesh */ - bvhtree_from_mesh_looptri_setup_data( - data, tree, true, epsilon, - mvert, vert_allocated, - mloop, loop_allocated, - looptri, looptri_allocated); + if (tree) { + /* Setup BVHTreeFromMesh */ + bvhtree_from_mesh_looptri_setup_data( + data, tree, true, epsilon, + mvert, vert_allocated, + mloop, loop_allocated, + looptri, looptri_allocated); + } + else { + if (vert_allocated) { + MEM_freeN(mvert); + } + if (loop_allocated) { + MEM_freeN(mloop); + } + if (looptri_allocated) { + MEM_freeN((void *)looptri); + } + memset(data, 0, sizeof(*data)); + } - return data->tree; + return tree; } BVHTree *bvhtree_from_mesh_looptri_ex( @@ -1184,7 +1231,7 @@ BVHTree *bvhtree_from_mesh_looptri_ex( mloop, loop_allocated, looptri, looptri_allocated); - return data->tree; + return tree; } /** \} */ @@ -1204,29 +1251,27 @@ void free_bvhtree_from_editmesh(struct BVHTreeFromEditMesh *data) /* Frees data allocated by a call to bvhtree_from_mesh_*. */ void free_bvhtree_from_mesh(struct BVHTreeFromMesh *data) { - if (data->tree) { - if (!data->cached) { - BLI_bvhtree_free(data->tree); - } - - if (data->vert_allocated) { - MEM_freeN((void *)data->vert); - } - if (data->edge_allocated) { - MEM_freeN((void *)data->edge); - } - if (data->face_allocated) { - MEM_freeN((void *)data->face); - } - if (data->loop_allocated) { - MEM_freeN((void *)data->loop); - } - if (data->looptri_allocated) { - MEM_freeN((void *)data->looptri); - } + if (data->tree && !data->cached) { + BLI_bvhtree_free(data->tree); + } - memset(data, 0, sizeof(*data)); + if (data->vert_allocated) { + MEM_freeN((void *)data->vert); + } + if (data->edge_allocated) { + MEM_freeN((void *)data->edge); } + if (data->face_allocated) { + MEM_freeN((void *)data->face); + } + if (data->loop_allocated) { + MEM_freeN((void *)data->loop); + } + if (data->looptri_allocated) { + MEM_freeN((void *)data->looptri); + } + + memset(data, 0, sizeof(*data)); } diff --git a/source/blender/blenkernel/intern/cdderivedmesh.c b/source/blender/blenkernel/intern/cdderivedmesh.c index fc3e358cb25..7042b46330b 100644 --- a/source/blender/blenkernel/intern/cdderivedmesh.c +++ b/source/blender/blenkernel/intern/cdderivedmesh.c @@ -2398,13 +2398,16 @@ DerivedMesh *CDDM_from_editbmesh(BMEditMesh *em, const bool use_mdisps, const bo use_tessface, em->tottri, (const BMLoop *(*)[3])em->looptris); } -static DerivedMesh *cddm_copy_ex(DerivedMesh *source, int faces_from_tessfaces) +static DerivedMesh *cddm_copy_ex(DerivedMesh *source, + const bool need_tessface_data, + const bool faces_from_tessfaces) { + const bool copy_tessface_data = (faces_from_tessfaces || need_tessface_data); CDDerivedMesh *cddm = cdDM_create("CDDM_copy cddm"); DerivedMesh *dm = &cddm->dm; int numVerts = source->numVertData; int numEdges = source->numEdgeData; - int numTessFaces = source->numTessFaceData; + int numTessFaces = copy_tessface_data ? source->numTessFaceData : 0; int numLoops = source->numLoopData; int numPolys = source->numPolyData; @@ -2414,20 +2417,29 @@ static DerivedMesh *cddm_copy_ex(DerivedMesh *source, int faces_from_tessfaces) source->getVertDataArray(source, CD_ORIGINDEX); source->getEdgeDataArray(source, CD_ORIGINDEX); source->getPolyDataArray(source, CD_ORIGINDEX); + if (copy_tessface_data) { + source->getTessFaceDataArray(source, CD_ORIGINDEX); + } /* this initializes dm, and copies all non mvert/medge/mface layers */ - DM_from_template(dm, source, DM_TYPE_CDDM, numVerts, numEdges, - faces_from_tessfaces ? numTessFaces : 0, + DM_from_template(dm, source, DM_TYPE_CDDM, numVerts, numEdges, numTessFaces, numLoops, numPolys); dm->deformedOnly = source->deformedOnly; dm->cd_flag = source->cd_flag; dm->dirty = source->dirty; - /* Tessellation data is never copied, so tag it here. */ - dm->dirty |= DM_DIRTY_TESS_CDLAYERS; + /* Tessellation data is never copied, so tag it here. + * Only tag dirty layers if we really ignored tessellation faces. + */ + if (!copy_tessface_data) { + dm->dirty |= DM_DIRTY_TESS_CDLAYERS; + } CustomData_copy_data(&source->vertData, &dm->vertData, 0, 0, numVerts); CustomData_copy_data(&source->edgeData, &dm->edgeData, 0, 0, numEdges); + if (copy_tessface_data) { + CustomData_copy_data(&source->faceData, &dm->faceData, 0, 0, numTessFaces); + } /* now add mvert/medge/mface layers */ cddm->mvert = source->dupVertArray(source); @@ -2435,17 +2447,16 @@ static DerivedMesh *cddm_copy_ex(DerivedMesh *source, int faces_from_tessfaces) CustomData_add_layer(&dm->vertData, CD_MVERT, CD_ASSIGN, cddm->mvert, numVerts); CustomData_add_layer(&dm->edgeData, CD_MEDGE, CD_ASSIGN, cddm->medge, numEdges); - - if (!faces_from_tessfaces) { - DM_DupPolys(source, dm); - } - else { - source->getTessFaceDataArray(source, CD_ORIGINDEX); - CustomData_copy_data(&source->faceData, &dm->faceData, 0, 0, numTessFaces); + if (faces_from_tessfaces || copy_tessface_data) { cddm->mface = source->dupTessFaceArray(source); CustomData_add_layer(&dm->faceData, CD_MFACE, CD_ASSIGN, cddm->mface, numTessFaces); + } + if (!faces_from_tessfaces) { + DM_DupPolys(source, dm); + } + else { CDDM_tessfaces_to_faces(dm); } @@ -2457,12 +2468,17 @@ static DerivedMesh *cddm_copy_ex(DerivedMesh *source, int faces_from_tessfaces) DerivedMesh *CDDM_copy(DerivedMesh *source) { - return cddm_copy_ex(source, 0); + return cddm_copy_ex(source, false, false); } DerivedMesh *CDDM_copy_from_tessface(DerivedMesh *source) { - return cddm_copy_ex(source, 1); + return cddm_copy_ex(source, false, true); +} + +DerivedMesh *CDDM_copy_with_tessface(DerivedMesh *source) +{ + return cddm_copy_ex(source, true, false); } /* note, the CD_ORIGINDEX layers are all 0, so if there is a direct diff --git a/source/blender/blenkernel/intern/cloth.c b/source/blender/blenkernel/intern/cloth.c index 28ef3f6f248..ee0fde1ea61 100644 --- a/source/blender/blenkernel/intern/cloth.c +++ b/source/blender/blenkernel/intern/cloth.c @@ -869,12 +869,6 @@ static int cloth_from_object(Object *ob, ClothModifierData *clmd, DerivedMesh *d return 0; } - for ( i = 0; i < dm->getNumVerts(dm); i++) { - if ((!(cloth->verts[i].flags & CLOTH_VERT_FLAG_PINNED)) && (cloth->verts[i].goal > ALMOST_ZERO)) { - cloth_add_spring (clmd, i, i, 0.0, CLOTH_SPRING_TYPE_GOAL); - } - } - // init our solver BPH_cloth_solver_init(ob, clmd); @@ -944,37 +938,6 @@ BLI_INLINE void spring_verts_ordered_set(ClothSpring *spring, int v0, int v1) } } -// be careful: implicit solver has to be resettet when using this one! -// --> only for implicit handling of this spring! -int cloth_add_spring(ClothModifierData *clmd, unsigned int indexA, unsigned int indexB, float restlength, int spring_type) -{ - Cloth *cloth = clmd->clothObject; - ClothSpring *spring = NULL; - - if (cloth) { - // TODO: look if this spring is already there - - spring = (ClothSpring *)MEM_callocN ( sizeof ( ClothSpring ), "cloth spring" ); - - if (!spring) - return 0; - - spring->ij = indexA; - spring->kl = indexB; - spring->restlen = restlength; - spring->type = spring_type; - spring->flags = 0; - spring->stiffness = 0; - - cloth->numsprings++; - - BLI_linklist_prepend ( &cloth->springs, spring ); - - return 1; - } - return 0; -} - static void cloth_free_edgelist(LinkNodePair *edgelist, unsigned int mvert_num) { if (edgelist) { diff --git a/source/blender/blenkernel/intern/colortools.c b/source/blender/blenkernel/intern/colortools.c index 4f3ffed41bc..b3a0895d063 100644 --- a/source/blender/blenkernel/intern/colortools.c +++ b/source/blender/blenkernel/intern/colortools.c @@ -508,7 +508,7 @@ static void calchandle_curvemap( if ((bezt->h2 == HD_AUTO_ANIM) && next && prev) { /* keep horizontal if extrema */ const float ydiff1 = prev->vec[1][1] - bezt->vec[1][1]; const float ydiff2 = next->vec[1][1] - bezt->vec[1][1]; - if ((ydiff1 <= 0.0f && ydiff2 <= 0.0f)|| + if ((ydiff1 <= 0.0f && ydiff2 <= 0.0f) || (ydiff1 >= 0.0f && ydiff2 >= 0.0f)) { bezt->vec[2][1] = bezt->vec[1][1]; diff --git a/source/blender/blenkernel/intern/curve.c b/source/blender/blenkernel/intern/curve.c index 90a514781d7..439abb1d593 100644 --- a/source/blender/blenkernel/intern/curve.c +++ b/source/blender/blenkernel/intern/curve.c @@ -89,20 +89,33 @@ void BKE_curve_editfont_free(Curve *cu) } } -void BKE_curve_editNurb_keyIndex_free(EditNurb *editnurb) +static void curve_editNurb_keyIndex_cv_free_cb(void *val) { - if (!editnurb->keyindex) { + CVKeyIndex *index = val; + MEM_freeN(index->orig_cv); + MEM_freeN(val); +} + +void BKE_curve_editNurb_keyIndex_delCV(GHash *keyindex, const void *cv) +{ + BLI_assert(keyindex != NULL); + BLI_ghash_remove(keyindex, cv, NULL, curve_editNurb_keyIndex_cv_free_cb); +} + +void BKE_curve_editNurb_keyIndex_free(GHash **keyindex) +{ + if (!(*keyindex)) { return; } - BLI_ghash_free(editnurb->keyindex, NULL, MEM_freeN); - editnurb->keyindex = NULL; + BLI_ghash_free(*keyindex, NULL, curve_editNurb_keyIndex_cv_free_cb); + *keyindex = NULL; } void BKE_curve_editNurb_free(Curve *cu) { if (cu->editnurb) { BKE_nurbList_free(&cu->editnurb->nurbs); - BKE_curve_editNurb_keyIndex_free(cu->editnurb); + BKE_curve_editNurb_keyIndex_free(&cu->editnurb->keyindex); MEM_freeN(cu->editnurb); cu->editnurb = NULL; } diff --git a/source/blender/blenkernel/intern/customdata.c b/source/blender/blenkernel/intern/customdata.c index 98d37fb07bf..7c3f0ac630d 100644 --- a/source/blender/blenkernel/intern/customdata.c +++ b/source/blender/blenkernel/intern/customdata.c @@ -33,14 +33,10 @@ /** \file blender/blenkernel/intern/customdata.c * \ingroup bke */ - - -#include <math.h> -#include <string.h> -#include <assert.h> #include "MEM_guardedalloc.h" +#include "DNA_customdata_types.h" #include "DNA_meshdata_types.h" #include "DNA_ID.h" @@ -62,20 +58,16 @@ #include "BKE_mesh_remap.h" #include "BKE_multires.h" -#include "data_transfer_intern.h" - #include "bmesh.h" -#include <math.h> -#include <string.h> +/* only for customdata_data_transfer_interp_normal_normals */ +#include "data_transfer_intern.h" /* number of layers to add when growing a CustomData object */ #define CUSTOMDATA_GROW 5 /* ensure typemap size is ok */ -BLI_STATIC_ASSERT(sizeof(((CustomData *)NULL)->typemap) / - sizeof(((CustomData *)NULL)->typemap[0]) == CD_NUMTYPES, - "size mismatch"); +BLI_STATIC_ASSERT(ARRAY_SIZE(((CustomData *)NULL)->typemap) == CD_NUMTYPES, "size mismatch"); /********************* Layer type information **********************/ @@ -805,18 +797,15 @@ static void layerInterp_mloopcol( const float *sub_weights, int count, void *dest) { MLoopCol *mc = dest; - int i; - const float *sub_weight; struct { float a; float r; float g; float b; - } col; - col.a = col.r = col.g = col.b = 0; + } col = {0}; - sub_weight = sub_weights; - for (i = 0; i < count; ++i) { + const float *sub_weight = sub_weights; + for (int i = 0; i < count; ++i) { float weight = weights ? weights[i] : 1; const MLoopCol *src = sources[i]; if (sub_weights) { @@ -833,19 +822,16 @@ static void layerInterp_mloopcol( col.a += src->a * weight; } } - + + /* Subdivide smooth or fractal can cause problems without clamping * although weights should also not cause this situation */ - CLAMP(col.a, 0.0f, 255.0f); - CLAMP(col.r, 0.0f, 255.0f); - CLAMP(col.g, 0.0f, 255.0f); - CLAMP(col.b, 0.0f, 255.0f); - /* delay writing to the destination incase dest is in sources */ - mc->r = (int)col.r; - mc->g = (int)col.g; - mc->b = (int)col.b; - mc->a = (int)col.a; + /* also delay writing to the destination incase dest is in sources */ + mc->r = CLAMPIS(iroundf(col.r), 0, 255); + mc->g = CLAMPIS(iroundf(col.g), 0, 255); + mc->b = CLAMPIS(iroundf(col.b), 0, 255); + mc->a = CLAMPIS(iroundf(col.a), 0, 255); } static int layerMaxNum_mloopcol(void) @@ -1068,15 +1054,10 @@ static void layerInterp_mcol( /* Subdivide smooth or fractal can cause problems without clamping * although weights should also not cause this situation */ - CLAMP(col[j].a, 0.0f, 255.0f); - CLAMP(col[j].r, 0.0f, 255.0f); - CLAMP(col[j].g, 0.0f, 255.0f); - CLAMP(col[j].b, 0.0f, 255.0f); - - mc[j].a = (int)col[j].a; - mc[j].r = (int)col[j].r; - mc[j].g = (int)col[j].g; - mc[j].b = (int)col[j].b; + mc[j].a = CLAMPIS(iroundf(col[j].a), 0, 255); + mc[j].r = CLAMPIS(iroundf(col[j].r), 0, 255); + mc[j].g = CLAMPIS(iroundf(col[j].g), 0, 255); + mc[j].b = CLAMPIS(iroundf(col[j].b), 0, 255); } } @@ -2595,7 +2576,7 @@ bool CustomData_from_bmeshpoly_test(CustomData *fdata, CustomData *pdata, Custom if (!LAYER_CMP(ldata, CD_TANGENT, fdata, CD_TANGENT)) return false; -#undef TEST_RET +#undef LAYER_CMP /* if no layers are on either CustomData's, * then there was nothing to do... */ diff --git a/source/blender/blenkernel/intern/depsgraph.c b/source/blender/blenkernel/intern/depsgraph.c index 294a4ce76b7..678dc92a5f2 100644 --- a/source/blender/blenkernel/intern/depsgraph.c +++ b/source/blender/blenkernel/intern/depsgraph.c @@ -544,10 +544,16 @@ static void build_dag_object(DagForest *dag, DagNode *scenenode, Main *bmain, Sc if (ct->tar->type == OB_MESH) node3->customdata_mask |= CD_MASK_MDEFORMVERT; } - else if (ELEM(con->type, CONSTRAINT_TYPE_FOLLOWPATH, CONSTRAINT_TYPE_CLAMPTO, CONSTRAINT_TYPE_SPLINEIK)) + else if (ELEM(con->type, CONSTRAINT_TYPE_FOLLOWPATH, + CONSTRAINT_TYPE_CLAMPTO, + CONSTRAINT_TYPE_SPLINEIK, + CONSTRAINT_TYPE_SHRINKWRAP)) + { dag_add_relation(dag, node3, node, DAG_RL_DATA_DATA | DAG_RL_OB_DATA, cti->name); - else + } + else { dag_add_relation(dag, node3, node, DAG_RL_OB_DATA, cti->name); + } } } @@ -881,8 +887,12 @@ static void build_dag_object(DagForest *dag, DagNode *scenenode, Main *bmain, Sc if (obt->type == OB_MESH) node2->customdata_mask |= CD_MASK_MDEFORMVERT; } - else + else if (cti->type == CONSTRAINT_TYPE_SHRINKWRAP) { + dag_add_relation(dag, node2, node, DAG_RL_DATA_DATA | DAG_RL_OB_DATA, cti->name); + } + else { dag_add_relation(dag, node2, node, DAG_RL_OB_OB, cti->name); + } } addtoroot = 0; } diff --git a/source/blender/blenkernel/intern/displist.c b/source/blender/blenkernel/intern/displist.c index 49db75a0474..f8a9d57f579 100644 --- a/source/blender/blenkernel/intern/displist.c +++ b/source/blender/blenkernel/intern/displist.c @@ -819,7 +819,7 @@ static void curve_calc_modifiers_pre(Scene *scene, Object *ob, ListBase *nurb, if (editmode) required_mode |= eModifierMode_Editmode; - if (cu->editnurb == NULL) { + if (!editmode) { keyVerts = BKE_key_evaluate_object(ob, &numVerts); if (keyVerts) { diff --git a/source/blender/blenkernel/intern/dynamicpaint.c b/source/blender/blenkernel/intern/dynamicpaint.c index dc9f3b57f1f..c525e75b825 100644 --- a/source/blender/blenkernel/intern/dynamicpaint.c +++ b/source/blender/blenkernel/intern/dynamicpaint.c @@ -2460,8 +2460,7 @@ static void dynamic_paint_find_island_border( const unsigned int *other_loop_idx = mlooptri[lt_index].tri; /* Check edges for match, looping in the same order as the outer loop. */ - for (int j = 0; j < 3; j++) - { + for (int j = 0; j < 3; j++) { const int overt0 = mloop[other_loop_idx[(j + 0)]].v; const int overt1 = mloop[other_loop_idx[(j + 1) % 3]].v; @@ -2525,8 +2524,7 @@ static void dynamic_paint_find_island_border( int final_pixel[2] = { (int)floorf(tgt_pixel[0] * w), (int)floorf(tgt_pixel[1] * h) }; /* If current pixel uv is outside of texture */ - if (final_pixel[0] < 0 || final_pixel[0] >= w || final_pixel[1] < 0 || final_pixel[1] >= h) - { + if (final_pixel[0] < 0 || final_pixel[0] >= w || final_pixel[1] < 0 || final_pixel[1] >= h) { if (bdata->best_index == NOT_FOUND) bdata->best_index = OUT_OF_TEXTURE; diff --git a/source/blender/blenkernel/intern/editderivedmesh.c b/source/blender/blenkernel/intern/editderivedmesh.c index e7c0e69b1cb..2fffa0dea28 100644 --- a/source/blender/blenkernel/intern/editderivedmesh.c +++ b/source/blender/blenkernel/intern/editderivedmesh.c @@ -642,10 +642,10 @@ static void emDM_recalcLoopTri(DerivedMesh *dm) MLoopTri *lt = &mlooptri[i]; ARRAY_SET_ITEMS( - lt->tri, - BM_elem_index_get(ltri[0]), - BM_elem_index_get(ltri[1]), - BM_elem_index_get(ltri[2])); + lt->tri, + BM_elem_index_get(ltri[0]), + BM_elem_index_get(ltri[1]), + BM_elem_index_get(ltri[2])); lt->poly = BM_elem_index_get(ltri[0]->f); } } diff --git a/source/blender/blenkernel/intern/effect.c b/source/blender/blenkernel/intern/effect.c index fe8f5ebdca6..4eee24b378f 100644 --- a/source/blender/blenkernel/intern/effect.c +++ b/source/blender/blenkernel/intern/effect.c @@ -848,6 +848,14 @@ static void do_physical_effector(EffectorCache *eff, EffectorData *efd, Effected break; case PFIELD_FORCE: normalize_v3(force); + if (pd->flag & PFIELD_GRAVITATION){ /* Option: Multiply by 1/distance^2 */ + if (efd->distance < FLT_EPSILON){ + strength = 0.0f; + } + else { + strength *= powf(efd->distance, -2.0f); + } + } mul_v3_fl(force, strength * efd->falloff); break; case PFIELD_VORTEX: diff --git a/source/blender/blenkernel/intern/fcurve.c b/source/blender/blenkernel/intern/fcurve.c index c67a61a5aad..7dbc43e0a32 100644 --- a/source/blender/blenkernel/intern/fcurve.c +++ b/source/blender/blenkernel/intern/fcurve.c @@ -1217,6 +1217,7 @@ bool driver_get_variable_property( return true; } +#if 0 /* Helper function to obtain a pointer to a Pose Channel (for evaluating drivers) */ static bPoseChannel *dtar_get_pchan_ptr(ChannelDriver *driver, DriverTarget *dtar) { @@ -1239,6 +1240,32 @@ static bPoseChannel *dtar_get_pchan_ptr(ChannelDriver *driver, DriverTarget *dta return NULL; } } +#endif + +static short driver_check_valid_targets(ChannelDriver *driver, DriverVar *dvar) +{ + short valid_targets = 0; + + DRIVER_TARGETS_USED_LOOPER(dvar) + { + Object *ob = (Object *)dtar_id_ensure_proxy_from(dtar->id); + + /* check if this target has valid data */ + if ((ob == NULL) || (GS(ob->id.name) != ID_OB)) { + /* invalid target, so will not have enough targets */ + driver->flag |= DRIVER_FLAG_INVALID; + dtar->flag |= DTAR_FLAG_INVALID; + } + else { + /* target seems to be OK now... */ + dtar->flag &= ~DTAR_FLAG_INVALID; + valid_targets++; + } + } + DRIVER_TARGETS_LOOPER_END + + return valid_targets; +} /* ......... */ @@ -1252,62 +1279,54 @@ static float dvar_eval_singleProp(ChannelDriver *driver, DriverVar *dvar) /* evaluate 'rotation difference' driver variable */ static float dvar_eval_rotDiff(ChannelDriver *driver, DriverVar *dvar) { - DriverTarget *dtar1 = &dvar->targets[0]; - DriverTarget *dtar2 = &dvar->targets[1]; - bPoseChannel *pchan, *pchan2; - float q1[4], q2[4], quat[4], angle; - - /* get pose channels, and check if we've got two */ - pchan = dtar_get_pchan_ptr(driver, dtar1); - pchan2 = dtar_get_pchan_ptr(driver, dtar2); - - if (ELEM(NULL, pchan, pchan2)) { - /* disable this driver, since it doesn't work correctly... */ - driver->flag |= DRIVER_FLAG_INVALID; - - /* check what the error was */ - if ((pchan == NULL) && (pchan2 == NULL)) { - if (G.debug & G_DEBUG) { - printf("Driver Evaluation Error: Rotational difference failed - first 2 targets invalid\n"); - } - - dtar1->flag |= DTAR_FLAG_INVALID; - dtar2->flag |= DTAR_FLAG_INVALID; - } - else if (pchan == NULL) { - if (G.debug & G_DEBUG) { - printf("Driver Evaluation Error: Rotational difference failed - first target not valid PoseChannel\n"); - } - - dtar1->flag |= DTAR_FLAG_INVALID; - dtar2->flag &= ~DTAR_FLAG_INVALID; - } - else if (pchan2 == NULL) { - if (G.debug & G_DEBUG) { - printf("Driver Evaluation Error: Rotational difference failed - second target not valid PoseChannel\n"); - } - - dtar1->flag &= ~DTAR_FLAG_INVALID; - dtar2->flag |= DTAR_FLAG_INVALID; + short valid_targets = driver_check_valid_targets(driver, dvar); + + /* make sure we have enough valid targets to use - all or nothing for now... */ + if (driver_check_valid_targets(driver, dvar) != 2) { + if (G.debug & G_DEBUG) { + printf("RotDiff DVar: not enough valid targets (n = %d) (a = %p, b = %p)\n", + valid_targets, dvar->targets[0].id, dvar->targets[1].id); } - - /* stop here... */ return 0.0f; } - else { - dtar1->flag &= ~DTAR_FLAG_INVALID; - dtar2->flag &= ~DTAR_FLAG_INVALID; + + float (*mat[2])[4]; + + /* NOTE: for now, these are all just worldspace */ + for (int i = 0; i < 2; i++) { + /* get pointer to loc values to store in */ + DriverTarget *dtar = &dvar->targets[i]; + Object *ob = (Object *)dtar_id_ensure_proxy_from(dtar->id); + bPoseChannel *pchan; + + /* after the checks above, the targets should be valid here... */ + BLI_assert((ob != NULL) && (GS(ob->id.name) == ID_OB)); + + /* try to get posechannel */ + pchan = BKE_pose_channel_find_name(ob->pose, dtar->pchan_name); + + /* check if object or bone */ + if (pchan) { + /* bone */ + mat[i] = pchan->pose_mat; + } + else { + /* object */ + mat[i] = ob->obmat; + } } - + + float q1[4], q2[4], quat[4], angle; + /* use the final posed locations */ - mat4_to_quat(q1, pchan->pose_mat); - mat4_to_quat(q2, pchan2->pose_mat); - + mat4_to_quat(q1, mat[0]); + mat4_to_quat(q2, mat[1]); + invert_qt_normalized(q1); mul_qt_qtqt(quat, q1, q2); angle = 2.0f * (saacos(quat[0])); angle = ABS(angle); - + return (angle > (float)M_PI) ? (float)((2.0f * (float)M_PI) - angle) : (float)(angle); } @@ -1317,32 +1336,8 @@ static float dvar_eval_locDiff(ChannelDriver *driver, DriverVar *dvar) { float loc1[3] = {0.0f, 0.0f, 0.0f}; float loc2[3] = {0.0f, 0.0f, 0.0f}; - short valid_targets = 0; - - /* Perform two passes - * - * FIRST PASS - to just check that everything works... - * NOTE: we use loops here to reduce code duplication, though in practice, - * there can only be 2 items or else we run into some problems later - */ - DRIVER_TARGETS_USED_LOOPER(dvar) - { - Object *ob = (Object *)dtar_id_ensure_proxy_from(dtar->id); - - /* check if this target has valid data */ - if ((ob == NULL) || (GS(ob->id.name) != ID_OB)) { - /* invalid target, so will not have enough targets */ - driver->flag |= DRIVER_FLAG_INVALID; - dtar->flag |= DTAR_FLAG_INVALID; - } - else { - /* target seems to be OK now... */ - dtar->flag &= ~DTAR_FLAG_INVALID; - valid_targets++; - } - } - DRIVER_TARGETS_LOOPER_END - + short valid_targets = driver_check_valid_targets(driver, dvar); + /* make sure we have enough valid targets to use - all or nothing for now... */ if (valid_targets < dvar->num_targets) { if (G.debug & G_DEBUG) { @@ -1351,8 +1346,7 @@ static float dvar_eval_locDiff(ChannelDriver *driver, DriverVar *dvar) } return 0.0f; } - - + /* SECOND PASS: get two location values */ /* NOTE: for now, these are all just worldspace */ DRIVER_TARGETS_USED_LOOPER(dvar) @@ -1570,7 +1564,7 @@ static DriverVarTypeInfo dvar_types[MAX_DVAR_TYPES] = { BEGIN_DVAR_TYPEDEF(DVAR_TYPE_ROT_DIFF) dvar_eval_rotDiff, /* eval callback */ 2, /* number of targets used */ - {"Bone 1", "Bone 2"}, /* UI names for targets */ + {"Object/Bone 1", "Object/Bone 2"}, /* UI names for targets */ {DTAR_FLAG_STRUCT_REF | DTAR_FLAG_ID_OB_ONLY, DTAR_FLAG_STRUCT_REF | DTAR_FLAG_ID_OB_ONLY} /* flags */ END_DVAR_TYPEDEF, diff --git a/source/blender/blenkernel/intern/idprop.c b/source/blender/blenkernel/intern/idprop.c index b2641b110f8..a596869e5e8 100644 --- a/source/blender/blenkernel/intern/idprop.c +++ b/source/blender/blenkernel/intern/idprop.c @@ -130,18 +130,22 @@ void IDP_FreeIDPArray(IDProperty *prop) MEM_freeN(prop->data.pointer); } -/*shallow copies item*/ +/* shallow copies item */ void IDP_SetIndexArray(IDProperty *prop, int index, IDProperty *item) { IDProperty *old; BLI_assert(prop->type == IDP_IDPARRAY); + if (index >= prop->len || index < 0) + return; + old = GETPROP(prop, index); - if (index >= prop->len || index < 0) return; - if (item != old) IDP_FreeProperty(old); - - memcpy(GETPROP(prop, index), item, sizeof(IDProperty)); + if (item != old) { + IDP_FreeProperty(old); + + memcpy(old, item, sizeof(IDProperty)); + } } IDProperty *IDP_GetIndexArray(IDProperty *prop, int index) diff --git a/source/blender/blenkernel/intern/image.c b/source/blender/blenkernel/intern/image.c index a2d94ccc478..318f6480aaf 100644 --- a/source/blender/blenkernel/intern/image.c +++ b/source/blender/blenkernel/intern/image.c @@ -3159,7 +3159,7 @@ static ImBuf *load_sequence_single(Image *ima, ImageUser *iuser, int frame, cons struct ImBuf *ibuf; char name[FILE_MAX]; int flag; - ImageUser iuser_t; + ImageUser iuser_t = {0}; /* XXX temp stuff? */ if (ima->lastframe != frame) @@ -3167,8 +3167,12 @@ static ImBuf *load_sequence_single(Image *ima, ImageUser *iuser, int frame, cons ima->lastframe = frame; - if (iuser) + if (iuser) { iuser_t = *iuser; + } + else { + /* TODO(sergey): Do we need to initialize something here? */ + } iuser_t.view = view_id; BKE_image_user_file_path(&iuser_t, ima, name); diff --git a/source/blender/blenkernel/intern/lattice.c b/source/blender/blenkernel/intern/lattice.c index b0671f33094..54cfae620a1 100644 --- a/source/blender/blenkernel/intern/lattice.c +++ b/source/blender/blenkernel/intern/lattice.c @@ -67,7 +67,7 @@ #include "BKE_deform.h" -/* Workaround for cyclic depenndnecy with curves. +/* Workaround for cyclic dependency with curves. * In such case curve_cache might not be ready yet, */ #define CYCLIC_DEPENDENCY_WORKAROUND diff --git a/source/blender/blenkernel/intern/library.c b/source/blender/blenkernel/intern/library.c index 6b4b492b212..e380b5c17e0 100644 --- a/source/blender/blenkernel/intern/library.c +++ b/source/blender/blenkernel/intern/library.c @@ -1863,7 +1863,7 @@ void BKE_library_make_local( /* Special case for objects because we don't want proxy pointers to be * cleared yet. This will happen down the road in this function. */ - BKE_object_make_local_ex(bmain, (Object*)id, true, false); + BKE_object_make_local_ex(bmain, (Object *)id, true, false); } else { id_make_local(bmain, id, false, true); diff --git a/source/blender/blenkernel/intern/library_remap.c b/source/blender/blenkernel/intern/library_remap.c index a408b498f18..b6f4621a0b3 100644 --- a/source/blender/blenkernel/intern/library_remap.c +++ b/source/blender/blenkernel/intern/library_remap.c @@ -179,6 +179,7 @@ static int foreach_libblock_remap_callback(void *user_data, ID *id_self, ID **id * on the other hand since they get reset to lib data on file open/reload it is indirect too... * Edit Mode is also a 'skip direct' case. */ const bool is_obj = (GS(id->name) == ID_OB); + const bool is_obj_proxy = (is_obj && (((Object *)id)->proxy || ((Object *)id)->proxy_group)); const bool is_obj_editmode = (is_obj && BKE_object_is_in_editmode((Object *)id)); const bool is_never_null = ((cb_flag & IDWALK_CB_NEVER_NULL) && (new_id == NULL) && (id_remap_data->flag & ID_REMAP_FORCE_NEVER_NULL_USAGE) == 0); @@ -231,7 +232,7 @@ static int foreach_libblock_remap_callback(void *user_data, ID *id_self, ID **id /* We cannot affect old_id->us directly, LIB_TAG_EXTRAUSER(_SET) are assumed to be set as needed, * that extra user is processed in final handling... */ } - if (!is_indirect) { + if (!is_indirect || is_obj_proxy) { id_remap_data->status |= ID_REMAP_IS_LINKED_DIRECT; } } diff --git a/source/blender/blenkernel/intern/mesh.c b/source/blender/blenkernel/intern/mesh.c index af02e02b017..c12890a354e 100644 --- a/source/blender/blenkernel/intern/mesh.c +++ b/source/blender/blenkernel/intern/mesh.c @@ -39,7 +39,9 @@ #include "BLI_utildefines.h" #include "BLI_math.h" +#include "BLI_linklist.h" #include "BLI_listbase.h" +#include "BLI_memarena.h" #include "BLI_edgehash.h" #include "BLI_string.h" @@ -66,6 +68,11 @@ #include "DEG_depsgraph.h" +/* Define for cases when you want extra validation of mesh + * after certain modifications. + */ +// #undef VALIDATE_MESH + enum { MESHCMP_DVERT_WEIGHTMISMATCH = 1, MESHCMP_DVERT_GROUPMISMATCH, @@ -2048,7 +2055,7 @@ void BKE_mesh_mselect_active_set(Mesh *me, int index, int type) (me->mselect[me->totselect - 1].type == type)); } -void BKE_mesh_calc_normals_split(Mesh *mesh) +void BKE_mesh_calc_normals_split_ex(Mesh *mesh, MLoopNorSpaceArray *r_lnors_spacearr) { float (*r_loopnors)[3]; float (*polynors)[3]; @@ -2083,113 +2090,339 @@ void BKE_mesh_calc_normals_split(Mesh *mesh) BKE_mesh_normals_loop_split( mesh->mvert, mesh->totvert, mesh->medge, mesh->totedge, mesh->mloop, r_loopnors, mesh->totloop, mesh->mpoly, (const float (*)[3])polynors, mesh->totpoly, - (mesh->flag & ME_AUTOSMOOTH) != 0, mesh->smoothresh, NULL, clnors, NULL); + (mesh->flag & ME_AUTOSMOOTH) != 0, mesh->smoothresh, r_lnors_spacearr, clnors, NULL); if (free_polynors) { MEM_freeN(polynors); } } -/* Spli faces based on the edge angle. - * Matches behavior of face splitting in render engines. - */ -void BKE_mesh_split_faces(Mesh *mesh) +void BKE_mesh_calc_normals_split(Mesh *mesh) { - const int num_verts = mesh->totvert; - const int num_edges = mesh->totedge; - const int num_polys = mesh->totpoly; + BKE_mesh_calc_normals_split_ex(mesh, NULL); +} + +/* Split faces helper functions. */ + +typedef struct SplitFaceNewVert { + struct SplitFaceNewVert *next; + int new_index; + int orig_index; + float *vnor; +} SplitFaceNewVert; + +typedef struct SplitFaceNewEdge { + struct SplitFaceNewEdge *next; + int new_index; + int orig_index; + int v1; + int v2; +} SplitFaceNewEdge; + +/* Detect needed new vertices, and update accordingly loops' vertex indices. + * WARNING! Leaves mesh in invalid state. */ +static int split_faces_prepare_new_verts( + const Mesh *mesh, MLoopNorSpaceArray *lnors_spacearr, SplitFaceNewVert **new_verts, MemArena *memarena, + bool *r_need_vnors_recalc) +{ + /* Note: if lnors_spacearr is NULL, ther is no autosmooth handling, and we only split out flat polys. */ + const int num_loops = mesh->totloop; + int num_verts = mesh->totvert; MVert *mvert = mesh->mvert; - MEdge *medge = mesh->medge; MLoop *mloop = mesh->mloop; - MPoly *mpoly = mesh->mpoly; - float (*lnors)[3]; - int poly, num_new_verts = 0; - if ((mesh->flag & ME_AUTOSMOOTH) == 0) { - return; - } - BKE_mesh_tessface_clear(mesh); - /* Compute loop normals if needed. */ - if (!CustomData_has_layer(&mesh->ldata, CD_NORMAL)) { - BKE_mesh_calc_normals_split(mesh); - } - lnors = CustomData_get_layer(&mesh->ldata, CD_NORMAL); - /* Count number of vertices to be split. */ - for (poly = 0; poly < num_polys; poly++) { - MPoly *mp = &mpoly[poly]; - int loop; - for (loop = 0; loop < mp->totloop; loop++) { - MLoop *ml = &mloop[mp->loopstart + loop]; - MVert *mv = &mvert[ml->v]; - float vn[3]; - normal_short_to_float_v3(vn, mv->no); - if (!equals_v3v3(vn, lnors[mp->loopstart + loop])) { - num_new_verts++; + + BLI_bitmap *verts_used = BLI_BITMAP_NEW(num_verts, __func__); + + if (lnors_spacearr) { + BLI_bitmap *done_loops = BLI_BITMAP_NEW(num_loops, __func__); + + MLoop *ml = mloop; + MLoopNorSpace **lnor_space = lnors_spacearr->lspacearr; + for (int loop_idx = 0; loop_idx < num_loops; loop_idx++, ml++, lnor_space++) { + if (!BLI_BITMAP_TEST(done_loops, loop_idx)) { + const int vert_idx = ml->v; + const bool vert_used = BLI_BITMAP_TEST_BOOL(verts_used, vert_idx); + /* If vert is already used by another smooth fan, we need a new vert for this one. */ + const int new_vert_idx = vert_used ? num_verts++ : vert_idx; + + BLI_assert(*lnor_space); + + if ((*lnor_space)->loops) { + for (LinkNode *lnode = (*lnor_space)->loops; lnode; lnode = lnode->next) { + const int ml_fan_idx = GET_INT_FROM_POINTER(lnode->link); + BLI_BITMAP_ENABLE(done_loops, ml_fan_idx); + if (vert_used) { + mloop[ml_fan_idx].v = new_vert_idx; + } + } + } + else { + /* Single loop in this fan... */ + BLI_BITMAP_ENABLE(done_loops, loop_idx); + if (vert_used) { + ml->v = new_vert_idx; + } + } + + if (!vert_used) { + BLI_BITMAP_ENABLE(verts_used, vert_idx); + /* We need to update that vertex's normal here, we won't go over it again. */ + /* This is important! *DO NOT* set vnor to final computed lnor, vnor should always be defined to + * 'automatic normal' value computed from its polys, not some custom normal. + * Fortunately, that's the loop normal space's 'lnor' reference vector. ;) */ + normal_float_to_short_v3(mvert[vert_idx].no, (*lnor_space)->vec_lnor); + } + else { + /* Add new vert to list. */ + SplitFaceNewVert *new_vert = BLI_memarena_alloc(memarena, sizeof(*new_vert)); + new_vert->orig_index = vert_idx; + new_vert->new_index = new_vert_idx; + new_vert->vnor = (*lnor_space)->vec_lnor; /* See note above. */ + new_vert->next = *new_verts; + *new_verts = new_vert; + } } } + + MEM_freeN(done_loops); } - if (num_new_verts == 0) { - /* No new vertices are to be added, can do early exit. */ - return; - } - /* Reallocate all vert and edge related data. */ - mesh->totvert += num_new_verts; - mesh->totedge += 2 * num_new_verts; - CustomData_realloc(&mesh->vdata, mesh->totvert); - CustomData_realloc(&mesh->edata, mesh->totedge); - /* Update pointers to a newly allocated memory. */ - BKE_mesh_update_customdata_pointers(mesh, false); - mvert = mesh->mvert; - medge = mesh->medge; - /* Perform actual vertex split. */ - num_new_verts = 0; - for (poly = 0; poly < num_polys; poly++) { - MPoly *mp = &mpoly[poly]; - int loop; - for (loop = 0; loop < mp->totloop; loop++) { - int poly_loop = mp->loopstart + loop; - MLoop *ml = &mloop[poly_loop]; - MVert *mv = &mvert[ml->v]; - float vn[3]; - normal_short_to_float_v3(vn, mv->no); - if (!equals_v3v3(vn, lnors[mp->loopstart + loop])) { - int poly_loop_prev = mp->loopstart + (loop + mp->totloop - 1) % mp->totloop; - MLoop *ml_prev = &mloop[poly_loop_prev]; - int new_edge_prev, new_edge; - /* Cretae new vertex. */ - int new_vert = num_verts + num_new_verts; - CustomData_copy_data(&mesh->vdata, &mesh->vdata, - ml->v, new_vert, 1); - normal_float_to_short_v3(mvert[new_vert].no, - lnors[poly_loop]); - /* Create new edges. */ - new_edge_prev = num_edges + 2 * num_new_verts; - new_edge = num_edges + 2 * num_new_verts + 1; - CustomData_copy_data(&mesh->edata, &mesh->edata, - ml_prev->e, new_edge_prev, 1); - CustomData_copy_data(&mesh->edata, &mesh->edata, - ml->e, new_edge, 1); - if (medge[new_edge_prev].v1 == ml->v) { - medge[new_edge_prev].v1 = new_vert; + else { + /* No loop normal spaces available, we only split out flat polys. */ + const int num_polys = mesh->totpoly; + const MPoly *mpoly = mesh->mpoly; + + /* We do that in two loops, to keep original edges/verts to smooth polys preferencially. */ + const MPoly *mp = mpoly; + for (int i = 0; i < num_polys; i++, mp++) { + if (mp->flag & ME_SMOOTH) { + const MLoop *ml = &mloop[mp->loopstart]; + for (int j = 0; j < mp->totloop; j++, ml++) { + /* Just mark the vertex as used/reserved, that way neighbor flat polys, if any, + * will have to create their own. */ + BLI_BITMAP_ENABLE(verts_used, ml->v); } - else { - medge[new_edge_prev].v2 = new_vert; + } + } + + mp = mpoly; + for (int i = 0; i < num_polys; i++, mp++) { + if (!(mp->flag & ME_SMOOTH)) { + MLoop *ml = &mloop[mp->loopstart]; + for (int j = 0; j < mp->totloop; j++, ml++) { + const int vert_idx = ml->v; + + if (BLI_BITMAP_TEST(verts_used, vert_idx)) { + /* Add new vert to list. */ + const int new_vert_idx = num_verts++; + ml->v = new_vert_idx; + + SplitFaceNewVert *new_vert = BLI_memarena_alloc(memarena, sizeof(*new_vert)); + new_vert->orig_index = vert_idx; + new_vert->new_index = new_vert_idx; + new_vert->vnor = NULL; /* See note below about normals. */ + new_vert->next = *new_verts; + *new_verts = new_vert; + } + else { + BLI_BITMAP_ENABLE(verts_used, vert_idx); + } } - if (medge[new_edge].v1 == ml->v) { - medge[new_edge].v1 = new_vert; + /* Note: there is no way to get new normals for smooth vertices here (and we don't have direct access + * to poly normals either for flat ones), so we'll have to recompute all vnors at the end... */ + *r_need_vnors_recalc = true; + } + } + } + + MEM_freeN(verts_used); + + return num_verts - mesh->totvert; +} + +/* Detect needed new edges, and update accordingly loops' edge indices. + * WARNING! Leaves mesh in invalid state. */ +static int split_faces_prepare_new_edges( + const Mesh *mesh, SplitFaceNewEdge **new_edges, MemArena *memarena) +{ + const int num_polys = mesh->totpoly; + int num_edges = mesh->totedge; + MEdge *medge = mesh->medge; + MLoop *mloop = mesh->mloop; + const MPoly *mpoly = mesh->mpoly; + + BLI_bitmap *edges_used = BLI_BITMAP_NEW(num_edges, __func__); + EdgeHash *edges_hash = BLI_edgehash_new_ex(__func__, num_edges); + + const MPoly *mp = mpoly; + for (int poly_idx = 0; poly_idx < num_polys; poly_idx++, mp++) { + MLoop *ml_prev = &mloop[mp->loopstart + mp->totloop - 1]; + MLoop *ml = &mloop[mp->loopstart]; + for (int loop_idx = 0; loop_idx < mp->totloop; loop_idx++, ml++) { + void **eval; + if (!BLI_edgehash_ensure_p(edges_hash, ml_prev->v, ml->v, &eval)) { + const int edge_idx = ml_prev->e; + + /* That edge has not been encountered yet, define it. */ + if (BLI_BITMAP_TEST(edges_used, edge_idx)) { + /* Original edge has already been used, we need to define a new one. */ + const int new_edge_idx = num_edges++; + *eval = SET_INT_IN_POINTER(new_edge_idx); + ml_prev->e = new_edge_idx; + + SplitFaceNewEdge *new_edge = BLI_memarena_alloc(memarena, sizeof(*new_edge)); + new_edge->orig_index = edge_idx; + new_edge->new_index = new_edge_idx; + new_edge->v1 = ml_prev->v; + new_edge->v2 = ml->v; + new_edge->next = *new_edges; + *new_edges = new_edge; } else { - medge[new_edge].v2 = new_vert; + /* We can re-use original edge. */ + medge[edge_idx].v1 = ml_prev->v; + medge[edge_idx].v2 = ml->v; + *eval = SET_INT_IN_POINTER(edge_idx); + BLI_BITMAP_ENABLE(edges_used, edge_idx); } - - ml->v = new_vert; - ml_prev->e = new_edge_prev; - ml->e = new_edge; - num_new_verts++; } + else { + /* Edge already known, just update loop's edge index. */ + ml_prev->e = GET_INT_FROM_POINTER(*eval); + } + + ml_prev = ml; + } + } + + MEM_freeN(edges_used); + BLI_edgehash_free(edges_hash, NULL); + + return num_edges - mesh->totedge; +} + +/* Perform actual split of vertices. */ +static void split_faces_split_new_verts( + Mesh *mesh, SplitFaceNewVert *new_verts, const int num_new_verts) +{ + const int num_verts = mesh->totvert - num_new_verts; + MVert *mvert = mesh->mvert; + + /* Remember new_verts is a single linklist, so its items are in reversed order... */ + MVert *new_mv = &mvert[mesh->totvert - 1]; + for (int i = mesh->totvert - 1; i >= num_verts ; i--, new_mv--, new_verts = new_verts->next) { + BLI_assert(new_verts->new_index == i); + BLI_assert(new_verts->new_index != new_verts->orig_index); + CustomData_copy_data(&mesh->vdata, &mesh->vdata, new_verts->orig_index, i, 1); + if (new_verts->vnor) { + normal_float_to_short_v3(new_mv->no, new_verts->vnor); } } } +/* Perform actual split of edges. */ +static void split_faces_split_new_edges( + Mesh *mesh, SplitFaceNewEdge *new_edges, const int num_new_edges) +{ + const int num_edges = mesh->totedge - num_new_edges; + MEdge *medge = mesh->medge; + + /* Remember new_edges is a single linklist, so its items are in reversed order... */ + MEdge *new_med = &medge[mesh->totedge - 1]; + for (int i = mesh->totedge - 1; i >= num_edges ; i--, new_med--, new_edges = new_edges->next) { + BLI_assert(new_edges->new_index == i); + BLI_assert(new_edges->new_index != new_edges->orig_index); + CustomData_copy_data(&mesh->edata, &mesh->edata, new_edges->orig_index, i, 1); + new_med->v1 = new_edges->v1; + new_med->v2 = new_edges->v2; + } +} + +/* Split faces based on the edge angle and loop normals. + * Matches behavior of face splitting in render engines. + * + * NOTE: Will leave CD_NORMAL loop data layer which is + * used by render engines to set shading up. + */ +void BKE_mesh_split_faces(Mesh *mesh, bool free_loop_normals) +{ + const int num_polys = mesh->totpoly; + + if (num_polys == 0) { + return; + } + BKE_mesh_tessface_clear(mesh); + + MLoopNorSpaceArray *lnors_spacearr = NULL; + MemArena *memarena; + bool need_vnors_recalc = false; + + if (mesh->flag & ME_AUTOSMOOTH) { + lnors_spacearr = MEM_callocN(sizeof(*lnors_spacearr), __func__); + /* Compute loop normals and loop normal spaces (a.k.a. smooth fans of faces around vertices). */ + BKE_mesh_calc_normals_split_ex(mesh, lnors_spacearr); + /* Stealing memarena from loop normals space array. */ + memarena = lnors_spacearr->mem; + } + else { + /* We still have to split out flat faces... */ + memarena = BLI_memarena_new(BLI_MEMARENA_STD_BUFSIZE, __func__); + } + + SplitFaceNewVert *new_verts = NULL; + SplitFaceNewEdge *new_edges = NULL; + + /* Detect loop normal spaces (a.k.a. smooth fans) that will need a new vert. */ + const int num_new_verts = split_faces_prepare_new_verts(mesh, lnors_spacearr, &new_verts, memarena, &need_vnors_recalc); + + if (num_new_verts > 0) { + /* Reminder: beyond this point, there is no way out, mesh is in invalid state (due to early-reassignment of + * loops' vertex and edge indices to new, to-be-created split ones). */ + + const int num_new_edges = split_faces_prepare_new_edges(mesh, &new_edges, memarena); + /* We can have to split a vertex without having to add a single new edge... */ + const bool do_edges = (num_new_edges > 0); + + /* Reallocate all vert and edge related data. */ + mesh->totvert += num_new_verts; + mesh->totedge += num_new_edges; + CustomData_realloc(&mesh->vdata, mesh->totvert); + if (do_edges) { + CustomData_realloc(&mesh->edata, mesh->totedge); + } + /* Update pointers to a newly allocated memory. */ + BKE_mesh_update_customdata_pointers(mesh, false); + + /* Perform actual split of vertices and edges. */ + split_faces_split_new_verts(mesh, new_verts, num_new_verts); + if (do_edges) { + split_faces_split_new_edges(mesh, new_edges, num_new_edges); + } + } + + /* Note: after this point mesh is expected to be valid again. */ + + /* CD_NORMAL is expected to be temporary only. */ + if (free_loop_normals) { + CustomData_free_layers(&mesh->ldata, CD_NORMAL, mesh->totloop); + } + + if (lnors_spacearr) { + /* Also frees new_verts/edges temp data, since we used its memarena to allocate them. */ + BKE_lnor_spacearr_free(lnors_spacearr); + MEM_freeN(lnors_spacearr); + } + else { + BLI_memarena_free(memarena); + } + + if (need_vnors_recalc) { + BKE_mesh_calc_normals(mesh); + } +#ifdef VALIDATE_MESH + BKE_mesh_validate(mesh, true, true); +#endif +} + /* settings: 1 - preview, 2 - render */ Mesh *BKE_mesh_new_from_object( Main *bmain, Scene *sce, Object *ob, diff --git a/source/blender/blenkernel/intern/mesh_evaluate.c b/source/blender/blenkernel/intern/mesh_evaluate.c index f9eba118383..0d0055113b7 100644 --- a/source/blender/blenkernel/intern/mesh_evaluate.c +++ b/source/blender/blenkernel/intern/mesh_evaluate.c @@ -304,13 +304,13 @@ void BKE_mesh_calc_normals_poly( void BKE_mesh_calc_normals(Mesh *mesh) { #ifdef DEBUG_TIME - TIMEIT_START(BKE_mesh_calc_normals); + TIMEIT_START_AVERAGED(BKE_mesh_calc_normals); #endif BKE_mesh_calc_normals_poly(mesh->mvert, NULL, mesh->totvert, mesh->mloop, mesh->mpoly, mesh->totloop, mesh->totpoly, NULL, false); #ifdef DEBUG_TIME - TIMEIT_END(BKE_mesh_calc_normals); + TIMEIT_END_AVERAGED(BKE_mesh_calc_normals); #endif } @@ -630,7 +630,6 @@ typedef struct LoopSplitTaskDataCommon { * Note we do not need to protect it, though, since two different tasks will *always* affect different * elements in the arrays. */ MLoopNorSpaceArray *lnors_spacearr; - BLI_bitmap *sharp_verts; float (*loopnors)[3]; short (*clnors_data)[2]; @@ -643,11 +642,8 @@ typedef struct LoopSplitTaskDataCommon { const int *loop_to_poly; const float (*polynors)[3]; + int numLoops; int numPolys; - - /* ***** Workers communication. ***** */ - ThreadQueue *task_queue; - } LoopSplitTaskDataCommon; #define INDEX_UNSET INT_MIN @@ -655,6 +651,50 @@ typedef struct LoopSplitTaskDataCommon { /* See comment about edge_to_loops below. */ #define IS_EDGE_SHARP(_e2l) (ELEM((_e2l)[1], INDEX_UNSET, INDEX_INVALID)) +static void loop_manifold_fan_around_vert_next( + const MLoop *mloops, const MPoly *mpolys, + const int *loop_to_poly, const int *e2lfan_curr, const uint mv_pivot_index, + const MLoop **r_mlfan_curr, int *r_mlfan_curr_index, int *r_mlfan_vert_index, int *r_mpfan_curr_index) +{ + const MLoop *mlfan_next; + const MPoly *mpfan_next; + + /* Warning! This is rather complex! + * We have to find our next edge around the vertex (fan mode). + * First we find the next loop, which is either previous or next to mlfan_curr_index, depending + * whether both loops using current edge are in the same direction or not, and whether + * mlfan_curr_index actually uses the vertex we are fanning around! + * mlfan_curr_index is the index of mlfan_next here, and mlfan_next is not the real next one + * (i.e. not the future mlfan_curr)... + */ + *r_mlfan_curr_index = (e2lfan_curr[0] == *r_mlfan_curr_index) ? e2lfan_curr[1] : e2lfan_curr[0]; + *r_mpfan_curr_index = loop_to_poly[*r_mlfan_curr_index]; + + BLI_assert(*r_mlfan_curr_index >= 0); + BLI_assert(*r_mpfan_curr_index >= 0); + + mlfan_next = &mloops[*r_mlfan_curr_index]; + mpfan_next = &mpolys[*r_mpfan_curr_index]; + if (((*r_mlfan_curr)->v == mlfan_next->v && (*r_mlfan_curr)->v == mv_pivot_index) || + ((*r_mlfan_curr)->v != mlfan_next->v && (*r_mlfan_curr)->v != mv_pivot_index)) + { + /* We need the previous loop, but current one is our vertex's loop. */ + *r_mlfan_vert_index = *r_mlfan_curr_index; + if (--(*r_mlfan_curr_index) < mpfan_next->loopstart) { + *r_mlfan_curr_index = mpfan_next->loopstart + mpfan_next->totloop - 1; + } + } + else { + /* We need the next loop, which is also our vertex's loop. */ + if (++(*r_mlfan_curr_index) >= mpfan_next->loopstart + mpfan_next->totloop) { + *r_mlfan_curr_index = mpfan_next->loopstart; + } + *r_mlfan_vert_index = *r_mlfan_curr_index; + } + *r_mlfan_curr = &mloops[*r_mlfan_curr_index]; + /* And now we are back in sync, mlfan_curr_index is the index of mlfan_curr! Pff! */ +} + static void split_loop_nor_single_do(LoopSplitTaskDataCommon *common_data, LoopSplitTaskData *data) { MLoopNorSpaceArray *lnors_spacearr = common_data->lnors_spacearr; @@ -680,7 +720,7 @@ static void split_loop_nor_single_do(LoopSplitTaskDataCommon *common_data, LoopS */ copy_v3_v3(*lnor, polynors[mp_index]); - /* printf("BASIC: handling loop %d / edge %d / vert %d / poly %d\n", ml_curr_index, ml_curr->e, ml_curr->v, mp_index); */ +// printf("BASIC: handling loop %d / edge %d / vert %d / poly %d\n", ml_curr_index, ml_curr->e, ml_curr->v, mp_index); /* If needed, generate this (simple!) lnor space. */ if (lnors_spacearr) { @@ -747,8 +787,7 @@ static void split_loop_nor_fan_do(LoopSplitTaskDataCommon *common_data, LoopSpli const MEdge *me_org = &medges[ml_curr->e]; /* ml_curr would be mlfan_prev if we needed that one */ const int *e2lfan_curr; float vec_curr[3], vec_prev[3], vec_org[3]; - const MLoop *mlfan_curr, *mlfan_next; - const MPoly *mpfan_next; + const MLoop *mlfan_curr; float lnor[3] = {0.0f, 0.0f, 0.0f}; /* mlfan_vert_index: the loop of our current edge might not be the loop of our current vertex! */ int mlfan_curr_index, mlfan_vert_index, mpfan_curr_index; @@ -787,7 +826,7 @@ static void split_loop_nor_fan_do(LoopSplitTaskDataCommon *common_data, LoopSpli } } - /* printf("FAN: vert %d, start edge %d\n", mv_pivot_index, ml_curr->e); */ +// printf("FAN: vert %d, start edge %d\n", mv_pivot_index, ml_curr->e); while (true) { const MEdge *me_curr = &medges[mlfan_curr->e]; @@ -803,7 +842,7 @@ static void split_loop_nor_fan_do(LoopSplitTaskDataCommon *common_data, LoopSpli normalize_v3(vec_curr); } - /* printf("\thandling edge %d / loop %d\n", mlfan_curr->e, mlfan_curr_index); */ +// printf("\thandling edge %d / loop %d\n", mlfan_curr->e, mlfan_curr_index); { /* Code similar to accumulate_vertex_normals_poly. */ @@ -845,46 +884,16 @@ static void split_loop_nor_fan_do(LoopSplitTaskDataCommon *common_data, LoopSpli /* Current edge is sharp and we have finished with this fan of faces around this vert, * or this vert is smooth, and we have completed a full turn around it. */ - /* printf("FAN: Finished!\n"); */ +// printf("FAN: Finished!\n"); break; } copy_v3_v3(vec_prev, vec_curr); - /* Warning! This is rather complex! - * We have to find our next edge around the vertex (fan mode). - * First we find the next loop, which is either previous or next to mlfan_curr_index, depending - * whether both loops using current edge are in the same direction or not, and whether - * mlfan_curr_index actually uses the vertex we are fanning around! - * mlfan_curr_index is the index of mlfan_next here, and mlfan_next is not the real next one - * (i.e. not the future mlfan_curr)... - */ - mlfan_curr_index = (e2lfan_curr[0] == mlfan_curr_index) ? e2lfan_curr[1] : e2lfan_curr[0]; - mpfan_curr_index = loop_to_poly[mlfan_curr_index]; - - BLI_assert(mlfan_curr_index >= 0); - BLI_assert(mpfan_curr_index >= 0); - - mlfan_next = &mloops[mlfan_curr_index]; - mpfan_next = &mpolys[mpfan_curr_index]; - if ((mlfan_curr->v == mlfan_next->v && mlfan_curr->v == mv_pivot_index) || - (mlfan_curr->v != mlfan_next->v && mlfan_curr->v != mv_pivot_index)) - { - /* We need the previous loop, but current one is our vertex's loop. */ - mlfan_vert_index = mlfan_curr_index; - if (--mlfan_curr_index < mpfan_next->loopstart) { - mlfan_curr_index = mpfan_next->loopstart + mpfan_next->totloop - 1; - } - } - else { - /* We need the next loop, which is also our vertex's loop. */ - if (++mlfan_curr_index >= mpfan_next->loopstart + mpfan_next->totloop) { - mlfan_curr_index = mpfan_next->loopstart; - } - mlfan_vert_index = mlfan_curr_index; - } - mlfan_curr = &mloops[mlfan_curr_index]; - /* And now we are back in sync, mlfan_curr_index is the index of mlfan_curr! Pff! */ + /* Find next loop of the smooth fan. */ + loop_manifold_fan_around_vert_next( + mloops, mpolys, loop_to_poly, e2lfan_curr, mv_pivot_index, + &mlfan_curr, &mlfan_curr_index, &mlfan_vert_index, &mpfan_curr_index); e2lfan_curr = edge_to_loops[mlfan_curr->e]; } @@ -955,31 +964,25 @@ static void loop_split_worker_do( } } -static void loop_split_worker(TaskPool * __restrict UNUSED(pool), void *taskdata, int UNUSED(threadid)) +static void loop_split_worker(TaskPool * __restrict pool, void *taskdata, int UNUSED(threadid)) { - LoopSplitTaskDataCommon *common_data = taskdata; - LoopSplitTaskData *data_buff; + LoopSplitTaskDataCommon *common_data = BLI_task_pool_userdata(pool); + LoopSplitTaskData *data = taskdata; /* Temp edge vectors stack, only used when computing lnor spacearr. */ BLI_Stack *edge_vectors = common_data->lnors_spacearr ? BLI_stack_new(sizeof(float[3]), __func__) : NULL; #ifdef DEBUG_TIME - TIMEIT_START(loop_split_worker); + TIMEIT_START_AVERAGED(loop_split_worker); #endif - while ((data_buff = BLI_thread_queue_pop(common_data->task_queue))) { - LoopSplitTaskData *data = data_buff; - int i; - - for (i = 0; i < LOOP_SPLIT_TASK_BLOCK_SIZE; i++, data++) { - /* A NULL ml_curr is used to tag ended data! */ - if (data->ml_curr == NULL) { - break; - } - loop_split_worker_do(common_data, data, edge_vectors); + for (int i = 0; i < LOOP_SPLIT_TASK_BLOCK_SIZE; i++, data++) { + /* A NULL ml_curr is used to tag ended data! */ + if (data->ml_curr == NULL) { + break; } - MEM_freeN(data_buff); + loop_split_worker_do(common_data, data, edge_vectors); } if (edge_vectors) { @@ -987,49 +990,117 @@ static void loop_split_worker(TaskPool * __restrict UNUSED(pool), void *taskdata } #ifdef DEBUG_TIME - TIMEIT_END(loop_split_worker); + TIMEIT_END_AVERAGED(loop_split_worker); #endif } -/* Note we use data_buff to detect whether we are in threaded context or not, in later case it is NULL. */ -static void loop_split_generator_do(LoopSplitTaskDataCommon *common_data, const bool threaded) +/* Check whether gievn loop is part of an unknown-so-far cyclic smooth fan, or not. + * Needed because cyclic smooth fans have no obvious 'entry point', and yet we need to walk them once, and only once. */ +static bool loop_split_generator_check_cyclic_smooth_fan( + const MLoop *mloops, const MPoly *mpolys, + const int (*edge_to_loops)[2], const int *loop_to_poly, const int *e2l_prev, BLI_bitmap *skip_loops, + const MLoop *ml_curr, const MLoop *ml_prev, const int ml_curr_index, const int ml_prev_index, + const int mp_curr_index) +{ + const unsigned int mv_pivot_index = ml_curr->v; /* The vertex we are "fanning" around! */ + const int *e2lfan_curr; + const MLoop *mlfan_curr; + /* mlfan_vert_index: the loop of our current edge might not be the loop of our current vertex! */ + int mlfan_curr_index, mlfan_vert_index, mpfan_curr_index; + + e2lfan_curr = e2l_prev; + if (IS_EDGE_SHARP(e2lfan_curr)) { + /* Sharp loop, so not a cyclic smooth fan... */ + return false; + } + + mlfan_curr = ml_prev; + mlfan_curr_index = ml_prev_index; + mlfan_vert_index = ml_curr_index; + mpfan_curr_index = mp_curr_index; + + BLI_assert(mlfan_curr_index >= 0); + BLI_assert(mlfan_vert_index >= 0); + BLI_assert(mpfan_curr_index >= 0); + + BLI_assert(!BLI_BITMAP_TEST(skip_loops, mlfan_vert_index)); + BLI_BITMAP_ENABLE(skip_loops, mlfan_vert_index); + + while(true) { + /* Find next loop of the smooth fan. */ + loop_manifold_fan_around_vert_next( + mloops, mpolys, loop_to_poly, e2lfan_curr, mv_pivot_index, + &mlfan_curr, &mlfan_curr_index, &mlfan_vert_index, &mpfan_curr_index); + + e2lfan_curr = edge_to_loops[mlfan_curr->e]; + + if (IS_EDGE_SHARP(e2lfan_curr)) { + /* Sharp loop/edge, so not a cyclic smooth fan... */ + return false; + } + /* Smooth loop/edge... */ + else if (BLI_BITMAP_TEST(skip_loops, mlfan_vert_index)) { + if (mlfan_vert_index == ml_curr_index) { + /* We walked around a whole cyclic smooth fan without finding any already-processed loop, means we can + * use initial ml_curr/ml_prev edge as start for this smooth fan. */ + return true; + } + /* ... already checked in some previous looping, we can abort. */ + return false; + } + else { + /* ... we can skip it in future, and keep checking the smooth fan. */ + BLI_BITMAP_ENABLE(skip_loops, mlfan_vert_index); + } + } +} + +static void loop_split_generator(TaskPool *pool, LoopSplitTaskDataCommon *common_data) { MLoopNorSpaceArray *lnors_spacearr = common_data->lnors_spacearr; - BLI_bitmap *sharp_verts = common_data->sharp_verts; float (*loopnors)[3] = common_data->loopnors; const MLoop *mloops = common_data->mloops; const MPoly *mpolys = common_data->mpolys; + const int *loop_to_poly = common_data->loop_to_poly; const int (*edge_to_loops)[2] = common_data->edge_to_loops; + const int numLoops = common_data->numLoops; const int numPolys = common_data->numPolys; const MPoly *mp; int mp_index; - LoopSplitTaskData *data, *data_buff = NULL, data_mem; + const MLoop *ml_curr; + const MLoop *ml_prev; + int ml_curr_index; + int ml_prev_index; + + BLI_bitmap *skip_loops = BLI_BITMAP_NEW(numLoops, __func__); + + LoopSplitTaskData *data_buff = NULL; int data_idx = 0; /* Temp edge vectors stack, only used when computing lnor spacearr (and we are not multi-threading). */ - BLI_Stack *edge_vectors = (lnors_spacearr && !data_buff) ? BLI_stack_new(sizeof(float[3]), __func__) : NULL; + BLI_Stack *edge_vectors = NULL; #ifdef DEBUG_TIME - TIMEIT_START(loop_split_generator); + TIMEIT_START_AVERAGED(loop_split_generator); #endif - if (!threaded) { - memset(&data_mem, 0, sizeof(data_mem)); - data = &data_mem; + if (!pool) { + if (lnors_spacearr) { + edge_vectors = BLI_stack_new(sizeof(float[3]), __func__); + } } /* We now know edges that can be smoothed (with their vector, and their two loops), and edges that will be hard! * Now, time to generate the normals. */ for (mp = mpolys, mp_index = 0; mp_index < numPolys; mp++, mp_index++) { - const MLoop *ml_curr, *ml_prev; float (*lnors)[3]; const int ml_last_index = (mp->loopstart + mp->totloop) - 1; - int ml_curr_index = mp->loopstart; - int ml_prev_index = ml_last_index; + ml_curr_index = mp->loopstart; + ml_prev_index = ml_last_index; ml_curr = &mloops[ml_curr_index]; ml_prev = &mloops[ml_prev_index]; @@ -1039,23 +1110,40 @@ static void loop_split_generator_do(LoopSplitTaskDataCommon *common_data, const const int *e2l_curr = edge_to_loops[ml_curr->e]; const int *e2l_prev = edge_to_loops[ml_prev->e]; - if (!IS_EDGE_SHARP(e2l_curr) && (!lnors_spacearr || BLI_BITMAP_TEST_BOOL(sharp_verts, ml_curr->v))) { - /* A smooth edge, and we are not generating lnor_spacearr, or the related vertex is sharp. - * We skip it because it is either: - * - in the middle of a 'smooth fan' already computed (or that will be as soon as we hit - * one of its ends, i.e. one of its two sharp edges), or... - * - the related vertex is a "full smooth" one, in which case pre-populated normals from vertex - * are just fine (or it has already be handled in a previous loop in case of needed lnors spacearr)! - */ - /* printf("Skipping loop %d / edge %d / vert %d(%d)\n", ml_curr_index, ml_curr->e, ml_curr->v, sharp_verts[ml_curr->v]); */ +// printf("Checking loop %d / edge %u / vert %u (sharp edge: %d, skiploop: %d)...", +// ml_curr_index, ml_curr->e, ml_curr->v, IS_EDGE_SHARP(e2l_curr), BLI_BITMAP_TEST_BOOL(skip_loops, ml_curr_index)); + + /* A smooth edge, we have to check for cyclic smooth fan case. + * If we find a new, never-processed cyclic smooth fan, we can do it now using that loop/edge as + * 'entry point', otherwise we can skip it. */ + /* Note: In theory, we could make loop_split_generator_check_cyclic_smooth_fan() store + * mlfan_vert_index'es and edge indexes in two stacks, to avoid having to fan again around the vert during + * actual computation of clnor & clnorspace. However, this would complicate the code, add more memory usage, + * and despite its logical complexity, loop_manifold_fan_around_vert_next() is quite cheap in term of + * CPU cycles, so really think it's not worth it. */ + if (!IS_EDGE_SHARP(e2l_curr) && + (BLI_BITMAP_TEST(skip_loops, ml_curr_index) || + !loop_split_generator_check_cyclic_smooth_fan( + mloops, mpolys, edge_to_loops, loop_to_poly, e2l_prev, skip_loops, + ml_curr, ml_prev, ml_curr_index, ml_prev_index, mp_index))) + { +// printf("SKIPPING!\n"); } else { - if (threaded) { + LoopSplitTaskData *data, data_local; + +// printf("PROCESSING!\n"); + + if (pool) { if (data_idx == 0) { data_buff = MEM_callocN(sizeof(*data_buff) * LOOP_SPLIT_TASK_BLOCK_SIZE, __func__); } data = &data_buff[data_idx]; } + else { + data = &data_local; + memset(data, 0, sizeof(*data)); + } if (IS_EDGE_SHARP(e2l_curr) && IS_EDGE_SHARP(e2l_prev)) { data->lnor = lnors; @@ -1091,22 +1179,18 @@ static void loop_split_generator_do(LoopSplitTaskDataCommon *common_data, const data->mp_index = mp_index; if (lnors_spacearr) { data->lnor_space = BKE_lnor_space_create(lnors_spacearr); - /* Tag related vertex as sharp, to avoid fanning around it again (in case it was a smooth one). - * This *has* to be done outside of workers tasks! */ - BLI_BITMAP_ENABLE(sharp_verts, ml_curr->v); } } - if (threaded) { + if (pool) { data_idx++; if (data_idx == LOOP_SPLIT_TASK_BLOCK_SIZE) { - BLI_thread_queue_push(common_data->task_queue, data_buff); + BLI_task_pool_push(pool, loop_split_worker, data_buff, true, TASK_PRIORITY_LOW); data_idx = 0; } } else { loop_split_worker_do(common_data, data, edge_vectors); - memset(data, 0, sizeof(data_mem)); } } @@ -1115,44 +1199,32 @@ static void loop_split_generator_do(LoopSplitTaskDataCommon *common_data, const } } - if (threaded) { - /* Last block of data... Since it is calloc'ed and we use first NULL item as stopper, everything is fine. */ - if (LIKELY(data_idx)) { - BLI_thread_queue_push(common_data->task_queue, data_buff); - } - - /* This will signal all other worker threads to wake up and finish! */ - BLI_thread_queue_nowait(common_data->task_queue); + /* Last block of data... Since it is calloc'ed and we use first NULL item as stopper, everything is fine. */ + if (pool && data_idx) { + BLI_task_pool_push(pool, loop_split_worker, data_buff, true, TASK_PRIORITY_LOW); } if (edge_vectors) { BLI_stack_free(edge_vectors); } + MEM_freeN(skip_loops); #ifdef DEBUG_TIME - TIMEIT_END(loop_split_generator); + TIMEIT_END_AVERAGED(loop_split_generator); #endif } -static void loop_split_generator(TaskPool * __restrict UNUSED(pool), void *taskdata, int UNUSED(threadid)) -{ - LoopSplitTaskDataCommon *common_data = taskdata; - - loop_split_generator_do(common_data, true); -} - /** * Compute split normals, i.e. vertex normals associated with each poly (hence 'loop normals'). * Useful to materialize sharp edges (or non-smooth faces) without actually modifying the geometry (splitting edges). */ void BKE_mesh_normals_loop_split( - const MVert *mverts, const int numVerts, MEdge *medges, const int numEdges, + const MVert *mverts, const int UNUSED(numVerts), MEdge *medges, const int numEdges, MLoop *mloops, float (*r_loopnors)[3], const int numLoops, MPoly *mpolys, const float (*polynors)[3], const int numPolys, const bool use_split_normals, float split_angle, MLoopNorSpaceArray *r_lnors_spacearr, short (*clnors_data)[2], int *r_loop_to_poly) { - /* For now this is not supported. If we do not use split normals, we do not generate anything fancy! */ BLI_assert(use_split_normals || !(r_lnors_spacearr)); @@ -1185,8 +1257,6 @@ void BKE_mesh_normals_loop_split( return; } - { - /* Mapping edge -> loops. * If that edge is used by more than two loops (polys), it is always sharp (and tagged as such, see below). * We also use the second loop index as a kind of flag: smooth edge: > 0, @@ -1196,33 +1266,25 @@ void BKE_mesh_normals_loop_split( * store the negated value of loop index instead of INDEX_INVALID to retrieve the real value later in code). * Note also that lose edges always have both values set to 0! */ - int (*edge_to_loops)[2] = MEM_callocN(sizeof(int[2]) * (size_t)numEdges, __func__); + int (*edge_to_loops)[2] = MEM_callocN(sizeof(*edge_to_loops) * (size_t)numEdges, __func__); /* Simple mapping from a loop to its polygon index. */ - int *loop_to_poly = r_loop_to_poly ? r_loop_to_poly : MEM_mallocN(sizeof(int) * (size_t)numLoops, __func__); + int *loop_to_poly = r_loop_to_poly ? r_loop_to_poly : MEM_mallocN(sizeof(*loop_to_poly) * (size_t)numLoops, __func__); MPoly *mp; - int mp_index, me_index; - bool check_angle = (split_angle < (float)M_PI); - int i; + int mp_index; - BLI_bitmap *sharp_verts = NULL; - MLoopNorSpaceArray _lnors_spacearr = {NULL}; + /* When using custom loop normals, disable the angle feature! */ + const bool check_angle = (split_angle < (float)M_PI) && (clnors_data == NULL); - LoopSplitTaskDataCommon common_data = {NULL}; + MLoopNorSpaceArray _lnors_spacearr = {NULL}; #ifdef DEBUG_TIME - TIMEIT_START(BKE_mesh_normals_loop_split); + TIMEIT_START_AVERAGED(BKE_mesh_normals_loop_split); #endif if (check_angle) { - /* When using custom loop normals, disable the angle feature! */ - if (clnors_data) { - check_angle = false; - } - else { - split_angle = cosf(split_angle); - } + split_angle = cosf(split_angle); } if (!r_lnors_spacearr && clnors_data) { @@ -1231,7 +1293,6 @@ void BKE_mesh_normals_loop_split( } if (r_lnors_spacearr) { BKE_lnor_spacearr_init(r_lnors_spacearr, numLoops); - sharp_verts = BLI_BITMAP_NEW((size_t)numVerts, __func__); } /* This first loop check which edges are actually smooth, and compute edge vectors. */ @@ -1285,60 +1346,38 @@ void BKE_mesh_normals_loop_split( } } - if (r_lnors_spacearr) { - /* Tag vertices that have at least one sharp edge as 'sharp' (used for the lnor spacearr computation). - * XXX This third loop over edges is a bit disappointing, could not find any other way yet. - * Not really performance-critical anyway. - */ - for (me_index = 0; me_index < numEdges; me_index++) { - const int *e2l = edge_to_loops[me_index]; - const MEdge *me = &medges[me_index]; - if (IS_EDGE_SHARP(e2l)) { - BLI_BITMAP_ENABLE(sharp_verts, me->v1); - BLI_BITMAP_ENABLE(sharp_verts, me->v2); - } - } - } - /* Init data common to all tasks. */ - common_data.lnors_spacearr = r_lnors_spacearr; - common_data.loopnors = r_loopnors; - common_data.clnors_data = clnors_data; - - common_data.mverts = mverts; - common_data.medges = medges; - common_data.mloops = mloops; - common_data.mpolys = mpolys; - common_data.sharp_verts = sharp_verts; - common_data.edge_to_loops = (const int(*)[2])edge_to_loops; - common_data.loop_to_poly = loop_to_poly; - common_data.polynors = polynors; - common_data.numPolys = numPolys; + LoopSplitTaskDataCommon common_data = { + .lnors_spacearr = r_lnors_spacearr, + .loopnors = r_loopnors, + .clnors_data = clnors_data, + .mverts = mverts, + .medges = medges, + .mloops = mloops, + .mpolys = mpolys, + .edge_to_loops = (const int(*)[2])edge_to_loops, + .loop_to_poly = loop_to_poly, + .polynors = polynors, + .numLoops = numLoops, + .numPolys = numPolys, + }; if (numLoops < LOOP_SPLIT_TASK_BLOCK_SIZE * 8) { /* Not enough loops to be worth the whole threading overhead... */ - loop_split_generator_do(&common_data, false); + loop_split_generator(NULL, &common_data); } else { TaskScheduler *task_scheduler; TaskPool *task_pool; - int nbr_workers; - - common_data.task_queue = BLI_thread_queue_init(); task_scheduler = BLI_task_scheduler_get(); - task_pool = BLI_task_pool_create(task_scheduler, NULL); + task_pool = BLI_task_pool_create(task_scheduler, &common_data); + + loop_split_generator(task_pool, &common_data); - nbr_workers = max_ii(2, BLI_task_scheduler_num_threads(task_scheduler)); - for (i = 1; i < nbr_workers; i++) { - BLI_task_pool_push(task_pool, loop_split_worker, &common_data, false, TASK_PRIORITY_HIGH); - } - BLI_task_pool_push(task_pool, loop_split_generator, &common_data, false, TASK_PRIORITY_HIGH); BLI_task_pool_work_and_wait(task_pool); BLI_task_pool_free(task_pool); - - BLI_thread_queue_free(common_data.task_queue); } MEM_freeN(edge_to_loops); @@ -1347,17 +1386,14 @@ void BKE_mesh_normals_loop_split( } if (r_lnors_spacearr) { - MEM_freeN(sharp_verts); if (r_lnors_spacearr == &_lnors_spacearr) { BKE_lnor_spacearr_free(r_lnors_spacearr); } } #ifdef DEBUG_TIME - TIMEIT_END(BKE_mesh_normals_loop_split); + TIMEIT_END_AVERAGED(BKE_mesh_normals_loop_split); #endif - - } } #undef INDEX_UNSET diff --git a/source/blender/blenkernel/intern/object.c b/source/blender/blenkernel/intern/object.c index ff8be5892e9..6e754755cf3 100644 --- a/source/blender/blenkernel/intern/object.c +++ b/source/blender/blenkernel/intern/object.c @@ -2236,66 +2236,6 @@ void BKE_boundbox_minmax(const BoundBox *bb, float obmat[4][4], float r_min[3], } } -/** - * Returns a BBox which each dimensions are at least epsilon. - * \note In case a given dimension needs to be enlarged, its final value will be in [epsilon, 3 * epsilon] range. - * - * \param bb the input bbox to check. - * \param bb_temp the temp bbox to modify (\a bb content is never changed). - * \param epsilon the minimum dimension to ensure. - * \return either bb (if nothing needed to be changed) or bb_temp. - */ -BoundBox *BKE_boundbox_ensure_minimum_dimensions(BoundBox *bb, BoundBox *bb_temp, const float epsilon) -{ - if (fabsf(bb->vec[0][0] - bb->vec[4][0]) < epsilon) { - /* Flat along X axis... */ - *bb_temp = *bb; - bb = bb_temp; - bb->vec[0][0] -= epsilon; - bb->vec[1][0] -= epsilon; - bb->vec[2][0] -= epsilon; - bb->vec[3][0] -= epsilon; - bb->vec[4][0] += epsilon; - bb->vec[5][0] += epsilon; - bb->vec[6][0] += epsilon; - bb->vec[7][0] += epsilon; - } - - if (fabsf(bb->vec[0][1] - bb->vec[3][1]) < epsilon) { - /* Flat along Y axis... */ - if (bb != bb_temp) { - *bb_temp = *bb; - bb = bb_temp; - } - bb->vec[0][1] -= epsilon; - bb->vec[1][1] -= epsilon; - bb->vec[4][1] -= epsilon; - bb->vec[5][1] -= epsilon; - bb->vec[2][1] += epsilon; - bb->vec[3][1] += epsilon; - bb->vec[6][1] += epsilon; - bb->vec[7][1] += epsilon; - } - - if (fabsf(bb->vec[0][2] - bb->vec[1][2]) < epsilon) { - /* Flat along Z axis... */ - if (bb != bb_temp) { - *bb_temp = *bb; - bb = bb_temp; - } - bb->vec[0][2] -= epsilon; - bb->vec[3][2] -= epsilon; - bb->vec[4][2] -= epsilon; - bb->vec[7][2] -= epsilon; - bb->vec[1][2] += epsilon; - bb->vec[2][2] += epsilon; - bb->vec[5][2] += epsilon; - bb->vec[6][2] += epsilon; - } - - return bb; -} - BoundBox *BKE_object_boundbox_get(Object *ob) { BoundBox *bb = NULL; diff --git a/source/blender/blenkernel/intern/object_update.c b/source/blender/blenkernel/intern/object_update.c index 5cb704e4737..a531466294e 100644 --- a/source/blender/blenkernel/intern/object_update.c +++ b/source/blender/blenkernel/intern/object_update.c @@ -54,6 +54,7 @@ #include "BKE_editmesh.h" #include "BKE_object.h" #include "BKE_particle.h" +#include "BKE_pointcache.h" #include "BKE_scene.h" #include "BKE_material.h" #include "BKE_image.h" @@ -347,3 +348,9 @@ void BKE_object_eval_uber_data(EvaluationContext *eval_ctx, ob->recalc &= ~(OB_RECALC_DATA | OB_RECALC_TIME); } + +void BKE_object_eval_cloth(EvaluationContext *UNUSED(eval_ctx), Scene *scene, Object *object) +{ + DEBUG_PRINT("%s on %s\n", __func__, object->id.name); + BKE_ptcache_object_reset(scene, object, PTCACHE_RESET_DEPSGRAPH); +} diff --git a/source/blender/blenkernel/intern/particle.c b/source/blender/blenkernel/intern/particle.c index 1ea27558545..228ae6afaba 100644 --- a/source/blender/blenkernel/intern/particle.c +++ b/source/blender/blenkernel/intern/particle.c @@ -633,8 +633,9 @@ void psys_render_set(Object *ob, ParticleSystem *psys, float viewmat[4][4], floa data->childcachebufs.last = psys->childcachebufs.last; data->totchildcache = psys->totchildcache; - if (psmd->dm_final) - data->dm = CDDM_copy(psmd->dm_final); + if (psmd->dm_final) { + data->dm = CDDM_copy_with_tessface(psmd->dm_final); + } data->totdmvert = psmd->totdmvert; data->totdmedge = psmd->totdmedge; data->totdmface = psmd->totdmface; diff --git a/source/blender/blenkernel/intern/scene.c b/source/blender/blenkernel/intern/scene.c index 56bfe5d7ff1..906fa0134a0 100644 --- a/source/blender/blenkernel/intern/scene.c +++ b/source/blender/blenkernel/intern/scene.c @@ -1510,8 +1510,6 @@ static void scene_update_object_func(TaskPool * __restrict pool, void *taskdata, if (add_to_stats) { StatisicsEntry *entry; - BLI_assert(threadid < BLI_pool_get_num_threads(pool)); - entry = MEM_mallocN(sizeof(StatisicsEntry), "update thread statistics"); entry->object = object; entry->start_time = start_time; @@ -1631,10 +1629,11 @@ static bool scene_need_update_objects(Main *bmain) static void scene_update_objects(EvaluationContext *eval_ctx, Main *bmain, Scene *scene, Scene *scene_parent) { - TaskScheduler *task_scheduler = BLI_task_scheduler_get(); + TaskScheduler *task_scheduler; TaskPool *task_pool; ThreadedObjectUpdateState state; bool need_singlethread_pass; + bool need_free_scheduler; /* Early check for whether we need to invoke all the task-based * things (spawn new ppol, traverse dependency graph and so on). @@ -1651,6 +1650,15 @@ static void scene_update_objects(EvaluationContext *eval_ctx, Main *bmain, Scene state.scene = scene; state.scene_parent = scene_parent; + if (G.debug & G_DEBUG_DEPSGRAPH_NO_THREADS) { + task_scheduler = BLI_task_scheduler_create(1); + need_free_scheduler = true; + } + else { + task_scheduler = BLI_task_scheduler_get(); + need_free_scheduler = false; + } + /* Those are only needed when blender is run with --debug argument. */ if (G.debug & G_DEBUG_DEPSGRAPH) { const int tot_thread = BLI_task_scheduler_num_threads(task_scheduler); @@ -1665,9 +1673,6 @@ static void scene_update_objects(EvaluationContext *eval_ctx, Main *bmain, Scene #endif task_pool = BLI_task_pool_create(task_scheduler, &state); - if (G.debug & G_DEBUG_DEPSGRAPH_NO_THREADS) { - BLI_pool_set_num_threads(task_pool, 1); - } DAG_threaded_update_begin(scene, scene_update_object_add_task, task_pool); BLI_task_pool_work_and_wait(task_pool); @@ -1700,6 +1705,10 @@ static void scene_update_objects(EvaluationContext *eval_ctx, Main *bmain, Scene if (need_singlethread_pass) { scene_update_all_bases(eval_ctx, scene, scene_parent); } + + if (need_free_scheduler) { + BLI_task_scheduler_free(task_scheduler); + } } static void scene_update_tagged_recursive(EvaluationContext *eval_ctx, Main *bmain, Scene *scene, Scene *scene_parent) diff --git a/source/blender/blenkernel/intern/sequencer.c b/source/blender/blenkernel/intern/sequencer.c index 1d2f5aee440..2de7cd49901 100644 --- a/source/blender/blenkernel/intern/sequencer.c +++ b/source/blender/blenkernel/intern/sequencer.c @@ -480,55 +480,74 @@ void BKE_sequencer_editing_free(Scene *scene) static void sequencer_imbuf_assign_spaces(Scene *scene, ImBuf *ibuf) { - if (ibuf->rect_float) { + if (ibuf->rect != NULL) { + IMB_colormanagement_assign_rect_colorspace(ibuf, scene->sequencer_colorspace_settings.name); + } + if (ibuf->rect_float != NULL) { IMB_colormanagement_assign_float_colorspace(ibuf, scene->sequencer_colorspace_settings.name); } } void BKE_sequencer_imbuf_to_sequencer_space(Scene *scene, ImBuf *ibuf, bool make_float) { - const char *from_colorspace = IMB_colormanagement_role_colorspace_name_get(COLOR_ROLE_SCENE_LINEAR); + /* Early output check: if both buffers are NULL we have nothing to convert. */ + if (ibuf->rect_float == NULL && ibuf->rect == NULL) { + return; + } + /* Get common conversion settings. */ const char *to_colorspace = scene->sequencer_colorspace_settings.name; - const char *float_colorspace = IMB_colormanagement_get_float_colorspace(ibuf); - - if (!ibuf->rect_float) { - if (ibuf->rect) { - const char *byte_colorspace = IMB_colormanagement_get_rect_colorspace(ibuf); - if (make_float || !STREQ(to_colorspace, byte_colorspace)) { - /* If byte space is not in sequencer's working space, we deliver float color space, - * this is to to prevent data loss. - */ - - /* when converting byte buffer to float in sequencer we need to make float - * buffer be in sequencer's working space, which is currently only doable - * from linear space. - */ - - /* - * OCIO_TODO: would be nice to support direct single transform from byte to sequencer's - */ - - IMB_float_from_rect(ibuf); - } - else { - return; - } + /* Perform actual conversion logic. */ + if (ibuf->rect_float == NULL) { + /* We are not requested to give float buffer and byte buffer is already + * in thee required colorspace. Can skip doing anything here. + */ + const char *from_colorspace = IMB_colormanagement_get_rect_colorspace(ibuf); + if (!make_float && STREQ(from_colorspace, to_colorspace)) { + return; + } + if (false) { + /* The idea here is to provide as fast playback as possible and + * enforcing float buffer here (a) uses more cache memory (b) might + * make some other effects slower to apply. + * + * However, this might also have negative effect by adding weird + * artifacts which will then not happen in final render. + */ + IMB_colormanagement_transform_byte_threaded( + (unsigned char *)ibuf->rect, ibuf->x, ibuf->y, ibuf->channels, + from_colorspace, to_colorspace); } else { - return; + /* We perform conversion to a float buffer so we don't worry about + * precision loss. + */ + imb_addrectfloatImBuf(ibuf); + IMB_colormanagement_transform_from_byte_threaded( + ibuf->rect_float, (unsigned char *)ibuf->rect, + ibuf->x, ibuf->y, ibuf->channels, + from_colorspace, to_colorspace); + /* We don't need byte buffer anymore. */ + imb_freerectImBuf(ibuf); } } - - if (from_colorspace && from_colorspace[0] != '\0') { - if (ibuf->rect) + else { + const char *from_colorspace = IMB_colormanagement_get_float_colorspace(ibuf); + /* Unknown input color space, can't perform conversion. */ + if (from_colorspace == NULL || from_colorspace[0] == '\0') { + return; + } + /* We don't want both byte and float buffers around: they'll either run + * out of sync or conversion of byte buffer will loose precision in there. + */ + if (ibuf->rect != NULL) { imb_freerectImBuf(ibuf); - - if (!STREQ(float_colorspace, to_colorspace)) { - IMB_colormanagement_transform_threaded(ibuf->rect_float, ibuf->x, ibuf->y, ibuf->channels, - from_colorspace, to_colorspace, true); - sequencer_imbuf_assign_spaces(scene, ibuf); } + IMB_colormanagement_transform_threaded(ibuf->rect_float, + ibuf->x, ibuf->y, ibuf->channels, + from_colorspace, to_colorspace, + true); } + sequencer_imbuf_assign_spaces(scene, ibuf); } void BKE_sequencer_imbuf_from_sequencer_space(Scene *scene, ImBuf *ibuf) diff --git a/source/blender/blenkernel/intern/text.c b/source/blender/blenkernel/intern/text.c index 88575c7d3be..298790a9e9b 100644 --- a/source/blender/blenkernel/intern/text.c +++ b/source/blender/blenkernel/intern/text.c @@ -1935,7 +1935,7 @@ void txt_do_undo(Text *text) int op = text->undo_buf[text->undo_pos]; int prev_flags; unsigned int linep; - unsigned int uchar; + unsigned int uni_char; unsigned int curln, selln; unsigned short curc, selc; unsigned short charp; @@ -1971,14 +1971,14 @@ void txt_do_undo(Text *text) case UNDO_BS_3: case UNDO_BS_4: charp = op - UNDO_BS_1 + 1; - uchar = txt_undo_read_unicode(text->undo_buf, &text->undo_pos, charp); + uni_char = txt_undo_read_unicode(text->undo_buf, &text->undo_pos, charp); /* get and restore the cursors */ txt_undo_read_cur(text->undo_buf, &text->undo_pos, &curln, &curc); txt_move_to(text, curln, curc, 0); txt_move_to(text, curln, curc, 1); - txt_add_char(text, uchar); + txt_add_char(text, uni_char); text->undo_pos--; break; @@ -1988,14 +1988,14 @@ void txt_do_undo(Text *text) case UNDO_DEL_3: case UNDO_DEL_4: charp = op - UNDO_DEL_1 + 1; - uchar = txt_undo_read_unicode(text->undo_buf, &text->undo_pos, charp); + uni_char = txt_undo_read_unicode(text->undo_buf, &text->undo_pos, charp); /* get and restore the cursors */ txt_undo_read_cur(text->undo_buf, &text->undo_pos, &curln, &curc); txt_move_to(text, curln, curc, 0); txt_move_to(text, curln, curc, 1); - txt_add_char(text, uchar); + txt_add_char(text, uni_char); txt_move_left(text, 0); @@ -2163,7 +2163,7 @@ void txt_do_redo(Text *text) char *buf; unsigned int linep; unsigned short charp; - unsigned int uchar; + unsigned int uni_uchar; unsigned int curln, selln; unsigned short curc, selc; @@ -2190,9 +2190,9 @@ void txt_do_redo(Text *text) txt_move_to(text, curln, curc, 1); charp = op - UNDO_INSERT_1 + 1; - uchar = txt_redo_read_unicode(text->undo_buf, &text->undo_pos, charp); + uni_uchar = txt_redo_read_unicode(text->undo_buf, &text->undo_pos, charp); - txt_add_char(text, uchar); + txt_add_char(text, uni_uchar); break; case UNDO_BS_1: diff --git a/source/blender/blenkernel/intern/texture.c b/source/blender/blenkernel/intern/texture.c index 2d3ecad19ad..60990c03b0e 100644 --- a/source/blender/blenkernel/intern/texture.c +++ b/source/blender/blenkernel/intern/texture.c @@ -1485,9 +1485,11 @@ bool BKE_texture_dependsOnTime(const struct Tex *texture) /* ------------------------------------------------------------------------- */ -void BKE_texture_get_value( +void BKE_texture_get_value_ex( const Scene *scene, Tex *texture, - float *tex_co, TexResult *texres, bool use_color_management) + float *tex_co, TexResult *texres, + struct ImagePool *pool, + bool use_color_management) { int result_type; bool do_color_manage = false; @@ -1497,7 +1499,7 @@ void BKE_texture_get_value( } /* no node textures for now */ - result_type = multitex_ext_safe(texture, tex_co, texres, NULL, do_color_manage, false); + result_type = multitex_ext_safe(texture, tex_co, texres, pool, do_color_manage, false); /* if the texture gave an RGB value, we assume it didn't give a valid * intensity, since this is in the context of modifiers don't use perceptual color conversion. @@ -1510,3 +1512,10 @@ void BKE_texture_get_value( copy_v3_fl(&texres->tr, texres->tin); } } + +void BKE_texture_get_value( + const Scene *scene, Tex *texture, + float *tex_co, TexResult *texres, bool use_color_management) +{ + BKE_texture_get_value_ex(scene, texture, tex_co, texres, NULL, use_color_management); +} diff --git a/source/blender/blenkernel/intern/tracking_stabilize.c b/source/blender/blenkernel/intern/tracking_stabilize.c index 36b24fbb2dc..3dfaa1ed77d 100644 --- a/source/blender/blenkernel/intern/tracking_stabilize.c +++ b/source/blender/blenkernel/intern/tracking_stabilize.c @@ -755,7 +755,8 @@ static void average_marker_positions(StabContext *ctx, int framenr, float r_ref_ if (ok) { r_ref_pos[0] /= weight_sum; r_ref_pos[1] /= weight_sum; - } else { + } + else { /* No usable tracking data on any track on this frame. * Use data from neighbouring frames to extrapolate... */ @@ -784,7 +785,8 @@ static void average_marker_positions(StabContext *ctx, int framenr, float r_ref_ * Also default to this frame when we're in a gap */ average_marker_positions(ctx, next_lower, r_ref_pos); - } else if (next_higher < MAXFRAME) { + } + else if (next_higher < MAXFRAME) { average_marker_positions(ctx, next_higher, r_ref_pos); } use_values_from_fcurves(ctx, false); @@ -1167,7 +1169,8 @@ static void stabilization_calculate_data(StabContext *ctx, if (ctx->stab->flag & TRACKING_STABILIZE_SCALE) { *r_scale = expf(scale_step * scaleinf); /* Averaged in log scale */ - } else { + } + else { *r_scale = 1.0f; } @@ -1180,8 +1183,8 @@ static void stabilization_calculate_data(StabContext *ctx, */ get_animated_target_pos(ctx, framenr, target_pos); sub_v2_v2(r_translation, target_pos); - *r_angle -= get_animated_target_rot(ctx,framenr); - target_scale = get_animated_target_scale(ctx,framenr); + *r_angle -= get_animated_target_rot(ctx, framenr); + target_scale = get_animated_target_scale(ctx, framenr); if (target_scale != 0.0f) { *r_scale /= target_scale; /* target_scale is an expected/intended reference zoom value */ diff --git a/source/blender/blenkernel/intern/writeffmpeg.c b/source/blender/blenkernel/intern/writeffmpeg.c index 9994d479ce7..156b74f5c3d 100644 --- a/source/blender/blenkernel/intern/writeffmpeg.c +++ b/source/blender/blenkernel/intern/writeffmpeg.c @@ -568,7 +568,8 @@ static AVStream *alloc_video_stream(FFMpegContext *context, RenderData *rd, int if (context->ffmpeg_crf >= 0) { ffmpeg_dict_set_int(&opts, "crf", context->ffmpeg_crf); - } else { + } + else { c->bit_rate = context->ffmpeg_video_bitrate * 1000; c->rc_max_rate = rd->ffcodecdata.rc_max_rate * 1000; c->rc_min_rate = rd->ffcodecdata.rc_min_rate * 1000; @@ -576,8 +577,8 @@ static AVStream *alloc_video_stream(FFMpegContext *context, RenderData *rd, int } if (context->ffmpeg_preset) { - char const * preset_name; - switch(context->ffmpeg_preset) { + char const *preset_name; + switch (context->ffmpeg_preset) { case FFM_PRESET_ULTRAFAST: preset_name = "ultrafast"; break; case FFM_PRESET_SUPERFAST: preset_name = "superfast"; break; case FFM_PRESET_VERYFAST: preset_name = "veryfast"; break; diff --git a/source/blender/blenlib/BLI_kdopbvh.h b/source/blender/blenlib/BLI_kdopbvh.h index 91d39801645..ba565fca522 100644 --- a/source/blender/blenlib/BLI_kdopbvh.h +++ b/source/blender/blenlib/BLI_kdopbvh.h @@ -95,10 +95,6 @@ typedef void (*BVHTree_NearestPointCallback)(void *userdata, int index, const fl /* callback must update hit in case it finds a nearest successful hit */ typedef void (*BVHTree_RayCastCallback)(void *userdata, int index, const BVHTreeRay *ray, BVHTreeRayHit *hit); -/* callback must update nearest in case it finds a nearest result */ -typedef void (*BVHTree_NearestToRayCallback)(void *userdata, const float ray_co[3], const float ray_dir[3], - const float scale[3], int index, BVHTreeNearest *nearest); - /* callback to check if 2 nodes overlap (use thread if intersection results need to be stored) */ typedef bool (*BVHTree_OverlapCallback)(void *userdata, int index_a, int index_b, int thread); @@ -143,18 +139,6 @@ int BLI_bvhtree_find_nearest( BVHTree *tree, const float co[3], BVHTreeNearest *nearest, BVHTree_NearestPointCallback callback, void *userdata); -int BLI_bvhtree_find_nearest_to_ray_angle( - BVHTree *tree, const float co[3], const float dir[3], - const bool ray_is_normalized, const float scale[3], - BVHTreeNearest *nearest, - BVHTree_NearestToRayCallback callback, void *userdata); - -int BLI_bvhtree_find_nearest_to_ray( - BVHTree *tree, const float co[3], const float dir[3], - const bool ray_is_normalized, const float scale[3], - BVHTreeNearest *nearest, - BVHTree_NearestToRayCallback callback, void *userdata); - int BLI_bvhtree_ray_cast_ex( BVHTree *tree, const float co[3], const float dir[3], float radius, BVHTreeRayHit *hit, BVHTree_RayCastCallback callback, void *userdata, diff --git a/source/blender/blenlib/BLI_math_geom.h b/source/blender/blenlib/BLI_math_geom.h index 4a85e859c16..f1d9c9571f2 100644 --- a/source/blender/blenlib/BLI_math_geom.h +++ b/source/blender/blenlib/BLI_math_geom.h @@ -298,23 +298,6 @@ bool isect_ray_aabb_v3_simple( const float bb_min[3], const float bb_max[3], float *tmin, float *tmax); -struct NearestRayToAABB_Precalc { - float ray_origin[3]; - float ray_direction[3]; - float ray_inv_dir[3]; - float cdot_axis[3]; - float idiag_sq[3]; - bool sign[3]; -}; - -void dist_squared_ray_to_aabb_v3_precalc( - struct NearestRayToAABB_Precalc *data, - const float ray_origin[3], const float ray_direction[3]); -float dist_squared_ray_to_aabb_v3( - const struct NearestRayToAABB_Precalc *data, - const float bb_min[3], const float bb_max[3], - bool r_axis_closest[3]); - /* other */ bool isect_sweeping_sphere_tri_v3(const float p1[3], const float p2[3], const float radius, const float v0[3], const float v1[3], const float v2[3], float *r_lambda, float ipoint[3]); diff --git a/source/blender/blenlib/BLI_path_util.h b/source/blender/blenlib/BLI_path_util.h index baa1f792018..b6a55d34d14 100644 --- a/source/blender/blenlib/BLI_path_util.h +++ b/source/blender/blenlib/BLI_path_util.h @@ -39,14 +39,6 @@ extern "C" { struct ListBase; -#ifdef WIN32 -#define SEP '\\' -#define ALTSEP '/' -#else -#define SEP '/' -#define ALTSEP '\\' -#endif - void BLI_setenv(const char *env, const char *val) ATTR_NONNULL(1); void BLI_setenv_if_new(const char *env, const char *val) ATTR_NONNULL(1); @@ -60,7 +52,13 @@ void BLI_path_append(char *__restrict dst, const size_t maxlen, const char *__restrict file) ATTR_NONNULL(); void BLI_join_dirfile(char *__restrict string, const size_t maxlen, const char *__restrict dir, const char *__restrict file) ATTR_NONNULL(); +size_t BLI_path_join( + char *__restrict dst, const size_t dst_len, + const char *path_first, ...) ATTR_NONNULL(1, 3) ATTR_SENTINEL(0); const char *BLI_path_basename(const char *path) ATTR_NONNULL() ATTR_WARN_UNUSED_RESULT; +bool BLI_path_name_at_index( + const char *__restrict path, const int index, + int *__restrict r_offset, int *__restrict r_len) ATTR_NONNULL() ATTR_WARN_UNUSED_RESULT; #if 0 typedef enum bli_rebase_state { @@ -83,7 +81,6 @@ bool BLI_path_program_extensions_add_win32(char *name, const size_t maxlen); #endif bool BLI_path_program_search(char *fullname, const size_t maxlen, const char *name); -void BLI_getlastdir(const char *dir, char *last, const size_t maxlen); bool BLI_testextensie(const char *str, const char *ext) ATTR_NONNULL() ATTR_WARN_UNUSED_RESULT; bool BLI_testextensie_n(const char *str, ...) ATTR_NONNULL(1) ATTR_SENTINEL(0); bool BLI_testextensie_array(const char *str, const char **ext_array) ATTR_NONNULL() ATTR_WARN_UNUSED_RESULT; @@ -143,6 +140,18 @@ bool BLI_path_suffix(char *string, size_t maxlen, const char *suffix, const char # define FILE_MAX 1024 #endif +#ifdef WIN32 +# define SEP '\\' +# define ALTSEP '/' +# define SEP_STR "\\" +# define ALTSEP_STR "/" +#else +# define SEP '/' +# define ALTSEP '\\' +# define SEP_STR "/" +# define ALTSEP_STR "\\" +#endif + /* Parent and current dir helpers. */ #define FILENAME_PARENT ".." #define FILENAME_CURRENT "." diff --git a/source/blender/blenlib/BLI_rect.h b/source/blender/blenlib/BLI_rect.h index 59bf3644912..041679ef876 100644 --- a/source/blender/blenlib/BLI_rect.h +++ b/source/blender/blenlib/BLI_rect.h @@ -47,6 +47,8 @@ bool BLI_rcti_is_empty(const struct rcti *rect); bool BLI_rctf_is_empty(const struct rctf *rect); void BLI_rctf_init(struct rctf *rect, float xmin, float xmax, float ymin, float ymax); void BLI_rcti_init(struct rcti *rect, int xmin, int xmax, int ymin, int ymax); +void BLI_rctf_init_pt_radius(struct rctf *rect, const float xy[2], float size); +void BLI_rcti_init_pt_radius(struct rcti *rect, const int xy[2], int size); void BLI_rcti_init_minmax(struct rcti *rect); void BLI_rctf_init_minmax(struct rctf *rect); void BLI_rcti_do_minmax_v(struct rcti *rect, const int xy[2]); diff --git a/source/blender/blenlib/BLI_sys_types.h b/source/blender/blenlib/BLI_sys_types.h index 7929e1d6551..9d8222c575d 100644 --- a/source/blender/blenlib/BLI_sys_types.h +++ b/source/blender/blenlib/BLI_sys_types.h @@ -80,6 +80,11 @@ typedef uint64_t u_int64_t; #include <stddef.h> /* size_t define */ #include <stdbool.h> +typedef unsigned int uint; +typedef unsigned short ushort; +typedef unsigned long ulong; +typedef unsigned char uchar; + #ifdef __cplusplus } #endif diff --git a/source/blender/blenlib/BLI_task.h b/source/blender/blenlib/BLI_task.h index 967e0be6d0a..c3c587275e1 100644 --- a/source/blender/blenlib/BLI_task.h +++ b/source/blender/blenlib/BLI_task.h @@ -81,6 +81,7 @@ typedef void (*TaskFreeFunction)(TaskPool *__restrict pool, void *taskdata, int TaskPool *BLI_task_pool_create(TaskScheduler *scheduler, void *userdata); TaskPool *BLI_task_pool_create_background(TaskScheduler *scheduler, void *userdata); +TaskPool *BLI_task_pool_create_suspended(TaskScheduler *scheduler, void *userdata); void BLI_task_pool_free(TaskPool *pool); void BLI_task_pool_push_ex( @@ -95,14 +96,6 @@ void BLI_task_pool_push_from_thread(TaskPool *pool, TaskRunFunction run, void BLI_task_pool_work_and_wait(TaskPool *pool); /* cancel all tasks, keep worker threads running */ void BLI_task_pool_cancel(TaskPool *pool); -/* stop all worker threads */ -void BLI_task_pool_stop(TaskPool *pool); - -/* get number of threads allowed to be used by this pool */ -int BLI_pool_get_num_threads(TaskPool *pool); - -/* set number of threads allowed to be used by this pool */ -void BLI_pool_set_num_threads(TaskPool *pool, int num_threads); /* for worker threads, test if canceled */ bool BLI_task_pool_canceled(TaskPool *pool); @@ -113,9 +106,6 @@ void *BLI_task_pool_userdata(TaskPool *pool); /* optional mutex to use from run function */ ThreadMutex *BLI_task_pool_user_mutex(TaskPool *pool); -/* number of tasks done, for stats, don't use this to make decisions */ -size_t BLI_task_pool_tasks_done(TaskPool *pool); - /* Parallel for routines */ typedef void (*TaskParallelRangeFunc)(void *userdata, const int iter); typedef void (*TaskParallelRangeFuncEx)(void *userdata, void *userdata_chunk, const int iter, const int thread_id); diff --git a/source/blender/blenlib/PIL_time_utildefines.h b/source/blender/blenlib/PIL_time_utildefines.h index 9157e04a7bf..412cfb3a090 100644 --- a/source/blender/blenlib/PIL_time_utildefines.h +++ b/source/blender/blenlib/PIL_time_utildefines.h @@ -80,9 +80,10 @@ } \ const float _delta_##var = TIMEIT_VALUE(var); \ _sum_##var += _delta_##var; \ + _num_##var++; \ printf("time end (" #var "): %.6f" " " AT "\n", _delta_##var); \ - printf("time averaged (" #var "): %.6f" " " AT "\n", \ - (_sum_##var / ++_num_##var)); \ + printf("time averaged (" #var "): %.6f (total: %.6f, in %d runs)\n", \ + (_sum_##var / _num_##var), _sum_##var, (int)_num_##var); \ fflush(stdout); \ } (void)0 diff --git a/source/blender/blenlib/intern/BLI_kdopbvh.c b/source/blender/blenlib/intern/BLI_kdopbvh.c index b14007a88cb..19d9711922e 100644 --- a/source/blender/blenlib/intern/BLI_kdopbvh.c +++ b/source/blender/blenlib/intern/BLI_kdopbvh.c @@ -159,29 +159,6 @@ typedef struct BVHRayCastData { BVHTreeRayHit hit; } BVHRayCastData; -typedef struct BVHNearestRayData { - BVHTree *tree; - BVHTree_NearestToRayCallback callback; - void *userdata; - - struct { - bool sign[3]; - float origin[3]; - float direction[3]; - - float direction_scaled_square[3]; - float inv_dir[3]; - - float cdot_axis[3]; - } ray; - - bool pick_smallest[3]; - - BVHTreeNearest nearest; - - float scale[3]; -} BVHNearestRayData; - /** \} */ @@ -1900,453 +1877,6 @@ void BLI_bvhtree_ray_cast_all( /* -------------------------------------------------------------------- */ -/** \name BLI_bvhtree_find_nearest_to_ray functions - * - * \{ */ - -static void dist_squared_ray_to_aabb_scaled_v3_precalc( - BVHNearestRayData *data, - const float ray_origin[3], const float ray_direction[3], - const bool ray_is_normalized, const float scale[3]) -{ - if (scale) { - copy_v3_v3(data->scale, scale); - } - else { - copy_v3_fl(data->scale, 1.0f); - } - /* un-normalize ray */ - if (ray_is_normalized && scale && - (data->scale[0] != 1.0f || data->scale[1] != 1.0f || data->scale[2] != 1.0f)) - { - data->ray.direction[0] = ray_direction[0] * data->scale[0]; - data->ray.direction[1] = ray_direction[1] * data->scale[1]; - data->ray.direction[2] = ray_direction[2] * data->scale[2]; - - mul_v3_v3fl(data->ray.direction, ray_direction, 1 / len_v3(data->ray.direction)); - } - else { - copy_v3_v3(data->ray.direction, ray_direction); - } - - float dir_sq[3]; - - for (int i = 0; i < 3; i++) { - data->ray.origin[i] = ray_origin[i]; - data->ray.inv_dir[i] = (data->ray.direction[i] != 0.0f) ? - (1.0f / data->ray.direction[i]) : FLT_MAX; - /* It has to be in function of `ray.inv_dir`, - * since the division of 1 by 0.0f, can be -inf or +inf */ - data->ray.sign[i] = (data->ray.inv_dir[i] < 0.0f); - - data->ray.direction_scaled_square[i] = data->ray.direction[i] * data->scale[i]; - - dir_sq[i] = SQUARE(data->ray.direction_scaled_square[i]); - - data->ray.direction_scaled_square[i] *= data->scale[i]; - } - - /* `diag_sq` Length square of each face diagonal */ - float diag_sq[3] = { - dir_sq[1] + dir_sq[2], - dir_sq[0] + dir_sq[2], - dir_sq[0] + dir_sq[1], - }; - - data->ray.cdot_axis[0] = (diag_sq[0] != 0.0f) ? data->ray.direction[0] / diag_sq[0] : FLT_MAX; - data->ray.cdot_axis[1] = (diag_sq[1] != 0.0f) ? data->ray.direction[1] / diag_sq[1] : FLT_MAX; - data->ray.cdot_axis[2] = (diag_sq[2] != 0.0f) ? data->ray.direction[2] / diag_sq[2] : FLT_MAX; -} - -/** - * Returns the squared distance from a ray to a bound-box `AABB`. - * It is based on `fast_ray_nearest_hit` solution to obtain - * the coordinates of the nearest edge of Bound Box to the ray - */ -MINLINE float dist_squared_ray_to_aabb_scaled_v3__impl( - const BVHNearestRayData *data, - const float bv[6], float *r_depth_sq, bool r_axis_closest[3]) -{ - - /* `tmin` is a vector that has the smaller distances to each of the - * infinite planes of the `AABB` faces (hit in nearest face X plane, - * nearest face Y plane and nearest face Z plane) */ - float local_bvmin[3], local_bvmax[3]; - - if (data->ray.sign[0]) { - local_bvmin[0] = bv[1]; - local_bvmax[0] = bv[0]; - } - else { - local_bvmin[0] = bv[0]; - local_bvmax[0] = bv[1]; - } - - if (data->ray.sign[1]) { - local_bvmin[1] = bv[3]; - local_bvmax[1] = bv[2]; - } - else { - local_bvmin[1] = bv[2]; - local_bvmax[1] = bv[3]; - } - - if (data->ray.sign[2]) { - local_bvmin[2] = bv[5]; - local_bvmax[2] = bv[4]; - } - else { - local_bvmin[2] = bv[4]; - local_bvmax[2] = bv[5]; - } - - sub_v3_v3(local_bvmin, data->ray.origin); - sub_v3_v3(local_bvmax, data->ray.origin); - - const float tmin[3] = { - local_bvmin[0] * data->ray.inv_dir[0], - local_bvmin[1] * data->ray.inv_dir[1], - local_bvmin[2] * data->ray.inv_dir[2], - }; - - /* `tmax` is a vector that has the longer distances to each of the - * infinite planes of the `AABB` faces (hit in farthest face X plane, - * farthest face Y plane and farthest face Z plane) */ - const float tmax[3] = { - local_bvmax[0] * data->ray.inv_dir[0], - local_bvmax[1] * data->ray.inv_dir[1], - local_bvmax[2] * data->ray.inv_dir[2], - }; - /* `v1` and `v3` is be the coordinates of the nearest `AABB` edge to the ray*/ - float v1[3], v2[3]; - /* `rtmin` is the highest value of the smaller distances. == max_axis_v3(tmin) - * `rtmax` is the lowest value of longer distances. == min_axis_v3(tmax)*/ - float rtmin, rtmax, mul; - /* `main_axis` is the axis equivalent to edge close to the ray */ - int main_axis; - - r_axis_closest[0] = false; - r_axis_closest[1] = false; - r_axis_closest[2] = false; - - /* *** min_axis_v3(tmax) *** */ - if ((tmax[0] <= tmax[1]) && (tmax[0] <= tmax[2])) { - // printf("# Hit in X %s\n", data->sign[0] ? "min", "max"); - rtmax = tmax[0]; - v1[0] = v2[0] = local_bvmax[0]; - mul = local_bvmax[0] * data->ray.direction_scaled_square[0]; - main_axis = 3; - r_axis_closest[0] = data->ray.sign[0]; - } - else if ((tmax[1] <= tmax[0]) && (tmax[1] <= tmax[2])) { - // printf("# Hit in Y %s\n", data->sign[1] ? "min", "max"); - rtmax = tmax[1]; - v1[1] = v2[1] = local_bvmax[1]; - mul = local_bvmax[1] * data->ray.direction_scaled_square[1]; - main_axis = 2; - r_axis_closest[1] = data->ray.sign[1]; - } - else { - // printf("# Hit in Z %s\n", data->sign[2] ? "min", "max"); - rtmax = tmax[2]; - v1[2] = v2[2] = local_bvmax[2]; - mul = local_bvmax[2] * data->ray.direction_scaled_square[2]; - main_axis = 1; - r_axis_closest[2] = data->ray.sign[2]; - } - - /* *** max_axis_v3(tmin) *** */ - if ((tmin[0] >= tmin[1]) && (tmin[0] >= tmin[2])) { - // printf("# To X %s\n", data->sign[0] ? "max", "min"); - rtmin = tmin[0]; - v1[0] = v2[0] = local_bvmin[0]; - mul += local_bvmin[0] * data->ray.direction_scaled_square[0]; - main_axis -= 3; - r_axis_closest[0] = !data->ray.sign[0]; - } - else if ((tmin[1] >= tmin[0]) && (tmin[1] >= tmin[2])) { - // printf("# To Y %s\n", data->sign[1] ? "max", "min"); - rtmin = tmin[1]; - v1[1] = v2[1] = local_bvmin[1]; - mul += local_bvmin[1] * data->ray.direction_scaled_square[1]; - main_axis -= 1; - r_axis_closest[1] = !data->ray.sign[1]; - } - else { - // printf("# To Z %s\n", data->sign[2] ? "max", "min"); - rtmin = tmin[2]; - v1[2] = v2[2] = local_bvmin[2]; - mul += local_bvmin[2] * data->ray.direction_scaled_square[2]; - main_axis -= 2; - r_axis_closest[2] = !data->ray.sign[2]; - } - /* *** end min/max axis *** */ - - if (main_axis < 0) - main_axis += 3; - - /* if rtmin < rtmax, ray intersect `AABB` */ - if (rtmin <= rtmax) { -#ifdef IGNORE_BEHIND_RAY - /* `if rtmax < depth_min`, the whole `AABB` is behind us */ - if (rtmax < min_depth) { - return fallback; - } -#endif - const float proj = rtmin * data->ray.direction[main_axis]; - - if (data->ray.sign[main_axis]) - r_axis_closest[main_axis] = (proj - local_bvmax[main_axis]) < (local_bvmin[main_axis] - proj); - else - r_axis_closest[main_axis] = (proj - local_bvmin[main_axis]) < (local_bvmax[main_axis] - proj); - - //if (r_depth_sq) - // *r_depth_sq = SQUARE(rtmin); - - return 0.0f; - } -#ifdef IGNORE_BEHIND_RAY - /* `if rtmin < depth_min`, the whole `AABB` is behing us */ - else if (rtmin < min_depth) { - return fallback; - } -#endif - - if (data->ray.sign[main_axis]) { - v1[main_axis] = local_bvmax[main_axis]; - v2[main_axis] = local_bvmin[main_axis]; - } - else { - v1[main_axis] = local_bvmin[main_axis]; - v2[main_axis] = local_bvmax[main_axis]; - } - { - /* `proj` equals to nearest point on the ray closest to the edge `v1 v2` of the `AABB`. */ - const float proj = mul * data->ray.cdot_axis[main_axis]; - float depth_sq, r_point[3]; - if (v1[main_axis] > proj) { /* the nearest point to the ray is the point v1 */ - r_axis_closest[main_axis] = true; - /* `depth` is equivalent the distance of the the projection of v1 on the ray */ - depth_sq = mul + data->ray.direction_scaled_square[main_axis] * v1[main_axis]; - - copy_v3_v3(r_point, v1); - } - else if (v2[main_axis] < proj) { /* the nearest point of the ray is the point v2 */ - r_axis_closest[main_axis] = false; - - depth_sq = mul + data->ray.direction_scaled_square[main_axis] * v2[main_axis]; - - copy_v3_v3(r_point, v2); - } - else { /* the nearest point of the ray is on the edge of the `AABB`. */ - r_axis_closest[main_axis] = (proj - v1[main_axis]) < (v2[main_axis] - proj); - - depth_sq = mul + data->ray.direction_scaled_square[main_axis] * proj; -#if 0 - r_point[0] = main_axis == 0 ? proj : v2[0]; - r_point[1] = main_axis == 1 ? proj : v2[1]; - r_point[2] = main_axis == 2 ? proj : v2[2]; -#else - v2[main_axis] = proj; - copy_v3_v3(r_point, v2); -#endif - } - depth_sq *= depth_sq; - - if (r_depth_sq) - *r_depth_sq = depth_sq; - - /* TODO: scale can be optional */ - r_point[0] *= data->scale[0]; - r_point[1] *= data->scale[1]; - r_point[2] *= data->scale[2]; - - return len_squared_v3(r_point) - depth_sq; - } -} - -/** - * <pre> - * + r_point - * | - * | dist - * | - * +----depth----+orig <-- dir - * - * tangent = dist/depth - * </pre> - */ -static float calc_tangent_sq(BVHNearestRayData *data, BVHNode *node) -{ - float depth_sq; - const float dist_sq = dist_squared_ray_to_aabb_scaled_v3__impl( - data, node->bv, &depth_sq, data->pick_smallest); - - return (dist_sq != 0.0f) ? (dist_sq / depth_sq) : 0.0f; -} - -static float calc_dist_sq_to_ray(BVHNearestRayData *data, BVHNode *node) -{ - return dist_squared_ray_to_aabb_scaled_v3__impl( - data, node->bv, NULL, - data->pick_smallest); -} - -static void dfs_find_lowest_tangent_dfs(BVHNearestRayData *data, BVHNode *node) -{ - if (node->totnode == 0) { - if (data->callback) { - data->callback(data->userdata, data->ray.origin, data->ray.direction, - data->scale, node->index, &data->nearest); - } - else { - data->nearest.index = node->index; - data->nearest.dist_sq = calc_tangent_sq(data, node); - /* TODO: return a value to the data->nearest.co - * not urgent however since users currently define own callbacks */ - } - } - else { - int i; - /* First pick the closest node to dive on */ - if (data->pick_smallest[node->main_axis]) { - for (i = 0; i != node->totnode; i++) { - if (calc_tangent_sq(data, node->children[i]) < data->nearest.dist_sq) { - dfs_find_lowest_tangent_dfs(data, node->children[i]); - } - } - } - else { - for (i = node->totnode - 1; i >= 0; i--) { - if (calc_tangent_sq(data, node->children[i]) < data->nearest.dist_sq) { - dfs_find_lowest_tangent_dfs(data, node->children[i]); - } - } - } - } -} - -static void dfs_find_nearest_to_ray_dfs(BVHNearestRayData *data, BVHNode *node) -{ - if (node->totnode == 0) { - if (data->callback) { - data->callback(data->userdata, data->ray.origin, data->ray.direction, - data->scale, node->index, &data->nearest); - } - else { - data->nearest.index = node->index; - data->nearest.dist_sq = calc_dist_sq_to_ray(data, node); - /* TODO: return a value to the data->nearest.co - * not urgent however since users currently define own callbacks */ - } - } - else { - int i; - /* First pick the closest node to dive on */ - if (data->pick_smallest[node->main_axis]) { - for (i = 0; i != node->totnode; i++) { - if (calc_dist_sq_to_ray(data, node->children[i]) < data->nearest.dist_sq) { - dfs_find_nearest_to_ray_dfs(data, node->children[i]); - } - } - } - else { - for (i = node->totnode - 1; i >= 0; i--) { - if (calc_dist_sq_to_ray(data, node->children[i]) < data->nearest.dist_sq) { - dfs_find_nearest_to_ray_dfs(data, node->children[i]); - } - } - } - } -} - -/** - * Returns the point whose tangent defined by the angle between the point and ray is the lowest - * nearest.dist_sq returns the angle's tangent - */ -int BLI_bvhtree_find_nearest_to_ray_angle( - BVHTree *tree, const float co[3], const float dir[3], - const bool ray_is_normalized, const float scale[3], - BVHTreeNearest *nearest, - BVHTree_NearestToRayCallback callback, void *userdata) -{ - BVHNearestRayData data; - BVHNode *root = tree->nodes[tree->totleaf]; - - data.tree = tree; - - data.callback = callback; - data.userdata = userdata; - - dist_squared_ray_to_aabb_scaled_v3_precalc(&data, co, dir, ray_is_normalized, scale); - - if (nearest) { - memcpy(&data.nearest, nearest, sizeof(*nearest)); - } - else { - data.nearest.index = -1; - data.nearest.dist_sq = FLT_MAX; - } - - /* dfs search */ - if (root) { - if (calc_tangent_sq(&data, root) < data.nearest.dist_sq) - dfs_find_lowest_tangent_dfs(&data, root); - } - - /* copy back results */ - if (nearest) { - memcpy(nearest, &data.nearest, sizeof(*nearest)); - } - - return data.nearest.index; -} - -/* return the nearest point to ray */ -int BLI_bvhtree_find_nearest_to_ray( - BVHTree *tree, const float co[3], const float dir[3], - const bool ray_is_normalized, const float scale[3], - BVHTreeNearest *nearest, - BVHTree_NearestToRayCallback callback, void *userdata) -{ - BVHNearestRayData data; - BVHNode *root = tree->nodes[tree->totleaf]; - - data.tree = tree; - - data.callback = callback; - data.userdata = userdata; - - dist_squared_ray_to_aabb_scaled_v3_precalc(&data, co, dir, ray_is_normalized, scale); - - if (nearest) { - memcpy(&data.nearest, nearest, sizeof(*nearest)); - } - else { - data.nearest.index = -1; - data.nearest.dist_sq = FLT_MAX; - } - - /* dfs search */ - if (root) { - if (calc_dist_sq_to_ray(&data, root) < data.nearest.dist_sq) { - dfs_find_nearest_to_ray_dfs(&data, root); - } - } - - /* copy back results */ - if (nearest) { - memcpy(nearest, &data.nearest, sizeof(*nearest)); - } - - return data.nearest.index; -} - -/** \} */ - - -/* -------------------------------------------------------------------- */ - /** \name BLI_bvhtree_range_query * * Allocs and fills an array with the indexs of node that are on the given spherical range (center, radius). diff --git a/source/blender/blenlib/intern/array_store.c b/source/blender/blenlib/intern/array_store.c index 21ddddad32e..295b39c1a2f 100644 --- a/source/blender/blenlib/intern/array_store.c +++ b/source/blender/blenlib/intern/array_store.c @@ -217,15 +217,12 @@ /** \name Internal Structs * \{ */ -typedef unsigned int uint; -typedef unsigned char ubyte; - typedef uint64_t hash_key; typedef struct BArrayInfo { size_t chunk_stride; - uint chunk_count; + // uint chunk_count; /* UNUSED (other values are derived from this) */ /* pre-calculated */ size_t chunk_byte_size; @@ -291,7 +288,7 @@ typedef struct BChunkList { /* a chunk of an array */ typedef struct BChunk { - const ubyte *data; + const uchar *data; size_t data_len; /** number of #BChunkList using this. */ int users; @@ -332,7 +329,7 @@ static size_t bchunk_list_size(const BChunkList *chunk_list); * \{ */ static BChunk *bchunk_new( - BArrayMemory *bs_mem, const ubyte *data, const size_t data_len) + BArrayMemory *bs_mem, const uchar *data, const size_t data_len) { BChunk *chunk = BLI_mempool_alloc(bs_mem->chunk); chunk->data = data; @@ -345,9 +342,9 @@ static BChunk *bchunk_new( } static BChunk *bchunk_new_copydata( - BArrayMemory *bs_mem, const ubyte *data, const size_t data_len) + BArrayMemory *bs_mem, const uchar *data, const size_t data_len) { - ubyte *data_copy = MEM_mallocN(data_len, __func__); + uchar *data_copy = MEM_mallocN(data_len, __func__); memcpy(data_copy, data, data_len); return bchunk_new(bs_mem, data_copy, data_len); } @@ -367,7 +364,7 @@ static void bchunk_decref( static bool bchunk_data_compare( const BChunk *chunk, - const ubyte *data_base, const size_t data_base_len, + const uchar *data_base, const size_t data_base_len, const size_t offset) { if (offset + (size_t)chunk->data_len <= data_base_len) { @@ -426,14 +423,14 @@ static void bchunk_list_decref( #ifdef USE_VALIDATE_LIST_DATA_PARTIAL static size_t bchunk_list_data_check( - const BChunkList *chunk_list, const ubyte *data) + const BChunkList *chunk_list, const uchar *data) { - size_t total_size = 0; + size_t offset = 0; for (BChunkRef *cref = chunk_list->chunk_refs.first; cref; cref = cref->next) { - if (memcmp(&data[total_size], cref->link->data, cref->link->data_len) != 0) { + if (memcmp(&data[offset], cref->link->data, cref->link->data_len) != 0) { return false; } - total_size += cref->link->data_len; + offset += cref->link->data_len; } return true; } @@ -466,7 +463,7 @@ static void bchunk_list_ensure_min_size_last( chunk_list->chunk_refs.last = cref->prev; chunk_list->chunk_refs_len -= 1; - ubyte *data_merge = MEM_mallocN(data_merge_len, __func__); + uchar *data_merge = MEM_mallocN(data_merge_len, __func__); memcpy(data_merge, chunk_prev->data, chunk_prev->data_len); memcpy(&data_merge[chunk_prev->data_len], chunk_curr->data, chunk_curr->data_len); @@ -487,8 +484,8 @@ static void bchunk_list_ensure_min_size_last( /* merge and split */ const size_t data_prev_len = split; const size_t data_curr_len = data_merge_len - split; - ubyte *data_prev = MEM_mallocN(data_prev_len, __func__); - ubyte *data_curr = MEM_mallocN(data_curr_len, __func__); + uchar *data_prev = MEM_mallocN(data_prev_len, __func__); + uchar *data_curr = MEM_mallocN(data_curr_len, __func__); if (data_prev_len <= chunk_prev->data_len) { const size_t data_curr_shrink_len = chunk_prev->data_len - data_prev_len; @@ -597,11 +594,10 @@ static void bchunk_list_append_only( static void bchunk_list_append_data( const BArrayInfo *info, BArrayMemory *bs_mem, BChunkList *chunk_list, - const ubyte *data, const size_t data_len) + const uchar *data, const size_t data_len) { BLI_assert(data_len != 0); - // printf("data_len: %d\n", data_len); #ifdef USE_MERGE_CHUNKS BLI_assert(data_len <= info->chunk_byte_size_max); @@ -613,13 +609,13 @@ static void bchunk_list_append_data( const size_t data_merge_len = chunk_prev->data_len + data_len; /* realloc for single user */ if (cref->link->users == 1) { - ubyte *data_merge = MEM_reallocN((void *)cref->link->data, data_merge_len); + uchar *data_merge = MEM_reallocN((void *)cref->link->data, data_merge_len); memcpy(&data_merge[chunk_prev->data_len], data, data_len); cref->link->data = data_merge; cref->link->data_len = data_merge_len; } else { - ubyte *data_merge = MEM_mallocN(data_merge_len, __func__); + uchar *data_merge = MEM_mallocN(data_merge_len, __func__); memcpy(data_merge, chunk_prev->data, chunk_prev->data_len); memcpy(&data_merge[chunk_prev->data_len], data, data_len); cref->link = bchunk_new(bs_mem, data_merge, data_merge_len); @@ -639,7 +635,7 @@ static void bchunk_list_append_data( /* don't run this, instead preemptively avoid creating a chunk only to merge it (above). */ #if 0 #ifdef USE_MERGE_CHUNKS - bchunk_list_ensure_min_size_last(info, bs_mem, chunk_list, chunk_size_min); + bchunk_list_ensure_min_size_last(info, bs_mem, chunk_list); #endif #endif } @@ -654,7 +650,7 @@ static void bchunk_list_append_data( static void bchunk_list_append_data_n( const BArrayInfo *info, BArrayMemory *bs_mem, BChunkList *chunk_list, - const ubyte *data, size_t data_len) + const uchar *data, size_t data_len) { size_t data_trim_len, data_last_chunk_len; bchunk_list_calc_trim_len(info, data_len, &data_trim_len, &data_last_chunk_len); @@ -714,7 +710,7 @@ static void bchunk_list_append( static void bchunk_list_fill_from_array( const BArrayInfo *info, BArrayMemory *bs_mem, BChunkList *chunk_list, - const ubyte *data, + const uchar *data, const size_t data_len) { BLI_assert(BLI_listbase_is_empty(&chunk_list->chunk_refs)); @@ -765,13 +761,13 @@ static void bchunk_list_fill_from_array( #define HASH_INIT (5381) -BLI_INLINE uint hash_data_single(const ubyte p) +BLI_INLINE uint hash_data_single(const uchar p) { return (HASH_INIT << 5) + HASH_INIT + (unsigned int)p; } /* hash bytes, from BLI_ghashutil_strhash_n */ -static uint hash_data(const ubyte *key, size_t n) +static uint hash_data(const uchar *key, size_t n) { const signed char *p; unsigned int h = HASH_INIT; @@ -788,7 +784,7 @@ static uint hash_data(const ubyte *key, size_t n) #ifdef USE_HASH_TABLE_ACCUMULATE static void hash_array_from_data( - const BArrayInfo *info, const ubyte *data_slice, const size_t data_slice_len, + const BArrayInfo *info, const uchar *data_slice, const size_t data_slice_len, hash_key *hash_array) { if (info->chunk_stride != 1) { @@ -877,7 +873,7 @@ static void hash_accum_single(hash_key *hash_array, const size_t hash_array_len, static hash_key key_from_chunk_ref( const BArrayInfo *info, const BChunkRef *cref, - /* avoid reallicating each time */ + /* avoid reallocating each time */ hash_key *hash_store, const size_t hash_store_len) { /* in C, will fill in a reusable array */ @@ -899,7 +895,7 @@ static hash_key key_from_chunk_ref( key = hash_store[0]; /* cache the key */ - if (key == HASH_TABLE_KEY_UNSET) { + if (UNLIKELY(key == HASH_TABLE_KEY_UNSET)) { key = HASH_TABLE_KEY_FALLBACK; } chunk->key = key; @@ -929,12 +925,12 @@ static hash_key key_from_chunk_ref( static const BChunkRef *table_lookup( const BArrayInfo *info, BTableRef **table, const size_t table_len, const size_t i_table_start, - const ubyte *data, const size_t data_len, const size_t offset, const hash_key *table_hash_array) + const uchar *data, const size_t data_len, const size_t offset, const hash_key *table_hash_array) { size_t size_left = data_len - offset; hash_key key = table_hash_array[((offset - i_table_start) / info->chunk_stride)]; size_t key_index = (size_t)(key % (hash_key)table_len); - for (BTableRef *tref = table[key_index]; tref; tref = tref->next) { + for (const BTableRef *tref = table[key_index]; tref; tref = tref->next) { const BChunkRef *cref = tref->cref; #ifdef USE_HASH_TABLE_KEY_CACHE if (cref->link->key == key) @@ -985,7 +981,7 @@ static hash_key key_from_chunk_ref(const BArrayInfo *info, const BChunkRef *cref static const BChunkRef *table_lookup( const BArrayInfo *info, BTableRef **table, const size_t table_len, const uint UNUSED(i_table_start), - const ubyte *data, const size_t data_len, const size_t offset, const hash_key *UNUSED(table_hash_array)) + const uchar *data, const size_t data_len, const size_t offset, const hash_key *UNUSED(table_hash_array)) { const size_t data_hash_len = BCHUNK_HASH_LEN * info->chunk_stride; /* TODO, cache */ @@ -1025,7 +1021,7 @@ static const BChunkRef *table_lookup( */ static BChunkList *bchunk_list_from_data_merge( const BArrayInfo *info, BArrayMemory *bs_mem, - const ubyte *data, const size_t data_len_original, + const uchar *data, const size_t data_len_original, const BChunkList *chunk_list_reference) { ASSERT_CHUNKLIST_SIZE(chunk_list_reference, chunk_list_reference->total_size); @@ -1042,10 +1038,8 @@ static BChunkList *bchunk_list_from_data_merge( size_t i_prev = 0; #ifdef USE_FASTPATH_CHUNKS_FIRST - bool full_match = false; - { - full_match = true; + bool full_match = true; const BChunkRef *cref = chunk_list_reference->chunk_refs.first; while (i_prev < data_len_original) { @@ -1433,7 +1427,7 @@ BArrayStore *BLI_array_store_create( BArrayStore *bs = MEM_callocN(sizeof(BArrayStore), __func__); bs->info.chunk_stride = stride; - bs->info.chunk_count = chunk_count; + // bs->info.chunk_count = chunk_count; bs->info.chunk_byte_size = chunk_count * stride; #ifdef USE_MERGE_CHUNKS @@ -1579,7 +1573,7 @@ BArrayState *BLI_array_store_state_add( if (state_reference) { chunk_list = bchunk_list_from_data_merge( &bs->info, &bs->memory, - (const ubyte *)data, data_len, + (const uchar *)data, data_len, /* re-use reference chunks */ state_reference->chunk_list); } @@ -1588,7 +1582,7 @@ BArrayState *BLI_array_store_state_add( bchunk_list_fill_from_array( &bs->info, &bs->memory, chunk_list, - (const ubyte *)data, data_len); + (const uchar *)data, data_len); } chunk_list->users += 1; @@ -1655,7 +1649,7 @@ void BLI_array_store_state_data_get( BLI_assert(data_test_len == state->chunk_list->total_size); #endif - ubyte *data_step = (ubyte *)data; + uchar *data_step = (uchar *)data; for (BChunkRef *cref = state->chunk_list->chunk_refs.first; cref; cref = cref->next) { BLI_assert(cref->link->users > 0); memcpy(data_step, cref->link->data, cref->link->data_len); diff --git a/source/blender/blenlib/intern/math_geom.c b/source/blender/blenlib/intern/math_geom.c index aeb6a550cd9..58699a0593b 100644 --- a/source/blender/blenlib/intern/math_geom.c +++ b/source/blender/blenlib/intern/math_geom.c @@ -2337,224 +2337,6 @@ bool isect_ray_aabb_v3_simple( } } -void dist_squared_ray_to_aabb_v3_precalc( - struct NearestRayToAABB_Precalc *data, - const float ray_origin[3], const float ray_direction[3]) -{ - float dir_sq[3]; - - for (int i = 0; i < 3; i++) { - data->ray_origin[i] = ray_origin[i]; - data->ray_direction[i] = ray_direction[i]; - data->ray_inv_dir[i] = (data->ray_direction[i] != 0.0f) ? (1.0f / data->ray_direction[i]) : FLT_MAX; - /* It has to be a function of `ray_inv_dir`, - * since the division of 1 by 0.0f, can be -inf or +inf */ - data->sign[i] = (data->ray_inv_dir[i] < 0.0f); - - dir_sq[i] = SQUARE(data->ray_direction[i]); - } - - /* `diag_sq` Length square of each face diagonal */ - float diag_sq[3] = { - dir_sq[1] + dir_sq[2], - dir_sq[0] + dir_sq[2], - dir_sq[0] + dir_sq[1], - }; - data->idiag_sq[0] = (diag_sq[0] > FLT_EPSILON) ? (1.0f / diag_sq[0]) : FLT_MAX; - data->idiag_sq[1] = (diag_sq[1] > FLT_EPSILON) ? (1.0f / diag_sq[1]) : FLT_MAX; - data->idiag_sq[2] = (diag_sq[2] > FLT_EPSILON) ? (1.0f / diag_sq[2]) : FLT_MAX; - - data->cdot_axis[0] = data->ray_direction[0] * data->idiag_sq[0]; - data->cdot_axis[1] = data->ray_direction[1] * data->idiag_sq[1]; - data->cdot_axis[2] = data->ray_direction[2] * data->idiag_sq[2]; -} - -/** - * Returns the squared distance from a ray to a bound-box `AABB`. - * It is based on `fast_ray_nearest_hit` solution to obtain - * the coordinates of the nearest edge of Bound Box to the ray - */ -float dist_squared_ray_to_aabb_v3( - const struct NearestRayToAABB_Precalc *data, - const float bb_min[3], const float bb_max[3], - bool r_axis_closest[3]) -{ - /* `tmin` is a vector that has the smaller distances to each of the - * infinite planes of the `AABB` faces (hit in nearest face X plane, - * nearest face Y plane and nearest face Z plane) */ - float local_bvmin[3], local_bvmax[3]; - - if (data->sign[0] == 0) { - local_bvmin[0] = bb_min[0] - data->ray_origin[0]; - local_bvmax[0] = bb_max[0] - data->ray_origin[0]; - } - else { - local_bvmin[0] = bb_max[0] - data->ray_origin[0]; - local_bvmax[0] = bb_min[0] - data->ray_origin[0]; - } - - if (data->sign[1] == 0) { - local_bvmin[1] = bb_min[1] - data->ray_origin[1]; - local_bvmax[1] = bb_max[1] - data->ray_origin[1]; - } - else { - local_bvmin[1] = bb_max[1] - data->ray_origin[1]; - local_bvmax[1] = bb_min[1] - data->ray_origin[1]; - } - - if (data->sign[2] == 0) { - local_bvmin[2] = bb_min[2] - data->ray_origin[2]; - local_bvmax[2] = bb_max[2] - data->ray_origin[2]; - } - else { - local_bvmin[2] = bb_max[2] - data->ray_origin[2]; - local_bvmax[2] = bb_min[2] - data->ray_origin[2]; - } - - const float tmin[3] = { - local_bvmin[0] * data->ray_inv_dir[0], - local_bvmin[1] * data->ray_inv_dir[1], - local_bvmin[2] * data->ray_inv_dir[2], - }; - - /* `tmax` is a vector that has the longer distances to each of the - * infinite planes of the `AABB` faces (hit in farthest face X plane, - * farthest face Y plane and farthest face Z plane) */ - const float tmax[3] = { - local_bvmax[0] * data->ray_inv_dir[0], - local_bvmax[1] * data->ray_inv_dir[1], - local_bvmax[2] * data->ray_inv_dir[2], - }; - /* `v1` and `v3` is be the coordinates of the nearest `AABB` edge to the ray*/ - float v1[3], v2[3]; - /* `rtmin` is the highest value of the smaller distances. == max_axis_v3(tmin) - * `rtmax` is the lowest value of longer distances. == min_axis_v3(tmax)*/ - float rtmin, rtmax, mul, rdist; - /* `main_axis` is the axis equivalent to edge close to the ray */ - int main_axis; - - r_axis_closest[0] = false; - r_axis_closest[1] = false; - r_axis_closest[2] = false; - - /* *** min_axis_v3(tmax) *** */ - if ((tmax[0] <= tmax[1]) && (tmax[0] <= tmax[2])) { - // printf("# Hit in X %s\n", data->sign[0] ? "min", "max"); - rtmax = tmax[0]; - v1[0] = v2[0] = local_bvmax[0]; - mul = local_bvmax[0] * data->ray_direction[0]; - main_axis = 3; - r_axis_closest[0] = data->sign[0]; - } - else if ((tmax[1] <= tmax[0]) && (tmax[1] <= tmax[2])) { - // printf("# Hit in Y %s\n", data->sign[1] ? "min", "max"); - rtmax = tmax[1]; - v1[1] = v2[1] = local_bvmax[1]; - mul = local_bvmax[1] * data->ray_direction[1]; - main_axis = 2; - r_axis_closest[1] = data->sign[1]; - } - else { - // printf("# Hit in Z %s\n", data->sign[2] ? "min", "max"); - rtmax = tmax[2]; - v1[2] = v2[2] = local_bvmax[2]; - mul = local_bvmax[2] * data->ray_direction[2]; - main_axis = 1; - r_axis_closest[2] = data->sign[2]; - } - - /* *** max_axis_v3(tmin) *** */ - if ((tmin[0] >= tmin[1]) && (tmin[0] >= tmin[2])) { - // printf("# To X %s\n", data->sign[0] ? "max", "min"); - rtmin = tmin[0]; - v1[0] = v2[0] = local_bvmin[0]; - mul += local_bvmin[0] * data->ray_direction[0]; - main_axis -= 3; - r_axis_closest[0] = !data->sign[0]; - } - else if ((tmin[1] >= tmin[0]) && (tmin[1] >= tmin[2])) { - // printf("# To Y %s\n", data->sign[1] ? "max", "min"); - rtmin = tmin[1]; - v1[1] = v2[1] = local_bvmin[1]; - mul += local_bvmin[1] * data->ray_direction[1]; - main_axis -= 1; - r_axis_closest[1] = !data->sign[1]; - } - else { - // printf("# To Z %s\n", data->sign[2] ? "max", "min"); - rtmin = tmin[2]; - v1[2] = v2[2] = local_bvmin[2]; - mul += local_bvmin[2] * data->ray_direction[2]; - main_axis -= 2; - r_axis_closest[2] = !data->sign[2]; - } - /* *** end min/max axis *** */ - - - /* `if rtmax < 0`, the whole `AABB` is behing us */ - if ((rtmax < 0.0f) && (rtmin < 0.0f)) { - return FLT_MAX; - } - - if (main_axis < 0) { - main_axis += 3; - } - - if (data->sign[main_axis] == 0) { - v1[main_axis] = local_bvmin[main_axis]; - v2[main_axis] = local_bvmax[main_axis]; - } - else { - v1[main_axis] = local_bvmax[main_axis]; - v2[main_axis] = local_bvmin[main_axis]; - } - - /* if rtmin < rtmax, ray intersect `AABB` */ - if (rtmin <= rtmax) { - const float proj = rtmin * data->ray_direction[main_axis]; - rdist = 0.0f; - r_axis_closest[main_axis] = (proj - v1[main_axis]) < (v2[main_axis] - proj); - } - else { - /* `proj` equals to nearest point on the ray closest to the edge `v1 v2` of the `AABB`. */ - const float proj = mul * data->cdot_axis[main_axis]; - float depth; - if (v1[main_axis] > proj) { /* the nearest point to the ray is the point v1 */ - /* `depth` is equivalent the distance from the origin to the point v1, - * Here's a faster way to calculate the dot product of v1 and ray - * (depth = dot_v3v3(v1, data->ray.direction))*/ - depth = mul + data->ray_direction[main_axis] * v1[main_axis]; - rdist = len_squared_v3(v1) - SQUARE(depth); - r_axis_closest[main_axis] = true; - } - else if (v2[main_axis] < proj) { /* the nearest point of the ray is the point v2 */ - depth = mul + data->ray_direction[main_axis] * v2[main_axis]; - rdist = len_squared_v3(v2) - SQUARE(depth); - r_axis_closest[main_axis] = false; - } - else { /* the nearest point of the ray is on the edge of the `AABB`. */ - float v[2]; - mul *= data->idiag_sq[main_axis]; - if (main_axis == 0) { - v[0] = (mul * data->ray_direction[1]) - v1[1]; - v[1] = (mul * data->ray_direction[2]) - v1[2]; - } - else if (main_axis == 1) { - v[0] = (mul * data->ray_direction[0]) - v1[0]; - v[1] = (mul * data->ray_direction[2]) - v1[2]; - } - else { - v[0] = (mul * data->ray_direction[0]) - v1[0]; - v[1] = (mul * data->ray_direction[1]) - v1[1]; - } - rdist = len_squared_v2(v); - r_axis_closest[main_axis] = (proj - v1[main_axis]) < (v2[main_axis] - proj); - } - } - - return rdist; -} - /* find closest point to p on line through (l1, l2) and return lambda, * where (0 <= lambda <= 1) when cp is in the line segment (l1, l2) */ @@ -3069,6 +2851,9 @@ bool barycentric_coords_v2(const float v1[2], const float v2[2], const float v3[ /** * \note: using #cross_tri_v2 means locations outside the triangle are correctly weighted + * + * \note This is *exactly* the same calculation as #resolve_tri_uv_v2, + * although it has double precision and is used for texture baking, so keep both. */ void barycentric_weights_v2(const float v1[2], const float v2[2], const float v3[2], const float co[2], float w[3]) { @@ -3108,9 +2893,11 @@ void barycentric_weights_v2_persp(const float v1[4], const float v2[4], const fl } } -/* same as #barycentric_weights_v2 but works with a quad, +/** + * same as #barycentric_weights_v2 but works with a quad, * note: untested for values outside the quad's bounds - * this is #interp_weights_poly_v2 expanded for quads only */ + * this is #interp_weights_poly_v2 expanded for quads only + */ void barycentric_weights_v2_quad(const float v1[2], const float v2[2], const float v3[2], const float v4[2], const float co[2], float w[4]) { @@ -3563,6 +3350,8 @@ void interp_cubic_v3(float x[3], float v[3], const float x1[3], const float v1[3 * Barycentric reverse * * Compute coordinates (u, v) for point \a st with respect to triangle (\a st0, \a st1, \a st2) + * + * \note same basic result as #barycentric_weights_v2, see it's comment for details. */ void resolve_tri_uv_v2(float r_uv[2], const float st[2], const float st0[2], const float st1[2], const float st2[2]) diff --git a/source/blender/blenlib/intern/path_util.c b/source/blender/blenlib/intern/path_util.c index 6644e6605a1..7b765cfa939 100644 --- a/source/blender/blenlib/intern/path_util.c +++ b/source/blender/blenlib/intern/path_util.c @@ -1167,30 +1167,6 @@ bool BLI_path_program_search( } /** - * Copies into *last the part of *dir following the second-last slash. - */ -void BLI_getlastdir(const char *dir, char *last, const size_t maxlen) -{ - const char *s = dir; - const char *lslash = NULL; - const char *prevslash = NULL; - while (*s) { - if ((*s == '\\') || (*s == '/')) { - prevslash = lslash; - lslash = s; - } - s++; - } - if (prevslash) { - BLI_strncpy(last, prevslash + 1, maxlen); - } - else { - BLI_strncpy(last, dir, maxlen); - } -} - - -/** * Sets the specified environment variable to the specified value, * and clears it if val == NULL. */ @@ -1615,6 +1591,90 @@ void BLI_join_dirfile(char *__restrict dst, const size_t maxlen, const char *__r } /** + * Join multiple strings into a path, ensuring only a single path separator between each, + * and trailing slash is kept. + * + * \note If you want a trailing slash, add ``SEP_STR`` as the last path argument, + * duplicate slashes will be cleaned up. + */ +size_t BLI_path_join(char *__restrict dst, const size_t dst_len, const char *path, ...) +{ + if (UNLIKELY(dst_len == 0)) { + return 0; + } + const size_t dst_last = dst_len - 1; + size_t ofs = BLI_strncpy_rlen(dst, path, dst_len); + + if (ofs == dst_last) { + return ofs; + } + + /* remove trailing slashes, unless there are _only_ trailing slashes + * (allow "//" as the first argument). */ + bool has_trailing_slash = false; + if (ofs != 0) { + size_t len = ofs; + while ((len != 0) && ELEM(path[len - 1], SEP, ALTSEP)) { + len -= 1; + } + if (len != 0) { + ofs = len; + } + has_trailing_slash = (path[len] != '\0'); + } + + va_list args; + va_start(args, path); + while ((path = (const char *) va_arg(args, const char *))) { + has_trailing_slash = false; + const char *path_init = path; + while (ELEM(path[0], SEP, ALTSEP)) { + path++; + } + size_t len = strlen(path); + if (len != 0) { + while ((len != 0) && ELEM(path[len - 1], SEP, ALTSEP)) { + len -= 1; + } + + if (len != 0) { + /* the very first path may have a slash at the end */ + if (ofs && !ELEM(dst[ofs - 1], SEP, ALTSEP)) { + dst[ofs++] = SEP; + if (ofs == dst_last) { + break; + } + } + has_trailing_slash = (path[len] != '\0'); + if (ofs + len >= dst_last) { + len = dst_last - ofs; + } + memcpy(&dst[ofs], path, len); + ofs += len; + if (ofs == dst_last) { + break; + } + } + } + else { + has_trailing_slash = (path_init != path); + } + } + va_end(args); + + if (has_trailing_slash) { + if ((ofs != dst_last) && (ofs != 0) && (ELEM(dst[ofs - 1], SEP, ALTSEP) == 0)) { + dst[ofs++] = SEP; + } + } + + BLI_assert(ofs <= dst_last); + dst[ofs] = '\0'; + + return ofs; +} + +/** * like pythons os.path.basename() * * \return The pointer into \a path string immediately after last slash, @@ -1626,6 +1686,71 @@ const char *BLI_path_basename(const char *path) return filename ? filename + 1 : path; } +/** + * Get an element of the path at an index, eg: + * "/some/path/file.txt" where an index of... + * - 0 or -3: "some" + * - 1 or -2: "path" + * - 2 or -1: "file.txt" + * + * Ignores multiple slashes at any point in the path (including start/end). + */ +bool BLI_path_name_at_index(const char *path, const int index, int *r_offset, int *r_len) +{ + if (index >= 0) { + int index_step = 0; + int prev = -1; + int i = 0; + while (true) { + const char c = path[i]; + if (ELEM(c, SEP, ALTSEP, '\0')) { + if (prev + 1 != i) { + prev += 1; + if (index_step == index) { + *r_offset = prev; + *r_len = i - prev; + /* printf("!!! %d %d\n", start, end); */ + return true; + } + index_step += 1; + } + if (c == '\0') { + break; + } + prev = i; + } + i += 1; + } + return false; + } + else { + /* negative number, reverse where -1 is the last element */ + int index_step = -1; + int prev = strlen(path); + int i = prev - 1; + while (true) { + const char c = i >= 0 ? path[i] : '\0'; + if (ELEM(c, SEP, ALTSEP, '\0')) { + if (prev - 1 != i) { + i += 1; + if (index_step == index) { + *r_offset = i; + *r_len = prev - i; + return true; + } + index_step -= 1; + } + if (c == '\0') { + break; + } + prev = i; + } + i -= 1; + } + return false; + } +} + /* UNUSED */ #if 0 /** diff --git a/source/blender/blenlib/intern/polyfill2d.c b/source/blender/blenlib/intern/polyfill2d.c index 8d9881e4539..2969b0eccf4 100644 --- a/source/blender/blenlib/intern/polyfill2d.c +++ b/source/blender/blenlib/intern/polyfill2d.c @@ -21,8 +21,15 @@ /** \file blender/blenlib/intern/polyfill2d.c * \ingroup bli * - * A simple implementation of the ear cutting algorithm - * to triangulate simple polygons without holes. + * An ear clipping algorithm to triangulate single boundary polygons. + * + * Details: + * + * - The algorithm guarantees all triangles are assigned (number of coords - 2) + * and that triangles will have non-overlapping indices (even for degenerate geometry). + * - Self-intersections are considered degenerate (resulting triangles will overlap). + * - While multiple polygons aren't supported, holes can still be defined using *key-holes* + * (where the polygon doubles back on its self with *exactly* matching coordinates). * * \note * @@ -74,6 +81,12 @@ typedef signed char eSign; #ifdef USE_KDTREE /** + * Spatial optimization for point-in-triangle intersection checks. + * The simple version of this algorithm is ``O(n^2)`` complexity + * (every point needing to check the triangle defined by every other point), + * Using a binary-tree reduces the complexity to ``O(n log n)`` + * plus some overhead of creating the tree. + * * This is a single purpose KDTree based on BLI_kdtree with some modifications * to better suit polyfill2d. * diff --git a/source/blender/blenlib/intern/rct.c b/source/blender/blenlib/intern/rct.c index ac73a981b45..fd24a00156d 100644 --- a/source/blender/blenlib/intern/rct.c +++ b/source/blender/blenlib/intern/rct.c @@ -351,6 +351,22 @@ void BLI_rcti_init(rcti *rect, int xmin, int xmax, int ymin, int ymax) } } +void BLI_rctf_init_pt_radius(rctf *rect, const float xy[2], float size) +{ + rect->xmin = xy[0] - size; + rect->xmax = xy[0] + size; + rect->ymin = xy[1] - size; + rect->ymax = xy[1] + size; +} + +void BLI_rcti_init_pt_radius(rcti *rect, const int xy[2], int size) +{ + rect->xmin = xy[0] - size; + rect->xmax = xy[0] + size; + rect->ymin = xy[1] - size; + rect->ymax = xy[1] + size; +} + void BLI_rcti_init_minmax(rcti *rect) { rect->xmin = rect->ymin = INT_MAX; diff --git a/source/blender/blenlib/intern/task.c b/source/blender/blenlib/intern/task.c index fc2d9674c2f..17e20f8fa18 100644 --- a/source/blender/blenlib/intern/task.c +++ b/source/blender/blenlib/intern/task.c @@ -48,6 +48,32 @@ */ #define MEMPOOL_SIZE 256 +/* Number of tasks which are pushed directly to local thread queue. + * + * This allows thread to fetch next task without locking the whole queue. + */ +#define LOCALQUEUE_SIZE 1 + +#ifndef NDEBUG +# define ASSERT_THREAD_ID(scheduler, thread_id) \ + do { \ + if (!BLI_thread_is_main()) { \ + TaskThread *thread = pthread_getspecific(scheduler->tls_id_key); \ + if (thread == NULL) { \ + BLI_assert(thread_id == 0); \ + } \ + else { \ + BLI_assert(thread_id == thread->id); \ + } \ + } \ + else { \ + BLI_assert(thread_id == 0); \ + } \ + } while (false) +#else +# define ASSERT_THREAD_ID(scheduler, thread_id) +#endif + typedef struct Task { struct Task *next, *prev; @@ -102,13 +128,16 @@ typedef struct TaskMemPoolStats { } TaskMemPoolStats; #endif +typedef struct TaskThreadLocalStorage { + TaskMemPool task_mempool; + int num_local_queue; + Task *local_queue[LOCALQUEUE_SIZE]; +} TaskThreadLocalStorage; + struct TaskPool { TaskScheduler *scheduler; volatile size_t num; - volatile size_t done; - size_t num_threads; - size_t currently_running_tasks; ThreadMutex num_mutex; ThreadCondition num_cond; @@ -116,6 +145,11 @@ struct TaskPool { ThreadMutex user_mutex; volatile bool do_cancel; + volatile bool do_work; + + volatile bool is_suspended; + ListBase suspended_queue; + size_t num_suspended; /* If set, this pool may never be work_and_wait'ed, which means TaskScheduler * has to use its special background fallback thread in case we are in @@ -123,16 +157,10 @@ struct TaskPool { */ bool run_in_background; - /* This pool is used for caching task pointers for thread id 0. - * This could either point to a global scheduler's task_mempool[0] if the - * pool is handled form the main thread or point to task_mempool_local - * otherwise. - * - * This way we solve possible threading conflicts accessing same global - * memory pool from multiple threads from which wait_work() is called. + /* This is a task scheduler's ID of a thread at which pool was constructed. + * It will be used to access task TLS. */ - TaskMemPool *task_mempool; - TaskMemPool task_mempool_local; + int thread_id; #ifdef DEBUG_STATS TaskMemPoolStats *mempool_stats; @@ -142,7 +170,6 @@ struct TaskPool { struct TaskScheduler { pthread_t *threads; struct TaskThread *task_threads; - TaskMemPool *task_mempool; int num_threads; bool background_thread_only; @@ -151,15 +178,19 @@ struct TaskScheduler { ThreadCondition queue_cond; volatile bool do_exit; + + /* NOTE: In pthread's TLS we store the whole TaskThread structure. */ + pthread_key_t tls_id_key; }; typedef struct TaskThread { TaskScheduler *scheduler; int id; + TaskThreadLocalStorage tls; } TaskThread; /* Helper */ -static void task_data_free(Task *task, const int thread_id) +BLI_INLINE void task_data_free(Task *task, const int thread_id) { if (task->free_taskdata) { if (task->freedata) { @@ -171,28 +202,42 @@ static void task_data_free(Task *task, const int thread_id) } } -BLI_INLINE TaskMemPool *get_task_mempool(TaskPool *pool, const int thread_id) +BLI_INLINE TaskThreadLocalStorage *get_task_tls(TaskPool *pool, + const int thread_id) { + TaskScheduler *scheduler = pool->scheduler; + BLI_assert(thread_id >= 0); + BLI_assert(thread_id <= scheduler->num_threads); if (thread_id == 0) { - return pool->task_mempool; + return &scheduler->task_threads[pool->thread_id].tls; + } + return &scheduler->task_threads[thread_id].tls; +} + +BLI_INLINE void free_task_tls(TaskThreadLocalStorage *tls) +{ + TaskMemPool *task_mempool = &tls->task_mempool; + for (int i = 0; i < task_mempool->num_tasks; ++i) { + MEM_freeN(task_mempool->tasks[i]); } - return &pool->scheduler->task_mempool[thread_id]; } static Task *task_alloc(TaskPool *pool, const int thread_id) { - assert(thread_id <= pool->scheduler->num_threads); + BLI_assert(thread_id <= pool->scheduler->num_threads); if (thread_id != -1) { - assert(thread_id >= 0); - TaskMemPool *mem_pool = get_task_mempool(pool, thread_id); + BLI_assert(thread_id >= 0); + BLI_assert(thread_id <= pool->scheduler->num_threads); + TaskThreadLocalStorage *tls = get_task_tls(pool, thread_id); + TaskMemPool *task_mempool = &tls->task_mempool; /* Try to re-use task memory from a thread local storage. */ - if (mem_pool->num_tasks > 0) { - --mem_pool->num_tasks; + if (task_mempool->num_tasks > 0) { + --task_mempool->num_tasks; /* Success! We've just avoided task allocation. */ #ifdef DEBUG_STATS pool->mempool_stats[thread_id].num_reuse++; #endif - return mem_pool->tasks[mem_pool->num_tasks]; + return task_mempool->tasks[task_mempool->num_tasks]; } /* We are doomed to allocate new task data. */ #ifdef DEBUG_STATS @@ -205,13 +250,14 @@ static Task *task_alloc(TaskPool *pool, const int thread_id) static void task_free(TaskPool *pool, Task *task, const int thread_id) { task_data_free(task, thread_id); - assert(thread_id >= 0); - assert(thread_id <= pool->scheduler->num_threads); - TaskMemPool *mem_pool = get_task_mempool(pool, thread_id); - if (mem_pool->num_tasks < MEMPOOL_SIZE - 1) { + BLI_assert(thread_id >= 0); + BLI_assert(thread_id <= pool->scheduler->num_threads); + TaskThreadLocalStorage *tls = get_task_tls(pool, thread_id); + TaskMemPool *task_mempool = &tls->task_mempool; + if (task_mempool->num_tasks < MEMPOOL_SIZE - 1) { /* Successfully allowed the task to be re-used later. */ - mem_pool->tasks[mem_pool->num_tasks] = task; - ++mem_pool->num_tasks; + task_mempool->tasks[task_mempool->num_tasks] = task; + ++task_mempool->num_tasks; } else { /* Local storage saturated, no other way than just discard @@ -237,8 +283,6 @@ static void task_pool_num_decrease(TaskPool *pool, size_t done) BLI_assert(pool->num >= done); pool->num -= done; - atomic_sub_and_fetch_z(&pool->currently_running_tasks, done); - pool->done += done; if (pool->num == 0) BLI_condition_notify_all(&pool->num_cond); @@ -246,11 +290,11 @@ static void task_pool_num_decrease(TaskPool *pool, size_t done) BLI_mutex_unlock(&pool->num_mutex); } -static void task_pool_num_increase(TaskPool *pool) +static void task_pool_num_increase(TaskPool *pool, size_t new) { BLI_mutex_lock(&pool->num_mutex); - pool->num++; + pool->num += new; BLI_condition_notify_all(&pool->num_cond); BLI_mutex_unlock(&pool->num_mutex); @@ -292,17 +336,10 @@ static bool task_scheduler_thread_wait_pop(TaskScheduler *scheduler, Task **task continue; } - if (atomic_add_and_fetch_z(&pool->currently_running_tasks, 1) <= pool->num_threads || - pool->num_threads == 0) - { - *task = current_task; - found_task = true; - BLI_remlink(&scheduler->queue, *task); - break; - } - else { - atomic_sub_and_fetch_z(&pool->currently_running_tasks, 1); - } + *task = current_task; + found_task = true; + BLI_remlink(&scheduler->queue, *task); + break; } if (!found_task) BLI_condition_wait(&scheduler->queue_cond, &scheduler->queue_mutex); @@ -313,13 +350,34 @@ static bool task_scheduler_thread_wait_pop(TaskScheduler *scheduler, Task **task return true; } +BLI_INLINE void handle_local_queue(TaskThreadLocalStorage *tls, + const int thread_id) +{ + while (tls->num_local_queue > 0) { + /* We pop task from queue before handling it so handler of the task can + * push next job to the local queue. + */ + tls->num_local_queue--; + Task *local_task = tls->local_queue[tls->num_local_queue]; + /* TODO(sergey): Double-check work_and_wait() doesn't handle other's + * pool tasks. + */ + TaskPool *local_pool = local_task->pool; + local_task->run(local_pool, local_task->taskdata, thread_id); + task_free(local_pool, local_task, thread_id); + } +} + static void *task_scheduler_thread_run(void *thread_p) { TaskThread *thread = (TaskThread *) thread_p; + TaskThreadLocalStorage *tls = &thread->tls; TaskScheduler *scheduler = thread->scheduler; int thread_id = thread->id; Task *task; + pthread_setspecific(scheduler->tls_id_key, thread); + /* keep popping off tasks */ while (task_scheduler_thread_wait_pop(scheduler, &task)) { TaskPool *pool = task->pool; @@ -330,6 +388,9 @@ static void *task_scheduler_thread_run(void *thread_p) /* delete task */ task_free(pool, task, thread_id); + /* Handle all tasks from local queue. */ + handle_local_queue(tls, thread_id); + /* notify pool task was done */ task_pool_num_decrease(pool, 1); } @@ -359,20 +420,24 @@ TaskScheduler *BLI_task_scheduler_create(int num_threads) /* Add background-only thread if needed. */ if (num_threads == 0) { - scheduler->background_thread_only = true; - num_threads = 1; + scheduler->background_thread_only = true; + num_threads = 1; } + scheduler->task_threads = MEM_callocN(sizeof(TaskThread) * (num_threads + 1), + "TaskScheduler task threads"); + + pthread_key_create(&scheduler->tls_id_key, NULL); + /* launch threads that will be waiting for work */ if (num_threads > 0) { int i; scheduler->num_threads = num_threads; scheduler->threads = MEM_callocN(sizeof(pthread_t) * num_threads, "TaskScheduler threads"); - scheduler->task_threads = MEM_callocN(sizeof(TaskThread) * num_threads, "TaskScheduler task threads"); for (i = 0; i < num_threads; i++) { - TaskThread *thread = &scheduler->task_threads[i]; + TaskThread *thread = &scheduler->task_threads[i + 1]; thread->scheduler = scheduler; thread->id = i + 1; @@ -380,9 +445,6 @@ TaskScheduler *BLI_task_scheduler_create(int num_threads) fprintf(stderr, "TaskScheduler failed to launch thread %d/%d\n", i, num_threads); } } - - scheduler->task_mempool = MEM_callocN(sizeof(*scheduler->task_mempool) * (num_threads + 1), - "TaskScheduler task_mempool"); } return scheduler; @@ -398,6 +460,8 @@ void BLI_task_scheduler_free(TaskScheduler *scheduler) BLI_condition_notify_all(&scheduler->queue_cond); BLI_mutex_unlock(&scheduler->queue_mutex); + pthread_key_delete(scheduler->tls_id_key); + /* delete threads */ if (scheduler->threads) { int i; @@ -412,17 +476,12 @@ void BLI_task_scheduler_free(TaskScheduler *scheduler) /* Delete task thread data */ if (scheduler->task_threads) { - MEM_freeN(scheduler->task_threads); - } - - /* Delete task memory pool */ - if (scheduler->task_mempool) { - for (int i = 0; i <= scheduler->num_threads; ++i) { - for (int j = 0; j < scheduler->task_mempool[i].num_tasks; ++j) { - MEM_freeN(scheduler->task_mempool[i].tasks[j]); - } + for (int i = 0; i < scheduler->num_threads + 1; ++i) { + TaskThreadLocalStorage *tls = &scheduler->task_threads[i].tls; + free_task_tls(tls); } - MEM_freeN(scheduler->task_mempool); + + MEM_freeN(scheduler->task_threads); } /* delete leftover tasks */ @@ -445,7 +504,7 @@ int BLI_task_scheduler_num_threads(TaskScheduler *scheduler) static void task_scheduler_push(TaskScheduler *scheduler, Task *task, TaskPriority priority) { - task_pool_num_increase(task->pool); + task_pool_num_increase(task->pool, 1); /* add task to queue */ BLI_mutex_lock(&scheduler->queue_mutex); @@ -471,7 +530,7 @@ static void task_scheduler_clear(TaskScheduler *scheduler, TaskPool *pool) nexttask = task->next; if (task->pool == pool) { - task_data_free(task, 0); + task_data_free(task, pool->thread_id); BLI_freelinkN(&scheduler->queue, task); done++; @@ -486,7 +545,10 @@ static void task_scheduler_clear(TaskScheduler *scheduler, TaskPool *pool) /* Task Pool */ -static TaskPool *task_pool_create_ex(TaskScheduler *scheduler, void *userdata, const bool is_background) +static TaskPool *task_pool_create_ex(TaskScheduler *scheduler, + void *userdata, + const bool is_background, + const bool is_suspended) { TaskPool *pool = MEM_mallocN(sizeof(TaskPool), "TaskPool"); @@ -504,10 +566,11 @@ static TaskPool *task_pool_create_ex(TaskScheduler *scheduler, void *userdata, c pool->scheduler = scheduler; pool->num = 0; - pool->done = 0; - pool->num_threads = 0; - pool->currently_running_tasks = 0; pool->do_cancel = false; + pool->do_work = false; + pool->is_suspended = is_suspended; + pool->num_suspended = 0; + pool->suspended_queue.first = pool->suspended_queue.last = NULL; pool->run_in_background = is_background; BLI_mutex_init(&pool->num_mutex); @@ -517,11 +580,21 @@ static TaskPool *task_pool_create_ex(TaskScheduler *scheduler, void *userdata, c BLI_mutex_init(&pool->user_mutex); if (BLI_thread_is_main()) { - pool->task_mempool = scheduler->task_mempool; + pool->thread_id = 0; } else { - pool->task_mempool = &pool->task_mempool_local; - pool->task_mempool_local.num_tasks = 0; + TaskThread *thread = pthread_getspecific(scheduler->tls_id_key); + /* NOTE: It is possible that pool is created from non-main thread + * which isn't a scheduler thread. In this case pthread's TLS will + * be NULL and we can safely consider thread id 0 for the main + * thread of this pool (the one which does wort_and_wait()). + */ + if (thread == NULL) { + pool->thread_id = 0; + } + else { + pool->thread_id = thread->id; + } } #ifdef DEBUG_STATS @@ -548,7 +621,7 @@ static TaskPool *task_pool_create_ex(TaskScheduler *scheduler, void *userdata, c */ TaskPool *BLI_task_pool_create(TaskScheduler *scheduler, void *userdata) { - return task_pool_create_ex(scheduler, userdata, false); + return task_pool_create_ex(scheduler, userdata, false, false); } /** @@ -563,25 +636,28 @@ TaskPool *BLI_task_pool_create(TaskScheduler *scheduler, void *userdata) */ TaskPool *BLI_task_pool_create_background(TaskScheduler *scheduler, void *userdata) { - return task_pool_create_ex(scheduler, userdata, true); + return task_pool_create_ex(scheduler, userdata, true, false); +} + +/** + * Similar to BLI_task_pool_create() but does not schedule any tasks for execution + * for until BLI_task_pool_work_and_wait() is called. This helps reducing therading + * overhead when pushing huge amount of small initial tasks from the main thread. + */ +TaskPool *BLI_task_pool_create_suspended(TaskScheduler *scheduler, void *userdata) +{ + return task_pool_create_ex(scheduler, userdata, false, true); } void BLI_task_pool_free(TaskPool *pool) { - BLI_task_pool_stop(pool); + BLI_task_pool_cancel(pool); BLI_mutex_end(&pool->num_mutex); BLI_condition_end(&pool->num_cond); BLI_mutex_end(&pool->user_mutex); - /* Free local memory pool, those pointers are lost forever. */ - if (pool->task_mempool == &pool->task_mempool_local) { - for (int i = 0; i < pool->task_mempool_local.num_tasks; i++) { - MEM_freeN(pool->task_mempool_local.tasks[i]); - } - } - #ifdef DEBUG_STATS printf("Thread ID Allocated Reused Discarded\n"); for (int i = 0; i < pool->scheduler->num_threads + 1; ++i) { @@ -612,6 +688,25 @@ static void task_pool_push( task->freedata = freedata; task->pool = pool; + if (pool->is_suspended) { + BLI_addhead(&pool->suspended_queue, task); + atomic_fetch_and_add_z(&pool->num_suspended, 1); + return; + } + + if (thread_id != -1 && + (thread_id != pool->thread_id || pool->do_work)) + { + ASSERT_THREAD_ID(pool->scheduler, thread_id); + + TaskThreadLocalStorage *tls = get_task_tls(pool, thread_id); + if (tls->num_local_queue < LOCALQUEUE_SIZE) { + tls->local_queue[tls->num_local_queue] = task; + tls->num_local_queue++; + return; + } + } + task_scheduler_push(pool->scheduler, task, priority); } @@ -636,8 +731,27 @@ void BLI_task_pool_push_from_thread(TaskPool *pool, TaskRunFunction run, void BLI_task_pool_work_and_wait(TaskPool *pool) { + TaskThreadLocalStorage *tls = get_task_tls(pool, pool->thread_id); TaskScheduler *scheduler = pool->scheduler; + if (atomic_fetch_and_and_uint8((uint8_t *)&pool->is_suspended, 0)) { + if (pool->num_suspended) { + task_pool_num_increase(pool, pool->num_suspended); + BLI_mutex_lock(&scheduler->queue_mutex); + + BLI_movelisttolist(&scheduler->queue, &pool->suspended_queue); + + BLI_condition_notify_all(&scheduler->queue_cond); + BLI_mutex_unlock(&scheduler->queue_mutex); + + } + pool->is_suspended = false; + } + + pool->do_work = true; + + ASSERT_THREAD_ID(pool->scheduler, pool->thread_id); + BLI_mutex_lock(&pool->num_mutex); while (pool->num != 0) { @@ -651,16 +765,12 @@ void BLI_task_pool_work_and_wait(TaskPool *pool) /* find task from this pool. if we get a task from another pool, * we can get into deadlock */ - if (pool->num_threads == 0 || - pool->currently_running_tasks < pool->num_threads) - { - for (task = scheduler->queue.first; task; task = task->next) { - if (task->pool == pool) { - work_task = task; - found_task = true; - BLI_remlink(&scheduler->queue, task); - break; - } + for (task = scheduler->queue.first; task; task = task->next) { + if (task->pool == pool) { + work_task = task; + found_task = true; + BLI_remlink(&scheduler->queue, task); + break; } } @@ -669,11 +779,13 @@ void BLI_task_pool_work_and_wait(TaskPool *pool) /* if found task, do it, otherwise wait until other tasks are done */ if (found_task) { /* run task */ - atomic_add_and_fetch_z(&pool->currently_running_tasks, 1); - work_task->run(pool, work_task->taskdata, 0); + work_task->run(pool, work_task->taskdata, pool->thread_id); /* delete task */ - task_free(pool, task, 0); + task_free(pool, task, pool->thread_id); + + /* Handle all tasks from local queue. */ + handle_local_queue(tls, pool->thread_id); /* notify pool task was done */ task_pool_num_decrease(pool, 1); @@ -688,22 +800,8 @@ void BLI_task_pool_work_and_wait(TaskPool *pool) } BLI_mutex_unlock(&pool->num_mutex); -} -int BLI_pool_get_num_threads(TaskPool *pool) -{ - if (pool->num_threads != 0) { - return pool->num_threads; - } - else { - return BLI_task_scheduler_num_threads(pool->scheduler); - } -} - -void BLI_pool_set_num_threads(TaskPool *pool, int num_threads) -{ - /* NOTE: Don't try to modify threads while tasks are running! */ - pool->num_threads = num_threads; + handle_local_queue(tls, pool->thread_id); } void BLI_task_pool_cancel(TaskPool *pool) @@ -721,13 +819,6 @@ void BLI_task_pool_cancel(TaskPool *pool) pool->do_cancel = false; } -void BLI_task_pool_stop(TaskPool *pool) -{ - task_scheduler_clear(pool->scheduler, pool); - - BLI_assert(pool->num == 0); -} - bool BLI_task_pool_canceled(TaskPool *pool) { return pool->do_cancel; @@ -743,11 +834,6 @@ ThreadMutex *BLI_task_pool_user_mutex(TaskPool *pool) return &pool->user_mutex; } -size_t BLI_task_pool_tasks_done(TaskPool *pool) -{ - return pool->done; -} - /* Parallel range routines */ /** @@ -783,7 +869,7 @@ BLI_INLINE bool parallel_range_next_iter_get( int * __restrict iter, int * __restrict count) { uint32_t uval = atomic_fetch_and_add_uint32((uint32_t *)(&state->iter), state->chunk_size); - int previter = *(int32_t*)&uval; + int previter = *(int32_t *)&uval; *iter = previter; *count = max_ii(0, min_ii(state->chunk_size, state->stop - previter)); @@ -918,7 +1004,8 @@ static void task_parallel_range_ex( BLI_task_pool_push_from_thread(task_pool, parallel_range_func, userdata_chunk_local, false, - TASK_PRIORITY_HIGH, 0); + TASK_PRIORITY_HIGH, + task_pool->thread_id); } BLI_task_pool_work_and_wait(task_pool); @@ -1124,7 +1211,8 @@ void BLI_task_parallel_listbase( BLI_task_pool_push_from_thread(task_pool, parallel_listbase_func, NULL, false, - TASK_PRIORITY_HIGH, 0); + TASK_PRIORITY_HIGH, + task_pool->thread_id); } BLI_task_pool_work_and_wait(task_pool); diff --git a/source/blender/blenlib/intern/threads.c b/source/blender/blenlib/intern/threads.c index b60981802aa..77da3be0600 100644 --- a/source/blender/blenlib/intern/threads.c +++ b/source/blender/blenlib/intern/threads.c @@ -54,6 +54,8 @@ # include <sys/time.h> #endif +#include "atomic_ops.h" + #if defined(__APPLE__) && defined(_OPENMP) && (__GNUC__ == 4) && (__GNUC_MINOR__ == 2) && !defined(__clang__) # define USE_APPLE_OMP_FIX #endif @@ -124,7 +126,7 @@ static pthread_mutex_t _colormanage_lock = PTHREAD_MUTEX_INITIALIZER; static pthread_mutex_t _fftw_lock = PTHREAD_MUTEX_INITIALIZER; static pthread_mutex_t _view3d_lock = PTHREAD_MUTEX_INITIALIZER; static pthread_t mainid; -static int thread_levels = 0; /* threads can be invoked inside threads */ +static unsigned int thread_levels = 0; /* threads can be invoked inside threads */ static int num_threads_override = 0; /* just a max for security reasons */ @@ -198,9 +200,9 @@ void BLI_init_threads(ListBase *threadbase, void *(*do_thread)(void *), int tot) tslot->avail = 1; } } - - BLI_spin_lock(&_malloc_lock); - if (thread_levels == 0) { + + unsigned int level = atomic_fetch_and_add_u(&thread_levels, 1); + if (level == 0) { MEM_set_lock_callback(BLI_lock_malloc_thread, BLI_unlock_malloc_thread); #ifdef USE_APPLE_OMP_FIX @@ -210,9 +212,6 @@ void BLI_init_threads(ListBase *threadbase, void *(*do_thread)(void *), int tot) thread_tls_data = pthread_getspecific(gomp_tls_key); #endif } - - thread_levels++; - BLI_spin_unlock(&_malloc_lock); } /* amount of available threads */ @@ -331,11 +330,10 @@ void BLI_end_threads(ListBase *threadbase) BLI_freelistN(threadbase); } - BLI_spin_lock(&_malloc_lock); - thread_levels--; - if (thread_levels == 0) + unsigned int level = atomic_sub_and_fetch_u(&thread_levels, 1); + if (level == 0) { MEM_set_lock_callback(NULL, NULL); - BLI_spin_unlock(&_malloc_lock); + } } /* System Information */ @@ -812,26 +810,17 @@ void BLI_thread_queue_wait_finish(ThreadQueue *queue) void BLI_begin_threaded_malloc(void) { - /* Used for debug only */ - /* BLI_assert(thread_levels >= 0); */ - - BLI_spin_lock(&_malloc_lock); - if (thread_levels == 0) { + unsigned int level = atomic_fetch_and_add_u(&thread_levels, 1); + if (level == 0) { MEM_set_lock_callback(BLI_lock_malloc_thread, BLI_unlock_malloc_thread); } - thread_levels++; - BLI_spin_unlock(&_malloc_lock); } void BLI_end_threaded_malloc(void) { - /* Used for debug only */ - /* BLI_assert(thread_levels >= 0); */ - - BLI_spin_lock(&_malloc_lock); - thread_levels--; - if (thread_levels == 0) + unsigned int level = atomic_sub_and_fetch_u(&thread_levels, 1); + if (level == 0) { MEM_set_lock_callback(NULL, NULL); - BLI_spin_unlock(&_malloc_lock); + } } diff --git a/source/blender/blenloader/BLO_readfile.h b/source/blender/blenloader/BLO_readfile.h index c85cf128643..e6fc4703248 100644 --- a/source/blender/blenloader/BLO_readfile.h +++ b/source/blender/blenloader/BLO_readfile.h @@ -71,11 +71,25 @@ typedef struct BlendFileData { BlenFileType type; } BlendFileData; -BlendFileData *BLO_read_from_file(const char *filepath, struct ReportList *reports); -BlendFileData *BLO_read_from_memory(const void *mem, int memsize, struct ReportList *reports); + +/* skip reading some data-block types (may want to skip screen data too). */ +typedef enum eBLOReadSkip { + BLO_READ_SKIP_NONE = 0, + BLO_READ_SKIP_USERDEF = (1 << 0), + BLO_READ_SKIP_DATA = (1 << 1), +} eBLOReadSkip; +#define BLO_READ_SKIP_ALL \ + (BLO_READ_SKIP_USERDEF | BLO_READ_SKIP_DATA) + +BlendFileData *BLO_read_from_file( + const char *filepath, + struct ReportList *reports, eBLOReadSkip skip_flag); +BlendFileData *BLO_read_from_memory( + const void *mem, int memsize, + struct ReportList *reports, eBLOReadSkip skip_flag); BlendFileData *BLO_read_from_memfile( struct Main *oldmain, const char *filename, struct MemFile *memfile, - struct ReportList *reports); + struct ReportList *reports, eBLOReadSkip skip_flag); void BLO_blendfiledata_free(BlendFileData *bfd); diff --git a/source/blender/blenloader/CMakeLists.txt b/source/blender/blenloader/CMakeLists.txt index 8cb9ef837b2..3d2e8a306de 100644 --- a/source/blender/blenloader/CMakeLists.txt +++ b/source/blender/blenloader/CMakeLists.txt @@ -69,6 +69,12 @@ if(WITH_BUILDINFO) add_definitions(-DWITH_BUILDINFO) endif() +if(WITH_PYTHON) + if(WITH_PYTHON_SECURITY) + add_definitions(-DWITH_PYTHON_SECURITY) + endif() +endif() + if(WITH_INTERNATIONAL) add_definitions(-DWITH_INTERNATIONAL) endif() diff --git a/source/blender/blenloader/intern/readblenentry.c b/source/blender/blenloader/intern/readblenentry.c index be893177b3b..73109413271 100644 --- a/source/blender/blenloader/intern/readblenentry.c +++ b/source/blender/blenloader/intern/readblenentry.c @@ -317,7 +317,9 @@ void BLO_blendhandle_close(BlendHandle *bh) * \param reports If the return value is NULL, errors indicating the cause of the failure. * \return The data of the file. */ -BlendFileData *BLO_read_from_file(const char *filepath, ReportList *reports) +BlendFileData *BLO_read_from_file( + const char *filepath, + ReportList *reports, eBLOReadSkip skip_flags) { BlendFileData *bfd = NULL; FileData *fd; @@ -325,6 +327,7 @@ BlendFileData *BLO_read_from_file(const char *filepath, ReportList *reports) fd = blo_openblenderfile(filepath, reports); if (fd) { fd->reports = reports; + fd->skip_flags = skip_flags; bfd = blo_read_file_internal(fd, filepath); blo_freefiledata(fd); } @@ -341,7 +344,9 @@ BlendFileData *BLO_read_from_file(const char *filepath, ReportList *reports) * \param reports If the return value is NULL, errors indicating the cause of the failure. * \return The data of the file. */ -BlendFileData *BLO_read_from_memory(const void *mem, int memsize, ReportList *reports) +BlendFileData *BLO_read_from_memory( + const void *mem, int memsize, + ReportList *reports, eBLOReadSkip skip_flags) { BlendFileData *bfd = NULL; FileData *fd; @@ -349,6 +354,7 @@ BlendFileData *BLO_read_from_memory(const void *mem, int memsize, ReportList *re fd = blo_openblendermemory(mem, memsize, reports); if (fd) { fd->reports = reports; + fd->skip_flags = skip_flags; bfd = blo_read_file_internal(fd, ""); blo_freefiledata(fd); } @@ -362,7 +368,9 @@ BlendFileData *BLO_read_from_memory(const void *mem, int memsize, ReportList *re * \param oldmain old main, from which we will keep libraries and other datablocks that should not have changed. * \param filename current file, only for retrieving library data. */ -BlendFileData *BLO_read_from_memfile(Main *oldmain, const char *filename, MemFile *memfile, ReportList *reports) +BlendFileData *BLO_read_from_memfile( + Main *oldmain, const char *filename, MemFile *memfile, + ReportList *reports, eBLOReadSkip skip_flags) { BlendFileData *bfd = NULL; FileData *fd; @@ -371,6 +379,7 @@ BlendFileData *BLO_read_from_memfile(Main *oldmain, const char *filename, MemFil fd = blo_openblendermemfile(memfile, reports); if (fd) { fd->reports = reports; + fd->skip_flags = skip_flags; BLI_strncpy(fd->relabase, filename, sizeof(fd->relabase)); /* clear ob->proxy_from pointers in old main */ diff --git a/source/blender/blenloader/intern/readfile.c b/source/blender/blenloader/intern/readfile.c index 6264336c096..09c88ac945a 100644 --- a/source/blender/blenloader/intern/readfile.c +++ b/source/blender/blenloader/intern/readfile.c @@ -2192,19 +2192,19 @@ static void direct_link_curvemapping(FileData *fd, CurveMapping *cumap) /* library brush linking after fileread */ static void lib_link_brush(FileData *fd, Main *main) { - Brush *brush; - /* only link ID pointers */ - for (brush = main->brush.first; brush; brush = brush->id.next) { + for (Brush *brush = main->brush.first; brush; brush = brush->id.next) { if (brush->id.tag & LIB_TAG_NEED_LINK) { - brush->id.tag &= ~LIB_TAG_NEED_LINK; - + IDP_LibLinkProperty(brush->id.properties, (fd->flags & FD_FLAGS_SWITCH_ENDIAN), fd); + /* brush->(mask_)mtex.obj is ignored on purpose? */ brush->mtex.tex = newlibadr_us(fd, brush->id.lib, brush->mtex.tex); brush->mask_mtex.tex = newlibadr_us(fd, brush->id.lib, brush->mask_mtex.tex); brush->clone.image = newlibadr(fd, brush->id.lib, brush->clone.image); brush->toggle_brush = newlibadr(fd, brush->id.lib, brush->toggle_brush); brush->paint_curve = newlibadr_us(fd, brush->id.lib, brush->paint_curve); + + brush->id.tag &= ~LIB_TAG_NEED_LINK; } } } @@ -2227,13 +2227,13 @@ static void direct_link_brush(FileData *fd, Brush *brush) } /* ************ READ Palette *************** */ -static void lib_link_palette(FileData *UNUSED(fd), Main *main) +static void lib_link_palette(FileData *fd, Main *main) { - Palette *palette; - /* only link ID pointers */ - for (palette = main->palettes.first; palette; palette = palette->id.next) { + for (Palette *palette = main->palettes.first; palette; palette = palette->id.next) { if (palette->id.tag & LIB_TAG_NEED_LINK) { + IDP_LibLinkProperty(palette->id.properties, (fd->flags & FD_FLAGS_SWITCH_ENDIAN), fd); + palette->id.tag &= ~LIB_TAG_NEED_LINK; } } @@ -2245,13 +2245,13 @@ static void direct_link_palette(FileData *fd, Palette *palette) link_list(fd, &palette->colors); } -static void lib_link_paint_curve(FileData *UNUSED(fd), Main *main) +static void lib_link_paint_curve(FileData *fd, Main *main) { - PaintCurve *pc; - /* only link ID pointers */ - for (pc = main->paintcurves.first; pc; pc = pc->id.next) { + for (PaintCurve *pc = main->paintcurves.first; pc; pc = pc->id.next) { if (pc->id.tag & LIB_TAG_NEED_LINK) { + IDP_LibLinkProperty(pc->id.properties, (fd->flags & FD_FLAGS_SWITCH_ENDIAN), fd); + pc->id.tag &= ~LIB_TAG_NEED_LINK; } } @@ -2503,15 +2503,12 @@ static void direct_link_fcurves(FileData *fd, ListBase *list) static void lib_link_action(FileData *fd, Main *main) { - bAction *act; - bActionChannel *chan; - - for (act = main->action.first; act; act = act->id.next) { + for (bAction *act = main->action.first; act; act = act->id.next) { if (act->id.tag & LIB_TAG_NEED_LINK) { - act->id.tag &= ~LIB_TAG_NEED_LINK; + IDP_LibLinkProperty(act->id.properties, (fd->flags & FD_FLAGS_SWITCH_ENDIAN), fd); // XXX deprecated - old animation system <<< - for (chan=act->chanbase.first; chan; chan=chan->next) { + for (bActionChannel *chan = act->chanbase.first; chan; chan = chan->next) { chan->ipo = newlibadr_us(fd, act->id.lib, chan->ipo); lib_link_constraint_channels(fd, &act->id, &chan->constraintChannels); } @@ -2524,6 +2521,8 @@ static void lib_link_action(FileData *fd, Main *main) marker->camera = newlibadr(fd, act->id.lib, marker->camera); } } + + act->id.tag &= ~LIB_TAG_NEED_LINK; } } } @@ -2710,26 +2709,20 @@ static void direct_link_animdata(FileData *fd, AnimData *adt) static void lib_link_cachefiles(FileData *fd, Main *bmain) { - CacheFile *cache_file; - /* only link ID pointers */ - for (cache_file = bmain->cachefiles.first; cache_file; cache_file = cache_file->id.next) { + for (CacheFile *cache_file = bmain->cachefiles.first; cache_file; cache_file = cache_file->id.next) { if (cache_file->id.tag & LIB_TAG_NEED_LINK) { - cache_file->id.tag &= ~LIB_TAG_NEED_LINK; - } - - BLI_listbase_clear(&cache_file->object_paths); - cache_file->handle = NULL; - cache_file->handle_mutex = NULL; - - if (cache_file->adt) { + IDP_LibLinkProperty(cache_file->id.properties, (fd->flags & FD_FLAGS_SWITCH_ENDIAN), fd); lib_link_animdata(fd, &cache_file->id, cache_file->adt); + + cache_file->id.tag &= ~LIB_TAG_NEED_LINK; } } } static void direct_link_cachefile(FileData *fd, CacheFile *cache_file) { + BLI_listbase_clear(&cache_file->object_paths); cache_file->handle = NULL; cache_file->handle_mutex = NULL; @@ -2766,6 +2759,7 @@ static void lib_link_ntree(FileData *fd, ID *id, bNodeTree *ntree) bNode *node; bNodeSocket *sock; + IDP_LibLinkProperty(ntree->id.properties, (fd->flags & FD_FLAGS_SWITCH_ENDIAN), fd); lib_link_animdata(fd, &ntree->id, ntree->adt); ntree->gpd = newlibadr_us(fd, id->lib, ntree->gpd); @@ -2775,30 +2769,37 @@ static void lib_link_ntree(FileData *fd, ID *id, bNodeTree *ntree) * of library blocks that implement this.*/ IDP_LibLinkProperty(node->prop, (fd->flags & FD_FLAGS_SWITCH_ENDIAN), fd); - node->id= newlibadr_us(fd, id->lib, node->id); + node->id = newlibadr_us(fd, id->lib, node->id); - for (sock = node->inputs.first; sock; sock = sock->next) + for (sock = node->inputs.first; sock; sock = sock->next) { + IDP_LibLinkProperty(sock->prop, (fd->flags & FD_FLAGS_SWITCH_ENDIAN), fd); lib_link_node_socket(fd, id, sock); - for (sock = node->outputs.first; sock; sock = sock->next) + } + for (sock = node->outputs.first; sock; sock = sock->next) { + IDP_LibLinkProperty(sock->prop, (fd->flags & FD_FLAGS_SWITCH_ENDIAN), fd); lib_link_node_socket(fd, id, sock); + } } - for (sock = ntree->inputs.first; sock; sock = sock->next) + for (sock = ntree->inputs.first; sock; sock = sock->next) { + IDP_LibLinkProperty(sock->prop, (fd->flags & FD_FLAGS_SWITCH_ENDIAN), fd); lib_link_node_socket(fd, id, sock); - for (sock = ntree->outputs.first; sock; sock = sock->next) + } + for (sock = ntree->outputs.first; sock; sock = sock->next) { + IDP_LibLinkProperty(sock->prop, (fd->flags & FD_FLAGS_SWITCH_ENDIAN), fd); lib_link_node_socket(fd, id, sock); + } } /* library ntree linking after fileread */ static void lib_link_nodetree(FileData *fd, Main *main) { - bNodeTree *ntree; - /* only link ID pointers */ - for (ntree = main->nodetree.first; ntree; ntree = ntree->id.next) { + for (bNodeTree *ntree = main->nodetree.first; ntree; ntree = ntree->id.next) { if (ntree->id.tag & LIB_TAG_NEED_LINK) { - ntree->id.tag &= ~LIB_TAG_NEED_LINK; lib_link_ntree(fd, &ntree->id, ntree); + + ntree->id.tag &= ~LIB_TAG_NEED_LINK; } } } @@ -3313,11 +3314,11 @@ static void lib_link_pose(FileData *fd, Main *bmain, Object *ob, bPose *pose) static void lib_link_armature(FileData *fd, Main *main) { - bArmature *arm; - - for (arm = main->armature.first; arm; arm = arm->id.next) { + for (bArmature *arm = main->armature.first; arm; arm = arm->id.next) { if (arm->id.tag & LIB_TAG_NEED_LINK) { + IDP_LibLinkProperty(arm->id.properties, (fd->flags & FD_FLAGS_SWITCH_ENDIAN), fd); lib_link_animdata(fd, &arm->id, arm->adt); + arm->id.tag &= ~LIB_TAG_NEED_LINK; } } @@ -3362,14 +3363,13 @@ static void direct_link_armature(FileData *fd, bArmature *arm) static void lib_link_camera(FileData *fd, Main *main) { - Camera *ca; - - for (ca = main->camera.first; ca; ca = ca->id.next) { + for (Camera *ca = main->camera.first; ca; ca = ca->id.next) { if (ca->id.tag & LIB_TAG_NEED_LINK) { + IDP_LibLinkProperty(ca->id.properties, (fd->flags & FD_FLAGS_SWITCH_ENDIAN), fd); lib_link_animdata(fd, &ca->id, ca->adt); ca->ipo = newlibadr_us(fd, ca->id.lib, ca->ipo); // XXX deprecated - old animation system - + ca->dof_ob = newlibadr(fd, ca->id.lib, ca->dof_ob); ca->id.tag &= ~LIB_TAG_NEED_LINK; @@ -3388,16 +3388,13 @@ static void direct_link_camera(FileData *fd, Camera *ca) static void lib_link_lamp(FileData *fd, Main *main) { - Lamp *la; - MTex *mtex; - int a; - - for (la = main->lamp.first; la; la = la->id.next) { + for (Lamp *la = main->lamp.first; la; la = la->id.next) { if (la->id.tag & LIB_TAG_NEED_LINK) { + IDP_LibLinkProperty(la->id.properties, (fd->flags & FD_FLAGS_SWITCH_ENDIAN), fd); lib_link_animdata(fd, &la->id, la->adt); - for (a = 0; a < MAX_MTEX; a++) { - mtex = la->mtex[a]; + for (int a = 0; a < MAX_MTEX; a++) { + MTex *mtex = la->mtex[a]; if (mtex) { mtex->tex = newlibadr_us(fd, la->id.lib, mtex->tex); mtex->object = newlibadr(fd, la->id.lib, mtex->object); @@ -3454,17 +3451,11 @@ void blo_do_versions_key_uidgen(Key *key) static void lib_link_key(FileData *fd, Main *main) { - Key *key; - - for (key = main->key.first; key; key = key->id.next) { - /*check if we need to generate unique ids for the shapekeys*/ - if (!key->uidgen) { - blo_do_versions_key_uidgen(key); - } - + for (Key *key = main->key.first; key; key = key->id.next) { BLI_assert((key->id.tag & LIB_TAG_EXTERN) == 0); if (key->id.tag & LIB_TAG_NEED_LINK) { + IDP_LibLinkProperty(key->id.properties, (fd->flags & FD_FLAGS_SWITCH_ENDIAN), fd); lib_link_animdata(fd, &key->id, key->adt); key->ipo = newlibadr_us(fd, key->id.lib, key->ipo); // XXX deprecated - old animation system @@ -3527,15 +3518,14 @@ static void direct_link_key(FileData *fd, Key *key) static void lib_link_mball(FileData *fd, Main *main) { - MetaBall *mb; - int a; - - for (mb = main->mball.first; mb; mb = mb->id.next) { + for (MetaBall *mb = main->mball.first; mb; mb = mb->id.next) { if (mb->id.tag & LIB_TAG_NEED_LINK) { + IDP_LibLinkProperty(mb->id.properties, (fd->flags & FD_FLAGS_SWITCH_ENDIAN), fd); lib_link_animdata(fd, &mb->id, mb->adt); - for (a = 0; a < mb->totcol; a++) + for (int a = 0; a < mb->totcol; a++) { mb->mat[a] = newlibadr_us(fd, mb->id.lib, mb->mat[a]); + } mb->ipo = newlibadr_us(fd, mb->id.lib, mb->ipo); // XXX deprecated - old animation system @@ -3564,18 +3554,15 @@ static void direct_link_mball(FileData *fd, MetaBall *mb) static void lib_link_world(FileData *fd, Main *main) { - World *wrld; - MTex *mtex; - int a; - - for (wrld = main->world.first; wrld; wrld = wrld->id.next) { + for (World *wrld = main->world.first; wrld; wrld = wrld->id.next) { if (wrld->id.tag & LIB_TAG_NEED_LINK) { + IDP_LibLinkProperty(wrld->id.properties, (fd->flags & FD_FLAGS_SWITCH_ENDIAN), fd); lib_link_animdata(fd, &wrld->id, wrld->adt); wrld->ipo = newlibadr_us(fd, wrld->id.lib, wrld->ipo); // XXX deprecated - old animation system - for (a=0; a < MAX_MTEX; a++) { - mtex = wrld->mtex[a]; + for (int a = 0; a < MAX_MTEX; a++) { + MTex *mtex = wrld->mtex[a]; if (mtex) { mtex->tex = newlibadr_us(fd, wrld->id.lib, mtex->tex); mtex->object = newlibadr(fd, wrld->id.lib, mtex->object); @@ -3616,12 +3603,12 @@ static void direct_link_world(FileData *fd, World *wrld) /* ************ READ VFONT ***************** */ -static void lib_link_vfont(FileData *UNUSED(fd), Main *main) +static void lib_link_vfont(FileData *fd, Main *main) { - VFont *vf; - - for (vf = main->vfont.first; vf; vf = vf->id.next) { + for (VFont *vf = main->vfont.first; vf; vf = vf->id.next) { if (vf->id.tag & LIB_TAG_NEED_LINK) { + IDP_LibLinkProperty(vf->id.properties, (fd->flags & FD_FLAGS_SWITCH_ENDIAN), fd); + vf->id.tag &= ~LIB_TAG_NEED_LINK; } } @@ -3636,12 +3623,12 @@ static void direct_link_vfont(FileData *fd, VFont *vf) /* ************ READ TEXT ****************** */ -static void lib_link_text(FileData *UNUSED(fd), Main *main) +static void lib_link_text(FileData *fd, Main *main) { - Text *text; - - for (text = main->text.first; text; text = text->id.next) { + for (Text *text = main->text.first; text; text = text->id.next) { if (text->id.tag & LIB_TAG_NEED_LINK) { + IDP_LibLinkProperty(text->id.properties, (fd->flags & FD_FLAGS_SWITCH_ENDIAN), fd); + text->id.tag &= ~LIB_TAG_NEED_LINK; } } @@ -3690,9 +3677,7 @@ static void direct_link_text(FileData *fd, Text *text) static void lib_link_image(FileData *fd, Main *main) { - Image *ima; - - for (ima = main->image.first; ima; ima = ima->id.next) { + for (Image *ima = main->image.first; ima; ima = ima->id.next) { if (ima->id.tag & LIB_TAG_NEED_LINK) { IDP_LibLinkProperty(ima->id.properties, (fd->flags & FD_FLAGS_SWITCH_ENDIAN), fd); @@ -3759,15 +3744,14 @@ static void direct_link_image(FileData *fd, Image *ima) static void lib_link_curve(FileData *fd, Main *main) { - Curve *cu; - int a; - - for (cu = main->curve.first; cu; cu = cu->id.next) { + for (Curve *cu = main->curve.first; cu; cu = cu->id.next) { if (cu->id.tag & LIB_TAG_NEED_LINK) { + IDP_LibLinkProperty(cu->id.properties, (fd->flags & FD_FLAGS_SWITCH_ENDIAN), fd); lib_link_animdata(fd, &cu->id, cu->adt); - for (a = 0; a < cu->totcol; a++) + for (int a = 0; a < cu->totcol; a++) { cu->mat[a] = newlibadr_us(fd, cu->id.lib, cu->mat[a]); + } cu->bevobj = newlibadr(fd, cu->id.lib, cu->bevobj); cu->taperobj = newlibadr(fd, cu->id.lib, cu->taperobj); @@ -3852,10 +3836,9 @@ static void direct_link_curve(FileData *fd, Curve *cu) static void lib_link_texture(FileData *fd, Main *main) { - Tex *tex; - - for (tex = main->tex.first; tex; tex = tex->id.next) { + for (Tex *tex = main->tex.first; tex; tex = tex->id.next) { if (tex->id.tag & LIB_TAG_NEED_LINK) { + IDP_LibLinkProperty(tex->id.properties, (fd->flags & FD_FLAGS_SWITCH_ENDIAN), fd); lib_link_animdata(fd, &tex->id, tex->adt); tex->ima = newlibadr_us(fd, tex->id.lib, tex->ima); @@ -3931,12 +3914,9 @@ static void direct_link_texture(FileData *fd, Tex *tex) static void lib_link_material(FileData *fd, Main *main) { - Material *ma; - MTex *mtex; - int a; - - for (ma = main->mat.first; ma; ma = ma->id.next) { + for (Material *ma = main->mat.first; ma; ma = ma->id.next) { if (ma->id.tag & LIB_TAG_NEED_LINK) { + IDP_LibLinkProperty(ma->id.properties, (fd->flags & FD_FLAGS_SWITCH_ENDIAN), fd); lib_link_animdata(fd, &ma->id, ma->adt); /* Link ID Properties -- and copy this comment EXACTLY for easy finding @@ -3946,8 +3926,8 @@ static void lib_link_material(FileData *fd, Main *main) ma->ipo = newlibadr_us(fd, ma->id.lib, ma->ipo); // XXX deprecated - old animation system ma->group = newlibadr_us(fd, ma->id.lib, ma->group); - for (a = 0; a < MAX_MTEX; a++) { - mtex = ma->mtex[a]; + for (int a = 0; a < MAX_MTEX; a++) { + MTex *mtex = ma->mtex[a]; if (mtex) { mtex->tex = newlibadr_us(fd, ma->id.lib, mtex->tex); mtex->object = newlibadr(fd, ma->id.lib, mtex->object); @@ -4078,14 +4058,11 @@ static void lib_link_partdeflect(FileData *fd, ID *id, PartDeflect *pd) static void lib_link_particlesettings(FileData *fd, Main *main) { - ParticleSettings *part; - ParticleDupliWeight *dw; - MTex *mtex; - int a; - - for (part = main->particle.first; part; part = part->id.next) { + for (ParticleSettings *part = main->particle.first; part; part = part->id.next) { if (part->id.tag & LIB_TAG_NEED_LINK) { + IDP_LibLinkProperty(part->id.properties, (fd->flags & FD_FLAGS_SWITCH_ENDIAN), fd); lib_link_animdata(fd, &part->id, part->adt); + part->ipo = newlibadr_us(fd, part->id.lib, part->ipo); // XXX deprecated - old animation system part->dup_ob = newlibadr(fd, part->id.lib, part->dup_ob); @@ -4105,6 +4082,7 @@ static void lib_link_particlesettings(FileData *fd, Main *main) } if (part->dupliweights.first && part->dup_group) { + ParticleDupliWeight *dw; int index_ok = 0; /* check for old files without indices (all indexes 0) */ if (BLI_listbase_is_single(&part->dupliweights)) { @@ -4169,8 +4147,8 @@ static void lib_link_particlesettings(FileData *fd, Main *main) } } - for (a = 0; a < MAX_MTEX; a++) { - mtex= part->mtex[a]; + for (int a = 0; a < MAX_MTEX; a++) { + MTex *mtex= part->mtex[a]; if (mtex) { mtex->tex = newlibadr_us(fd, part->id.lib, mtex->tex); mtex->object = newlibadr(fd, part->id.lib, mtex->object); @@ -4677,10 +4655,9 @@ static void direct_link_mesh(FileData *fd, Mesh *mesh) static void lib_link_latt(FileData *fd, Main *main) { - Lattice *lt; - - for (lt = main->latt.first; lt; lt = lt->id.next) { + for (Lattice *lt = main->latt.first; lt; lt = lt->id.next) { if (lt->id.tag & LIB_TAG_NEED_LINK) { + IDP_LibLinkProperty(lt->id.properties, (fd->flags & FD_FLAGS_SWITCH_ENDIAN), fd); lib_link_animdata(fd, <->id, lt->adt); lt->ipo = newlibadr_us(fd, lt->id.lib, lt->ipo); // XXX deprecated - old animation system @@ -4724,16 +4701,12 @@ static void lib_link_modifiers(FileData *fd, Object *ob) static void lib_link_object(FileData *fd, Main *main) { - Object *ob; - PartEff *paf; - bSensor *sens; - bController *cont; - bActuator *act; - void *poin; - int warn=0, a; - - for (ob = main->object.first; ob; ob = ob->id.next) { + bool warn = false; + + for (Object *ob = main->object.first; ob; ob = ob->id.next) { if (ob->id.tag & LIB_TAG_NEED_LINK) { + int a; + IDP_LibLinkProperty(ob->id.properties, (fd->flags & FD_FLAGS_SWITCH_ENDIAN), fd); lib_link_animdata(fd, &ob->id, ob->adt); @@ -4766,17 +4739,17 @@ static void lib_link_object(FileData *fd, Main *main) } ob->proxy_group = newlibadr(fd, ob->id.lib, ob->proxy_group); - poin = ob->data; + void *poin = ob->data; ob->data = newlibadr_us(fd, ob->id.lib, ob->data); - if (ob->data==NULL && poin!=NULL) { + if (ob->data == NULL && poin != NULL) { if (ob->id.lib) printf("Can't find obdata of %s lib %s\n", ob->id.name + 2, ob->id.lib->name); else printf("Object %s lost data.\n", ob->id.name + 2); ob->type = OB_EMPTY; - warn = 1; + warn = true; if (ob->pose) { /* we can't call #BKE_pose_free() here because of library linking @@ -4822,13 +4795,13 @@ static void lib_link_object(FileData *fd, Main *main) lib_link_nlastrips(fd, &ob->id, &ob->nlastrips); // >>> XXX deprecated - old animation system - for (paf = ob->effect.first; paf; paf = paf->next) { + for (PartEff *paf = ob->effect.first; paf; paf = paf->next) { if (paf->type == EFF_PARTICLE) { paf->group = newlibadr_us(fd, ob->id.lib, paf->group); } } - for (sens = ob->sensors.first; sens; sens = sens->next) { + for (bSensor *sens = ob->sensors.first; sens; sens = sens->next) { for (a = 0; a < sens->totlinks; a++) sens->links[a] = newglobadr(fd, sens->links[a]); @@ -4839,7 +4812,7 @@ static void lib_link_object(FileData *fd, Main *main) } } - for (cont = ob->controllers.first; cont; cont = cont->next) { + for (bController *cont = ob->controllers.first; cont; cont = cont->next) { for (a=0; a < cont->totlinks; a++) cont->links[a] = newglobadr(fd, cont->links[a]); @@ -4851,86 +4824,117 @@ static void lib_link_object(FileData *fd, Main *main) cont->totslinks = 0; } - for (act = ob->actuators.first; act; act = act->next) { - if (act->type == ACT_SOUND) { - bSoundActuator *sa = act->data; - sa->sound= newlibadr_us(fd, ob->id.lib, sa->sound); - } - else if (act->type == ACT_GAME) { - /* bGameActuator *ga= act->data; */ - } - else if (act->type == ACT_CAMERA) { - bCameraActuator *ca = act->data; - ca->ob= newlibadr(fd, ob->id.lib, ca->ob); - } - /* leave this one, it's obsolete but necessary to read for conversion */ - else if (act->type == ACT_ADD_OBJECT) { - bAddObjectActuator *eoa = act->data; - if (eoa) eoa->ob= newlibadr(fd, ob->id.lib, eoa->ob); - } - else if (act->type == ACT_OBJECT) { - bObjectActuator *oa = act->data; - if (oa == NULL) { - init_actuator(act); + for (bActuator *act = ob->actuators.first; act; act = act->next) { + switch (act->type) { + case ACT_SOUND: + { + bSoundActuator *sa = act->data; + sa->sound = newlibadr_us(fd, ob->id.lib, sa->sound); + break; } - else { - oa->reference = newlibadr(fd, ob->id.lib, oa->reference); + case ACT_GAME: + /* bGameActuator *ga= act->data; */ + break; + case ACT_CAMERA: + { + bCameraActuator *ca = act->data; + ca->ob = newlibadr(fd, ob->id.lib, ca->ob); + break; } - } - else if (act->type == ACT_EDIT_OBJECT) { - bEditObjectActuator *eoa = act->data; - if (eoa == NULL) { - init_actuator(act); + /* leave this one, it's obsolete but necessary to read for conversion */ + case ACT_ADD_OBJECT: + { + bAddObjectActuator *eoa = act->data; + if (eoa) + eoa->ob = newlibadr(fd, ob->id.lib, eoa->ob); + break; } - else { - eoa->ob= newlibadr(fd, ob->id.lib, eoa->ob); - eoa->me= newlibadr(fd, ob->id.lib, eoa->me); + case ACT_OBJECT: + { + bObjectActuator *oa = act->data; + if (oa == NULL) { + init_actuator(act); + } + else { + oa->reference = newlibadr(fd, ob->id.lib, oa->reference); + } + break; } - } - else if (act->type == ACT_SCENE) { - bSceneActuator *sa = act->data; - sa->camera= newlibadr(fd, ob->id.lib, sa->camera); - sa->scene= newlibadr(fd, ob->id.lib, sa->scene); - } - else if (act->type == ACT_ACTION) { - bActionActuator *aa = act->data; - aa->act= newlibadr_us(fd, ob->id.lib, aa->act); - } - else if (act->type == ACT_SHAPEACTION) { - bActionActuator *aa = act->data; - aa->act= newlibadr_us(fd, ob->id.lib, aa->act); - } - else if (act->type == ACT_PROPERTY) { - bPropertyActuator *pa = act->data; - pa->ob= newlibadr(fd, ob->id.lib, pa->ob); - } - else if (act->type == ACT_MESSAGE) { - bMessageActuator *ma = act->data; - ma->toObject= newlibadr(fd, ob->id.lib, ma->toObject); - } - else if (act->type == ACT_2DFILTER) { - bTwoDFilterActuator *_2dfa = act->data; - _2dfa->text= newlibadr(fd, ob->id.lib, _2dfa->text); - } - else if (act->type == ACT_PARENT) { - bParentActuator *parenta = act->data; - parenta->ob = newlibadr(fd, ob->id.lib, parenta->ob); - } - else if (act->type == ACT_STATE) { - /* bStateActuator *statea = act->data; */ - } - else if (act->type == ACT_ARMATURE) { - bArmatureActuator *arma= act->data; - arma->target= newlibadr(fd, ob->id.lib, arma->target); - arma->subtarget= newlibadr(fd, ob->id.lib, arma->subtarget); - } - else if (act->type == ACT_STEERING) { - bSteeringActuator *steeringa = act->data; - steeringa->target = newlibadr(fd, ob->id.lib, steeringa->target); - steeringa->navmesh = newlibadr(fd, ob->id.lib, steeringa->navmesh); - } - else if (act->type == ACT_MOUSE) { - /* bMouseActuator *moa= act->data; */ + case ACT_EDIT_OBJECT: + { + bEditObjectActuator *eoa = act->data; + if (eoa == NULL) { + init_actuator(act); + } + else { + eoa->ob = newlibadr(fd, ob->id.lib, eoa->ob); + eoa->me = newlibadr(fd, ob->id.lib, eoa->me); + } + break; + } + case ACT_SCENE: + { + bSceneActuator *sa = act->data; + sa->camera = newlibadr(fd, ob->id.lib, sa->camera); + sa->scene = newlibadr(fd, ob->id.lib, sa->scene); + break; + } + case ACT_ACTION: + { + bActionActuator *aa = act->data; + aa->act = newlibadr_us(fd, ob->id.lib, aa->act); + break; + } + case ACT_SHAPEACTION: + { + bActionActuator *aa = act->data; + aa->act = newlibadr_us(fd, ob->id.lib, aa->act); + break; + } + case ACT_PROPERTY: + { + bPropertyActuator *pa = act->data; + pa->ob = newlibadr(fd, ob->id.lib, pa->ob); + break; + } + case ACT_MESSAGE: + { + bMessageActuator *ma = act->data; + ma->toObject = newlibadr(fd, ob->id.lib, ma->toObject); + break; + } + case ACT_2DFILTER: + { + bTwoDFilterActuator *_2dfa = act->data; + _2dfa->text = newlibadr(fd, ob->id.lib, _2dfa->text); + break; + } + case ACT_PARENT: + { + bParentActuator *parenta = act->data; + parenta->ob = newlibadr(fd, ob->id.lib, parenta->ob); + break; + } + case ACT_STATE: + /* bStateActuator *statea = act->data; */ + break; + case ACT_ARMATURE: + { + bArmatureActuator *arma= act->data; + arma->target = newlibadr(fd, ob->id.lib, arma->target); + arma->subtarget = newlibadr(fd, ob->id.lib, arma->subtarget); + break; + } + case ACT_STEERING: + { + bSteeringActuator *steeringa = act->data; + steeringa->target = newlibadr(fd, ob->id.lib, steeringa->target); + steeringa->navmesh = newlibadr(fd, ob->id.lib, steeringa->navmesh); + break; + } + case ACT_MOUSE: + /* bMouseActuator *moa = act->data; */ + break; } } @@ -5317,6 +5321,37 @@ static void direct_link_modifiers(FileData *fd, ListBase *lb) MeshSeqCacheModifierData *msmcd = (MeshSeqCacheModifierData *)md; msmcd->reader = NULL; } + else if (md->type == eModifierType_SurfaceDeform) { + SurfaceDeformModifierData *smd = (SurfaceDeformModifierData *)md; + + smd->verts = newdataadr(fd, smd->verts); + + if (smd->verts) { + for (int i = 0; i < smd->numverts; i++) { + smd->verts[i].binds = newdataadr(fd, smd->verts[i].binds); + + if (smd->verts[i].binds) { + for (int j = 0; j < smd->verts[i].numbinds; j++) { + smd->verts[i].binds[j].vert_inds = newdataadr(fd, smd->verts[i].binds[j].vert_inds); + smd->verts[i].binds[j].vert_weights = newdataadr(fd, smd->verts[i].binds[j].vert_weights); + + if (fd->flags & FD_FLAGS_SWITCH_ENDIAN) { + if (smd->verts[i].binds[j].vert_inds) + BLI_endian_switch_uint32_array(smd->verts[i].binds[j].vert_inds, smd->verts[i].binds[j].numverts); + + if (smd->verts[i].binds[j].vert_weights) { + if (smd->verts[i].binds[j].mode == MOD_SDEF_MODE_CENTROID || + smd->verts[i].binds[j].mode == MOD_SDEF_MODE_LOOPTRI) + BLI_endian_switch_float_array(smd->verts[i].binds[j].vert_weights, 3); + else + BLI_endian_switch_float_array(smd->verts[i].binds[j].vert_weights, smd->verts[i].binds[j].numverts); + } + } + } + } + } + } + } } } @@ -5631,19 +5666,12 @@ static bool scene_validate_setscene__liblink(Scene *sce, const int totscene) static void lib_link_scene(FileData *fd, Main *main) { - Scene *sce; - Base *base, *next; - Sequence *seq; - SceneRenderLayer *srl; - FreestyleModuleConfig *fmc; - FreestyleLineSet *fls; - #ifdef USE_SETSCENE_CHECK bool need_check_set = false; int totscene = 0; #endif - for (sce = main->scene.first; sce; sce = sce->id.next) { + for (Scene *sce = main->scene.first; sce; sce = sce->id.next) { if (sce->id.tag & LIB_TAG_NEED_LINK) { /* Link ID Properties -- and copy this comment EXACTLY for easy finding * of library blocks that implement this.*/ @@ -5683,7 +5711,7 @@ static void lib_link_scene(FileData *fd, Main *main) sce->toolsettings->particle.shape_object = newlibadr(fd, sce->id.lib, sce->toolsettings->particle.shape_object); - for (base = sce->base.first; base; base = next) { + for (Base *next, *base = sce->base.first; base; base = next) { next = base->next; base->object = newlibadr_us(fd, sce->id.lib, base->object); @@ -5697,6 +5725,7 @@ static void lib_link_scene(FileData *fd, Main *main) } } + Sequence *seq; SEQ_BEGIN (sce->ed, seq) { if (seq->ipo) seq->ipo = newlibadr_us(fd, sce->id.lib, seq->ipo); // XXX deprecated - old animation system @@ -5762,13 +5791,13 @@ static void lib_link_scene(FileData *fd, Main *main) composite_patch(sce->nodetree, sce); } - for (srl = sce->r.layers.first; srl; srl = srl->next) { + for (SceneRenderLayer *srl = sce->r.layers.first; srl; srl = srl->next) { srl->mat_override = newlibadr_us(fd, sce->id.lib, srl->mat_override); srl->light_override = newlibadr_us(fd, sce->id.lib, srl->light_override); - for (fmc = srl->freestyleConfig.modules.first; fmc; fmc = fmc->next) { + for (FreestyleModuleConfig *fmc = srl->freestyleConfig.modules.first; fmc; fmc = fmc->next) { fmc->script = newlibadr(fd, sce->id.lib, fmc->script); } - for (fls = srl->freestyleConfig.linesets.first; fls; fls = fls->next) { + for (FreestyleLineSet *fls = srl->freestyleConfig.linesets.first; fls; fls = fls->next) { fls->linestyle = newlibadr_us(fd, sce->id.lib, fls->linestyle); fls->group = newlibadr_us(fd, sce->id.lib, fls->group); } @@ -5802,7 +5831,7 @@ static void lib_link_scene(FileData *fd, Main *main) #ifdef USE_SETSCENE_CHECK if (need_check_set) { - for (sce = main->scene.first; sce; sce = sce->id.next) { + for (Scene *sce = main->scene.first; sce; sce = sce->id.next) { if (sce->id.tag & LIB_TAG_NEED_LINK) { sce->id.tag &= ~LIB_TAG_NEED_LINK; if (!scene_validate_setscene__liblink(sce, totscene)) { @@ -6212,8 +6241,10 @@ static void lib_link_windowmanager(FileData *fd, Main *main) for (wm = main->wm.first; wm; wm = wm->id.next) { if (wm->id.tag & LIB_TAG_NEED_LINK) { - for (win = wm->windows.first; win; win = win->next) + /* Note: WM IDProperties are never written to file, hence no need to read/link them here. */ + for (win = wm->windows.first; win; win = win->next) { win->screen = newlibadr(fd, NULL, win->screen); + } wm->id.tag &= ~LIB_TAG_NEED_LINK; } @@ -6225,13 +6256,12 @@ static void lib_link_windowmanager(FileData *fd, Main *main) /* relink's grease pencil data's refs */ static void lib_link_gpencil(FileData *fd, Main *main) { - bGPdata *gpd; - - for (gpd = main->gpencil.first; gpd; gpd = gpd->id.next) { + for (bGPdata *gpd = main->gpencil.first; gpd; gpd = gpd->id.next) { if (gpd->id.tag & LIB_TAG_NEED_LINK) { - gpd->id.tag &= ~LIB_TAG_NEED_LINK; - + IDP_LibLinkProperty(gpd->id.properties, (fd->flags & FD_FLAGS_SWITCH_ENDIAN), fd); lib_link_animdata(fd, &gpd->id, gpd->adt); + + gpd->id.tag &= ~LIB_TAG_NEED_LINK; } } } @@ -6293,12 +6323,11 @@ static void direct_link_gpencil(FileData *fd, bGPdata *gpd) * check lib pointers in call below */ static void lib_link_screen(FileData *fd, Main *main) { - bScreen *sc; - ScrArea *sa; - - for (sc = main->screen.first; sc; sc = sc->id.next) { + for (bScreen *sc = main->screen.first; sc; sc = sc->id.next) { if (sc->id.tag & LIB_TAG_NEED_LINK) { + IDP_LibLinkProperty(sc->id.properties, (fd->flags & FD_FLAGS_SWITCH_ENDIAN), fd); id_us_ensure_real(&sc->id); + sc->scene = newlibadr(fd, sc->id.lib, sc->scene); /* this should not happen, but apparently it does somehow. Until we figure out the cause, @@ -6309,177 +6338,204 @@ static void lib_link_screen(FileData *fd, Main *main) sc->animtimer = NULL; /* saved in rare cases */ sc->scrubbing = false; - for (sa = sc->areabase.first; sa; sa = sa->next) { - SpaceLink *sl; - + for (ScrArea *sa = sc->areabase.first; sa; sa = sa->next) { sa->full = newlibadr(fd, sc->id.lib, sa->full); - for (sl = sa->spacedata.first; sl; sl= sl->next) { - if (sl->spacetype == SPACE_VIEW3D) { - View3D *v3d = (View3D*) sl; - BGpic *bgpic = NULL; - - v3d->camera= newlibadr(fd, sc->id.lib, v3d->camera); - v3d->ob_centre= newlibadr(fd, sc->id.lib, v3d->ob_centre); - - /* should be do_versions but not easy adding into the listbase */ - if (v3d->bgpic) { - v3d->bgpic = newlibadr(fd, sc->id.lib, v3d->bgpic); - BLI_addtail(&v3d->bgpicbase, bgpic); - v3d->bgpic = NULL; + for (SpaceLink *sl = sa->spacedata.first; sl; sl= sl->next) { + switch (sl->spacetype) { + case SPACE_VIEW3D: + { + View3D *v3d = (View3D*) sl; + BGpic *bgpic = NULL; + + v3d->camera= newlibadr(fd, sc->id.lib, v3d->camera); + v3d->ob_centre= newlibadr(fd, sc->id.lib, v3d->ob_centre); + + /* should be do_versions but not easy adding into the listbase */ + if (v3d->bgpic) { + v3d->bgpic = newlibadr(fd, sc->id.lib, v3d->bgpic); + BLI_addtail(&v3d->bgpicbase, bgpic); + v3d->bgpic = NULL; + } + + for (bgpic = v3d->bgpicbase.first; bgpic; bgpic = bgpic->next) { + bgpic->ima = newlibadr_us(fd, sc->id.lib, bgpic->ima); + bgpic->clip = newlibadr_us(fd, sc->id.lib, bgpic->clip); + } + if (v3d->localvd) { + v3d->localvd->camera = newlibadr(fd, sc->id.lib, v3d->localvd->camera); + } + break; } - - for (bgpic = v3d->bgpicbase.first; bgpic; bgpic = bgpic->next) { - bgpic->ima = newlibadr_us(fd, sc->id.lib, bgpic->ima); - bgpic->clip = newlibadr_us(fd, sc->id.lib, bgpic->clip); + case SPACE_IPO: + { + SpaceIpo *sipo = (SpaceIpo *)sl; + bDopeSheet *ads = sipo->ads; + + if (ads) { + ads->source = newlibadr(fd, sc->id.lib, ads->source); + ads->filter_grp = newlibadr(fd, sc->id.lib, ads->filter_grp); + } + break; } - if (v3d->localvd) { - v3d->localvd->camera = newlibadr(fd, sc->id.lib, v3d->localvd->camera); + case SPACE_BUTS: + { + SpaceButs *sbuts = (SpaceButs *)sl; + sbuts->pinid = newlibadr(fd, sc->id.lib, sbuts->pinid); + if (sbuts->pinid == NULL) { + sbuts->flag &= ~SB_PIN_CONTEXT; + } + break; } - } - else if (sl->spacetype == SPACE_IPO) { - SpaceIpo *sipo = (SpaceIpo *)sl; - bDopeSheet *ads = sipo->ads; - - if (ads) { - ads->source = newlibadr(fd, sc->id.lib, ads->source); - ads->filter_grp = newlibadr(fd, sc->id.lib, ads->filter_grp); + case SPACE_FILE: + break; + case SPACE_ACTION: + { + SpaceAction *saction = (SpaceAction *)sl; + bDopeSheet *ads = &saction->ads; + + if (ads) { + ads->source = newlibadr(fd, sc->id.lib, ads->source); + ads->filter_grp = newlibadr(fd, sc->id.lib, ads->filter_grp); + } + + saction->action = newlibadr(fd, sc->id.lib, saction->action); + break; } - } - else if (sl->spacetype == SPACE_BUTS) { - SpaceButs *sbuts = (SpaceButs *)sl; - sbuts->pinid = newlibadr(fd, sc->id.lib, sbuts->pinid); - if (sbuts->pinid == NULL) { - sbuts->flag &= ~SB_PIN_CONTEXT; + case SPACE_IMAGE: + { + SpaceImage *sima = (SpaceImage *)sl; + + sima->image = newlibadr_real_us(fd, sc->id.lib, sima->image); + sima->mask_info.mask = newlibadr_real_us(fd, sc->id.lib, sima->mask_info.mask); + + /* NOTE: pre-2.5, this was local data not lib data, but now we need this as lib data + * so fingers crossed this works fine! + */ + sima->gpd = newlibadr_us(fd, sc->id.lib, sima->gpd); + break; } - } - else if (sl->spacetype == SPACE_FILE) { - ; - } - else if (sl->spacetype == SPACE_ACTION) { - SpaceAction *saction = (SpaceAction *)sl; - bDopeSheet *ads = &saction->ads; - - if (ads) { - ads->source = newlibadr(fd, sc->id.lib, ads->source); - ads->filter_grp = newlibadr(fd, sc->id.lib, ads->filter_grp); + case SPACE_SEQ: + { + SpaceSeq *sseq = (SpaceSeq *)sl; + + /* NOTE: pre-2.5, this was local data not lib data, but now we need this as lib data + * so fingers crossed this works fine! + */ + sseq->gpd = newlibadr_us(fd, sc->id.lib, sseq->gpd); + break; } - - saction->action = newlibadr(fd, sc->id.lib, saction->action); - } - else if (sl->spacetype == SPACE_IMAGE) { - SpaceImage *sima = (SpaceImage *)sl; - - sima->image = newlibadr_real_us(fd, sc->id.lib, sima->image); - sima->mask_info.mask = newlibadr_real_us(fd, sc->id.lib, sima->mask_info.mask); - - /* NOTE: pre-2.5, this was local data not lib data, but now we need this as lib data - * so fingers crossed this works fine! - */ - sima->gpd = newlibadr_us(fd, sc->id.lib, sima->gpd); - } - else if (sl->spacetype == SPACE_SEQ) { - SpaceSeq *sseq = (SpaceSeq *)sl; - - /* NOTE: pre-2.5, this was local data not lib data, but now we need this as lib data - * so fingers crossed this works fine! - */ - sseq->gpd = newlibadr_us(fd, sc->id.lib, sseq->gpd); + case SPACE_NLA: + { + SpaceNla *snla= (SpaceNla *)sl; + bDopeSheet *ads= snla->ads; + + if (ads) { + ads->source = newlibadr(fd, sc->id.lib, ads->source); + ads->filter_grp = newlibadr(fd, sc->id.lib, ads->filter_grp); + } + break; + } + case SPACE_TEXT: + { + SpaceText *st= (SpaceText *)sl; - } - else if (sl->spacetype == SPACE_NLA) { - SpaceNla *snla= (SpaceNla *)sl; - bDopeSheet *ads= snla->ads; - - if (ads) { - ads->source = newlibadr(fd, sc->id.lib, ads->source); - ads->filter_grp = newlibadr(fd, sc->id.lib, ads->filter_grp); + st->text= newlibadr(fd, sc->id.lib, st->text); + break; } - } - else if (sl->spacetype == SPACE_TEXT) { - SpaceText *st= (SpaceText *)sl; - - st->text= newlibadr(fd, sc->id.lib, st->text); - } - else if (sl->spacetype == SPACE_SCRIPT) { - SpaceScript *scpt = (SpaceScript *)sl; - /*scpt->script = NULL; - 2.45 set to null, better re-run the script */ - if (scpt->script) { - scpt->script = newlibadr(fd, sc->id.lib, scpt->script); + case SPACE_SCRIPT: + { + SpaceScript *scpt = (SpaceScript *)sl; + /*scpt->script = NULL; - 2.45 set to null, better re-run the script */ if (scpt->script) { - SCRIPT_SET_NULL(scpt->script); + scpt->script = newlibadr(fd, sc->id.lib, scpt->script); + if (scpt->script) { + SCRIPT_SET_NULL(scpt->script); + } } + break; } - } - else if (sl->spacetype == SPACE_OUTLINER) { - SpaceOops *so= (SpaceOops *)sl; - so->search_tse.id = newlibadr(fd, NULL, so->search_tse.id); - - if (so->treestore) { - TreeStoreElem *tselem; - BLI_mempool_iter iter; + case SPACE_OUTLINER: + { + SpaceOops *so= (SpaceOops *)sl; + so->search_tse.id = newlibadr(fd, NULL, so->search_tse.id); + + if (so->treestore) { + TreeStoreElem *tselem; + BLI_mempool_iter iter; + + BLI_mempool_iternew(so->treestore, &iter); + while ((tselem = BLI_mempool_iterstep(&iter))) { + tselem->id = newlibadr(fd, NULL, tselem->id); + } + if (so->treehash) { + /* rebuild hash table, because it depends on ids too */ + so->storeflag |= SO_TREESTORE_REBUILD; + } + } + break; + } + case SPACE_NODE: + { + SpaceNode *snode = (SpaceNode *)sl; + bNodeTreePath *path, *path_next; + bNodeTree *ntree; + + /* node tree can be stored locally in id too, link this first */ + snode->id = newlibadr(fd, sc->id.lib, snode->id); + snode->from = newlibadr(fd, sc->id.lib, snode->from); + + ntree = snode->id ? ntreeFromID(snode->id) : NULL; + snode->nodetree = ntree ? ntree : newlibadr_us(fd, sc->id.lib, snode->nodetree); + + for (path = snode->treepath.first; path; path = path->next) { + if (path == snode->treepath.first) { + /* first nodetree in path is same as snode->nodetree */ + path->nodetree = snode->nodetree; + } + else + path->nodetree = newlibadr_us(fd, sc->id.lib, path->nodetree); - BLI_mempool_iternew(so->treestore, &iter); - while ((tselem = BLI_mempool_iterstep(&iter))) { - tselem->id = newlibadr(fd, NULL, tselem->id); + if (!path->nodetree) + break; } - if (so->treehash) { - /* rebuild hash table, because it depends on ids too */ - so->storeflag |= SO_TREESTORE_REBUILD; + + /* remaining path entries are invalid, remove */ + for (; path; path = path_next) { + path_next = path->next; + + BLI_remlink(&snode->treepath, path); + MEM_freeN(path); } - } - } - else if (sl->spacetype == SPACE_NODE) { - SpaceNode *snode = (SpaceNode *)sl; - bNodeTreePath *path, *path_next; - bNodeTree *ntree; - - /* node tree can be stored locally in id too, link this first */ - snode->id = newlibadr(fd, sc->id.lib, snode->id); - snode->from = newlibadr(fd, sc->id.lib, snode->from); - - ntree = snode->id ? ntreeFromID(snode->id) : NULL; - snode->nodetree = ntree ? ntree : newlibadr_us(fd, sc->id.lib, snode->nodetree); - - for (path = snode->treepath.first; path; path = path->next) { - if (path == snode->treepath.first) { - /* first nodetree in path is same as snode->nodetree */ - path->nodetree = snode->nodetree; + + /* edittree is just the last in the path, + * set this directly since the path may have been shortened above */ + if (snode->treepath.last) { + path = snode->treepath.last; + snode->edittree = path->nodetree; } - else - path->nodetree = newlibadr_us(fd, sc->id.lib, path->nodetree); - - if (!path->nodetree) - break; + else { + snode->edittree = NULL; + } + break; } - - /* remaining path entries are invalid, remove */ - for (; path; path = path_next) { - path_next = path->next; - - BLI_remlink(&snode->treepath, path); - MEM_freeN(path); + case SPACE_CLIP: + { + SpaceClip *sclip = (SpaceClip *)sl; + + sclip->clip = newlibadr_real_us(fd, sc->id.lib, sclip->clip); + sclip->mask_info.mask = newlibadr_real_us(fd, sc->id.lib, sclip->mask_info.mask); + break; } - - /* edittree is just the last in the path, - * set this directly since the path may have been shortened above */ - if (snode->treepath.last) { - path = snode->treepath.last; - snode->edittree = path->nodetree; + case SPACE_LOGIC: + { + SpaceLogic *slogic = (SpaceLogic *)sl; + + slogic->gpd = newlibadr_us(fd, sc->id.lib, slogic->gpd); + break; } - else - snode->edittree = NULL; - } - else if (sl->spacetype == SPACE_CLIP) { - SpaceClip *sclip = (SpaceClip *)sl; - - sclip->clip = newlibadr_real_us(fd, sc->id.lib, sclip->clip); - sclip->mask_info.mask = newlibadr_real_us(fd, sc->id.lib, sclip->mask_info.mask); - } - else if (sl->spacetype == SPACE_LOGIC) { - SpaceLogic *slogic = (SpaceLogic *)sl; - - slogic->gpd = newlibadr_us(fd, sc->id.lib, slogic->gpd); + default: + break; } } } @@ -7353,13 +7409,13 @@ static void fix_relpaths_library(const char *basepath, Main *main) static void lib_link_speaker(FileData *fd, Main *main) { - Speaker *spk; - - for (spk = main->speaker.first; spk; spk = spk->id.next) { + for (Speaker *spk = main->speaker.first; spk; spk = spk->id.next) { if (spk->id.tag & LIB_TAG_NEED_LINK) { + IDP_LibLinkProperty(spk->id.properties, (fd->flags & FD_FLAGS_SWITCH_ENDIAN), fd); lib_link_animdata(fd, &spk->id, spk->adt); spk->sound = newlibadr_us(fd, spk->id.lib, spk->sound); + spk->id.tag &= ~LIB_TAG_NEED_LINK; } } @@ -7409,14 +7465,15 @@ static void direct_link_sound(FileData *fd, bSound *sound) static void lib_link_sound(FileData *fd, Main *main) { - bSound *sound; - - for (sound = main->sound.first; sound; sound = sound->id.next) { + for (bSound *sound = main->sound.first; sound; sound = sound->id.next) { if (sound->id.tag & LIB_TAG_NEED_LINK) { - sound->id.tag &= ~LIB_TAG_NEED_LINK; + IDP_LibLinkProperty(sound->id.properties, (fd->flags & FD_FLAGS_SWITCH_ENDIAN), fd); + sound->ipo = newlibadr_us(fd, sound->id.lib, sound->ipo); // XXX deprecated - old animation system BKE_sound_load(main, sound); + + sound->id.tag &= ~LIB_TAG_NEED_LINK; } } } @@ -7431,17 +7488,13 @@ static void direct_link_group(FileData *fd, Group *group) static void lib_link_group(FileData *fd, Main *main) { - Group *group; - GroupObject *go; - bool add_us; - - for (group = main->group.first; group; group = group->id.next) { + for (Group *group = main->group.first; group; group = group->id.next) { if (group->id.tag & LIB_TAG_NEED_LINK) { - group->id.tag &= ~LIB_TAG_NEED_LINK; + IDP_LibLinkProperty(group->id.properties, (fd->flags & FD_FLAGS_SWITCH_ENDIAN), fd); - add_us = false; + bool add_us = false; - for (go = group->gobject.first; go; go = go->next) { + for (GroupObject *go = group->gobject.first; go; go = go->next) { go->ob = newlibadr_real_us(fd, group->id.lib, go->ob); if (go->ob) { go->ob->flag |= OB_FROMGROUP; @@ -7453,6 +7506,8 @@ static void lib_link_group(FileData *fd, Main *main) id_us_ensure_real(&group->id); } BKE_group_object_unlink(group, NULL, NULL, NULL); /* removes NULL entries */ + + group->id.tag &= ~LIB_TAG_NEED_LINK; } } } @@ -7557,13 +7612,11 @@ static void lib_link_moviePlaneTracks(FileData *fd, MovieClip *clip, ListBase *t static void lib_link_movieclip(FileData *fd, Main *main) { - MovieClip *clip; - - for (clip = main->movieclip.first; clip; clip = clip->id.next) { + for (MovieClip *clip = main->movieclip.first; clip; clip = clip->id.next) { if (clip->id.tag & LIB_TAG_NEED_LINK) { MovieTracking *tracking = &clip->tracking; - MovieTrackingObject *object; + IDP_LibLinkProperty(clip->id.properties, (fd->flags & FD_FLAGS_SWITCH_ENDIAN), fd); lib_link_animdata(fd, &clip->id, clip->adt); clip->gpd = newlibadr_us(fd, clip->id.lib, clip->gpd); @@ -7571,7 +7624,7 @@ static void lib_link_movieclip(FileData *fd, Main *main) lib_link_movieTracks(fd, clip, &tracking->tracks); lib_link_moviePlaneTracks(fd, clip, &tracking->plane_tracks); - for (object = tracking->objects.first; object; object = object->next) { + for (MovieTrackingObject *object = tracking->objects.first; object; object = object->next) { lib_link_movieTracks(fd, clip, &object->tracks); lib_link_moviePlaneTracks(fd, clip, &object->plane_tracks); } @@ -7648,16 +7701,12 @@ static void lib_link_mask_parent(FileData *fd, Mask *mask, MaskParent *parent) static void lib_link_mask(FileData *fd, Main *main) { - Mask *mask; - - mask = main->mask.first; - while (mask) { + for (Mask *mask = main->mask.first; mask; mask = mask->id.next) { if (mask->id.tag & LIB_TAG_NEED_LINK) { - MaskLayer *masklay; - + IDP_LibLinkProperty(mask->id.properties, (fd->flags & FD_FLAGS_SWITCH_ENDIAN), fd); lib_link_animdata(fd, &mask->id, mask->adt); - for (masklay = mask->masklayers.first; masklay; masklay = masklay->next) { + for (MaskLayer *masklay = mask->masklayers.first; masklay; masklay = masklay->next) { MaskSpline *spline; spline = masklay->splines.first; @@ -7678,7 +7727,6 @@ static void lib_link_mask(FileData *fd, Main *main) mask->id.tag &= ~LIB_TAG_NEED_LINK; } - mask = mask->id.next; } } @@ -7686,18 +7734,13 @@ static void lib_link_mask(FileData *fd, Main *main) static void lib_link_linestyle(FileData *fd, Main *main) { - FreestyleLineStyle *linestyle; - LineStyleModifier *m; - MTex *mtex; - int a; - - linestyle = main->linestyle.first; - while (linestyle) { + for (FreestyleLineStyle *linestyle = main->linestyle.first; linestyle; linestyle = linestyle->id.next) { if (linestyle->id.tag & LIB_TAG_NEED_LINK) { - linestyle->id.tag &= ~LIB_TAG_NEED_LINK; + LineStyleModifier *m; IDP_LibLinkProperty(linestyle->id.properties, (fd->flags & FD_FLAGS_SWITCH_ENDIAN), fd); lib_link_animdata(fd, &linestyle->id, linestyle->adt); + for (m = linestyle->color_modifiers.first; m; m = m->next) { switch (m->type) { case LS_MODIFIER_DISTANCE_FROM_OBJECT: @@ -7728,8 +7771,8 @@ static void lib_link_linestyle(FileData *fd, Main *main) break; } } - for (a=0; a < MAX_MTEX; a++) { - mtex = linestyle->mtex[a]; + for (int a = 0; a < MAX_MTEX; a++) { + MTex *mtex = linestyle->mtex[a]; if (mtex) { mtex->tex = newlibadr_us(fd, linestyle->id.lib, mtex->tex); mtex->object = newlibadr(fd, linestyle->id.lib, mtex->object); @@ -7739,8 +7782,9 @@ static void lib_link_linestyle(FileData *fd, Main *main) lib_link_ntree(fd, &linestyle->id, linestyle->nodetree); linestyle->nodetree->id.lib = linestyle->id.lib; } + + linestyle->id.tag &= ~LIB_TAG_NEED_LINK; } - linestyle = linestyle->id.next; } } @@ -8411,12 +8455,13 @@ static void lib_link_all(FileData *fd, Main *main) lib_link_screen(fd, main); lib_link_scene(fd, main); lib_link_object(fd, main); + lib_link_mesh(fd, main); lib_link_curve(fd, main); lib_link_mball(fd, main); lib_link_material(fd, main); lib_link_texture(fd, main); lib_link_image(fd, main); - lib_link_ipo(fd, main); // XXX deprecated... still needs to be maintained for version patches still + lib_link_ipo(fd, main); /* XXX deprecated... still needs to be maintained for version patches still */ lib_link_key(fd, main); lib_link_world(fd, main); lib_link_lamp(fd, main); @@ -8429,7 +8474,7 @@ static void lib_link_all(FileData *fd, Main *main) lib_link_armature(fd, main); lib_link_action(fd, main); lib_link_vfont(fd, main); - lib_link_nodetree(fd, main); /* has to be done after scene/materials, this will verify group nodes */ + lib_link_nodetree(fd, main); /* has to be done after scene/materials, this will verify group nodes */ lib_link_brush(fd, main); lib_link_palette(fd, main); lib_link_paint_curve(fd, main); @@ -8440,9 +8485,7 @@ static void lib_link_all(FileData *fd, Main *main) lib_link_gpencil(fd, main); lib_link_cachefiles(fd, main); - lib_link_mesh(fd, main); /* as last: tpage images with users at zero */ - - lib_link_library(fd, main); /* only init users */ + lib_link_library(fd, main); /* only init users */ } static void direct_link_keymapitem(FileData *fd, wmKeyMapItem *kmi) @@ -8566,7 +8609,12 @@ BlendFileData *blo_read_file_internal(FileData *fd, const char *filepath) bhead = read_global(bfd, fd, bhead); break; case USER: - bhead = read_userdef(bfd, fd, bhead); + if (fd->skip_flags & BLO_READ_SKIP_USERDEF) { + bhead = blo_nextbhead(fd, bhead); + } + else { + bhead = read_userdef(bfd, fd, bhead); + } break; case ENDB: bhead = NULL; @@ -8575,15 +8623,24 @@ BlendFileData *blo_read_file_internal(FileData *fd, const char *filepath) case ID_ID: /* Always adds to the most recently loaded ID_LI block, see direct_link_library. * This is part of the file format definition. */ - bhead = read_libblock(fd, mainlist.last, bhead, LIB_TAG_READ | LIB_TAG_EXTERN, NULL); + if (fd->skip_flags & BLO_READ_SKIP_DATA) { + bhead = blo_nextbhead(fd, bhead); + } + else { + bhead = read_libblock(fd, mainlist.last, bhead, LIB_TAG_READ | LIB_TAG_EXTERN, NULL); + } break; - /* in 2.50+ files, the file identifier for screens is patched, forward compatibility */ case ID_SCRN: bhead->code = ID_SCR; /* deliberate pass on to default */ default: - bhead = read_libblock(fd, bfd->main, bhead, LIB_TAG_LOCAL, NULL); + if (fd->skip_flags & BLO_READ_SKIP_DATA) { + bhead = blo_nextbhead(fd, bhead); + } + else { + bhead = read_libblock(fd, bfd->main, bhead, LIB_TAG_LOCAL, NULL); + } } } @@ -10395,6 +10452,9 @@ static void read_libraries(FileData *basefd, ListBase *mainlist) else { mainptr->curlib->filedata = NULL; mainptr->curlib->id.tag |= LIB_TAG_MISSING; + /* Set lib version to current main one... Makes assert later happy. */ + mainptr->versionfile = mainptr->curlib->versionfile = mainl->versionfile; + mainptr->subversionfile = mainptr->curlib->subversionfile = mainl->subversionfile; } if (fd == NULL) { diff --git a/source/blender/blenloader/intern/readfile.h b/source/blender/blenloader/intern/readfile.h index d97bef13a78..62ce15a640e 100644 --- a/source/blender/blenloader/intern/readfile.h +++ b/source/blender/blenloader/intern/readfile.h @@ -81,6 +81,8 @@ typedef struct FileData { int id_name_offs; /* used to retrieve ID names from (bhead+1) */ int globalf, fileflags; /* for do_versions patching */ + eBLOReadSkip skip_flags; /* skip some data-blocks */ + struct OldNewMap *datamap; struct OldNewMap *globmap; struct OldNewMap *libmap; diff --git a/source/blender/blenloader/intern/versioning_270.c b/source/blender/blenloader/intern/versioning_270.c index 6235ebf28ee..d3f33cf725f 100644 --- a/source/blender/blenloader/intern/versioning_270.c +++ b/source/blender/blenloader/intern/versioning_270.c @@ -1575,6 +1575,40 @@ void blo_do_versions_270(FileData *fd, Library *UNUSED(lib), Main *main) } } + /* Fix for T50736, Glare comp node using same var for two different things. */ + if (!DNA_struct_elem_find(fd->filesdna, "NodeGlare", "char", "star_45")) { + FOREACH_NODETREE(main, ntree, id) { + if (ntree->type == NTREE_COMPOSIT) { + ntreeSetTypes(NULL, ntree); + for (bNode *node = ntree->nodes.first; node; node = node->next) { + if (node->type == CMP_NODE_GLARE) { + NodeGlare *ndg = node->storage; + switch (ndg->type) { + case 2: /* Grrrr! magic numbers :( */ + ndg->streaks = ndg->angle; + break; + case 0: + ndg->star_45 = ndg->angle != 0; + break; + default: + break; + } + } + } + } + } FOREACH_NODETREE_END + } + + if (!DNA_struct_elem_find(fd->filesdna, "SurfaceDeformModifierData", "float", "mat[4][4]")) { + for (Object *ob = main->object.first; ob; ob = ob->id.next) { + for (ModifierData *md = ob->modifiers.first; md; md = md->next) { + if (md->type == eModifierType_SurfaceDeform) { + SurfaceDeformModifierData *smd = (SurfaceDeformModifierData *)md; + unit_m4(smd->mat); + } + } + } + } } } diff --git a/source/blender/blenloader/intern/versioning_defaults.c b/source/blender/blenloader/intern/versioning_defaults.c index 99d9e140481..e34f12b1cf9 100644 --- a/source/blender/blenloader/intern/versioning_defaults.c +++ b/source/blender/blenloader/intern/versioning_defaults.c @@ -68,6 +68,18 @@ void BLO_update_defaults_userpref_blend(void) * but take care since some hardware has driver bugs here (T46962). * Further hardware workarounds should be made in gpu_extensions.c */ U.glalphaclip = (1.0f / 255); + + /* default so DPI is detected automatically */ + U.dpi = 0; + U.ui_scale = 1.0f; + +#ifdef WITH_PYTHON_SECURITY + /* use alternative setting for security nuts + * otherwise we'd need to patch the binary blob - startup.blend.c */ + U.flag |= USER_SCRIPT_AUTOEXEC_DISABLE; +#else + U.flag &= ~USER_SCRIPT_AUTOEXEC_DISABLE; +#endif } /** diff --git a/source/blender/blenloader/intern/writefile.c b/source/blender/blenloader/intern/writefile.c index 7b8b95f0005..57be237be6f 100644 --- a/source/blender/blenloader/intern/writefile.c +++ b/source/blender/blenloader/intern/writefile.c @@ -78,7 +78,7 @@ * - write #TEST (#RenderInfo struct. 128x128 blend file preview is optional). * - write #GLOB (#FileGlobal struct) (some global vars). * - write #DNA1 (#SDNA struct) - * - write #USER (#UserDef struct) if filename is ``~/X.XX/config/startup.blend``. + * - write #USER (#UserDef struct) if filename is ``~/.config/blender/X.XX/config/startup.blend``. */ @@ -797,30 +797,22 @@ static void write_fcurves(WriteData *wd, ListBase *fcurves) } } -static void write_actions(WriteData *wd, ListBase *idbase) +static void write_action(WriteData *wd, bAction *act) { - bAction *act; - bActionGroup *grp; - TimeMarker *marker; - - for (act = idbase->first; act; act = act->id.next) { - if (act->id.us > 0 || wd->current) { - writestruct(wd, ID_AC, bAction, 1, act); - write_iddata(wd, &act->id); + if (act->id.us > 0 || wd->current) { + writestruct(wd, ID_AC, bAction, 1, act); + write_iddata(wd, &act->id); - write_fcurves(wd, &act->curves); + write_fcurves(wd, &act->curves); - for (grp = act->groups.first; grp; grp = grp->next) { - writestruct(wd, DATA, bActionGroup, 1, grp); - } + for (bActionGroup *grp = act->groups.first; grp; grp = grp->next) { + writestruct(wd, DATA, bActionGroup, 1, grp); + } - for (marker = act->markers.first; marker; marker = marker->next) { - writestruct(wd, DATA, TimeMarker, 1, marker); - } + for (TimeMarker *marker = act->markers.first; marker; marker = marker->next) { + writestruct(wd, DATA, TimeMarker, 1, marker); } } - - mywrite_flush(wd); } static void write_keyingsets(WriteData *wd, ListBase *list) @@ -969,7 +961,7 @@ static void write_node_socket_interface(WriteData *wd, bNodeTree *UNUSED(ntree), } } /* this is only direct data, tree itself should have been written */ -static void write_nodetree(WriteData *wd, bNodeTree *ntree) +static void write_nodetree_nolib(WriteData *wd, bNodeTree *ntree) { bNode *node; bNodeSocket *sock; @@ -1030,6 +1022,25 @@ static void write_nodetree(WriteData *wd, bNodeTree *ntree) { /* pass */ } + else if ((ntree->type == NTREE_COMPOSIT) && (node->type == CMP_NODE_GLARE)) { + /* Simple forward compat for fix for T50736. + * Not ideal (there is no ideal solution here), but should do for now. */ + NodeGlare *ndg = node->storage; + /* Not in undo case. */ + if (!wd->current) { + switch (ndg->type) { + case 2: /* Grrrr! magic numbers :( */ + ndg->angle = ndg->streaks; + break; + case 0: + ndg->angle = ndg->star_45; + break; + default: + break; + } + } + writestruct_id(wd, DATA, node->typeinfo->storagename, 1, node->storage); + } else { writestruct_id(wd, DATA, node->typeinfo->storagename, 1, node->storage); } @@ -1278,68 +1289,60 @@ static void write_pointcaches(WriteData *wd, ListBase *ptcaches) } } } -static void write_particlesettings(WriteData *wd, ListBase *idbase) -{ - ParticleSettings *part; - ParticleDupliWeight *dw; - GroupObject *go; - int a; - part = idbase->first; - while (part) { - if (part->id.us > 0 || wd->current) { - /* write LibData */ - writestruct(wd, ID_PA, ParticleSettings, 1, part); - write_iddata(wd, &part->id); +static void write_particlesettings(WriteData *wd, ParticleSettings *part) +{ + if (part->id.us > 0 || wd->current) { + /* write LibData */ + writestruct(wd, ID_PA, ParticleSettings, 1, part); + write_iddata(wd, &part->id); - if (part->adt) { - write_animdata(wd, part->adt); - } - writestruct(wd, DATA, PartDeflect, 1, part->pd); - writestruct(wd, DATA, PartDeflect, 1, part->pd2); - writestruct(wd, DATA, EffectorWeights, 1, part->effector_weights); + if (part->adt) { + write_animdata(wd, part->adt); + } + writestruct(wd, DATA, PartDeflect, 1, part->pd); + writestruct(wd, DATA, PartDeflect, 1, part->pd2); + writestruct(wd, DATA, EffectorWeights, 1, part->effector_weights); - if (part->clumpcurve) { - write_curvemapping(wd, part->clumpcurve); - } - if (part->roughcurve) { - write_curvemapping(wd, part->roughcurve); - } + if (part->clumpcurve) { + write_curvemapping(wd, part->clumpcurve); + } + if (part->roughcurve) { + write_curvemapping(wd, part->roughcurve); + } - dw = part->dupliweights.first; - for (; dw; dw = dw->next) { - /* update indices, but only if dw->ob is set (can be NULL after loading e.g.) */ - if (dw->ob != NULL) { - dw->index = 0; - if (part->dup_group) { /* can be NULL if lining fails or set to None */ - for (go = part->dup_group->gobject.first; go && go->ob != dw->ob; go = go->next, dw->index++); - } + for (ParticleDupliWeight *dw = part->dupliweights.first; dw; dw = dw->next) { + /* update indices, but only if dw->ob is set (can be NULL after loading e.g.) */ + if (dw->ob != NULL) { + dw->index = 0; + if (part->dup_group) { /* can be NULL if lining fails or set to None */ + for (GroupObject *go = part->dup_group->gobject.first; + go && go->ob != dw->ob; + go = go->next, dw->index++); } - writestruct(wd, DATA, ParticleDupliWeight, 1, dw); } + writestruct(wd, DATA, ParticleDupliWeight, 1, dw); + } - if (part->boids && part->phystype == PART_PHYS_BOIDS) { - BoidState *state = part->boids->states.first; - - writestruct(wd, DATA, BoidSettings, 1, part->boids); + if (part->boids && part->phystype == PART_PHYS_BOIDS) { + writestruct(wd, DATA, BoidSettings, 1, part->boids); - for (; state; state = state->next) { - write_boid_state(wd, state); - } - } - if (part->fluid && part->phystype == PART_PHYS_FLUID) { - writestruct(wd, DATA, SPHFluidSettings, 1, part->fluid); + for (BoidState *state = part->boids->states.first; state; state = state->next) { + write_boid_state(wd, state); } + } + if (part->fluid && part->phystype == PART_PHYS_FLUID) { + writestruct(wd, DATA, SPHFluidSettings, 1, part->fluid); + } - for (a = 0; a < MAX_MTEX; a++) { - if (part->mtex[a]) { - writestruct(wd, DATA, MTex, 1, part->mtex[a]); - } + for (int a = 0; a < MAX_MTEX; a++) { + if (part->mtex[a]) { + writestruct(wd, DATA, MTex, 1, part->mtex[a]); } } - part = part->id.next; } } + static void write_particlesystems(WriteData *wd, ListBase *particles) { ParticleSystem *psys = particles->first; @@ -1830,242 +1833,207 @@ static void write_modifiers(WriteData *wd, ListBase *modbase) writedata(wd, DATA, sizeof(float[3]) * csmd->bind_coords_num, csmd->bind_coords); } } - } -} + else if (md->type == eModifierType_SurfaceDeform) { + SurfaceDeformModifierData *smd = (SurfaceDeformModifierData *)md; -static void write_objects(WriteData *wd, ListBase *idbase) -{ - Object *ob; + writestruct(wd, DATA, SDefVert, smd->numverts, smd->verts); - ob = idbase->first; - while (ob) { - if (ob->id.us > 0 || wd->current) { - /* write LibData */ - writestruct(wd, ID_OB, Object, 1, ob); - write_iddata(wd, &ob->id); + if (smd->verts) { + for (int i = 0; i < smd->numverts; i++) { + writestruct(wd, DATA, SDefBind, smd->verts[i].numbinds, smd->verts[i].binds); - if (ob->adt) { - write_animdata(wd, ob->adt); - } + if (smd->verts[i].binds) { + for (int j = 0; j < smd->verts[i].numbinds; j++) { + writedata(wd, DATA, sizeof(int) * smd->verts[i].binds[j].numverts, smd->verts[i].binds[j].vert_inds); - /* direct data */ - writedata(wd, DATA, sizeof(void *) * ob->totcol, ob->mat); - writedata(wd, DATA, sizeof(char) * ob->totcol, ob->matbits); - /* write_effects(wd, &ob->effect); */ /* not used anymore */ - write_properties(wd, &ob->prop); - write_sensors(wd, &ob->sensors); - write_controllers(wd, &ob->controllers); - write_actuators(wd, &ob->actuators); - - if (ob->type == OB_ARMATURE) { - bArmature *arm = ob->data; - if (arm && ob->pose && arm->act_bone) { - BLI_strncpy(ob->pose->proxy_act_bone, arm->act_bone->name, sizeof(ob->pose->proxy_act_bone)); + if (smd->verts[i].binds[j].mode == MOD_SDEF_MODE_CENTROID || + smd->verts[i].binds[j].mode == MOD_SDEF_MODE_LOOPTRI) + { + writedata(wd, DATA, sizeof(float) * 3, smd->verts[i].binds[j].vert_weights); + } + else { + writedata(wd, DATA, sizeof(float) * smd->verts[i].binds[j].numverts, smd->verts[i].binds[j].vert_weights); + } + } + } } } + } + } +} - write_pose(wd, ob->pose); - write_defgroups(wd, &ob->defbase); - write_constraints(wd, &ob->constraints); - write_motionpath(wd, ob->mpath); +static void write_object(WriteData *wd, Object *ob) +{ + if (ob->id.us > 0 || wd->current) { + /* write LibData */ + writestruct(wd, ID_OB, Object, 1, ob); + write_iddata(wd, &ob->id); - writestruct(wd, DATA, PartDeflect, 1, ob->pd); - writestruct(wd, DATA, SoftBody, 1, ob->soft); - if (ob->soft) { - write_pointcaches(wd, &ob->soft->ptcaches); - writestruct(wd, DATA, EffectorWeights, 1, ob->soft->effector_weights); - } - writestruct(wd, DATA, BulletSoftBody, 1, ob->bsoft); + if (ob->adt) { + write_animdata(wd, ob->adt); + } - if (ob->rigidbody_object) { - /* TODO: if any extra data is added to handle duplis, will need separate function then */ - writestruct(wd, DATA, RigidBodyOb, 1, ob->rigidbody_object); - } - if (ob->rigidbody_constraint) { - writestruct(wd, DATA, RigidBodyCon, 1, ob->rigidbody_constraint); - } + /* direct data */ + writedata(wd, DATA, sizeof(void *) * ob->totcol, ob->mat); + writedata(wd, DATA, sizeof(char) * ob->totcol, ob->matbits); + /* write_effects(wd, &ob->effect); */ /* not used anymore */ + write_properties(wd, &ob->prop); + write_sensors(wd, &ob->sensors); + write_controllers(wd, &ob->controllers); + write_actuators(wd, &ob->actuators); - if (ob->type == OB_EMPTY && ob->empty_drawtype == OB_EMPTY_IMAGE) { - writestruct(wd, DATA, ImageUser, 1, ob->iuser); + if (ob->type == OB_ARMATURE) { + bArmature *arm = ob->data; + if (arm && ob->pose && arm->act_bone) { + BLI_strncpy(ob->pose->proxy_act_bone, arm->act_bone->name, sizeof(ob->pose->proxy_act_bone)); } + } - write_particlesystems(wd, &ob->particlesystem); - write_modifiers(wd, &ob->modifiers); + write_pose(wd, ob->pose); + write_defgroups(wd, &ob->defbase); + write_constraints(wd, &ob->constraints); + write_motionpath(wd, ob->mpath); - writelist(wd, DATA, LinkData, &ob->pc_ids); - writelist(wd, DATA, LodLevel, &ob->lodlevels); + writestruct(wd, DATA, PartDeflect, 1, ob->pd); + writestruct(wd, DATA, SoftBody, 1, ob->soft); + if (ob->soft) { + write_pointcaches(wd, &ob->soft->ptcaches); + writestruct(wd, DATA, EffectorWeights, 1, ob->soft->effector_weights); } + writestruct(wd, DATA, BulletSoftBody, 1, ob->bsoft); - write_previews(wd, ob->preview); + if (ob->rigidbody_object) { + /* TODO: if any extra data is added to handle duplis, will need separate function then */ + writestruct(wd, DATA, RigidBodyOb, 1, ob->rigidbody_object); + } + if (ob->rigidbody_constraint) { + writestruct(wd, DATA, RigidBodyCon, 1, ob->rigidbody_constraint); + } - ob = ob->id.next; - } + if (ob->type == OB_EMPTY && ob->empty_drawtype == OB_EMPTY_IMAGE) { + writestruct(wd, DATA, ImageUser, 1, ob->iuser); + } - mywrite_flush(wd); -} + write_particlesystems(wd, &ob->particlesystem); + write_modifiers(wd, &ob->modifiers); + writelist(wd, DATA, LinkData, &ob->pc_ids); + writelist(wd, DATA, LodLevel, &ob->lodlevels); -static void write_vfonts(WriteData *wd, ListBase *idbase) -{ - VFont *vf; - PackedFile *pf; + write_previews(wd, ob->preview); + } +} - vf = idbase->first; - while (vf) { - if (vf->id.us > 0 || wd->current) { - /* write LibData */ - writestruct(wd, ID_VF, VFont, 1, vf); - write_iddata(wd, &vf->id); - /* direct data */ +static void write_vfont(WriteData *wd, VFont *vf) +{ + if (vf->id.us > 0 || wd->current) { + /* write LibData */ + writestruct(wd, ID_VF, VFont, 1, vf); + write_iddata(wd, &vf->id); - if (vf->packedfile) { - pf = vf->packedfile; - writestruct(wd, DATA, PackedFile, 1, pf); - writedata(wd, DATA, pf->size, pf->data); - } + /* direct data */ + if (vf->packedfile) { + PackedFile *pf = vf->packedfile; + writestruct(wd, DATA, PackedFile, 1, pf); + writedata(wd, DATA, pf->size, pf->data); } - - vf = vf->id.next; } - - mywrite_flush(wd); } -static void write_keys(WriteData *wd, ListBase *idbase) +static void write_key(WriteData *wd, Key *key) { - Key *key; - KeyBlock *kb; - - key = idbase->first; - while (key) { - if (key->id.us > 0 || wd->current) { - /* write LibData */ - writestruct(wd, ID_KE, Key, 1, key); - write_iddata(wd, &key->id); - - if (key->adt) { - write_animdata(wd, key->adt); - } + if (key->id.us > 0 || wd->current) { + /* write LibData */ + writestruct(wd, ID_KE, Key, 1, key); + write_iddata(wd, &key->id); - /* direct data */ - kb = key->block.first; - while (kb) { - writestruct(wd, DATA, KeyBlock, 1, kb); - if (kb->data) { - writedata(wd, DATA, kb->totelem * key->elemsize, kb->data); - } - kb = kb->next; - } + if (key->adt) { + write_animdata(wd, key->adt); } - key = key->id.next; + /* direct data */ + for (KeyBlock *kb = key->block.first; kb; kb = kb->next) { + writestruct(wd, DATA, KeyBlock, 1, kb); + if (kb->data) { + writedata(wd, DATA, kb->totelem * key->elemsize, kb->data); + } + } } - - mywrite_flush(wd); } -static void write_cameras(WriteData *wd, ListBase *idbase) +static void write_camera(WriteData *wd, Camera *cam) { - Camera *cam; - - cam = idbase->first; - while (cam) { - if (cam->id.us > 0 || wd->current) { - /* write LibData */ - writestruct(wd, ID_CA, Camera, 1, cam); - write_iddata(wd, &cam->id); + if (cam->id.us > 0 || wd->current) { + /* write LibData */ + writestruct(wd, ID_CA, Camera, 1, cam); + write_iddata(wd, &cam->id); - if (cam->adt) { - write_animdata(wd, cam->adt); - } + if (cam->adt) { + write_animdata(wd, cam->adt); } - - cam = cam->id.next; } } -static void write_mballs(WriteData *wd, ListBase *idbase) +static void write_mball(WriteData *wd, MetaBall *mb) { - MetaBall *mb; - MetaElem *ml; - - mb = idbase->first; - while (mb) { - if (mb->id.us > 0 || wd->current) { - /* write LibData */ - writestruct(wd, ID_MB, MetaBall, 1, mb); - write_iddata(wd, &mb->id); + if (mb->id.us > 0 || wd->current) { + /* write LibData */ + writestruct(wd, ID_MB, MetaBall, 1, mb); + write_iddata(wd, &mb->id); - /* direct data */ - writedata(wd, DATA, sizeof(void *) * mb->totcol, mb->mat); - if (mb->adt) { - write_animdata(wd, mb->adt); - } + /* direct data */ + writedata(wd, DATA, sizeof(void *) * mb->totcol, mb->mat); + if (mb->adt) { + write_animdata(wd, mb->adt); + } - ml = mb->elems.first; - while (ml) { - writestruct(wd, DATA, MetaElem, 1, ml); - ml = ml->next; - } + for (MetaElem *ml = mb->elems.first; ml; ml = ml->next) { + writestruct(wd, DATA, MetaElem, 1, ml); } - mb = mb->id.next; } } -static void write_curves(WriteData *wd, ListBase *idbase) +static void write_curve(WriteData *wd, Curve *cu) { - Curve *cu; - Nurb *nu; - - cu = idbase->first; - while (cu) { - if (cu->id.us > 0 || wd->current) { - /* write LibData */ - writestruct(wd, ID_CU, Curve, 1, cu); - write_iddata(wd, &cu->id); + if (cu->id.us > 0 || wd->current) { + /* write LibData */ + writestruct(wd, ID_CU, Curve, 1, cu); + write_iddata(wd, &cu->id); - /* direct data */ - writedata(wd, DATA, sizeof(void *) * cu->totcol, cu->mat); - if (cu->adt) { - write_animdata(wd, cu->adt); - } + /* direct data */ + writedata(wd, DATA, sizeof(void *) * cu->totcol, cu->mat); + if (cu->adt) { + write_animdata(wd, cu->adt); + } - if (cu->vfont) { - writedata(wd, DATA, cu->len + 1, cu->str); - writestruct(wd, DATA, CharInfo, cu->len_wchar + 1, cu->strinfo); - writestruct(wd, DATA, TextBox, cu->totbox, cu->tb); + if (cu->vfont) { + writedata(wd, DATA, cu->len + 1, cu->str); + writestruct(wd, DATA, CharInfo, cu->len_wchar + 1, cu->strinfo); + writestruct(wd, DATA, TextBox, cu->totbox, cu->tb); + } + else { + /* is also the order of reading */ + for (Nurb *nu = cu->nurb.first; nu; nu = nu->next) { + writestruct(wd, DATA, Nurb, 1, nu); } - else { - /* is also the order of reading */ - nu = cu->nurb.first; - while (nu) { - writestruct(wd, DATA, Nurb, 1, nu); - nu = nu->next; + for (Nurb *nu = cu->nurb.first; nu; nu = nu->next) { + if (nu->type == CU_BEZIER) { + writestruct(wd, DATA, BezTriple, nu->pntsu, nu->bezt); } - nu = cu->nurb.first; - while (nu) { - if (nu->type == CU_BEZIER) { - writestruct(wd, DATA, BezTriple, nu->pntsu, nu->bezt); + else { + writestruct(wd, DATA, BPoint, nu->pntsu * nu->pntsv, nu->bp); + if (nu->knotsu) { + writedata(wd, DATA, KNOTSU(nu) * sizeof(float), nu->knotsu); } - else { - writestruct(wd, DATA, BPoint, nu->pntsu * nu->pntsv, nu->bp); - if (nu->knotsu) { - writedata(wd, DATA, KNOTSU(nu) * sizeof(float), nu->knotsu); - } - if (nu->knotsv) { - writedata(wd, DATA, KNOTSV(nu) * sizeof(float), nu->knotsv); - } + if (nu->knotsv) { + writedata(wd, DATA, KNOTSV(nu) * sizeof(float), nu->knotsv); } - nu = nu->next; } } } - cu = cu->id.next; } - - mywrite_flush(wd); } static void write_dverts(WriteData *wd, int count, MDeformVert *dvlist) @@ -2182,407 +2150,349 @@ static void write_customdata( } } -static void write_meshes(WriteData *wd, ListBase *idbase) +static void write_mesh(WriteData *wd, Mesh *mesh) { - Mesh *mesh; - bool save_for_old_blender = false; - #ifdef USE_BMESH_SAVE_AS_COMPAT - save_for_old_blender = wd->use_mesh_compat; /* option to save with older mesh format */ + const bool save_for_old_blender = wd->use_mesh_compat; /* option to save with older mesh format */ +#else + const bool save_for_old_blender = false; #endif - mesh = idbase->first; - while (mesh) { - CustomDataLayer *vlayers = NULL, vlayers_buff[CD_TEMP_CHUNK_SIZE]; - CustomDataLayer *elayers = NULL, elayers_buff[CD_TEMP_CHUNK_SIZE]; - CustomDataLayer *flayers = NULL, flayers_buff[CD_TEMP_CHUNK_SIZE]; - CustomDataLayer *llayers = NULL, llayers_buff[CD_TEMP_CHUNK_SIZE]; - CustomDataLayer *players = NULL, players_buff[CD_TEMP_CHUNK_SIZE]; - - if (mesh->id.us > 0 || wd->current) { - /* write LibData */ - if (!save_for_old_blender) { - /* write a copy of the mesh, don't modify in place because it is - * not thread safe for threaded renders that are reading this */ - Mesh *old_mesh = mesh; - Mesh copy_mesh = *mesh; - mesh = ©_mesh; + CustomDataLayer *vlayers = NULL, vlayers_buff[CD_TEMP_CHUNK_SIZE]; + CustomDataLayer *elayers = NULL, elayers_buff[CD_TEMP_CHUNK_SIZE]; + CustomDataLayer *flayers = NULL, flayers_buff[CD_TEMP_CHUNK_SIZE]; + CustomDataLayer *llayers = NULL, llayers_buff[CD_TEMP_CHUNK_SIZE]; + CustomDataLayer *players = NULL, players_buff[CD_TEMP_CHUNK_SIZE]; + + if (mesh->id.us > 0 || wd->current) { + /* write LibData */ + if (!save_for_old_blender) { + /* write a copy of the mesh, don't modify in place because it is + * not thread safe for threaded renders that are reading this */ + Mesh *old_mesh = mesh; + Mesh copy_mesh = *mesh; + mesh = ©_mesh; #ifdef USE_BMESH_SAVE_WITHOUT_MFACE - /* cache only - don't write */ - mesh->mface = NULL; - mesh->totface = 0; - memset(&mesh->fdata, 0, sizeof(mesh->fdata)); + /* cache only - don't write */ + mesh->mface = NULL; + mesh->totface = 0; + memset(&mesh->fdata, 0, sizeof(mesh->fdata)); #endif /* USE_BMESH_SAVE_WITHOUT_MFACE */ - /** - * Those calls: - * - Reduce mesh->xdata.totlayer to number of layers to write. - * - Fill xlayers with those layers to be written. - * Note that mesh->xdata is from now on invalid for Blender, but this is why the whole mesh is - * a temp local copy! - */ - CustomData_file_write_prepare(&mesh->vdata, &vlayers, vlayers_buff, ARRAY_SIZE(vlayers_buff)); - CustomData_file_write_prepare(&mesh->edata, &elayers, elayers_buff, ARRAY_SIZE(elayers_buff)); + /** + * Those calls: + * - Reduce mesh->xdata.totlayer to number of layers to write. + * - Fill xlayers with those layers to be written. + * Note that mesh->xdata is from now on invalid for Blender, but this is why the whole mesh is + * a temp local copy! + */ + CustomData_file_write_prepare(&mesh->vdata, &vlayers, vlayers_buff, ARRAY_SIZE(vlayers_buff)); + CustomData_file_write_prepare(&mesh->edata, &elayers, elayers_buff, ARRAY_SIZE(elayers_buff)); #ifndef USE_BMESH_SAVE_WITHOUT_MFACE /* Do not copy org fdata in this case!!! */ - CustomData_file_write_prepare(&mesh->fdata, &flayers, flayers_buff, ARRAY_SIZE(flayers_buff)); + CustomData_file_write_prepare(&mesh->fdata, &flayers, flayers_buff, ARRAY_SIZE(flayers_buff)); #else - flayers = flayers_buff; + flayers = flayers_buff; #endif - CustomData_file_write_prepare(&mesh->ldata, &llayers, llayers_buff, ARRAY_SIZE(llayers_buff)); - CustomData_file_write_prepare(&mesh->pdata, &players, players_buff, ARRAY_SIZE(players_buff)); + CustomData_file_write_prepare(&mesh->ldata, &llayers, llayers_buff, ARRAY_SIZE(llayers_buff)); + CustomData_file_write_prepare(&mesh->pdata, &players, players_buff, ARRAY_SIZE(players_buff)); - writestruct_at_address(wd, ID_ME, Mesh, 1, old_mesh, mesh); - write_iddata(wd, &mesh->id); + writestruct_at_address(wd, ID_ME, Mesh, 1, old_mesh, mesh); + write_iddata(wd, &mesh->id); - /* direct data */ - if (mesh->adt) { - write_animdata(wd, mesh->adt); - } + /* direct data */ + if (mesh->adt) { + write_animdata(wd, mesh->adt); + } - writedata(wd, DATA, sizeof(void *) * mesh->totcol, mesh->mat); - writedata(wd, DATA, sizeof(MSelect) * mesh->totselect, mesh->mselect); + writedata(wd, DATA, sizeof(void *) * mesh->totcol, mesh->mat); + writedata(wd, DATA, sizeof(MSelect) * mesh->totselect, mesh->mselect); - write_customdata(wd, &mesh->id, mesh->totvert, &mesh->vdata, vlayers, -1, 0); - write_customdata(wd, &mesh->id, mesh->totedge, &mesh->edata, elayers, -1, 0); - /* fdata is really a dummy - written so slots align */ - write_customdata(wd, &mesh->id, mesh->totface, &mesh->fdata, flayers, -1, 0); - write_customdata(wd, &mesh->id, mesh->totloop, &mesh->ldata, llayers, -1, 0); - write_customdata(wd, &mesh->id, mesh->totpoly, &mesh->pdata, players, -1, 0); + write_customdata(wd, &mesh->id, mesh->totvert, &mesh->vdata, vlayers, -1, 0); + write_customdata(wd, &mesh->id, mesh->totedge, &mesh->edata, elayers, -1, 0); + /* fdata is really a dummy - written so slots align */ + write_customdata(wd, &mesh->id, mesh->totface, &mesh->fdata, flayers, -1, 0); + write_customdata(wd, &mesh->id, mesh->totloop, &mesh->ldata, llayers, -1, 0); + write_customdata(wd, &mesh->id, mesh->totpoly, &mesh->pdata, players, -1, 0); - /* restore pointer */ - mesh = old_mesh; - } - else { + /* restore pointer */ + mesh = old_mesh; + } + else { #ifdef USE_BMESH_SAVE_AS_COMPAT - /* write a copy of the mesh, don't modify in place because it is - * not thread safe for threaded renders that are reading this */ - Mesh *old_mesh = mesh; - Mesh copy_mesh = *mesh; - mesh = ©_mesh; - - mesh->mpoly = NULL; - mesh->mface = NULL; - mesh->totface = 0; - mesh->totpoly = 0; - mesh->totloop = 0; - CustomData_reset(&mesh->fdata); - CustomData_reset(&mesh->pdata); - CustomData_reset(&mesh->ldata); - mesh->edit_btmesh = NULL; - - /* now fill in polys to mfaces */ - /* XXX This breaks writing design, by using temp allocated memory, which will likely generate - * duplicates in stored 'old' addresses. - * This is very bad, but do not see easy way to avoid this, aside from generating those data - * outside of save process itself. - * Maybe we can live with this, though? - */ - mesh->totface = BKE_mesh_mpoly_to_mface(&mesh->fdata, &old_mesh->ldata, &old_mesh->pdata, - mesh->totface, old_mesh->totloop, old_mesh->totpoly); + /* write a copy of the mesh, don't modify in place because it is + * not thread safe for threaded renders that are reading this */ + Mesh *old_mesh = mesh; + Mesh copy_mesh = *mesh; + mesh = ©_mesh; + + mesh->mpoly = NULL; + mesh->mface = NULL; + mesh->totface = 0; + mesh->totpoly = 0; + mesh->totloop = 0; + CustomData_reset(&mesh->fdata); + CustomData_reset(&mesh->pdata); + CustomData_reset(&mesh->ldata); + mesh->edit_btmesh = NULL; + + /* now fill in polys to mfaces */ + /* XXX This breaks writing design, by using temp allocated memory, which will likely generate + * duplicates in stored 'old' addresses. + * This is very bad, but do not see easy way to avoid this, aside from generating those data + * outside of save process itself. + * Maybe we can live with this, though? + */ + mesh->totface = BKE_mesh_mpoly_to_mface(&mesh->fdata, &old_mesh->ldata, &old_mesh->pdata, + mesh->totface, old_mesh->totloop, old_mesh->totpoly); - BKE_mesh_update_customdata_pointers(mesh, false); + BKE_mesh_update_customdata_pointers(mesh, false); - CustomData_file_write_prepare(&mesh->vdata, &vlayers, vlayers_buff, ARRAY_SIZE(vlayers_buff)); - CustomData_file_write_prepare(&mesh->edata, &elayers, elayers_buff, ARRAY_SIZE(elayers_buff)); - CustomData_file_write_prepare(&mesh->fdata, &flayers, flayers_buff, ARRAY_SIZE(flayers_buff)); + CustomData_file_write_prepare(&mesh->vdata, &vlayers, vlayers_buff, ARRAY_SIZE(vlayers_buff)); + CustomData_file_write_prepare(&mesh->edata, &elayers, elayers_buff, ARRAY_SIZE(elayers_buff)); + CustomData_file_write_prepare(&mesh->fdata, &flayers, flayers_buff, ARRAY_SIZE(flayers_buff)); #if 0 - CustomData_file_write_prepare(&mesh->ldata, &llayers, llayers_buff, ARRAY_SIZE(llayers_buff)); - CustomData_file_write_prepare(&mesh->pdata, &players, players_buff, ARRAY_SIZE(players_buff)); + CustomData_file_write_prepare(&mesh->ldata, &llayers, llayers_buff, ARRAY_SIZE(llayers_buff)); + CustomData_file_write_prepare(&mesh->pdata, &players, players_buff, ARRAY_SIZE(players_buff)); #endif - writestruct_at_address(wd, ID_ME, Mesh, 1, old_mesh, mesh); - write_iddata(wd, &mesh->id); + writestruct_at_address(wd, ID_ME, Mesh, 1, old_mesh, mesh); + write_iddata(wd, &mesh->id); - /* direct data */ - if (mesh->adt) { - write_animdata(wd, mesh->adt); - } + /* direct data */ + if (mesh->adt) { + write_animdata(wd, mesh->adt); + } - writedata(wd, DATA, sizeof(void *) * mesh->totcol, mesh->mat); - /* writedata(wd, DATA, sizeof(MSelect) * mesh->totselect, mesh->mselect); */ /* pre-bmesh NULL's */ + writedata(wd, DATA, sizeof(void *) * mesh->totcol, mesh->mat); + /* writedata(wd, DATA, sizeof(MSelect) * mesh->totselect, mesh->mselect); */ /* pre-bmesh NULL's */ - write_customdata(wd, &mesh->id, mesh->totvert, &mesh->vdata, vlayers, -1, 0); - write_customdata(wd, &mesh->id, mesh->totedge, &mesh->edata, elayers, -1, 0); - write_customdata(wd, &mesh->id, mesh->totface, &mesh->fdata, flayers, -1, 0); - /* harmless for older blender versioins but _not_ writing these keeps file size down */ + write_customdata(wd, &mesh->id, mesh->totvert, &mesh->vdata, vlayers, -1, 0); + write_customdata(wd, &mesh->id, mesh->totedge, &mesh->edata, elayers, -1, 0); + write_customdata(wd, &mesh->id, mesh->totface, &mesh->fdata, flayers, -1, 0); + /* harmless for older blender versioins but _not_ writing these keeps file size down */ #if 0 - write_customdata(wd, &mesh->id, mesh->totloop, &mesh->ldata, llayers, -1, 0); - write_customdata(wd, &mesh->id, mesh->totpoly, &mesh->pdata, players, -1, 0); + write_customdata(wd, &mesh->id, mesh->totloop, &mesh->ldata, llayers, -1, 0); + write_customdata(wd, &mesh->id, mesh->totpoly, &mesh->pdata, players, -1, 0); #endif - CustomData_free(&mesh->fdata, mesh->totface); - flayers = NULL; + CustomData_free(&mesh->fdata, mesh->totface); + flayers = NULL; - /* restore pointer */ - mesh = old_mesh; + /* restore pointer */ + mesh = old_mesh; #endif /* USE_BMESH_SAVE_AS_COMPAT */ - } - } - - if (vlayers && vlayers != vlayers_buff) { - MEM_freeN(vlayers); } - if (elayers && elayers != elayers_buff) { - MEM_freeN(elayers); - } - if (flayers && flayers != flayers_buff) { - MEM_freeN(flayers); - } - if (llayers && llayers != llayers_buff) { - MEM_freeN(llayers); - } - if (players && players != players_buff) { - MEM_freeN(players); - } - - mesh = mesh->id.next; } - mywrite_flush(wd); + if (vlayers && vlayers != vlayers_buff) { + MEM_freeN(vlayers); + } + if (elayers && elayers != elayers_buff) { + MEM_freeN(elayers); + } + if (flayers && flayers != flayers_buff) { + MEM_freeN(flayers); + } + if (llayers && llayers != llayers_buff) { + MEM_freeN(llayers); + } + if (players && players != players_buff) { + MEM_freeN(players); + } } -static void write_lattices(WriteData *wd, ListBase *idbase) +static void write_lattice(WriteData *wd, Lattice *lt) { - Lattice *lt; - - lt = idbase->first; - while (lt) { - if (lt->id.us > 0 || wd->current) { - /* write LibData */ - writestruct(wd, ID_LT, Lattice, 1, lt); - write_iddata(wd, <->id); - - /* write animdata */ - if (lt->adt) { - write_animdata(wd, lt->adt); - } + if (lt->id.us > 0 || wd->current) { + /* write LibData */ + writestruct(wd, ID_LT, Lattice, 1, lt); + write_iddata(wd, <->id); - /* direct data */ - writestruct(wd, DATA, BPoint, lt->pntsu * lt->pntsv * lt->pntsw, lt->def); + /* write animdata */ + if (lt->adt) { + write_animdata(wd, lt->adt); + } - write_dverts(wd, lt->pntsu * lt->pntsv * lt->pntsw, lt->dvert); + /* direct data */ + writestruct(wd, DATA, BPoint, lt->pntsu * lt->pntsv * lt->pntsw, lt->def); - } - lt = lt->id.next; + write_dverts(wd, lt->pntsu * lt->pntsv * lt->pntsw, lt->dvert); } - - mywrite_flush(wd); } -static void write_images(WriteData *wd, ListBase *idbase) +static void write_image(WriteData *wd, Image *ima) { - Image *ima; - PackedFile *pf; - ImageView *iv; - ImagePackedFile *imapf; - - ima = idbase->first; - while (ima) { - if (ima->id.us > 0 || wd->current) { - /* Some trickery to keep forward compatibility of packed images. */ - BLI_assert(ima->packedfile == NULL); - if (ima->packedfiles.first != NULL) { - imapf = ima->packedfiles.first; - ima->packedfile = imapf->packedfile; - } - - /* write LibData */ - writestruct(wd, ID_IM, Image, 1, ima); - write_iddata(wd, &ima->id); + if (ima->id.us > 0 || wd->current) { + ImagePackedFile *imapf; - for (imapf = ima->packedfiles.first; imapf; imapf = imapf->next) { - writestruct(wd, DATA, ImagePackedFile, 1, imapf); - if (imapf->packedfile) { - pf = imapf->packedfile; - writestruct(wd, DATA, PackedFile, 1, pf); - writedata(wd, DATA, pf->size, pf->data); - } - } + /* Some trickery to keep forward compatibility of packed images. */ + BLI_assert(ima->packedfile == NULL); + if (ima->packedfiles.first != NULL) { + imapf = ima->packedfiles.first; + ima->packedfile = imapf->packedfile; + } - write_previews(wd, ima->preview); + /* write LibData */ + writestruct(wd, ID_IM, Image, 1, ima); + write_iddata(wd, &ima->id); - for (iv = ima->views.first; iv; iv = iv->next) { - writestruct(wd, DATA, ImageView, 1, iv); + for (imapf = ima->packedfiles.first; imapf; imapf = imapf->next) { + writestruct(wd, DATA, ImagePackedFile, 1, imapf); + if (imapf->packedfile) { + PackedFile *pf = imapf->packedfile; + writestruct(wd, DATA, PackedFile, 1, pf); + writedata(wd, DATA, pf->size, pf->data); } - writestruct(wd, DATA, Stereo3dFormat, 1, ima->stereo3d_format); + } - ima->packedfile = NULL; + write_previews(wd, ima->preview); + + for (ImageView *iv = ima->views.first; iv; iv = iv->next) { + writestruct(wd, DATA, ImageView, 1, iv); } - ima = ima->id.next; - } + writestruct(wd, DATA, Stereo3dFormat, 1, ima->stereo3d_format); - mywrite_flush(wd); + ima->packedfile = NULL; + } } -static void write_textures(WriteData *wd, ListBase *idbase) +static void write_texture(WriteData *wd, Tex *tex) { - Tex *tex; - - tex = idbase->first; - while (tex) { - if (tex->id.us > 0 || wd->current) { - /* write LibData */ - writestruct(wd, ID_TE, Tex, 1, tex); - write_iddata(wd, &tex->id); + if (tex->id.us > 0 || wd->current) { + /* write LibData */ + writestruct(wd, ID_TE, Tex, 1, tex); + write_iddata(wd, &tex->id); - if (tex->adt) { - write_animdata(wd, tex->adt); - } + if (tex->adt) { + write_animdata(wd, tex->adt); + } - /* direct data */ - if (tex->coba) { - writestruct(wd, DATA, ColorBand, 1, tex->coba); - } - if (tex->type == TEX_ENVMAP && tex->env) { - writestruct(wd, DATA, EnvMap, 1, tex->env); - } - if (tex->type == TEX_POINTDENSITY && tex->pd) { - writestruct(wd, DATA, PointDensity, 1, tex->pd); - if (tex->pd->coba) { - writestruct(wd, DATA, ColorBand, 1, tex->pd->coba); - } - if (tex->pd->falloff_curve) { - write_curvemapping(wd, tex->pd->falloff_curve); - } - } - if (tex->type == TEX_VOXELDATA) { - writestruct(wd, DATA, VoxelData, 1, tex->vd); - } - if (tex->type == TEX_OCEAN && tex->ot) { - writestruct(wd, DATA, OceanTex, 1, tex->ot); + /* direct data */ + if (tex->coba) { + writestruct(wd, DATA, ColorBand, 1, tex->coba); + } + if (tex->type == TEX_ENVMAP && tex->env) { + writestruct(wd, DATA, EnvMap, 1, tex->env); + } + if (tex->type == TEX_POINTDENSITY && tex->pd) { + writestruct(wd, DATA, PointDensity, 1, tex->pd); + if (tex->pd->coba) { + writestruct(wd, DATA, ColorBand, 1, tex->pd->coba); } - - /* nodetree is integral part of texture, no libdata */ - if (tex->nodetree) { - writestruct(wd, DATA, bNodeTree, 1, tex->nodetree); - write_nodetree(wd, tex->nodetree); + if (tex->pd->falloff_curve) { + write_curvemapping(wd, tex->pd->falloff_curve); } + } + if (tex->type == TEX_VOXELDATA) { + writestruct(wd, DATA, VoxelData, 1, tex->vd); + } + if (tex->type == TEX_OCEAN && tex->ot) { + writestruct(wd, DATA, OceanTex, 1, tex->ot); + } - write_previews(wd, tex->preview); + /* nodetree is integral part of texture, no libdata */ + if (tex->nodetree) { + writestruct(wd, DATA, bNodeTree, 1, tex->nodetree); + write_nodetree_nolib(wd, tex->nodetree); } - tex = tex->id.next; - } - mywrite_flush(wd); + write_previews(wd, tex->preview); + } } -static void write_materials(WriteData *wd, ListBase *idbase) +static void write_material(WriteData *wd, Material *ma) { - Material *ma; - int a; - - ma = idbase->first; - while (ma) { - if (ma->id.us > 0 || wd->current) { - /* write LibData */ - writestruct(wd, ID_MA, Material, 1, ma); - write_iddata(wd, &ma->id); - - if (ma->adt) { - write_animdata(wd, ma->adt); - } + if (ma->id.us > 0 || wd->current) { + /* write LibData */ + writestruct(wd, ID_MA, Material, 1, ma); + write_iddata(wd, &ma->id); - for (a = 0; a < MAX_MTEX; a++) { - if (ma->mtex[a]) { - writestruct(wd, DATA, MTex, 1, ma->mtex[a]); - } - } + if (ma->adt) { + write_animdata(wd, ma->adt); + } - if (ma->ramp_col) { - writestruct(wd, DATA, ColorBand, 1, ma->ramp_col); - } - if (ma->ramp_spec) { - writestruct(wd, DATA, ColorBand, 1, ma->ramp_spec); + for (int a = 0; a < MAX_MTEX; a++) { + if (ma->mtex[a]) { + writestruct(wd, DATA, MTex, 1, ma->mtex[a]); } + } - /* nodetree is integral part of material, no libdata */ - if (ma->nodetree) { - writestruct(wd, DATA, bNodeTree, 1, ma->nodetree); - write_nodetree(wd, ma->nodetree); - } + if (ma->ramp_col) { + writestruct(wd, DATA, ColorBand, 1, ma->ramp_col); + } + if (ma->ramp_spec) { + writestruct(wd, DATA, ColorBand, 1, ma->ramp_spec); + } - write_previews(wd, ma->preview); + /* nodetree is integral part of material, no libdata */ + if (ma->nodetree) { + writestruct(wd, DATA, bNodeTree, 1, ma->nodetree); + write_nodetree_nolib(wd, ma->nodetree); } - ma = ma->id.next; + + write_previews(wd, ma->preview); } } -static void write_worlds(WriteData *wd, ListBase *idbase) +static void write_world(WriteData *wd, World *wrld) { - World *wrld; - int a; - - wrld = idbase->first; - while (wrld) { - if (wrld->id.us > 0 || wd->current) { - /* write LibData */ - writestruct(wd, ID_WO, World, 1, wrld); - write_iddata(wd, &wrld->id); - - if (wrld->adt) { - write_animdata(wd, wrld->adt); - } + if (wrld->id.us > 0 || wd->current) { + /* write LibData */ + writestruct(wd, ID_WO, World, 1, wrld); + write_iddata(wd, &wrld->id); - for (a = 0; a < MAX_MTEX; a++) { - if (wrld->mtex[a]) { - writestruct(wd, DATA, MTex, 1, wrld->mtex[a]); - } - } + if (wrld->adt) { + write_animdata(wd, wrld->adt); + } - /* nodetree is integral part of world, no libdata */ - if (wrld->nodetree) { - writestruct(wd, DATA, bNodeTree, 1, wrld->nodetree); - write_nodetree(wd, wrld->nodetree); + for (int a = 0; a < MAX_MTEX; a++) { + if (wrld->mtex[a]) { + writestruct(wd, DATA, MTex, 1, wrld->mtex[a]); } + } - write_previews(wd, wrld->preview); + /* nodetree is integral part of world, no libdata */ + if (wrld->nodetree) { + writestruct(wd, DATA, bNodeTree, 1, wrld->nodetree); + write_nodetree_nolib(wd, wrld->nodetree); } - wrld = wrld->id.next; + + write_previews(wd, wrld->preview); } } -static void write_lamps(WriteData *wd, ListBase *idbase) +static void write_lamp(WriteData *wd, Lamp *la) { - Lamp *la; - int a; - - la = idbase->first; - while (la) { - if (la->id.us > 0 || wd->current) { - /* write LibData */ - writestruct(wd, ID_LA, Lamp, 1, la); - write_iddata(wd, &la->id); - - if (la->adt) { - write_animdata(wd, la->adt); - } - - /* direct data */ - for (a = 0; a < MAX_MTEX; a++) { - if (la->mtex[a]) { - writestruct(wd, DATA, MTex, 1, la->mtex[a]); - } - } + if (la->id.us > 0 || wd->current) { + /* write LibData */ + writestruct(wd, ID_LA, Lamp, 1, la); + write_iddata(wd, &la->id); - if (la->curfalloff) { - write_curvemapping(wd, la->curfalloff); - } + if (la->adt) { + write_animdata(wd, la->adt); + } - /* nodetree is integral part of lamps, no libdata */ - if (la->nodetree) { - writestruct(wd, DATA, bNodeTree, 1, la->nodetree); - write_nodetree(wd, la->nodetree); + /* direct data */ + for (int a = 0; a < MAX_MTEX; a++) { + if (la->mtex[a]) { + writestruct(wd, DATA, MTex, 1, la->mtex[a]); } + } - write_previews(wd, la->preview); + if (la->curfalloff) { + write_curvemapping(wd, la->curfalloff); + } + /* nodetree is integral part of lamps, no libdata */ + if (la->nodetree) { + writestruct(wd, DATA, bNodeTree, 1, la->nodetree); + write_nodetree_nolib(wd, la->nodetree); } - la = la->id.next; - } - mywrite_flush(wd); + write_previews(wd, la->preview); + } } static void write_sequence_modifiers(WriteData *wd, ListBase *modbase) @@ -2626,294 +2536,253 @@ static void write_paint(WriteData *wd, Paint *p) } } -static void write_scenes(WriteData *wd, ListBase *scebase) +static void write_scene(WriteData *wd, Scene *sce) { - Scene *sce; - Base *base; - Editing *ed; - Sequence *seq; - MetaStack *ms; - Strip *strip; - TimeMarker *marker; - TransformOrientation *ts; - SceneRenderLayer *srl; - SceneRenderView *srv; - ToolSettings *tos; - FreestyleModuleConfig *fmc; - FreestyleLineSet *fls; - - sce = scebase->first; - while (sce) { - /* write LibData */ - writestruct(wd, ID_SCE, Scene, 1, sce); - write_iddata(wd, &sce->id); + /* write LibData */ + writestruct(wd, ID_SCE, Scene, 1, sce); + write_iddata(wd, &sce->id); - if (sce->adt) { - write_animdata(wd, sce->adt); - } - write_keyingsets(wd, &sce->keyingsets); + if (sce->adt) { + write_animdata(wd, sce->adt); + } + write_keyingsets(wd, &sce->keyingsets); - /* direct data */ - base = sce->base.first; - while (base) { - writestruct(wd, DATA, Base, 1, base); - base = base->next; - } - - tos = sce->toolsettings; - writestruct(wd, DATA, ToolSettings, 1, tos); - if (tos->vpaint) { - writestruct(wd, DATA, VPaint, 1, tos->vpaint); - write_paint(wd, &tos->vpaint->paint); - } - if (tos->wpaint) { - writestruct(wd, DATA, VPaint, 1, tos->wpaint); - write_paint(wd, &tos->wpaint->paint); - } - if (tos->sculpt) { - writestruct(wd, DATA, Sculpt, 1, tos->sculpt); - write_paint(wd, &tos->sculpt->paint); - } - if (tos->uvsculpt) { - writestruct(wd, DATA, UvSculpt, 1, tos->uvsculpt); - write_paint(wd, &tos->uvsculpt->paint); - } - /* write grease-pencil drawing brushes to file */ - writelist(wd, DATA, bGPDbrush, &tos->gp_brushes); - for (bGPDbrush *brush = tos->gp_brushes.first; brush; brush = brush->next) { - if (brush->cur_sensitivity) { - write_curvemapping(wd, brush->cur_sensitivity); - } - if (brush->cur_strength) { - write_curvemapping(wd, brush->cur_strength); - } - if (brush->cur_jitter) { - write_curvemapping(wd, brush->cur_jitter); - } + /* direct data */ + for (Base *base = sce->base.first; base; base = base->next) { + writestruct(wd, DATA, Base, 1, base); + } + + ToolSettings *tos = sce->toolsettings; + writestruct(wd, DATA, ToolSettings, 1, tos); + if (tos->vpaint) { + writestruct(wd, DATA, VPaint, 1, tos->vpaint); + write_paint(wd, &tos->vpaint->paint); + } + if (tos->wpaint) { + writestruct(wd, DATA, VPaint, 1, tos->wpaint); + write_paint(wd, &tos->wpaint->paint); + } + if (tos->sculpt) { + writestruct(wd, DATA, Sculpt, 1, tos->sculpt); + write_paint(wd, &tos->sculpt->paint); + } + if (tos->uvsculpt) { + writestruct(wd, DATA, UvSculpt, 1, tos->uvsculpt); + write_paint(wd, &tos->uvsculpt->paint); + } + /* write grease-pencil drawing brushes to file */ + writelist(wd, DATA, bGPDbrush, &tos->gp_brushes); + for (bGPDbrush *brush = tos->gp_brushes.first; brush; brush = brush->next) { + if (brush->cur_sensitivity) { + write_curvemapping(wd, brush->cur_sensitivity); } - /* write grease-pencil custom ipo curve to file */ - if (tos->gp_interpolate.custom_ipo) { - write_curvemapping(wd, tos->gp_interpolate.custom_ipo); + if (brush->cur_strength) { + write_curvemapping(wd, brush->cur_strength); } - + if (brush->cur_jitter) { + write_curvemapping(wd, brush->cur_jitter); + } + } + /* write grease-pencil custom ipo curve to file */ + if (tos->gp_interpolate.custom_ipo) { + write_curvemapping(wd, tos->gp_interpolate.custom_ipo); + } - write_paint(wd, &tos->imapaint.paint); - ed = sce->ed; - if (ed) { - writestruct(wd, DATA, Editing, 1, ed); + write_paint(wd, &tos->imapaint.paint); - /* reset write flags too */ + Editing *ed = sce->ed; + if (ed) { + Sequence *seq; - SEQ_BEGIN(ed, seq) - { - if (seq->strip) { - seq->strip->done = false; - } - writestruct(wd, DATA, Sequence, 1, seq); - } - SEQ_END + writestruct(wd, DATA, Editing, 1, ed); - SEQ_BEGIN(ed, seq) - { - if (seq->strip && seq->strip->done == 0) { - /* write strip with 'done' at 0 because readfile */ - - if (seq->effectdata) { - switch (seq->type) { - case SEQ_TYPE_COLOR: - writestruct(wd, DATA, SolidColorVars, 1, seq->effectdata); - break; - case SEQ_TYPE_SPEED: - writestruct(wd, DATA, SpeedControlVars, 1, seq->effectdata); - break; - case SEQ_TYPE_WIPE: - writestruct(wd, DATA, WipeVars, 1, seq->effectdata); - break; - case SEQ_TYPE_GLOW: - writestruct(wd, DATA, GlowVars, 1, seq->effectdata); - break; - case SEQ_TYPE_TRANSFORM: - writestruct(wd, DATA, TransformVars, 1, seq->effectdata); - break; - case SEQ_TYPE_GAUSSIAN_BLUR: - writestruct(wd, DATA, GaussianBlurVars, 1, seq->effectdata); - break; - case SEQ_TYPE_TEXT: - writestruct(wd, DATA, TextVars, 1, seq->effectdata); - break; - } - } + /* reset write flags too */ - writestruct(wd, DATA, Stereo3dFormat, 1, seq->stereo3d_format); + SEQ_BEGIN(ed, seq) + { + if (seq->strip) { + seq->strip->done = false; + } + writestruct(wd, DATA, Sequence, 1, seq); + } + SEQ_END - strip = seq->strip; - writestruct(wd, DATA, Strip, 1, strip); - if (seq->flag & SEQ_USE_CROP && strip->crop) { - writestruct(wd, DATA, StripCrop, 1, strip->crop); - } - if (seq->flag & SEQ_USE_TRANSFORM && strip->transform) { - writestruct(wd, DATA, StripTransform, 1, strip->transform); - } - if (seq->flag & SEQ_USE_PROXY && strip->proxy) { - writestruct(wd, DATA, StripProxy, 1, strip->proxy); - } - if (seq->type == SEQ_TYPE_IMAGE) { - writestruct(wd, DATA, StripElem, - MEM_allocN_len(strip->stripdata) / sizeof(struct StripElem), - strip->stripdata); - } - else if (ELEM(seq->type, SEQ_TYPE_MOVIE, SEQ_TYPE_SOUND_RAM, SEQ_TYPE_SOUND_HD)) { - writestruct(wd, DATA, StripElem, 1, strip->stripdata); + SEQ_BEGIN(ed, seq) + { + if (seq->strip && seq->strip->done == 0) { + /* write strip with 'done' at 0 because readfile */ + + if (seq->effectdata) { + switch (seq->type) { + case SEQ_TYPE_COLOR: + writestruct(wd, DATA, SolidColorVars, 1, seq->effectdata); + break; + case SEQ_TYPE_SPEED: + writestruct(wd, DATA, SpeedControlVars, 1, seq->effectdata); + break; + case SEQ_TYPE_WIPE: + writestruct(wd, DATA, WipeVars, 1, seq->effectdata); + break; + case SEQ_TYPE_GLOW: + writestruct(wd, DATA, GlowVars, 1, seq->effectdata); + break; + case SEQ_TYPE_TRANSFORM: + writestruct(wd, DATA, TransformVars, 1, seq->effectdata); + break; + case SEQ_TYPE_GAUSSIAN_BLUR: + writestruct(wd, DATA, GaussianBlurVars, 1, seq->effectdata); + break; + case SEQ_TYPE_TEXT: + writestruct(wd, DATA, TextVars, 1, seq->effectdata); + break; } - - strip->done = true; } - if (seq->prop) { - IDP_WriteProperty(seq->prop, wd); + writestruct(wd, DATA, Stereo3dFormat, 1, seq->stereo3d_format); + + Strip *strip = seq->strip; + writestruct(wd, DATA, Strip, 1, strip); + if (seq->flag & SEQ_USE_CROP && strip->crop) { + writestruct(wd, DATA, StripCrop, 1, strip->crop); + } + if (seq->flag & SEQ_USE_TRANSFORM && strip->transform) { + writestruct(wd, DATA, StripTransform, 1, strip->transform); + } + if (seq->flag & SEQ_USE_PROXY && strip->proxy) { + writestruct(wd, DATA, StripProxy, 1, strip->proxy); + } + if (seq->type == SEQ_TYPE_IMAGE) { + writestruct(wd, DATA, StripElem, + MEM_allocN_len(strip->stripdata) / sizeof(struct StripElem), + strip->stripdata); + } + else if (ELEM(seq->type, SEQ_TYPE_MOVIE, SEQ_TYPE_SOUND_RAM, SEQ_TYPE_SOUND_HD)) { + writestruct(wd, DATA, StripElem, 1, strip->stripdata); } - write_sequence_modifiers(wd, &seq->modifiers); + strip->done = true; } - SEQ_END - /* new; meta stack too, even when its nasty restore code */ - for (ms = ed->metastack.first; ms; ms = ms->next) { - writestruct(wd, DATA, MetaStack, 1, ms); + if (seq->prop) { + IDP_WriteProperty(seq->prop, wd); } + + write_sequence_modifiers(wd, &seq->modifiers); } + SEQ_END - if (sce->r.avicodecdata) { - writestruct(wd, DATA, AviCodecData, 1, sce->r.avicodecdata); - if (sce->r.avicodecdata->lpFormat) { - writedata(wd, DATA, sce->r.avicodecdata->cbFormat, sce->r.avicodecdata->lpFormat); - } - if (sce->r.avicodecdata->lpParms) { - writedata(wd, DATA, sce->r.avicodecdata->cbParms, sce->r.avicodecdata->lpParms); - } + /* new; meta stack too, even when its nasty restore code */ + for (MetaStack *ms = ed->metastack.first; ms; ms = ms->next) { + writestruct(wd, DATA, MetaStack, 1, ms); } + } - if (sce->r.qtcodecdata) { - writestruct(wd, DATA, QuicktimeCodecData, 1, sce->r.qtcodecdata); - if (sce->r.qtcodecdata->cdParms) { - writedata(wd, DATA, sce->r.qtcodecdata->cdSize, sce->r.qtcodecdata->cdParms); - } + if (sce->r.avicodecdata) { + writestruct(wd, DATA, AviCodecData, 1, sce->r.avicodecdata); + if (sce->r.avicodecdata->lpFormat) { + writedata(wd, DATA, sce->r.avicodecdata->cbFormat, sce->r.avicodecdata->lpFormat); } - if (sce->r.ffcodecdata.properties) { - IDP_WriteProperty(sce->r.ffcodecdata.properties, wd); + if (sce->r.avicodecdata->lpParms) { + writedata(wd, DATA, sce->r.avicodecdata->cbParms, sce->r.avicodecdata->lpParms); } + } - /* writing dynamic list of TimeMarkers to the blend file */ - for (marker = sce->markers.first; marker; marker = marker->next) { - writestruct(wd, DATA, TimeMarker, 1, marker); + if (sce->r.qtcodecdata) { + writestruct(wd, DATA, QuicktimeCodecData, 1, sce->r.qtcodecdata); + if (sce->r.qtcodecdata->cdParms) { + writedata(wd, DATA, sce->r.qtcodecdata->cdSize, sce->r.qtcodecdata->cdParms); } + } + if (sce->r.ffcodecdata.properties) { + IDP_WriteProperty(sce->r.ffcodecdata.properties, wd); + } - /* writing dynamic list of TransformOrientations to the blend file */ - for (ts = sce->transform_spaces.first; ts; ts = ts->next) { - writestruct(wd, DATA, TransformOrientation, 1, ts); - } + /* writing dynamic list of TimeMarkers to the blend file */ + for (TimeMarker *marker = sce->markers.first; marker; marker = marker->next) { + writestruct(wd, DATA, TimeMarker, 1, marker); + } - for (srl = sce->r.layers.first; srl; srl = srl->next) { - writestruct(wd, DATA, SceneRenderLayer, 1, srl); - for (fmc = srl->freestyleConfig.modules.first; fmc; fmc = fmc->next) { - writestruct(wd, DATA, FreestyleModuleConfig, 1, fmc); - } - for (fls = srl->freestyleConfig.linesets.first; fls; fls = fls->next) { - writestruct(wd, DATA, FreestyleLineSet, 1, fls); - } - } + /* writing dynamic list of TransformOrientations to the blend file */ + for (TransformOrientation *ts = sce->transform_spaces.first; ts; ts = ts->next) { + writestruct(wd, DATA, TransformOrientation, 1, ts); + } - /* writing MultiView to the blend file */ - for (srv = sce->r.views.first; srv; srv = srv->next) { - writestruct(wd, DATA, SceneRenderView, 1, srv); + for (SceneRenderLayer *srl = sce->r.layers.first; srl; srl = srl->next) { + writestruct(wd, DATA, SceneRenderLayer, 1, srl); + for (FreestyleModuleConfig *fmc = srl->freestyleConfig.modules.first; fmc; fmc = fmc->next) { + writestruct(wd, DATA, FreestyleModuleConfig, 1, fmc); } - - if (sce->nodetree) { - writestruct(wd, DATA, bNodeTree, 1, sce->nodetree); - write_nodetree(wd, sce->nodetree); + for (FreestyleLineSet *fls = srl->freestyleConfig.linesets.first; fls; fls = fls->next) { + writestruct(wd, DATA, FreestyleLineSet, 1, fls); } + } - write_view_settings(wd, &sce->view_settings); + /* writing MultiView to the blend file */ + for (SceneRenderView *srv = sce->r.views.first; srv; srv = srv->next) { + writestruct(wd, DATA, SceneRenderView, 1, srv); + } - /* writing RigidBodyWorld data to the blend file */ - if (sce->rigidbody_world) { - writestruct(wd, DATA, RigidBodyWorld, 1, sce->rigidbody_world); - writestruct(wd, DATA, EffectorWeights, 1, sce->rigidbody_world->effector_weights); - write_pointcaches(wd, &(sce->rigidbody_world->ptcaches)); - } + if (sce->nodetree) { + writestruct(wd, DATA, bNodeTree, 1, sce->nodetree); + write_nodetree_nolib(wd, sce->nodetree); + } - write_previews(wd, sce->preview); - write_curvemapping_curves(wd, &sce->r.mblur_shutter_curve); + write_view_settings(wd, &sce->view_settings); - sce = sce->id.next; + /* writing RigidBodyWorld data to the blend file */ + if (sce->rigidbody_world) { + writestruct(wd, DATA, RigidBodyWorld, 1, sce->rigidbody_world); + writestruct(wd, DATA, EffectorWeights, 1, sce->rigidbody_world->effector_weights); + write_pointcaches(wd, &(sce->rigidbody_world->ptcaches)); } - mywrite_flush(wd); + write_previews(wd, sce->preview); + write_curvemapping_curves(wd, &sce->r.mblur_shutter_curve); } -static void write_gpencils(WriteData *wd, ListBase *lb) +static void write_gpencil(WriteData *wd, bGPdata *gpd) { - bGPdata *gpd; - bGPDlayer *gpl; - bGPDframe *gpf; - bGPDstroke *gps; - bGPDpalette *palette; - - for (gpd = lb->first; gpd; gpd = gpd->id.next) { - if (gpd->id.us > 0 || wd->current) { - /* write gpd data block to file */ - writestruct(wd, ID_GD, bGPdata, 1, gpd); - write_iddata(wd, &gpd->id); - - if (gpd->adt) { - write_animdata(wd, gpd->adt); - } - - /* write grease-pencil layers to file */ - writelist(wd, DATA, bGPDlayer, &gpd->layers); - for (gpl = gpd->layers.first; gpl; gpl = gpl->next) { - - /* write this layer's frames to file */ - writelist(wd, DATA, bGPDframe, &gpl->frames); - for (gpf = gpl->frames.first; gpf; gpf = gpf->next) { - - /* write strokes */ - writelist(wd, DATA, bGPDstroke, &gpf->strokes); - for (gps = gpf->strokes.first; gps; gps = gps->next) { - writestruct(wd, DATA, bGPDspoint, gps->totpoints, gps->points); - } + if (gpd->id.us > 0 || wd->current) { + /* write gpd data block to file */ + writestruct(wd, ID_GD, bGPdata, 1, gpd); + write_iddata(wd, &gpd->id); + + if (gpd->adt) { + write_animdata(wd, gpd->adt); + } + + /* write grease-pencil layers to file */ + writelist(wd, DATA, bGPDlayer, &gpd->layers); + for (bGPDlayer *gpl = gpd->layers.first; gpl; gpl = gpl->next) { + /* write this layer's frames to file */ + writelist(wd, DATA, bGPDframe, &gpl->frames); + for (bGPDframe *gpf = gpl->frames.first; gpf; gpf = gpf->next) { + /* write strokes */ + writelist(wd, DATA, bGPDstroke, &gpf->strokes); + for (bGPDstroke *gps = gpf->strokes.first; gps; gps = gps->next) { + writestruct(wd, DATA, bGPDspoint, gps->totpoints, gps->points); } } - /* write grease-pencil palettes */ - writelist(wd, DATA, bGPDpalette, &gpd->palettes); - for (palette = gpd->palettes.first; palette; palette = palette->next) { - writelist(wd, DATA, bGPDpalettecolor, &palette->colors); - } } - } - mywrite_flush(wd); + /* write grease-pencil palettes */ + writelist(wd, DATA, bGPDpalette, &gpd->palettes); + for (bGPDpalette *palette = gpd->palettes.first; palette; palette = palette->next) { + writelist(wd, DATA, bGPDpalettecolor, &palette->colors); + } + } } -static void write_windowmanagers(WriteData *wd, ListBase *lb) +static void write_windowmanager(WriteData *wd, wmWindowManager *wm) { - wmWindowManager *wm; - wmWindow *win; - - for (wm = lb->first; wm; wm = wm->id.next) { - writestruct(wd, ID_WM, wmWindowManager, 1, wm); - write_iddata(wd, &wm->id); + writestruct(wd, ID_WM, wmWindowManager, 1, wm); + write_iddata(wd, &wm->id); - for (win = wm->windows.first; win; win = win->next) { - writestruct(wd, DATA, wmWindow, 1, win); - writestruct(wd, DATA, Stereo3dFormat, 1, win->stereo3d_format); - } + for (wmWindow *win = wm->windows.first; win; win = win->next) { + writestruct(wd, DATA, wmWindow, 1, win); + writestruct(wd, DATA, Stereo3dFormat, 1, win->stereo3d_format); } - - /* typically flushing wouldn't be needed however this data _always_ changes, - * so flush here for more efficient undo. */ - mywrite_flush(wd); } static void write_region(WriteData *wd, ARegion *ar, int spacetype) @@ -2996,184 +2865,168 @@ static void write_soops(WriteData *wd, SpaceOops *so) } } -static void write_screens(WriteData *wd, ListBase *scrbase) +static void write_screen(WriteData *wd, bScreen *sc) { - bScreen *sc; - ScrArea *sa; - ScrVert *sv; - ScrEdge *se; - - sc = scrbase->first; - while (sc) { + /* write LibData */ + /* in 2.50+ files, the file identifier for screens is patched, forward compatibility */ + writestruct(wd, ID_SCRN, bScreen, 1, sc); + write_iddata(wd, &sc->id); - /* write LibData */ - /* in 2.50+ files, the file identifier for screens is patched, forward compatibility */ - writestruct(wd, ID_SCRN, bScreen, 1, sc); - write_iddata(wd, &sc->id); + /* direct data */ + for (ScrVert *sv = sc->vertbase.first; sv; sv = sv->next) { + writestruct(wd, DATA, ScrVert, 1, sv); + } - /* direct data */ - for (sv = sc->vertbase.first; sv; sv = sv->next) { - writestruct(wd, DATA, ScrVert, 1, sv); - } + for (ScrEdge *se = sc->edgebase.first; se; se = se->next) { + writestruct(wd, DATA, ScrEdge, 1, se); + } - for (se = sc->edgebase.first; se; se = se->next) { - writestruct(wd, DATA, ScrEdge, 1, se); - } + for (ScrArea *sa = sc->areabase.first; sa; sa = sa->next) { + SpaceLink *sl; + Panel *pa; + uiList *ui_list; + uiPreview *ui_preview; + PanelCategoryStack *pc_act; + ARegion *ar; - for (sa = sc->areabase.first; sa; sa = sa->next) { - SpaceLink *sl; - Panel *pa; - uiList *ui_list; - uiPreview *ui_preview; - PanelCategoryStack *pc_act; - ARegion *ar; + writestruct(wd, DATA, ScrArea, 1, sa); - writestruct(wd, DATA, ScrArea, 1, sa); + for (ar = sa->regionbase.first; ar; ar = ar->next) { + write_region(wd, ar, sa->spacetype); - for (ar = sa->regionbase.first; ar; ar = ar->next) { - write_region(wd, ar, sa->spacetype); + for (pa = ar->panels.first; pa; pa = pa->next) { + writestruct(wd, DATA, Panel, 1, pa); + } - for (pa = ar->panels.first; pa; pa = pa->next) { - writestruct(wd, DATA, Panel, 1, pa); - } + for (pc_act = ar->panels_category_active.first; pc_act; pc_act = pc_act->next) { + writestruct(wd, DATA, PanelCategoryStack, 1, pc_act); + } - for (pc_act = ar->panels_category_active.first; pc_act; pc_act = pc_act->next) { - writestruct(wd, DATA, PanelCategoryStack, 1, pc_act); - } + for (ui_list = ar->ui_lists.first; ui_list; ui_list = ui_list->next) { + write_uilist(wd, ui_list); + } - for (ui_list = ar->ui_lists.first; ui_list; ui_list = ui_list->next) { - write_uilist(wd, ui_list); - } + for (ui_preview = ar->ui_previews.first; ui_preview; ui_preview = ui_preview->next) { + writestruct(wd, DATA, uiPreview, 1, ui_preview); + } + } - for (ui_preview = ar->ui_previews.first; ui_preview; ui_preview = ui_preview->next) { - writestruct(wd, DATA, uiPreview, 1, ui_preview); - } + for (sl = sa->spacedata.first; sl; sl = sl->next) { + for (ar = sl->regionbase.first; ar; ar = ar->next) { + write_region(wd, ar, sl->spacetype); } - sl = sa->spacedata.first; - while (sl) { - for (ar = sl->regionbase.first; ar; ar = ar->next) { - write_region(wd, ar, sl->spacetype); + if (sl->spacetype == SPACE_VIEW3D) { + View3D *v3d = (View3D *)sl; + BGpic *bgpic; + writestruct(wd, DATA, View3D, 1, v3d); + for (bgpic = v3d->bgpicbase.first; bgpic; bgpic = bgpic->next) { + writestruct(wd, DATA, BGpic, 1, bgpic); } - - if (sl->spacetype == SPACE_VIEW3D) { - View3D *v3d = (View3D *)sl; - BGpic *bgpic; - writestruct(wd, DATA, View3D, 1, v3d); - for (bgpic = v3d->bgpicbase.first; bgpic; bgpic = bgpic->next) { - writestruct(wd, DATA, BGpic, 1, bgpic); - } - if (v3d->localvd) { - writestruct(wd, DATA, View3D, 1, v3d->localvd); - } - - if (v3d->fx_settings.ssao) { - writestruct(wd, DATA, GPUSSAOSettings, 1, v3d->fx_settings.ssao); - } - if (v3d->fx_settings.dof) { - writestruct(wd, DATA, GPUDOFSettings, 1, v3d->fx_settings.dof); - } + if (v3d->localvd) { + writestruct(wd, DATA, View3D, 1, v3d->localvd); } - else if (sl->spacetype == SPACE_IPO) { - SpaceIpo *sipo = (SpaceIpo *)sl; - ListBase tmpGhosts = sipo->ghostCurves; - - /* temporarily disable ghost curves when saving */ - sipo->ghostCurves.first = sipo->ghostCurves.last = NULL; - writestruct(wd, DATA, SpaceIpo, 1, sl); - if (sipo->ads) { - writestruct(wd, DATA, bDopeSheet, 1, sipo->ads); - } - - /* reenable ghost curves */ - sipo->ghostCurves = tmpGhosts; + if (v3d->fx_settings.ssao) { + writestruct(wd, DATA, GPUSSAOSettings, 1, v3d->fx_settings.ssao); } - else if (sl->spacetype == SPACE_BUTS) { - writestruct(wd, DATA, SpaceButs, 1, sl); + if (v3d->fx_settings.dof) { + writestruct(wd, DATA, GPUDOFSettings, 1, v3d->fx_settings.dof); } - else if (sl->spacetype == SPACE_FILE) { - SpaceFile *sfile = (SpaceFile *)sl; + } + else if (sl->spacetype == SPACE_IPO) { + SpaceIpo *sipo = (SpaceIpo *)sl; + ListBase tmpGhosts = sipo->ghostCurves; - writestruct(wd, DATA, SpaceFile, 1, sl); - if (sfile->params) { - writestruct(wd, DATA, FileSelectParams, 1, sfile->params); - } - } - else if (sl->spacetype == SPACE_SEQ) { - writestruct(wd, DATA, SpaceSeq, 1, sl); - } - else if (sl->spacetype == SPACE_OUTLINER) { - SpaceOops *so = (SpaceOops *)sl; - write_soops(wd, so); - } - else if (sl->spacetype == SPACE_IMAGE) { - writestruct(wd, DATA, SpaceImage, 1, sl); - } - else if (sl->spacetype == SPACE_TEXT) { - writestruct(wd, DATA, SpaceText, 1, sl); - } - else if (sl->spacetype == SPACE_SCRIPT) { - SpaceScript *scr = (SpaceScript *)sl; - scr->but_refs = NULL; - writestruct(wd, DATA, SpaceScript, 1, sl); - } - else if (sl->spacetype == SPACE_ACTION) { - writestruct(wd, DATA, SpaceAction, 1, sl); - } - else if (sl->spacetype == SPACE_NLA) { - SpaceNla *snla = (SpaceNla *)sl; + /* temporarily disable ghost curves when saving */ + sipo->ghostCurves.first = sipo->ghostCurves.last = NULL; - writestruct(wd, DATA, SpaceNla, 1, snla); - if (snla->ads) { - writestruct(wd, DATA, bDopeSheet, 1, snla->ads); - } - } - else if (sl->spacetype == SPACE_TIME) { - writestruct(wd, DATA, SpaceTime, 1, sl); + writestruct(wd, DATA, SpaceIpo, 1, sl); + if (sipo->ads) { + writestruct(wd, DATA, bDopeSheet, 1, sipo->ads); } - else if (sl->spacetype == SPACE_NODE) { - SpaceNode *snode = (SpaceNode *)sl; - bNodeTreePath *path; - writestruct(wd, DATA, SpaceNode, 1, snode); - for (path = snode->treepath.first; path; path = path->next) { - writestruct(wd, DATA, bNodeTreePath, 1, path); - } - } - else if (sl->spacetype == SPACE_LOGIC) { - writestruct(wd, DATA, SpaceLogic, 1, sl); - } - else if (sl->spacetype == SPACE_CONSOLE) { - SpaceConsole *con = (SpaceConsole *)sl; - ConsoleLine *cl; - - for (cl = con->history.first; cl; cl = cl->next) { - /* 'len_alloc' is invalid on write, set from 'len' on read */ - writestruct(wd, DATA, ConsoleLine, 1, cl); - writedata(wd, DATA, cl->len + 1, cl->line); - } - writestruct(wd, DATA, SpaceConsole, 1, sl); + /* reenable ghost curves */ + sipo->ghostCurves = tmpGhosts; + } + else if (sl->spacetype == SPACE_BUTS) { + writestruct(wd, DATA, SpaceButs, 1, sl); + } + else if (sl->spacetype == SPACE_FILE) { + SpaceFile *sfile = (SpaceFile *)sl; + writestruct(wd, DATA, SpaceFile, 1, sl); + if (sfile->params) { + writestruct(wd, DATA, FileSelectParams, 1, sfile->params); } - else if (sl->spacetype == SPACE_USERPREF) { - writestruct(wd, DATA, SpaceUserPref, 1, sl); + } + else if (sl->spacetype == SPACE_SEQ) { + writestruct(wd, DATA, SpaceSeq, 1, sl); + } + else if (sl->spacetype == SPACE_OUTLINER) { + SpaceOops *so = (SpaceOops *)sl; + write_soops(wd, so); + } + else if (sl->spacetype == SPACE_IMAGE) { + writestruct(wd, DATA, SpaceImage, 1, sl); + } + else if (sl->spacetype == SPACE_TEXT) { + writestruct(wd, DATA, SpaceText, 1, sl); + } + else if (sl->spacetype == SPACE_SCRIPT) { + SpaceScript *scr = (SpaceScript *)sl; + scr->but_refs = NULL; + writestruct(wd, DATA, SpaceScript, 1, sl); + } + else if (sl->spacetype == SPACE_ACTION) { + writestruct(wd, DATA, SpaceAction, 1, sl); + } + else if (sl->spacetype == SPACE_NLA) { + SpaceNla *snla = (SpaceNla *)sl; + + writestruct(wd, DATA, SpaceNla, 1, snla); + if (snla->ads) { + writestruct(wd, DATA, bDopeSheet, 1, snla->ads); } - else if (sl->spacetype == SPACE_CLIP) { - writestruct(wd, DATA, SpaceClip, 1, sl); + } + else if (sl->spacetype == SPACE_TIME) { + writestruct(wd, DATA, SpaceTime, 1, sl); + } + else if (sl->spacetype == SPACE_NODE) { + SpaceNode *snode = (SpaceNode *)sl; + bNodeTreePath *path; + writestruct(wd, DATA, SpaceNode, 1, snode); + + for (path = snode->treepath.first; path; path = path->next) { + writestruct(wd, DATA, bNodeTreePath, 1, path); } - else if (sl->spacetype == SPACE_INFO) { - writestruct(wd, DATA, SpaceInfo, 1, sl); + } + else if (sl->spacetype == SPACE_LOGIC) { + writestruct(wd, DATA, SpaceLogic, 1, sl); + } + else if (sl->spacetype == SPACE_CONSOLE) { + SpaceConsole *con = (SpaceConsole *)sl; + ConsoleLine *cl; + + for (cl = con->history.first; cl; cl = cl->next) { + /* 'len_alloc' is invalid on write, set from 'len' on read */ + writestruct(wd, DATA, ConsoleLine, 1, cl); + writedata(wd, DATA, cl->len + 1, cl->line); } + writestruct(wd, DATA, SpaceConsole, 1, sl); - sl = sl->next; + } + else if (sl->spacetype == SPACE_USERPREF) { + writestruct(wd, DATA, SpaceUserPref, 1, sl); + } + else if (sl->spacetype == SPACE_CLIP) { + writestruct(wd, DATA, SpaceClip, 1, sl); + } + else if (sl->spacetype == SPACE_INFO) { + writestruct(wd, DATA, SpaceInfo, 1, sl); } } - - sc = sc->id.next; } - - mywrite_flush(wd); } static void write_bone(WriteData *wd, Bone *bone) @@ -3196,156 +3049,101 @@ static void write_bone(WriteData *wd, Bone *bone) } } -static void write_armatures(WriteData *wd, ListBase *idbase) +static void write_armature(WriteData *wd, bArmature *arm) { - bArmature *arm; - Bone *bone; - - arm = idbase->first; - while (arm) { - if (arm->id.us > 0 || wd->current) { - writestruct(wd, ID_AR, bArmature, 1, arm); - write_iddata(wd, &arm->id); + if (arm->id.us > 0 || wd->current) { + writestruct(wd, ID_AR, bArmature, 1, arm); + write_iddata(wd, &arm->id); - if (arm->adt) { - write_animdata(wd, arm->adt); - } + if (arm->adt) { + write_animdata(wd, arm->adt); + } - /* Direct data */ - bone = arm->bonebase.first; - while (bone) { - write_bone(wd, bone); - bone = bone->next; - } + /* Direct data */ + for (Bone *bone = arm->bonebase.first; bone; bone = bone->next) { + write_bone(wd, bone); } - arm = arm->id.next; } - - mywrite_flush(wd); } -static void write_texts(WriteData *wd, ListBase *idbase) +static void write_text(WriteData *wd, Text *text) { - Text *text; - TextLine *tmp; + if ((text->flags & TXT_ISMEM) && (text->flags & TXT_ISEXT)) { + text->flags &= ~TXT_ISEXT; + } - text = idbase->first; - while (text) { - if ( (text->flags & TXT_ISMEM) && (text->flags & TXT_ISEXT)) { - text->flags &= ~TXT_ISEXT; - } + /* write LibData */ + writestruct(wd, ID_TXT, Text, 1, text); + write_iddata(wd, &text->id); - /* write LibData */ - writestruct(wd, ID_TXT, Text, 1, text); - write_iddata(wd, &text->id); + if (text->name) { + writedata(wd, DATA, strlen(text->name) + 1, text->name); + } - if (text->name) { - writedata(wd, DATA, strlen(text->name) + 1, text->name); + if (!(text->flags & TXT_ISEXT)) { + /* now write the text data, in two steps for optimization in the readfunction */ + for (TextLine *tmp = text->lines.first; tmp; tmp = tmp->next) { + writestruct(wd, DATA, TextLine, 1, tmp); } - if (!(text->flags & TXT_ISEXT)) { - /* now write the text data, in two steps for optimization in the readfunction */ - tmp = text->lines.first; - while (tmp) { - writestruct(wd, DATA, TextLine, 1, tmp); - tmp = tmp->next; - } - - tmp = text->lines.first; - while (tmp) { - writedata(wd, DATA, tmp->len + 1, tmp->line); - tmp = tmp->next; - } + for (TextLine *tmp = text->lines.first; tmp; tmp = tmp->next) { + writedata(wd, DATA, tmp->len + 1, tmp->line); } - - - text = text->id.next; } - - mywrite_flush(wd); } -static void write_speakers(WriteData *wd, ListBase *idbase) +static void write_speaker(WriteData *wd, Speaker *spk) { - Speaker *spk; - - spk = idbase->first; - while (spk) { - if (spk->id.us > 0 || wd->current) { - /* write LibData */ - writestruct(wd, ID_SPK, Speaker, 1, spk); - write_iddata(wd, &spk->id); + if (spk->id.us > 0 || wd->current) { + /* write LibData */ + writestruct(wd, ID_SPK, Speaker, 1, spk); + write_iddata(wd, &spk->id); - if (spk->adt) { - write_animdata(wd, spk->adt); - } + if (spk->adt) { + write_animdata(wd, spk->adt); } - spk = spk->id.next; } } -static void write_sounds(WriteData *wd, ListBase *idbase) +static void write_sound(WriteData *wd, bSound *sound) { - bSound *sound; - - PackedFile *pf; - - sound = idbase->first; - while (sound) { - if (sound->id.us > 0 || wd->current) { - /* write LibData */ - writestruct(wd, ID_SO, bSound, 1, sound); - write_iddata(wd, &sound->id); + if (sound->id.us > 0 || wd->current) { + /* write LibData */ + writestruct(wd, ID_SO, bSound, 1, sound); + write_iddata(wd, &sound->id); - if (sound->packedfile) { - pf = sound->packedfile; - writestruct(wd, DATA, PackedFile, 1, pf); - writedata(wd, DATA, pf->size, pf->data); - } + if (sound->packedfile) { + PackedFile *pf = sound->packedfile; + writestruct(wd, DATA, PackedFile, 1, pf); + writedata(wd, DATA, pf->size, pf->data); } - sound = sound->id.next; } - - mywrite_flush(wd); } -static void write_groups(WriteData *wd, ListBase *idbase) +static void write_group(WriteData *wd, Group *group) { - Group *group; - GroupObject *go; - - for (group = idbase->first; group; group = group->id.next) { - if (group->id.us > 0 || wd->current) { - /* write LibData */ - writestruct(wd, ID_GR, Group, 1, group); - write_iddata(wd, &group->id); + if (group->id.us > 0 || wd->current) { + /* write LibData */ + writestruct(wd, ID_GR, Group, 1, group); + write_iddata(wd, &group->id); - write_previews(wd, group->preview); + write_previews(wd, group->preview); - go = group->gobject.first; - while (go) { - writestruct(wd, DATA, GroupObject, 1, go); - go = go->next; - } + for (GroupObject *go = group->gobject.first; go; go = go->next) { + writestruct(wd, DATA, GroupObject, 1, go); } } - - mywrite_flush(wd); } -static void write_nodetrees(WriteData *wd, ListBase *idbase) +static void write_nodetree(WriteData *wd, bNodeTree *ntree) { - bNodeTree *ntree; - - for (ntree = idbase->first; ntree; ntree = ntree->id.next) { - if (ntree->id.us > 0 || wd->current) { - writestruct(wd, ID_NT, bNodeTree, 1, ntree); - /* Note that trees directly used by other IDs (materials etc.) are not 'real' ID, they cannot - * be linked, etc., so we write actual id data here only, for 'real' ID trees. */ - write_iddata(wd, &ntree->id); + if (ntree->id.us > 0 || wd->current) { + writestruct(wd, ID_NT, bNodeTree, 1, ntree); + /* Note that trees directly used by other IDs (materials etc.) are not 'real' ID, they cannot + * be linked, etc., so we write actual id data here only, for 'real' ID trees. */ + write_iddata(wd, &ntree->id); - write_nodetree(wd, ntree); - } + write_nodetree_nolib(wd, ntree); } } @@ -3419,53 +3217,41 @@ static void customnodes_free_deprecated_data(Main *mainvar) } #endif -static void write_brushes(WriteData *wd, ListBase *idbase) +static void write_brush(WriteData *wd, Brush *brush) { - Brush *brush; - - for (brush = idbase->first; brush; brush = brush->id.next) { - if (brush->id.us > 0 || wd->current) { - writestruct(wd, ID_BR, Brush, 1, brush); - write_iddata(wd, &brush->id); + if (brush->id.us > 0 || wd->current) { + writestruct(wd, ID_BR, Brush, 1, brush); + write_iddata(wd, &brush->id); - if (brush->curve) { - write_curvemapping(wd, brush->curve); - } - if (brush->gradient) { - writestruct(wd, DATA, ColorBand, 1, brush->gradient); - } + if (brush->curve) { + write_curvemapping(wd, brush->curve); + } + if (brush->gradient) { + writestruct(wd, DATA, ColorBand, 1, brush->gradient); } } } -static void write_palettes(WriteData *wd, ListBase *idbase) +static void write_palette(WriteData *wd, Palette *palette) { - Palette *palette; - - for (palette = idbase->first; palette; palette = palette->id.next) { - if (palette->id.us > 0 || wd->current) { - PaletteColor *color; - writestruct(wd, ID_PAL, Palette, 1, palette); - write_iddata(wd, &palette->id); + if (palette->id.us > 0 || wd->current) { + PaletteColor *color; + writestruct(wd, ID_PAL, Palette, 1, palette); + write_iddata(wd, &palette->id); - for (color = palette->colors.first; color; color = color->next) { - writestruct(wd, DATA, PaletteColor, 1, color); - } + for (color = palette->colors.first; color; color = color->next) { + writestruct(wd, DATA, PaletteColor, 1, color); } } } -static void write_paintcurves(WriteData *wd, ListBase *idbase) +static void write_paintcurve(WriteData *wd, PaintCurve *pc) { - PaintCurve *pc; - - for (pc = idbase->first; pc; pc = pc->id.next) { - if (pc->id.us > 0 || wd->current) { - writestruct(wd, ID_PC, PaintCurve, 1, pc); - write_iddata(wd, &pc->id); + if (pc->id.us > 0 || wd->current) { + writestruct(wd, ID_PC, PaintCurve, 1, pc); + write_iddata(wd, &pc->id); - writestruct(wd, DATA, PaintCurvePoint, pc->tot_points, pc->points); - } + writestruct(wd, DATA, PaintCurvePoint, pc->tot_points, pc->points); } } @@ -3507,103 +3293,85 @@ static void write_movieReconstruction(WriteData *wd, MovieTrackingReconstruction } } -static void write_movieclips(WriteData *wd, ListBase *idbase) +static void write_movieclip(WriteData *wd, MovieClip *clip) { - MovieClip *clip; - - clip = idbase->first; - while (clip) { - if (clip->id.us > 0 || wd->current) { - MovieTracking *tracking = &clip->tracking; - MovieTrackingObject *object; + if (clip->id.us > 0 || wd->current) { + MovieTracking *tracking = &clip->tracking; + MovieTrackingObject *object; - writestruct(wd, ID_MC, MovieClip, 1, clip); - write_iddata(wd, &clip->id); + writestruct(wd, ID_MC, MovieClip, 1, clip); + write_iddata(wd, &clip->id); - if (clip->adt) { - write_animdata(wd, clip->adt); - } + if (clip->adt) { + write_animdata(wd, clip->adt); + } - write_movieTracks(wd, &tracking->tracks); - write_moviePlaneTracks(wd, &tracking->plane_tracks); - write_movieReconstruction(wd, &tracking->reconstruction); + write_movieTracks(wd, &tracking->tracks); + write_moviePlaneTracks(wd, &tracking->plane_tracks); + write_movieReconstruction(wd, &tracking->reconstruction); - object = tracking->objects.first; - while (object) { - writestruct(wd, DATA, MovieTrackingObject, 1, object); + object = tracking->objects.first; + while (object) { + writestruct(wd, DATA, MovieTrackingObject, 1, object); - write_movieTracks(wd, &object->tracks); - write_moviePlaneTracks(wd, &object->plane_tracks); - write_movieReconstruction(wd, &object->reconstruction); + write_movieTracks(wd, &object->tracks); + write_moviePlaneTracks(wd, &object->plane_tracks); + write_movieReconstruction(wd, &object->reconstruction); - object = object->next; - } + object = object->next; } - - clip = clip->id.next; } - - mywrite_flush(wd); } -static void write_masks(WriteData *wd, ListBase *idbase) +static void write_mask(WriteData *wd, Mask *mask) { - Mask *mask; - - mask = idbase->first; - while (mask) { - if (mask->id.us > 0 || wd->current) { - MaskLayer *masklay; + if (mask->id.us > 0 || wd->current) { + MaskLayer *masklay; - writestruct(wd, ID_MSK, Mask, 1, mask); - write_iddata(wd, &mask->id); + writestruct(wd, ID_MSK, Mask, 1, mask); + write_iddata(wd, &mask->id); - if (mask->adt) { - write_animdata(wd, mask->adt); - } + if (mask->adt) { + write_animdata(wd, mask->adt); + } - for (masklay = mask->masklayers.first; masklay; masklay = masklay->next) { - MaskSpline *spline; - MaskLayerShape *masklay_shape; + for (masklay = mask->masklayers.first; masklay; masklay = masklay->next) { + MaskSpline *spline; + MaskLayerShape *masklay_shape; - writestruct(wd, DATA, MaskLayer, 1, masklay); + writestruct(wd, DATA, MaskLayer, 1, masklay); - for (spline = masklay->splines.first; spline; spline = spline->next) { - int i; + for (spline = masklay->splines.first; spline; spline = spline->next) { + int i; - void *points_deform = spline->points_deform; - spline->points_deform = NULL; + void *points_deform = spline->points_deform; + spline->points_deform = NULL; - writestruct(wd, DATA, MaskSpline, 1, spline); - writestruct(wd, DATA, MaskSplinePoint, spline->tot_point, spline->points); + writestruct(wd, DATA, MaskSpline, 1, spline); + writestruct(wd, DATA, MaskSplinePoint, spline->tot_point, spline->points); - spline->points_deform = points_deform; + spline->points_deform = points_deform; - for (i = 0; i < spline->tot_point; i++) { - MaskSplinePoint *point = &spline->points[i]; + for (i = 0; i < spline->tot_point; i++) { + MaskSplinePoint *point = &spline->points[i]; - if (point->tot_uw) { - writestruct(wd, DATA, MaskSplinePointUW, point->tot_uw, point->uw); - } + if (point->tot_uw) { + writestruct(wd, DATA, MaskSplinePointUW, point->tot_uw, point->uw); } } + } - for (masklay_shape = masklay->splines_shapes.first; - masklay_shape; - masklay_shape = masklay_shape->next) - { - writestruct(wd, DATA, MaskLayerShape, 1, masklay_shape); - writedata(wd, DATA, - masklay_shape->tot_vert * sizeof(float) * MASK_OBJECT_SHAPE_ELEM_SIZE, - masklay_shape->data); - } + for (masklay_shape = masklay->splines_shapes.first; + masklay_shape; + masklay_shape = masklay_shape->next) + { + writestruct(wd, DATA, MaskLayerShape, 1, masklay_shape); + writedata(wd, DATA, + masklay_shape->tot_vert * sizeof(float) * MASK_OBJECT_SHAPE_ELEM_SIZE, + masklay_shape->data); } } - - mask = mask->id.next; } - - mywrite_flush(wd); } static void write_linestyle_color_modifiers(WriteData *wd, ListBase *modifiers) @@ -3860,48 +3628,39 @@ static void write_linestyle_geometry_modifiers(WriteData *wd, ListBase *modifier } } -static void write_linestyles(WriteData *wd, ListBase *idbase) +static void write_linestyle(WriteData *wd, FreestyleLineStyle *linestyle) { - FreestyleLineStyle *linestyle; - int a; + if (linestyle->id.us > 0 || wd->current) { + writestruct(wd, ID_LS, FreestyleLineStyle, 1, linestyle); + write_iddata(wd, &linestyle->id); - for (linestyle = idbase->first; linestyle; linestyle = linestyle->id.next) { - if (linestyle->id.us > 0 || wd->current) { - writestruct(wd, ID_LS, FreestyleLineStyle, 1, linestyle); - write_iddata(wd, &linestyle->id); - - if (linestyle->adt) { - write_animdata(wd, linestyle->adt); - } + if (linestyle->adt) { + write_animdata(wd, linestyle->adt); + } - write_linestyle_color_modifiers(wd, &linestyle->color_modifiers); - write_linestyle_alpha_modifiers(wd, &linestyle->alpha_modifiers); - write_linestyle_thickness_modifiers(wd, &linestyle->thickness_modifiers); - write_linestyle_geometry_modifiers(wd, &linestyle->geometry_modifiers); - for (a = 0; a < MAX_MTEX; a++) { - if (linestyle->mtex[a]) { - writestruct(wd, DATA, MTex, 1, linestyle->mtex[a]); - } - } - if (linestyle->nodetree) { - writestruct(wd, DATA, bNodeTree, 1, linestyle->nodetree); - write_nodetree(wd, linestyle->nodetree); + write_linestyle_color_modifiers(wd, &linestyle->color_modifiers); + write_linestyle_alpha_modifiers(wd, &linestyle->alpha_modifiers); + write_linestyle_thickness_modifiers(wd, &linestyle->thickness_modifiers); + write_linestyle_geometry_modifiers(wd, &linestyle->geometry_modifiers); + for (int a = 0; a < MAX_MTEX; a++) { + if (linestyle->mtex[a]) { + writestruct(wd, DATA, MTex, 1, linestyle->mtex[a]); } } + if (linestyle->nodetree) { + writestruct(wd, DATA, bNodeTree, 1, linestyle->nodetree); + write_nodetree_nolib(wd, linestyle->nodetree); + } } } -static void write_cachefiles(WriteData *wd, ListBase *idbase) +static void write_cachefile(WriteData *wd, CacheFile *cache_file) { - CacheFile *cache_file; + if (cache_file->id.us > 0 || wd->current) { + writestruct(wd, ID_CF, CacheFile, 1, cache_file); - for (cache_file = idbase->first; cache_file; cache_file = cache_file->id.next) { - if (cache_file->id.us > 0 || wd->current) { - writestruct(wd, ID_CF, CacheFile, 1, cache_file); - - if (cache_file->adt) { - write_animdata(wd, cache_file->adt); - } + if (cache_file->adt) { + write_animdata(wd, cache_file->adt); } } } @@ -3915,7 +3674,6 @@ static void write_libraries(WriteData *wd, Main *main) bool found_one; for (; main; main = main->next) { - a = tot = set_listbasepointers(main, lbarray); /* test: is lib being used */ @@ -3924,16 +3682,13 @@ static void write_libraries(WriteData *wd, Main *main) } else { found_one = false; - while (tot--) { + while (!found_one && tot--) { for (id = lbarray[tot]->first; id; id = id->next) { if (id->us > 0 && (id->tag & LIB_TAG_EXTERN)) { found_one = true; break; } } - if (found_one) { - break; - } } } @@ -4072,38 +3827,131 @@ static bool write_file_handle( * avoid thumbnail detecting changes because of this. */ mywrite_flush(wd); - write_windowmanagers(wd, &mainvar->wm); - write_screens(wd, &mainvar->screen); - write_movieclips(wd, &mainvar->movieclip); - write_masks(wd, &mainvar->mask); - write_scenes(wd, &mainvar->scene); - write_curves(wd, &mainvar->curve); - write_mballs(wd, &mainvar->mball); - write_images(wd, &mainvar->image); - write_cameras(wd, &mainvar->camera); - write_lamps(wd, &mainvar->lamp); - write_lattices(wd, &mainvar->latt); - write_vfonts(wd, &mainvar->vfont); - write_keys(wd, &mainvar->key); - write_worlds(wd, &mainvar->world); - write_texts(wd, &mainvar->text); - write_speakers(wd, &mainvar->speaker); - write_sounds(wd, &mainvar->sound); - write_groups(wd, &mainvar->group); - write_armatures(wd, &mainvar->armature); - write_actions(wd, &mainvar->action); - write_objects(wd, &mainvar->object); - write_materials(wd, &mainvar->mat); - write_textures(wd, &mainvar->tex); - write_meshes(wd, &mainvar->mesh); - write_particlesettings(wd, &mainvar->particle); - write_nodetrees(wd, &mainvar->nodetree); - write_brushes(wd, &mainvar->brush); - write_palettes(wd, &mainvar->palettes); - write_paintcurves(wd, &mainvar->paintcurves); - write_gpencils(wd, &mainvar->gpencil); - write_linestyles(wd, &mainvar->linestyle); - write_cachefiles(wd, &mainvar->cachefiles); + ListBase *lbarray[MAX_LIBARRAY]; + int a = set_listbasepointers(mainvar, lbarray); + while (a--) { + ID *id = lbarray[a]->first; + + if (id && GS(id->name) == ID_LI) { + continue; /* Libraries are handled separately below. */ + } + + for (; id; id = id->next) { + switch ((ID_Type)GS(id->name)) { + case ID_WM: + write_windowmanager(wd, (wmWindowManager *)id); + break; + case ID_SCR: + write_screen(wd, (bScreen *)id); + break; + case ID_MC: + write_movieclip(wd, (MovieClip *)id); + break; + case ID_MSK: + write_mask(wd, (Mask *)id); + break; + case ID_SCE: + write_scene(wd, (Scene *)id); + break; + case ID_CU: + write_curve(wd,(Curve *)id); + break; + case ID_MB: + write_mball(wd, (MetaBall *)id); + break; + case ID_IM: + write_image(wd, (Image *)id); + break; + case ID_CA: + write_camera(wd, (Camera *)id); + break; + case ID_LA: + write_lamp(wd, (Lamp *)id); + break; + case ID_LT: + write_lattice(wd, (Lattice *)id); + break; + case ID_VF: + write_vfont(wd, (VFont *)id); + break; + case ID_KE: + write_key(wd, (Key *)id); + break; + case ID_WO: + write_world(wd, (World *)id); + break; + case ID_TXT: + write_text(wd, (Text *)id); + break; + case ID_SPK: + write_speaker(wd, (Speaker *)id); + break; + case ID_SO: + write_sound(wd, (bSound *)id); + break; + case ID_GR: + write_group(wd, (Group *)id); + break; + case ID_AR: + write_armature(wd, (bArmature *)id); + break; + case ID_AC: + write_action(wd, (bAction *)id); + break; + case ID_OB: + write_object(wd, (Object *)id); + break; + case ID_MA: + write_material(wd, (Material *)id); + break; + case ID_TE: + write_texture(wd, (Tex *)id); + break; + case ID_ME: + write_mesh(wd, (Mesh *)id); + break; + case ID_PA: + write_particlesettings(wd, (ParticleSettings *)id); + break; + case ID_NT: + write_nodetree(wd, (bNodeTree *)id); + break; + case ID_BR: + write_brush(wd, (Brush *)id); + break; + case ID_PAL: + write_palette(wd, (Palette *)id); + break; + case ID_PC: + write_paintcurve(wd, (PaintCurve *)id); + break; + case ID_GD: + write_gpencil(wd, (bGPdata *)id); + break; + case ID_LS: + write_linestyle(wd, (FreestyleLineStyle *)id); + break; + case ID_CF: + write_cachefile(wd, (CacheFile *)id); + break; + case ID_LI: + /* Do nothing, handled below - and should never be reached. */ + BLI_assert(0); + break; + case ID_IP: + /* Do nothing, deprecated. */ + break; + default: + /* Should never be reached. */ + BLI_assert(0); + break; + } + } + + mywrite_flush(wd); + } + + /* Special handling, operating over split Mains... */ write_libraries(wd, mainvar->next); /* So changes above don't cause a 'DNA1' to be detected as changed on undo. */ diff --git a/source/blender/bmesh/CMakeLists.txt b/source/blender/bmesh/CMakeLists.txt index 30fefe37f0e..ea24da86626 100644 --- a/source/blender/bmesh/CMakeLists.txt +++ b/source/blender/bmesh/CMakeLists.txt @@ -152,6 +152,8 @@ set(SRC tools/bmesh_path_region.h tools/bmesh_region_match.c tools/bmesh_region_match.h + tools/bmesh_separate.c + tools/bmesh_separate.h tools/bmesh_triangulate.c tools/bmesh_triangulate.h tools/bmesh_wireframe.c diff --git a/source/blender/bmesh/bmesh.h b/source/blender/bmesh/bmesh.h index f29d280d071..b84a3d5e559 100644 --- a/source/blender/bmesh/bmesh.h +++ b/source/blender/bmesh/bmesh.h @@ -192,9 +192,10 @@ * * These conventions should be used throughout the bmesh module. * - * - ``BM_***()`` - High level BMesh API function for use anywhere. - * - ``bmesh_***()`` - Low level API function. + * - ``bmesh_kernel_*()`` - Low level API, for primitive functions that others are built ontop of. + * - ``bmesh_***()`` - Low level API function. * - ``bm_***()`` - 'static' functions, not apart of the API at all, but use prefix since they operate on BMesh data. + * - ``BM_***()`` - High level BMesh API function for use anywhere. * - ``BMO_***()`` - High level operator API function for use anywhere. * - ``bmo_***()`` - Low level / internal operator API functions. * - ``_bm_***()`` - Functions which are called via macros only. diff --git a/source/blender/bmesh/bmesh_tools.h b/source/blender/bmesh/bmesh_tools.h index 23212dd085e..a537c3b872c 100644 --- a/source/blender/bmesh/bmesh_tools.h +++ b/source/blender/bmesh/bmesh_tools.h @@ -43,6 +43,7 @@ extern "C" { #include "tools/bmesh_path.h" #include "tools/bmesh_path_region.h" #include "tools/bmesh_region_match.h" +#include "tools/bmesh_separate.h" #include "tools/bmesh_triangulate.h" #ifdef __cplusplus diff --git a/source/blender/bmesh/intern/bmesh_core.c b/source/blender/bmesh/intern/bmesh_core.c index d1178a198dc..cee5450a37d 100644 --- a/source/blender/bmesh/intern/bmesh_core.c +++ b/source/blender/bmesh/intern/bmesh_core.c @@ -1021,7 +1021,7 @@ static int UNUSED_FUNCTION(bm_loop_length)(BMLoop *l) * \param use_loop_mdisp_flip: When set, flip the Z-depth of the mdisp, * (use when flipping normals, disable when mirroring, eg: symmetrize). */ -void bmesh_loop_reverse( +void bmesh_kernel_loop_reverse( BMesh *bm, BMFace *f, const int cd_loop_mdisp_offset, const bool use_loop_mdisp_flip) { @@ -1438,7 +1438,7 @@ static BMFace *bm_face_create__sfme(BMesh *bm, BMFace *f_example) * * \return A BMFace pointer */ -BMFace *bmesh_sfme( +BMFace *bmesh_kernel_split_face_make_edge( BMesh *bm, BMFace *f, BMLoop *l_v1, BMLoop *l_v2, BMLoop **r_l, #ifdef USE_BMESH_HOLES @@ -1584,7 +1584,7 @@ BMFace *bmesh_sfme( * * \return The newly created BMVert pointer. */ -BMVert *bmesh_semv(BMesh *bm, BMVert *tv, BMEdge *e, BMEdge **r_e) +BMVert *bmesh_kernel_split_edge_make_vert(BMesh *bm, BMVert *tv, BMEdge *e, BMEdge **r_e) { BMLoop *l_next; BMEdge *e_new; @@ -1766,7 +1766,7 @@ BMVert *bmesh_semv(BMesh *bm, BMVert *tv, BMEdge *e, BMEdge **r_e) * faces with just 2 edges. It is up to the caller to decide what to do with * these faces. */ -BMEdge *bmesh_jekv( +BMEdge *bmesh_kernel_join_edge_kill_vert( BMesh *bm, BMEdge *e_kill, BMVert *v_kill, const bool do_del, const bool check_edge_double, const bool kill_degenerate_faces) @@ -1920,7 +1920,7 @@ BMEdge *bmesh_jekv( * * Collapse an edge, merging surrounding data. * - * Unlike #BM_vert_collapse_edge & #bmesh_jekv which only handle 2 valence verts, + * Unlike #BM_vert_collapse_edge & #bmesh_kernel_join_edge_kill_vert which only handle 2 valence verts, * this can handle any number of connected edges/faces. * * <pre> @@ -1932,7 +1932,7 @@ BMEdge *bmesh_jekv( * +-+-+-+ +-+-+-+ * </pre> */ -BMVert *bmesh_jvke( +BMVert *bmesh_kernel_join_vert_kill_edge( BMesh *bm, BMEdge *e_kill, BMVert *v_kill, const bool do_del, const bool check_edge_double, const bool kill_degenerate_faces) @@ -2035,7 +2035,7 @@ BMVert *bmesh_jvke( * In the example A, faces \a f1 and \a f2 are joined by a single edge, * and the euler can safely be used. * In example B however, \a f1 and \a f2 are joined by multiple edges and will produce an error. - * The caller in this case should call #bmesh_jekv on the extra edges + * The caller in this case should call #bmesh_kernel_join_edge_kill_vert on the extra edges * before attempting to fuse \a f1 and \a f2. * * \note The order of arguments decides whether or not certain per-face attributes are present @@ -2044,7 +2044,7 @@ BMVert *bmesh_jvke( * * \return A BMFace pointer */ -BMFace *bmesh_jfke(BMesh *bm, BMFace *f1, BMFace *f2, BMEdge *e) +BMFace *bmesh_kernel_join_face_kill_edge(BMesh *bm, BMFace *f1, BMFace *f2, BMEdge *e) { BMLoop *l_iter, *l_f1 = NULL, *l_f2 = NULL; int newlen = 0, i, f1len = 0, f2len = 0; @@ -2249,7 +2249,7 @@ bool BM_vert_splice(BMesh *bm, BMVert *v_dst, BMVert *v_src) } -/** \name BM_vert_separate, bmesh_vert_separate and friends +/** \name BM_vert_separate, bmesh_kernel_vert_separate and friends * \{ */ /* BM_edge_face_count(e) >= 1 */ @@ -2269,7 +2269,7 @@ BLI_INLINE bool bm_edge_supports_separate(const BMEdge *e) * * \return Success */ -void bmesh_vert_separate( +void bmesh_kernel_vert_separate( BMesh *bm, BMVert *v, BMVert ***r_vout, int *r_vout_len, const bool copy_select) { @@ -2385,7 +2385,7 @@ void bmesh_vert_separate( * * Takes a list of edges, which have been split from their original. * - * Any edges which failed to split off in #bmesh_vert_separate will be merged back into the original edge. + * Any edges which failed to split off in #bmesh_kernel_vert_separate will be merged back into the original edge. * * \param edges_separate * A list-of-lists, each list is from a single original edge (the first edge is the original), @@ -2398,7 +2398,7 @@ void bmesh_vert_separate( * \note this function looks like it could become slow, * but in common cases its only going to iterate a few times. */ -static void bmesh_vert_separate__cleanup(BMesh *bm, LinkNode *edges_separate) +static void bmesh_kernel_vert_separate__cleanup(BMesh *bm, LinkNode *edges_separate) { do { LinkNode *n_orig = edges_separate->link; @@ -2418,7 +2418,7 @@ static void bmesh_vert_separate__cleanup(BMesh *bm, LinkNode *edges_separate) } /** - * High level function which wraps both #bmesh_vert_separate and #bmesh_edge_separate + * High level function which wraps both #bmesh_kernel_vert_separate and #bmesh_kernel_edge_separate */ void BM_vert_separate( BMesh *bm, BMVert *v, @@ -2435,7 +2435,7 @@ void BM_vert_separate( LinkNode *edges_orig = NULL; do { BMLoop *l_sep = e->l; - bmesh_edge_separate(bm, e, l_sep, copy_select); + bmesh_kernel_edge_separate(bm, e, l_sep, copy_select); BLI_linklist_prepend_alloca(&edges_orig, l_sep->e); BLI_assert(e != l_sep->e); } while (bm_edge_supports_separate(e)); @@ -2444,10 +2444,10 @@ void BM_vert_separate( } } - bmesh_vert_separate(bm, v, r_vout, r_vout_len, copy_select); + bmesh_kernel_vert_separate(bm, v, r_vout, r_vout_len, copy_select); if (edges_separate) { - bmesh_vert_separate__cleanup(bm, edges_separate); + bmesh_kernel_vert_separate__cleanup(bm, edges_separate); } } @@ -2472,7 +2472,7 @@ void BM_vert_separate_hflag( LinkNode *edges_orig = NULL; do { BMLoop *l_sep = e->l; - bmesh_edge_separate(bm, e, l_sep, copy_select); + bmesh_kernel_edge_separate(bm, e, l_sep, copy_select); /* trick to avoid looping over separated edges */ if (edges_separate == NULL && edges_orig == NULL) { e_first = l_sep->e; @@ -2486,10 +2486,10 @@ void BM_vert_separate_hflag( } } while ((e_iter = BM_DISK_EDGE_NEXT(e_iter, v)) != e_first); - bmesh_vert_separate(bm, v, r_vout, r_vout_len, copy_select); + bmesh_kernel_vert_separate(bm, v, r_vout, r_vout_len, copy_select); if (edges_separate) { - bmesh_vert_separate__cleanup(bm, edges_separate); + bmesh_kernel_vert_separate__cleanup(bm, edges_separate); } } @@ -2574,7 +2574,7 @@ bool BM_edge_splice(BMesh *bm, BMEdge *e_dst, BMEdge *e_src) * \note Does nothing if \a l_sep is already the only loop in the * edge radial. */ -void bmesh_edge_separate( +void bmesh_kernel_edge_separate( BMesh *bm, BMEdge *e, BMLoop *l_sep, const bool copy_select) { @@ -2620,7 +2620,7 @@ void bmesh_edge_separate( * * \note Will be a no-op and return original vertex if only two edges at that vertex. */ -BMVert *bmesh_urmv_loop(BMesh *bm, BMLoop *l_sep) +BMVert *bmesh_kernel_unglue_region_make_vert(BMesh *bm, BMLoop *l_sep) { BMVert *v_new = NULL; BMVert *v_sep = l_sep->v; @@ -2630,10 +2630,12 @@ BMVert *bmesh_urmv_loop(BMesh *bm, BMLoop *l_sep) /* peel the face from the edge radials on both sides of the * loop vert, disconnecting the face from its fan */ - if (!BM_edge_is_boundary(l_sep->e)) - bmesh_edge_separate(bm, l_sep->e, l_sep, false); - if (!BM_edge_is_boundary(l_sep->prev->e)) - bmesh_edge_separate(bm, l_sep->prev->e, l_sep->prev, false); + if (!BM_edge_is_boundary(l_sep->e)) { + bmesh_kernel_edge_separate(bm, l_sep->e, l_sep, false); + } + if (!BM_edge_is_boundary(l_sep->prev->e)) { + bmesh_kernel_edge_separate(bm, l_sep->prev->e, l_sep->prev, false); + } /* do inline, below */ #if 0 @@ -2681,21 +2683,23 @@ BMVert *bmesh_urmv_loop(BMesh *bm, BMLoop *l_sep) } /** - * A version of #bmesh_urmv_loop that disconnects multiple loops at once. + * A version of #bmesh_kernel_unglue_region_make_vert that disconnects multiple loops at once. * The loops must all share the same vertex, can be in any order * and are all moved to use a single new vertex - which is returned. * * This function handles the details of finding fans boundaries. */ -BMVert *bmesh_urmv_loop_multi( +BMVert *bmesh_kernel_unglue_region_make_vert_multi( BMesh *bm, BMLoop **larr, int larr_len) { BMVert *v_sep = larr[0]->v; BMVert *v_new; + int edges_len = 0; int i; - bool is_mixed_any = false; - - BLI_SMALLSTACK_DECLARE(edges, BMEdge *); + /* any edges not owned by 'larr' loops connected to 'v_sep'? */ + bool is_mixed_edge_any = false; + /* any loops not owned by 'larr' radially connected to 'larr' loop edges? */ + bool is_mixed_loop_any = false; #define LOOP_VISIT _FLAG_WALK #define EDGE_VISIT _FLAG_WALK @@ -2713,58 +2717,74 @@ BMVert *bmesh_urmv_loop_multi( * while doing a radial loop (where loops may be adjacent) */ BM_ELEM_API_FLAG_ENABLE(l_sep->next, LOOP_VISIT); BM_ELEM_API_FLAG_ENABLE(l_sep->prev, LOOP_VISIT); - } - - for (i = 0; i < larr_len; i++) { - BMLoop *l_sep = larr[i]; BMLoop *loop_pair[2] = {l_sep, l_sep->prev}; - int j; - for (j = 0; j < ARRAY_SIZE(loop_pair); j++) { + for (int j = 0; j < ARRAY_SIZE(loop_pair); j++) { BMEdge *e = loop_pair[j]->e; if (!BM_ELEM_API_FLAG_TEST(e, EDGE_VISIT)) { - BMLoop *l_iter, *l_first; - bool is_mixed = false; - BM_ELEM_API_FLAG_ENABLE(e, EDGE_VISIT); + edges_len += 1; + } + } + } - l_iter = l_first = e->l; + BMEdge **edges = BLI_array_alloca(edges, edges_len); + STACK_DECLARE(edges); + + STACK_INIT(edges, edges_len); + + { + BMEdge *e_first, *e_iter; + e_iter = e_first = v_sep->e; + do { + if (BM_ELEM_API_FLAG_TEST(e_iter, EDGE_VISIT)) { + BMLoop *l_iter, *l_first; + bool is_mixed_loop = false; + + l_iter = l_first = e_iter->l; do { if (!BM_ELEM_API_FLAG_TEST(l_iter, LOOP_VISIT)) { - is_mixed = true; - is_mixed_any = true; + is_mixed_loop = true; break; } } while ((l_iter = l_iter->radial_next) != l_first); - if (is_mixed) { + if (is_mixed_loop) { /* ensure the first loop is one we don't own so we can do a quick check below * on the edge's loop-flag to see if the edge is mixed or not. */ - e->l = l_iter; + e_iter->l = l_iter; + + is_mixed_loop_any = true; } - BLI_SMALLSTACK_PUSH(edges, e); + + STACK_PUSH(edges, e_iter); } - } + else { + /* at least one edge attached isn't connected to our loops */ + is_mixed_edge_any = true; + } + } while ((e_iter = bmesh_disk_edge_next(e_iter, v_sep)) != e_first); } - if (is_mixed_any == false) { + BLI_assert(edges_len == STACK_SIZE(edges)); + + if (is_mixed_loop_any == false && is_mixed_edge_any == false) { /* all loops in 'larr' are the sole owners of their edges. * nothing to split away from, this is a no-op */ v_new = v_sep; } else { - BMEdge *e; - - BLI_assert(!BLI_SMALLSTACK_IS_EMPTY(edges)); - v_new = BM_vert_create(bm, v_sep->co, v_sep, BM_CREATE_NOP); - while ((e = BLI_SMALLSTACK_POP(edges))) { + + for (i = 0; i < STACK_SIZE(edges); i++) { + BMEdge *e = edges[i]; BMLoop *l_iter, *l_first, *l_next; BMEdge *e_new; /* disable so copied edge isn't left dirty (loop edges are cleared last too) */ BM_ELEM_API_FLAG_DISABLE(e, EDGE_VISIT); + /* will always be false when (is_mixed_loop_any == false) */ if (!BM_ELEM_API_FLAG_TEST(e->l, LOOP_VISIT)) { /* edge has some loops owned by us, some owned by other loops */ BMVert *e_new_v_pair[2]; @@ -2853,9 +2873,9 @@ static void bmesh_edge_vert_swap__recursive(BMEdge *e, BMVert *v_dst, BMVert *v_ /** * This function assumes l_sep is apart of a larger fan which has already been - * isolated by calling bmesh_edge_separate to segregate it radially. + * isolated by calling #bmesh_kernel_edge_separate to segregate it radially. */ -BMVert *bmesh_urmv_loop_region(BMesh *bm, BMLoop *l_sep) +BMVert *bmesh_kernel_unglue_region_make_vert_multi_isolated(BMesh *bm, BMLoop *l_sep) { BMVert *v_new = BM_vert_create(bm, l_sep->v->co, l_sep->v, BM_CREATE_NOP); /* passing either 'l_sep->e', 'l_sep->prev->e' will work */ @@ -2864,20 +2884,6 @@ BMVert *bmesh_urmv_loop_region(BMesh *bm, BMLoop *l_sep) return v_new; } - -/** - * \brief Unglue Region Make Vert (URMV) - * - * Disconnects f_sep from the vertex fan at \a v_sep - * - * \return The newly created BMVert - */ -BMVert *bmesh_urmv(BMesh *bm, BMFace *f_sep, BMVert *v_sep) -{ - BMLoop *l = BM_face_vert_share_loop(f_sep, v_sep); - return bmesh_urmv_loop(bm, l); -} - /** * Avoid calling this where possible, * low level function so both face pointers remain intact but point to swapped data. diff --git a/source/blender/bmesh/intern/bmesh_core.h b/source/blender/bmesh/intern/bmesh_core.h index f72e9d7b198..fb6b66809f3 100644 --- a/source/blender/bmesh/intern/bmesh_core.h +++ b/source/blender/bmesh/intern/bmesh_core.h @@ -64,21 +64,16 @@ void BM_face_kill(BMesh *bm, BMFace *f); void BM_edge_kill(BMesh *bm, BMEdge *e); void BM_vert_kill(BMesh *bm, BMVert *v); -void bmesh_edge_separate( - BMesh *bm, BMEdge *e, BMLoop *l_sep, - const bool copy_select); bool BM_edge_splice(BMesh *bm, BMEdge *e_dst, BMEdge *e_src); bool BM_vert_splice(BMesh *bm, BMVert *v_dst, BMVert *v_src); bool BM_vert_splice_check_double(BMVert *v_a, BMVert *v_b); -void bmesh_vert_separate( - BMesh *bm, BMVert *v, BMVert ***r_vout, int *r_vout_len, - const bool copy_select); - -void bmesh_loop_reverse( +void bmesh_kernel_loop_reverse( BMesh *bm, BMFace *f, const int cd_loop_mdisp_offset, const bool use_loop_mdisp_flip); +void bmesh_face_swap_data(BMFace *f_a, BMFace *f_b); + BMFace *BM_faces_join(BMesh *bm, BMFace **faces, int totface, const bool do_del); void BM_vert_separate( BMesh *bm, BMVert *v, BMEdge **e_in, int e_in_len, const bool copy_select, @@ -90,34 +85,43 @@ void BM_vert_separate_wire_hflag( BMesh *bm, BMVert *v_dst, BMVert *v_src, const char hflag); -/* EULER API - For modifying structure */ -BMFace *bmesh_sfme( +/** + * BMesh Kernel: For modifying structure. + * + * Names are on the verbose side but these are only for low-level access. + */ +void bmesh_kernel_vert_separate( + BMesh *bm, BMVert *v, BMVert ***r_vout, int *r_vout_len, + const bool copy_select); +void bmesh_kernel_edge_separate( + BMesh *bm, BMEdge *e, BMLoop *l_sep, + const bool copy_select); + +BMFace *bmesh_kernel_split_face_make_edge( BMesh *bm, BMFace *f, BMLoop *l1, BMLoop *l2, BMLoop **r_l, #ifdef USE_BMESH_HOLES - ListBase *holes, + ListBase *holes, #endif - BMEdge *example, - const bool no_double - ); + BMEdge *example, + const bool no_double + ); -BMVert *bmesh_semv(BMesh *bm, BMVert *tv, BMEdge *e, BMEdge **r_e); -BMEdge *bmesh_jekv( +BMVert *bmesh_kernel_split_edge_make_vert( + BMesh *bm, BMVert *tv, BMEdge *e, BMEdge **r_e); +BMEdge *bmesh_kernel_join_edge_kill_vert( BMesh *bm, BMEdge *e_kill, BMVert *v_kill, const bool do_del, const bool check_edge_splice, const bool kill_degenerate_faces); -BMVert *bmesh_jvke( +BMVert *bmesh_kernel_join_vert_kill_edge( BMesh *bm, BMEdge *e_kill, BMVert *v_kill, const bool do_del, const bool check_edge_double, const bool kill_degenerate_faces); -BMFace *bmesh_jfke(BMesh *bm, BMFace *f1, BMFace *f2, BMEdge *e); -BMVert *bmesh_urmv(BMesh *bm, BMFace *f_sep, BMVert *v_sep); -BMVert *bmesh_urmv_loop(BMesh *bm, BMLoop *l_sep); -BMVert *bmesh_urmv_loop_multi( - BMesh *bm, BMLoop **larr, int larr_len); -BMVert *bmesh_urmv_loop_region(BMesh *bm, BMLoop *l_sep); +BMFace *bmesh_kernel_join_face_kill_edge(BMesh *bm, BMFace *f1, BMFace *f2, BMEdge *e); -void bmesh_face_swap_data(BMFace *f_a, BMFace *f_b); +BMVert *bmesh_kernel_unglue_region_make_vert(BMesh *bm, BMLoop *l_sep); +BMVert *bmesh_kernel_unglue_region_make_vert_multi(BMesh *bm, BMLoop **larr, int larr_len); +BMVert *bmesh_kernel_unglue_region_make_vert_multi_isolated(BMesh *bm, BMLoop *l_sep); #endif /* __BMESH_CORE_H__ */ diff --git a/source/blender/bmesh/intern/bmesh_mesh.c b/source/blender/bmesh/intern/bmesh_mesh.c index 57a6d8d2e1a..e9fb6a28154 100644 --- a/source/blender/bmesh/intern/bmesh_mesh.c +++ b/source/blender/bmesh/intern/bmesh_mesh.c @@ -486,8 +486,7 @@ static void bm_mesh_edges_sharp_tag( BMesh *bm, const float (*vnos)[3], const float (*fnos)[3], float split_angle, float (*r_lnos)[3]) { - BMIter eiter, viter; - BMVert *v; + BMIter eiter; BMEdge *e; int i; @@ -498,19 +497,13 @@ static void bm_mesh_edges_sharp_tag( } { - char htype = BM_LOOP; + char htype = BM_VERT | BM_LOOP; if (fnos) { htype |= BM_FACE; } BM_mesh_elem_index_ensure(bm, htype); } - /* Clear all vertices' tags (means they are all smooth for now). */ - BM_ITER_MESH_INDEX (v, &viter, bm, BM_VERTS_OF_MESH, i) { - BM_elem_index_set(v, i); /* set_inline */ - BM_elem_flag_disable(v, BM_ELEM_TAG); - } - /* This first loop checks which edges are actually smooth, and pre-populate lnos with vnos (as if they were * all smooth). */ @@ -551,20 +544,45 @@ static void bm_mesh_edges_sharp_tag( no = vnos ? vnos[BM_elem_index_get(l_b->v)] : l_b->v->no; copy_v3_v3(r_lnos[BM_elem_index_get(l_b)], no); } - else { - /* Sharp edge, tag its verts as such. */ - BM_elem_flag_enable(e->v1, BM_ELEM_TAG); - BM_elem_flag_enable(e->v2, BM_ELEM_TAG); + } + } + + bm->elem_index_dirty &= ~BM_EDGE; +} + +/* Check whether gievn loop is part of an unknown-so-far cyclic smooth fan, or not. + * Needed because cyclic smooth fans have no obvious 'entry point', and yet we need to walk them once, and only once. */ +static bool bm_mesh_loop_check_cyclic_smooth_fan(BMLoop *l_curr) +{ + BMLoop *lfan_pivot_next = l_curr; + BMEdge *e_next = l_curr->e; + + BLI_assert(!BM_elem_flag_test(lfan_pivot_next, BM_ELEM_TAG)); + BM_elem_flag_enable(lfan_pivot_next, BM_ELEM_TAG); + + while (true) { + /* Much simpler than in sibling code with basic Mesh data! */ + lfan_pivot_next = BM_vert_step_fan_loop(lfan_pivot_next, &e_next); + + if (!lfan_pivot_next || !BM_elem_flag_test(e_next, BM_ELEM_TAG)) { + /* Sharp loop/edge, so not a cyclic smooth fan... */ + return false; + } + /* Smooth loop/edge... */ + else if (BM_elem_flag_test(lfan_pivot_next, BM_ELEM_TAG)) { + if (lfan_pivot_next == l_curr) { + /* We walked around a whole cyclic smooth fan without finding any already-processed loop, means we can + * use initial l_curr/l_prev edge as start for this smooth fan. */ + return true; } + /* ... already checked in some previous looping, we can abort. */ + return false; } else { - /* Sharp edge, tag its verts as such. */ - BM_elem_flag_enable(e->v1, BM_ELEM_TAG); - BM_elem_flag_enable(e->v2, BM_ELEM_TAG); + /* ... we can skip it in future, and keep checking the smooth fan. */ + BM_elem_flag_enable(lfan_pivot_next, BM_ELEM_TAG); } } - - bm->elem_index_dirty &= ~(BM_EDGE | BM_VERT); } /* BMesh version of BKE_mesh_normals_loop_split() in mesh_evaluate.c @@ -587,13 +605,11 @@ static void bm_mesh_loops_calc_normals( BLI_Stack *edge_vectors = NULL; { - char htype = BM_LOOP; + char htype = 0; if (vcos) { htype |= BM_VERT; } - if (fnos) { - htype |= BM_FACE; - } + /* Face/Loop indices are set inline below. */ BM_mesh_elem_index_ensure(bm, htype); } @@ -606,6 +622,21 @@ static void bm_mesh_loops_calc_normals( edge_vectors = BLI_stack_new(sizeof(float[3]), __func__); } + /* Clear all loops' tags (means none are to be skipped for now). */ + int index_face, index_loop = 0; + BM_ITER_MESH_INDEX (f_curr, &fiter, bm, BM_FACES_OF_MESH, index_face) { + BMLoop *l_curr, *l_first; + + BM_elem_index_set(f_curr, index_face); /* set_inline */ + + l_curr = l_first = BM_FACE_FIRST_LOOP(f_curr); + do { + BM_elem_index_set(l_curr, index_loop++); /* set_inline */ + BM_elem_flag_disable(l_curr, BM_ELEM_TAG); + } while ((l_curr = l_curr->next) != l_first); + } + bm->elem_index_dirty &= ~(BM_FACE|BM_LOOP); + /* We now know edges that can be smoothed (they are tagged), and edges that will be hard (they aren't). * Now, time to generate the normals. */ @@ -614,16 +645,16 @@ static void bm_mesh_loops_calc_normals( l_curr = l_first = BM_FACE_FIRST_LOOP(f_curr); do { + /* A smooth edge, we have to check for cyclic smooth fan case. + * If we find a new, never-processed cyclic smooth fan, we can do it now using that loop/edge as + * 'entry point', otherwise we can skip it. */ + /* Note: In theory, we could make bm_mesh_loop_check_cyclic_smooth_fan() store mlfan_pivot's in a stack, + * to avoid having to fan again around the vert during actual computation of clnor & clnorspace. + * However, this would complicate the code, add more memory usage, and BM_vert_step_fan_loop() + * is quite cheap in term of CPU cycles, so really think it's not worth it. */ if (BM_elem_flag_test(l_curr->e, BM_ELEM_TAG) && - (!r_lnors_spacearr || BM_elem_flag_test(l_curr->v, BM_ELEM_TAG))) + (BM_elem_flag_test(l_curr, BM_ELEM_TAG) || !bm_mesh_loop_check_cyclic_smooth_fan(l_curr))) { - /* A smooth edge, and we are not generating lnors_spacearr, or the related vertex is sharp. - * We skip it because it is either: - * - in the middle of a 'smooth fan' already computed (or that will be as soon as we hit - * one of its ends, i.e. one of its two sharp edges), or... - * - the related vertex is a "full smooth" one, in which case pre-populated normals from vertex - * are just fine! - */ } else if (!BM_elem_flag_test(l_curr->e, BM_ELEM_TAG) && !BM_elem_flag_test(l_curr->prev->e, BM_ELEM_TAG)) @@ -1481,23 +1512,6 @@ int BM_mesh_elem_count(BMesh *bm, const char htype) } } -/** - * Special case: Python uses custom-data layers to hold PyObject references. - * These have to be kept in-place, else the PyObject's we point to, wont point back to us. - * - * \note ``ele_src`` Is a duplicate, so we don't need to worry about getting in a feedback loop. - * - * \note If there are other customdata layers which need this functionality, it should be generalized. - * However #BM_mesh_remap is currently the only place where this is done. - */ -static void bm_mesh_remap_cd_update( - BMHeader *ele_dst, BMHeader *ele_src, - const int cd_elem_pyptr) -{ - void **pyptr_dst_p = BM_ELEM_CD_GET_VOID_P(((BMElem *)ele_dst), cd_elem_pyptr); - void **pyptr_src_p = BM_ELEM_CD_GET_VOID_P(((BMElem *)ele_src), cd_elem_pyptr); - *pyptr_dst_p = *pyptr_src_p; -} /** * Remaps the vertices, edges and/or faces of the bmesh as indicated by vert/edge/face_idx arrays @@ -1539,6 +1553,8 @@ void BM_mesh_remap( BMVert **verts_pool, *verts_copy, **vep; int i, totvert = bm->totvert; const unsigned int *new_idx; + /* Special case: Python uses custom - data layers to hold PyObject references. + * These have to be kept in - place, else the PyObject's we point to, wont point back to us. */ const int cd_vert_pyptr = CustomData_get_offset(&bm->vdata, CD_BM_ELEM_PYPTR); /* Init the old-to-new vert pointers mapping */ @@ -1547,9 +1563,14 @@ void BM_mesh_remap( /* Make a copy of all vertices. */ verts_pool = bm->vtable; verts_copy = MEM_mallocN(sizeof(BMVert) * totvert, "BM_mesh_remap verts copy"); + void **pyptrs = (cd_vert_pyptr != -1) ? MEM_mallocN(sizeof(void *) * totvert, __func__) : NULL; for (i = totvert, ve = verts_copy + totvert - 1, vep = verts_pool + totvert - 1; i--; ve--, vep--) { *ve = **vep; /* printf("*vep: %p, verts_pool[%d]: %p\n", *vep, i, verts_pool[i]);*/ + if (cd_vert_pyptr != -1) { + void **pyptr = BM_ELEM_CD_GET_VOID_P(((BMElem *)ve), cd_vert_pyptr); + pyptrs[i] = *pyptr; + } } /* Copy back verts to their new place, and update old2new pointers mapping. */ @@ -1562,13 +1583,17 @@ void BM_mesh_remap( /* printf("mapping vert from %d to %d (%p/%p to %p)\n", i, *new_idx, *vep, verts_pool[i], new_vep);*/ BLI_ghash_insert(vptr_map, *vep, new_vep); if (cd_vert_pyptr != -1) { - bm_mesh_remap_cd_update(&(*vep)->head, &new_vep->head, cd_vert_pyptr); + void **pyptr = BM_ELEM_CD_GET_VOID_P(((BMElem *)new_vep), cd_vert_pyptr); + *pyptr = pyptrs[*new_idx]; } } bm->elem_index_dirty |= BM_VERT; bm->elem_table_dirty |= BM_VERT; MEM_freeN(verts_copy); + if (pyptrs) { + MEM_freeN(pyptrs); + } } /* Remap Edges */ @@ -1576,6 +1601,8 @@ void BM_mesh_remap( BMEdge **edges_pool, *edges_copy, **edp; int i, totedge = bm->totedge; const unsigned int *new_idx; + /* Special case: Python uses custom - data layers to hold PyObject references. + * These have to be kept in - place, else the PyObject's we point to, wont point back to us. */ const int cd_edge_pyptr = CustomData_get_offset(&bm->edata, CD_BM_ELEM_PYPTR); /* Init the old-to-new vert pointers mapping */ @@ -1584,8 +1611,13 @@ void BM_mesh_remap( /* Make a copy of all vertices. */ edges_pool = bm->etable; edges_copy = MEM_mallocN(sizeof(BMEdge) * totedge, "BM_mesh_remap edges copy"); + void **pyptrs = (cd_edge_pyptr != -1) ? MEM_mallocN(sizeof(void *) * totedge, __func__) : NULL; for (i = totedge, ed = edges_copy + totedge - 1, edp = edges_pool + totedge - 1; i--; ed--, edp--) { *ed = **edp; + if (cd_edge_pyptr != -1) { + void **pyptr = BM_ELEM_CD_GET_VOID_P(((BMElem *)ed), cd_edge_pyptr); + pyptrs[i] = *pyptr; + } } /* Copy back verts to their new place, and update old2new pointers mapping. */ @@ -1598,13 +1630,17 @@ void BM_mesh_remap( BLI_ghash_insert(eptr_map, *edp, new_edp); /* printf("mapping edge from %d to %d (%p/%p to %p)\n", i, *new_idx, *edp, edges_pool[i], new_edp);*/ if (cd_edge_pyptr != -1) { - bm_mesh_remap_cd_update(&(*edp)->head, &new_edp->head, cd_edge_pyptr); + void **pyptr = BM_ELEM_CD_GET_VOID_P(((BMElem *)new_edp), cd_edge_pyptr); + *pyptr = pyptrs[*new_idx]; } } bm->elem_index_dirty |= BM_EDGE; bm->elem_table_dirty |= BM_EDGE; MEM_freeN(edges_copy); + if (pyptrs) { + MEM_freeN(pyptrs); + } } /* Remap Faces */ @@ -1612,6 +1648,8 @@ void BM_mesh_remap( BMFace **faces_pool, *faces_copy, **fap; int i, totface = bm->totface; const unsigned int *new_idx; + /* Special case: Python uses custom - data layers to hold PyObject references. + * These have to be kept in - place, else the PyObject's we point to, wont point back to us. */ const int cd_poly_pyptr = CustomData_get_offset(&bm->pdata, CD_BM_ELEM_PYPTR); /* Init the old-to-new vert pointers mapping */ @@ -1620,8 +1658,13 @@ void BM_mesh_remap( /* Make a copy of all vertices. */ faces_pool = bm->ftable; faces_copy = MEM_mallocN(sizeof(BMFace) * totface, "BM_mesh_remap faces copy"); + void **pyptrs = (cd_poly_pyptr != -1) ? MEM_mallocN(sizeof(void *) * totface, __func__) : NULL; for (i = totface, fa = faces_copy + totface - 1, fap = faces_pool + totface - 1; i--; fa--, fap--) { *fa = **fap; + if (cd_poly_pyptr != -1) { + void **pyptr = BM_ELEM_CD_GET_VOID_P(((BMElem *)fa), cd_poly_pyptr); + pyptrs[i] = *pyptr; + } } /* Copy back verts to their new place, and update old2new pointers mapping. */ @@ -1633,7 +1676,8 @@ void BM_mesh_remap( *new_fap = *fa; BLI_ghash_insert(fptr_map, *fap, new_fap); if (cd_poly_pyptr != -1) { - bm_mesh_remap_cd_update(&(*fap)->head, &new_fap->head, cd_poly_pyptr); + void **pyptr = BM_ELEM_CD_GET_VOID_P(((BMElem *)new_fap), cd_poly_pyptr); + *pyptr = pyptrs[*new_idx]; } } @@ -1641,6 +1685,9 @@ void BM_mesh_remap( bm->elem_table_dirty |= BM_FACE; MEM_freeN(faces_copy); + if (pyptrs) { + MEM_freeN(pyptrs); + } } /* And now, fix all vertices/edges/faces/loops pointers! */ @@ -2008,4 +2055,4 @@ void BM_mesh_toolflags_set(BMesh *bm, bool use_toolflags) vpool_dst, epool_dst, NULL, fpool_dst); bm->use_toolflags = use_toolflags; -}
\ No newline at end of file +} diff --git a/source/blender/bmesh/intern/bmesh_mods.c b/source/blender/bmesh/intern/bmesh_mods.c index 500da6b8788..1cd51528e06 100644 --- a/source/blender/bmesh/intern/bmesh_mods.c +++ b/source/blender/bmesh/intern/bmesh_mods.c @@ -234,7 +234,7 @@ BMFace *BM_faces_join_pair(BMesh *bm, BMLoop *l_a, BMLoop *l_b, const bool do_de if (l_a->v == l_b->v) { const int cd_loop_mdisp_offset = CustomData_get_offset(&bm->ldata, CD_MDISPS); - bmesh_loop_reverse(bm, l_b->f, cd_loop_mdisp_offset, true); + bmesh_kernel_loop_reverse(bm, l_b->f, cd_loop_mdisp_offset, true); } BMFace *faces[2] = {l_a->f, l_b->f}; @@ -288,9 +288,9 @@ BMFace *BM_face_split( } #ifdef USE_BMESH_HOLES - f_new = bmesh_sfme(bm, f, l_a, l_b, r_l, NULL, example, no_double); + f_new = bmesh_kernel_split_face_make_edge(bm, f, l_a, l_b, r_l, NULL, example, no_double); #else - f_new = bmesh_sfme(bm, f, l_a, l_b, r_l, example, no_double); + f_new = bmesh_kernel_split_face_make_edge(bm, f, l_a, l_b, r_l, example, no_double); #endif if (f_new) { @@ -370,19 +370,19 @@ BMFace *BM_face_split_n( f_tmp = BM_face_copy(bm, bm, f, true, true); #ifdef USE_BMESH_HOLES - f_new = bmesh_sfme(bm, f, l_a, l_b, &l_new, NULL, example, false); + f_new = bmesh_kernel_split_face_make_edge(bm, f, l_a, l_b, &l_new, NULL, example, false); #else - f_new = bmesh_sfme(bm, f, l_a, l_b, &l_new, example, false); + f_new = bmesh_kernel_split_face_make_edge(bm, f, l_a, l_b, &l_new, example, false); #endif - /* bmesh_sfme returns in 'l_new' a Loop for f_new going from 'v_a' to 'v_b'. + /* bmesh_kernel_split_face_make_edge returns in 'l_new' a Loop for f_new going from 'v_a' to 'v_b'. * The radial_next is for 'f' and goes from 'v_b' to 'v_a' */ if (f_new) { e = l_new->e; for (i = 0; i < n; i++) { - v_new = bmesh_semv(bm, v_b, e, &e_new); + v_new = bmesh_kernel_split_edge_make_vert(bm, v_b, e, &e_new); BLI_assert(v_new != NULL); - /* bmesh_semv returns in 'e_new' the edge going from 'v_new' to 'v_b' */ + /* bmesh_kernel_split_edge_make_vert returns in 'e_new' the edge going from 'v_new' to 'v_b' */ copy_v3_v3(v_new->co, cos[i]); /* interpolate the loop data for the loops with (v == v_new), using orig face */ @@ -507,7 +507,7 @@ BMEdge *BM_vert_collapse_faces( /* single face or no faces */ /* same as BM_vert_collapse_edge() however we already * have vars to perform this operation so don't call. */ - e_new = bmesh_jekv(bm, e_kill, v_kill, do_del, true, kill_degenerate_faces); + e_new = bmesh_kernel_join_edge_kill_vert(bm, e_kill, v_kill, do_del, true, kill_degenerate_faces); /* e_new = BM_edge_exists(tv, tv2); */ /* same as return above */ } @@ -542,7 +542,7 @@ BMEdge *BM_vert_collapse_edge( BMVert *tv2 = BM_edge_other_vert(e2, v_kill); if (tv2) { /* only action, other calls here only get the edge to return */ - e_new = bmesh_jekv(bm, e_kill, v_kill, do_del, true, kill_degenerate_faces); + e_new = bmesh_kernel_join_edge_kill_vert(bm, e_kill, v_kill, do_del, true, kill_degenerate_faces); } } } @@ -564,7 +564,7 @@ BMVert *BM_edge_collapse( BMesh *bm, BMEdge *e_kill, BMVert *v_kill, const bool do_del, const bool kill_degenerate_faces) { - return bmesh_jvke(bm, e_kill, v_kill, do_del, true, kill_degenerate_faces); + return bmesh_kernel_join_vert_kill_edge(bm, e_kill, v_kill, do_del, true, kill_degenerate_faces); } /** @@ -616,7 +616,7 @@ BMVert *BM_edge_split(BMesh *bm, BMEdge *e, BMVert *v, BMEdge **r_e, float fac) } v_other = BM_edge_other_vert(e, v); - v_new = bmesh_semv(bm, v, e, &e_new); + v_new = bmesh_kernel_split_edge_make_vert(bm, v, e, &e_new); if (r_e != NULL) { *r_e = e_new; } @@ -1090,23 +1090,18 @@ BMEdge *BM_edge_rotate(BMesh *bm, BMEdge *e, const bool ccw, const short check_f /** * \brief Rip a single face from a vertex fan */ -BMVert *BM_face_vert_separate(BMesh *bm, BMFace *sf, BMVert *sv) +BMVert *BM_face_loop_separate(BMesh *bm, BMLoop *l_sep) { - return bmesh_urmv(bm, sf, sv); + return bmesh_kernel_unglue_region_make_vert(bm, l_sep); } -/** - * \brief Rip a single face from a vertex fan - * - * \note same as #BM_face_vert_separate but faster (avoids a loop lookup) - */ -BMVert *BM_face_loop_separate(BMesh *bm, BMLoop *sl) +BMVert *BM_face_loop_separate_multi_isolated(BMesh *bm, BMLoop *l_sep) { - return bmesh_urmv_loop(bm, sl); + return bmesh_kernel_unglue_region_make_vert_multi_isolated(bm, l_sep); } -BMVert *BM_face_loop_separate_multi( - BMesh *bm, BMLoop **larr, int larr_len) +BMVert *BM_face_loop_separate_multi(BMesh *bm, BMLoop **larr, int larr_len) { - return bmesh_urmv_loop_multi(bm, larr, larr_len); + return bmesh_kernel_unglue_region_make_vert_multi(bm, larr, larr_len); } + diff --git a/source/blender/bmesh/intern/bmesh_mods.h b/source/blender/bmesh/intern/bmesh_mods.h index 5e95e9a2cc7..330a714418d 100644 --- a/source/blender/bmesh/intern/bmesh_mods.h +++ b/source/blender/bmesh/intern/bmesh_mods.h @@ -86,9 +86,8 @@ enum { }; -BMVert *BM_face_vert_separate(BMesh *bm, BMFace *sf, BMVert *sv); -BMVert *BM_face_loop_separate(BMesh *bm, BMLoop *sl); -BMVert *BM_face_loop_separate_multi( - BMesh *bm, BMLoop **larr, int larr_len); +BMVert *BM_face_loop_separate(BMesh *bm, BMLoop *l_sep); +BMVert *BM_face_loop_separate_multi_isolated(BMesh *bm, BMLoop *l_sep); +BMVert *BM_face_loop_separate_multi(BMesh *bm, BMLoop **larr, int larr_len); #endif /* __BMESH_MODS_H__ */ diff --git a/source/blender/bmesh/intern/bmesh_opdefines.c b/source/blender/bmesh/intern/bmesh_opdefines.c index 0d0fdda2c4c..6b388a75436 100644 --- a/source/blender/bmesh/intern/bmesh_opdefines.c +++ b/source/blender/bmesh/intern/bmesh_opdefines.c @@ -1284,7 +1284,7 @@ static BMOpDefine bmo_bisect_plane_def = { {"clear_inner", BMO_OP_SLOT_BOOL}, /* when enabled. remove all geometry on the negative side of the plane */ {{'\0'}}, }, - {{"geom_cut.out", BMO_OP_SLOT_ELEMENT_BUF, {BM_VERT | BM_EDGE}}, /* output new geometry from the cut */ + {{"geom_cut.out", BMO_OP_SLOT_ELEMENT_BUF, {BM_VERT | BM_EDGE}}, /* output geometry aligned with the plane (new and existing) */ {"geom.out", BMO_OP_SLOT_ELEMENT_BUF, {BM_VERT | BM_EDGE | BM_FACE}}, /* input and output geometry (result of cut) */ {{'\0'}}}, bmo_bisect_plane_exec, diff --git a/source/blender/bmesh/intern/bmesh_polygon.c b/source/blender/bmesh/intern/bmesh_polygon.c index 6acd790fc0c..78ee7589d23 100644 --- a/source/blender/bmesh/intern/bmesh_polygon.c +++ b/source/blender/bmesh/intern/bmesh_polygon.c @@ -844,7 +844,7 @@ void BM_face_normal_flip_ex( BMesh *bm, BMFace *f, const int cd_loop_mdisp_offset, const bool use_loop_mdisp_flip) { - bmesh_loop_reverse(bm, f, cd_loop_mdisp_offset, use_loop_mdisp_flip); + bmesh_kernel_loop_reverse(bm, f, cd_loop_mdisp_offset, use_loop_mdisp_flip); negate_v3(f->no); } diff --git a/source/blender/bmesh/operators/bmo_bisect_plane.c b/source/blender/bmesh/operators/bmo_bisect_plane.c index bed1ea5cb94..2c80ff651b8 100644 --- a/source/blender/bmesh/operators/bmo_bisect_plane.c +++ b/source/blender/bmesh/operators/bmo_bisect_plane.c @@ -38,7 +38,8 @@ #include "intern/bmesh_operators_private.h" /* own include */ #define ELE_NEW 1 -#define ELE_INPUT 2 +#define ELE_CUT 2 +#define ELE_INPUT 4 void bmo_bisect_plane_exec(BMesh *bm, BMOperator *op) { @@ -69,7 +70,7 @@ void bmo_bisect_plane_exec(BMesh *bm, BMOperator *op) BM_mesh_bisect_plane(bm, plane, use_snap_center, true, - ELE_NEW, dist); + ELE_CUT, ELE_NEW, dist); if (clear_outer || clear_inner) { @@ -108,5 +109,5 @@ void bmo_bisect_plane_exec(BMesh *bm, BMOperator *op) } BMO_slot_buffer_from_enabled_flag(bm, op, op->slots_out, "geom.out", BM_ALL_NOLOOP, ELE_NEW | ELE_INPUT); - BMO_slot_buffer_from_enabled_flag(bm, op, op->slots_out, "geom_cut.out", BM_VERT | BM_EDGE, ELE_NEW); + BMO_slot_buffer_from_enabled_flag(bm, op, op->slots_out, "geom_cut.out", BM_VERT | BM_EDGE, ELE_CUT); } diff --git a/source/blender/bmesh/operators/bmo_dupe.c b/source/blender/bmesh/operators/bmo_dupe.c index 56639a097b6..8048add84d4 100644 --- a/source/blender/bmesh/operators/bmo_dupe.c +++ b/source/blender/bmesh/operators/bmo_dupe.c @@ -378,6 +378,10 @@ void BMO_dupe_from_flag(BMesh *bm, int htype, const char hflag) * BMOP_DUPE_VOUTPUT: Buffer containing pointers to the split mesh vertices * BMOP_DUPE_EOUTPUT: Buffer containing pointers to the split mesh edges * BMOP_DUPE_FOUTPUT: Buffer containing pointers to the split mesh faces + * + * \note Lower level uses of this operator may want to use #BM_mesh_separate_faces + * Since it's faster for the 'use_only_faces' case. + * */ void bmo_split_exec(BMesh *bm, BMOperator *op) { diff --git a/source/blender/bmesh/operators/bmo_inset.c b/source/blender/bmesh/operators/bmo_inset.c index e2ff09669d7..f0796e425fa 100644 --- a/source/blender/bmesh/operators/bmo_inset.c +++ b/source/blender/bmesh/operators/bmo_inset.c @@ -663,7 +663,7 @@ void bmo_inset_region_exec(BMesh *bm, BMOperator *op) /* run the separate arg */ if (!BM_edge_is_boundary(es->e_old)) { - bmesh_edge_separate(bm, es->e_old, es->l, false); + bmesh_kernel_edge_separate(bm, es->e_old, es->l, false); } /* calc edge-split info */ @@ -742,7 +742,7 @@ void bmo_inset_region_exec(BMesh *bm, BMOperator *op) /* disable touching twice, this _will_ happen if the flags not disabled */ BM_elem_flag_disable(v, BM_ELEM_TAG); - bmesh_vert_separate(bm, v, &vout, &r_vout_len, false); + bmesh_kernel_vert_separate(bm, v, &vout, &r_vout_len, false); v = NULL; /* don't use again */ /* in some cases the edge doesn't split off */ diff --git a/source/blender/bmesh/operators/bmo_offset_edgeloops.c b/source/blender/bmesh/operators/bmo_offset_edgeloops.c index 7a6f779b34f..a9840a72fc9 100644 --- a/source/blender/bmesh/operators/bmo_offset_edgeloops.c +++ b/source/blender/bmesh/operators/bmo_offset_edgeloops.c @@ -270,7 +270,7 @@ void bmo_offset_edgeloops_exec(BMesh *bm, BMOperator *op) v_other = BM_edge_other_vert(e, v); if (BM_elem_index_get(v_other) == -1) { if (BM_vert_is_edge_pair(v_other)) { - /* defer bmesh_jekv to avoid looping over data we're removing */ + /* defer bmesh_kernel_join_edge_kill_vert to avoid looping over data we're removing */ v_other->e = e; STACK_PUSH(varr, v_other); } @@ -278,7 +278,7 @@ void bmo_offset_edgeloops_exec(BMesh *bm, BMOperator *op) } while ((v = STACK_POP(varr))) { - bmesh_jekv(bm, v->e, v, true, false, false); + bmesh_kernel_join_edge_kill_vert(bm, v->e, v, true, false, false); } } } diff --git a/source/blender/bmesh/operators/bmo_primitive.c b/source/blender/bmesh/operators/bmo_primitive.c index 8408169d85e..723e0b168e0 100644 --- a/source/blender/bmesh/operators/bmo_primitive.c +++ b/source/blender/bmesh/operators/bmo_primitive.c @@ -1122,7 +1122,7 @@ static void bm_mesh_calc_uvs_sphere_face(BMFace *f, const int cd_loop_uv_offset) } /* Shift borderline coordinates to the left. */ - if (fabsf(theta - M_PI) < 0.0001f) { + if (fabsf(theta - (float)M_PI) < 0.0001f) { theta = -M_PI; } diff --git a/source/blender/bmesh/tools/bmesh_bisect_plane.c b/source/blender/bmesh/tools/bmesh_bisect_plane.c index 51b92a3c45e..828c50c39fd 100644 --- a/source/blender/bmesh/tools/bmesh_bisect_plane.c +++ b/source/blender/bmesh/tools/bmesh_bisect_plane.c @@ -110,7 +110,7 @@ static int bm_vert_sortval_cb(const void *v_a_v, const void *v_b_v) } -static void bm_face_bisect_verts(BMesh *bm, BMFace *f, const float plane[4], const short oflag_center) +static void bm_face_bisect_verts(BMesh *bm, BMFace *f, const float plane[4], const short oflag_center, const short oflag_new) { /* unlikely more than 2 verts are needed */ const unsigned int f_len_orig = (unsigned int)f->len; @@ -154,10 +154,11 @@ static void bm_face_bisect_verts(BMesh *bm, BMFace *f, const float plane[4], con /* common case, just cut the face once */ BM_face_split(bm, f, l_a, l_b, &l_new, NULL, true); if (l_new) { - if (oflag_center) { - BMO_edge_flag_enable(bm, l_new->e, oflag_center); - BMO_face_flag_enable(bm, l_new->f, oflag_center); - BMO_face_flag_enable(bm, f, oflag_center); + if (oflag_center | oflag_new) { + BMO_edge_flag_enable(bm, l_new->e, oflag_center | oflag_new); + } + if (oflag_new) { + BMO_face_flag_enable(bm, l_new->f, oflag_new); } } } @@ -269,10 +270,11 @@ static void bm_face_bisect_verts(BMesh *bm, BMFace *f, const float plane[4], con f_tmp = BM_face_split(bm, face_split_arr[j], l_a, l_b, &l_new, NULL, true); if (l_new) { - if (oflag_center) { - BMO_edge_flag_enable(bm, l_new->e, oflag_center); - BMO_face_flag_enable(bm, l_new->f, oflag_center); - BMO_face_flag_enable(bm, face_split_arr[j], oflag_center); + if (oflag_center | oflag_new) { + BMO_edge_flag_enable(bm, l_new->e, oflag_center | oflag_new); + } + if (oflag_new) { + BMO_face_flag_enable(bm, l_new->f, oflag_new); } } @@ -307,7 +309,7 @@ finally: void BM_mesh_bisect_plane( BMesh *bm, const float plane[4], const bool use_snap_center, const bool use_tag, - const short oflag_center, const float eps) + const short oflag_center, const short oflag_new, const float eps) { unsigned int einput_len; unsigned int i; @@ -390,7 +392,7 @@ void BM_mesh_bisect_plane( const float dist[2] = {BM_VERT_DIST(e->v1), BM_VERT_DIST(e->v2)}; if (side[0] && side[1] && (side[0] != side[1])) { - const float e_fac = fabsf(dist[0]) / fabsf(dist[0] - dist[1]); + const float e_fac = dist[0] / (dist[0] - dist[1]); BMVert *v_new; if (e->l) { @@ -404,10 +406,17 @@ void BM_mesh_bisect_plane( } while ((l_iter = l_iter->radial_next) != l_first); } - v_new = BM_edge_split(bm, e, e->v1, NULL, e_fac); + { + BMEdge *e_new; + v_new = BM_edge_split(bm, e, e->v1, &e_new, e_fac); + if (oflag_new) { + BMO_edge_flag_enable(bm, e_new, oflag_new); + } + } + vert_is_center_enable(v_new); - if (oflag_center) { - BMO_vert_flag_enable(bm, v_new, oflag_center); + if (oflag_new | oflag_center) { + BMO_vert_flag_enable(bm, v_new, oflag_new | oflag_center); } BM_VERT_DIR(v_new) = 0; @@ -448,7 +457,7 @@ void BM_mesh_bisect_plane( MEM_freeN(edges_arr); while ((f = BLI_LINKSTACK_POP(face_stack))) { - bm_face_bisect_verts(bm, f, plane, oflag_center); + bm_face_bisect_verts(bm, f, plane, oflag_center, oflag_new); } /* now we have all faces to split in the stack */ diff --git a/source/blender/bmesh/tools/bmesh_bisect_plane.h b/source/blender/bmesh/tools/bmesh_bisect_plane.h index 7f3a97c4c79..fb99a1c8214 100644 --- a/source/blender/bmesh/tools/bmesh_bisect_plane.h +++ b/source/blender/bmesh/tools/bmesh_bisect_plane.h @@ -30,6 +30,6 @@ void BM_mesh_bisect_plane( BMesh *bm, const float plane[4], const bool use_snap_center, const bool use_tag, - const short oflag_center, const float eps); + const short oflag_center, const short oflag_new, const float eps); #endif /* __BMESH_BISECT_PLANE_H__ */ diff --git a/source/blender/bmesh/tools/bmesh_intersect.c b/source/blender/bmesh/tools/bmesh_intersect.c index 58234ddf3bd..2cb82d0fc02 100644 --- a/source/blender/bmesh/tools/bmesh_intersect.c +++ b/source/blender/bmesh/tools/bmesh_intersect.c @@ -986,7 +986,7 @@ bool BM_mesh_intersect( struct BMLoop *(*looptris)[3], const int looptris_tot, int (*test_fn)(BMFace *f, void *user_data), void *user_data, const bool use_self, const bool use_separate, const bool use_dissolve, const bool use_island_connect, - const int boolean_mode, + const bool use_edge_tag, const int boolean_mode, const float eps) { struct ISectState s; @@ -1526,7 +1526,7 @@ bool BM_mesh_intersect( BM_mesh_edgesplit(bm, false, true, false); } - else if (boolean_mode != BMESH_ISECT_BOOLEAN_NONE) { + else if (boolean_mode != BMESH_ISECT_BOOLEAN_NONE || use_edge_tag) { GSetIterator gs_iter; /* no need to clear for boolean */ diff --git a/source/blender/bmesh/tools/bmesh_intersect.h b/source/blender/bmesh/tools/bmesh_intersect.h index d0cc41654eb..51926a01710 100644 --- a/source/blender/bmesh/tools/bmesh_intersect.h +++ b/source/blender/bmesh/tools/bmesh_intersect.h @@ -30,7 +30,7 @@ bool BM_mesh_intersect( struct BMLoop *(*looptris)[3], const int looptris_tot, int (*test_fn)(BMFace *f, void *user_data), void *user_data, const bool use_self, const bool use_separate, const bool use_dissolve, const bool use_island_connect, - const int boolean_mode, + const bool use_edge_tag, const int boolean_mode, const float eps); enum { diff --git a/source/blender/bmesh/tools/bmesh_separate.c b/source/blender/bmesh/tools/bmesh_separate.c new file mode 100644 index 00000000000..ff28ed4c156 --- /dev/null +++ b/source/blender/bmesh/tools/bmesh_separate.c @@ -0,0 +1,133 @@ +/* + * ***** BEGIN GPL LICENSE BLOCK ***** + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License + * as published by the Free Software Foundation; either version 2 + * of the License, or (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software Foundation, + * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. + * + * ***** END GPL LICENSE BLOCK ***** + */ + +/** \file blender/bmesh/tools/bmesh_separate.c + * \ingroup bmesh + * + * BMesh separate, disconnects a set of faces from all others, + * so they don't share any vertices/edges with other faces. + */ + +#include <limits.h> + +#include "MEM_guardedalloc.h" + +#include "BLI_utildefines.h" +#include "BLI_buffer.h" + +#include "bmesh.h" +#include "intern/bmesh_private.h" +#include "bmesh_separate.h" /* own include */ + +/** + * Split all faces that match `filter_fn`. + * \note + */ +void BM_mesh_separate_faces( + BMesh *bm, + BMFaceFilterFunc filter_fn, void *user_data) +{ + BMFace **faces_array_all = MEM_mallocN(bm->totface * sizeof(BMFace *), __func__); + /* + * - Create an array of faces based on 'filter_fn'. + * First part of array for match, for non-match. + * + * - Enable all vertex tags, then clear all tagged vertices from 'faces_b'. + * + * - Loop over 'faces_a', checking each vertex, + * splitting out any which aren't tagged (and therefor shared), disabling tags as we go. + */ + + BMFace *f; + BMIter iter; + + unsigned int faces_a_len = 0; + unsigned int faces_b_len = 0; + { + int i_a = 0; + int i_b = bm->totface; + BM_ITER_MESH (f, &iter, bm, BM_FACES_OF_MESH) { + faces_array_all[filter_fn(f, user_data) ? i_a++ : --i_b] = f; + } + faces_a_len = i_a; + faces_b_len = bm->totface - i_a; + } + + BMFace **faces_a = faces_array_all; + BMFace **faces_b = faces_array_all + faces_a_len; + + /* Enable for all */ + BM_mesh_elem_hflag_enable_all(bm, BM_VERT, BM_ELEM_TAG, false); + + /* Disable vert tag on faces_b */ + for (unsigned int i = 0; i < faces_b_len; i++) { + BMLoop *l_iter, *l_first; + l_iter = l_first = BM_FACE_FIRST_LOOP(faces_b[i]); + do { + BM_elem_flag_disable(l_iter->v, BM_ELEM_TAG); + } while ((l_iter = l_iter->next) != l_first); + } + + + BLI_buffer_declare_static(BMLoop **, loop_split, 0, 128); + + /* Check shared verts ('faces_a' tag and disable) */ + for (unsigned int i = 0; i < faces_a_len; i++) { + BMLoop *l_iter, *l_first; + l_iter = l_first = BM_FACE_FIRST_LOOP(faces_a[i]); + do { + if (!BM_elem_flag_test(l_iter->v, BM_ELEM_TAG)) { + BMVert *v = l_iter->v; + /* Enable, since we may visit this vertex again on other faces */ + BM_elem_flag_enable(v, BM_ELEM_TAG); + + /* We know the vertex is shared, collect all vertices and split them off. */ + + /* Fill 'loop_split' */ + { + BMEdge *e_first, *e_iter; + e_iter = e_first = l_iter->e; + do { + if (e_iter->l != NULL) { + BMLoop *l_radial_first, *l_radial_iter; + l_radial_first = l_radial_iter = e_iter->l; + do { + if (l_radial_iter->v == v) { + if (filter_fn(l_radial_iter->f, user_data)) { + BLI_buffer_append(&loop_split, BMLoop *, l_radial_iter); + } + } + } while ((l_radial_iter = l_radial_iter->radial_next) != l_radial_first); + } + } while ((e_iter = bmesh_disk_edge_next(e_iter, v)) != e_first); + } + + /* Perform the split */ + BM_face_loop_separate_multi(bm, loop_split.data, loop_split.count); + + BLI_buffer_empty(&loop_split); + } + } while ((l_iter = l_iter->next) != l_first); + } + + BLI_buffer_free(&loop_split); + + MEM_freeN(faces_array_all); +} diff --git a/source/blender/bmesh/tools/bmesh_separate.h b/source/blender/bmesh/tools/bmesh_separate.h new file mode 100644 index 00000000000..91b2b71c872 --- /dev/null +++ b/source/blender/bmesh/tools/bmesh_separate.h @@ -0,0 +1,32 @@ +/* + * ***** BEGIN GPL LICENSE BLOCK ***** + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License + * as published by the Free Software Foundation; either version 2 + * of the License, or (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software Foundation, + * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. + * + * ***** END GPL LICENSE BLOCK ***** + */ + +#ifndef __BMESH_SEPARATE_H__ +#define __BMESH_SEPARATE_H__ + +/** \file blender/bmesh/tools/bmesh_separate.h + * \ingroup bmesh + */ + +void BM_mesh_separate_faces( + BMesh *bm, + BMFaceFilterFunc filter_fn, void *user_data); + +#endif /* __BMESH_SEPARATE_H__ */ diff --git a/source/blender/collada/AnimationImporter.cpp b/source/blender/collada/AnimationImporter.cpp index 3801c9300df..bd47ee0214d 100644 --- a/source/blender/collada/AnimationImporter.cpp +++ b/source/blender/collada/AnimationImporter.cpp @@ -1934,7 +1934,7 @@ Object *AnimationImporter::get_joint_object(COLLADAFW::Node *root, COLLADAFW::No mul_m4_m4m4(mat, ipar, temp); } - TransformBase::decompose(mat, job->loc, NULL, job->quat, job->size); + bc_decompose(mat, job->loc, NULL, job->quat, job->size); if (par_job) { job->parent = par_job; diff --git a/source/blender/collada/ArmatureExporter.cpp b/source/blender/collada/ArmatureExporter.cpp index 9c26ba83b44..49722873a91 100644 --- a/source/blender/collada/ArmatureExporter.cpp +++ b/source/blender/collada/ArmatureExporter.cpp @@ -156,11 +156,6 @@ void ArmatureExporter::find_objects_using_armature(Object *ob_arm, std::vector<O } #endif -std::string ArmatureExporter::get_joint_sid(Bone *bone, Object *ob_arm) -{ - return get_joint_id(bone, ob_arm); -} - // parent_mat is armature-space void ArmatureExporter::add_bone_node(Bone *bone, Object *ob_arm, Scene *sce, SceneExporter *se, @@ -261,47 +256,57 @@ void ArmatureExporter::add_bone_transform(Object *ob_arm, Bone *bone, COLLADASW: //bPoseChannel *pchan = BKE_pose_channel_find_name(ob_arm->pose, bone->name); float mat[4][4]; + float bone_rest_mat[4][4]; /* derived from bone->arm_mat */ + float parent_rest_mat[4][4]; /* derived from bone->parent->arm_mat */ - if (bone->parent) { - // get bone-space matrix from parent pose - /*bPoseChannel *parchan = BKE_pose_channel_find_name(ob_arm->pose, bone->parent->name); - float invpar[4][4]; - invert_m4_m4(invpar, parchan->pose_mat); - mul_m4_m4m4(mat, invpar, pchan->pose_mat);*/ - - float invpar[4][4]; - invert_m4_m4(invpar, bone->parent->arm_mat); - mul_m4_m4m4(mat, invpar, bone->arm_mat); + bool has_restmat = bc_get_property_matrix(bone, "rest_mat", mat); - } - else { + if (!has_restmat) { + + /* Have no restpose matrix stored, try old style <= Blender 2.78 */ - //copy_m4_m4(mat, pchan->pose_mat); - //pose mat is object space - //New change: export bone->arm_mat - copy_m4_m4(mat, bone->arm_mat); - } + bc_create_restpose_mat(this->export_settings, bone, bone_rest_mat, bone->arm_mat, true); - // OPEN_SIM_COMPATIBILITY - if (export_settings->open_sim) { - // Remove rotations vs armature from transform - // parent_rest_rot * mat * irest_rot - float temp[4][4]; - copy_m4_m4(temp, bone->arm_mat); - temp[3][0] = temp[3][1] = temp[3][2] = 0.0f; - invert_m4(temp); + if (bone->parent) { + // get bone-space matrix from parent pose + /*bPoseChannel *parchan = BKE_pose_channel_find_name(ob_arm->pose, bone->parent->name); + float invpar[4][4]; + invert_m4_m4(invpar, parchan->pose_mat); + mul_m4_m4m4(mat, invpar, pchan->pose_mat);*/ + float invpar[4][4]; + bc_create_restpose_mat(this->export_settings, bone->parent, parent_rest_mat, bone->parent->arm_mat, true); - mul_m4_m4m4(mat, mat, temp); + invert_m4_m4(invpar, parent_rest_mat); + mul_m4_m4m4(mat, invpar, bone_rest_mat); - if (bone->parent) { - copy_m4_m4(temp, bone->parent->arm_mat); + } + else { + copy_m4_m4(mat, bone_rest_mat); + } + + // OPEN_SIM_COMPATIBILITY + if (export_settings->open_sim) { + // Remove rotations vs armature from transform + // parent_rest_rot * mat * irest_rot + float temp[4][4]; + copy_m4_m4(temp, bone_rest_mat); temp[3][0] = temp[3][1] = temp[3][2] = 0.0f; + invert_m4(temp); + + mul_m4_m4m4(mat, mat, temp); - mul_m4_m4m4(mat, temp, mat); + if (bone->parent) { + copy_m4_m4(temp, parent_rest_mat); + temp[3][0] = temp[3][1] = temp[3][2] = 0.0f; + + mul_m4_m4m4(mat, temp, mat); + } } } + bc_sanitize_mat(mat, 6); // XXX: Make this optional ? TransformWriter::add_node_transform(node, mat, NULL); + } std::string ArmatureExporter::get_controller_id(Object *ob_arm, Object *ob) diff --git a/source/blender/collada/ArmatureExporter.h b/source/blender/collada/ArmatureExporter.h index 883a6aca847..d271b505aa9 100644 --- a/source/blender/collada/ArmatureExporter.h +++ b/source/blender/collada/ArmatureExporter.h @@ -83,8 +83,6 @@ private: void find_objects_using_armature(Object *ob_arm, std::vector<Object *>& objects, Scene *sce); #endif - std::string get_joint_sid(Bone *bone, Object *ob_arm); - // Scene, SceneExporter and the list of child_objects // are required for writing bone parented objects void add_bone_node(Bone *bone, Object *ob_arm, Scene *sce, SceneExporter *se, diff --git a/source/blender/collada/ArmatureImporter.cpp b/source/blender/collada/ArmatureImporter.cpp index 17334ca326c..0ea8324ed7c 100644 --- a/source/blender/collada/ArmatureImporter.cpp +++ b/source/blender/collada/ArmatureImporter.cpp @@ -32,6 +32,7 @@ #include "COLLADAFWUniqueId.h" +extern "C" { #include "BKE_action.h" #include "BKE_depsgraph.h" #include "BKE_object.h" @@ -39,7 +40,9 @@ #include "BLI_string.h" #include "BLI_listbase.h" #include "ED_armature.h" +} +#include "collada_utils.h" #include "ArmatureImporter.h" // use node name, or fall back to original id if not present (name is optional) @@ -91,6 +94,7 @@ int ArmatureImporter::create_bone(SkinInfo *skin, COLLADAFW::Node *node, EditBon { float mat[4][4]; float joint_inv_bind_mat[4][4]; + float joint_bind_mat[4][4]; int chain_length = 0; //Checking if bone is already made. @@ -106,7 +110,7 @@ int ArmatureImporter::create_bone(SkinInfo *skin, COLLADAFW::Node *node, EditBon */ std::map<COLLADAFW::UniqueId, SkinInfo>::iterator skin_it; - bool bone_is_not_skinned = true; + bool bone_is_skinned = false; for (skin_it = skin_by_data_uid.begin(); skin_it != skin_by_data_uid.end(); skin_it++) { SkinInfo *b = &skin_it->second; @@ -114,7 +118,7 @@ int ArmatureImporter::create_bone(SkinInfo *skin, COLLADAFW::Node *node, EditBon // get original world-space matrix invert_m4_m4(mat, joint_inv_bind_mat); - + copy_m4_m4(joint_bind_mat, mat); // And make local to armature Object *ob_arm = skin->BKE_armature_from_object(); if (ob_arm) { @@ -123,24 +127,14 @@ int ArmatureImporter::create_bone(SkinInfo *skin, COLLADAFW::Node *node, EditBon mul_m4_m4m4(mat, invmat, mat); } - bone_is_not_skinned = false; + bone_is_skinned = true; break; } } // create a bone even if there's no joint data for it (i.e. it has no influence) - if (bone_is_not_skinned) { - float obmat[4][4]; - // bone-space - get_node_mat(obmat, node, NULL, NULL); - - // get world-space - if (parent) { - mul_m4_m4m4(mat, parent_mat, obmat); - } - else { - copy_m4_m4(mat, obmat); - } + if (!bone_is_skinned) { + get_node_mat(mat, node, NULL, NULL, parent_mat); } if (parent) bone->parent = parent; @@ -156,10 +150,11 @@ int ArmatureImporter::create_bone(SkinInfo *skin, COLLADAFW::Node *node, EditBon int use_connect = be.get_use_connect(); switch (use_connect) { - case 1: bone->flag |= BONE_CONNECTED; - break; - case 0: bone->flag &= ~BONE_CONNECTED; - case -1: break; // not defined + case 1: bone->flag |= BONE_CONNECTED; + break; + case -1:/* Connect type not specified */ + case 0: bone->flag &= ~BONE_CONNECTED; + break; } if (be.has_roll()) { @@ -172,6 +167,15 @@ int ArmatureImporter::create_bone(SkinInfo *skin, COLLADAFW::Node *node, EditBon bone->roll = angle; } copy_v3_v3(bone->head, mat[3]); + + if (bone_is_skinned) + { + float rest_mat[4][4]; + get_node_mat(rest_mat, node, NULL, NULL, NULL); + bc_set_IDPropertyMatrix(bone, "bind_mat", joint_bind_mat); + bc_set_IDPropertyMatrix(bone, "rest_mat", rest_mat); + } + add_v3_v3v3(bone->tail, bone->head, tail); //tail must be non zero /* find smallest bone length in armature (used later for leaf bone length) */ @@ -273,7 +277,6 @@ void ArmatureImporter::fix_parent_connect(bArmature *armature, Bone *bone) } - void ArmatureImporter::connect_bone_chains(bArmature *armature, Bone *parentbone, int clip) { BoneExtensionMap &extended_bones = bone_extension_manager.getExtensionMap(armature); @@ -289,12 +292,13 @@ void ArmatureImporter::connect_bone_chains(bArmature *armature, Bone *parentbone for (; child; child = child->next) { BoneExtended *be = extended_bones[child->name]; if (be != NULL) { - if (be->get_chain_length() <= clip) { - if (be->get_chain_length() > maxlen) { + int chain_len = be->get_chain_length(); + if (chain_len <= clip) { + if (chain_len > maxlen) { dominant_child = be; - maxlen = be->get_chain_length(); + maxlen = chain_len; } - else if (be->get_chain_length() == maxlen) { + else if (chain_len == maxlen) { dominant_child = NULL; } } @@ -308,7 +312,6 @@ void ArmatureImporter::connect_bone_chains(bArmature *armature, Bone *parentbone EditBone *pebone = bc_get_edit_bone(armature, parentbone->name); EditBone *cebone = bc_get_edit_bone(armature, dominant_child->get_name()); if (pebone && !(cebone->flag & BONE_CONNECTED)) { - float vec[3]; sub_v3_v3v3(vec, cebone->head, pebone->head); @@ -321,14 +324,16 @@ void ArmatureImporter::connect_bone_chains(bArmature *armature, Bone *parentbone if (len_squared_v3(vec) > MINIMUM_BONE_LENGTH) { - pebone->tail[0] = cebone->head[0]; - pebone->tail[1] = cebone->head[1]; - pebone->tail[2] = cebone->head[2]; - + copy_v3_v3(pebone->tail, cebone->head); + pbe->set_tail(pebone->tail); /* to make fix_leafbone happy ...*/ if (pbe && pbe->get_chain_length() >= this->import_settings->min_chain_length) { + + BoneExtended *cbe = extended_bones[cebone->name]; + cbe->set_use_connect(true); + cebone->flag |= BONE_CONNECTED; - printf("Connecting chain: parent %s --> %s (child)\n", pebone->name, cebone->name); pbe->set_leaf_bone(false); + printf("Connect Bone chain: parent (%s --> %s) child)\n", pebone->name, cebone->name); } } } diff --git a/source/blender/collada/ControllerExporter.cpp b/source/blender/collada/ControllerExporter.cpp index 06e151c363b..5444c9dfa6b 100644 --- a/source/blender/collada/ControllerExporter.cpp +++ b/source/blender/collada/ControllerExporter.cpp @@ -157,11 +157,6 @@ void ArmatureExporter::find_objects_using_armature(Object *ob_arm, std::vector<O } #endif -std::string ControllerExporter::get_joint_sid(Bone *bone, Object *ob_arm) -{ - return get_joint_id(bone, ob_arm); -} - std::string ControllerExporter::get_controller_id(Object *ob_arm, Object *ob) { return translate_id(id_name(ob_arm)) + "_" + translate_id(id_name(ob)) + SKIN_CONTROLLER_ID_SUFFIX; @@ -468,81 +463,6 @@ std::string ControllerExporter::add_joints_source(Object *ob_arm, ListBase *defb return source_id; } -static float get_property(Bone *bone, const char *key, float def) -{ - float result = def; - if (bone->prop) { - IDProperty *property = IDP_GetPropertyFromGroup(bone->prop, key); - if (property) { - switch (property->type) { - case IDP_INT: - result = (float)(IDP_Int(property)); - break; - case IDP_FLOAT: - result = (float)(IDP_Float(property)); - break; - case IDP_DOUBLE: - result = (float)(IDP_Double(property)); - break; - default: - result = def; - } - } - } - return result; -} - -/** - * This function creates an arbitrary rest pose matrix from - * data provided as custom properties. This is a workaround - * for support of maya's restpose matrix which can be arbitrary - * in opposition to Blender where the Rest pose Matrix is always - * the Identity matrix. - * - * The custom properties are: - * - * restpose_scale_x - * restpose_scale_y - * restpose_scale_z - * - * restpose_rot_x - * restpose_rot_y - * restpose_rot_z - * - * restpose_loc_x - * restpose_loc_y - * restpose_loc_z - * - * The matrix is only setup if the scale AND the rot properties are defined. - * The presence of the loc properties is optional. - * - * This feature has been implemented to support Second Life "Fitted Mesh" - * TODO: Check if an arbitrary rest pose matrix makes sense within Blender. - * Eventually leverage the custom property data into an "official" - * Edit_bone Property - */ -static void create_restpose_mat(Bone *bone, float mat[4][4]) -{ - float loc[3] = { - get_property(bone, "restpose_loc_x", 0.0), - get_property(bone, "restpose_loc_y", 0.0), - get_property(bone, "restpose_loc_z", 0.0) - }; - - float rot[3] = { - DEG2RADF(get_property(bone, "restpose_rot_x", 0.0)), - DEG2RADF(get_property(bone, "restpose_rot_y", 0.0)), - DEG2RADF(get_property(bone, "restpose_rot_z", 0.0)) - }; - - float scale[3] = { - get_property(bone, "restpose_scale_x", 1.0), - get_property(bone, "restpose_scale_y", 1.0), - get_property(bone, "restpose_scale_z", 1.0) - }; - - loc_eulO_size_to_mat4(mat, loc, rot, scale, 6); -} std::string ControllerExporter::add_inv_bind_mats_source(Object *ob_arm, ListBase *defbase, const std::string& controller_id) { @@ -585,30 +505,35 @@ std::string ControllerExporter::add_inv_bind_mats_source(Object *ob_arm, ListBas float world[4][4]; float inv_bind_mat[4][4]; + float bind_mat[4][4]; /* derived from bone->arm_mat */ + + bool has_bindmat = bc_get_property_matrix(pchan->bone, "bind_mat", bind_mat); - // SL/OPEN_SIM COMPATIBILITY - if (export_settings->open_sim) { - // Only translations, no rotation vs armature - float temp[4][4]; - unit_m4(temp); - copy_v3_v3(temp[3], pchan->bone->arm_mat[3]); - mul_m4_m4m4(world, ob_arm->obmat, temp); - - // Add Maya restpose matrix (if defined as properties) - float restpose_mat[4][4]; - create_restpose_mat(pchan->bone, restpose_mat); - mul_m4_m4m4(world, world, restpose_mat); + if (!has_bindmat) { - } - else { - // make world-space matrix, arm_mat is armature-space - mul_m4_m4m4(world, ob_arm->obmat, pchan->bone->arm_mat); + /* Have no bind matrix stored, try old style <= Blender 2.78 */ + + bc_create_restpose_mat(this->export_settings, pchan->bone, bind_mat, pchan->bone->arm_mat, true); + + // SL/OPEN_SIM COMPATIBILITY + if (export_settings->open_sim) { + + float loc[3]; + float rot[3] = { 0, 0, 0 }; + float scale[3]; + bc_decompose(bind_mat, loc, NULL, NULL, scale); + + // Only translations, no rotation vs armature + loc_eulO_size_to_mat4(bind_mat, loc, rot, scale, 6); + } } + // make world-space matrix (bind_mat is armature-space) + mul_m4_m4m4(world, ob_arm->obmat, bind_mat); invert_m4_m4(mat, world); converter.mat4_to_dae(inv_bind_mat, mat); - + bc_sanitize_mat(inv_bind_mat, 6); // XXX: Make this optional ? source.appendValues(inv_bind_mat); } } diff --git a/source/blender/collada/ControllerExporter.h b/source/blender/collada/ControllerExporter.h index 0be51187f6f..80b858ca6dd 100644 --- a/source/blender/collada/ControllerExporter.h +++ b/source/blender/collada/ControllerExporter.h @@ -84,8 +84,6 @@ private: void find_objects_using_armature(Object *ob_arm, std::vector<Object *>& objects, Scene *sce); #endif - std::string get_joint_sid(Bone *bone, Object *ob_arm); - std::string get_controller_id(Object *ob_arm, Object *ob); std::string get_controller_id(Key *key, Object *ob); diff --git a/source/blender/collada/ExportSettings.h b/source/blender/collada/ExportSettings.h index 9451cac9dae..03e380dc198 100644 --- a/source/blender/collada/ExportSettings.h +++ b/source/blender/collada/ExportSettings.h @@ -51,7 +51,9 @@ public: bool use_blender_profile; bool sort_by_name; BC_export_transformation_type export_transformation_type; + bool open_sim; + bool keep_bind_info; char *filepath; LinkNode *export_set; diff --git a/source/blender/collada/ImportSettings.h b/source/blender/collada/ImportSettings.h index 2c52d73e756..4a2d4e8046a 100644 --- a/source/blender/collada/ImportSettings.h +++ b/source/blender/collada/ImportSettings.h @@ -37,6 +37,7 @@ public: bool fix_orientation; int min_chain_length; char *filepath; + bool keep_bind_info; }; #endif diff --git a/source/blender/collada/SkinInfo.cpp b/source/blender/collada/SkinInfo.cpp index 7242a24523c..71875d6274a 100644 --- a/source/blender/collada/SkinInfo.cpp +++ b/source/blender/collada/SkinInfo.cpp @@ -230,7 +230,6 @@ void SkinInfo::link_armature(bContext *C, Object *ob, std::map<COLLADAFW::Unique ModifierData *md = ED_object_modifier_add(NULL, bmain, scene, ob, NULL, eModifierType_Armature); ArmatureModifierData *amd = (ArmatureModifierData *)md; amd->object = ob_arm; - struct bArmature *armature = (bArmature *)ob_arm->data; #if 1 bc_set_parent(ob, ob_arm, C); diff --git a/source/blender/collada/TransformReader.cpp b/source/blender/collada/TransformReader.cpp index f8f31304d28..7f742be7e30 100644 --- a/source/blender/collada/TransformReader.cpp +++ b/source/blender/collada/TransformReader.cpp @@ -34,7 +34,21 @@ TransformReader::TransformReader(UnitConverter *conv) : unit_converter(conv) /* pass */ } -void TransformReader::get_node_mat(float mat[4][4], COLLADAFW::Node *node, std::map<COLLADAFW::UniqueId, Animation> *animation_map, Object *ob) +void TransformReader::get_node_mat( + float mat[4][4], + COLLADAFW::Node *node, + std::map<COLLADAFW::UniqueId, Animation> *animation_map, + Object *ob) +{ + get_node_mat(mat, node, animation_map, ob, NULL); +} + +void TransformReader::get_node_mat( + float mat[4][4], + COLLADAFW::Node *node, + std::map<COLLADAFW::UniqueId, Animation> *animation_map, + Object *ob, + float parent_mat[4][4]) { float cur[4][4]; float copy[4][4]; @@ -52,6 +66,9 @@ void TransformReader::get_node_mat(float mat[4][4], COLLADAFW::Node *node, std:: // then this is considered as redundant information. // So if we find a Matrix we use that and return. dae_matrix_to_mat4(tm, mat); + if (parent_mat) { + mul_m4_m4m4(mat, parent_mat, mat); + } return; case COLLADAFW::Transformation::TRANSLATE: dae_translate_to_mat4(tm, cur); @@ -80,6 +97,10 @@ void TransformReader::get_node_mat(float mat[4][4], COLLADAFW::Node *node, std:: (*animation_map)[anim_list_id] = anim; } } + + if (parent_mat) { + mul_m4_m4m4(mat, parent_mat, mat); + } } void TransformReader::dae_rotate_to_mat4(COLLADAFW::Transformation *tm, float m[4][4]) diff --git a/source/blender/collada/TransformReader.h b/source/blender/collada/TransformReader.h index ab974b9ba85..08bb17ccac1 100644 --- a/source/blender/collada/TransformReader.h +++ b/source/blender/collada/TransformReader.h @@ -43,7 +43,7 @@ //struct Object; -class TransformReader : public TransformBase +class TransformReader { protected: @@ -59,7 +59,8 @@ public: TransformReader(UnitConverter *conv); void get_node_mat(float mat[4][4], COLLADAFW::Node *node, std::map<COLLADAFW::UniqueId, Animation> *animation_map, Object *ob); - + void get_node_mat(float mat[4][4], COLLADAFW::Node *node, std::map<COLLADAFW::UniqueId, Animation> *animation_map, Object *ob, float parent_mat[4][4]); + void dae_rotate_to_mat4(COLLADAFW::Transformation *tm, float m[4][4]); void dae_translate_to_mat4(COLLADAFW::Transformation *tm, float m[4][4]); void dae_scale_to_mat4(COLLADAFW::Transformation *tm, float m[4][4]); diff --git a/source/blender/collada/TransformWriter.cpp b/source/blender/collada/TransformWriter.cpp index 908111ebae6..b7eeff3b074 100644 --- a/source/blender/collada/TransformWriter.cpp +++ b/source/blender/collada/TransformWriter.cpp @@ -27,11 +27,10 @@ #include "BKE_object.h" +#include "BLI_math.h" #include "TransformWriter.h" -#include "BLI_math.h" - void TransformWriter::add_node_transform(COLLADASW::Node& node, float mat[4][4], float parent_mat[4][4]) { float loc[3], rot[3], scale[3]; @@ -51,7 +50,7 @@ void TransformWriter::add_node_transform(COLLADASW::Node& node, float mat[4][4], converter->mat4_to_dae_double(dmat, local); delete converter; - TransformBase::decompose(local, loc, rot, NULL, scale); + bc_decompose(local, loc, rot, NULL, scale); if (node.getType() == COLLADASW::Node::JOINT) { // XXX Why are joints handled differently ? @@ -116,7 +115,7 @@ void TransformWriter::add_node_transform_ob(COLLADASW::Node& node, Object *ob, B case BC_TRANSFORMATION_TYPE_TRANSROTLOC: { float loc[3], rot[3], scale[3]; - TransformBase::decompose(f_obmat, loc, rot, NULL, scale); + bc_decompose(f_obmat, loc, rot, NULL, scale); add_transform(node, loc, rot, scale); break; } diff --git a/source/blender/collada/TransformWriter.h b/source/blender/collada/TransformWriter.h index 7f69a4b9c95..5bb13d4aac9 100644 --- a/source/blender/collada/TransformWriter.h +++ b/source/blender/collada/TransformWriter.h @@ -33,9 +33,10 @@ #include "DNA_object_types.h" #include "collada_internal.h" +#include "collada_utils.h" #include "collada.h" -class TransformWriter : protected TransformBase +class TransformWriter { protected: void add_node_transform(COLLADASW::Node& node, float mat[4][4], float parent_mat[4][4]); diff --git a/source/blender/collada/collada.cpp b/source/blender/collada/collada.cpp index fe8b1d2320a..c0784ca8eb5 100644 --- a/source/blender/collada/collada.cpp +++ b/source/blender/collada/collada.cpp @@ -48,7 +48,8 @@ int collada_import(bContext *C, int find_chains, int auto_connect, int fix_orientation, - int min_chain_length) + int min_chain_length, + int keep_bind_info) { ImportSettings import_settings; @@ -58,6 +59,7 @@ int collada_import(bContext *C, import_settings.find_chains = find_chains != 0; import_settings.fix_orientation = fix_orientation != 0; import_settings.min_chain_length = min_chain_length; + import_settings.keep_bind_info = keep_bind_info; DocumentImporter imp(C, &import_settings); if (imp.import()) return 1; @@ -87,7 +89,9 @@ int collada_export(Scene *sce, int use_blender_profile, int sort_by_name, BC_export_transformation_type export_transformation_type, - int open_sim) + int open_sim, + + int keep_bind_info) { ExportSettings export_settings; @@ -113,6 +117,7 @@ int collada_export(Scene *sce, export_settings.export_transformation_type = export_transformation_type; export_settings.open_sim = open_sim != 0; + export_settings.keep_bind_info = keep_bind_info; int includeFilter = OB_REL_NONE; if (export_settings.include_armatures) includeFilter |= OB_REL_MOD_ARMATURE; diff --git a/source/blender/collada/collada.h b/source/blender/collada/collada.h index a4416608584..09acc4a064b 100644 --- a/source/blender/collada/collada.h +++ b/source/blender/collada/collada.h @@ -58,7 +58,9 @@ int collada_import(struct bContext *C, int find_chains, int auto_connect, int fix_orientation, - int min_chain_length); + int min_chain_length, + + int keep_bind_info); int collada_export(struct Scene *sce, const char *filepath, @@ -81,9 +83,9 @@ int collada_export(struct Scene *sce, int use_blender_profile, int sort_by_name, BC_export_transformation_type export_transformation_type, - int open_sim); - + int open_sim, + int keep_bind_info); #ifdef __cplusplus } diff --git a/source/blender/collada/collada_internal.cpp b/source/blender/collada/collada_internal.cpp index e1a13559b08..6ebde6bd773 100644 --- a/source/blender/collada/collada_internal.cpp +++ b/source/blender/collada/collada_internal.cpp @@ -162,18 +162,6 @@ void UnitConverter::calculate_scale(Scene &sce) size_to_mat4(scale_mat4, rescale); } -void TransformBase::decompose(float mat[4][4], float *loc, float eul[3], float quat[4], float *size) -{ - mat4_to_size(size, mat); - if (eul) { - mat4_to_eul(eul, mat); - } - if (quat) { - mat4_to_quat(quat, mat); - } - copy_v3_v3(loc, mat[3]); -} - /** * Translation map. * Used to translate every COLLADA id to a valid id, no matter what "wrong" letters may be @@ -341,7 +329,12 @@ std::string get_light_id(Object *ob) std::string get_joint_id(Bone *bone, Object *ob_arm) { - return translate_id(/*id_name(ob_arm) + "_" +*/ bone->name); + return translate_id(id_name(ob_arm) + "_" + bone->name); +} + +std::string get_joint_sid(Bone *bone, Object *ob_arm) +{ + return translate_id(bone->name); } std::string get_camera_id(Object *ob) diff --git a/source/blender/collada/collada_internal.h b/source/blender/collada/collada_internal.h index 4aa637a6876..1c7aa160f57 100644 --- a/source/blender/collada/collada_internal.h +++ b/source/blender/collada/collada_internal.h @@ -85,12 +85,6 @@ public: }; -class TransformBase -{ -public: - void decompose(float mat[4][4], float *loc, float eul[3], float quat[4], float *size); -}; - extern void clear_global_id_map(); /** Look at documentation of translate_map */ extern std::string translate_id(const std::string &id); @@ -104,6 +98,7 @@ extern std::string get_geometry_id(Object *ob, bool use_instantiation); extern std::string get_light_id(Object *ob); extern std::string get_joint_id(Bone *bone, Object *ob_arm); +extern std::string get_joint_sid(Bone *bone, Object *ob_arm); extern std::string get_camera_id(Object *ob); diff --git a/source/blender/collada/collada_utils.cpp b/source/blender/collada/collada_utils.cpp index 2efa8b21d81..ac4395e1430 100644 --- a/source/blender/collada/collada_utils.cpp +++ b/source/blender/collada/collada_utils.cpp @@ -32,8 +32,6 @@ #include "COLLADAFWMeshPrimitive.h" #include "COLLADAFWMeshVertexData.h" -#include "collada_utils.h" - extern "C" { #include "DNA_modifier_types.h" #include "DNA_customdata_types.h" @@ -63,6 +61,9 @@ extern "C" { #include "bmesh_tools.h" } +#include "collada_utils.h" +#include "ExportSettings.h" + float bc_get_float_value(const COLLADAFW::FloatOrDoubleArray& array, unsigned int index) { if (index >= array.getValuesCount()) @@ -352,6 +353,28 @@ void bc_match_scale(std::vector<Object *> *objects_done, } } +/* + Convenience function to get only the needed components of a matrix +*/ +void bc_decompose(float mat[4][4], float *loc, float eul[3], float quat[4], float *size) +{ + if (size) { + mat4_to_size(size, mat); + } + + if (eul) { + mat4_to_eul(eul, mat); + } + + if (quat) { + mat4_to_quat(quat, mat); + } + + if (loc) { + copy_v3_v3(loc, mat[3]); + } +} + void bc_triangulate_mesh(Mesh *me) { bool use_beauty = false; @@ -612,3 +635,200 @@ int BoneExtended::get_use_connect() { return this->use_connect; } + +/** +* Stores a 4*4 matrix as a custom bone property array of size 16 +*/ +void bc_set_IDPropertyMatrix(EditBone *ebone, const char *key, float mat[4][4]) +{ + IDProperty *idgroup = (IDProperty *)ebone->prop; + if (idgroup == NULL) + { + IDPropertyTemplate val = { 0 }; + idgroup = IDP_New(IDP_GROUP, &val, "RNA_EditBone ID properties"); + ebone->prop = idgroup; + } + + IDPropertyTemplate val = { 0 }; + val.array.len = 16; + val.array.type = IDP_FLOAT; + + IDProperty *data = IDP_New(IDP_ARRAY, &val, key); + float *array = (float *)IDP_Array(data); + for (int i = 0; i < 4; i++) + for (int j = 0; j < 4; j++) + array[4 * i + j] = mat[i][j]; + + IDP_AddToGroup(idgroup, data); +} + +#if 0 +/** +* Stores a Float value as a custom bone property +* +* Note: This function is currently not needed. Keep for future usage +*/ +static void bc_set_IDProperty(EditBone *ebone, const char *key, float value) +{ + if (ebone->prop == NULL) + { + IDPropertyTemplate val = { 0 }; + ebone->prop = IDP_New(IDP_GROUP, &val, "RNA_EditBone ID properties"); + } + + IDProperty *pgroup = (IDProperty *)ebone->prop; + IDPropertyTemplate val = { 0 }; + IDProperty *prop = IDP_New(IDP_FLOAT, &val, key); + IDP_Float(prop) = value; + IDP_AddToGroup(pgroup, prop); + +} +#endif + +/* +* Get a custom property when it exists. +* This function is also used to check if a property exists. +*/ +IDProperty *bc_get_IDProperty(Bone *bone, std::string key) +{ + return (bone->prop == NULL) ? NULL : IDP_GetPropertyFromGroup(bone->prop, key.c_str()); +} + +/** +* Read a custom bone property and convert to float +* Return def if the property does not exist. +*/ +float bc_get_property(Bone *bone, std::string key, float def) +{ + float result = def; + IDProperty *property = bc_get_IDProperty(bone, key); + if (property) { + switch (property->type) { + case IDP_INT: + result = (float)(IDP_Int(property)); + break; + case IDP_FLOAT: + result = (float)(IDP_Float(property)); + break; + case IDP_DOUBLE: + result = (float)(IDP_Double(property)); + break; + default: + result = def; + } + } + return result; +} + +/** +* Read a custom bone property and convert to matrix +* Return true if conversion was succesfull +* +* Return false if: +* - the property does not exist +* - is not an array of size 16 +*/ +bool bc_get_property_matrix(Bone *bone, std::string key, float mat[4][4]) +{ + IDProperty *property = bc_get_IDProperty(bone, key); + if (property && property->type == IDP_ARRAY && property->len == 16) { + float *array = (float *)IDP_Array(property); + for (int i = 0; i < 4; i++) + for (int j = 0; j < 4; j++) + mat[i][j] = array[4 * i + j]; + return true; + } + return false; +} + +/** +* get a vector that is stored in 3 custom properties (used in Blender <= 2.78) +*/ +void bc_get_property_vector(Bone *bone, std::string key, float val[3], const float def[3]) +{ + val[0] = bc_get_property(bone, key + "_x", def[0]); + val[1] = bc_get_property(bone, key + "_y", def[1]); + val[2] = bc_get_property(bone, key + "_z", def[2]); +} + +/** +* Check if vector exist stored in 3 custom properties (used in Blender <= 2.78) +*/ +static bool has_custom_props(Bone *bone, bool enabled, std::string key) +{ + if (!enabled) + return false; + + return (bc_get_IDProperty(bone, key + "_x") + || bc_get_IDProperty(bone, key + "_y") + || bc_get_IDProperty(bone, key + "_z")); + +} + +/** +* Check if custom information about bind matrix exists and modify the from_mat +* accordingly. +* +* Note: This is old style for Blender <= 2.78 only kept for compatibility +*/ +void bc_create_restpose_mat(const ExportSettings *export_settings, Bone *bone, float to_mat[4][4], float from_mat[4][4], bool use_local_space) +{ + float loc[3]; + float rot[3]; + float scale[3]; + static const float V0[3] = { 0, 0, 0 }; + + if (!has_custom_props(bone, export_settings->keep_bind_info, "restpose_loc") && + !has_custom_props(bone, export_settings->keep_bind_info, "restpose_rot") && + !has_custom_props(bone, export_settings->keep_bind_info, "restpose_scale")) + { + /* No need */ + copy_m4_m4(to_mat, from_mat); + return; + } + + bc_decompose(from_mat, loc, rot, NULL, scale); + loc_eulO_size_to_mat4(to_mat, loc, rot, scale, 6); + + if (export_settings->keep_bind_info) { + bc_get_property_vector(bone, "restpose_loc", loc, loc); + + if (use_local_space && bone->parent) { + Bone *b = bone; + while (b->parent) { + b = b->parent; + float ploc[3]; + bc_get_property_vector(b, "restpose_loc", ploc, V0); + loc[0] += ploc[0]; + loc[1] += ploc[1]; + loc[2] += ploc[2]; + } + } + } + + if (export_settings->keep_bind_info) { + if (bc_get_IDProperty(bone, "restpose_rot_x")) + rot[0] = DEG2RADF(bc_get_property(bone, "restpose_rot_x", 0)); + if (bc_get_IDProperty(bone, "restpose_rot_y")) + rot[1] = DEG2RADF(bc_get_property(bone, "restpose_rot_y", 0)); + if (bc_get_IDProperty(bone, "restpose_rot_z")) + rot[2] = DEG2RADF(bc_get_property(bone, "restpose_rot_z", 0)); + } + + if (export_settings->keep_bind_info) { + bc_get_property_vector(bone, "restpose_scale", scale, scale); + } + + loc_eulO_size_to_mat4(to_mat, loc, rot, scale, 6); + +} + +/* + Make 4*4 matrices better readable +*/ +void bc_sanitize_mat(float mat[4][4], int precision) +{ + for (int i = 0; i < 4; i++) + for (int j = 0; j < 4; j++) + mat[i][j] = double_round(mat[i][j], precision); +}
\ No newline at end of file diff --git a/source/blender/collada/collada_utils.h b/source/blender/collada/collada_utils.h index 7fdbef3b6cb..38c0bd5096a 100644 --- a/source/blender/collada/collada_utils.h +++ b/source/blender/collada/collada_utils.h @@ -53,8 +53,10 @@ extern "C" { #include "BKE_object.h" #include "BKE_DerivedMesh.h" #include "BKE_scene.h" +#include "BKE_idprop.h" } +#include "ImportSettings.h" #include "ExportSettings.h" #include "collada_internal.h" @@ -88,11 +90,24 @@ extern std::string bc_url_encode(std::string data); extern void bc_match_scale(Object *ob, UnitConverter &bc_unit, bool scale_to_scene); extern void bc_match_scale(std::vector<Object *> *objects_done, UnitConverter &unit_converter, bool scale_to_scene); +extern void bc_decompose(float mat[4][4], float *loc, float eul[3], float quat[4], float *size); + extern void bc_triangulate_mesh(Mesh *me); extern bool bc_is_leaf_bone(Bone *bone); extern EditBone *bc_get_edit_bone(bArmature * armature, char *name); extern int bc_set_layer(int bitfield, int layer, bool enable); extern int bc_set_layer(int bitfield, int layer); +extern void bc_sanitize_mat(float mat[4][4], int precision); + +extern IDProperty *bc_get_IDProperty(Bone *bone, std::string key); +extern void bc_set_IDProperty(EditBone *ebone, const char *key, float value); +extern void bc_set_IDPropertyMatrix(EditBone *ebone, const char *key, float mat[4][4]); + +extern float bc_get_property(Bone *bone, std::string key, float def); +extern void bc_get_property_vector(Bone *bone, std::string key, float val[3], const float def[3]); +extern bool bc_get_property_matrix(Bone *bone, std::string key, float mat[4][4]); + +extern void bc_create_restpose_mat(const ExportSettings *export_settings, Bone *bone, float to_mat[4][4], float world[4][4], bool use_local_space); class BCPolygonNormalsIndices { diff --git a/source/blender/compositor/operations/COM_ConvolutionEdgeFilterOperation.cpp b/source/blender/compositor/operations/COM_ConvolutionEdgeFilterOperation.cpp index e1ada9a8c39..5f78067220a 100644 --- a/source/blender/compositor/operations/COM_ConvolutionEdgeFilterOperation.cpp +++ b/source/blender/compositor/operations/COM_ConvolutionEdgeFilterOperation.cpp @@ -94,4 +94,10 @@ void ConvolutionEdgeFilterOperation::executePixel(float output[4], int x, int y, output[2] = output[2] * value[0] + in2[2] * mval; output[3] = in2[3]; + + /* Make sure we don't return negative color. */ + output[0] = max(output[0], 0.0f); + output[1] = max(output[1], 0.0f); + output[2] = max(output[2], 0.0f); + output[3] = max(output[3], 0.0f); } diff --git a/source/blender/compositor/operations/COM_ConvolutionFilterOperation.cpp b/source/blender/compositor/operations/COM_ConvolutionFilterOperation.cpp index 68ec2be5ebd..6ac1ff9a1eb 100644 --- a/source/blender/compositor/operations/COM_ConvolutionFilterOperation.cpp +++ b/source/blender/compositor/operations/COM_ConvolutionFilterOperation.cpp @@ -107,6 +107,12 @@ void ConvolutionFilterOperation::executePixel(float output[4], int x, int y, voi output[1] = output[1] * value[0] + in2[1] * mval; output[2] = output[2] * value[0] + in2[2] * mval; output[3] = output[3] * value[0] + in2[3] * mval; + + /* Make sure we don't return negative color. */ + output[0] = max(output[0], 0.0f); + output[1] = max(output[1], 0.0f); + output[2] = max(output[2], 0.0f); + output[3] = max(output[3], 0.0f); } bool ConvolutionFilterOperation::determineDependingAreaOfInterest(rcti *input, ReadBufferOperation *readOperation, rcti *output) diff --git a/source/blender/compositor/operations/COM_GlareSimpleStarOperation.cpp b/source/blender/compositor/operations/COM_GlareSimpleStarOperation.cpp index 957ac5af748..57aa3a1bac2 100644 --- a/source/blender/compositor/operations/COM_GlareSimpleStarOperation.cpp +++ b/source/blender/compositor/operations/COM_GlareSimpleStarOperation.cpp @@ -44,18 +44,18 @@ void GlareSimpleStarOperation::generateGlare(float *data, MemoryBuffer *inputTil xp = x + i; tbuf1->read(c, x, y); mul_v3_fl(c, f1); - tbuf1->read(tc, (settings->angle ? xm : x), ym); + tbuf1->read(tc, (settings->star_45 ? xm : x), ym); madd_v3_v3fl(c, tc, f2); - tbuf1->read(tc, (settings->angle ? xp : x), yp); + tbuf1->read(tc, (settings->star_45 ? xp : x), yp); madd_v3_v3fl(c, tc, f2); c[3] = 1.0f; tbuf1->writePixel(x, y, c); tbuf2->read(c, x, y); mul_v3_fl(c, f1); - tbuf2->read(tc, xm, (settings->angle ? yp : y)); + tbuf2->read(tc, xm, (settings->star_45 ? yp : y)); madd_v3_v3fl(c, tc, f2); - tbuf2->read(tc, xp, (settings->angle ? ym : y)); + tbuf2->read(tc, xp, (settings->star_45 ? ym : y)); madd_v3_v3fl(c, tc, f2); c[3] = 1.0f; tbuf2->writePixel(x, y, c); @@ -73,18 +73,18 @@ void GlareSimpleStarOperation::generateGlare(float *data, MemoryBuffer *inputTil xp = x + i; tbuf1->read(c, x, y); mul_v3_fl(c, f1); - tbuf1->read(tc, (settings->angle ? xm : x), ym); + tbuf1->read(tc, (settings->star_45 ? xm : x), ym); madd_v3_v3fl(c, tc, f2); - tbuf1->read(tc, (settings->angle ? xp : x), yp); + tbuf1->read(tc, (settings->star_45 ? xp : x), yp); madd_v3_v3fl(c, tc, f2); c[3] = 1.0f; tbuf1->writePixel(x, y, c); tbuf2->read(c, x, y); mul_v3_fl(c, f1); - tbuf2->read(tc, xm, (settings->angle ? yp : y)); + tbuf2->read(tc, xm, (settings->star_45 ? yp : y)); madd_v3_v3fl(c, tc, f2); - tbuf2->read(tc, xp, (settings->angle ? ym : y)); + tbuf2->read(tc, xp, (settings->star_45 ? ym : y)); madd_v3_v3fl(c, tc, f2); c[3] = 1.0f; tbuf2->writePixel(x, y, c); diff --git a/source/blender/compositor/operations/COM_GlareStreaksOperation.cpp b/source/blender/compositor/operations/COM_GlareStreaksOperation.cpp index da6076337b4..535f2952e5d 100644 --- a/source/blender/compositor/operations/COM_GlareStreaksOperation.cpp +++ b/source/blender/compositor/operations/COM_GlareStreaksOperation.cpp @@ -28,7 +28,7 @@ void GlareStreaksOperation::generateGlare(float *data, MemoryBuffer *inputTile, int x, y, n; unsigned int nump = 0; float c1[4], c2[4], c3[4], c4[4]; - float a, ang = DEG2RADF(360.0f) / (float)settings->angle; + float a, ang = DEG2RADF(360.0f) / (float)settings->streaks; int size = inputTile->getWidth() * inputTile->getHeight(); int size4 = size * 4; diff --git a/source/blender/depsgraph/intern/builder/deg_builder.cc b/source/blender/depsgraph/intern/builder/deg_builder.cc index cb2f057a090..828da6cb056 100644 --- a/source/blender/depsgraph/intern/builder/deg_builder.cc +++ b/source/blender/depsgraph/intern/builder/deg_builder.cc @@ -77,28 +77,8 @@ static bool check_object_needs_evaluation(Object *object) return false; } -void deg_graph_build_finalize(Depsgraph *graph) +void deg_graph_build_flush_layers(Depsgraph *graph) { - /* STEP 1: Make sure new invisible dependencies are ready for use. - * - * TODO(sergey): This might do a bit of extra tagging, but it's kinda nice - * to do it ahead of a time and don't spend time on flushing updates on - * every frame change. - */ - GHASH_FOREACH_BEGIN(IDDepsNode *, id_node, graph->id_hash) - { - if (id_node->layers == 0) { - ID *id = id_node->id; - if (GS(id->name) == ID_OB) { - Object *object = (Object *)id; - if (check_object_needs_evaluation(object)) { - id_node->tag_update(graph); - } - } - } - } - GHASH_FOREACH_END(); - /* STEP 2: Flush visibility layers from children to parent. */ std::stack<OperationDepsNode *> stack; foreach (OperationDepsNode *node, graph->operations) { IDDepsNode *id_node = node->owner->owner; @@ -143,6 +123,31 @@ void deg_graph_build_finalize(Depsgraph *graph) } } } +} + +void deg_graph_build_finalize(Depsgraph *graph) +{ + /* STEP 1: Make sure new invisible dependencies are ready for use. + * + * TODO(sergey): This might do a bit of extra tagging, but it's kinda nice + * to do it ahead of a time and don't spend time on flushing updates on + * every frame change. + */ + GHASH_FOREACH_BEGIN(IDDepsNode *, id_node, graph->id_hash) + { + if (id_node->layers == 0) { + ID *id = id_node->id; + if (GS(id->name) == ID_OB) { + Object *object = (Object *)id; + if (check_object_needs_evaluation(object)) { + id_node->tag_update(graph); + } + } + } + } + GHASH_FOREACH_END(); + /* STEP 2: Flush visibility layers from children to parent. */ + deg_graph_build_flush_layers(graph); /* STEP 3: Re-tag IDs for update if it was tagged before the relations * update tag. */ @@ -154,7 +159,7 @@ void deg_graph_build_finalize(Depsgraph *graph) } GHASH_FOREACH_END(); - if ((id_node->layers & graph->layers) != 0) { + if ((id_node->layers & graph->layers) != 0 || graph->layers == 0) { ID *id = id_node->id; if ((id->tag & LIB_TAG_ID_RECALC_ALL) && (id->tag & LIB_TAG_DOIT)) diff --git a/source/blender/depsgraph/intern/builder/deg_builder.h b/source/blender/depsgraph/intern/builder/deg_builder.h index bdc030e3810..3cc51a2d7db 100644 --- a/source/blender/depsgraph/intern/builder/deg_builder.h +++ b/source/blender/depsgraph/intern/builder/deg_builder.h @@ -42,5 +42,6 @@ struct Depsgraph; string deg_fcurve_id_name(const FCurve *fcu); void deg_graph_build_finalize(struct Depsgraph *graph); +void deg_graph_build_flush_layers(struct Depsgraph *graph); } // namespace DEG diff --git a/source/blender/depsgraph/intern/builder/deg_builder_nodes.cc b/source/blender/depsgraph/intern/builder/deg_builder_nodes.cc index 6b7fb5246ca..7bcaee61fe0 100644 --- a/source/blender/depsgraph/intern/builder/deg_builder_nodes.cc +++ b/source/blender/depsgraph/intern/builder/deg_builder_nodes.cc @@ -432,7 +432,7 @@ void DepsgraphNodeBuilder::build_object(Scene *scene, Base *base, Object *ob) } /* Object data. */ - if (ob->data) { + if (ob->data != NULL) { /* type-specific data... */ switch (ob->type) { case OB_MESH: /* Geometry */ @@ -490,14 +490,25 @@ void DepsgraphNodeBuilder::build_object(Scene *scene, Base *base, Object *ob) build_animdata(&ob->id); /* particle systems */ - if (ob->particlesystem.first) { + if (ob->particlesystem.first != NULL) { build_particles(scene, ob); } - /* grease pencil */ - if (ob->gpd) { + /* Grease pencil. */ + if (ob->gpd != NULL) { build_gpencil(ob->gpd); } + + /* Object that this is a proxy for. */ + if (ob->proxy) { + ob->proxy->proxy_from = ob; + build_object(scene, base, ob->proxy); + } + + /* Object dupligroup. */ + if (ob->dup_group != NULL) { + build_group(scene, base, ob->dup_group); + } } void DepsgraphNodeBuilder::build_object_transform(Scene *scene, Object *ob) @@ -734,7 +745,8 @@ void DepsgraphNodeBuilder::build_particles(Scene *scene, Object *ob) */ /* component for all particle systems */ - ComponentDepsNode *psys_comp = add_component_node(&ob->id, DEPSNODE_TYPE_EVAL_PARTICLES); + ComponentDepsNode *psys_comp = + add_component_node(&ob->id, DEPSNODE_TYPE_EVAL_PARTICLES); /* particle systems */ LINKLIST_FOREACH (ParticleSystem *, psys, &ob->particlesystem) { @@ -747,11 +759,12 @@ void DepsgraphNodeBuilder::build_particles(Scene *scene, Object *ob) /* this particle system */ // TODO: for now, this will just be a placeholder "ubereval" node add_operation_node(psys_comp, - DEPSOP_TYPE_EXEC, function_bind(BKE_particle_system_eval, - _1, - scene, - ob, - psys), + DEPSOP_TYPE_EXEC, + function_bind(BKE_particle_system_eval, + _1, + scene, + ob, + psys), DEG_OPCODE_PSYS_EVAL, psys->name); } @@ -760,6 +773,20 @@ void DepsgraphNodeBuilder::build_particles(Scene *scene, Object *ob) // TODO... } +void DepsgraphNodeBuilder::build_cloth(Scene *scene, Object *object) +{ + ComponentDepsNode *cache_comp = add_component_node(&object->id, + DEPSNODE_TYPE_CACHE); + add_operation_node(cache_comp, + DEPSOP_TYPE_EXEC, + function_bind(BKE_object_eval_cloth, + _1, + scene, + object), + DEG_OPCODE_PLACEHOLDER, + "Cloth Modifier"); +} + /* Shapekeys */ void DepsgraphNodeBuilder::build_shapekeys(Key *key) { @@ -821,6 +848,9 @@ void DepsgraphNodeBuilder::build_obdata_geom(Scene *scene, Object *ob) md), DEG_OPCODE_GEOMETRY_MODIFIER, md->name); + if (md->type == eModifierType_Cloth) { + build_cloth(scene, ob); + } } /* materials */ diff --git a/source/blender/depsgraph/intern/builder/deg_builder_nodes.h b/source/blender/depsgraph/intern/builder/deg_builder_nodes.h index c5035f35f6e..10b586342dd 100644 --- a/source/blender/depsgraph/intern/builder/deg_builder_nodes.h +++ b/source/blender/depsgraph/intern/builder/deg_builder_nodes.h @@ -133,6 +133,7 @@ struct DepsgraphNodeBuilder { void build_pose_constraints(Object *ob, bPoseChannel *pchan); void build_rigidbody(Scene *scene); void build_particles(Scene *scene, Object *ob); + void build_cloth(Scene *scene, Object *object); void build_animdata(ID *id); OperationDepsNode *build_driver(ID *id, FCurve *fcurve); void build_ik_pose(Scene *scene, diff --git a/source/blender/depsgraph/intern/builder/deg_builder_nodes_scene.cc b/source/blender/depsgraph/intern/builder/deg_builder_nodes_scene.cc index 99e61692231..7dd694cb570 100644 --- a/source/blender/depsgraph/intern/builder/deg_builder_nodes_scene.cc +++ b/source/blender/depsgraph/intern/builder/deg_builder_nodes_scene.cc @@ -81,21 +81,7 @@ void DepsgraphNodeBuilder::build_scene(Main *bmain, Scene *scene) /* scene objects */ LINKLIST_FOREACH (Base *, base, &scene->base) { Object *ob = base->object; - - /* object itself */ build_object(scene, base, ob); - - /* object that this is a proxy for */ - // XXX: the way that proxies work needs to be completely reviewed! - if (ob->proxy) { - ob->proxy->proxy_from = ob; - build_object(scene, base, ob->proxy); - } - - /* Object dupligroup. */ - if (ob->dup_group) { - build_group(scene, base, ob->dup_group); - } } /* rigidbody */ diff --git a/source/blender/depsgraph/intern/builder/deg_builder_relations.cc b/source/blender/depsgraph/intern/builder/deg_builder_relations.cc index 94cff521edc..7f2c6d8d9a1 100644 --- a/source/blender/depsgraph/intern/builder/deg_builder_relations.cc +++ b/source/blender/depsgraph/intern/builder/deg_builder_relations.cc @@ -396,30 +396,52 @@ void DepsgraphRelationBuilder::build_object(Main *bmain, Scene *scene, Object *o OperationKey ob_ubereval_key(&ob->id, DEPSNODE_TYPE_TRANSFORM, DEG_OPCODE_OBJECT_UBEREVAL); /* parenting */ - if (ob->parent) { + if (ob->parent != NULL) { /* parent relationship */ build_object_parent(ob); /* local -> parent */ - add_relation(local_transform_key, parent_transform_key, DEPSREL_TYPE_COMPONENT_ORDER, "[ObLocal -> ObParent]"); + add_relation(local_transform_key, + parent_transform_key, + DEPSREL_TYPE_COMPONENT_ORDER, + "[ObLocal -> ObParent]"); } /* object constraints */ - if (ob->constraints.first) { - OperationKey constraint_key(&ob->id, DEPSNODE_TYPE_TRANSFORM, DEG_OPCODE_TRANSFORM_CONSTRAINTS); + if (ob->constraints.first != NULL) { + OperationKey constraint_key(&ob->id, + DEPSNODE_TYPE_TRANSFORM, + DEG_OPCODE_TRANSFORM_CONSTRAINTS); /* constraint relations */ // TODO: provide base op // XXX: this is broken - build_constraints(scene, &ob->id, DEPSNODE_TYPE_TRANSFORM, "", &ob->constraints, NULL); + build_constraints(scene, + &ob->id, + DEPSNODE_TYPE_TRANSFORM, + "", + &ob->constraints, + NULL); /* operation order */ - add_relation(base_op_key, constraint_key, DEPSREL_TYPE_COMPONENT_ORDER, "[ObBase-> Constraint Stack]"); - add_relation(constraint_key, final_transform_key, DEPSREL_TYPE_COMPONENT_ORDER, "[ObConstraints -> Done]"); + add_relation(base_op_key, + constraint_key, + DEPSREL_TYPE_COMPONENT_ORDER, + "[ObBase-> Constraint Stack]"); + add_relation(constraint_key, + final_transform_key, + DEPSREL_TYPE_COMPONENT_ORDER, + "[ObConstraints -> Done]"); // XXX - add_relation(constraint_key, ob_ubereval_key, DEPSREL_TYPE_COMPONENT_ORDER, "Temp Ubereval"); - add_relation(ob_ubereval_key, final_transform_key, DEPSREL_TYPE_COMPONENT_ORDER, "Temp Ubereval"); + add_relation(constraint_key, + ob_ubereval_key, + DEPSREL_TYPE_COMPONENT_ORDER, + "Temp Ubereval"); + add_relation(ob_ubereval_key, + final_transform_key, + DEPSREL_TYPE_COMPONENT_ORDER, + "Temp Ubereval"); } else { /* NOTE: Keep an eye here, we skip some relations here to "streamline" @@ -448,7 +470,10 @@ void DepsgraphRelationBuilder::build_object(Main *bmain, Scene *scene, Object *o // XXX: This should be hooked up by the build_animdata code if (needs_animdata_node(&ob->id)) { ComponentKey adt_key(&ob->id, DEPSNODE_TYPE_ANIMATION); - add_relation(adt_key, local_transform_key, DEPSREL_TYPE_OPERATION, "Object Animation"); + add_relation(adt_key, + local_transform_key, + DEPSREL_TYPE_OPERATION, + "Object Animation"); } @@ -494,25 +519,48 @@ void DepsgraphRelationBuilder::build_object(Main *bmain, Scene *scene, Object *o if (key != NULL) { ComponentKey geometry_key((ID *)ob->data, DEPSNODE_TYPE_GEOMETRY); ComponentKey key_key(&key->id, DEPSNODE_TYPE_GEOMETRY); - add_relation(key_key, geometry_key, DEPSREL_TYPE_GEOMETRY_EVAL, "Shapekeys"); + add_relation(key_key, + geometry_key, + DEPSREL_TYPE_GEOMETRY_EVAL, + "Shapekeys"); } } - /* particle systems */ - if (ob->particlesystem.first) { + /* Particle systems. */ + if (ob->particlesystem.first != NULL) { build_particles(scene, ob); } - /* grease pencil */ - if (ob->gpd) { + /* Grease pencil. */ + if (ob->gpd != NULL) { build_gpencil(ob->gpd); } + + /* Object that this is a proxy for. */ + if (ob->proxy != NULL) { + ob->proxy->proxy_from = ob; + build_object(bmain, scene, ob->proxy); + /* TODO(sergey): This is an inverted relation, matches old depsgraph + * behavior and need to be investigated if it still need to be inverted. + */ + ComponentKey ob_pose_key(&ob->id, DEPSNODE_TYPE_EVAL_POSE); + ComponentKey proxy_pose_key(&ob->proxy->id, DEPSNODE_TYPE_EVAL_POSE); + add_relation(ob_pose_key, proxy_pose_key, DEPSREL_TYPE_TRANSFORM, "Proxy"); + } + + /* Object dupligroup. */ + if (ob->dup_group != NULL) { + build_group(bmain, scene, ob, ob->dup_group); + } } void DepsgraphRelationBuilder::build_object_parent(Object *ob) { - /* XXX: for now, need to use the component key (not just direct to the parent op), or else the matrix doesn't get reset */ - // XXX: @sergey - it would be good if we got that backwards flushing working when tagging for updates + /* XXX: for now, need to use the component key (not just direct to the parent op), + * or else the matrix doesn't get reset/ + */ + // XXX: @sergey - it would be good if we got that backwards flushing working + // when tagging for updates. //OperationKey ob_key(&ob->id, DEPSNODE_TYPE_TRANSFORM, DEG_OPCODE_TRANSFORM_PARENT); ComponentKey ob_key(&ob->id, DEPSNODE_TYPE_TRANSFORM); @@ -1312,6 +1360,25 @@ void DepsgraphRelationBuilder::build_particles(Scene *scene, Object *ob) // TODO... } +void DepsgraphRelationBuilder::build_cloth(Scene * /*scene*/, + Object *object, + ModifierData *md) +{ + OperationKey cache_key(&object->id, + DEPSNODE_TYPE_CACHE, + DEG_OPCODE_PLACEHOLDER, + "Cloth Modifier"); + /* Cache component affects on modifier. */ + OperationKey modifier_key(&object->id, + DEPSNODE_TYPE_GEOMETRY, + DEG_OPCODE_GEOMETRY_MODIFIER, + md->name); + add_relation(cache_key, + modifier_key, + DEPSREL_TYPE_TIME, + "Cloth Cache -> Cloth"); +} + /* Shapekeys */ void DepsgraphRelationBuilder::build_shapekeys(ID *obdata, Key *key) { @@ -1413,6 +1480,10 @@ void DepsgraphRelationBuilder::build_obdata_geom(Main *bmain, Scene *scene, Obje } } + if (md->type == eModifierType_Cloth) { + build_cloth(scene, ob, md); + } + prev_mod_key = mod_key; } } diff --git a/source/blender/depsgraph/intern/builder/deg_builder_relations.h b/source/blender/depsgraph/intern/builder/deg_builder_relations.h index 054e4103290..4ca95bebe3f 100644 --- a/source/blender/depsgraph/intern/builder/deg_builder_relations.h +++ b/source/blender/depsgraph/intern/builder/deg_builder_relations.h @@ -58,6 +58,7 @@ struct Main; struct Mask; struct Material; struct MTex; +struct ModifierData; struct MovieClip; struct bNodeTree; struct Object; @@ -205,6 +206,7 @@ struct DepsgraphRelationBuilder void build_world(World *world); void build_rigidbody(Scene *scene); void build_particles(Scene *scene, Object *ob); + void build_cloth(Scene *scene, Object *object, ModifierData *md); void build_ik_pose(Object *ob, bPoseChannel *pchan, bConstraint *con, diff --git a/source/blender/depsgraph/intern/builder/deg_builder_relations_scene.cc b/source/blender/depsgraph/intern/builder/deg_builder_relations_scene.cc index 8a3476cff45..3bf435c37e0 100644 --- a/source/blender/depsgraph/intern/builder/deg_builder_relations_scene.cc +++ b/source/blender/depsgraph/intern/builder/deg_builder_relations_scene.cc @@ -76,26 +76,7 @@ void DepsgraphRelationBuilder::build_scene(Main *bmain, Scene *scene) /* scene objects */ LINKLIST_FOREACH (Base *, base, &scene->base) { Object *ob = base->object; - - /* object itself */ build_object(bmain, scene, ob); - - /* object that this is a proxy for */ - if (ob->proxy) { - ob->proxy->proxy_from = ob; - build_object(bmain, scene, ob->proxy); - /* TODO(sergey): This is an inverted relation, matches old depsgraph - * behavior and need to be investigated if it still need to be inverted. - */ - ComponentKey ob_pose_key(&ob->id, DEPSNODE_TYPE_EVAL_POSE); - ComponentKey proxy_pose_key(&ob->proxy->id, DEPSNODE_TYPE_EVAL_POSE); - add_relation(ob_pose_key, proxy_pose_key, DEPSREL_TYPE_TRANSFORM, "Proxy"); - } - - /* Object dupligroup. */ - if (ob->dup_group) { - build_group(bmain, scene, ob, ob->dup_group); - } } /* rigidbody */ diff --git a/source/blender/depsgraph/intern/depsgraph_tag.cc b/source/blender/depsgraph/intern/depsgraph_tag.cc index e8ed03666a6..cfc9005a1e3 100644 --- a/source/blender/depsgraph/intern/depsgraph_tag.cc +++ b/source/blender/depsgraph/intern/depsgraph_tag.cc @@ -56,8 +56,8 @@ extern "C" { #include "DEG_depsgraph.h" } /* extern "C" */ +#include "intern/builder/deg_builder.h" #include "intern/eval/deg_eval_flush.h" - #include "intern/nodes/deg_node.h" #include "intern/nodes/deg_node_component.h" #include "intern/nodes/deg_node_operation.h" @@ -346,6 +346,27 @@ void DEG_graph_on_visible_update(Main *bmain, Scene *scene) GHASH_FOREACH_END(); } scene->lay_updated |= graph->layers; + /* Special trick to get local view to work. */ + LINKLIST_FOREACH (Base *, base, &scene->base) { + Object *object = base->object; + DEG::IDDepsNode *id_node = graph->find_id_node(&object->id); + id_node->layers = 0; + } + LINKLIST_FOREACH (Base *, base, &scene->base) { + Object *object = base->object; + DEG::IDDepsNode *id_node = graph->find_id_node(&object->id); + id_node->layers |= base->lay; + } + DEG::deg_graph_build_flush_layers(graph); + LINKLIST_FOREACH (Base *, base, &scene->base) { + Object *object = base->object; + DEG::IDDepsNode *id_node = graph->find_id_node(&object->id); + GHASH_FOREACH_BEGIN(DEG::ComponentDepsNode *, comp, id_node->components) + { + id_node->layers |= comp->layers; + } + GHASH_FOREACH_END(); + } } void DEG_on_visible_update(Main *bmain, const bool UNUSED(do_time)) diff --git a/source/blender/depsgraph/intern/depsgraph_types.h b/source/blender/depsgraph/intern/depsgraph_types.h index effd34a0eb9..c9c4329769d 100644 --- a/source/blender/depsgraph/intern/depsgraph_types.h +++ b/source/blender/depsgraph/intern/depsgraph_types.h @@ -44,9 +44,6 @@ #include <string> #include <vector> -using std::string; -using std::vector; - struct bAction; struct ChannelDriver; struct ModifierData; @@ -56,6 +53,9 @@ struct FCurve; namespace DEG { +using std::string; +using std::vector; + /* Evaluation Operation for atomic operation */ // XXX: move this to another header that can be exposed? typedef function<void(struct EvaluationContext *)> DepsEvalOperationCb; @@ -89,52 +89,52 @@ typedef enum eDepsNode_Type { /* **** Generic Types **** */ /* "Current Scene" - basically whatever kicks off the evaluation process. */ - DEPSNODE_TYPE_ROOT = 1, + DEPSNODE_TYPE_ROOT, /* Time-Source */ - DEPSNODE_TYPE_TIMESOURCE = 2, + DEPSNODE_TYPE_TIMESOURCE, /* ID-Block reference - used as landmarks/collection point for components, * but not usually part of main graph. */ - DEPSNODE_TYPE_ID_REF = 3, + DEPSNODE_TYPE_ID_REF, /* Isolated sub-graph - used for keeping instanced data separate from * instances using them. */ - DEPSNODE_TYPE_SUBGRAPH = 4, + DEPSNODE_TYPE_SUBGRAPH, /* **** Outer Types **** */ /* Parameters Component - Default when nothing else fits * (i.e. just SDNA property setting). */ - DEPSNODE_TYPE_PARAMETERS = 11, + DEPSNODE_TYPE_PARAMETERS, /* Generic "Proxy-Inherit" Component * XXX: Also for instancing of subgraphs? */ - DEPSNODE_TYPE_PROXY = 12, + DEPSNODE_TYPE_PROXY, /* Animation Component * * XXX: merge in with parameters? */ - DEPSNODE_TYPE_ANIMATION = 13, + DEPSNODE_TYPE_ANIMATION, /* Transform Component (Parenting/Constraints) */ - DEPSNODE_TYPE_TRANSFORM = 14, + DEPSNODE_TYPE_TRANSFORM, /* Geometry Component (DerivedMesh/Displist) */ - DEPSNODE_TYPE_GEOMETRY = 15, + DEPSNODE_TYPE_GEOMETRY, /* Sequencer Component (Scene Only) */ - DEPSNODE_TYPE_SEQUENCER = 16, + DEPSNODE_TYPE_SEQUENCER, /* **** Evaluation-Related Outer Types (with Subdata) **** */ /* Pose Component - Owner/Container of Bones Eval */ - DEPSNODE_TYPE_EVAL_POSE = 21, + DEPSNODE_TYPE_EVAL_POSE, /* Bone Component - Child/Subcomponent of Pose */ - DEPSNODE_TYPE_BONE = 22, + DEPSNODE_TYPE_BONE, /* Particle Systems Component */ - DEPSNODE_TYPE_EVAL_PARTICLES = 23, + DEPSNODE_TYPE_EVAL_PARTICLES, /* Material Shading Component */ - DEPSNODE_TYPE_SHADING = 24, + DEPSNODE_TYPE_SHADING, /* Cache Component */ - DEPSNODE_TYPE_CACHE = 25, + DEPSNODE_TYPE_CACHE, } eDepsNode_Type; /* Identifiers for common operations (as an enum). */ diff --git a/source/blender/depsgraph/intern/eval/deg_eval.cc b/source/blender/depsgraph/intern/eval/deg_eval.cc index 3a042535d26..e739bc9dbb5 100644 --- a/source/blender/depsgraph/intern/eval/deg_eval.cc +++ b/source/blender/depsgraph/intern/eval/deg_eval.cc @@ -95,105 +95,38 @@ static void deg_task_run_func(TaskPool *pool, /* Should only be the case for NOOPs, which never get to this point. */ BLI_assert(node->evaluate); - while (true) { - /* Get context. */ - /* TODO: Who initialises this? "Init" operations aren't able to - * initialise it!!! - */ - /* TODO(sergey): We don't use component contexts at this moment. */ - /* ComponentDepsNode *comp = node->owner; */ - BLI_assert(node->owner != NULL); - - /* Since we're not leaving the thread for until the graph branches it is - * possible to have NO-OP on the way. for which evaluate() will be NULL. - * but that's all fine, we'll just scheduler it's children. - */ - if (node->evaluate) { + /* Get context. */ + /* TODO: Who initialises this? "Init" operations aren't able to + * initialise it!!! + */ + /* TODO(sergey): We don't use component contexts at this moment. */ + /* ComponentDepsNode *comp = node->owner; */ + BLI_assert(node->owner != NULL); + + /* Since we're not leaving the thread for until the graph branches it is + * possible to have NO-OP on the way. for which evaluate() will be NULL. + * but that's all fine, we'll just scheduler it's children. + */ + if (node->evaluate) { /* Take note of current time. */ #ifdef USE_DEBUGGER - double start_time = PIL_check_seconds_timer(); - DepsgraphDebug::task_started(state->graph, node); + double start_time = PIL_check_seconds_timer(); + DepsgraphDebug::task_started(state->graph, node); #endif - /* Perform operation. */ - node->evaluate(state->eval_ctx); + /* Perform operation. */ + node->evaluate(state->eval_ctx); /* Note how long this took. */ #ifdef USE_DEBUGGER - double end_time = PIL_check_seconds_timer(); - DepsgraphDebug::task_completed(state->graph, - node, - end_time - start_time); + double end_time = PIL_check_seconds_timer(); + DepsgraphDebug::task_completed(state->graph, + node, + end_time - start_time); #endif - } - - /* If there's only one outgoing link we try to immediately switch to - * that node evaluation, without leaving the thread. - * - * It's only doable if the child don't have extra relations or all they - * are satisfied. - * - * TODO(sergey): Checks here can be de-duplicated with the ones from - * schedule_node(), however, how to do it nicely? - */ - if (node->outlinks.size() == 1) { - DepsRelation *rel = node->outlinks[0]; - OperationDepsNode *child = (OperationDepsNode *)rel->to; - BLI_assert(child->type == DEPSNODE_TYPE_OPERATION); - if (!child->scheduled) { - unsigned int id_layers = child->owner->owner->layers; - if (!((child->flag & DEPSOP_FLAG_NEEDS_UPDATE) != 0 && - (id_layers & state->layers) != 0)) - { - /* Node does not need an update, so can;t continue with the - * chain and need to switch to another one by leaving the - * thread. - */ - break; - } - if ((rel->flag & DEPSREL_FLAG_CYCLIC) == 0) { - BLI_assert(child->num_links_pending > 0); - atomic_sub_and_fetch_uint32(&child->num_links_pending, 1); - } - if (child->num_links_pending == 0) { - bool is_scheduled = atomic_fetch_and_or_uint8( - (uint8_t *)&child->scheduled, (uint8_t)true); - if (!is_scheduled) { - /* Node was not scheduled, switch to it! */ - node = child; - } - else { - /* Someone else scheduled the node, leaving us - * unemployed in this thread, we're leaving. - */ - break; - } - } - else { - /* There are other dependencies on the child, can't do - * anything in the current thread. - */ - break; - } - } - else { - /* Happens when having cyclic dependencies. - * - * Nothing to do here, single child was already scheduled, we - * can leave the thread now. - */ - break; - } - } - else { - /* TODO(sergey): It's possible to use one of the outgoing relations - * as a chain which we'll try to keep alive, but it's a bit more - * involved change. - */ - schedule_children(pool, state->graph, node, state->layers, thread_id); - break; - } } + + schedule_children(pool, state->graph, node, state->layers, thread_id); } typedef struct CalculatePengindData { @@ -378,12 +311,19 @@ void deg_evaluate_on_refresh(EvaluationContext *eval_ctx, state.graph = graph; state.layers = layers; - TaskScheduler *task_scheduler = BLI_task_scheduler_get(); - TaskPool *task_pool = BLI_task_pool_create(task_scheduler, &state); + TaskScheduler *task_scheduler; + bool need_free_scheduler; if (G.debug & G_DEBUG_DEPSGRAPH_NO_THREADS) { - BLI_pool_set_num_threads(task_pool, 1); + task_scheduler = BLI_task_scheduler_create(1); + need_free_scheduler = true; } + else { + task_scheduler = BLI_task_scheduler_get(); + need_free_scheduler = false; + } + + TaskPool *task_pool = BLI_task_pool_create_suspended(task_scheduler, &state); calculate_pending_parents(graph, layers); @@ -410,6 +350,10 @@ void deg_evaluate_on_refresh(EvaluationContext *eval_ctx, /* Clear any uncleared tags - just in case. */ deg_graph_clear_tags(graph); + + if (need_free_scheduler) { + BLI_task_scheduler_free(task_scheduler); + } } } // namespace DEG diff --git a/source/blender/depsgraph/intern/eval/deg_eval_flush.cc b/source/blender/depsgraph/intern/eval/deg_eval_flush.cc index 7c6c25bef0d..e10f86f6e95 100644 --- a/source/blender/depsgraph/intern/eval/deg_eval_flush.cc +++ b/source/blender/depsgraph/intern/eval/deg_eval_flush.cc @@ -180,6 +180,11 @@ void deg_graph_flush_updates(Main *bmain, Depsgraph *graph) comp_node->done = 1; /* Flush to nodes along links... */ + /* TODO(sergey): This is mainly giving speedup due ot less queue pushes, which + * reduces number of memory allocations. + * + * We should try solve the allocation issue instead of doing crazy things here. + */ if (node->outlinks.size() == 1) { OperationDepsNode *to_node = (OperationDepsNode *)node->outlinks[0]->to; if (to_node->scheduled == false) { diff --git a/source/blender/editors/animation/anim_channels_defines.c b/source/blender/editors/animation/anim_channels_defines.c index 57302c18a88..5852ee72b19 100644 --- a/source/blender/editors/animation/anim_channels_defines.c +++ b/source/blender/editors/animation/anim_channels_defines.c @@ -2775,7 +2775,7 @@ static bAnimChannelType ACF_DSMCLIP = acf_generic_indention_1, /* indent level */ acf_generic_basic_offset, /* offset */ - acf_generic_idblock_name , /* name */ + acf_generic_idblock_name, /* name */ acf_generic_idfill_name_prop, /* name prop */ acf_dsmclip_icon, /* icon */ @@ -3856,7 +3856,8 @@ void ANIM_channel_draw(bAnimContext *ac, bAnimListElem *ale, float yminc, float if (ac->sl) { if ((ac->spacetype == SPACE_IPO) && (acf->has_setting(ac, ale, ACHANNEL_SETTING_VISIBLE) || - acf->has_setting(ac, ale, ACHANNEL_SETTING_ALWAYS_VISIBLE))) { + acf->has_setting(ac, ale, ACHANNEL_SETTING_ALWAYS_VISIBLE))) + { /* for F-Curves, draw color-preview of curve behind checkbox */ if (ELEM(ale->type, ANIMTYPE_FCURVE, ANIMTYPE_NLACURVE)) { FCurve *fcu = (FCurve *)ale->data; diff --git a/source/blender/editors/animation/anim_draw.c b/source/blender/editors/animation/anim_draw.c index 33e44d73894..98900812bb2 100644 --- a/source/blender/editors/animation/anim_draw.c +++ b/source/blender/editors/animation/anim_draw.c @@ -46,6 +46,8 @@ #include "BLI_dlrbTree.h" #include "BKE_context.h" +#include "BKE_curve.h" +#include "BKE_fcurve.h" #include "BKE_global.h" #include "BKE_nla.h" #include "BKE_mask.h" @@ -115,7 +117,8 @@ void ANIM_draw_cfra(const bContext *C, View2D *v2d, short flag) /* Draw a light green line to indicate current frame */ UI_ThemeColor(TH_CFRAME); - const float x = (float)(scene->r.cfra * scene->r.framelen); + const float time = scene->r.cfra + scene->r.subframe; + const float x = (float)(time * scene->r.framelen); glLineWidth((flag & DRAWCFRA_WIDE) ? 3.0 : 2.0); @@ -308,7 +311,8 @@ static float normalization_factor_get(Scene *scene, FCurve *fcu, short flag, flo fcu->prev_norm_factor = 1.0f; if (fcu->bezt) { - BezTriple *bezt; + const bool use_preview_only = PRVRANGEON; + const BezTriple *bezt; int i; float max_coord = -FLT_MAX; float min_coord = FLT_MAX; @@ -318,28 +322,77 @@ static float normalization_factor_get(Scene *scene, FCurve *fcu, short flag, flo return 1.0f; } - if (PRVRANGEON) { - for (i = 0, bezt = fcu->bezt; i < fcu->totvert; i++, bezt++) { - if (IN_RANGE_INCL(bezt->vec[1][0], scene->r.psfra, scene->r.pefra)) { - max_coord = max_ff(max_coord, bezt->vec[0][1]); - max_coord = max_ff(max_coord, bezt->vec[1][1]); - max_coord = max_ff(max_coord, bezt->vec[2][1]); - - min_coord = min_ff(min_coord, bezt->vec[0][1]); - min_coord = min_ff(min_coord, bezt->vec[1][1]); - min_coord = min_ff(min_coord, bezt->vec[2][1]); - } + for (i = 0, bezt = fcu->bezt; i < fcu->totvert; i++, bezt++) { + if (use_preview_only && !IN_RANGE_INCL(bezt->vec[1][0], + scene->r.psfra, + scene->r.pefra)) + { + continue; } - } - else { - for (i = 0, bezt = fcu->bezt; i < fcu->totvert; i++, bezt++) { - max_coord = max_ff(max_coord, bezt->vec[0][1]); + + if (i == 0) { + /* We ignore extrapolation flags and handle here, and use the + * control point position only. so we normalize "interesting" + * part of the curve. + * + * Here we handle left extrapolation. + */ max_coord = max_ff(max_coord, bezt->vec[1][1]); - max_coord = max_ff(max_coord, bezt->vec[2][1]); - min_coord = min_ff(min_coord, bezt->vec[0][1]); min_coord = min_ff(min_coord, bezt->vec[1][1]); - min_coord = min_ff(min_coord, bezt->vec[2][1]); + } + else { + const BezTriple *prev_bezt = bezt - 1; + if (prev_bezt->ipo == BEZT_IPO_CONST) { + /* Constant interpolation: previous CV value is used up + * to the current keyframe. + */ + max_coord = max_ff(max_coord, bezt->vec[1][1]); + min_coord = min_ff(min_coord, bezt->vec[1][1]); + } + else if (prev_bezt->ipo == BEZT_IPO_LIN) { + /* Linear interpolation: min/max using both previous and + * and current CV. + */ + max_coord = max_ff(max_coord, bezt->vec[1][1]); + min_coord = min_ff(min_coord, bezt->vec[1][1]); + max_coord = max_ff(max_coord, prev_bezt->vec[1][1]); + min_coord = min_ff(min_coord, prev_bezt->vec[1][1]); + } + else if (prev_bezt->ipo == BEZT_IPO_BEZ) { + const int resol = fcu->driver + ? 32 + : min_ii((int)(5.0f * len_v2v2(bezt->vec[1], prev_bezt->vec[1])), 32); + if (resol < 2) { + max_coord = max_ff(max_coord, prev_bezt->vec[1][1]); + min_coord = min_ff(min_coord, prev_bezt->vec[1][1]); + } + else { + float data[120]; + float v1[2], v2[2], v3[2], v4[2]; + + v1[0] = prev_bezt->vec[1][0]; + v1[1] = prev_bezt->vec[1][1]; + v2[0] = prev_bezt->vec[2][0]; + v2[1] = prev_bezt->vec[2][1]; + + v3[0] = bezt->vec[0][0]; + v3[1] = bezt->vec[0][1]; + v4[0] = bezt->vec[1][0]; + v4[1] = bezt->vec[1][1]; + + correct_bezpart(v1, v2, v3, v4); + + BKE_curve_forward_diff_bezier(v1[0], v2[0], v3[0], v4[0], data, resol, sizeof(float) * 3); + BKE_curve_forward_diff_bezier(v1[1], v2[1], v3[1], v4[1], data + 1, resol, sizeof(float) * 3); + + for (int j = 0; j <= resol; ++j) { + const float *fp = &data[j * 3]; + max_coord = max_ff(max_coord, fp[1]); + min_coord = min_ff(min_coord, fp[1]); + } + } + } } } diff --git a/source/blender/editors/animation/anim_ops.c b/source/blender/editors/animation/anim_ops.c index c0d6963acbb..0eb6508f7b2 100644 --- a/source/blender/editors/animation/anim_ops.c +++ b/source/blender/editors/animation/anim_ops.c @@ -95,7 +95,7 @@ static void change_frame_apply(bContext *C, wmOperator *op) { Main *bmain = CTX_data_main(C); Scene *scene = CTX_data_scene(C); - int frame = RNA_int_get(op->ptr, "frame"); + float frame = RNA_float_get(op->ptr, "frame"); bool do_snap = RNA_boolean_get(op->ptr, "snap"); if (do_snap && CTX_wm_space_seq(C)) { @@ -103,10 +103,16 @@ static void change_frame_apply(bContext *C, wmOperator *op) } /* set the new frame number */ - CFRA = frame; + if (scene->r.flag & SCER_SHOW_SUBFRAME) { + CFRA = (int)frame; + SUBFRA = frame - (int)frame; + } + else { + CFRA = iroundf(frame); + SUBFRA = 0.0f; + } FRAMENUMBER_MIN_CLAMP(CFRA); - SUBFRA = 0.0f; - + /* do updates */ BKE_sound_seek_scene(bmain, scene); WM_event_add_notifier(C, NC_SCENE | ND_FRAME, scene); @@ -125,19 +131,16 @@ static int change_frame_exec(bContext *C, wmOperator *op) /* ---- */ /* Get frame from mouse coordinates */ -static int frame_from_event(bContext *C, const wmEvent *event) +static float frame_from_event(bContext *C, const wmEvent *event) { ARegion *region = CTX_wm_region(C); Scene *scene = CTX_data_scene(C); - float viewx; - int frame; + float frame; /* convert from region coordinates to View2D 'tot' space */ - viewx = UI_view2d_region_to_view_x(®ion->v2d, event->mval[0]); - - /* round result to nearest int (frames are ints!) */ - frame = iroundf(viewx); + frame = UI_view2d_region_to_view_x(®ion->v2d, event->mval[0]); + /* respect preview range restrictions (if only allowed to move around within that range) */ if (scene->r.flag & SCER_LOCK_FRAME_SELECTION) { CLAMP(frame, PSFRA, PEFRA); } @@ -187,7 +190,7 @@ static int change_frame_invoke(bContext *C, wmOperator *op, const wmEvent *event * as user could click on a single frame (jump to frame) as well as * click-dragging over a range (modal scrubbing). */ - RNA_int_set(op->ptr, "frame", frame_from_event(C, event)); + RNA_float_set(op->ptr, "frame", frame_from_event(C, event)); change_frame_seq_preview_begin(C, event); @@ -215,7 +218,7 @@ static int change_frame_modal(bContext *C, wmOperator *op, const wmEvent *event) break; case MOUSEMOVE: - RNA_int_set(op->ptr, "frame", frame_from_event(C, event)); + RNA_float_set(op->ptr, "frame", frame_from_event(C, event)); change_frame_apply(C, op); break; @@ -268,7 +271,7 @@ static void ANIM_OT_change_frame(wmOperatorType *ot) ot->undo_group = "FRAME_CHANGE"; /* rna */ - ot->prop = RNA_def_int(ot->srna, "frame", 0, MINAFRAME, MAXFRAME, "Frame", "", MINAFRAME, MAXFRAME); + ot->prop = RNA_def_float(ot->srna, "frame", 0, MINAFRAME, MAXFRAME, "Frame", "", MINAFRAME, MAXFRAME); prop = RNA_def_boolean(ot->srna, "snap", false, "Snap", ""); RNA_def_property_flag(prop, PROP_SKIP_SAVE); } diff --git a/source/blender/editors/armature/armature_intern.h b/source/blender/editors/armature/armature_intern.h index b39b4bd81ee..190b0610059 100644 --- a/source/blender/editors/armature/armature_intern.h +++ b/source/blender/editors/armature/armature_intern.h @@ -247,8 +247,10 @@ void armature_select_mirrored_ex(struct bArmature *arm, const int flag); void armature_select_mirrored(struct bArmature *arm); void armature_tag_unselect(struct bArmature *arm); -void *get_nearest_bone(struct bContext *C, short findunsel, int x, int y); -void *get_bone_from_selectbuffer(struct Scene *scene, struct Base *base, unsigned int *buffer, short hits, short findunsel, bool do_nearest); +void *get_nearest_bone(struct bContext *C, const int xy[2], bool findunsel); +void *get_bone_from_selectbuffer( + struct Scene *scene, struct Base *base, const unsigned int *buffer, short hits, + bool findunsel, bool do_nearest); int bone_looper(struct Object *ob, struct Bone *bone, void *data, int (*bone_func)(struct Object *, struct Bone *, void *)); diff --git a/source/blender/editors/armature/armature_naming.c b/source/blender/editors/armature/armature_naming.c index fa192ed6f36..c928508237d 100644 --- a/source/blender/editors/armature/armature_naming.c +++ b/source/blender/editors/armature/armature_naming.c @@ -362,7 +362,7 @@ static int armature_flip_names_exec(bContext *C, wmOperator *UNUSED(op)) arm = ob->data; - ListBase bones_names= {NULL}; + ListBase bones_names = {NULL}; CTX_DATA_BEGIN(C, EditBone *, ebone, selected_editable_bones) { diff --git a/source/blender/editors/armature/armature_select.c b/source/blender/editors/armature/armature_select.c index e9946abba0b..ed44214591c 100644 --- a/source/blender/editors/armature/armature_select.c +++ b/source/blender/editors/armature/armature_select.c @@ -74,7 +74,9 @@ Bone *get_indexed_bone(Object *ob, int index) /* See if there are any selected bones in this buffer */ /* only bones from base are checked on */ -void *get_bone_from_selectbuffer(Scene *scene, Base *base, unsigned int *buffer, short hits, short findunsel, bool do_nearest) +void *get_bone_from_selectbuffer( + Scene *scene, Base *base, const unsigned int *buffer, short hits, + bool findunsel, bool do_nearest) { Object *obedit = scene->obedit; // XXX get from context Bone *bone; @@ -103,8 +105,8 @@ void *get_bone_from_selectbuffer(Scene *scene, Base *base, unsigned int *buffer, sel = (bone->flag & BONE_SELECTED); else sel = !(bone->flag & BONE_SELECTED); - - data = bone; + + data = bone; } else { data = NULL; @@ -162,7 +164,7 @@ void *get_bone_from_selectbuffer(Scene *scene, Base *base, unsigned int *buffer, /* used by posemode as well editmode */ /* only checks scene->basact! */ /* x and y are mouse coords (area space) */ -void *get_nearest_bone(bContext *C, short findunsel, int x, int y) +void *get_nearest_bone(bContext *C, const int xy[2], bool findunsel) { ViewContext vc; rcti rect; @@ -172,10 +174,10 @@ void *get_nearest_bone(bContext *C, short findunsel, int x, int y) view3d_set_viewcontext(C, &vc); // rect.xmin = ... mouseco! - rect.xmin = rect.xmax = x; - rect.ymin = rect.ymax = y; + rect.xmin = rect.xmax = xy[0]; + rect.ymin = rect.ymax = xy[1]; - hits = view3d_opengl_select(&vc, buffer, MAXPICKBUF, &rect, true); + hits = view3d_opengl_select(&vc, buffer, MAXPICKBUF, &rect, VIEW3D_SELECT_PICK_NEAREST); if (hits > 0) return get_bone_from_selectbuffer(vc.scene, vc.scene->basact, buffer, hits, findunsel, true); @@ -197,10 +199,7 @@ static int armature_select_linked_invoke(bContext *C, wmOperator *op, const wmEv view3d_operator_needs_opengl(C); - if (extend) - bone = get_nearest_bone(C, 0, event->mval[0], event->mval[1]); - else - bone = get_nearest_bone(C, 1, event->mval[0], event->mval[1]); + bone = get_nearest_bone(C, event->mval, !extend); if (!bone) return OPERATOR_CANCELLED; @@ -276,10 +275,24 @@ void ARMATURE_OT_select_linked(wmOperatorType *ot) RNA_def_boolean(ot->srna, "extend", false, "Extend", "Extend selection instead of deselecting everything first"); } +/* utility function for get_nearest_editbonepoint */ +static int selectbuffer_ret_hits_12(unsigned int *UNUSED(buffer), const int hits12) +{ + return hits12; +} + +static int selectbuffer_ret_hits_5(unsigned int *buffer, const int hits12, const int hits5) +{ + const int offs = 4 * hits12; + memcpy(buffer, buffer + offs, 4 * hits5 * sizeof(unsigned int)); + return hits5; +} + /* does bones and points */ /* note that BONE ROOT only gets drawn for root bones (or without IK) */ -static EditBone *get_nearest_editbonepoint(ViewContext *vc, const int mval[2], - ListBase *edbo, int findunsel, int *selmask) +static EditBone *get_nearest_editbonepoint( + ViewContext *vc, const int mval[2], + ListBase *edbo, bool findunsel, bool use_cycle, int *r_selmask) { bArmature *arm = (bArmature *)vc->obedit->data; EditBone *ebone_next_act = arm->act_edbone; @@ -289,7 +302,9 @@ static EditBone *get_nearest_editbonepoint(ViewContext *vc, const int mval[2], unsigned int buffer[MAXPICKBUF]; unsigned int hitresult, besthitresult = BONESEL_NOSEL; int i, mindep = 5; - short hits; + int hits12, hits5 = 0; + + static int last_mval[2] = {-100, -100}; /* find the bone after the current active bone, so as to bump up its chances in selection. * this way overlapping bones will cycle selection state as with objects. */ @@ -303,22 +318,59 @@ static EditBone *get_nearest_editbonepoint(ViewContext *vc, const int mval[2], ebone_next_act = NULL; } - rect.xmin = mval[0] - 5; - rect.xmax = mval[0] + 5; - rect.ymin = mval[1] - 5; - rect.ymax = mval[1] + 5; - - hits = view3d_opengl_select(vc, buffer, MAXPICKBUF, &rect, true); - if (hits == 0) { - rect.xmin = mval[0] - 12; - rect.xmax = mval[0] + 12; - rect.ymin = mval[1] - 12; - rect.ymax = mval[1] + 12; - hits = view3d_opengl_select(vc, buffer, MAXPICKBUF, &rect, true); + bool do_nearest = false; + + /* define if we use solid nearest select or not */ + if (use_cycle) { + if (vc->v3d->drawtype > OB_WIRE) { + do_nearest = true; + if (len_manhattan_v2v2_int(mval, last_mval) < 3) { + do_nearest = false; + } + } + copy_v2_v2_int(last_mval, mval); + } + else { + if (vc->v3d->drawtype > OB_WIRE) { + do_nearest = true; + } } + + /* matching logic from 'mixed_bones_object_selectbuffer' */ + const int select_mode = (do_nearest ? VIEW3D_SELECT_PICK_NEAREST : VIEW3D_SELECT_PICK_ALL); + int hits = 0; + + /* we _must_ end cache before return, use 'goto cache_end' */ + view3d_opengl_select_cache_begin(); + + BLI_rcti_init_pt_radius(&rect, mval, 12); + hits12 = view3d_opengl_select(vc, buffer, MAXPICKBUF, &rect, select_mode); + if (hits12 == 1) { + hits = selectbuffer_ret_hits_12(buffer, hits12); + goto cache_end; + } + else if (hits12 > 0) { + int offs; + + offs = 4 * hits12; + BLI_rcti_init_pt_radius(&rect, mval, 5); + hits5 = view3d_opengl_select(vc, buffer + offs, MAXPICKBUF - offs, &rect, select_mode); + + if (hits5 == 1) { + hits = selectbuffer_ret_hits_5(buffer, hits12, hits5); + goto cache_end; + } + + if (hits5 > 0) { hits = selectbuffer_ret_hits_5(buffer, hits12, hits5); goto cache_end; } + else { hits = selectbuffer_ret_hits_12(buffer, hits12); goto cache_end; } + } + +cache_end: + view3d_opengl_select_cache_end(); + /* See if there are any selected bones in this group */ if (hits > 0) { - + if (hits == 1) { if (!(buffer[3] & BONESEL_NOSEL)) besthitresult = buffer[3]; @@ -375,17 +427,17 @@ static EditBone *get_nearest_editbonepoint(ViewContext *vc, const int mval[2], ebone = BLI_findlink(edbo, besthitresult & ~BONESEL_ANY); - *selmask = 0; + *r_selmask = 0; if (besthitresult & BONESEL_ROOT) - *selmask |= BONE_ROOTSEL; + *r_selmask |= BONE_ROOTSEL; if (besthitresult & BONESEL_TIP) - *selmask |= BONE_TIPSEL; + *r_selmask |= BONE_TIPSEL; if (besthitresult & BONESEL_BONE) - *selmask |= BONE_SELECTED; + *r_selmask |= BONE_SELECTED; return ebone; } } - *selmask = 0; + *r_selmask = 0; return NULL; } @@ -439,8 +491,8 @@ bool ED_armature_select_pick(bContext *C, const int mval[2], bool extend, bool d if (BIF_sk_selectStroke(C, mval, extend)) { return true; } - - nearBone = get_nearest_editbonepoint(&vc, mval, arm->edbo, 1, &selmask); + + nearBone = get_nearest_editbonepoint(&vc, mval, arm->edbo, true, true, &selmask); if (nearBone) { if (!extend && !deselect && !toggle) { @@ -1202,7 +1254,7 @@ static int armature_shortest_path_pick_invoke(bContext *C, wmOperator *op, const view3d_operator_needs_opengl(C); ebone_src = arm->act_edbone; - ebone_dst = get_nearest_bone(C, 0, event->mval[0], event->mval[1]); + ebone_dst = get_nearest_bone(C, event->mval, false); /* fallback to object selection */ if (ELEM(NULL, ebone_src, ebone_dst) || (ebone_src == ebone_dst)) { diff --git a/source/blender/editors/armature/editarmature_sketch.c b/source/blender/editors/armature/editarmature_sketch.c index f6c04e9570a..bba486bc65c 100644 --- a/source/blender/editors/armature/editarmature_sketch.c +++ b/source/blender/editors/armature/editarmature_sketch.c @@ -1907,12 +1907,9 @@ static bool sk_selectStroke(bContext *C, SK_Sketch *sketch, const int mval[2], c view3d_set_viewcontext(C, &vc); - rect.xmin = mval[0] - 5; - rect.xmax = mval[0] + 5; - rect.ymin = mval[1] - 5; - rect.ymax = mval[1] + 5; + BLI_rcti_init_pt_radius(&rect, mval, 5); - hits = view3d_opengl_select(&vc, buffer, MAXPICKBUF, &rect, true); + hits = view3d_opengl_select(&vc, buffer, MAXPICKBUF, &rect, VIEW3D_SELECT_PICK_NEAREST); if (hits > 0) { int besthitresult = -1; diff --git a/source/blender/editors/armature/pose_select.c b/source/blender/editors/armature/pose_select.c index 44470c1f827..6e328552411 100644 --- a/source/blender/editors/armature/pose_select.c +++ b/source/blender/editors/armature/pose_select.c @@ -132,8 +132,9 @@ void ED_pose_bone_select(Object *ob, bPoseChannel *pchan, bool select) /* called from editview.c, for mode-less pose selection */ /* assumes scene obact and basact is still on old situation */ -int ED_do_pose_selectbuffer(Scene *scene, Base *base, unsigned int *buffer, short hits, - bool extend, bool deselect, bool toggle, bool do_nearest) +bool ED_do_pose_selectbuffer( + Scene *scene, Base *base, const unsigned int *buffer, short hits, + bool extend, bool deselect, bool toggle, bool do_nearest) { Object *ob = base->object; Bone *nearBone; @@ -280,12 +281,9 @@ static int pose_select_connected_invoke(bContext *C, wmOperator *op, const wmEve const bool extend = RNA_boolean_get(op->ptr, "extend"); view3d_operator_needs_opengl(C); - - if (extend) - bone = get_nearest_bone(C, 0, event->mval[0], event->mval[1]); - else - bone = get_nearest_bone(C, 1, event->mval[0], event->mval[1]); - + + bone = get_nearest_bone(C, event->mval, !extend); + if (!bone) return OPERATOR_CANCELLED; diff --git a/source/blender/editors/armature/pose_slide.c b/source/blender/editors/armature/pose_slide.c index 8e8345d34c9..21cb405c32b 100644 --- a/source/blender/editors/armature/pose_slide.c +++ b/source/blender/editors/armature/pose_slide.c @@ -98,8 +98,11 @@ typedef struct tPoseSlideOp { int prevFrame; /* frame before current frame (blend-from) */ int nextFrame; /* frame after current frame (blend-to) */ - int mode; /* sliding mode (ePoseSlide_Modes) */ - int flag; /* unused for now, but can later get used for storing runtime settings.... */ + short mode; /* sliding mode (ePoseSlide_Modes) */ + short flag; /* unused for now, but can later get used for storing runtime settings.... */ + + short channels; /* which transforms/channels are affected (ePoseSlide_Channels) */ + short axislock; /* axis-limits for transforms (ePoseSlide_AxisLock) */ float percentage; /* 0-1 value for determining the influence of whatever is relevant */ @@ -113,6 +116,49 @@ typedef enum ePoseSlide_Modes { POSESLIDE_BREAKDOWN, /* slide between the endpoint poses, finding a 'soft' spot */ } ePoseSlide_Modes; + +/* Transforms/Channels to Affect */ +typedef enum ePoseSlide_Channels { + PS_TFM_ALL = 0, /* All transforms and properties */ + + PS_TFM_LOC, /* Loc/Rot/Scale */ + PS_TFM_ROT, + PS_TFM_SIZE, + + PS_TFM_BBONE_SHAPE, /* Bendy Bones */ + + PS_TFM_PROPS /* Custom Properties */ +} ePoseSlide_Channels; + +/* Property enum for ePoseSlide_Channels */ +static EnumPropertyItem prop_channels_types[] = { + {PS_TFM_ALL, "ALL", 0, "All Properties", + "All properties, including transforms, bendy bone shape, and custom properties"}, + {PS_TFM_LOC, "LOC", 0, "Location", "Location only"}, + {PS_TFM_ROT, "ROT", 0, "Rotation", "Rotation only"}, + {PS_TFM_SIZE, "SIZE", 0, "Scale", "Scale only"}, + {PS_TFM_BBONE_SHAPE, "BBONE", 0, "Bendy Bone", "Bendy Bone shape properties"}, + {PS_TFM_PROPS, "CUSTOM", 0, "Custom Properties", "Custom properties"}, + {0, NULL, 0, NULL, NULL} +}; + +/* Axis Locks */ +typedef enum ePoseSlide_AxisLock { + PS_LOCK_X = (1 << 0), + PS_LOCK_Y = (1 << 1), + PS_LOCK_Z = (1 << 2) +} ePoseSlide_AxisLock; + +/* Property enum for ePoseSlide_AxisLock */ +static EnumPropertyItem prop_axis_lock_types[] = { + {0, "FREE", 0, "Free", "All axes are affected"}, + {PS_LOCK_X, "X", 0, "X", "Only X-axis transforms are affected"}, + {PS_LOCK_Y, "Y", 0, "Y", "Only Y-axis transforms are affected"}, + {PS_LOCK_Z, "Z", 0, "Z", "Only Z-axis transforms are affected"}, + /* TODO: Combinations? */ + {0, NULL, 0, NULL, NULL} +}; + /* ------------------------------------ */ /* operator init */ @@ -139,6 +185,10 @@ static int pose_slide_init(bContext *C, wmOperator *op, short mode) pso->prevFrame = RNA_int_get(op->ptr, "prev_frame"); pso->nextFrame = RNA_int_get(op->ptr, "next_frame"); + /* get the set of properties/axes that can be operated on */ + pso->channels = RNA_enum_get(op->ptr, "channels"); + pso->axislock = RNA_enum_get(op->ptr, "axis_lock"); + /* check the settings from the context */ if (ELEM(NULL, pso->ob, pso->arm, pso->ob->adt, pso->ob->adt->action)) return 0; @@ -293,10 +343,20 @@ static void pose_slide_apply_vec3(tPoseSlideOp *pso, tPChanFCurveLink *pfl, floa /* using this path, find each matching F-Curve for the variables we're interested in */ while ( (ld = poseAnim_mapping_getNextFCurve(&pfl->fcurves, ld, path)) ) { FCurve *fcu = (FCurve *)ld->data; - - /* just work on these channels one by one... there's no interaction between values */ + const int idx = fcu->array_index; + const int lock = pso->axislock; + + /* check if this F-Curve is ok given the current axis locks */ BLI_assert(fcu->array_index < 3); - pose_slide_apply_val(pso, fcu, &vec[fcu->array_index]); + + if ((lock == 0) || + ((lock & PS_LOCK_X) && (idx == 0)) || + ((lock & PS_LOCK_Y) && (idx == 1)) || + ((lock & PS_LOCK_Z) && (idx == 2))) + { + /* just work on these channels one by one... there's no interaction between values */ + pose_slide_apply_val(pso, fcu, &vec[fcu->array_index]); + } } /* free the temp path we got */ @@ -494,17 +554,17 @@ static void pose_slide_apply(bContext *C, tPoseSlideOp *pso) */ bPoseChannel *pchan = pfl->pchan; - if (pchan->flag & POSE_LOC) { + if (ELEM(pso->channels, PS_TFM_ALL, PS_TFM_LOC) && (pchan->flag & POSE_LOC)) { /* calculate these for the 'location' vector, and use location curves */ pose_slide_apply_vec3(pso, pfl, pchan->loc, "location"); } - if (pchan->flag & POSE_SIZE) { + if (ELEM(pso->channels, PS_TFM_ALL, PS_TFM_SIZE) && (pchan->flag & POSE_SIZE)) { /* calculate these for the 'scale' vector, and use scale curves */ pose_slide_apply_vec3(pso, pfl, pchan->size, "scale"); } - if (pchan->flag & POSE_ROT) { + if (ELEM(pso->channels, PS_TFM_ALL, PS_TFM_ROT) && (pchan->flag & POSE_ROT)) { /* everything depends on the rotation mode */ if (pchan->rotmode > 0) { /* eulers - so calculate these for the 'eul' vector, and use euler_rotation curves */ @@ -519,12 +579,12 @@ static void pose_slide_apply(bContext *C, tPoseSlideOp *pso) } } - if (pchan->flag & POSE_BBONE_SHAPE) { + if (ELEM(pso->channels, PS_TFM_ALL, PS_TFM_BBONE_SHAPE) && (pchan->flag & POSE_BBONE_SHAPE)) { /* bbone properties - they all start a "bbone_" prefix */ pose_slide_apply_props(pso, pfl, "bbone_"); } - if (pfl->oldprops) { + if (ELEM(pso->channels, PS_TFM_ALL, PS_TFM_PROPS) && (pfl->oldprops)) { /* not strictly a transform, but custom properties contribute to the pose produced in many rigs * (e.g. the facial rigs used in Sintel) */ @@ -553,9 +613,12 @@ static void pose_slide_reset(tPoseSlideOp *pso) /* ------------------------------------ */ /* draw percentage indicator in header */ +// TODO: Include hints about locks here... static void pose_slide_draw_status(tPoseSlideOp *pso) { char status_str[UI_MAX_DRAW_STR]; + char limits_str[UI_MAX_DRAW_STR]; + char axis_str[50]; char mode_str[32]; switch (pso->mode) { @@ -575,16 +638,58 @@ static void pose_slide_draw_status(tPoseSlideOp *pso) break; } + switch (pso->axislock) { + case PS_LOCK_X: + BLI_strncpy(axis_str, "[X]/Y/Z axis only (X to clear)", sizeof(axis_str)); + break; + case PS_LOCK_Y: + BLI_strncpy(axis_str, "X/[Y]/Z axis only (Y to clear)", sizeof(axis_str)); + break; + case PS_LOCK_Z: + BLI_strncpy(axis_str, "X/Y/[Z] axis only (Z to clear)", sizeof(axis_str)); + break; + + default: + if (ELEM(pso->channels, PS_TFM_LOC, PS_TFM_ROT, PS_TFM_SIZE)) { + BLI_strncpy(axis_str, "X/Y/Z = Axis Constraint", sizeof(axis_str)); + } + else { + axis_str[0] = '\0'; + } + break; + } + + switch (pso->channels) { + case PS_TFM_LOC: + BLI_snprintf(limits_str, sizeof(limits_str), "[G]/R/S/B/C - Location only (G to clear) | %s", axis_str); + break; + case PS_TFM_ROT: + BLI_snprintf(limits_str, sizeof(limits_str), "G/[R]/S/B/C - Rotation only (R to clear) | %s", axis_str); + break; + case PS_TFM_SIZE: + BLI_snprintf(limits_str, sizeof(limits_str), "G/R/[S]/B/C - Scale only (S to clear) | %s", axis_str); + break; + case PS_TFM_BBONE_SHAPE: + BLI_strncpy(limits_str, "G/R/S/[B]/C - Bendy Bone properties only (B to clear) | %s", sizeof(limits_str)); + break; + case PS_TFM_PROPS: + BLI_strncpy(limits_str, "G/R/S/B/[C] - Custom Properties only (C to clear) | %s", sizeof(limits_str)); + break; + default: + BLI_strncpy(limits_str, "G/R/S/B/C - Limit to Transform/Property Set", sizeof(limits_str)); + break; + } + if (hasNumInput(&pso->num)) { Scene *scene = pso->scene; char str_offs[NUM_STR_REP_LEN]; outputNumInput(&pso->num, str_offs, &scene->unit); - BLI_snprintf(status_str, sizeof(status_str), "%s: %s", mode_str, str_offs); + BLI_snprintf(status_str, sizeof(status_str), "%s: %s | %s", mode_str, str_offs, limits_str); } else { - BLI_snprintf(status_str, sizeof(status_str), "%s: %d %%", mode_str, (int)(pso->percentage * 100.0f)); + BLI_snprintf(status_str, sizeof(status_str), "%s: %d %% | %s", mode_str, (int)(pso->percentage * 100.0f), limits_str); } ED_area_headerprint(pso->sa, status_str); @@ -675,11 +780,58 @@ static void pose_slide_mouse_update_percentage(tPoseSlideOp *pso, wmOperator *op RNA_float_set(op->ptr, "percentage", pso->percentage); } +/* handle an event to toggle channels mode */ +static void pose_slide_toggle_channels_mode(wmOperator *op, tPoseSlideOp *pso, ePoseSlide_Channels channel) +{ + /* Turn channel on or off? */ + if (pso->channels == channel) { + /* Already limiting to transform only, so pressing this again turns it off */ + pso->channels = PS_TFM_ALL; + } + else { + /* Only this set of channels */ + pso->channels = channel; + } + RNA_enum_set(op->ptr, "channels", pso->channels); + + + /* Reset axis limits too for good measure */ + pso->axislock = 0; + RNA_enum_set(op->ptr, "axis_lock", pso->axislock); +} + +/* handle an event to toggle axis locks - returns whether any change in state is needed */ +static bool pose_slide_toggle_axis_locks(wmOperator *op, tPoseSlideOp *pso, ePoseSlide_AxisLock axis) +{ + /* Axis can only be set when a transform is set - it doesn't make sense otherwise */ + if (ELEM(pso->channels, PS_TFM_ALL, PS_TFM_BBONE_SHAPE, PS_TFM_PROPS)) { + pso->axislock = 0; + RNA_enum_set(op->ptr, "axis_lock", pso->axislock); + return false; + } + + /* Turn on or off? */ + if (pso->axislock == axis) { + /* Already limiting on this axis, so turn off */ + pso->axislock = 0; + } + else { + /* Only this axis */ + pso->axislock = axis; + } + RNA_enum_set(op->ptr, "axis_lock", pso->axislock); + + /* Setting changed, so pose update is needed */ + return true; +} + /* common code for modal() */ static int pose_slide_modal(bContext *C, wmOperator *op, const wmEvent *event) { tPoseSlideOp *pso = op->customdata; wmWindow *win = CTX_wm_window(C); + bool do_pose_update = false; + const bool has_numinput = hasNumInput(&pso->num); switch (event->type) { @@ -718,7 +870,8 @@ static int pose_slide_modal(bContext *C, wmOperator *op, const wmEvent *event) /* canceled! */ return OPERATOR_CANCELLED; } - + + /* Percentage Chane... */ case MOUSEMOVE: /* calculate new position */ { /* only handle mousemove if not doing numinput */ @@ -726,14 +879,8 @@ static int pose_slide_modal(bContext *C, wmOperator *op, const wmEvent *event) /* update percentage based on position of mouse */ pose_slide_mouse_update_percentage(pso, op, event); - /* update percentage indicator in header */ - pose_slide_draw_status(pso); - - /* reset transforms (to avoid accumulation errors) */ - pose_slide_reset(pso); - - /* apply... */ - pose_slide_apply(C, pso); + /* update pose to reflect the new values (see below) */ + do_pose_update = true; } break; } @@ -751,16 +898,75 @@ static int pose_slide_modal(bContext *C, wmOperator *op, const wmEvent *event) CLAMP(pso->percentage, 0.0f, 1.0f); RNA_float_set(op->ptr, "percentage", pso->percentage); - /* update percentage indicator in header */ - pose_slide_draw_status(pso); - - /* reset transforms (to avoid accumulation errors) */ - pose_slide_reset(pso); - - /* apply... */ - pose_slide_apply(C, pso); + /* Update pose to reflect the new values (see below) */ + do_pose_update = true; break; } + else if (event->val == KM_PRESS) { + switch (event->type) { + /* Transform Channel Limits */ + /* XXX: Replace these hardcoded hotkeys with a modalmap that can be customised */ + case GKEY: /* Location */ + { + pose_slide_toggle_channels_mode(op, pso, PS_TFM_LOC); + do_pose_update = true; + break; + } + case RKEY: /* Rotation */ + { + pose_slide_toggle_channels_mode(op, pso, PS_TFM_ROT); + do_pose_update = true; + break; + } + case SKEY: /* Scale */ + { + pose_slide_toggle_channels_mode(op, pso, PS_TFM_SIZE); + do_pose_update = true; + break; + } + case BKEY: /* Bendy Bones */ + { + pose_slide_toggle_channels_mode(op, pso, PS_TFM_BBONE_SHAPE); + do_pose_update = true; + break; + } + case CKEY: /* Custom Properties */ + { + pose_slide_toggle_channels_mode(op, pso, PS_TFM_PROPS); + do_pose_update = true; + break; + } + + + /* Axis Locks */ + /* XXX: Hardcoded... */ + case XKEY: + { + if (pose_slide_toggle_axis_locks(op, pso, PS_LOCK_X)) { + do_pose_update = true; + } + break; + } + case YKEY: + { + if (pose_slide_toggle_axis_locks(op, pso, PS_LOCK_Y)) { + do_pose_update = true; + } + break; + } + case ZKEY: + { + if (pose_slide_toggle_axis_locks(op, pso, PS_LOCK_Z)) { + do_pose_update = true; + } + break; + } + + + default: /* Some other unhandled key... */ + break; + } + } else { /* unhandled event - maybe it was some view manip? */ /* allow to pass through */ @@ -768,6 +974,19 @@ static int pose_slide_modal(bContext *C, wmOperator *op, const wmEvent *event) } } + + /* perform pose updates - in response to some user action (e.g. pressing a key or moving the mouse) */ + if (do_pose_update) { + /* update percentage indicator in header */ + pose_slide_draw_status(pso); + + /* reset transforms (to avoid accumulation errors) */ + pose_slide_reset(pso); + + /* apply... */ + pose_slide_apply(C, pso); + } + /* still running... */ return OPERATOR_RUNNING_MODAL; } @@ -795,11 +1014,16 @@ static int pose_slide_exec_common(bContext *C, wmOperator *op, tPoseSlideOp *pso } /* common code for defining RNA properties */ +/* TODO: Skip save on these? */ static void pose_slide_opdef_properties(wmOperatorType *ot) { + RNA_def_float_percentage(ot->srna, "percentage", 0.5f, 0.0f, 1.0f, "Percentage", "Weighting factor for which keyframe is favored more", 0.3, 0.7); + RNA_def_int(ot->srna, "prev_frame", 0, MINAFRAME, MAXFRAME, "Previous Keyframe", "Frame number of keyframe immediately before the current frame", 0, 50); RNA_def_int(ot->srna, "next_frame", 0, MINAFRAME, MAXFRAME, "Next Keyframe", "Frame number of keyframe immediately after the current frame", 0, 50); - RNA_def_float_percentage(ot->srna, "percentage", 0.5f, 0.0f, 1.0f, "Percentage", "Weighting factor for the sliding operation", 0.3, 0.7); + + RNA_def_enum(ot->srna, "channels", prop_channels_types, PS_TFM_ALL, "Channels", "Set of properties that are affected"); + RNA_def_enum(ot->srna, "axis_lock", prop_axis_lock_types, 0, "Axis Lock", "Transform axis to restrict effects to"); } /* ------------------------------------ */ diff --git a/source/blender/editors/curve/editcurve.c b/source/blender/editors/curve/editcurve.c index e9fd5fb5a43..47f42ab5321 100644 --- a/source/blender/editors/curve/editcurve.c +++ b/source/blender/editors/curve/editcurve.c @@ -91,14 +91,6 @@ typedef struct { int flag; } UndoCurve; -/* Definitions needed for shape keys */ -typedef struct { - void *orig_cv; - int key_index, nu_index, pt_index, vertex_index; - bool switched; - Nurb *orig_nu; -} CVKeyIndex; - void selectend_nurb(Object *obedit, enum eEndPoint_Types selfirst, bool doswap, bool selstatus); static void adduplicateflagNurb(Object *obedit, ListBase *newnurb, const short flag, const bool split); static int curve_delete_segments(Object *obedit, const bool split); @@ -138,9 +130,9 @@ void printknots(Object *obedit) /* ********************* Shape keys *************** */ -static CVKeyIndex *init_cvKeyIndex(void *cv, int key_index, int nu_index, int pt_index, int vertex_index, Nurb *orig_nu) +static CVKeyIndex *init_cvKeyIndex(void *cv, int key_index, int nu_index, int pt_index, int vertex_index) { - CVKeyIndex *cvIndex = MEM_callocN(sizeof(CVKeyIndex), "init_cvKeyIndex"); + CVKeyIndex *cvIndex = MEM_callocN(sizeof(CVKeyIndex), __func__); cvIndex->orig_cv = cv; cvIndex->key_index = key_index; @@ -148,7 +140,6 @@ static CVKeyIndex *init_cvKeyIndex(void *cv, int key_index, int nu_index, int pt cvIndex->pt_index = pt_index; cvIndex->vertex_index = vertex_index; cvIndex->switched = false; - cvIndex->orig_nu = orig_nu; return cvIndex; } @@ -174,7 +165,12 @@ static void init_editNurb_keyIndex(EditNurb *editnurb, ListBase *origBase) origbezt = orignu->bezt; pt_index = 0; while (a--) { - keyIndex = init_cvKeyIndex(origbezt, key_index, nu_index, pt_index, vertex_index, orignu); + /* We cannot keep *any* reference to curve obdata, + * it might be replaced and freed while editcurve remain in use (in viewport render case e.g.). + * Note that we could use a pool to avoid lots of malloc's here, but... not really a problem for now. */ + BezTriple *origbezt_cpy = MEM_mallocN(sizeof(*origbezt), __func__); + *origbezt_cpy = *origbezt; + keyIndex = init_cvKeyIndex(origbezt_cpy, key_index, nu_index, pt_index, vertex_index); BLI_ghash_insert(gh, bezt, keyIndex); key_index += 12; vertex_index += 3; @@ -189,7 +185,12 @@ static void init_editNurb_keyIndex(EditNurb *editnurb, ListBase *origBase) origbp = orignu->bp; pt_index = 0; while (a--) { - keyIndex = init_cvKeyIndex(origbp, key_index, nu_index, pt_index, vertex_index, orignu); + /* We cannot keep *any* reference to curve obdata, + * it might be replaced and freed while editcurve remain in use (in viewport render case e.g.). + * Note that we could use a pool to avoid lots of malloc's here, but... not really a problem for now. */ + BPoint *origbp_cpy = MEM_mallocN(sizeof(*origbp_cpy), __func__); + *origbp_cpy = *origbp; + keyIndex = init_cvKeyIndex(origbp_cpy, key_index, nu_index, pt_index, vertex_index); BLI_ghash_insert(gh, bp, keyIndex); key_index += 4; bp++; @@ -250,23 +251,22 @@ static int getKeyIndexOrig_keyIndex(EditNurb *editnurb, void *cv) return index->key_index; } -static void keyIndex_delCV(EditNurb *editnurb, const void *cv) +static void keyIndex_delBezt(EditNurb *editnurb, BezTriple *bezt) { if (!editnurb->keyindex) { return; } - BLI_ghash_remove(editnurb->keyindex, cv, NULL, MEM_freeN); -} - -static void keyIndex_delBezt(EditNurb *editnurb, BezTriple *bezt) -{ - keyIndex_delCV(editnurb, bezt); + BKE_curve_editNurb_keyIndex_delCV(editnurb->keyindex, bezt); } static void keyIndex_delBP(EditNurb *editnurb, BPoint *bp) { - keyIndex_delCV(editnurb, bp); + if (!editnurb->keyindex) { + return; + } + + BKE_curve_editNurb_keyIndex_delCV(editnurb->keyindex, bp); } static void keyIndex_delNurb(EditNurb *editnurb, Nurb *nu) @@ -282,7 +282,7 @@ static void keyIndex_delNurb(EditNurb *editnurb, Nurb *nu) a = nu->pntsu; while (a--) { - BLI_ghash_remove(editnurb->keyindex, bezt, NULL, MEM_freeN); + BKE_curve_editNurb_keyIndex_delCV(editnurb->keyindex, bezt); bezt++; } } @@ -291,7 +291,7 @@ static void keyIndex_delNurb(EditNurb *editnurb, Nurb *nu) a = nu->pntsu * nu->pntsv; while (a--) { - BLI_ghash_remove(editnurb->keyindex, bp, NULL, MEM_freeN); + BKE_curve_editNurb_keyIndex_delCV(editnurb->keyindex, bp); bp++; } } @@ -535,6 +535,7 @@ static GHash *dupli_keyIndexHash(GHash *keyindex) CVKeyIndex *newIndex = MEM_mallocN(sizeof(CVKeyIndex), "dupli_keyIndexHash index"); memcpy(newIndex, index, sizeof(CVKeyIndex)); + newIndex->orig_cv = MEM_dupallocN(index->orig_cv); BLI_ghash_insert(gh, cv, newIndex); } @@ -624,7 +625,7 @@ static void calc_keyHandles(ListBase *nurb, float *key) } } -static void calc_shapeKeys(Object *obedit) +static void calc_shapeKeys(Object *obedit, ListBase *newnurbs) { Curve *cu = (Curve *)obedit->data; @@ -636,7 +637,7 @@ static void calc_shapeKeys(Object *obedit) KeyBlock *actkey = BLI_findlink(&cu->key->block, editnurb->shapenr - 1); BezTriple *bezt, *oldbezt; BPoint *bp, *oldbp; - Nurb *nu; + Nurb *nu, *newnu; int totvert = BKE_nurbList_verts_count(&editnurb->nurbs); float (*ofs)[3] = NULL; @@ -706,20 +707,25 @@ static void calc_shapeKeys(Object *obedit) currkey = cu->key->block.first; while (currkey) { - int apply_offset = (ofs && (currkey != actkey) && (editnurb->shapenr - 1 == currkey->relative)); + const bool apply_offset = (ofs && (currkey != actkey) && (editnurb->shapenr - 1 == currkey->relative)); float *fp = newkey = MEM_callocN(cu->key->elemsize * totvert, "currkey->data"); ofp = oldkey = currkey->data; nu = editnurb->nurbs.first; + /* We need to restore to original curve into newnurb, *not* editcurve's nurbs. + * Otherwise, in case we update obdata *without* leaving editmode (e.g. viewport render), we would + * invalidate editcurve. */ + newnu = newnurbs->first; i = 0; while (nu) { if (currkey == actkey) { - int restore = actkey != cu->key->refkey; + const bool restore = actkey != cu->key->refkey; if (nu->bezt) { bezt = nu->bezt; a = nu->pntsu; + BezTriple *newbezt = newnu->bezt; while (a--) { int j; oldbezt = getKeyIndexOrig_bezt(editnurb, bezt); @@ -728,7 +734,7 @@ static void calc_shapeKeys(Object *obedit) copy_v3_v3(fp, bezt->vec[j]); if (restore && oldbezt) { - copy_v3_v3(bezt->vec[j], oldbezt->vec[j]); + copy_v3_v3(newbezt->vec[j], oldbezt->vec[j]); } fp += 3; @@ -736,16 +742,18 @@ static void calc_shapeKeys(Object *obedit) fp[0] = bezt->alfa; if (restore && oldbezt) { - bezt->alfa = oldbezt->alfa; + newbezt->alfa = oldbezt->alfa; } fp += 3; ++i; /* alphas */ bezt++; + newbezt++; } } else { bp = nu->bp; a = nu->pntsu * nu->pntsv; + BPoint *newbp = newnu->bp; while (a--) { oldbp = getKeyIndexOrig_bp(editnurb, bp); @@ -754,12 +762,13 @@ static void calc_shapeKeys(Object *obedit) fp[3] = bp->alfa; if (restore && oldbp) { - copy_v3_v3(bp->vec, oldbp->vec); - bp->alfa = oldbp->alfa; + copy_v3_v3(newbp->vec, oldbp->vec); + newbp->alfa = oldbp->alfa; } fp += 4; bp++; + newbp++; i += 2; } } @@ -1204,9 +1213,13 @@ void ED_curve_editnurb_load(Object *obedit) } } + /* We have to pass also new copied nurbs, since we want to restore original curve (without edited shapekey) + * on obdata, but *not* on editcurve itself (ED_curve_editnurb_load call does not always implies freeing + * of editcurve, e.g. when called to generate render data...). */ + calc_shapeKeys(obedit, &newnurb); + cu->nurb = newnurb; - calc_shapeKeys(obedit); ED_curve_updateAnimPaths(obedit->data); BKE_nurbList_free(&oldnurb); @@ -1227,13 +1240,11 @@ void ED_curve_editnurb_make(Object *obedit) if (actkey) { // XXX strcpy(G.editModeTitleExtra, "(Key) "); undo_editmode_clear(); - BKE_keyblock_convert_to_curve(actkey, cu, &cu->nurb); } if (editnurb) { BKE_nurbList_free(&editnurb->nurbs); - BKE_curve_editNurb_keyIndex_free(editnurb); - editnurb->keyindex = NULL; + BKE_curve_editNurb_keyIndex_free(&editnurb->keyindex); } else { editnurb = MEM_callocN(sizeof(EditNurb), "editnurb"); @@ -1248,12 +1259,16 @@ void ED_curve_editnurb_make(Object *obedit) nu = nu->next; } - if (actkey) - editnurb->shapenr = obedit->shapenr; - /* animation could be added in editmode even if there was no animdata in * object mode hence we always need CVs index be created */ init_editNurb_keyIndex(editnurb, &cu->nurb); + + if (actkey) { + editnurb->shapenr = obedit->shapenr; + /* Apply shapekey to new nurbs of editnurb, not those of original curve (and *after* we generated keyIndex), + * else we do not have valid 'original' data to properly restore curve when leaving editmode. */ + BKE_keyblock_convert_to_curve(actkey, cu, &editnurb->nurbs); + } } } @@ -1309,8 +1324,7 @@ static int separate_exec(bContext *C, wmOperator *op) ED_curve_editnurb_make(newob); newedit = newcu->editnurb; BKE_nurbList_free(&newedit->nurbs); - BKE_curve_editNurb_keyIndex_free(newedit); - newedit->keyindex = NULL; + BKE_curve_editNurb_keyIndex_free(&newedit->keyindex); BLI_movelisttolist(&newedit->nurbs, &newnurb); /* 4. put old object out of editmode and delete separated geometry */ @@ -6110,7 +6124,7 @@ static void undoCurve_to_editCurve(void *ucu, void *UNUSED(edata), void *cu_v) BKE_nurbList_free(editbase); if (undoCurve->undoIndex) { - BLI_ghash_free(editnurb->keyindex, NULL, MEM_freeN); + BKE_curve_editNurb_keyIndex_free(&editnurb->keyindex); editnurb->keyindex = dupli_keyIndexHash(undoCurve->undoIndex); } @@ -6188,8 +6202,7 @@ static void free_undoCurve(void *ucv) BKE_nurbList_free(&undoCurve->nubase); - if (undoCurve->undoIndex) - BLI_ghash_free(undoCurve->undoIndex, NULL, MEM_freeN); + BKE_curve_editNurb_keyIndex_free(&undoCurve->undoIndex); free_fcurves(&undoCurve->fcurves); free_fcurves(&undoCurve->drivers); diff --git a/source/blender/editors/gpencil/gpencil_edit.c b/source/blender/editors/gpencil/gpencil_edit.c index 5b011b679a6..601a86b97cb 100644 --- a/source/blender/editors/gpencil/gpencil_edit.c +++ b/source/blender/editors/gpencil/gpencil_edit.c @@ -74,7 +74,6 @@ #include "ED_object.h" #include "ED_screen.h" #include "ED_view3d.h" -#include "ED_screen.h" #include "ED_space_api.h" #include "gpencil_intern.h" @@ -953,6 +952,9 @@ static int gp_dissolve_selected_points(bContext *C) /* skip strokes that are invalid for current view */ if (ED_gpencil_stroke_can_use(C, gps) == false) continue; + /* check if the color is editable */ + if (ED_gpencil_stroke_color_use(gpl, gps) == false) + continue; if (gps->flag & GP_STROKE_SELECT) { bGPDspoint *pt; @@ -1165,6 +1167,9 @@ static int gp_delete_selected_points(bContext *C) /* skip strokes that are invalid for current view */ if (ED_gpencil_stroke_can_use(C, gps) == false) continue; + /* check if the color is editable */ + if (ED_gpencil_stroke_color_use(gpl, gps) == false) + continue; if (gps->flag & GP_STROKE_SELECT) { @@ -1204,7 +1209,7 @@ static int gp_delete_exec(bContext *C, wmOperator *op) case GP_DELETEOP_POINTS: /* selected points (breaks the stroke into segments) */ result = gp_delete_selected_points(C); break; - + case GP_DELETEOP_FRAME: /* active frame */ result = gp_actframe_delete_exec(C, op); break; @@ -2121,10 +2126,10 @@ static int gp_count_subdivision_cuts(bGPDstroke *gps) int totnewpoints = 0; for (i = 0, pt = gps->points; i < gps->totpoints && pt; i++, pt++) { if (pt->flag & GP_SPOINT_SELECT) { - if (i + 1 < gps->totpoints){ + if (i + 1 < gps->totpoints) { if (gps->points[i + 1].flag & GP_SPOINT_SELECT) { ++totnewpoints; - }; + } } } } @@ -2179,7 +2184,7 @@ static int gp_stroke_subdivide_exec(bContext *C, wmOperator *op) /* if next point is selected add a half way point */ if (pt->flag & GP_SPOINT_SELECT) { - if (i + 1 < oldtotpoints){ + if (i + 1 < oldtotpoints) { if (temp_points[i + 1].flag & GP_SPOINT_SELECT) { pt_final = &gps->points[i2]; /* Interpolate all values */ @@ -2191,7 +2196,7 @@ static int gp_stroke_subdivide_exec(bContext *C, wmOperator *op) pt_final->time = interpf(pt->time, next->time, 0.5f); pt_final->flag |= GP_SPOINT_SELECT; ++i2; - }; + } } } } diff --git a/source/blender/editors/gpencil/gpencil_interpolate.c b/source/blender/editors/gpencil/gpencil_interpolate.c index 297058168a0..d2360fea672 100644 --- a/source/blender/editors/gpencil/gpencil_interpolate.c +++ b/source/blender/editors/gpencil/gpencil_interpolate.c @@ -86,13 +86,14 @@ /* Core/Shared Utilities */ /* Poll callback for interpolation operators */ -static int gpencil_interpolate_poll(bContext *C) +static int gpencil_view3d_poll(bContext *C) { bGPdata *gpd = CTX_data_gpencil_data(C); bGPDlayer *gpl = CTX_data_active_gpencil_layer(C); /* only 3D view */ - if (CTX_wm_area(C)->spacetype != SPACE_VIEW3D) { + ScrArea *sa = CTX_wm_area(C); + if (sa && sa->spacetype != SPACE_VIEW3D) { return 0; } @@ -673,7 +674,7 @@ void GPENCIL_OT_interpolate(wmOperatorType *ot) ot->invoke = gpencil_interpolate_invoke; ot->modal = gpencil_interpolate_modal; ot->cancel = gpencil_interpolate_cancel; - ot->poll = gpencil_interpolate_poll; + ot->poll = gpencil_view3d_poll; /* flags */ ot->flag = OPTYPE_UNDO | OPTYPE_BLOCKING; @@ -1017,7 +1018,7 @@ void GPENCIL_OT_interpolate_sequence(wmOperatorType *ot) /* api callbacks */ ot->exec = gpencil_interpolate_seq_exec; - ot->poll = gpencil_interpolate_poll; + ot->poll = gpencil_view3d_poll; /* flags */ ot->flag = OPTYPE_REGISTER | OPTYPE_UNDO; @@ -1025,24 +1026,14 @@ void GPENCIL_OT_interpolate_sequence(wmOperatorType *ot) /* ******************** Remove Breakdowns ************************ */ -/* Same as gpencil_interpolate_poll(), - * except we ALSO need to have an active frame that is a breakdown - */ static int gpencil_interpolate_reverse_poll(bContext *C) { - bGPdata *gpd = CTX_data_gpencil_data(C); - bGPDlayer *gpl = CTX_data_active_gpencil_layer(C); - - /* only 3D view */ - if (CTX_wm_area(C)->spacetype != SPACE_VIEW3D) { - return 0; - } - - /* need data to interpolate */ - if (ELEM(NULL, gpd, gpl)) { + if (!gpencil_view3d_poll(C)) { return 0; } - + + bGPDlayer *gpl = CTX_data_active_gpencil_layer(C); + /* need to be on a breakdown frame */ if ((gpl->actframe == NULL) || (gpl->actframe->key_type != BEZT_KEYTYPE_BREAKDOWN)) { CTX_wm_operator_poll_msg_set(C, "Expected current frame to be a breakdown"); diff --git a/source/blender/editors/include/ED_armature.h b/source/blender/editors/include/ED_armature.h index 7ad61671b1b..6b8943421bd 100644 --- a/source/blender/editors/include/ED_armature.h +++ b/source/blender/editors/include/ED_armature.h @@ -130,8 +130,9 @@ void ED_armature_ebone_listbase_temp_clear(struct ListBase *lb); void ED_armature_deselect_all(struct Object *obedit); void ED_armature_deselect_all_visible(struct Object *obedit); -int ED_do_pose_selectbuffer(struct Scene *scene, struct Base *base, unsigned int *buffer, - short hits, bool extend, bool deselect, bool toggle, bool do_nearest); +bool ED_do_pose_selectbuffer( + struct Scene *scene, struct Base *base, const unsigned int *buffer, short hits, + bool extend, bool deselect, bool toggle, bool do_nearest); bool ED_armature_select_pick(struct bContext *C, const int mval[2], bool extend, bool deselect, bool toggle); int join_armature_exec(struct bContext *C, struct wmOperator *op); struct Bone *get_indexed_bone(struct Object *ob, int index); diff --git a/source/blender/editors/include/ED_view3d.h b/source/blender/editors/include/ED_view3d.h index 79176d9e9cf..9ce71258573 100644 --- a/source/blender/editors/include/ED_view3d.h +++ b/source/blender/editors/include/ED_view3d.h @@ -47,6 +47,7 @@ struct Main; struct MetaElem; struct Nurb; struct Object; +struct RV3DMatrixStore; struct RegionView3D; struct Scene; struct ScrArea; @@ -301,7 +302,22 @@ bool ED_view3d_autodist_depth_seg(struct ARegion *ar, const int mval_sta[2], con /* select */ #define MAXPICKELEMS 2500 #define MAXPICKBUF (4 * MAXPICKELEMS) -short view3d_opengl_select(struct ViewContext *vc, unsigned int *buffer, unsigned int bufsize, const struct rcti *input, bool do_nearest); + +typedef enum { + /* all elements in the region, ignore depth */ + VIEW3D_SELECT_ALL = 0, + /* pick also depth sorts (only for small regions!) */ + VIEW3D_SELECT_PICK_ALL = 1, + /* sorts and only returns visible objects (only for small regions!) */ + VIEW3D_SELECT_PICK_NEAREST = 2, +} eV3DSelectMode; + +void view3d_opengl_select_cache_begin(void); +void view3d_opengl_select_cache_end(void); + +int view3d_opengl_select( + struct ViewContext *vc, unsigned int *buffer, unsigned int bufsize, const struct rcti *input, + eV3DSelectMode select_mode); /* view3d_select.c */ float ED_view3d_select_dist_px(void); @@ -330,8 +346,8 @@ void ED_view3d_check_mats_rv3d(struct RegionView3D *rv3d); #endif int ED_view3d_scene_layer_set(int lay, const int *values, int *active); -void *ED_view3d_mats_rv3d_backup(struct RegionView3D *rv3d); -void ED_view3d_mats_rv3d_restore(struct RegionView3D *rv3d, void *rv3dmat_pt); +struct RV3DMatrixStore *ED_view3d_mats_rv3d_backup(struct RegionView3D *rv3d); +void ED_view3d_mats_rv3d_restore(struct RegionView3D *rv3d, struct RV3DMatrixStore *rv3dmat); bool ED_view3d_context_activate(struct bContext *C); void ED_view3d_draw_offscreen_init(struct Scene *scene, struct View3D *v3d); diff --git a/source/blender/editors/include/UI_icons.h b/source/blender/editors/include/UI_icons.h index 8579778ff79..8420591aa3e 100644 --- a/source/blender/editors/include/UI_icons.h +++ b/source/blender/editors/include/UI_icons.h @@ -653,9 +653,9 @@ DEF_ICON(IPO_BACK) DEF_ICON(IPO_EASE_IN) DEF_ICON(IPO_EASE_OUT) DEF_ICON(IPO_EASE_IN_OUT) +DEF_ICON(NORMALIZE_FCURVES) #ifndef DEF_ICON_BLANK_SKIP /* available */ - DEF_ICON(BLANK203) DEF_ICON(BLANK204) DEF_ICON(BLANK205) DEF_ICON(BLANK206) diff --git a/source/blender/editors/include/UI_interface.h b/source/blender/editors/include/UI_interface.h index 9fbce7dd203..2ce4f3e2790 100644 --- a/source/blender/editors/include/UI_interface.h +++ b/source/blender/editors/include/UI_interface.h @@ -421,7 +421,7 @@ typedef void (*uiBlockCancelFunc)(struct bContext *C, void *arg1); void UI_popup_block_invoke(struct bContext *C, uiBlockCreateFunc func, void *arg); void UI_popup_block_invoke_ex(struct bContext *C, uiBlockCreateFunc func, void *arg, const char *opname, int opcontext); -void UI_popup_block_ex(struct bContext *C, uiBlockCreateFunc func, uiBlockHandleFunc popup_func, uiBlockCancelFunc cancel_func, void *arg); +void UI_popup_block_ex(struct bContext *C, uiBlockCreateFunc func, uiBlockHandleFunc popup_func, uiBlockCancelFunc cancel_func, void *arg, struct wmOperator *op); /* void uiPupBlockOperator(struct bContext *C, uiBlockCreateFunc func, struct wmOperator *op, int opcontext); */ /* UNUSED */ void UI_popup_block_close(struct bContext *C, struct wmWindow *win, uiBlock *block); diff --git a/source/blender/editors/interface/interface.c b/source/blender/editors/interface/interface.c index 682db20af55..7180e18ab92 100644 --- a/source/blender/editors/interface/interface.c +++ b/source/blender/editors/interface/interface.c @@ -1337,7 +1337,7 @@ static void ui_but_to_pixelrect(rcti *rect, const ARegion *ar, uiBlock *block, u rctf rectf; ui_block_to_window_rctf(ar, block, &rectf, (but) ? &but->rect : &block->rect); - BLI_rcti_rctf_copy_floor(rect, &rectf); + BLI_rcti_rctf_copy(rect, &rectf); BLI_rcti_translate(rect, -ar->winrct.xmin, -ar->winrct.ymin); } @@ -2288,7 +2288,7 @@ char *ui_but_string_get_dynamic(uiBut *but, int *r_str_size) #ifdef WITH_PYTHON -static bool ui_set_but_string_eval_num_unit(bContext *C, uiBut *but, const char *str, double *value) +static bool ui_set_but_string_eval_num_unit(bContext *C, uiBut *but, const char *str, double *r_value) { char str_unit_convert[256]; const int unit_type = UI_but_unit_type_get(but); @@ -2300,13 +2300,13 @@ static bool ui_set_but_string_eval_num_unit(bContext *C, uiBut *but, const char bUnit_ReplaceString(str_unit_convert, sizeof(str_unit_convert), but->drawstr, ui_get_but_scale_unit(but, 1.0), but->block->unit->system, RNA_SUBTYPE_UNIT_VALUE(unit_type)); - return BPY_execute_string_as_number(C, str_unit_convert, value, true); + return BPY_execute_string_as_number(C, str_unit_convert, true, r_value); } #endif /* WITH_PYTHON */ -bool ui_but_string_set_eval_num(bContext *C, uiBut *but, const char *str, double *value) +bool ui_but_string_set_eval_num(bContext *C, uiBut *but, const char *str, double *r_value) { bool ok = false; @@ -2315,13 +2315,13 @@ bool ui_but_string_set_eval_num(bContext *C, uiBut *but, const char *str, double if (str[0] != '\0') { bool is_unit_but = (ui_but_is_float(but) && ui_but_is_unit(but)); /* only enable verbose if we won't run again with units */ - if (BPY_execute_string_as_number(C, str, value, is_unit_but == false)) { + if (BPY_execute_string_as_number(C, str, is_unit_but == false, r_value)) { /* if the value parsed ok without unit conversion this button may still need a unit multiplier */ if (is_unit_but) { char str_new[128]; - BLI_snprintf(str_new, sizeof(str_new), "%f", *value); - ok = ui_set_but_string_eval_num_unit(C, but, str_new, value); + BLI_snprintf(str_new, sizeof(str_new), "%f", *r_value); + ok = ui_set_but_string_eval_num_unit(C, but, str_new, r_value); } else { ok = true; /* parse normal string via py (no unit conversion needed) */ @@ -2329,7 +2329,7 @@ bool ui_but_string_set_eval_num(bContext *C, uiBut *but, const char *str, double } else if (is_unit_but) { /* parse failed, this is a unit but so run replacements and parse again */ - ok = ui_set_but_string_eval_num_unit(C, but, str, value); + ok = ui_set_but_string_eval_num_unit(C, but, str, r_value); } } @@ -3913,6 +3913,8 @@ uiBut *uiDefIconTextButO_ptr(uiBlock *block, int type, wmOperatorType *ot, int o uiBut *uiDefIconTextButO(uiBlock *block, int type, const char *opname, int opcontext, int icon, const char *str, int x, int y, short width, short height, const char *tip) { wmOperatorType *ot = WM_operatortype_find(opname, 0); + if (str && str[0] == '\0') + return uiDefIconButO_ptr(block, type, ot, opcontext, icon, x, y, width, height, tip); return uiDefIconTextButO_ptr(block, type, ot, opcontext, icon, str, x, y, width, height, tip); } diff --git a/source/blender/editors/interface/interface_anim.c b/source/blender/editors/interface/interface_anim.c index 5da294302e9..a04360b3395 100644 --- a/source/blender/editors/interface/interface_anim.c +++ b/source/blender/editors/interface/interface_anim.c @@ -39,6 +39,7 @@ #include "BLI_utildefines.h" #include "BKE_context.h" +#include "BKE_depsgraph.h" #include "BKE_fcurve.h" #include "BKE_global.h" #include "BKE_nla.h" @@ -210,6 +211,7 @@ bool ui_but_anim_expression_create(uiBut *but, const char *str) /* updates */ driver->flag |= DRIVER_FLAG_RECOMPILE; + DAG_relations_tag_update(CTX_data_main(C)); WM_event_add_notifier(C, NC_ANIMATION | ND_KEYFRAME, NULL); ok = true; } diff --git a/source/blender/editors/interface/interface_eyedropper.c b/source/blender/editors/interface/interface_eyedropper.c index 5154a77ad21..f3859154dfb 100644 --- a/source/blender/editors/interface/interface_eyedropper.c +++ b/source/blender/editors/interface/interface_eyedropper.c @@ -1082,7 +1082,7 @@ static int depthdropper_poll(bContext *C) return 1; } } - else { + else { RegionView3D *rv3d = CTX_wm_region_view3d(C); if (rv3d && rv3d->persp == RV3D_CAMOB) { View3D *v3d = CTX_wm_view3d(C); diff --git a/source/blender/editors/interface/interface_handlers.c b/source/blender/editors/interface/interface_handlers.c index 734cd02a056..2974c2e9304 100644 --- a/source/blender/editors/interface/interface_handlers.c +++ b/source/blender/editors/interface/interface_handlers.c @@ -87,6 +87,7 @@ #include "WM_api.h" #include "WM_types.h" +#include "wm_event_system.h" #ifdef WITH_INPUT_IME # include "wm_window.h" @@ -380,6 +381,7 @@ typedef struct uiAfterFunc { void *butm_func_arg; int a2; + wmOperator *popup_op; wmOperatorType *optype; int opcontext; PointerRNA *opptr; @@ -635,13 +637,24 @@ PointerRNA *ui_handle_afterfunc_add_operator(wmOperatorType *ot, int opcontext, return ptr; } +static void popup_check(bContext *C, wmOperator *op) +{ + if (op && op->type->check && op->type->check(C, op)) { + /* check for popup and re-layout buttons */ + ARegion *ar_menu = CTX_wm_menu(C); + if (ar_menu) + ED_region_tag_refresh_ui(ar_menu); + } +} + /** * Check if a #uiAfterFunc is needed for this button. */ static bool ui_afterfunc_check(const uiBlock *block, const uiBut *but) { return (but->func || but->funcN || but->rename_func || but->optype || but->rnaprop || block->handle_func || - (but->type == UI_BTYPE_BUT_MENU && block->butm_func)); + (but->type == UI_BTYPE_BUT_MENU && block->butm_func) || + (block->handle && block->handle->popup_op)); } static void ui_apply_but_func(bContext *C, uiBut *but) @@ -682,6 +695,9 @@ static void ui_apply_but_func(bContext *C, uiBut *but) after->butm_func_arg = block->butm_func_arg; after->a2 = but->a2; } + + if (block->handle) + after->popup_op = block->handle->popup_op; after->optype = but->optype; after->opcontext = but->opcontext; @@ -766,6 +782,9 @@ static void ui_apply_but_funcs_after(bContext *C) if (after.context) CTX_store_set(C, after.context); + if (after.popup_op) + popup_check(C, after.popup_op); + if (after.opptr) { /* free in advance to avoid leak on exit */ opptr = *after.opptr; @@ -6665,7 +6684,7 @@ static void remove_shortcut_func(bContext *C, void *arg1, void *UNUSED(arg2)) static void popup_add_shortcut_func(bContext *C, void *arg1, void *UNUSED(arg2)) { uiBut *but = (uiBut *)arg1; - UI_popup_block_ex(C, menu_add_shortcut, NULL, menu_add_shortcut_cancel, but); + UI_popup_block_ex(C, menu_add_shortcut, NULL, menu_add_shortcut_cancel, but, NULL); } /** @@ -7005,20 +7024,17 @@ static bool ui_but_menu(bContext *C, uiBut *but) uiItemO(layout, CTX_IFACE_(BLT_I18NCONTEXT_OPERATOR_DEFAULT, "Online Manual"), ICON_URL, "WM_OT_doc_view_manual_ui_context"); - WM_operator_properties_create(&ptr_props, "WM_OT_doc_view"); + ptr_props = uiItemFullO(layout, "WM_OT_doc_view", + CTX_IFACE_(BLT_I18NCONTEXT_OPERATOR_DEFAULT, "Online Python Reference"), + ICON_NONE, NULL, WM_OP_EXEC_DEFAULT, UI_ITEM_O_RETURN_PROPS); RNA_string_set(&ptr_props, "doc_id", buf); - uiItemFullO(layout, "WM_OT_doc_view", - CTX_IFACE_(BLT_I18NCONTEXT_OPERATOR_DEFAULT, "Online Python Reference"), - ICON_NONE, ptr_props.data, WM_OP_EXEC_DEFAULT, 0); /* XXX inactive option, not for public! */ #if 0 - WM_operator_properties_create(&ptr_props, "WM_OT_doc_edit"); + ptr_props = uiItemFullO(layout, "WM_OT_doc_edit", + "Submit Description", ICON_NONE, NULL, WM_OP_INVOKE_DEFAULT, UI_ITEM_O_RETURN_PROPS); RNA_string_set(&ptr_props, "doc_id", buf); RNA_string_set(&ptr_props, "doc_new", RNA_property_description(but->rnaprop)); - - uiItemFullO(layout, "WM_OT_doc_edit", - "Submit Description", ICON_NONE, ptr_props.data, WM_OP_INVOKE_DEFAULT, 0); #endif } } @@ -7741,7 +7757,8 @@ static void button_activate_state(bContext *C, uiBut *but, uiHandleButtonState s if (ui_but_is_cursor_warp(but)) { #ifdef USE_CONT_MOUSE_CORRECT - if (data->ungrab_mval[0] != FLT_MAX) { + /* stereo3d has issues with changing cursor location so rather avoid */ + if (data->ungrab_mval[0] != FLT_MAX && !WM_stereo3d_enabled(data->window, false)) { int mouse_ungrab_xy[2]; ui_block_to_window_fl(data->region, but->block, &data->ungrab_mval[0], &data->ungrab_mval[1]); mouse_ungrab_xy[0] = data->ungrab_mval[0]; @@ -10181,6 +10198,25 @@ void UI_popup_handlers_add(bContext *C, ListBase *handlers, uiPopupBlockHandle * void UI_popup_handlers_remove(ListBase *handlers, uiPopupBlockHandle *popup) { + wmEventHandler *handler; + + for (handler = handlers->first; handler; handler = handler->next) { + if (handler->ui_handle == ui_popup_handler && + handler->ui_remove == ui_popup_handler_remove && + handler->ui_userdata == popup) + { + /* tag refresh parent popup */ + if (handler->next && + handler->next->ui_handle == ui_popup_handler && + handler->next->ui_remove == ui_popup_handler_remove) + { + uiPopupBlockHandle *parent_popup = handler->next->ui_userdata; + ED_region_tag_refresh_ui(parent_popup->region); + } + break; + } + } + WM_event_remove_ui_handler(handlers, ui_popup_handler, ui_popup_handler_remove, popup, false); } diff --git a/source/blender/editors/interface/interface_intern.h b/source/blender/editors/interface/interface_intern.h index d8f9fdcbaae..067279777ba 100644 --- a/source/blender/editors/interface/interface_intern.h +++ b/source/blender/editors/interface/interface_intern.h @@ -556,6 +556,7 @@ struct uiPopupBlockHandle { struct uiKeyNavLock keynav_state; /* for operator popups */ + struct wmOperator *popup_op; struct wmOperatorType *optype; ScrArea *ctx_area; ARegion *ctx_region; diff --git a/source/blender/editors/interface/interface_layout.c b/source/blender/editors/interface/interface_layout.c index ca2538022b0..9b6547cf8a1 100644 --- a/source/blender/editors/interface/interface_layout.c +++ b/source/blender/editors/interface/interface_layout.c @@ -189,7 +189,7 @@ static const char *ui_item_name_add_colon(const char *name, char namestr[UI_MAX_ return name; } -static int ui_item_fit(int item, int pos, int all, int available, bool is_last, int alignment) +static int ui_item_fit(int item, int pos, int all, int available, bool is_last, int alignment, float *extra_pixel) { /* available == 0 is unlimited */ if (available == 0) @@ -199,16 +199,22 @@ static int ui_item_fit(int item, int pos, int all, int available, bool is_last, /* contents is bigger than available space */ if (is_last) return available - pos; - else - return (item * available) / all; + else { + float width = *extra_pixel + (item * available) / (float)all; + *extra_pixel = width - (int)width; + return (int)width; + } } else { /* contents is smaller or equal to available space */ if (alignment == UI_LAYOUT_ALIGN_EXPAND) { if (is_last) return available - pos; - else - return (item * available) / all; + else { + float width = *extra_pixel + (item * available) / (float)all; + *extra_pixel = width - (int)width; + return (int)width; + } } else return item; @@ -302,6 +308,26 @@ static void ui_item_position(uiItem *item, int x, int y, int w, int h) } } +static void ui_item_move(uiItem *item, int delta_xmin, int delta_xmax) +{ + if (item->type == ITEM_BUTTON) { + uiButtonItem *bitem = (uiButtonItem *)item; + + bitem->but->rect.xmin += delta_xmin; + bitem->but->rect.xmax += delta_xmax; + + ui_but_update(bitem->but); /* for strlen */ + } + else { + uiLayout *litem = (uiLayout *)item; + + if (delta_xmin > 0) + litem->x += delta_xmin; + else + litem->w += delta_xmax; + } +} + /******************** Special RNA Items *********************/ static int ui_layout_local_dir(uiLayout *layout) @@ -2099,9 +2125,10 @@ static int ui_litem_min_width(int itemw) static void ui_litem_layout_row(uiLayout *litem) { - uiItem *item; + uiItem *item, *last_free_item = NULL; int x, y, w, tot, totw, neww, newtotw, itemw, minw, itemh, offset; int fixedw, freew, fixedx, freex, flag = 0, lastw = 0; + float extra_pixel; /* x = litem->x; */ /* UNUSED */ y = litem->y; @@ -2128,6 +2155,7 @@ static void ui_litem_layout_row(uiLayout *litem) x = 0; flag = 0; newtotw = totw; + extra_pixel = 0.0f; for (item = litem->items.first; item; item = item->next) { if (item->flag & UI_ITEM_FIXED) @@ -2137,13 +2165,23 @@ static void ui_litem_layout_row(uiLayout *litem) minw = ui_litem_min_width(itemw); if (w - lastw > 0) - neww = ui_item_fit(itemw, x, totw, w - lastw, !item->next, litem->alignment); + neww = ui_item_fit(itemw, x, totw, w - lastw, !item->next, litem->alignment, &extra_pixel); else neww = 0; /* no space left, all will need clamping to minimum size */ x += neww; - if ((neww < minw || itemw == minw || item->flag & UI_ITEM_MIN) && w != 0) { + bool min_flag = item->flag & UI_ITEM_MIN; + /* ignore min flag for rows with right or center alignment */ + if (item->type != ITEM_BUTTON && + ELEM(((uiLayout *)item)->alignment, UI_LAYOUT_ALIGN_RIGHT, UI_LAYOUT_ALIGN_CENTER) && + litem->alignment == UI_LAYOUT_ALIGN_EXPAND && + ((uiItem *)litem)->flag & UI_ITEM_MIN) + { + min_flag = false; + } + + if ((neww < minw || min_flag) && w != 0) { /* fixed size */ item->flag |= UI_ITEM_FIXED; if (item->type != ITEM_BUTTON && item->flag & UI_ITEM_MIN) { @@ -2166,6 +2204,7 @@ static void ui_litem_layout_row(uiLayout *litem) freex = 0; fixedx = 0; + extra_pixel = 0.0f; x = litem->x; for (item = litem->items.first; item; item = item->next) { @@ -2177,13 +2216,14 @@ static void ui_litem_layout_row(uiLayout *litem) if (item->type != ITEM_BUTTON && item->flag & UI_ITEM_MIN) { minw = itemw; } - itemw = ui_item_fit(minw, fixedx, fixedw, min_ii(w, fixedw), !item->next, litem->alignment); + itemw = ui_item_fit(minw, fixedx, fixedw, min_ii(w, fixedw), !item->next, litem->alignment, &extra_pixel); fixedx += itemw; } else { /* free size item */ - itemw = ui_item_fit(itemw, freex, freew, w - fixedw, !item->next, litem->alignment); + itemw = ui_item_fit(itemw, freex, freew, w - fixedw, !item->next, litem->alignment, &extra_pixel); freex += itemw; + last_free_item = item; } /* align right/center */ @@ -2205,6 +2245,17 @@ static void ui_litem_layout_row(uiLayout *litem) x += litem->space; } + /* add extra pixel */ + uiItem *last_item = litem->items.last; + extra_pixel = litem->w - (x - litem->x); + if (extra_pixel > 0 && litem->alignment == UI_LAYOUT_ALIGN_EXPAND && + last_free_item && last_item && last_item->flag & UI_ITEM_FIXED) + { + ui_item_move(last_free_item, 0, extra_pixel); + for (item = last_free_item->next; item; item = item->next) + ui_item_move(item, extra_pixel, extra_pixel); + } + litem->w = x - litem->x; litem->h = litem->y - y; litem->x = x; @@ -2212,7 +2263,7 @@ static void ui_litem_layout_row(uiLayout *litem) } /* single-column layout */ -static void ui_litem_estimate_column(uiLayout *litem) +static void ui_litem_estimate_column(uiLayout *litem, bool is_box) { uiItem *item; int itemw, itemh; @@ -2229,16 +2280,16 @@ static void ui_litem_estimate_column(uiLayout *litem) litem->w = MAX2(litem->w, itemw); litem->h += itemh; - if (item->next) + if (item->next && (!is_box || item != litem->items.first)) litem->h += litem->space; } - + if (min_size_flag) { litem->item.flag |= UI_ITEM_MIN; } } -static void ui_litem_layout_column(uiLayout *litem) +static void ui_litem_layout_column(uiLayout *litem, bool is_box) { uiItem *item; int itemh, x, y; @@ -2252,7 +2303,7 @@ static void ui_litem_layout_column(uiLayout *litem) y -= itemh; ui_item_position(item, x, y, litem->w, itemh); - if (item->next) + if (item->next && (!is_box || item != litem->items.first)) y -= litem->space; } @@ -2394,7 +2445,7 @@ static void ui_litem_layout_root(uiLayout *litem) else if (litem->root->type == UI_LAYOUT_PIEMENU) ui_litem_layout_root_radial(litem); else - ui_litem_layout_column(litem); + ui_litem_layout_column(litem, false); } /* box layout */ @@ -2402,9 +2453,10 @@ static void ui_litem_estimate_box(uiLayout *litem) { uiStyle *style = litem->root->style; - ui_litem_estimate_column(litem); + ui_litem_estimate_column(litem, true); + litem->item.flag &= ~UI_ITEM_MIN; litem->w += 2 * style->boxspace; - litem->h += style->boxspace; + litem->h += 2 * style->boxspace; } static void ui_litem_layout_box(uiLayout *litem) @@ -2418,17 +2470,18 @@ static void ui_litem_layout_box(uiLayout *litem) h = litem->h; litem->x += style->boxspace; + litem->y -= style->boxspace; if (w != 0) litem->w -= 2 * style->boxspace; if (h != 0) litem->h -= 2 * style->boxspace; - ui_litem_layout_column(litem); + ui_litem_layout_column(litem, true); litem->x -= style->boxspace; litem->y -= style->boxspace; if (w != 0) litem->w += 2 * style->boxspace; - if (h != 0) litem->h += style->boxspace; + if (h != 0) litem->h += 2 * style->boxspace; /* roundbox around the sublayout */ but = box->roundbox; @@ -2648,13 +2701,14 @@ static void ui_litem_layout_absolute(uiLayout *litem) static void ui_litem_estimate_split(uiLayout *litem) { ui_litem_estimate_row(litem); + litem->item.flag &= ~UI_ITEM_MIN; } static void ui_litem_layout_split(uiLayout *litem) { uiLayoutItemSplit *split = (uiLayoutItemSplit *)litem; uiItem *item; - float percentage; + float percentage, extra_pixel = 0.0f; const int tot = BLI_listbase_count(&litem->items); int itemh, x, y, w, colw = 0; @@ -2677,7 +2731,9 @@ static void ui_litem_layout_split(uiLayout *litem) x += colw; if (item->next) { - colw = (w - (int)(w * percentage)) / (tot - 1); + const float width = extra_pixel + (w - (int)(w * percentage)) / ((float)tot - 1); + extra_pixel = width - (int)width; + colw = (int)width; colw = MAX2(colw, 0); x += litem->space; @@ -3078,7 +3134,7 @@ static void ui_item_estimate(uiItem *item) switch (litem->item.type) { case ITEM_LAYOUT_COLUMN: - ui_litem_estimate_column(litem); + ui_litem_estimate_column(litem, false); break; case ITEM_LAYOUT_COLUMN_FLOW: ui_litem_estimate_column_flow(litem); @@ -3133,9 +3189,9 @@ static void ui_item_align(uiLayout *litem, short nr) } else if (item->type == ITEM_LAYOUT_BOX) { box = (uiLayoutItemBx *)item; - box->roundbox->alignnr = nr; - BLI_remlink(&litem->root->block->buttons, box->roundbox); - BLI_addhead(&litem->root->block->buttons, box->roundbox); + if (!box->roundbox->alignnr) { + box->roundbox->alignnr = nr; + } } else if (((uiLayout *)item)->align) { ui_item_align((uiLayout *)item, nr); @@ -3177,7 +3233,7 @@ static void ui_item_layout(uiItem *item) switch (litem->item.type) { case ITEM_LAYOUT_COLUMN: - ui_litem_layout_column(litem); + ui_litem_layout_column(litem, false); break; case ITEM_LAYOUT_COLUMN_FLOW: ui_litem_layout_column_flow(litem); @@ -3521,14 +3577,13 @@ void uiLayoutOperatorButs( row = uiLayoutRow(layout, true); uiItemM(row, (bContext *)C, "WM_MT_operator_presets", NULL, ICON_NONE); - WM_operator_properties_create(&op_ptr, "WM_OT_operator_preset_add"); + wmOperatorType *ot = WM_operatortype_find("WM_OT_operator_preset_add", false); + op_ptr = uiItemFullO_ptr(row, ot, "", ICON_ZOOMIN, NULL, WM_OP_INVOKE_DEFAULT, UI_ITEM_O_RETURN_PROPS); RNA_string_set(&op_ptr, "operator", op->type->idname); - uiItemFullO(row, "WM_OT_operator_preset_add", "", ICON_ZOOMIN, op_ptr.data, WM_OP_INVOKE_DEFAULT, 0); - WM_operator_properties_create(&op_ptr, "WM_OT_operator_preset_add"); + op_ptr = uiItemFullO_ptr(row, ot, "", ICON_ZOOMOUT, NULL, WM_OP_INVOKE_DEFAULT, UI_ITEM_O_RETURN_PROPS); RNA_string_set(&op_ptr, "operator", op->type->idname); RNA_boolean_set(&op_ptr, "remove_active", true); - uiItemFullO(row, "WM_OT_operator_preset_add", "", ICON_ZOOMOUT, op_ptr.data, WM_OP_INVOKE_DEFAULT, 0); } if (op->type->ui) { diff --git a/source/blender/editors/interface/interface_ops.c b/source/blender/editors/interface/interface_ops.c index 40ebc946e79..fb95cdf389b 100644 --- a/source/blender/editors/interface/interface_ops.c +++ b/source/blender/editors/interface/interface_ops.c @@ -739,6 +739,7 @@ static int editsource_text_edit( if (text == NULL) { text = BKE_text_load(bmain, filepath, bmain->name); + id_us_ensure_real(&text->id); } if (text == NULL) { diff --git a/source/blender/editors/interface/interface_regions.c b/source/blender/editors/interface/interface_regions.c index 466978272bc..534bd4278ca 100644 --- a/source/blender/editors/interface/interface_regions.c +++ b/source/blender/editors/interface/interface_regions.c @@ -1692,6 +1692,28 @@ static void ui_block_region_draw(const bContext *C, ARegion *ar) UI_block_draw(C, block); } +/** + * Use to refresh centered popups on screen resizing (for splash). + */ +static void ui_block_region_popup_window_listener( + bScreen *UNUSED(sc), ScrArea *UNUSED(sa), ARegion *ar, wmNotifier *wmn) +{ + switch (wmn->category) { + case NC_WINDOW: + { + switch (wmn->action) { + case NA_EDITED: + { + /* window resize */ + ED_region_tag_refresh_ui(ar); + break; + } + } + break; + } + } +} + static void ui_popup_block_clip(wmWindow *window, uiBlock *block) { uiBut *bt; @@ -2003,6 +2025,11 @@ uiPopupBlockHandle *ui_popup_block_create( block = ui_popup_block_refresh(C, handle, butregion, but); handle = block->handle; + /* keep centered on window resizing */ + if ((block->bounds_type == UI_BLOCK_BOUNDS_POPUP_CENTER) && handle->can_refresh) { + type.listener = ui_block_region_popup_window_listener; + } + return handle; } @@ -3285,7 +3312,7 @@ void UI_popup_block_invoke(bContext *C, uiBlockCreateFunc func, void *arg) UI_popup_block_invoke_ex(C, func, arg, NULL, WM_OP_INVOKE_DEFAULT); } -void UI_popup_block_ex(bContext *C, uiBlockCreateFunc func, uiBlockHandleFunc popup_func, uiBlockCancelFunc cancel_func, void *arg) +void UI_popup_block_ex(bContext *C, uiBlockCreateFunc func, uiBlockHandleFunc popup_func, uiBlockCancelFunc cancel_func, void *arg, wmOperator *op) { wmWindow *window = CTX_wm_window(C); uiPopupBlockHandle *handle; @@ -3294,6 +3321,7 @@ void UI_popup_block_ex(bContext *C, uiBlockCreateFunc func, uiBlockHandleFunc po handle->popup = true; handle->retvalue = 1; + handle->popup_op = op; handle->popup_arg = arg; handle->popup_func = popup_func; handle->cancel_func = cancel_func; diff --git a/source/blender/editors/interface/interface_templates.c b/source/blender/editors/interface/interface_templates.c index 4db1c845c23..131584dd405 100644 --- a/source/blender/editors/interface/interface_templates.c +++ b/source/blender/editors/interface/interface_templates.c @@ -1978,6 +1978,7 @@ static void curvemap_tools_dofunc(bContext *C, void *cumap_v, int event) case UICURVE_FUNC_HANDLE_AUTO_ANIM: /* set auto-clamped */ curvemap_handle_set(cuma, HD_AUTO_ANIM); curvemapping_changed(cumap, false); + break; case UICURVE_FUNC_EXTEND_HOZ: /* extend horiz */ cuma->flag &= ~CUMA_EXTEND_EXTRAPOLATE; curvemapping_changed(cumap, false); diff --git a/source/blender/editors/interface/interface_utils.c b/source/blender/editors/interface/interface_utils.c index df6f098ee81..636b7e4e9ce 100644 --- a/source/blender/editors/interface/interface_utils.c +++ b/source/blender/editors/interface/interface_utils.c @@ -127,12 +127,10 @@ uiBut *uiDefAutoButR(uiBlock *block, PointerRNA *ptr, PropertyRNA *prop, int ind break; case PROP_POINTER: { - PointerRNA pptr; - - pptr = RNA_property_pointer_get(ptr, prop); - if (!pptr.type) - pptr.type = RNA_property_pointer_type(ptr, prop); - icon = RNA_struct_ui_icon(pptr.type); + if (icon == 0) { + PointerRNA pptr = RNA_property_pointer_get(ptr, prop); + icon = RNA_struct_ui_icon(pptr.type ? pptr.type : RNA_property_pointer_type(ptr, prop)); + } if (icon == ICON_DOT) icon = 0; diff --git a/source/blender/editors/interface/interface_widgets.c b/source/blender/editors/interface/interface_widgets.c index d43a94c5514..f7f2b422724 100644 --- a/source/blender/editors/interface/interface_widgets.c +++ b/source/blender/editors/interface/interface_widgets.c @@ -874,21 +874,18 @@ static void widget_draw_icon( float ofs = 1.0f / aspect; if (but->drawflag & UI_BUT_ICON_LEFT) { - if (but->block->flag & UI_BLOCK_LOOP) { - if (but->type == UI_BTYPE_SEARCH_MENU) - xs = rect->xmin + 4.0f * ofs; - else - xs = rect->xmin + ofs; - } - else { + /* special case - icon_only pie buttons */ + if (ui_block_is_pie_menu(but->block) && but->type != UI_BTYPE_MENU && but->str && but->str[0] == '\0') + xs = rect->xmin + 2.0f * ofs; + else if (but->dt == UI_EMBOSS_NONE || but->type == UI_BTYPE_LABEL) + xs = rect->xmin + 2.0f * ofs; + else xs = rect->xmin + 4.0f * ofs; - } - ys = (rect->ymin + rect->ymax - height) / 2.0f; } else { xs = (rect->xmin + rect->xmax - height) / 2.0f; - ys = (rect->ymin + rect->ymax - height) / 2.0f; } + ys = (rect->ymin + rect->ymax - height) / 2.0f; /* force positions to integers, for zoom levels near 1. draws icons crisp. */ if (aspect > 0.95f && aspect < 1.05f) { @@ -1554,11 +1551,15 @@ static void widget_draw_text_icon(uiFontStyle *fstyle, uiWidgetColors *wcol, uiB /* Icons on the left with optional text label on the right */ else if (but->flag & UI_HAS_ICON || show_menu_icon) { const BIFIconID icon = (but->flag & UI_HAS_ICON) ? but->icon + but->iconadd : ICON_NONE; - const float icon_size = ICON_SIZE_FROM_BUTRECT(rect); + const float icon_size = ICON_DEFAULT_WIDTH_SCALE; /* menu item - add some more padding so menus don't feel cramped. it must * be part of the button so that this area is still clickable */ - if (ui_block_is_menu(but->block)) + if (ui_block_is_pie_menu(but->block)) { + if (but->dt == UI_EMBOSS_RADIAL) + rect->xmin += 0.3f * U.widget_unit; + } + else if (ui_block_is_menu(but->block)) rect->xmin += 0.3f * U.widget_unit; widget_draw_icon(but, icon, alpha, rect, show_menu_icon); @@ -1584,7 +1585,7 @@ static void widget_draw_text_icon(uiFontStyle *fstyle, uiWidgetColors *wcol, uiB temp.xmin = temp.xmax - (BLI_rcti_size_y(rect) * 1.08f); if (extra_icon_type == UI_BUT_ICONEXTRA_CLEAR) { - widget_draw_icon(but, ICON_X, alpha, &temp, false); + widget_draw_icon(but, ICON_PANEL_CLOSE, alpha, &temp, false); } else if (extra_icon_type == UI_BUT_ICONEXTRA_EYEDROPPER) { widget_draw_icon(but, ICON_EYEDROPPER, alpha, &temp, false); diff --git a/source/blender/editors/io/io_alembic.c b/source/blender/editors/io/io_alembic.c index a991f59e8e2..3f9eb33e239 100644 --- a/source/blender/editors/io/io_alembic.c +++ b/source/blender/editors/io/io_alembic.c @@ -124,9 +124,9 @@ static int wm_alembic_export_exec(bContext *C, wmOperator *op) .use_subdiv_schema = RNA_boolean_get(op->ptr, "subdiv_schema"), .compression_type = RNA_enum_get(op->ptr, "compression_type"), .packuv = RNA_boolean_get(op->ptr, "packuv"), - .triangulate = RNA_boolean_get(op->ptr, "triangulate"), - .quad_method = RNA_enum_get(op->ptr, "quad_method"), - .ngon_method = RNA_enum_get(op->ptr, "ngon_method"), + .triangulate = RNA_boolean_get(op->ptr, "triangulate"), + .quad_method = RNA_enum_get(op->ptr, "quad_method"), + .ngon_method = RNA_enum_get(op->ptr, "ngon_method"), .global_scale = RNA_float_get(op->ptr, "global_scale"), }; diff --git a/source/blender/editors/io/io_collada.c b/source/blender/editors/io/io_collada.c index baae92f962e..b8794afbf85 100644 --- a/source/blender/editors/io/io_collada.c +++ b/source/blender/editors/io/io_collada.c @@ -97,7 +97,9 @@ static int wm_collada_export_exec(bContext *C, wmOperator *op) int use_blender_profile; int sort_by_name; int export_transformation_type; + int open_sim; + int keep_bind_info; int export_count; @@ -148,6 +150,8 @@ static int wm_collada_export_exec(bContext *C, wmOperator *op) export_transformation_type = RNA_enum_get(op->ptr, "export_transformation_type_selection"); open_sim = RNA_boolean_get(op->ptr, "open_sim"); + keep_bind_info = RNA_boolean_get(op->ptr, "keep_bind_info"); + /* get editmode results */ ED_object_editmode_load(CTX_data_edit_object(C)); @@ -172,7 +176,10 @@ static int wm_collada_export_exec(bContext *C, wmOperator *op) use_blender_profile, sort_by_name, export_transformation_type, - open_sim); + + open_sim, + keep_bind_info + ); if (export_count == 0) { BKE_report(op->reports, RPT_WARNING, "No objects selected -- Created empty export file"); @@ -256,11 +263,11 @@ static void uiCollada_exportSettings(uiLayout *layout, PointerRNA *imfptr) uiItemL(row, IFACE_("Collada Options:"), ICON_MODIFIER); row = uiLayoutRow(box, false); - uiItemR(row, imfptr, "triangulate", 0, NULL, ICON_NONE); + uiItemR(row, imfptr, "triangulate", 1, NULL, ICON_NONE); row = uiLayoutRow(box, false); - uiItemR(row, imfptr, "use_object_instantiation", 0, NULL, ICON_NONE); + uiItemR(row, imfptr, "use_object_instantiation", 1, NULL, ICON_NONE); row = uiLayoutRow(box, false); - uiItemR(row, imfptr, "use_blender_profile", 0, NULL, ICON_NONE); + uiItemR(row, imfptr, "use_blender_profile", 1, NULL, ICON_NONE); row = uiLayoutRow(box, false); split = uiLayoutSplit(row, 0.6f, UI_LAYOUT_ALIGN_RIGHT); @@ -269,6 +276,8 @@ static void uiCollada_exportSettings(uiLayout *layout, PointerRNA *imfptr) row = uiLayoutRow(box, false); uiItemR(row, imfptr, "sort_by_name", 0, NULL, ICON_NONE); + row = uiLayoutRow(box, false); + uiItemR(row, imfptr, "keep_bind_info", 0, NULL, ICON_NONE); } @@ -296,6 +305,8 @@ static bool wm_collada_export_check(bContext *UNUSED(C), wmOperator *op) void WM_OT_collada_export(wmOperatorType *ot) { + struct StructRNA *func = ot->srna; + static EnumPropertyItem prop_bc_export_mesh_type[] = { {BC_MESH_TYPE_VIEW, "view", 0, "View", "Apply modifier's view settings"}, {BC_MESH_TYPE_RENDER, "render", 0, "Render", "Apply modifier's render settings"}, @@ -325,65 +336,69 @@ void WM_OT_collada_export(wmOperatorType *ot) ot, FILE_TYPE_FOLDER | FILE_TYPE_COLLADA, FILE_BLENDER, FILE_SAVE, WM_FILESEL_FILEPATH, FILE_DEFAULTDISPLAY, FILE_SORT_ALPHA); - RNA_def_boolean(ot->srna, + RNA_def_boolean(func, "apply_modifiers", 0, "Apply Modifiers", "Apply modifiers to exported mesh (non destructive))"); - RNA_def_int(ot->srna, "export_mesh_type", 0, INT_MIN, INT_MAX, + RNA_def_int(func, "export_mesh_type", 0, INT_MIN, INT_MAX, "Resolution", "Modifier resolution for export", INT_MIN, INT_MAX); - RNA_def_enum(ot->srna, "export_mesh_type_selection", prop_bc_export_mesh_type, 0, + RNA_def_enum(func, "export_mesh_type_selection", prop_bc_export_mesh_type, 0, "Resolution", "Modifier resolution for export"); - RNA_def_boolean(ot->srna, "selected", 0, "Selection Only", + RNA_def_boolean(func, "selected", 0, "Selection Only", "Export only selected elements"); - RNA_def_boolean(ot->srna, "include_children", 0, "Include Children", + RNA_def_boolean(func, "include_children", 0, "Include Children", "Export all children of selected objects (even if not selected)"); - RNA_def_boolean(ot->srna, "include_armatures", 0, "Include Armatures", + RNA_def_boolean(func, "include_armatures", 0, "Include Armatures", "Export related armatures (even if not selected)"); - RNA_def_boolean(ot->srna, "include_shapekeys", 1, "Include Shape Keys", + RNA_def_boolean(func, "include_shapekeys", 1, "Include Shape Keys", "Export all Shape Keys from Mesh Objects"); - RNA_def_boolean(ot->srna, "deform_bones_only", 0, "Deform Bones only", + RNA_def_boolean(func, "deform_bones_only", 0, "Deform Bones only", "Only export deforming bones with armatures"); - RNA_def_boolean(ot->srna, "active_uv_only", 0, "Only Selected UV Map", + RNA_def_boolean(func, "active_uv_only", 0, "Only Selected UV Map", "Export only the selected UV Map"); - RNA_def_boolean(ot->srna, "include_uv_textures", 0, "Include UV Textures", + RNA_def_boolean(func, "include_uv_textures", 0, "Include UV Textures", "Export textures assigned to the object UV Maps"); - RNA_def_boolean(ot->srna, "include_material_textures", 0, "Include Material Textures", + RNA_def_boolean(func, "include_material_textures", 0, "Include Material Textures", "Export textures assigned to the object Materials"); - RNA_def_boolean(ot->srna, "use_texture_copies", 1, "Copy", + RNA_def_boolean(func, "use_texture_copies", 1, "Copy", "Copy textures to same folder where the .dae file is exported"); - RNA_def_boolean(ot->srna, "triangulate", 1, "Triangulate", + RNA_def_boolean(func, "triangulate", 1, "Triangulate", "Export Polygons (Quads & NGons) as Triangles"); - RNA_def_boolean(ot->srna, "use_object_instantiation", 1, "Use Object Instances", + RNA_def_boolean(func, "use_object_instantiation", 1, "Use Object Instances", "Instantiate multiple Objects from same Data"); - RNA_def_boolean(ot->srna, "use_blender_profile", 1, "Use Blender Profile", + RNA_def_boolean(func, "use_blender_profile", 1, "Use Blender Profile", "Export additional Blender specific information (for material, shaders, bones, etc.)"); - RNA_def_boolean(ot->srna, "sort_by_name", 0, "Sort by Object name", + RNA_def_boolean(func, "sort_by_name", 0, "Sort by Object name", "Sort exported data by Object name"); - RNA_def_int(ot->srna, "export_transformation_type", 0, INT_MIN, INT_MAX, + RNA_def_int(func, "export_transformation_type", 0, INT_MIN, INT_MAX, "Transform", "Transformation type for translation, scale and rotation", INT_MIN, INT_MAX); - RNA_def_enum(ot->srna, "export_transformation_type_selection", prop_bc_export_transformation_type, 0, + RNA_def_enum(func, "export_transformation_type_selection", prop_bc_export_transformation_type, 0, "Transform", "Transformation type for translation, scale and rotation"); - RNA_def_boolean(ot->srna, "open_sim", 0, "Export to SL/OpenSim", + RNA_def_boolean(func, "open_sim", 0, "Export to SL/OpenSim", "Compatibility mode for SL, OpenSim and other compatible online worlds"); + + RNA_def_boolean(func, "keep_bind_info", 0, + "Keep Bind Info", "Store Bindpose information in custom bone properties for later use during Collada export"); + } @@ -395,7 +410,9 @@ static int wm_collada_import_exec(bContext *C, wmOperator *op) int find_chains; int auto_connect; int fix_orientation; - int min_chain_length; + int min_chain_length; + + int keep_bind_info; if (!RNA_struct_property_is_set(op->ptr, "filepath")) { BKE_report(op->reports, RPT_ERROR, "No filename given"); @@ -407,6 +424,9 @@ static int wm_collada_import_exec(bContext *C, wmOperator *op) find_chains = RNA_boolean_get(op->ptr, "find_chains"); auto_connect = RNA_boolean_get(op->ptr, "auto_connect"); fix_orientation = RNA_boolean_get(op->ptr, "fix_orientation"); + + keep_bind_info = RNA_boolean_get(op->ptr, "keep_bind_info"); + min_chain_length = RNA_int_get(op->ptr, "min_chain_length"); RNA_string_get(op->ptr, "filepath", filename); @@ -416,7 +436,8 @@ static int wm_collada_import_exec(bContext *C, wmOperator *op) find_chains, auto_connect, fix_orientation, - min_chain_length)) + min_chain_length, + keep_bind_info) ) { return OPERATOR_FINISHED; } @@ -453,6 +474,13 @@ static void uiCollada_importSettings(uiLayout *layout, PointerRNA *imfptr) row = uiLayoutRow(box, false); uiItemR(row, imfptr, "min_chain_length", 0, NULL, ICON_NONE); + + box = uiLayoutBox(layout); + row = uiLayoutRow(box, false); + + row = uiLayoutRow(box, false); + uiItemR(row, imfptr, "keep_bind_info", 0, NULL, ICON_NONE); + } static void wm_collada_import_draw(bContext *UNUSED(C), wmOperator *op) @@ -508,5 +536,9 @@ void WM_OT_collada_import(wmOperatorType *ot) 0, INT_MAX); + RNA_def_boolean(ot->srna, + "keep_bind_info", 0, "Keep Bind Info", + "Store Bindpose information in custom bone properties for later use during Collada export"); + } #endif diff --git a/source/blender/editors/mesh/editmesh_intersect.c b/source/blender/editors/mesh/editmesh_intersect.c index de93211bec4..49bfde77032 100644 --- a/source/blender/editors/mesh/editmesh_intersect.c +++ b/source/blender/editors/mesh/editmesh_intersect.c @@ -51,6 +51,7 @@ #include "mesh_intern.h" /* own include */ #include "tools/bmesh_intersect.h" +#include "tools/bmesh_separate.h" /* detect isolated holes and fill them */ @@ -137,6 +138,12 @@ enum { ISECT_SEL_UNSEL = 1, }; +enum { + ISECT_SEPARATE_ALL = 0, + ISECT_SEPARATE_CUT = 1, + ISECT_SEPARATE_NONE = 2, +}; + static int edbm_intersect_exec(bContext *C, wmOperator *op) { Object *obedit = CTX_data_edit_object(C); @@ -144,7 +151,9 @@ static int edbm_intersect_exec(bContext *C, wmOperator *op) BMesh *bm = em->bm; const int mode = RNA_enum_get(op->ptr, "mode"); int (*test_fn)(BMFace *, void *); - bool use_separate = RNA_boolean_get(op->ptr, "use_separate"); + bool use_separate_all = false; + bool use_separate_cut = false; + const int separate_mode = RNA_enum_get(op->ptr, "separate_mode"); const float eps = RNA_float_get(op->ptr, "threshold"); bool use_self; bool has_isect; @@ -160,15 +169,38 @@ static int edbm_intersect_exec(bContext *C, wmOperator *op) break; } + switch (separate_mode) { + case ISECT_SEPARATE_ALL: + use_separate_all = true; + break; + case ISECT_SEPARATE_CUT: + if (use_self == false) { + use_separate_cut = true; + } + else { + /* we could support this but would require more advanced logic inside 'BM_mesh_intersect' + * for now just separate all */ + use_separate_all = true; + } + break; + default: /* ISECT_SEPARATE_NONE */ + break; + } has_isect = BM_mesh_intersect( bm, em->looptris, em->tottri, test_fn, NULL, - use_self, use_separate, true, true, + use_self, use_separate_all, true, true, true, -1, eps); + if (use_separate_cut) { + /* detach selected/un-selected faces */ + BM_mesh_separate_faces( + bm, + BM_elem_cb_check_hflag_enabled_simple(const BMFace *, BM_ELEM_SELECT)); + } if (has_isect) { edbm_intersect_select(em); @@ -190,6 +222,16 @@ void MESH_OT_intersect(struct wmOperatorType *ot) {0, NULL, 0, NULL, NULL} }; + static EnumPropertyItem isect_separate_items[] = { + {ISECT_SEPARATE_ALL, "ALL", 0, "All", + "Separate all geometry from intersections"}, + {ISECT_SEPARATE_CUT, "CUT", 0, "Cut", + "Cut into geometry keeping each side separate (Selected/Unselected only)"}, + {ISECT_SEPARATE_NONE, "NONE", 0, "Merge", + "Merge all geometry from the intersection"}, + {0, NULL, 0, NULL, NULL} + }; + /* identifiers */ ot->name = "Intersect (Knife)"; ot->description = "Cut an intersection into faces"; @@ -201,7 +243,7 @@ void MESH_OT_intersect(struct wmOperatorType *ot) /* props */ RNA_def_enum(ot->srna, "mode", isect_mode_items, ISECT_SEL_UNSEL, "Source", ""); - RNA_def_boolean(ot->srna, "use_separate", true, "Separate", ""); + RNA_def_enum(ot->srna, "separate_mode", isect_separate_items, ISECT_SEPARATE_CUT, "Separate Mode", ""); RNA_def_float_distance(ot->srna, "threshold", 0.000001f, 0.0, 0.01, "Merge threshold", "", 0.0, 0.001); /* flags */ @@ -239,7 +281,7 @@ static int edbm_intersect_boolean_exec(bContext *C, wmOperator *op) bm, em->looptris, em->tottri, test_fn, NULL, - false, false, true, true, + false, false, true, true, true, boolean_operation, eps); diff --git a/source/blender/editors/mesh/editmesh_rip.c b/source/blender/editors/mesh/editmesh_rip.c index e05ce727e22..1a2f9fdb62b 100644 --- a/source/blender/editors/mesh/editmesh_rip.c +++ b/source/blender/editors/mesh/editmesh_rip.c @@ -596,7 +596,7 @@ static int edbm_rip_invoke__vert(bContext *C, wmOperator *op, const wmEvent *eve BMVert *v_new; BLI_assert(l_sep->v == v); - v_new = bmesh_urmv_loop_region(bm, l_sep); + v_new = BM_face_loop_separate_multi_isolated(bm, l_sep); BLI_assert(BM_vert_find_first_loop(v)); BM_vert_select_set(bm, v, false); @@ -665,7 +665,7 @@ static int edbm_rip_invoke__vert(bContext *C, wmOperator *op, const wmEvent *eve BM_vert_select_set(bm, v, false); - bmesh_vert_separate(bm, v, &vout, &vout_len, true); + bmesh_kernel_vert_separate(bm, v, &vout, &vout_len, true); if (vout_len < 2) { MEM_freeN(vout); diff --git a/source/blender/editors/metaball/mball_edit.c b/source/blender/editors/metaball/mball_edit.c index ed5bf4a92b4..bc42717b69f 100644 --- a/source/blender/editors/metaball/mball_edit.c +++ b/source/blender/editors/metaball/mball_edit.c @@ -592,12 +592,9 @@ bool ED_mball_select_pick(bContext *C, const int mval[2], bool extend, bool dese view3d_set_viewcontext(C, &vc); - rect.xmin = mval[0] - 12; - rect.xmax = mval[0] + 12; - rect.ymin = mval[1] - 12; - rect.ymax = mval[1] + 12; + BLI_rcti_init_pt_radius(&rect, mval, 12); - hits = view3d_opengl_select(&vc, buffer, MAXPICKBUF, &rect, true); + hits = view3d_opengl_select(&vc, buffer, MAXPICKBUF, &rect, VIEW3D_SELECT_PICK_NEAREST); /* does startelem exist? */ ml = mb->editelems->first; diff --git a/source/blender/editors/object/object_add.c b/source/blender/editors/object/object_add.c index 02b2d8492b4..20523264cd9 100644 --- a/source/blender/editors/object/object_add.c +++ b/source/blender/editors/object/object_add.c @@ -64,6 +64,7 @@ #include "BKE_armature.h" #include "BKE_camera.h" #include "BKE_context.h" +#include "BKE_constraint.h" #include "BKE_curve.h" #include "BKE_depsgraph.h" #include "BKE_DerivedMesh.h" @@ -1377,7 +1378,7 @@ static void make_object_duplilist_real(bContext *C, Scene *scene, Base *base, ob->proxy = NULL; ob->parent = NULL; - BLI_listbase_clear(&ob->constraints); + BKE_constraints_free(&ob->constraints); ob->curve_cache = NULL; ob->transflag &= ~OB_DUPLI; ob->lay = base->lay; @@ -1655,8 +1656,25 @@ static int convert_exec(bContext *C, wmOperator *op) } } - CTX_DATA_BEGIN (C, Base *, base, selected_editable_bases) + ListBase selected_editable_bases = CTX_data_collection_get(C, "selected_editable_bases"); + + /* Ensure we get all meshes calculated with a sufficient data-mask, + * needed since re-evaluating single modifiers causes bugs if they depend + * on other objects data masks too, see: T50950. */ { + for (CollectionPointerLink *link = selected_editable_bases.first; link; link = link->next) { + Base *base = link->ptr.data; + DAG_id_tag_update(&base->object->id, OB_RECALC_DATA); + } + + uint64_t customdata_mask_prev = scene->customdata_mask; + scene->customdata_mask |= CD_MASK_MESH; + BKE_scene_update_tagged(bmain->eval_ctx, bmain, scene); + scene->customdata_mask = customdata_mask_prev; + } + + for (CollectionPointerLink *link = selected_editable_bases.first; link; link = link->next) { + Base *base = link->ptr.data; ob = base->object; if (ob->flag & OB_DONE || !IS_TAGGED(ob->data)) { @@ -1699,7 +1717,7 @@ static int convert_exec(bContext *C, wmOperator *op) ED_rigidbody_object_remove(bmain, scene, newob); } } - else if (ob->type == OB_MESH && ob->modifiers.first) { /* converting a mesh with no modifiers causes a segfault */ + else if (ob->type == OB_MESH) { ob->flag |= OB_DONE; if (keep_original) { @@ -1723,7 +1741,6 @@ static int convert_exec(bContext *C, wmOperator *op) * cases this doesnt give correct results (when MDEF is used for eg) */ dm = mesh_get_derived_final(scene, newob, CD_MASK_MESH); - // dm = mesh_create_derived_no_deform(ob1, NULL); /* this was called original (instead of get_derived). man o man why! (ton) */ DM_to_mesh(dm, newob->data, newob, CD_MASK_MESH, true); @@ -1888,7 +1905,7 @@ static int convert_exec(bContext *C, wmOperator *op) ((ID *)ob->data)->tag &= ~LIB_TAG_DOIT; /* flag not to convert this datablock again */ } } - CTX_DATA_END; + BLI_freelistN(&selected_editable_bases); if (!keep_original) { if (mballConverted) { diff --git a/source/blender/editors/object/object_bake_api.c b/source/blender/editors/object/object_bake_api.c index fd95d6129ad..968081818a2 100644 --- a/source/blender/editors/object/object_bake_api.c +++ b/source/blender/editors/object/object_bake_api.c @@ -352,12 +352,17 @@ static bool is_noncolor_pass(ScenePassType pass_type) } /* if all is good tag image and return true */ -static bool bake_object_check(Object *ob, ReportList *reports) +static bool bake_object_check(Scene *scene, Object *ob, ReportList *reports) { Image *image; void *lock; int i; + if ((ob->lay & scene->lay) == 0) { + BKE_reportf(reports, RPT_ERROR, "Object \"%s\" is not on a scene layer", ob->id.name + 2); + return false; + } + if (ob->type != OB_MESH) { BKE_reportf(reports, RPT_ERROR, "Object \"%s\" is not a mesh", ob->id.name + 2); return false; @@ -491,7 +496,7 @@ static bool bake_pass_filter_check(ScenePassType pass_type, const int pass_filte } /* before even getting in the bake function we check for some basic errors */ -static bool bake_objects_check(Main *bmain, Object *ob, ListBase *selected_objects, +static bool bake_objects_check(Main *bmain, Scene *scene, Object *ob, ListBase *selected_objects, ReportList *reports, const bool is_selected_to_active) { CollectionPointerLink *link; @@ -502,7 +507,7 @@ static bool bake_objects_check(Main *bmain, Object *ob, ListBase *selected_objec if (is_selected_to_active) { int tot_objects = 0; - if (!bake_object_check(ob, reports)) + if (!bake_object_check(scene, ob, reports)) return false; for (link = selected_objects->first; link; link = link->next) { @@ -530,7 +535,7 @@ static bool bake_objects_check(Main *bmain, Object *ob, ListBase *selected_objec } for (link = selected_objects->first; link; link = link->next) { - if (!bake_object_check(link->ptr.data, reports)) + if (!bake_object_check(scene, link->ptr.data, reports)) return false; } } @@ -619,7 +624,7 @@ static Mesh *bake_mesh_new_from_object(Main *bmain, Scene *scene, Object *ob) ED_object_editmode_load(ob); Mesh *me = BKE_mesh_new_from_object(bmain, scene, ob, 1, 2, 0, 0); - BKE_mesh_split_faces(me); + BKE_mesh_split_faces(me, true); return me; } @@ -1179,7 +1184,7 @@ static int bake_exec(bContext *C, wmOperator *op) goto finally; } - if (!bake_objects_check(bkr.main, bkr.ob, &bkr.selected_objects, bkr.reports, bkr.is_selected_to_active)) { + if (!bake_objects_check(bkr.main, bkr.scene, bkr.ob, &bkr.selected_objects, bkr.reports, bkr.is_selected_to_active)) { goto finally; } @@ -1237,7 +1242,7 @@ static void bake_startjob(void *bkv, short *UNUSED(stop), short *do_update, floa return; } - if (!bake_objects_check(bkr->main, bkr->ob, &bkr->selected_objects, bkr->reports, bkr->is_selected_to_active)) { + if (!bake_objects_check(bkr->main, bkr->scene, bkr->ob, &bkr->selected_objects, bkr->reports, bkr->is_selected_to_active)) { bkr->result = OPERATOR_CANCELLED; return; } diff --git a/source/blender/editors/object/object_intern.h b/source/blender/editors/object/object_intern.h index 9710e4f843d..b8957bdedf9 100644 --- a/source/blender/editors/object/object_intern.h +++ b/source/blender/editors/object/object_intern.h @@ -186,6 +186,7 @@ void OBJECT_OT_skin_loose_mark_clear(struct wmOperatorType *ot); void OBJECT_OT_skin_radii_equalize(struct wmOperatorType *ot); void OBJECT_OT_skin_armature_create(struct wmOperatorType *ot); void OBJECT_OT_laplaciandeform_bind(struct wmOperatorType *ot); +void OBJECT_OT_surfacedeform_bind(struct wmOperatorType *ot); /* object_constraint.c */ void OBJECT_OT_constraint_add(struct wmOperatorType *ot); diff --git a/source/blender/editors/object/object_modifier.c b/source/blender/editors/object/object_modifier.c index 06f495fb9f1..06d306ded42 100644 --- a/source/blender/editors/object/object_modifier.c +++ b/source/blender/editors/object/object_modifier.c @@ -1484,7 +1484,6 @@ static int skin_root_mark_exec(bContext *C, wmOperator *UNUSED(op)) Object *ob = CTX_data_edit_object(C); BMEditMesh *em = BKE_editmesh_from_object(ob); BMesh *bm = em->bm; - const int cd_vert_skin_offset = CustomData_get_offset(&bm->vdata, CD_MVERT_SKIN); BMVert *bm_vert; BMIter bm_iter; GSet *visited; @@ -1493,6 +1492,8 @@ static int skin_root_mark_exec(bContext *C, wmOperator *UNUSED(op)) BKE_mesh_ensure_skin_customdata(ob->data); + const int cd_vert_skin_offset = CustomData_get_offset(&bm->vdata, CD_MVERT_SKIN); + BM_ITER_MESH (bm_vert, &bm_iter, bm, BM_VERTS_OF_MESH) { if (BM_elem_flag_test(bm_vert, BM_ELEM_SELECT) && BLI_gset_add(visited, bm_vert)) @@ -2294,3 +2295,56 @@ void OBJECT_OT_laplaciandeform_bind(wmOperatorType *ot) ot->flag = OPTYPE_REGISTER | OPTYPE_UNDO | OPTYPE_INTERNAL; edit_modifier_properties(ot); } + +/************************ sdef bind operator *********************/ + +static int surfacedeform_bind_poll(bContext *C) +{ + return edit_modifier_poll_generic(C, &RNA_SurfaceDeformModifier, 0); +} + +static int surfacedeform_bind_exec(bContext *C, wmOperator *op) +{ + Object *ob = ED_object_active_context(C); + SurfaceDeformModifierData *smd = (SurfaceDeformModifierData *)edit_modifier_property_get(op, ob, eModifierType_SurfaceDeform); + + if (!smd) + return OPERATOR_CANCELLED; + + if (smd->flags & MOD_SDEF_BIND) { + smd->flags &= ~MOD_SDEF_BIND; + } + else if (smd->target) { + smd->flags |= MOD_SDEF_BIND; + } + + DAG_id_tag_update(&ob->id, OB_RECALC_DATA); + WM_event_add_notifier(C, NC_OBJECT | ND_MODIFIER, ob); + + return OPERATOR_FINISHED; +} + +static int surfacedeform_bind_invoke(bContext *C, wmOperator *op, const wmEvent *UNUSED(event)) +{ + if (edit_modifier_invoke_properties(C, op)) + return surfacedeform_bind_exec(C, op); + else + return OPERATOR_CANCELLED; +} + +void OBJECT_OT_surfacedeform_bind(wmOperatorType *ot) +{ + /* identifiers */ + ot->name = "Surface Deform Bind"; + ot->description = "Bind mesh to target in surface deform modifier"; + ot->idname = "OBJECT_OT_surfacedeform_bind"; + + /* api callbacks */ + ot->poll = surfacedeform_bind_poll; + ot->invoke = surfacedeform_bind_invoke; + ot->exec = surfacedeform_bind_exec; + + /* flags */ + ot->flag = OPTYPE_REGISTER | OPTYPE_UNDO | OPTYPE_INTERNAL; + edit_modifier_properties(ot); +} diff --git a/source/blender/editors/object/object_ops.c b/source/blender/editors/object/object_ops.c index 7e7e1ef182c..5fe5a884354 100644 --- a/source/blender/editors/object/object_ops.c +++ b/source/blender/editors/object/object_ops.c @@ -255,6 +255,7 @@ void ED_operatortypes_object(void) WM_operatortype_append(OBJECT_OT_data_transfer); WM_operatortype_append(OBJECT_OT_datalayout_transfer); + WM_operatortype_append(OBJECT_OT_surfacedeform_bind); } void ED_operatormacros_object(void) diff --git a/source/blender/editors/physics/physics_ops.c b/source/blender/editors/physics/physics_ops.c index 0c907f19753..b1d708ebc07 100644 --- a/source/blender/editors/physics/physics_ops.c +++ b/source/blender/editors/physics/physics_ops.c @@ -138,13 +138,21 @@ static void keymap_particle(wmKeyConfig *keyconf) RNA_boolean_set(kmi->ptr, "unselected", true); /* Shift+LMB behavior first, so it has priority over KM_ANY item below. */ - kmi = WM_keymap_verify_item(keymap, "VIEW3D_OT_manipulator", LEFTMOUSE, KM_PRESS, KM_SHIFT, 0); + kmi = WM_keymap_add_item(keymap, "VIEW3D_OT_manipulator", LEFTMOUSE, KM_PRESS, KM_SHIFT, 0); RNA_boolean_set(kmi->ptr, "release_confirm", true); RNA_boolean_set(kmi->ptr, "use_planar_constraint", true); + RNA_boolean_set(kmi->ptr, "use_accurate", false); + + kmi = WM_keymap_add_item(keymap, "VIEW3D_OT_manipulator", LEFTMOUSE, KM_PRESS, KM_SHIFT, 0); + RNA_boolean_set(kmi->ptr, "release_confirm", true); + RNA_boolean_set(kmi->ptr, "use_planar_constraint", false); + RNA_boolean_set(kmi->ptr, "use_accurate", true); + /* Using KM_ANY here to allow holding modifiers before starting to transform. */ kmi = WM_keymap_add_item(keymap, "VIEW3D_OT_manipulator", LEFTMOUSE, KM_PRESS, KM_ANY, 0); RNA_boolean_set(kmi->ptr, "release_confirm", true); RNA_boolean_set(kmi->ptr, "use_planar_constraint", false); + RNA_boolean_set(kmi->ptr, "use_accurate", false); WM_keymap_add_item(keymap, "PARTICLE_OT_brush_edit", LEFTMOUSE, KM_PRESS, 0, 0); WM_keymap_add_item(keymap, "PARTICLE_OT_brush_edit", LEFTMOUSE, KM_PRESS, KM_SHIFT, 0); diff --git a/source/blender/editors/render/render_opengl.c b/source/blender/editors/render/render_opengl.c index 9097432a251..1d0f433ba38 100644 --- a/source/blender/editors/render/render_opengl.c +++ b/source/blender/editors/render/render_opengl.c @@ -315,7 +315,7 @@ static void screen_opengl_render_doit(OGLRender *oglrender, RenderResult *rr) RE_render_result_rect_from_ibuf(rr, &scene->r, out, oglrender->view_id); IMB_freeImBuf(out); } - else if (gpd){ + else if (gpd) { /* If there are no strips, Grease Pencil still needs a buffer to draw on */ ImBuf *out = IMB_allocImBuf(oglrender->sizex, oglrender->sizey, 32, IB_rect); RE_render_result_rect_from_ibuf(rr, &scene->r, out, oglrender->view_id); @@ -715,7 +715,6 @@ static bool screen_opengl_render_init(bContext *C, wmOperator *op) oglrender->task_scheduler = task_scheduler; oglrender->task_pool = BLI_task_pool_create_background(task_scheduler, oglrender); - BLI_pool_set_num_threads(oglrender->task_pool, 1); } else { oglrender->task_scheduler = NULL; @@ -747,6 +746,23 @@ static void screen_opengl_render_end(bContext *C, OGLRender *oglrender) int i; if (oglrender->is_animation) { + /* Trickery part for movie output: + * + * We MUST write frames in an exact order, so we only let background + * thread to work on that, and main thread is simply waits for that + * thread to do all the dirty work. + * + * After this loop is done work_and_wait() will have nothing to do, + * so we don't run into wrong order of frames written to the stream. + */ + if (BKE_imtype_is_movie(scene->r.im_format.imtype)) { + BLI_mutex_lock(&oglrender->task_mutex); + while (oglrender->num_scheduled_frames > 0) { + BLI_condition_wait(&oglrender->task_condition, + &oglrender->task_mutex); + } + BLI_mutex_unlock(&oglrender->task_mutex); + } BLI_task_pool_work_and_wait(oglrender->task_pool); BLI_task_pool_free(oglrender->task_pool); /* Depending on various things we might or might not use global scheduler. */ @@ -886,14 +902,15 @@ static void write_result_func(TaskPool * __restrict pool, */ ReportList reports; BKE_reports_init(&reports, oglrender->reports->flag & ~RPT_PRINT); - /* Do actual save logic here, depending on the file format. */ + /* Do actual save logic here, depending on the file format. + * + * NOTE: We have to construct temporary scene with proper scene->r.cfra. + * This is because underlying calls do not use r.cfra but use scene + * for that. + */ + Scene tmp_scene = *scene; + tmp_scene.r.cfra = cfra; if (is_movie) { - /* We have to construct temporary scene with proper scene->r.cfra. - * This is because underlying calls do not use r.cfra but use scene - * for that. - */ - Scene tmp_scene = *scene; - tmp_scene.r.cfra = cfra; ok = RE_WriteRenderViewsMovie(&reports, rr, &tmp_scene, @@ -917,8 +934,8 @@ static void write_result_func(TaskPool * __restrict pool, true, NULL); - BKE_render_result_stamp_info(scene, scene->camera, rr, false); - ok = RE_WriteRenderViewsImage(NULL, rr, scene, true, name); + BKE_render_result_stamp_info(&tmp_scene, tmp_scene.camera, rr, false); + ok = RE_WriteRenderViewsImage(NULL, rr, &tmp_scene, true, name); if (!ok) { BKE_reportf(&reports, RPT_ERROR, diff --git a/source/blender/editors/render/render_preview.c b/source/blender/editors/render/render_preview.c index 87c08dc6583..fa18b82507e 100644 --- a/source/blender/editors/render/render_preview.c +++ b/source/blender/editors/render/render_preview.c @@ -196,7 +196,7 @@ static Main *load_main_from_memory(const void *blend, int blend_size) BlendFileData *bfd; G.fileflags |= G_FILE_NO_UI; - bfd = BLO_read_from_memory(blend, blend_size, NULL); + bfd = BLO_read_from_memory(blend, blend_size, NULL, BLO_READ_SKIP_NONE); if (bfd) { bmain = bfd->main; diff --git a/source/blender/editors/screen/area.c b/source/blender/editors/screen/area.c index e6bb604d387..9cfaf3b4c1d 100644 --- a/source/blender/editors/screen/area.c +++ b/source/blender/editors/screen/area.c @@ -1749,7 +1749,7 @@ int ED_area_header_switchbutton(const bContext *C, uiBlock *block, int yco) RNA_pointer_create(&(scr->id), &RNA_Area, sa, &areaptr); - uiDefButR(block, UI_BTYPE_MENU, 0, "", xco, yco, 1.5 * U.widget_unit, U.widget_unit, + uiDefButR(block, UI_BTYPE_MENU, 0, "", xco, yco, 1.6 * U.widget_unit, U.widget_unit, &areaptr, "type", 0, 0.0f, 0.0f, 0.0f, 0.0f, ""); return xco + 1.7 * U.widget_unit; diff --git a/source/blender/editors/screen/screen_ops.c b/source/blender/editors/screen/screen_ops.c index a7a0a240259..c2e094fc161 100644 --- a/source/blender/editors/screen/screen_ops.c +++ b/source/blender/editors/screen/screen_ops.c @@ -2808,7 +2808,7 @@ static int screen_area_options_invoke(bContext *C, wmOperator *op, const wmEvent bScreen *sc = CTX_wm_screen(C); uiPopupMenu *pup; uiLayout *layout; - PointerRNA ptr1, ptr2; + PointerRNA ptr; ScrEdge *actedge; const int winsize_x = WM_window_pixels_x(win); const int winsize_y = WM_window_pixels_y(win); @@ -2820,22 +2820,17 @@ static int screen_area_options_invoke(bContext *C, wmOperator *op, const wmEvent pup = UI_popup_menu_begin(C, RNA_struct_ui_name(op->type->srna), ICON_NONE); layout = UI_popup_menu_layout(pup); - WM_operator_properties_create(&ptr1, "SCREEN_OT_area_join"); - - /* mouse cursor on edge, '4' can fail on wide edges... */ - RNA_int_set(&ptr1, "min_x", event->x + 4); - RNA_int_set(&ptr1, "min_y", event->y + 4); - RNA_int_set(&ptr1, "max_x", event->x - 4); - RNA_int_set(&ptr1, "max_y", event->y - 4); - - WM_operator_properties_create(&ptr2, "SCREEN_OT_area_split"); - + ptr = uiItemFullO(layout, "SCREEN_OT_area_split", NULL, ICON_NONE, NULL, WM_OP_INVOKE_DEFAULT, UI_ITEM_O_RETURN_PROPS); /* store initial mouse cursor position */ - RNA_int_set(&ptr2, "mouse_x", event->x); - RNA_int_set(&ptr2, "mouse_y", event->y); - - uiItemFullO(layout, "SCREEN_OT_area_split", NULL, ICON_NONE, ptr2.data, WM_OP_INVOKE_DEFAULT, 0); - uiItemFullO(layout, "SCREEN_OT_area_join", NULL, ICON_NONE, ptr1.data, WM_OP_INVOKE_DEFAULT, 0); + RNA_int_set(&ptr, "mouse_x", event->x); + RNA_int_set(&ptr, "mouse_y", event->y); + + ptr = uiItemFullO(layout, "SCREEN_OT_area_join", NULL, ICON_NONE, NULL, WM_OP_INVOKE_DEFAULT, UI_ITEM_O_RETURN_PROPS); + /* mouse cursor on edge, '4' can fail on wide edges... */ + RNA_int_set(&ptr, "min_x", event->x + 4); + RNA_int_set(&ptr, "min_y", event->y + 4); + RNA_int_set(&ptr, "max_x", event->x - 4); + RNA_int_set(&ptr, "max_y", event->y - 4); UI_popup_menu_end(C, pup); @@ -2903,10 +2898,23 @@ static void SCREEN_OT_spacedata_cleanup(wmOperatorType *ot) static int repeat_last_exec(bContext *C, wmOperator *UNUSED(op)) { - wmOperator *lastop = CTX_wm_manager(C)->operators.last; - - if (lastop) + wmWindowManager *wm = CTX_wm_manager(C); + wmOperator *lastop = wm->operators.last; + + /* Seek last registered operator */ + while (lastop) { + if (lastop->type->flag & OPTYPE_REGISTER) { + break; + } + else { + lastop = lastop->prev; + } + } + + if (lastop) { + WM_operator_free_all_after(wm, lastop); WM_operator_repeat(C, lastop); + } return OPERATOR_CANCELLED; } @@ -2941,8 +2949,9 @@ static int repeat_history_invoke(bContext *C, wmOperator *op, const wmEvent *UNU layout = UI_popup_menu_layout(pup); for (i = items - 1, lastop = wm->operators.last; lastop; lastop = lastop->prev, i--) - if (WM_operator_repeat_check(C, lastop)) + if ((lastop->type->flag & OPTYPE_REGISTER) && WM_operator_repeat_check(C, lastop)) { uiItemIntO(layout, RNA_struct_ui_name(lastop->type->srna), ICON_NONE, op->type->idname, "index", i); + } UI_popup_menu_end(C, pup); diff --git a/source/blender/editors/sculpt_paint/paint_image_proj.c b/source/blender/editors/sculpt_paint/paint_image_proj.c index f5d115442c6..d0f1cc99b8d 100644 --- a/source/blender/editors/sculpt_paint/paint_image_proj.c +++ b/source/blender/editors/sculpt_paint/paint_image_proj.c @@ -5711,21 +5711,16 @@ static bool proj_paint_add_slot(bContext *C, wmOperator *op) /* successful creation of mtex layer, now create set */ if (mtex) { int type = MAP_COL; - int type_id = 0; + char imagename_buff[MAX_ID_NAME - 2]; + const char *imagename = DATA_("Diffuse Color"); if (op) { - int i; type = RNA_enum_get(op->ptr, "type"); - - for (i = 0; i < ARRAY_SIZE(layer_type_items); i++) { - if (layer_type_items[i].value == type) { - type_id = i; - break; - } - } + RNA_string_get(op->ptr, "name", imagename_buff); + imagename = imagename_buff; } - mtex->tex = BKE_texture_add(bmain, DATA_(layer_type_items[type_id].name)); + mtex->tex = BKE_texture_add(bmain, imagename); mtex->mapto = type; if (mtex->tex) { diff --git a/source/blender/editors/sculpt_paint/sculpt.c b/source/blender/editors/sculpt_paint/sculpt.c index 84e98181dfb..44cc2720a32 100644 --- a/source/blender/editors/sculpt_paint/sculpt.c +++ b/source/blender/editors/sculpt_paint/sculpt.c @@ -5361,8 +5361,12 @@ static int sculpt_mode_toggle_exec(bContext *C, wmOperator *op) if (mmd) multires_force_update(ob); - if (flush_recalc || (ob->sculpt && ob->sculpt->bm)) + /* Always for now, so leaving sculpt mode always ensures scene is in + * a consistent state. + */ + if (true || flush_recalc || (ob->sculpt && ob->sculpt->bm)) { DAG_id_tag_update(&ob->id, OB_RECALC_DATA); + } if (me->flag & ME_SCULPT_DYNAMIC_TOPOLOGY) { /* Dynamic topology must be disabled before exiting sculpt diff --git a/source/blender/editors/space_clip/tracking_ops.c b/source/blender/editors/space_clip/tracking_ops.c index d28cbe5fb1d..169eb76399b 100644 --- a/source/blender/editors/space_clip/tracking_ops.c +++ b/source/blender/editors/space_clip/tracking_ops.c @@ -1534,7 +1534,8 @@ static int join_tracks_exec(bContext *C, wmOperator *op) update_stabilization = true; if ((act_track->flag & TRACK_USE_2D_STAB) == 0) { act_track->flag |= TRACK_USE_2D_STAB; - } else { + } + else { stab->tot_track--; } BLI_assert(0 <= stab->tot_track); @@ -1543,7 +1544,8 @@ static int join_tracks_exec(bContext *C, wmOperator *op) update_stabilization = true; if ((act_track->flag & TRACK_USE_2D_STAB_ROT) == 0) { act_track->flag |= TRACK_USE_2D_STAB_ROT; - } else { + } + else { stab->tot_rot_track--; } BLI_assert(0 <= stab->tot_rot_track); diff --git a/source/blender/editors/space_file/fsmenu.c b/source/blender/editors/space_file/fsmenu.c index 631ff06a77a..71d49e0dc2e 100644 --- a/source/blender/editors/space_file/fsmenu.c +++ b/source/blender/editors/space_file/fsmenu.c @@ -169,12 +169,15 @@ void ED_fsmenu_entry_set_path(struct FSMenuEntry *fsentry, const char *path) static void fsmenu_entry_generate_name(struct FSMenuEntry *fsentry, char *name, size_t name_size) { - char temp[FILE_MAX]; + int offset = 0; + int len = name_size; - BLI_strncpy(temp, fsentry->path, FILE_MAX); - BLI_add_slash(temp); - BLI_getlastdir(temp, name, name_size); - BLI_del_slash(name); + if (BLI_path_name_at_index(fsentry->path, -1, &offset, &len)) { + /* use as size */ + len += 1; + } + + BLI_strncpy(name, &fsentry->path[offset], MIN2(len, name_size)); if (!name[0]) { name[0] = '/'; name[1] = '\0'; @@ -537,28 +540,22 @@ void fsmenu_read_system(struct FSMenu *fsmenu, int read_bookmarks) /* Finally get user favorite places */ if (read_bookmarks) { UInt32 seed; - OSErr err = noErr; - CFArrayRef pathesArray; - LSSharedFileListRef list; - LSSharedFileListItemRef itemRef; - CFIndex i, pathesCount; - CFURLRef cfURL = NULL; - CFStringRef pathString = NULL; - list = LSSharedFileListCreate(NULL, kLSSharedFileListFavoriteItems, NULL); - pathesArray = LSSharedFileListCopySnapshot(list, &seed); - pathesCount = CFArrayGetCount(pathesArray); + LSSharedFileListRef list = LSSharedFileListCreate(NULL, kLSSharedFileListFavoriteItems, NULL); + CFArrayRef pathesArray = LSSharedFileListCopySnapshot(list, &seed); + CFIndex pathesCount = CFArrayGetCount(pathesArray); - for (i = 0; i < pathesCount; i++) { - itemRef = (LSSharedFileListItemRef)CFArrayGetValueAtIndex(pathesArray, i); + for (CFIndex i = 0; i < pathesCount; i++) { + LSSharedFileListItemRef itemRef = (LSSharedFileListItemRef)CFArrayGetValueAtIndex(pathesArray, i); - err = LSSharedFileListItemResolve(itemRef, - kLSSharedFileListNoUserInteraction | - kLSSharedFileListDoNotMountVolumes, - &cfURL, NULL); - if (err != noErr) + CFURLRef cfURL = NULL; + OSErr err = LSSharedFileListItemResolve(itemRef, + kLSSharedFileListNoUserInteraction | + kLSSharedFileListDoNotMountVolumes, + &cfURL, NULL); + if (err != noErr || !cfURL) continue; - pathString = CFURLCopyFileSystemPath(cfURL, kCFURLPOSIXPathStyle); + CFStringRef pathString = CFURLCopyFileSystemPath(cfURL, kCFURLPOSIXPathStyle); if (pathString == NULL || !CFStringGetCString(pathString, line, sizeof(line), kCFStringEncodingUTF8)) continue; diff --git a/source/blender/editors/space_graph/graph_buttons.c b/source/blender/editors/space_graph/graph_buttons.c index 516814b63b4..cbb8e98e7e0 100644 --- a/source/blender/editors/space_graph/graph_buttons.c +++ b/source/blender/editors/space_graph/graph_buttons.c @@ -616,30 +616,31 @@ static void graph_panel_driverVar__rotDiff(uiLayout *layout, ID *id, DriverVar * Object *ob2 = (Object *)dtar2->id; PointerRNA dtar_ptr, dtar2_ptr; uiLayout *col; - + /* initialize RNA pointer to the target */ - RNA_pointer_create(id, &RNA_DriverTarget, dtar, &dtar_ptr); - RNA_pointer_create(id, &RNA_DriverTarget, dtar2, &dtar2_ptr); - - /* Bone 1 */ + RNA_pointer_create(id, &RNA_DriverTarget, dtar, &dtar_ptr); + RNA_pointer_create(id, &RNA_DriverTarget, dtar2, &dtar2_ptr); + + /* Object 1 */ col = uiLayoutColumn(layout, true); uiLayoutSetRedAlert(col, (dtar->flag & DTAR_FLAG_INVALID)); /* XXX: per field... */ - uiItemR(col, &dtar_ptr, "id", 0, IFACE_("Bone 1"), ICON_NONE); - + uiItemR(col, &dtar_ptr, "id", 0, IFACE_("Object 1"), ICON_NONE); + if (dtar->id && GS(dtar->id->name) == ID_OB && ob1->pose) { PointerRNA tar_ptr; - + RNA_pointer_create(dtar->id, &RNA_Pose, ob1->pose, &tar_ptr); uiItemPointerR(col, &dtar_ptr, "bone_target", &tar_ptr, "bones", "", ICON_BONE_DATA); } - + + /* Object 2 */ col = uiLayoutColumn(layout, true); uiLayoutSetRedAlert(col, (dtar2->flag & DTAR_FLAG_INVALID)); /* XXX: per field... */ - uiItemR(col, &dtar2_ptr, "id", 0, IFACE_("Bone 2"), ICON_NONE); - + uiItemR(col, &dtar2_ptr, "id", 0, IFACE_("Object 2"), ICON_NONE); + if (dtar2->id && GS(dtar2->id->name) == ID_OB && ob2->pose) { PointerRNA tar_ptr; - + RNA_pointer_create(dtar2->id, &RNA_Pose, ob2->pose, &tar_ptr); uiItemPointerR(col, &dtar2_ptr, "bone_target", &tar_ptr, "bones", "", ICON_BONE_DATA); } @@ -658,8 +659,8 @@ static void graph_panel_driverVar__locDiff(uiLayout *layout, ID *id, DriverVar * /* initialize RNA pointer to the target */ RNA_pointer_create(id, &RNA_DriverTarget, dtar, &dtar_ptr); RNA_pointer_create(id, &RNA_DriverTarget, dtar2, &dtar2_ptr); - - /* Bone 1 */ + + /* Object 1 */ col = uiLayoutColumn(layout, true); uiLayoutSetRedAlert(col, (dtar->flag & DTAR_FLAG_INVALID)); /* XXX: per field... */ uiItemR(col, &dtar_ptr, "id", 0, IFACE_("Object 1"), ICON_NONE); @@ -673,7 +674,8 @@ static void graph_panel_driverVar__locDiff(uiLayout *layout, ID *id, DriverVar * uiLayoutSetRedAlert(col, false); /* we can clear it again now - it's only needed when creating the ID/Bone fields */ uiItemR(col, &dtar_ptr, "transform_space", 0, NULL, ICON_NONE); - + + /* Object 2 */ col = uiLayoutColumn(layout, true); uiLayoutSetRedAlert(col, (dtar2->flag & DTAR_FLAG_INVALID)); /* XXX: per field... */ uiItemR(col, &dtar2_ptr, "id", 0, IFACE_("Object 2"), ICON_NONE); diff --git a/source/blender/editors/space_nla/nla_draw.c b/source/blender/editors/space_nla/nla_draw.c index 5b3c062e16d..93dcdbb5c02 100644 --- a/source/blender/editors/space_nla/nla_draw.c +++ b/source/blender/editors/space_nla/nla_draw.c @@ -290,7 +290,8 @@ static void nla_draw_strip_curves(NlaStrip *strip, float yminc, float ymaxc) * - min y-val is yminc, max is y-maxc, so clamp in those regions */ for (cfra = strip->start; cfra <= strip->end; cfra += 1.0f) { - float y = evaluate_fcurve(fcu, cfra); // assume this to be in 0-1 range + float y = evaluate_fcurve(fcu, cfra); + CLAMP(y, 0.0f, 1.0f); glVertex2f(cfra, ((y * yheight) + yminc)); } glEnd(); // GL_LINE_STRIP diff --git a/source/blender/editors/space_node/drawnode.c b/source/blender/editors/space_node/drawnode.c index a6ee80d6feb..8dba0e39524 100644 --- a/source/blender/editors/space_node/drawnode.c +++ b/source/blender/editors/space_node/drawnode.c @@ -1321,9 +1321,6 @@ static void node_composit_buts_renderlayers(uiLayout *layout, bContext *C, Point PropertyRNA *prop; const char *layer_name; char scene_name[MAX_ID_NAME - 2]; - wmOperatorType *ot = WM_operatortype_find("RENDER_OT_render", 1); - - BLI_assert(ot != 0); uiTemplateID(layout, C, ptr, "scene", NULL, NULL, NULL); @@ -1340,11 +1337,9 @@ static void node_composit_buts_renderlayers(uiLayout *layout, bContext *C, Point scn_ptr = RNA_pointer_get(ptr, "scene"); RNA_string_get(&scn_ptr, "name", scene_name); - WM_operator_properties_create_ptr(&op_ptr, ot); + op_ptr = uiItemFullO(row, "RENDER_OT_render", "", ICON_RENDER_STILL, NULL, WM_OP_INVOKE_DEFAULT, UI_ITEM_O_RETURN_PROPS); RNA_string_set(&op_ptr, "layer", layer_name); RNA_string_set(&op_ptr, "scene", scene_name); - uiItemFullO_ptr(row, ot, "", ICON_RENDER_STILL, op_ptr.data, WM_OP_INVOKE_DEFAULT, 0); - } @@ -1793,6 +1788,7 @@ static void node_composit_buts_file_output_ex(uiLayout *layout, bContext *C, Poi Scene *scene = CTX_data_scene(C); PointerRNA imfptr = RNA_pointer_get(ptr, "format"); PointerRNA active_input_ptr, op_ptr; + wmOperatorType *ot; uiLayout *row, *col; int active_index; const bool multilayer = RNA_enum_get(&imfptr, "file_format") == R_IMF_IMTYPE_MULTILAYER; @@ -1831,11 +1827,10 @@ static void node_composit_buts_file_output_ex(uiLayout *layout, bContext *C, Poi active_input_ptr.id.data = ptr->id.data; col = uiLayoutColumn(row, true); - op_ptr = uiItemFullO(col, "NODE_OT_output_file_move_active_socket", "", - ICON_TRIA_UP, NULL, WM_OP_INVOKE_DEFAULT, UI_ITEM_O_RETURN_PROPS); + ot = WM_operatortype_find("NODE_OT_output_file_move_active_socket", false); + op_ptr = uiItemFullO_ptr(col, ot, "", ICON_TRIA_UP, NULL, WM_OP_INVOKE_DEFAULT, UI_ITEM_O_RETURN_PROPS); RNA_enum_set(&op_ptr, "direction", 1); - op_ptr = uiItemFullO(col, "NODE_OT_output_file_move_active_socket", "", - ICON_TRIA_DOWN, NULL, WM_OP_INVOKE_DEFAULT, UI_ITEM_O_RETURN_PROPS); + op_ptr = uiItemFullO_ptr(col, ot, "", ICON_TRIA_DOWN, NULL, WM_OP_INVOKE_DEFAULT, UI_ITEM_O_RETURN_PROPS); RNA_enum_set(&op_ptr, "direction", 2); if (active_input_ptr.data) { @@ -2167,14 +2162,7 @@ static void node_composit_buts_switch(uiLayout *layout, bContext *UNUSED(C), Poi static void node_composit_buts_switch_view_ex(uiLayout *layout, bContext *UNUSED(C), PointerRNA *UNUSED(ptr)) { - PointerRNA op_ptr; - wmOperatorType *ot = WM_operatortype_find("NODE_OT_switch_view_update", 1); - - BLI_assert(ot != 0); - - WM_operator_properties_create_ptr(&op_ptr, ot); - - uiItemFullO_ptr(layout, ot, "Update Views", ICON_FILE_REFRESH, op_ptr.data, WM_OP_INVOKE_DEFAULT, 0); + uiItemFullO(layout, "NODE_OT_switch_view_update", "Update Views", ICON_FILE_REFRESH, NULL, WM_OP_INVOKE_DEFAULT, 0); } static void node_composit_buts_boxmask(uiLayout *layout, bContext *UNUSED(C), PointerRNA *ptr) diff --git a/source/blender/editors/space_node/node_add.c b/source/blender/editors/space_node/node_add.c index d49df2afbd4..bcd0f6623e1 100644 --- a/source/blender/editors/space_node/node_add.c +++ b/source/blender/editors/space_node/node_add.c @@ -44,6 +44,8 @@ #include "BKE_main.h" #include "BKE_node.h" #include "BKE_report.h" +#include "BKE_scene.h" +#include "BKE_texture.h" #include "ED_node.h" /* own include */ #include "ED_screen.h" @@ -312,7 +314,10 @@ static int node_add_file_exec(bContext *C, wmOperator *op) switch (snode->nodetree->type) { case NTREE_SHADER: - type = SH_NODE_TEX_IMAGE; + if (BKE_scene_use_new_shading_nodes(CTX_data_scene(C))) + type = SH_NODE_TEX_IMAGE; + else + type = SH_NODE_TEXTURE; break; case NTREE_TEXTURE: type = TEX_NODE_IMAGE; @@ -333,7 +338,14 @@ static int node_add_file_exec(bContext *C, wmOperator *op) return OPERATOR_CANCELLED; } - node->id = (ID *)ima; + if (type == SH_NODE_TEXTURE) { + Tex *tex = BKE_texture_add(CTX_data_main(C), DATA_(ima->id.name)); + tex->ima = ima; + node->id = (ID *)tex; + WM_event_add_notifier(C, NC_TEXTURE | NA_ADDED, node->id); + } + else + node->id = (ID *)ima; /* When adding new image file via drag-drop we need to load imbuf in order * to get proper image source. diff --git a/source/blender/editors/space_node/node_buttons.c b/source/blender/editors/space_node/node_buttons.c index f0567924edd..925298451ce 100644 --- a/source/blender/editors/space_node/node_buttons.c +++ b/source/blender/editors/space_node/node_buttons.c @@ -133,6 +133,7 @@ static void node_tree_interface_panel(const bContext *C, Panel *pa) int in_out; uiLayout *layout = pa->layout, *row, *split, *col; PointerRNA ptr, sockptr, opptr; + wmOperatorType *ot; if (!ntree) return; @@ -146,23 +147,25 @@ static void node_tree_interface_panel(const bContext *C, Panel *pa) split = uiLayoutRow(row, true); col = uiLayoutColumn(split, true); + ot = WM_operatortype_find("NODE_OT_tree_socket_add", false); uiItemL(col, IFACE_("Inputs:"), ICON_NONE); uiTemplateList(col, (bContext *)C, "NODE_UL_interface_sockets", "inputs", &ptr, "inputs", &ptr, "active_input", NULL, 0, 0, 0, 0); - opptr = uiItemFullO(col, "NODE_OT_tree_socket_add", "", ICON_PLUS, NULL, WM_OP_EXEC_DEFAULT, UI_ITEM_O_RETURN_PROPS); + opptr = uiItemFullO_ptr(col, ot, "", ICON_PLUS, NULL, WM_OP_EXEC_DEFAULT, UI_ITEM_O_RETURN_PROPS); RNA_enum_set(&opptr, "in_out", SOCK_IN); col = uiLayoutColumn(split, true); uiItemL(col, IFACE_("Outputs:"), ICON_NONE); uiTemplateList(col, (bContext *)C, "NODE_UL_interface_sockets", "outputs", &ptr, "outputs", &ptr, "active_output", NULL, 0, 0, 0, 0); - opptr = uiItemFullO(col, "NODE_OT_tree_socket_add", "", ICON_PLUS, NULL, WM_OP_EXEC_DEFAULT, UI_ITEM_O_RETURN_PROPS); + opptr = uiItemFullO_ptr(col, ot, "", ICON_PLUS, NULL, WM_OP_EXEC_DEFAULT, UI_ITEM_O_RETURN_PROPS); RNA_enum_set(&opptr, "in_out", SOCK_OUT); + ot = WM_operatortype_find("NODE_OT_tree_socket_move", false); col = uiLayoutColumn(row, true); - opptr = uiItemFullO(col, "NODE_OT_tree_socket_move", "", ICON_TRIA_UP, NULL, WM_OP_EXEC_DEFAULT, UI_ITEM_O_RETURN_PROPS); + opptr = uiItemFullO_ptr(col, ot, "", ICON_TRIA_UP, NULL, WM_OP_EXEC_DEFAULT, UI_ITEM_O_RETURN_PROPS); RNA_enum_set(&opptr, "direction", 1); - opptr = uiItemFullO(col, "NODE_OT_tree_socket_move", "", ICON_TRIA_DOWN, NULL, WM_OP_EXEC_DEFAULT, UI_ITEM_O_RETURN_PROPS); + opptr = uiItemFullO_ptr(col, ot, "", ICON_TRIA_DOWN, NULL, WM_OP_EXEC_DEFAULT, UI_ITEM_O_RETURN_PROPS); RNA_enum_set(&opptr, "direction", 2); if (sock) { diff --git a/source/blender/editors/space_node/node_edit.c b/source/blender/editors/space_node/node_edit.c index ffe510016ff..fdfe316f5ed 100644 --- a/source/blender/editors/space_node/node_edit.c +++ b/source/blender/editors/space_node/node_edit.c @@ -582,7 +582,7 @@ void snode_set_context(const bContext *C) } } - if (snode->nodetree != ntree || snode->id != id || snode->from != from) { + if (snode->nodetree != ntree || snode->id != id || snode->from != from || snode->treepath.last == NULL) { ED_node_tree_start(snode, ntree, id, from); } @@ -1069,12 +1069,9 @@ int node_find_indicated_socket(SpaceNode *snode, bNode **nodep, bNodeSocket **so /* check if we click in a socket */ for (node = snode->edittree->nodes.first; node; node = node->next) { - - rect.xmin = cursor[0] - (NODE_SOCKSIZE + 4); - rect.ymin = cursor[1] - (NODE_SOCKSIZE + 4); - rect.xmax = cursor[0] + (NODE_SOCKSIZE + 4); - rect.ymax = cursor[1] + (NODE_SOCKSIZE + 4); - + + BLI_rctf_init_pt_radius(&rect, cursor, NODE_SOCKSIZE + 4); + if (!(node->flag & NODE_HIDDEN)) { /* extra padding inside and out - allow dragging on the text areas too */ if (in_out == SOCK_IN) { diff --git a/source/blender/editors/space_outliner/outliner_draw.c b/source/blender/editors/space_outliner/outliner_draw.c index 99242fd12f9..121fe812016 100644 --- a/source/blender/editors/space_outliner/outliner_draw.c +++ b/source/blender/editors/space_outliner/outliner_draw.c @@ -1040,9 +1040,6 @@ static void tselem_draw_icon(uiBlock *block, int xmax, float x, float y, TreeSto struct DrawIconArg arg; float aspect; - /* icons tiny bit away from text */ - x -= 0.15f * UI_UNIT_Y; - /* make function calls a bit compacter */ arg.block = block; arg.id = tselem->id; @@ -1053,8 +1050,10 @@ static void tselem_draw_icon(uiBlock *block, int xmax, float x, float y, TreeSto /* placement of icons, copied from interface_widgets.c */ aspect = (0.8f * UI_UNIT_Y) / ICON_DEFAULT_HEIGHT; - arg.x = x = x + 4.0f * aspect; - arg.y = y = y + 0.1f * UI_UNIT_Y; + x += 2.0f * aspect; + y += 2.0f * aspect; + arg.x = x; + arg.y = y; if (tselem->type) { switch (tselem->type) { @@ -1126,6 +1125,7 @@ static void tselem_draw_icon(uiBlock *block, int xmax, float x, float y, TreeSto case eModifierType_Cast: UI_icon_draw(x, y, ICON_MOD_CAST); break; case eModifierType_MeshDeform: + case eModifierType_SurfaceDeform: UI_icon_draw(x, y, ICON_MOD_MESHDEFORM); break; case eModifierType_Bevel: UI_icon_draw(x, y, ICON_MOD_BEVEL); break; @@ -1388,9 +1388,9 @@ static void outliner_draw_iconrow(bContext *C, uiBlock *block, Scene *scene, Spa UI_draw_roundbox_corner_set(UI_CNR_ALL); glColor4ub(255, 255, 255, 100); UI_draw_roundbox( - (float) *offsx - 1.0f * ufac, + (float) *offsx + 1.0f * ufac, (float)ys + 1.0f * ufac, - (float)*offsx + UI_UNIT_X - 2.0f * ufac, + (float)*offsx + UI_UNIT_X - 1.0f * ufac, (float)ys + UI_UNIT_Y - ufac, (float)UI_UNIT_Y / 2.0f - ufac); glEnable(GL_BLEND); /* roundbox disables */ @@ -1535,9 +1535,9 @@ static void outliner_draw_tree_element( if (active != OL_DRAWSEL_NONE) { UI_draw_roundbox_corner_set(UI_CNR_ALL); UI_draw_roundbox( - (float)startx + UI_UNIT_X, + (float)startx + UI_UNIT_X + 1.0f * ufac, (float)*starty + 1.0f * ufac, - (float)startx + 2.0f * UI_UNIT_X - 2.0f * ufac, + (float)startx + 2.0f * UI_UNIT_X - 1.0f * ufac, (float)*starty + UI_UNIT_Y - 1.0f * ufac, UI_UNIT_Y / 2.0f - 1.0f * ufac); glEnable(GL_BLEND); /* roundbox disables it */ @@ -1548,16 +1548,13 @@ static void outliner_draw_tree_element( /* open/close icon, only when sublevels, except for scene */ if (te->subtree.first || (tselem->type == 0 && te->idcode == ID_SCE) || (te->flag & TE_LAZY_CLOSED)) { int icon_x; - if (tselem->type == 0 && ELEM(te->idcode, ID_OB, ID_SCE)) - icon_x = startx; - else - icon_x = startx + 5 * ufac; + icon_x = startx; // icons a bit higher if (TSELEM_OPEN(tselem, soops)) - UI_icon_draw((float)icon_x, (float)*starty + 2 * ufac, ICON_DISCLOSURE_TRI_DOWN); + UI_icon_draw((float)icon_x + 2 * ufac, (float)*starty + 1 * ufac, ICON_DISCLOSURE_TRI_DOWN); else - UI_icon_draw((float)icon_x, (float)*starty + 2 * ufac, ICON_DISCLOSURE_TRI_RIGHT); + UI_icon_draw((float)icon_x + 2 * ufac, (float)*starty + 1 * ufac, ICON_DISCLOSURE_TRI_RIGHT); } offsx += UI_UNIT_X; @@ -1567,7 +1564,7 @@ static void outliner_draw_tree_element( tselem_draw_icon(block, xmax, (float)startx + offsx, (float)*starty, tselem, te, 1.0f); - offsx += UI_UNIT_X; + offsx += UI_UNIT_X + 2 * ufac; } else offsx += 2 * ufac; @@ -1575,16 +1572,16 @@ static void outliner_draw_tree_element( if (tselem->type == 0 && ID_IS_LINKED_DATABLOCK(tselem->id)) { glPixelTransferf(GL_ALPHA_SCALE, 0.5f); if (tselem->id->tag & LIB_TAG_MISSING) { - UI_icon_draw((float)startx + offsx, (float)*starty + 2 * ufac, ICON_LIBRARY_DATA_BROKEN); + UI_icon_draw((float)startx + offsx + 2 * ufac, (float)*starty + 2 * ufac, ICON_LIBRARY_DATA_BROKEN); } else if (tselem->id->tag & LIB_TAG_INDIRECT) { - UI_icon_draw((float)startx + offsx, (float)*starty + 2 * ufac, ICON_LIBRARY_DATA_INDIRECT); + UI_icon_draw((float)startx + offsx + 2 * ufac, (float)*starty + 2 * ufac, ICON_LIBRARY_DATA_INDIRECT); } else { - UI_icon_draw((float)startx + offsx, (float)*starty + 2 * ufac, ICON_LIBRARY_DATA_DIRECT); + UI_icon_draw((float)startx + offsx + 2 * ufac, (float)*starty + 2 * ufac, ICON_LIBRARY_DATA_DIRECT); } glPixelTransferf(GL_ALPHA_SCALE, 1.0f); - offsx += UI_UNIT_X; + offsx += UI_UNIT_X + 2 * ufac; } glDisable(GL_BLEND); @@ -1756,7 +1753,7 @@ static void outliner_draw_tree(bContext *C, uiBlock *block, Scene *scene, ARegio // gray hierarchy lines UI_ThemeColorBlend(TH_BACK, TH_TEXT, 0.4f); starty = (int)ar->v2d.tot.ymax - UI_UNIT_Y / 2 - OL_Y_OFFSET; - startx = 6; + startx = UI_UNIT_X / 2 - 1.0f; outliner_draw_hierarchy(soops, &soops->tree, startx, &starty); // items themselves diff --git a/source/blender/editors/space_outliner/outliner_edit.c b/source/blender/editors/space_outliner/outliner_edit.c index 345ac353c11..335eb95da0e 100644 --- a/source/blender/editors/space_outliner/outliner_edit.c +++ b/source/blender/editors/space_outliner/outliner_edit.c @@ -309,9 +309,12 @@ static void id_delete(bContext *C, ReportList *reports, TreeElement *te, TreeSto ID *id = tselem->id; BLI_assert(te->idcode != 0 && id != NULL); - BLI_assert(te->idcode != ID_LI || ((Library *)id)->parent == NULL); UNUSED_VARS_NDEBUG(te); + if (te->idcode == ID_LI && ((Library *)id)->parent != NULL) { + BKE_reportf(reports, RPT_WARNING, "Cannot delete indirectly linked library '%s'", id->name); + return; + } if (id->tag & LIB_TAG_INDIRECT) { BKE_reportf(reports, RPT_WARNING, "Cannot delete indirectly linked id '%s'", id->name); return; @@ -516,7 +519,8 @@ void OUTLINER_OT_id_remap(wmOperatorType *ot) ot->flag = 0; - RNA_def_enum(ot->srna, "id_type", rna_enum_id_type_items, ID_OB, "ID Type", ""); + prop = RNA_def_enum(ot->srna, "id_type", rna_enum_id_type_items, ID_OB, "ID Type", ""); + RNA_def_property_translation_context(prop, BLT_I18NCONTEXT_ID_ID); prop = RNA_def_enum(ot->srna, "old_id", DummyRNA_NULL_items, 0, "Old ID", "Old ID to replace"); RNA_def_property_enum_funcs_runtime(prop, NULL, NULL, outliner_id_itemf); @@ -2068,74 +2072,62 @@ static int parent_drop_invoke(bContext *C, wmOperator *op, const wmEvent *event) wmOperatorType *ot = WM_operatortype_find("OUTLINER_OT_parent_drop", false); uiPopupMenu *pup = UI_popup_menu_begin(C, IFACE_("Set Parent To"), ICON_NONE); uiLayout *layout = UI_popup_menu_layout(pup); - PointerRNA ptr; - WM_operator_properties_create_ptr(&ptr, ot); + /* Cannot use uiItemEnumO()... have multiple properties to set. */ + ptr = uiItemFullO_ptr(layout, ot, IFACE_("Object"), 0, NULL, WM_OP_EXEC_DEFAULT, UI_ITEM_O_RETURN_PROPS); RNA_string_set(&ptr, "parent", parname); RNA_string_set(&ptr, "child", childname); RNA_enum_set(&ptr, "type", PAR_OBJECT); - /* Cannot use uiItemEnumO()... have multiple properties to set. */ - uiItemFullO_ptr(layout, ot, IFACE_("Object"), 0, ptr.data, WM_OP_EXEC_DEFAULT, 0); - + /* par becomes parent, make the associated menus */ if (par->type == OB_ARMATURE) { - WM_operator_properties_create_ptr(&ptr, ot); + ptr = uiItemFullO_ptr(layout, ot, IFACE_("Armature Deform"), 0, NULL, WM_OP_EXEC_DEFAULT, UI_ITEM_O_RETURN_PROPS); RNA_string_set(&ptr, "parent", parname); RNA_string_set(&ptr, "child", childname); RNA_enum_set(&ptr, "type", PAR_ARMATURE); - uiItemFullO_ptr(layout, ot, IFACE_("Armature Deform"), 0, ptr.data, WM_OP_EXEC_DEFAULT, 0); - - WM_operator_properties_create_ptr(&ptr, ot); + + ptr = uiItemFullO_ptr(layout, ot, IFACE_(" With Empty Groups"), 0, NULL, WM_OP_EXEC_DEFAULT, UI_ITEM_O_RETURN_PROPS); RNA_string_set(&ptr, "parent", parname); RNA_string_set(&ptr, "child", childname); RNA_enum_set(&ptr, "type", PAR_ARMATURE_NAME); - uiItemFullO_ptr(layout, ot, IFACE_(" With Empty Groups"), 0, ptr.data, WM_OP_EXEC_DEFAULT, 0); - - WM_operator_properties_create_ptr(&ptr, ot); + + ptr = uiItemFullO_ptr(layout, ot, IFACE_(" With Envelope Weights"), 0, NULL, WM_OP_EXEC_DEFAULT, UI_ITEM_O_RETURN_PROPS); RNA_string_set(&ptr, "parent", parname); RNA_string_set(&ptr, "child", childname); RNA_enum_set(&ptr, "type", PAR_ARMATURE_ENVELOPE); - uiItemFullO_ptr(layout, ot, IFACE_(" With Envelope Weights"), 0, ptr.data, WM_OP_EXEC_DEFAULT, 0); - - WM_operator_properties_create_ptr(&ptr, ot); + + ptr = uiItemFullO_ptr(layout, ot, IFACE_(" With Automatic Weights"), 0, NULL, WM_OP_EXEC_DEFAULT, UI_ITEM_O_RETURN_PROPS); RNA_string_set(&ptr, "parent", parname); RNA_string_set(&ptr, "child", childname); RNA_enum_set(&ptr, "type", PAR_ARMATURE_AUTO); - uiItemFullO_ptr(layout, ot, IFACE_(" With Automatic Weights"), 0, ptr.data, WM_OP_EXEC_DEFAULT, 0); - - WM_operator_properties_create_ptr(&ptr, ot); + + ptr = uiItemFullO_ptr(layout, ot, IFACE_("Bone"), 0, NULL, WM_OP_EXEC_DEFAULT, UI_ITEM_O_RETURN_PROPS); RNA_string_set(&ptr, "parent", parname); RNA_string_set(&ptr, "child", childname); RNA_enum_set(&ptr, "type", PAR_BONE); - uiItemFullO_ptr(layout, ot, IFACE_("Bone"), - 0, ptr.data, WM_OP_EXEC_DEFAULT, 0); } else if (par->type == OB_CURVE) { - WM_operator_properties_create_ptr(&ptr, ot); + ptr = uiItemFullO_ptr(layout, ot, IFACE_("Curve Deform"), 0, NULL, WM_OP_EXEC_DEFAULT, UI_ITEM_O_RETURN_PROPS); RNA_string_set(&ptr, "parent", parname); RNA_string_set(&ptr, "child", childname); RNA_enum_set(&ptr, "type", PAR_CURVE); - uiItemFullO_ptr(layout, ot, IFACE_("Curve Deform"), 0, ptr.data, WM_OP_EXEC_DEFAULT, 0); - - WM_operator_properties_create_ptr(&ptr, ot); + + ptr = uiItemFullO_ptr(layout, ot, IFACE_("Follow Path"), 0, NULL, WM_OP_EXEC_DEFAULT, UI_ITEM_O_RETURN_PROPS); RNA_string_set(&ptr, "parent", parname); RNA_string_set(&ptr, "child", childname); RNA_enum_set(&ptr, "type", PAR_FOLLOW); - uiItemFullO_ptr(layout, ot, IFACE_("Follow Path"), 0, ptr.data, WM_OP_EXEC_DEFAULT, 0); - - WM_operator_properties_create_ptr(&ptr, ot); + + ptr = uiItemFullO_ptr(layout, ot, IFACE_("Path Constraint"), 0, NULL, WM_OP_EXEC_DEFAULT, UI_ITEM_O_RETURN_PROPS); RNA_string_set(&ptr, "parent", parname); RNA_string_set(&ptr, "child", childname); RNA_enum_set(&ptr, "type", PAR_PATH_CONST); - uiItemFullO_ptr(layout, ot, IFACE_("Path Constraint"), 0, ptr.data, WM_OP_EXEC_DEFAULT, 0); } else if (par->type == OB_LATTICE) { - WM_operator_properties_create_ptr(&ptr, ot); + ptr = uiItemFullO_ptr(layout, ot, IFACE_("Lattice Deform"), 0, NULL, WM_OP_EXEC_DEFAULT, UI_ITEM_O_RETURN_PROPS); RNA_string_set(&ptr, "parent", parname); RNA_string_set(&ptr, "child", childname); RNA_enum_set(&ptr, "type", PAR_LATTICE); - uiItemFullO_ptr(layout, ot, IFACE_("Lattice Deform"), 0, ptr.data, WM_OP_EXEC_DEFAULT, 0); } UI_popup_menu_end(C, pup); diff --git a/source/blender/editors/space_outliner/outliner_tools.c b/source/blender/editors/space_outliner/outliner_tools.c index 4eda7977622..086cc55366b 100644 --- a/source/blender/editors/space_outliner/outliner_tools.c +++ b/source/blender/editors/space_outliner/outliner_tools.c @@ -1254,6 +1254,7 @@ static int outliner_id_operation_exec(bContext *C, wmOperator *op) { if (idlevel > 0) { outliner_do_libdata_operation(C, op->reports, scene, soops, &soops->tree, id_delete_cb, NULL); + ED_undo_push(C, "Delete"); } break; } @@ -1261,6 +1262,7 @@ static int outliner_id_operation_exec(bContext *C, wmOperator *op) { if (idlevel > 0) { outliner_do_libdata_operation(C, op->reports, scene, soops, &soops->tree, id_remap_cb, NULL); + ED_undo_push(C, "Remap"); } break; } @@ -1369,18 +1371,20 @@ static int outliner_lib_operation_exec(bContext *C, wmOperator *op) outliner_do_libdata_operation(C, op->reports, scene, soops, &soops->tree, item_rename_cb, NULL); WM_event_add_notifier(C, NC_ID | NA_EDITED, NULL); - ED_undo_push(C, "Rename"); + ED_undo_push(C, "Rename Library"); break; } case OL_LIB_DELETE: { outliner_do_libdata_operation(C, op->reports, scene, soops, &soops->tree, id_delete_cb, NULL); + ED_undo_push(C, "Delete Library"); break; } case OL_LIB_RELOCATE: { /* rename */ outliner_do_libdata_operation(C, op->reports, scene, soops, &soops->tree, lib_relocate_cb, NULL); + ED_undo_push(C, "Relocate Library"); break; } case OL_LIB_RELOAD: diff --git a/source/blender/editors/space_sequencer/sequencer_add.c b/source/blender/editors/space_sequencer/sequencer_add.c index ede6b7ce469..46f212e3679 100644 --- a/source/blender/editors/space_sequencer/sequencer_add.c +++ b/source/blender/editors/space_sequencer/sequencer_add.c @@ -38,6 +38,8 @@ #include "BLI_math.h" #include "BLI_utildefines.h" +#include "BLT_translation.h" + #include "DNA_scene_types.h" #include "DNA_mask_types.h" @@ -445,6 +447,7 @@ void SEQUENCER_OT_movieclip_strip_add(struct wmOperatorType *ot) sequencer_generic_props__internal(ot, SEQPROP_STARTFRAME); prop = RNA_def_enum(ot->srna, "clip", DummyRNA_NULL_items, 0, "Clip", ""); RNA_def_enum_funcs(prop, RNA_movieclip_itemf); + RNA_def_property_translation_context(prop, BLT_I18NCONTEXT_ID_MOVIECLIP); RNA_def_property_flag(prop, PROP_ENUM_NO_TRANSLATE); ot->prop = prop; } diff --git a/source/blender/editors/space_sequencer/sequencer_draw.c b/source/blender/editors/space_sequencer/sequencer_draw.c index e1768e4aedc..70a6e6d83cb 100644 --- a/source/blender/editors/space_sequencer/sequencer_draw.c +++ b/source/blender/editors/space_sequencer/sequencer_draw.c @@ -545,7 +545,8 @@ static void draw_seq_text(View2D *v2d, SpaceSeq *sseq, Sequence *seq, float x1, if ((sseq->flag & SEQ_ALL_WAVEFORMS) || (seq->flag & SEQ_AUDIO_DRAW_WAVEFORM)) { str[0] = 0; str_len = 0; - } else if (seq->sound) { + } + else if (seq->sound) { str_len = BLI_snprintf(str, sizeof(str), "%s: %s | %d", name, seq->sound->name, seq->len); } diff --git a/source/blender/editors/space_text/text_draw.c b/source/blender/editors/space_text/text_draw.c index 81605a80f69..70b5feac280 100644 --- a/source/blender/editors/space_text/text_draw.c +++ b/source/blender/editors/space_text/text_draw.c @@ -926,6 +926,7 @@ static void draw_textscroll(const SpaceText *st, rcti *scroll, rcti *back) /*********************** draw documentation *******************************/ +#if 0 static void draw_documentation(const SpaceText *st, ARegion *ar) { TextDrawContext tdc = {0}; @@ -1014,6 +1015,7 @@ static void draw_documentation(const SpaceText *st, ARegion *ar) draw_documentation(st, ar); } } +#endif /*********************** draw suggestion list *******************************/ @@ -1487,7 +1489,7 @@ void draw_text_main(SpaceText *st, ARegion *ar) /* draw other stuff */ draw_brackets(st, &tdc, ar); draw_textscroll(st, &scroll, &back); - draw_documentation(st, ar); + /* draw_documentation(st, ar); - No longer supported */ draw_suggestion_list(st, &tdc, ar); text_font_end(&tdc); diff --git a/source/blender/editors/space_view3d/drawarmature.c b/source/blender/editors/space_view3d/drawarmature.c index 149144d969a..314b7bf3335 100644 --- a/source/blender/editors/space_view3d/drawarmature.c +++ b/source/blender/editors/space_view3d/drawarmature.c @@ -1265,7 +1265,7 @@ static void draw_b_bone(const short dt, int armflag, int boneflag, short constfl else { /* wire */ if (armflag & ARM_POSEMODE) { - if (constflag) { + if (constflag && ((G.f & G_PICKSEL) == 0)) { /* set constraint colors */ if (set_pchan_glColor(PCHAN_COLOR_CONSTS, boneflag, constflag)) { glEnable(GL_BLEND); @@ -1406,7 +1406,7 @@ static void draw_bone(const short dt, int armflag, int boneflag, short constflag set_ebone_glColor(boneflag); } else if (armflag & ARM_POSEMODE) { - if (constflag) { + if (constflag && ((G.f & G_PICKSEL) == 0)) { /* draw constraint colors */ if (set_pchan_glColor(PCHAN_COLOR_CONSTS, boneflag, constflag)) { glEnable(GL_BLEND); @@ -2360,7 +2360,6 @@ static void draw_ebones(View3D *v3d, ARegion *ar, Object *ob, const short dt) /* Draw name */ if (arm->flag & ARM_DRAWNAMES) { mid_v3_v3v3(vec, eBone->head, eBone->tail); - glRasterPos3fv(vec); view3d_cached_text_draw_add(vec, eBone->name, strlen(eBone->name), 10, 0, col); } /* Draw additional axes */ diff --git a/source/blender/editors/space_view3d/drawobject.c b/source/blender/editors/space_view3d/drawobject.c index 90d33dc5995..be2e4ab05e0 100644 --- a/source/blender/editors/space_view3d/drawobject.c +++ b/source/blender/editors/space_view3d/drawobject.c @@ -7171,8 +7171,9 @@ static void drawtexspace(Object *ob) } /* draws wire outline */ -static void drawObjectSelect(Scene *scene, View3D *v3d, ARegion *ar, Base *base, - const unsigned char ob_wire_col[4]) +static void draw_object_selected_outline( + Scene *scene, View3D *v3d, ARegion *ar, Base *base, + const unsigned char ob_wire_col[4]) { RegionView3D *rv3d = ar->regiondata; Object *ob = base->object; @@ -7645,7 +7646,7 @@ void draw_object(Scene *scene, ARegion *ar, View3D *v3d, Base *base, const short if ((v3d->flag & V3D_SELECT_OUTLINE) && !render_override && ob->type != OB_MESH) { if (dt > OB_WIRE && (ob->mode & OB_MODE_EDIT) == 0 && (dflag & DRAW_SCENESET) == 0) { if (!(ob->dtx & OB_DRAWWIRE) && (ob->flag & SELECT) && !(dflag & (DRAW_PICKING | DRAW_CONSTCOLOR))) { - drawObjectSelect(scene, v3d, ar, base, ob_wire_col); + draw_object_selected_outline(scene, v3d, ar, base, ob_wire_col); } } } @@ -8174,6 +8175,50 @@ void draw_object(Scene *scene, ARegion *ar, View3D *v3d, Base *base, const short ED_view3d_clear_mats_rv3d(rv3d); } + +/** + * Drawing for selection picking, + * caller must have called 'GPU_select_load_id(base->selcode)' first. + */ +void draw_object_select(Scene *scene, ARegion *ar, View3D *v3d, Base *base, const short dflag) +{ + BLI_assert(dflag & DRAW_PICKING && dflag & DRAW_CONSTCOLOR); + draw_object(scene, ar, v3d, base, dflag); + + /* we draw duplicators for selection too */ + if ((base->object->transflag & OB_DUPLI)) { + ListBase *lb; + DupliObject *dob; + Base tbase; + + tbase.flag = OB_FROMDUPLI; + lb = object_duplilist(G.main->eval_ctx, scene, base->object); + + for (dob = lb->first; dob; dob = dob->next) { + float omat[4][4]; + char dt; + short dtx; + + tbase.object = dob->ob; + copy_m4_m4(omat, dob->ob->obmat); + copy_m4_m4(dob->ob->obmat, dob->mat); + + /* extra service: draw the duplicator in drawtype of parent */ + /* MIN2 for the drawtype to allow bounding box objects in groups for lods */ + dt = tbase.object->dt; tbase.object->dt = MIN2(tbase.object->dt, base->object->dt); + dtx = tbase.object->dtx; tbase.object->dtx = base->object->dtx; + + draw_object(scene, ar, v3d, &tbase, dflag); + + tbase.object->dt = dt; + tbase.object->dtx = dtx; + + copy_m4_m4(dob->ob->obmat, omat); + } + free_object_duplilist(lb); + } +} + /* ***************** BACKBUF SEL (BBS) ********* */ static void bbs_obmode_mesh_verts__mapFunc(void *userData, int index, const float co[3], diff --git a/source/blender/editors/space_view3d/drawvolume.c b/source/blender/editors/space_view3d/drawvolume.c index 27ecbf83db5..c076bfb4aa4 100644 --- a/source/blender/editors/space_view3d/drawvolume.c +++ b/source/blender/editors/space_view3d/drawvolume.c @@ -737,7 +737,7 @@ static void add_streamline(float (*verts)[3], float(*colors)[3], float center[3] copy_v3_v3(verts[(*offset)++], center); } -typedef void (*vector_draw_func)(float(*)[3], float(*)[3], float*, float*, float, float, int*); +typedef void (*vector_draw_func)(float(*)[3], float(*)[3], float *, float *, float, float, int *); #endif /* WITH_SMOKE */ void draw_smoke_velocity(SmokeDomainSettings *domain, float viewnormal[3]) @@ -774,8 +774,8 @@ void draw_smoke_velocity(SmokeDomainSettings *domain, float viewnormal[3]) float min[3] = { domain->p0[0] - domain->cell_size[0] * domain->adapt_res, - domain->p0[1] - domain->cell_size[1] * domain->adapt_res, - domain->p0[2] - domain->cell_size[2] * domain->adapt_res, + domain->p0[1] - domain->cell_size[1] * domain->adapt_res, + domain->p0[2] - domain->cell_size[2] * domain->adapt_res, }; int num_points_v[3] = { diff --git a/source/blender/editors/space_view3d/space_view3d.c b/source/blender/editors/space_view3d/space_view3d.c index 075b1faf502..b8228c63209 100644 --- a/source/blender/editors/space_view3d/space_view3d.c +++ b/source/blender/editors/space_view3d/space_view3d.c @@ -180,8 +180,8 @@ bool ED_view3d_context_user_region(bContext *C, View3D **r_v3d, ARegion **r_ar) View3D *v3d = (View3D *)sa->spacedata.first; if (ar) { - RegionView3D *rv3d = ar->regiondata; - if (rv3d && (rv3d->viewlock & RV3D_LOCKED) == 0) { + RegionView3D *rv3d; + if ((ar->regiontype == RGN_TYPE_WINDOW) && (rv3d = ar->regiondata) && (rv3d->viewlock & RV3D_LOCKED) == 0) { *r_v3d = v3d; *r_ar = ar; return true; @@ -869,6 +869,7 @@ static void view3d_main_region_listener(bScreen *sc, ScrArea *sa, ARegion *ar, w case ND_CONSTRAINT: case ND_KEYS: case ND_PARTICLE: + case ND_POINTCACHE: case ND_LOD: ED_region_tag_redraw(ar); break; @@ -1256,21 +1257,6 @@ static void space_view3d_listener(bScreen *UNUSED(sc), ScrArea *sa, struct wmNot } break; } - - /* removed since BKE_image_user_frame_calc is now called in view3d_draw_bgpic because screen_ops doesnt call the notifier. */ -#if 0 - if (wmn->category == NC_SCENE && wmn->data == ND_FRAME) { - View3D *v3d = area->spacedata.first; - BGpic *bgpic = v3d->bgpicbase.first; - - for (; bgpic; bgpic = bgpic->next) { - if (bgpic->ima) { - Scene *scene = wmn->reference; - BKE_image_user_frame_calc(&bgpic->iuser, scene->r.cfra, 0); - } - } - } -#endif } const char *view3d_context_dir[] = { diff --git a/source/blender/editors/space_view3d/view3d_buttons.c b/source/blender/editors/space_view3d/view3d_buttons.c index 351c7ccec15..34e01405e7e 100644 --- a/source/blender/editors/space_view3d/view3d_buttons.c +++ b/source/blender/editors/space_view3d/view3d_buttons.c @@ -814,10 +814,6 @@ static void view3d_panel_vgroup(const bContext *C, Panel *pa) if (dv && dv->totweight) { ToolSettings *ts = scene->toolsettings; - wmOperatorType *ot_weight_set_active = WM_operatortype_find("OBJECT_OT_vertex_weight_set_active", true); - wmOperatorType *ot_weight_paste = WM_operatortype_find("OBJECT_OT_vertex_weight_paste", true); - wmOperatorType *ot_weight_delete = WM_operatortype_find("OBJECT_OT_vertex_weight_delete", true); - wmOperatorType *ot; PointerRNA op_ptr, tools_ptr; PointerRNA *but_ptr; @@ -856,7 +852,7 @@ static void view3d_panel_vgroup(const bContext *C, Panel *pa) /* The Weight Group Name */ - ot = ot_weight_set_active; + ot = WM_operatortype_find("OBJECT_OT_vertex_weight_set_active", true); but = uiDefButO_ptr(block, UI_BTYPE_BUT, ot, WM_OP_EXEC_DEFAULT, dg->name, xco, yco, (x = UI_UNIT_X * 5), UI_UNIT_Y, ""); but_ptr = UI_but_operator_ptr_get(but); @@ -882,23 +878,16 @@ static void view3d_panel_vgroup(const bContext *C, Panel *pa) xco += x; /* The weight group paste function */ - - ot = ot_weight_paste; - WM_operator_properties_create_ptr(&op_ptr, ot); - RNA_int_set(&op_ptr, "weight_group", i); icon = (locked) ? ICON_BLANK1 : ICON_PASTEDOWN; - uiItemFullO_ptr(row, ot, "", icon, op_ptr.data, WM_OP_INVOKE_DEFAULT, 0); + op_ptr = uiItemFullO(row, "OBJECT_OT_vertex_weight_paste", "", icon, NULL, WM_OP_INVOKE_DEFAULT, UI_ITEM_O_RETURN_PROPS); + RNA_int_set(&op_ptr, "weight_group", i); /* The weight entry delete function */ - - ot = ot_weight_delete; - WM_operator_properties_create_ptr(&op_ptr, ot); - RNA_int_set(&op_ptr, "weight_group", i); icon = (locked) ? ICON_LOCKED : ICON_X; - uiItemFullO_ptr(row, ot, "", icon, op_ptr.data, WM_OP_INVOKE_DEFAULT, 0); + op_ptr = uiItemFullO(row, "OBJECT_OT_vertex_weight_delete", "", icon, NULL, WM_OP_INVOKE_DEFAULT, UI_ITEM_O_RETURN_PROPS); + RNA_int_set(&op_ptr, "weight_group", i); yco -= UI_UNIT_Y; - } } } diff --git a/source/blender/editors/space_view3d/view3d_draw.c b/source/blender/editors/space_view3d/view3d_draw.c index f23e587e55d..3fb2761d40e 100644 --- a/source/blender/editors/space_view3d/view3d_draw.c +++ b/source/blender/editors/space_view3d/view3d_draw.c @@ -100,6 +100,7 @@ #include "GPU_material.h" #include "GPU_compositing.h" #include "GPU_extensions.h" +#include "GPU_select.h" #include "view3d_intern.h" /* own include */ @@ -2023,6 +2024,35 @@ static void view3d_draw_xraytransp(Scene *scene, ARegion *ar, View3D *v3d, const glDepthMask(GL_TRUE); } +/* clears zbuffer and draws it over, + * note that in the select version we don't care about transparent flag as with regular drawing */ +static void view3d_draw_xray_select(Scene *scene, ARegion *ar, View3D *v3d, bool *clear) +{ + /* Not ideal, but we need to read from the previous depths before clearing + * otherwise we could have a function to load the depths after drawing. + * + * Clearing the depth buffer isn't all that common between drawing objects so accept this for now. + */ + if (U.gpu_select_pick_deph) { + GPU_select_load_id(-1); + } + + View3DAfter *v3da; + if (*clear && v3d->zbuf) { + glClear(GL_DEPTH_BUFFER_BIT); + *clear = false; + } + + v3d->xray = true; + while ((v3da = BLI_pophead(&v3d->afterdraw_xray))) { + if (GPU_select_load_id(v3da->base->selcol)) { + draw_object_select(scene, ar, v3d, v3da->base, v3da->dflag); + } + MEM_freeN(v3da); + } + v3d->xray = false; +} + /* *********************** */ /* @@ -2487,6 +2517,58 @@ void ED_view3d_draw_depth(Scene *scene, ARegion *ar, View3D *v3d, bool alphaover U.obcenter_dia = obcenter_dia; } +void ED_view3d_draw_select_loop( + ViewContext *vc, Scene *scene, View3D *v3d, ARegion *ar, + bool use_obedit_skip, bool use_nearest) +{ + short code = 1; + const short dflag = DRAW_PICKING | DRAW_CONSTCOLOR; + + if (vc->obedit && vc->obedit->type == OB_MBALL) { + draw_object(scene, ar, v3d, BASACT, dflag); + } + else if ((vc->obedit && vc->obedit->type == OB_ARMATURE)) { + /* if not drawing sketch, draw bones */ + if (!BDR_drawSketchNames(vc)) { + draw_object(scene, ar, v3d, BASACT, dflag); + } + } + else { + Base *base; + + for (base = scene->base.first; base; base = base->next) { + if (base->lay & v3d->lay) { + + if ((base->object->restrictflag & OB_RESTRICT_SELECT) || + (use_obedit_skip && (scene->obedit->data == base->object->data))) + { + base->selcol = 0; + } + else { + base->selcol = code; + + if (use_nearest && (base->object->dtx & OB_DRAWXRAY)) { + ED_view3d_after_add(&v3d->afterdraw_xray, base, dflag); + } + else { + if (GPU_select_load_id(code)) { + draw_object_select(scene, ar, v3d, base, dflag); + } + } + code++; + } + } + } + + if (use_nearest) { + bool xrayclear = true; + if (v3d->afterdraw_xray.first) { + view3d_draw_xray_select(scene, ar, v3d, &xrayclear); + } + } + } +} + typedef struct View3DShadow { struct View3DShadow *next, *prev; GPULamp *lamp; @@ -2955,7 +3037,7 @@ struct RV3DMatrixStore { float pixsize; }; -void *ED_view3d_mats_rv3d_backup(struct RegionView3D *rv3d) +struct RV3DMatrixStore *ED_view3d_mats_rv3d_backup(struct RegionView3D *rv3d) { struct RV3DMatrixStore *rv3dmat = MEM_mallocN(sizeof(*rv3dmat), __func__); copy_m4_m4(rv3dmat->winmat, rv3d->winmat); @@ -2968,9 +3050,8 @@ void *ED_view3d_mats_rv3d_backup(struct RegionView3D *rv3d) return (void *)rv3dmat; } -void ED_view3d_mats_rv3d_restore(struct RegionView3D *rv3d, void *rv3dmat_pt) +void ED_view3d_mats_rv3d_restore(struct RegionView3D *rv3d, struct RV3DMatrixStore *rv3dmat) { - struct RV3DMatrixStore *rv3dmat = rv3dmat_pt; copy_m4_m4(rv3d->winmat, rv3dmat->winmat); copy_m4_m4(rv3d->viewmat, rv3dmat->viewmat); copy_m4_m4(rv3d->persmat, rv3dmat->persmat); diff --git a/source/blender/editors/space_view3d/view3d_edit.c b/source/blender/editors/space_view3d/view3d_edit.c index 2b53eb71d99..f07727f8118 100644 --- a/source/blender/editors/space_view3d/view3d_edit.c +++ b/source/blender/editors/space_view3d/view3d_edit.c @@ -90,19 +90,6 @@ bool ED_view3d_offset_lock_check(const View3D *v3d, const RegionView3D *rv3d) return (rv3d->persp != RV3D_CAMOB) && (v3d->ob_centre_cursor || v3d->ob_centre); } -static bool view3d_operator_offset_lock_check(bContext *C, wmOperator *op) -{ - View3D *v3d = CTX_wm_view3d(C); - RegionView3D *rv3d = CTX_wm_region_view3d(C); - if (ED_view3d_offset_lock_check(v3d, rv3d)) { - BKE_report(op->reports, RPT_WARNING, "View offset is locked"); - return true; - } - else { - return false; - } -} - /* ********************** view3d_edit: view manipulations ********************* */ /** @@ -2596,6 +2583,19 @@ void VIEW3D_OT_zoom(wmOperatorType *ot) /* ************************ viewdolly ******************************** */ +static bool viewdolly_offset_lock_check(bContext *C, wmOperator *op) +{ + View3D *v3d = CTX_wm_view3d(C); + RegionView3D *rv3d = CTX_wm_region_view3d(C); + if (ED_view3d_offset_lock_check(v3d, rv3d)) { + BKE_report(op->reports, RPT_WARNING, "Cannot dolly when the view offset is locked"); + return true; + } + else { + return false; + } +} + static void view_dolly_mouseloc(ARegion *ar, float orig_ofs[3], float dvec[3], float dfac) { RegionView3D *rv3d = ar->regiondata; @@ -2746,7 +2746,7 @@ static int viewdolly_invoke(bContext *C, wmOperator *op, const wmEvent *event) { ViewOpsData *vod; - if (view3d_operator_offset_lock_check(C, op)) + if (viewdolly_offset_lock_check(C, op)) return OPERATOR_CANCELLED; /* makes op->customdata */ @@ -4364,41 +4364,24 @@ static EnumPropertyItem prop_view_pan_items[] = { {0, NULL, 0, NULL, NULL} }; -static int viewpan_exec(bContext *C, wmOperator *op) +static int viewpan_invoke(bContext *C, wmOperator *op, const wmEvent *event) { - ScrArea *sa = CTX_wm_area(C); - ARegion *ar = CTX_wm_region(C); - View3D *v3d = CTX_wm_view3d(C); - RegionView3D *rv3d = CTX_wm_region_view3d(C); - float vec[3]; - const float co_zero[3] = {0.0f}; - float mval_f[2] = {0.0f, 0.0f}; - float zfac; - int pandir; + int x = 0, y = 0; + int pandir = RNA_enum_get(op->ptr, "type"); - if (view3d_operator_offset_lock_check(C, op)) - return OPERATOR_CANCELLED; + if (pandir == V3D_VIEW_PANRIGHT) { x = -32; } + else if (pandir == V3D_VIEW_PANLEFT) { x = 32; } + else if (pandir == V3D_VIEW_PANUP) { y = -25; } + else if (pandir == V3D_VIEW_PANDOWN) { y = 25; } - pandir = RNA_enum_get(op->ptr, "type"); - - ED_view3d_camera_lock_init(v3d, rv3d); - - zfac = ED_view3d_calc_zfac(rv3d, co_zero, NULL); - if (pandir == V3D_VIEW_PANRIGHT) { mval_f[0] = -32.0f; } - else if (pandir == V3D_VIEW_PANLEFT) { mval_f[0] = 32.0f; } - else if (pandir == V3D_VIEW_PANUP) { mval_f[1] = -25.0f; } - else if (pandir == V3D_VIEW_PANDOWN) { mval_f[1] = 25.0f; } - ED_view3d_win_to_delta(ar, mval_f, vec, zfac); - add_v3_v3(rv3d->ofs, vec); - - if (rv3d->viewlock & RV3D_BOXVIEW) - view3d_boxview_sync(sa, ar); - - ED_view3d_depth_tag_update(rv3d); + viewops_data_alloc(C, op); + viewops_data_create(C, op, event); + ViewOpsData *vod = op->customdata; - ED_view3d_camera_lock_sync(v3d, rv3d); + viewmove_apply(vod, vod->oldx + x, vod->oldy + y); - ED_region_tag_redraw(ar); + ED_view3d_depth_tag_update(vod->rv3d); + viewops_data_free(C, op); return OPERATOR_FINISHED; } @@ -4411,7 +4394,7 @@ void VIEW3D_OT_view_pan(wmOperatorType *ot) ot->idname = "VIEW3D_OT_view_pan"; /* api callbacks */ - ot->exec = viewpan_exec; + ot->invoke = viewpan_invoke; ot->poll = ED_operator_region_view3d_active; /* flags */ @@ -4798,6 +4781,7 @@ static int manipulator_invoke(bContext *C, wmOperator *op, const wmEvent *event) void VIEW3D_OT_manipulator(wmOperatorType *ot) { + PropertyRNA *prop; /* identifiers */ ot->name = "3D Manipulator"; @@ -4812,8 +4796,9 @@ void VIEW3D_OT_manipulator(wmOperatorType *ot) /* properties to pass to transform */ Transform_Properties(ot, P_CONSTRAINT); - RNA_def_boolean(ot->srna, "use_planar_constraint", false, "Planar Constraint", "Limit the transformation to the " - "two axes that have not been clicked (translate/scale only)"); + prop = RNA_def_boolean(ot->srna, "use_planar_constraint", false, "Planar Constraint", "Limit the transformation to the " + "two axes that have not been clicked (translate/scale only)"); + RNA_def_property_flag(prop, PROP_SKIP_SAVE | PROP_HIDDEN); } static int enable_manipulator_invoke(bContext *C, wmOperator *op, const wmEvent *UNUSED(event)) @@ -4902,11 +4887,7 @@ static float view_autodist_depth_margin(ARegion *ar, const int mval[2], int marg rect.ymax = mval[1] + 1; } else { - rect.xmax = mval[0] + margin; - rect.ymax = mval[1] + margin; - - rect.xmin = mval[0] - margin; - rect.ymin = mval[1] - margin; + BLI_rcti_init_pt_radius(&rect, mval, margin); } view3d_update_depths_rect(ar, &depth_temp, &rect); diff --git a/source/blender/editors/space_view3d/view3d_intern.h b/source/blender/editors/space_view3d/view3d_intern.h index b11f42bcfef..c2b8d1f8bda 100644 --- a/source/blender/editors/space_view3d/view3d_intern.h +++ b/source/blender/editors/space_view3d/view3d_intern.h @@ -143,6 +143,8 @@ void draw_motion_paths_cleanup(View3D *v3d); /* drawobject.c */ void draw_object(Scene *scene, struct ARegion *ar, View3D *v3d, Base *base, const short dflag); +void draw_object_select(Scene *scene, ARegion *ar, View3D *v3d, Base *base, const short dflag); + bool draw_glsl_material(Scene *scene, struct Object *ob, View3D *v3d, const char dt); void draw_object_instance(Scene *scene, View3D *v3d, RegionView3D *rv3d, struct Object *ob, const char dt, int outline); void draw_object_backbufsel(Scene *scene, View3D *v3d, RegionView3D *rv3d, struct Object *ob); @@ -195,7 +197,11 @@ void draw_sim_debug_data(Scene *scene, View3D *v3d, ARegion *ar); void view3d_main_region_draw(const struct bContext *C, struct ARegion *ar); void ED_view3d_draw_depth(Scene *scene, struct ARegion *ar, View3D *v3d, bool alphaoverride); void ED_view3d_draw_depth_gpencil(Scene *scene, ARegion *ar, View3D *v3d); -void ED_view3d_after_add(ListBase *lb, Base *base, const short dflag); +void ED_view3d_draw_select_loop( + ViewContext *vc, Scene *scene, View3D *v3d, ARegion *ar, + bool use_obedit_skip, bool use_nearest); + +void ED_view3d_after_add(ListBase *lb, Base *base, const short dflag);\ void circf(float x, float y, float rad); void circ(float x, float y, float rad); @@ -241,7 +247,7 @@ void ED_view3d_smooth_view_force_finish( struct bContext *C, struct View3D *v3d, struct ARegion *ar); -void view3d_winmatrix_set(ARegion *ar, const View3D *v3d, const rctf *rect); +void view3d_winmatrix_set(ARegion *ar, const View3D *v3d, const rcti *rect); void view3d_viewmatrix_set(Scene *scene, const View3D *v3d, RegionView3D *rv3d); void fly_modal_keymap(struct wmKeyConfig *keyconf); diff --git a/source/blender/editors/space_view3d/view3d_ops.c b/source/blender/editors/space_view3d/view3d_ops.c index 0fa6841fe27..d71639c35d2 100644 --- a/source/blender/editors/space_view3d/view3d_ops.c +++ b/source/blender/editors/space_view3d/view3d_ops.c @@ -241,13 +241,21 @@ void view3d_keymap(wmKeyConfig *keyconf) keymap = WM_keymap_find(keyconf, "3D View", SPACE_VIEW3D, 0); /* Shift+LMB behavior first, so it has priority over KM_ANY item below. */ - kmi = WM_keymap_verify_item(keymap, "VIEW3D_OT_manipulator", LEFTMOUSE, KM_PRESS, KM_SHIFT, 0); + kmi = WM_keymap_add_item(keymap, "VIEW3D_OT_manipulator", LEFTMOUSE, KM_PRESS, KM_SHIFT, 0); RNA_boolean_set(kmi->ptr, "release_confirm", true); RNA_boolean_set(kmi->ptr, "use_planar_constraint", true); + RNA_boolean_set(kmi->ptr, "use_accurate", false); + + kmi = WM_keymap_add_item(keymap, "VIEW3D_OT_manipulator", LEFTMOUSE, KM_PRESS, KM_SHIFT, 0); + RNA_boolean_set(kmi->ptr, "release_confirm", true); + RNA_boolean_set(kmi->ptr, "use_planar_constraint", false); + RNA_boolean_set(kmi->ptr, "use_accurate", true); + /* Using KM_ANY here to allow holding modifiers before starting to transform. */ kmi = WM_keymap_add_item(keymap, "VIEW3D_OT_manipulator", LEFTMOUSE, KM_PRESS, KM_ANY, 0); RNA_boolean_set(kmi->ptr, "release_confirm", true); RNA_boolean_set(kmi->ptr, "use_planar_constraint", false); + RNA_boolean_set(kmi->ptr, "use_accurate", false); WM_keymap_verify_item(keymap, "VIEW3D_OT_cursor3d", ACTIONMOUSE, KM_PRESS, 0, 0); diff --git a/source/blender/editors/space_view3d/view3d_select.c b/source/blender/editors/space_view3d/view3d_select.c index 3239d07553f..7d927766bbd 100644 --- a/source/blender/editors/space_view3d/view3d_select.c +++ b/source/blender/editors/space_view3d/view3d_select.c @@ -98,6 +98,8 @@ #include "view3d_intern.h" /* own include */ +// #include "PIL_time_utildefines.h" + float ED_view3d_select_dist_px(void) { return 75.0f * U.pixelsize; @@ -1087,7 +1089,9 @@ static void deselectall_except(Scene *scene, Base *b) /* deselect all except b } } -static Base *object_mouse_select_menu(bContext *C, ViewContext *vc, unsigned int *buffer, int hits, const int mval[2], short toggle) +static Base *object_mouse_select_menu( + bContext *C, ViewContext *vc, unsigned int *buffer, int hits, + const int mval[2], bool toggle) { short baseCount = 0; bool ok; @@ -1178,19 +1182,19 @@ static bool selectbuffer_has_bones(const unsigned int *buffer, const unsigned in } /* utility function for mixed_bones_object_selectbuffer */ -static short selectbuffer_ret_hits_15(unsigned int *UNUSED(buffer), const short hits15) +static int selectbuffer_ret_hits_15(unsigned int *UNUSED(buffer), const int hits15) { return hits15; } -static short selectbuffer_ret_hits_9(unsigned int *buffer, const short hits15, const short hits9) +static int selectbuffer_ret_hits_9(unsigned int *buffer, const int hits15, const int hits9) { const int offs = 4 * hits15; memcpy(buffer, buffer + offs, 4 * hits9 * sizeof(unsigned int)); return hits9; } -static short selectbuffer_ret_hits_5(unsigned int *buffer, const short hits15, const short hits9, const short hits5) +static int selectbuffer_ret_hits_5(unsigned int *buffer, const int hits15, const int hits9, const int hits5) { const int offs = 4 * hits15 + 4 * hits9; memcpy(buffer, buffer + offs, 4 * hits5 * sizeof(unsigned int)); @@ -1199,14 +1203,13 @@ static short selectbuffer_ret_hits_5(unsigned int *buffer, const short hits15, c /* we want a select buffer with bones, if there are... */ /* so check three selection levels and compare */ -static short mixed_bones_object_selectbuffer( +static int mixed_bones_object_selectbuffer( ViewContext *vc, unsigned int *buffer, const int mval[2], bool use_cycle, bool enumerate, bool *r_do_nearest) { rcti rect; - int offs; - short hits15, hits9 = 0, hits5 = 0; + int hits15, hits9 = 0, hits5 = 0; bool has_bones15 = false, has_bones9 = false, has_bones5 = false; static int last_mval[2] = {-100, -100}; bool do_nearest = false; @@ -1234,44 +1237,57 @@ static short mixed_bones_object_selectbuffer( do_nearest = do_nearest && !enumerate; - BLI_rcti_init(&rect, mval[0] - 14, mval[0] + 14, mval[1] - 14, mval[1] + 14); - hits15 = view3d_opengl_select(vc, buffer, MAXPICKBUF, &rect, do_nearest); + const int select_mode = (do_nearest ? VIEW3D_SELECT_PICK_NEAREST : VIEW3D_SELECT_PICK_ALL); + int hits = 0; + + /* we _must_ end cache before return, use 'goto finally' */ + view3d_opengl_select_cache_begin(); + + BLI_rcti_init_pt_radius(&rect, mval, 14); + hits15 = view3d_opengl_select(vc, buffer, MAXPICKBUF, &rect, select_mode); if (hits15 == 1) { - return selectbuffer_ret_hits_15(buffer, hits15); + hits = selectbuffer_ret_hits_15(buffer, hits15); + goto finally; } else if (hits15 > 0) { + int offs; has_bones15 = selectbuffer_has_bones(buffer, hits15); offs = 4 * hits15; - BLI_rcti_init(&rect, mval[0] - 9, mval[0] + 9, mval[1] - 9, mval[1] + 9); - hits9 = view3d_opengl_select(vc, buffer + offs, MAXPICKBUF - offs, &rect, do_nearest); + BLI_rcti_init_pt_radius(&rect, mval, 9); + hits9 = view3d_opengl_select(vc, buffer + offs, MAXPICKBUF - offs, &rect, select_mode); if (hits9 == 1) { - return selectbuffer_ret_hits_9(buffer, hits15, hits9); + hits = selectbuffer_ret_hits_9(buffer, hits15, hits9); + goto finally; } else if (hits9 > 0) { has_bones9 = selectbuffer_has_bones(buffer + offs, hits9); offs += 4 * hits9; - BLI_rcti_init(&rect, mval[0] - 5, mval[0] + 5, mval[1] - 5, mval[1] + 5); - hits5 = view3d_opengl_select(vc, buffer + offs, MAXPICKBUF - offs, &rect, do_nearest); + BLI_rcti_init_pt_radius(&rect, mval, 5); + hits5 = view3d_opengl_select(vc, buffer + offs, MAXPICKBUF - offs, &rect, select_mode); if (hits5 == 1) { - return selectbuffer_ret_hits_5(buffer, hits15, hits9, hits5); + hits = selectbuffer_ret_hits_5(buffer, hits15, hits9, hits5); + goto finally; } else if (hits5 > 0) { has_bones5 = selectbuffer_has_bones(buffer + offs, hits5); } } - if (has_bones5) return selectbuffer_ret_hits_5(buffer, hits15, hits9, hits5); - else if (has_bones9) return selectbuffer_ret_hits_9(buffer, hits15, hits9); - else if (has_bones15) return selectbuffer_ret_hits_15(buffer, hits15); - - if (hits5 > 0) return selectbuffer_ret_hits_5(buffer, hits15, hits9, hits5); - else if (hits9 > 0) return selectbuffer_ret_hits_9(buffer, hits15, hits9); - else return selectbuffer_ret_hits_15(buffer, hits15); + if (has_bones5) { hits = selectbuffer_ret_hits_5(buffer, hits15, hits9, hits5); goto finally; } + else if (has_bones9) { hits = selectbuffer_ret_hits_9(buffer, hits15, hits9); goto finally; } + else if (has_bones15) { hits = selectbuffer_ret_hits_15(buffer, hits15); goto finally; } + + if (hits5 > 0) { hits = selectbuffer_ret_hits_5(buffer, hits15, hits9, hits5); goto finally; } + else if (hits9 > 0) { hits = selectbuffer_ret_hits_9(buffer, hits15, hits9); goto finally; } + else { hits = selectbuffer_ret_hits_15(buffer, hits15); goto finally; } } - - return 0; + +finally: + view3d_opengl_select_cache_end(); + + return hits; } /* returns basact */ @@ -1412,7 +1428,7 @@ static bool ed_object_select_pick( bool is_obedit; float dist = ED_view3d_select_dist_px() * 1.3333f; bool retval = false; - short hits; + int hits; const float mval_fl[2] = {(float)mval[0], (float)mval[1]}; @@ -1464,10 +1480,13 @@ static bool ed_object_select_pick( unsigned int buffer[MAXPICKBUF]; bool do_nearest; + // TIMEIT_START(select_time); + /* if objects have posemode set, the bones are in the same selection buffer */ - hits = mixed_bones_object_selectbuffer(&vc, buffer, mval, true, enumerate, &do_nearest); - + + // TIMEIT_END(select_time); + if (hits > 0) { /* note: bundles are handling in the same way as bones */ const bool has_bones = selectbuffer_has_bones(buffer, hits); @@ -1904,9 +1923,9 @@ static int do_meta_box_select(ViewContext *vc, rcti *rect, bool select, bool ext int a; unsigned int buffer[MAXPICKBUF]; - short hits; + int hits; - hits = view3d_opengl_select(vc, buffer, MAXPICKBUF, rect, false); + hits = view3d_opengl_select(vc, buffer, MAXPICKBUF, rect, VIEW3D_SELECT_ALL); if (extend == false && select) BKE_mball_deselect_all(mb); @@ -1938,9 +1957,9 @@ static int do_armature_box_select(ViewContext *vc, rcti *rect, bool select, bool int a; unsigned int buffer[MAXPICKBUF]; - short hits; + int hits; - hits = view3d_opengl_select(vc, buffer, MAXPICKBUF, rect, false); + hits = view3d_opengl_select(vc, buffer, MAXPICKBUF, rect, VIEW3D_SELECT_ALL); /* clear flag we use to detect point was affected */ for (ebone = arm->edbo->first; ebone; ebone = ebone->next) @@ -2013,7 +2032,7 @@ static int do_object_pose_box_select(bContext *C, ViewContext *vc, rcti *rect, b int bone_only; int bone_selected = 0; int totobj = MAXPICKBUF; /* XXX solve later */ - short hits; + int hits; if ((ob) && (ob->mode & OB_MODE_POSE)) bone_only = 1; @@ -2037,7 +2056,7 @@ static int do_object_pose_box_select(bContext *C, ViewContext *vc, rcti *rect, b /* selection buffer now has bones potentially too, so we add MAXPICKBUF */ vbuffer = MEM_mallocN(4 * (totobj + MAXPICKELEMS) * sizeof(unsigned int), "selection buffer"); - hits = view3d_opengl_select(vc, vbuffer, 4 * (totobj + MAXPICKELEMS), rect, false); + hits = view3d_opengl_select(vc, vbuffer, 4 * (totobj + MAXPICKELEMS), rect, VIEW3D_SELECT_ALL); /* * LOGIC NOTES (theeth): * The buffer and ListBase have the same relative order, which makes the selection @@ -2577,7 +2596,7 @@ static void lattice_circle_select(ViewContext *vc, const bool select, const int /* NOTE: pose-bone case is copied from editbone case... */ -static short pchan_circle_doSelectJoint(void *userData, bPoseChannel *pchan, const float screen_co[2]) +static bool pchan_circle_doSelectJoint(void *userData, bPoseChannel *pchan, const float screen_co[2]) { CircleSelectUserData *data = userData; @@ -2655,7 +2674,7 @@ static void pose_circle_select(ViewContext *vc, const bool select, const int mva } } -static short armature_circle_doSelectJoint(void *userData, EditBone *ebone, const float screen_co[2], short head) +static bool armature_circle_doSelectJoint(void *userData, EditBone *ebone, const float screen_co[2], bool head) { CircleSelectUserData *data = userData; diff --git a/source/blender/editors/space_view3d/view3d_view.c b/source/blender/editors/space_view3d/view3d_view.c index 8582952d1a0..fe8274064e5 100644 --- a/source/blender/editors/space_view3d/view3d_view.c +++ b/source/blender/editors/space_view3d/view3d_view.c @@ -908,7 +908,7 @@ void ED_view3d_polygon_offset(const RegionView3D *rv3d, const float dist) /** * \param rect optional for picking (can be NULL). */ -void view3d_winmatrix_set(ARegion *ar, const View3D *v3d, const rctf *rect) +void view3d_winmatrix_set(ARegion *ar, const View3D *v3d, const rcti *rect) { RegionView3D *rv3d = ar->regiondata; rctf viewplane; @@ -1091,76 +1091,19 @@ void view3d_viewmatrix_set(Scene *scene, const View3D *v3d, RegionView3D *rv3d) } } -static void view3d_select_loop(ViewContext *vc, Scene *scene, View3D *v3d, ARegion *ar, bool use_obedit_skip) +/** + * Optionally cache data for multiple calls to #view3d_opengl_select + * + * just avoid GPU_select headers outside this file + */ +void view3d_opengl_select_cache_begin(void) { - short code = 1; - char dt; - short dtx; - - if (vc->obedit && vc->obedit->type == OB_MBALL) { - draw_object(scene, ar, v3d, BASACT, DRAW_PICKING | DRAW_CONSTCOLOR); - } - else if ((vc->obedit && vc->obedit->type == OB_ARMATURE)) { - /* if not drawing sketch, draw bones */ - if (!BDR_drawSketchNames(vc)) { - draw_object(scene, ar, v3d, BASACT, DRAW_PICKING | DRAW_CONSTCOLOR); - } - } - else { - Base *base; - - v3d->xray = true; /* otherwise it postpones drawing */ - for (base = scene->base.first; base; base = base->next) { - if (base->lay & v3d->lay) { - - if ((base->object->restrictflag & OB_RESTRICT_SELECT) || - (use_obedit_skip && (scene->obedit->data == base->object->data))) - { - base->selcol = 0; - } - else { - base->selcol = code; - - if (GPU_select_load_id(code)) { - draw_object(scene, ar, v3d, base, DRAW_PICKING | DRAW_CONSTCOLOR); - - /* we draw duplicators for selection too */ - if ((base->object->transflag & OB_DUPLI)) { - ListBase *lb; - DupliObject *dob; - Base tbase; - - tbase.flag = OB_FROMDUPLI; - lb = object_duplilist(G.main->eval_ctx, scene, base->object); - - for (dob = lb->first; dob; dob = dob->next) { - float omat[4][4]; - - tbase.object = dob->ob; - copy_m4_m4(omat, dob->ob->obmat); - copy_m4_m4(dob->ob->obmat, dob->mat); - - /* extra service: draw the duplicator in drawtype of parent */ - /* MIN2 for the drawtype to allow bounding box objects in groups for lods */ - dt = tbase.object->dt; tbase.object->dt = MIN2(tbase.object->dt, base->object->dt); - dtx = tbase.object->dtx; tbase.object->dtx = base->object->dtx; - - draw_object(scene, ar, v3d, &tbase, DRAW_PICKING | DRAW_CONSTCOLOR); - - tbase.object->dt = dt; - tbase.object->dtx = dtx; + GPU_select_cache_begin(); +} - copy_m4_m4(dob->ob->obmat, omat); - } - free_object_duplilist(lb); - } - } - code++; - } - } - } - v3d->xray = false; /* restore */ - } +void view3d_opengl_select_cache_end(void) +{ + GPU_select_cache_end(); } /** @@ -1170,29 +1113,65 @@ static void view3d_select_loop(ViewContext *vc, Scene *scene, View3D *v3d, ARegi * * \note (vc->obedit == NULL) can be set to explicitly skip edit-object selection. */ -short view3d_opengl_select(ViewContext *vc, unsigned int *buffer, unsigned int bufsize, const rcti *input, bool do_nearest) +int view3d_opengl_select( + ViewContext *vc, unsigned int *buffer, unsigned int bufsize, const rcti *input, + eV3DSelectMode select_mode) { Scene *scene = vc->scene; View3D *v3d = vc->v3d; ARegion *ar = vc->ar; - rctf rect; - short hits; + rcti rect; + int hits; const bool use_obedit_skip = (scene->obedit != NULL) && (vc->obedit == NULL); - const bool do_passes = do_nearest && GPU_select_query_check_active(); + const bool is_pick_select = (U.gpu_select_pick_deph != 0); + const bool do_passes = ( + (is_pick_select == false) && + (select_mode == VIEW3D_SELECT_PICK_NEAREST) && + GPU_select_query_check_active()); + const bool use_nearest = (is_pick_select && select_mode == VIEW3D_SELECT_PICK_NEAREST); + + char gpu_select_mode; - G.f |= G_PICKSEL; - /* case not a border select */ if (input->xmin == input->xmax) { - rect.xmin = input->xmin - 12; /* seems to be default value for bones only now */ - rect.xmax = input->xmin + 12; - rect.ymin = input->ymin - 12; - rect.ymax = input->ymin + 12; + /* seems to be default value for bones only now */ + BLI_rcti_init_pt_radius(&rect, (const int[2]){input->xmin, input->ymin}, 12); } else { - BLI_rctf_rcti_copy(&rect, input); + rect = *input; } - + + if (is_pick_select) { + if (is_pick_select && select_mode == VIEW3D_SELECT_PICK_NEAREST) { + gpu_select_mode = GPU_SELECT_PICK_NEAREST; + } + else if (is_pick_select && select_mode == VIEW3D_SELECT_PICK_ALL) { + gpu_select_mode = GPU_SELECT_PICK_ALL; + } + else { + gpu_select_mode = GPU_SELECT_ALL; + } + } + else { + if (do_passes) { + gpu_select_mode = GPU_SELECT_NEAREST_FIRST_PASS; + } + else { + gpu_select_mode = GPU_SELECT_ALL; + } + } + + /* Re-use cache (rect must be smaller then the cached) + * other context is assumed to be unchanged */ + if (GPU_select_is_cached()) { + GPU_select_begin(buffer, bufsize, &rect, gpu_select_mode, 0); + GPU_select_cache_load_id(); + hits = GPU_select_end(); + goto finally; + } + + G.f |= G_PICKSEL; + view3d_winmatrix_set(ar, v3d, &rect); mul_m4_m4m4(vc->rv3d->persmat, vc->rv3d->winmat, vc->rv3d->viewmat); @@ -1204,12 +1183,9 @@ short view3d_opengl_select(ViewContext *vc, unsigned int *buffer, unsigned int b if (vc->rv3d->rflag & RV3D_CLIPPING) ED_view3d_clipping_set(vc->rv3d); - if (do_passes) - GPU_select_begin(buffer, bufsize, &rect, GPU_SELECT_NEAREST_FIRST_PASS, 0); - else - GPU_select_begin(buffer, bufsize, &rect, GPU_SELECT_ALL, 0); + GPU_select_begin(buffer, bufsize, &rect, gpu_select_mode, 0); - view3d_select_loop(vc, scene, v3d, ar, use_obedit_skip); + ED_view3d_draw_select_loop(vc, scene, v3d, ar, use_obedit_skip, use_nearest); hits = GPU_select_end(); @@ -1217,7 +1193,7 @@ short view3d_opengl_select(ViewContext *vc, unsigned int *buffer, unsigned int b if (do_passes) { GPU_select_begin(buffer, bufsize, &rect, GPU_SELECT_NEAREST_SECOND_PASS, hits); - view3d_select_loop(vc, scene, v3d, ar, use_obedit_skip); + ED_view3d_draw_select_loop(vc, scene, v3d, ar, use_obedit_skip, use_nearest); GPU_select_end(); } @@ -1233,7 +1209,8 @@ short view3d_opengl_select(ViewContext *vc, unsigned int *buffer, unsigned int b if (vc->rv3d->rflag & RV3D_CLIPPING) ED_view3d_clipping_disable(); - + +finally: if (hits < 0) printf("Too many objects in select buffer\n"); /* XXX make error message */ return hits; @@ -1428,6 +1405,8 @@ static bool view3d_localview_init( } } + DAG_on_visible_update(bmain, false); + return ok; } diff --git a/source/blender/editors/transform/transform.c b/source/blender/editors/transform/transform.c index 1916f9b4dab..7d9063c3285 100644 --- a/source/blender/editors/transform/transform.c +++ b/source/blender/editors/transform/transform.c @@ -2176,7 +2176,14 @@ bool initTransform(bContext *C, TransInfo *t, wmOperator *op, const wmEvent *eve calculateCenter(t); if (event) { - initMouseInput(t, &t->mouse, t->center2d, event->mval, event->shift); + /* Initialize accurate transform to settings requested by keymap. */ + bool use_accurate = false; + if ((prop = RNA_struct_find_property(op->ptr, "use_accurate")) && RNA_property_is_set(op->ptr, prop)) { + if (RNA_property_boolean_get(op->ptr, prop)) { + use_accurate = true; + } + } + initMouseInput(t, &t->mouse, t->center2d, event->mval, use_accurate); } switch (mode) { diff --git a/source/blender/editors/transform/transform_conversions.c b/source/blender/editors/transform/transform_conversions.c index 091a5773cf0..5e67f304755 100644 --- a/source/blender/editors/transform/transform_conversions.c +++ b/source/blender/editors/transform/transform_conversions.c @@ -1973,9 +1973,12 @@ void flushTransParticles(TransInfo *t) /* ********************* mesh ****************** */ -static bool bmesh_test_dist_add(BMVert *v, BMVert *v_other, - float *dists, const float *dists_prev, - float mtx[3][3]) +static bool bmesh_test_dist_add( + BMVert *v, BMVert *v_other, + float *dists, const float *dists_prev, + /* optionally track original index */ + int *index, const int *index_prev, + float mtx[3][3]) { if ((BM_elem_flag_test(v_other, BM_ELEM_SELECT) == 0) && (BM_elem_flag_test(v_other, BM_ELEM_HIDDEN) == 0)) @@ -1990,6 +1993,9 @@ static bool bmesh_test_dist_add(BMVert *v, BMVert *v_other, dist_other = dists_prev[i] + len_v3(vec); if (dist_other < dists[i_other]) { dists[i_other] = dist_other; + if (index != NULL) { + index[i_other] = index_prev[i]; + } return true; } } @@ -1997,11 +2003,13 @@ static bool bmesh_test_dist_add(BMVert *v, BMVert *v_other, return false; } -static void editmesh_set_connectivity_distance(BMesh *bm, float mtx[3][3], float *dists) +/** + * \parm mtx: Measure disatnce in this space. + * \parm dists: Store the closest connected distance to selected vertices. + * \parm index: Optionally store the original index we're measuring the distance to (can be NULL). + */ +static void editmesh_set_connectivity_distance(BMesh *bm, float mtx[3][3], float *dists, int *index) { - /* need to be very careful of feedback loops here, store previous dist's to avoid feedback */ - float *dists_prev = MEM_mallocN(bm->totvert * sizeof(float), __func__); - BLI_LINKSTACK_DECLARE(queue, BMVert *); /* any BM_ELEM_TAG'd vertex is in 'queue_next', so we don't add in twice */ @@ -2022,17 +2030,27 @@ static void editmesh_set_connectivity_distance(BMesh *bm, float mtx[3][3], float if (BM_elem_flag_test(v, BM_ELEM_SELECT) == 0 || BM_elem_flag_test(v, BM_ELEM_HIDDEN)) { dist = FLT_MAX; + if (index != NULL) { + index[i] = i; + } } else { BLI_LINKSTACK_PUSH(queue, v); dist = 0.0f; + if (index != NULL) { + index[i] = i; + } } - dists[i] = dists_prev[i] = dist; + dists[i] = dist; } bm->elem_index_dirty &= ~BM_VERT; } + /* need to be very careful of feedback loops here, store previous dist's to avoid feedback */ + float *dists_prev = MEM_dupallocN(dists); + int *index_prev = MEM_dupallocN(index); /* may be NULL */ + do { BMVert *v; LinkNode *lnk; @@ -2061,7 +2079,7 @@ static void editmesh_set_connectivity_distance(BMesh *bm, float mtx[3][3], float /* edge distance */ { BMVert *v_other = BM_edge_other_vert(e_iter, v); - if (bmesh_test_dist_add(v, v_other, dists, dists_prev, mtx)) { + if (bmesh_test_dist_add(v, v_other, dists, dists_prev, index, index_prev, mtx)) { if (BM_elem_flag_test(v_other, BM_ELEM_TAG) == 0) { BM_elem_flag_enable(v_other, BM_ELEM_TAG); BLI_LINKSTACK_PUSH(queue_next, v_other); @@ -2086,7 +2104,7 @@ static void editmesh_set_connectivity_distance(BMesh *bm, float mtx[3][3], float (BM_elem_flag_test(l_iter_radial->f, BM_ELEM_HIDDEN) == 0)) { BMVert *v_other = l_iter_radial->next->next->v; - if (bmesh_test_dist_add(v, v_other, dists, dists_prev, mtx)) { + if (bmesh_test_dist_add(v, v_other, dists, dists_prev, index, index_prev, mtx)) { if (BM_elem_flag_test(v_other, BM_ELEM_TAG) == 0) { BM_elem_flag_enable(v_other, BM_ELEM_TAG); BLI_LINKSTACK_PUSH(queue_next, v_other); @@ -2110,6 +2128,9 @@ static void editmesh_set_connectivity_distance(BMesh *bm, float mtx[3][3], float /* keep in sync, avoid having to do full memcpy each iteration */ dists_prev[i] = dists[i]; + if (index != NULL) { + index_prev[i] = index[i]; + } } BLI_LINKSTACK_SWAP(queue, queue_next); @@ -2123,9 +2144,14 @@ static void editmesh_set_connectivity_distance(BMesh *bm, float mtx[3][3], float BLI_LINKSTACK_FREE(queue_next); MEM_freeN(dists_prev); + if (index_prev != NULL) { + MEM_freeN(index_prev); + } } -static struct TransIslandData *editmesh_islands_info_calc(BMEditMesh *em, int *r_island_tot, int **r_island_vert_map) +static struct TransIslandData *editmesh_islands_info_calc( + BMEditMesh *em, int *r_island_tot, int **r_island_vert_map, + bool calc_single_islands) { BMesh *bm = em->bm; struct TransIslandData *trans_islands; @@ -2237,6 +2263,42 @@ static struct TransIslandData *editmesh_islands_info_calc(BMEditMesh *em, int *r MEM_freeN(groups_array); MEM_freeN(group_index); + /* for PET we need islands of 1 so connected vertices can use it with V3D_AROUND_LOCAL_ORIGINS */ + if (calc_single_islands) { + BMIter viter; + BMVert *v; + int group_tot_single = 0; + + BM_ITER_MESH_INDEX (v, &viter, bm, BM_VERTS_OF_MESH, i) { + if (BM_elem_flag_test(v, BM_ELEM_SELECT) && (vert_map[i] == -1)) { + group_tot_single += 1; + } + } + + if (group_tot_single != 0) { + trans_islands = MEM_reallocN(trans_islands, group_tot + group_tot_single); + + BM_ITER_MESH_INDEX (v, &viter, bm, BM_VERTS_OF_MESH, i) { + if (BM_elem_flag_test(v, BM_ELEM_SELECT) && (vert_map[i] == -1)) { + struct TransIslandData *v_island = &trans_islands[group_tot]; + vert_map[i] = group_tot; + + copy_v3_v3(v_island->co, v->co); + + if (is_zero_v3(v->no) != 0.0f) { + axis_dominant_v3_to_m3(v_island->axismtx, v->no); + invert_m3(v_island->axismtx); + } + else { + unit_m3(v_island->axismtx); + } + + group_tot += 1; + } + } + } + } + *r_island_tot = group_tot; *r_island_vert_map = vert_map; @@ -2336,6 +2398,11 @@ static void createTransEditVerts(TransInfo *t) int island_info_tot; int *island_vert_map = NULL; + const bool is_island_center = (t->around == V3D_AROUND_LOCAL_ORIGINS) && (t->mode != TFM_TRANSLATION); + /* Original index of our connected vertex when connected distances are calculated. + * Optional, allocate if needed. */ + int *dists_index = NULL; + if (t->flag & T_MIRROR) { EDBM_verts_mirror_cache_begin(em, 0, false, (t->flag & T_PROP_EDIT) == 0, use_topology); mirror = 1; @@ -2367,8 +2434,12 @@ static void createTransEditVerts(TransInfo *t) t->total = count; /* allocating scratch arrays */ - if (prop_mode & T_PROP_CONNECTED) - dists = MEM_mallocN(em->bm->totvert * sizeof(float), "scratch nears"); + if (prop_mode & T_PROP_CONNECTED) { + dists = MEM_mallocN(em->bm->totvert * sizeof(float), __func__); + if (is_island_center) { + dists_index = MEM_mallocN(em->bm->totvert * sizeof(int), __func__); + } + } } else { t->total = bm->totvertsel; @@ -2390,7 +2461,7 @@ static void createTransEditVerts(TransInfo *t) pseudoinverse_m3_m3(smtx, mtx, PSEUDOINVERSE_EPSILON); if (prop_mode & T_PROP_CONNECTED) { - editmesh_set_connectivity_distance(em->bm, mtx, dists); + editmesh_set_connectivity_distance(em->bm, mtx, dists, dists_index); } /* Only in case of rotation and resize, we want the elements of the edited @@ -2398,8 +2469,14 @@ static void createTransEditVerts(TransInfo *t) * * TODO: use island_info to detect the closest point when the "Snap Target" * in Blender UI is "Closest" */ - if ((t->around == V3D_AROUND_LOCAL_ORIGINS) && (t->mode != TFM_TRANSLATION)) { - island_info = editmesh_islands_info_calc(em, &island_info_tot, &island_vert_map); + if (is_island_center) { + /* In this specific case, near-by vertices will need to know the island of the nearest connected vertex. */ + const bool calc_single_islands = ( + (prop_mode & T_PROP_CONNECTED) && + (t->around == V3D_AROUND_LOCAL_ORIGINS) && + (em->selectmode & SCE_SELECT_VERTEX)); + + island_info = editmesh_islands_info_calc(em, &island_info_tot, &island_vert_map, calc_single_islands); } /* detect CrazySpace [tm] */ @@ -2449,10 +2526,16 @@ static void createTransEditVerts(TransInfo *t) BM_ITER_MESH_INDEX (eve, &iter, bm, BM_VERTS_OF_MESH, a) { if (!BM_elem_flag_test(eve, BM_ELEM_HIDDEN)) { if (prop_mode || BM_elem_flag_test(eve, BM_ELEM_SELECT)) { - struct TransIslandData *v_island = (island_info && island_vert_map[a] != -1) ? - &island_info[island_vert_map[a]] : NULL; + struct TransIslandData *v_island = NULL; float *bweight = (cd_vert_bweight_offset != -1) ? BM_ELEM_CD_GET_VOID_P(eve, cd_vert_bweight_offset) : NULL; + if (island_info) { + const int connected_index = (dists_index && dists_index[a] != -1) ? dists_index[a] : a; + v_island = (island_vert_map[connected_index] != -1) ? + &island_info[island_vert_map[connected_index]] : NULL; + } + + VertsToTransData(t, tob, tx, em, eve, bweight, v_island); if (tx) tx++; @@ -2531,6 +2614,8 @@ cleanup: MEM_freeN(defmats); if (dists) MEM_freeN(dists); + if (dists_index) + MEM_freeN(dists_index); if (t->flag & T_MIRROR) { EDBM_verts_mirror_cache_end(em); @@ -8038,7 +8123,12 @@ void createTransData(bContext *C, TransInfo *t) if (t->data && t->flag & T_PROP_EDIT) { if (ELEM(t->obedit->type, OB_CURVE, OB_MESH)) { sort_trans_data(t); // makes selected become first in array - set_prop_dist(t, 0); + if ((t->obedit->type == OB_MESH) && (t->flag & T_PROP_CONNECTED)) { + /* already calculated by editmesh_set_connectivity_distance */ + } + else { + set_prop_dist(t, 0); + } sort_trans_data_dist(t); } else { diff --git a/source/blender/editors/transform/transform_manipulator.c b/source/blender/editors/transform/transform_manipulator.c index e141724f2df..0a984d90ae3 100644 --- a/source/blender/editors/transform/transform_manipulator.c +++ b/source/blender/editors/transform/transform_manipulator.c @@ -1724,14 +1724,14 @@ static int manipulator_selectbuf(ScrArea *sa, ARegion *ar, const int mval[2], fl { View3D *v3d = sa->spacedata.first; RegionView3D *rv3d = ar->regiondata; - rctf rect, selrect; + rcti rect; GLuint buffer[64]; // max 4 items per select, so large enuf short hits; const bool is_picksel = true; const bool do_passes = GPU_select_query_check_active(); /* XXX check a bit later on this... (ton) */ - extern void view3d_winmatrix_set(ARegion *ar, View3D *v3d, rctf *rect); + extern void view3d_winmatrix_set(ARegion *ar, View3D *v3d, const rcti *rect); /* when looking through a selected camera, the manipulator can be at the * exact same position as the view, skip so we don't break selection */ @@ -1743,15 +1743,13 @@ static int manipulator_selectbuf(ScrArea *sa, ARegion *ar, const int mval[2], fl rect.ymin = mval[1] - hotspot; rect.ymax = mval[1] + hotspot; - selrect = rect; - view3d_winmatrix_set(ar, v3d, &rect); mul_m4_m4m4(rv3d->persmat, rv3d->winmat, rv3d->viewmat); if (do_passes) - GPU_select_begin(buffer, 64, &selrect, GPU_SELECT_NEAREST_FIRST_PASS, 0); + GPU_select_begin(buffer, 64, &rect, GPU_SELECT_NEAREST_FIRST_PASS, 0); else - GPU_select_begin(buffer, 64, &selrect, GPU_SELECT_ALL, 0); + GPU_select_begin(buffer, 64, &rect, GPU_SELECT_ALL, 0); /* do the drawing */ if (v3d->twtype & V3D_MANIP_ROTATE) { @@ -1766,7 +1764,7 @@ static int manipulator_selectbuf(ScrArea *sa, ARegion *ar, const int mval[2], fl hits = GPU_select_end(); if (do_passes) { - GPU_select_begin(buffer, 64, &selrect, GPU_SELECT_NEAREST_SECOND_PASS, hits); + GPU_select_begin(buffer, 64, &rect, GPU_SELECT_NEAREST_SECOND_PASS, hits); /* do the drawing */ if (v3d->twtype & V3D_MANIP_ROTATE) { @@ -1826,6 +1824,23 @@ static int manipulator_selectbuf(ScrArea *sa, ARegion *ar, const int mval[2], fl return 0; } +static const char *manipulator_get_operator_name(int man_val) +{ + if (man_val & MAN_TRANS_C) { + return "TRANSFORM_OT_translate"; + } + else if (man_val == MAN_ROT_T) { + return "TRANSFORM_OT_trackball"; + } + else if (man_val & MAN_ROT_C) { + return "TRANSFORM_OT_rotate"; + } + else if (man_val & MAN_SCALE_C) { + return "TRANSFORM_OT_resize"; + } + + return NULL; +} /* return 0; nothing happened */ int BIF_do_manipulator(bContext *C, const struct wmEvent *event, wmOperator *op) @@ -1846,11 +1861,24 @@ int BIF_do_manipulator(bContext *C, const struct wmEvent *event, wmOperator *op) // find the hotspots first test narrow hotspot val = manipulator_selectbuf(sa, ar, event->mval, 0.5f * (float)U.tw_hotspot); if (val) { + wmOperatorType *ot; + PointerRNA props_ptr; + PropertyRNA *prop; + const char *opname; // drawflags still global, for drawing call above drawflags = manipulator_selectbuf(sa, ar, event->mval, 0.2f * (float)U.tw_hotspot); if (drawflags == 0) drawflags = val; + /* Planar constraint doesn't make sense for rotation, give other keymaps a chance */ + if ((drawflags & MAN_ROT_C) && use_planar) { + return 0; + } + + opname = manipulator_get_operator_name(drawflags); + ot = WM_operatortype_find(opname, true); + WM_operator_properties_create_ptr(&props_ptr, ot); + if (drawflags & MAN_TRANS_C) { switch (drawflags) { case MAN_TRANS_C: @@ -1880,8 +1908,7 @@ int BIF_do_manipulator(bContext *C, const struct wmEvent *event, wmOperator *op) constraint_axis[2] = 1; break; } - RNA_boolean_set_array(op->ptr, "constraint_axis", constraint_axis); - WM_operator_name_call(C, "TRANSFORM_OT_translate", WM_OP_INVOKE_DEFAULT, op->ptr); + RNA_boolean_set_array(&props_ptr, "constraint_axis", constraint_axis); } else if (drawflags & MAN_SCALE_C) { switch (drawflags) { @@ -1910,22 +1937,10 @@ int BIF_do_manipulator(bContext *C, const struct wmEvent *event, wmOperator *op) constraint_axis[2] = 1; break; } - RNA_boolean_set_array(op->ptr, "constraint_axis", constraint_axis); - WM_operator_name_call(C, "TRANSFORM_OT_resize", WM_OP_INVOKE_DEFAULT, op->ptr); + RNA_boolean_set_array(&props_ptr, "constraint_axis", constraint_axis); } - else if (drawflags == MAN_ROT_T) { /* trackball need special case, init is different */ - /* Do not pass op->ptr!!! trackball has no "constraint" properties! - * See [#34621], it's a miracle it did not cause more problems!!! */ - /* However, we need to copy the "release_confirm" property, but only if defined, see T41112. */ - PointerRNA props_ptr; - PropertyRNA *prop; - wmOperatorType *ot = WM_operatortype_find("TRANSFORM_OT_trackball", true); - WM_operator_properties_create_ptr(&props_ptr, ot); - if ((prop = RNA_struct_find_property(op->ptr, "release_confirm")) && RNA_property_is_set(op->ptr, prop)) { - RNA_property_boolean_set(&props_ptr, prop, RNA_property_boolean_get(op->ptr, prop)); - } - WM_operator_name_call_ptr(C, ot, WM_OP_INVOKE_DEFAULT, &props_ptr); - WM_operator_properties_free(&props_ptr); + else if (drawflags == MAN_ROT_T) { + /* pass */ } else if (drawflags & MAN_ROT_C) { switch (drawflags) { @@ -1939,9 +1954,25 @@ int BIF_do_manipulator(bContext *C, const struct wmEvent *event, wmOperator *op) constraint_axis[2] = 1; break; } - RNA_boolean_set_array(op->ptr, "constraint_axis", constraint_axis); - WM_operator_name_call(C, "TRANSFORM_OT_rotate", WM_OP_INVOKE_DEFAULT, op->ptr); + RNA_boolean_set_array(&props_ptr, "constraint_axis", constraint_axis); + } + + /* pass operator properties on to transform operators */ + prop = RNA_struct_find_property(op->ptr, "use_accurate"); + if (RNA_property_is_set(op->ptr, prop)) { + RNA_property_boolean_set(&props_ptr, prop, RNA_property_boolean_get(op->ptr, prop)); + } + prop = RNA_struct_find_property(op->ptr, "release_confirm"); + if (RNA_property_is_set(op->ptr, prop)) { + RNA_property_boolean_set(&props_ptr, prop, RNA_property_boolean_get(op->ptr, prop)); } + prop = RNA_struct_find_property(op->ptr, "constraint_orientation"); + if (RNA_property_is_set(op->ptr, prop)) { + RNA_property_enum_set(&props_ptr, prop, RNA_property_enum_get(op->ptr, prop)); + } + + WM_operator_name_call_ptr(C, ot, WM_OP_INVOKE_DEFAULT, &props_ptr); + WM_operator_properties_free(&props_ptr); } /* after transform, restore drawflags */ drawflags = 0xFFFF; diff --git a/source/blender/editors/transform/transform_ops.c b/source/blender/editors/transform/transform_ops.c index cbe58ddf586..2a97384cf7d 100644 --- a/source/blender/editors/transform/transform_ops.c +++ b/source/blender/editors/transform/transform_ops.c @@ -569,6 +569,9 @@ void Transform_Properties(struct wmOperatorType *ot, int flags) // Add confirm method all the time. At the end because it's not really that important and should be hidden only in log, not in keymap edit /*prop =*/ RNA_def_boolean(ot->srna, "release_confirm", 0, "Confirm on Release", "Always confirm operation when releasing button"); //RNA_def_property_flag(prop, PROP_HIDDEN); + + prop = RNA_def_boolean(ot->srna, "use_accurate", 0, "Accurate", "Use accurate transformation"); + RNA_def_property_flag(prop, PROP_HIDDEN); } } diff --git a/source/blender/editors/transform/transform_snap_object.c b/source/blender/editors/transform/transform_snap_object.c index 7c9dc43dbe4..b33528b4149 100644 --- a/source/blender/editors/transform/transform_snap_object.c +++ b/source/blender/editors/transform/transform_snap_object.c @@ -87,6 +87,8 @@ typedef struct SnapObjectData { typedef struct SnapObjectData_Mesh { SnapObjectData sd; BVHTreeFromMesh *bvh_trees[3]; + MPoly *mpoly; + bool poly_allocated; } SnapObjectData_Mesh; @@ -934,9 +936,10 @@ static bool snapEmpty( float tmp_co[3]; copy_v3_v3(tmp_co, obmat[3]); if (test_projected_vert_dist( - snapdata->depth_range, snapdata->mval, tmp_co, - snapdata->pmat, snapdata->win_half, is_persp, &dist_px_sq, - r_loc)) { + snapdata->depth_range, snapdata->mval, tmp_co, + snapdata->pmat, snapdata->win_half, is_persp, &dist_px_sq, + r_loc)) + { *dist_px = sqrtf(dist_px_sq); *ray_depth = depth_get(r_loc, snapdata->ray_start, snapdata->ray_dir); retval = true; @@ -1051,7 +1054,6 @@ static int dm_looptri_to_poly_index(DerivedMesh *dm, const MLoopTri *lt) static bool snapDerivedMesh( SnapObjectContext *sctx, SnapData *snapdata, Object *ob, DerivedMesh *dm, float obmat[4][4], const unsigned int ob_index, - bool do_bb, /* read/write args */ float *ray_depth, float *dist_px, /* return args */ @@ -1112,39 +1114,30 @@ static bool snapDerivedMesh( copy_v3_v3(ray_org_local, snapdata->ray_origin); mul_m4_v3(imat, ray_org_local); - if (do_bb) { - BoundBox *bb = BKE_object_boundbox_get(ob); - - if (bb) { - BoundBox bb_temp; - - /* We cannot afford a bounding box with some null dimension, which may happen in some cases... - * Threshold is rather high, but seems to be needed to get good behavior, see T46099. */ - bb = BKE_boundbox_ensure_minimum_dimensions(bb, &bb_temp, 1e-1f); - - /* In vertex and edges you need to get the pixel distance from ray to BoundBox, see T46816. */ - if (ELEM(snapdata->snap_to, SCE_SNAP_MODE_VERTEX, SCE_SNAP_MODE_EDGE)) { - float dist_px_sq = dist_squared_to_projected_aabb_simple( - lpmat, snapdata->win_half, ray_min_dist, snapdata->mval, - ray_org_local, ray_normal_local, bb->vec[0], bb->vec[6]); - if (dist_px_sq > SQUARE(*dist_px)) - { - return retval; - } + /* Test BoundBox */ + BoundBox *bb = BKE_object_boundbox_get(ob); + if (bb) { + /* In vertex and edges you need to get the pixel distance from ray to BoundBox, see: T46099, T46816 */ + if (ELEM(snapdata->snap_to, SCE_SNAP_MODE_VERTEX, SCE_SNAP_MODE_EDGE)) { + float dist_px_sq = dist_squared_to_projected_aabb_simple( + lpmat, snapdata->win_half, ray_min_dist, snapdata->mval, + ray_org_local, ray_normal_local, bb->vec[0], bb->vec[6]); + if (dist_px_sq > SQUARE(*dist_px)) { + return retval; } - else { - /* was BKE_boundbox_ray_hit_check, see: cf6ca226fa58 */ - if (!isect_ray_aabb_v3_simple( - ray_start_local, ray_normal_local, bb->vec[0], bb->vec[6], NULL, NULL)) - { - return retval; - } + } + else { + /* was BKE_boundbox_ray_hit_check, see: cf6ca226fa58 */ + if (!isect_ray_aabb_v3_simple( + ray_start_local, ray_normal_local, bb->vec[0], bb->vec[6], NULL, NULL)) + { + return retval; } - /* was local_depth, see: T47838 */ - len_diff = dist_aabb_to_plane(bb->vec[0], bb->vec[6], ray_start_local, ray_normal_local); - if (len_diff < 0) len_diff = 0.0f; - need_ray_start_correction_init = false; } + /* was local_depth, see: T47838 */ + len_diff = dist_aabb_to_plane(bb->vec[0], bb->vec[6], ray_start_local, ray_normal_local); + if (len_diff < 0) len_diff = 0.0f; + need_ray_start_correction_init = false; } SnapObjectData_Mesh *sod = NULL; @@ -1182,6 +1175,29 @@ static bool snapDerivedMesh( if (treedata->cached && !bvhcache_has_tree(dm->bvhCache, treedata->tree)) { free_bvhtree_from_mesh(treedata); } + else { + if (!treedata->vert_allocated) { + treedata->vert = DM_get_vert_array(dm, &treedata->vert_allocated); + } + if ((tree_index == 1) && !treedata->edge_allocated) { + treedata->edge = DM_get_edge_array(dm, &treedata->vert_allocated); + } + if (tree_index == 2) { + if (!treedata->loop_allocated) { + treedata->loop = DM_get_loop_array(dm, &treedata->loop_allocated); + } + if (!treedata->looptri_allocated) { + if (!sod->poly_allocated) { + sod->mpoly = DM_get_poly_array(dm, &sod->poly_allocated); + } + treedata->looptri = DM_get_looptri_array( + dm, treedata->vert, + sod->mpoly, dm->getNumPolys(dm), + treedata->loop, dm->getNumLoops(dm), + &treedata->looptri_allocated); + } + } + } } } @@ -1295,10 +1311,17 @@ static bool snapDerivedMesh( } /* SCE_SNAP_MODE_VERTEX or SCE_SNAP_MODE_EDGE */ else { + + /* Warning: the depth_max is currently being used only in perspective view. + * It is not correct to limit the maximum depth for elements obtained with nearest + * since this limitation depends on the normal and the size of the occlusion face. + * And more... ray_depth is being confused with Z-depth here... (varies only the precision) */ + const float ray_depth_max_global = *ray_depth + snapdata->depth_range[0]; + Nearest2dUserData neasrest2d = { .dist_px_sq = SQUARE(*dist_px), .r_axis_closest = {1.0f, 1.0f, 1.0f}, - .depth_range = {snapdata->depth_range[0], *ray_depth + snapdata->depth_range[0]}, + .depth_range = {snapdata->depth_range[0], ray_depth_max_global}, .userdata = treedata, .get_edge_verts = (Nearest2DGetEdgeVertsCallback)get_dm_edge_verts, .copy_vert_no = (Nearest2DCopyVertNoCallback)copy_dm_vert_no, @@ -1650,7 +1673,6 @@ static bool snapObject( } retval = snapDerivedMesh( sctx, snapdata, ob, dm, obmat, ob_index, - true, ray_depth, dist_px, r_loc, r_no, r_index, r_hit_list); @@ -1858,6 +1880,9 @@ static void snap_object_data_free(void *sod_v) free_bvhtree_from_mesh(sod->bvh_trees[i]); } } + if (sod->poly_allocated) { + MEM_freeN(sod->mpoly); + } break; } case SNAP_EDIT_MESH: @@ -2053,10 +2078,10 @@ static bool transform_snap_context_project_view3d_mixed_impl( for (int i = 0; i < 3; i++) { if (snap_to_flag & (1 << i)) { if (ED_transform_snap_object_project_view3d( - sctx, - elem_type[i], params, - mval, dist_px, &ray_depth, - r_co, r_no)) + sctx, + elem_type[i], params, + mval, dist_px, &ray_depth, + r_co, r_no)) { is_hit = true; break; diff --git a/source/blender/editors/util/numinput.c b/source/blender/editors/util/numinput.c index e07831358d6..3727be1842c 100644 --- a/source/blender/editors/util/numinput.c +++ b/source/blender/editors/util/numinput.c @@ -497,7 +497,7 @@ bool handleNumInput(bContext *C, NumInput *n, const wmEvent *event) n->unit_sys, n->unit_type[idx]); /* Note: with angles, we always get values as radians here... */ - if (BPY_execute_string_as_number(C, str_unit_convert, &val, false)) { + if (BPY_execute_string_as_number(C, str_unit_convert, false, &val)) { n->val[idx] = (float)val; n->val_flag[idx] &= ~NUM_INVALID; } diff --git a/source/blender/editors/util/undo.c b/source/blender/editors/util/undo.c index fab5b7e821f..4e021d4833e 100644 --- a/source/blender/editors/util/undo.c +++ b/source/blender/editors/util/undo.c @@ -419,6 +419,9 @@ int ED_undo_operator_repeat(bContext *C, struct wmOperator *op) if (G.debug & G_DEBUG) printf("redo_cb: operator redo %s\n", op->type->name); + + WM_operator_free_all_after(wm, op); + ED_undo_pop_op(C, op); if (op->type->check) { diff --git a/source/blender/gpu/CMakeLists.txt b/source/blender/gpu/CMakeLists.txt index 8885209ce01..885ff2ff159 100644 --- a/source/blender/gpu/CMakeLists.txt +++ b/source/blender/gpu/CMakeLists.txt @@ -57,6 +57,8 @@ set(SRC intern/gpu_init_exit.c intern/gpu_material.c intern/gpu_select.c + intern/gpu_select_pick.c + intern/gpu_select_sample_query.c intern/gpu_shader.c intern/gpu_texture.c @@ -97,6 +99,7 @@ set(SRC GPU_texture.h intern/gpu_codegen.h intern/gpu_private.h + intern/gpu_select_private.h ) data_to_c_simple(shaders/gpu_shader_geometry.glsl SRC) diff --git a/source/blender/gpu/GPU_select.h b/source/blender/gpu/GPU_select.h index 6a16b5b7456..cf5b8bf7d8f 100644 --- a/source/blender/gpu/GPU_select.h +++ b/source/blender/gpu/GPU_select.h @@ -30,19 +30,30 @@ #ifndef __GPU_SELECT_H__ #define __GPU_SELECT_H__ -#include "DNA_vec_types.h" /* rcft */ #include "BLI_sys_types.h" +struct rcti; + /* flags for mode of operation */ enum { GPU_SELECT_ALL = 1, + /* gpu_select_query */ GPU_SELECT_NEAREST_FIRST_PASS = 2, GPU_SELECT_NEAREST_SECOND_PASS = 3, + /* gpu_select_pick */ + GPU_SELECT_PICK_ALL = 4, + GPU_SELECT_PICK_NEAREST = 5, }; -void GPU_select_begin(unsigned int *buffer, unsigned int bufsize, rctf *input, char mode, int oldhits); +void GPU_select_begin(unsigned int *buffer, unsigned int bufsize, const struct rcti *input, char mode, int oldhits); bool GPU_select_load_id(unsigned int id); unsigned int GPU_select_end(void); bool GPU_select_query_check_active(void); +/* cache selection region */ +bool GPU_select_is_cached(void); +void GPU_select_cache_begin(void); +void GPU_select_cache_load_id(void); +void GPU_select_cache_end(void); + #endif diff --git a/source/blender/gpu/intern/gpu_codegen.c b/source/blender/gpu/intern/gpu_codegen.c index 211394e7932..c3896fbd659 100644 --- a/source/blender/gpu/intern/gpu_codegen.c +++ b/source/blender/gpu/intern/gpu_codegen.c @@ -360,7 +360,7 @@ static void codegen_print_datatype(DynStr *ds, const GPUType type, float *data) BLI_dynstr_appendf(ds, "%s(", GPU_DATATYPE_STR[type]); for (i = 0; i < type; i++) { - BLI_dynstr_appendf(ds, "%f", data[i]); + BLI_dynstr_appendf(ds, "%.12f", data[i]); if (i == type - 1) BLI_dynstr_append(ds, ")"); else diff --git a/source/blender/gpu/intern/gpu_debug.c b/source/blender/gpu/intern/gpu_debug.c index d632e767ca9..ba68d1a6a0f 100644 --- a/source/blender/gpu/intern/gpu_debug.c +++ b/source/blender/gpu/intern/gpu_debug.c @@ -161,7 +161,7 @@ const char *gpuErrorString(GLenum err) #endif -static const char* source_name(GLenum source) +static const char *source_name(GLenum source) { switch (source) { case GL_DEBUG_SOURCE_API: return "API"; @@ -174,7 +174,7 @@ static const char* source_name(GLenum source) } } -static const char* message_type_name(GLenum message) +static const char *message_type_name(GLenum message) { switch (message) { case GL_DEBUG_TYPE_ERROR: return "error"; @@ -188,7 +188,7 @@ static const char* message_type_name(GLenum message) } } -static const char* category_name_amd(GLenum category) +static const char *category_name_amd(GLenum category) { switch (category) { case GL_DEBUG_CATEGORY_API_ERROR_AMD: return "API error"; diff --git a/source/blender/gpu/intern/gpu_select.c b/source/blender/gpu/intern/gpu_select.c index 58582232cd5..9496ff137dc 100644 --- a/source/blender/gpu/intern/gpu_select.c +++ b/source/blender/gpu/intern/gpu_select.c @@ -29,109 +29,86 @@ * Interface for accessing gpu-related methods for selection. The semantics will be * similar to glRenderMode(GL_SELECT) since the goal is to maintain compatibility. */ +#include <stdlib.h> + #include "GPU_select.h" #include "GPU_extensions.h" #include "GPU_glew.h" - + #include "MEM_guardedalloc.h" #include "DNA_userdef_types.h" #include "BLI_utildefines.h" -/* Ad hoc number of queries to allocate to skip doing many glGenQueries */ -#define ALLOC_QUERIES 200 - -typedef struct GPUQueryState { +#include "gpu_select_private.h" + +/* Internal algorithm used */ +enum { + /** GL_SELECT, legacy OpenGL selection */ + ALGO_GL_LEGACY = 1, + /** glBegin/EndQuery(GL_SAMPLES_PASSED... ), `gpu_select_query.c` + * Only sets 4th component (ID) correctly. */ + ALGO_GL_QUERY = 2, + /** Read depth buffer for every drawing pass and extract depths, `gpu_select_pick.c` + * Only sets 4th component (ID) correctly. */ + ALGO_GL_PICK = 3, +}; + +typedef struct GPUSelectState { /* To ignore selection id calls when not initialized */ bool select_is_active; - /* Tracks whether a query has been issued so that gpu_load_id can end the previous one */ - bool query_issued; - /* array holding the OpenGL query identifiers */ - unsigned int *queries; - /* array holding the id corresponding to each query */ - unsigned int *id; - /* number of queries in *queries and *id */ - unsigned int num_of_queries; - /* index to the next query to start */ - unsigned int active_query; /* flag to cache user preference for occlusion based selection */ bool use_gpu_select; - /* cache on initialization */ - unsigned int *buffer; - /* buffer size (stores number of integers, for actual size multiply by sizeof integer)*/ - unsigned int bufsize; /* mode of operation */ char mode; - unsigned int index; - int oldhits; -} GPUQueryState; + /* internal algorithm for selection */ + char algorithm; + /* allow GPU_select_begin/end without drawing */ + bool use_cache; +} GPUSelectState; -static GPUQueryState g_query_state = {0}; +static GPUSelectState g_select_state = {0}; /** * initialize and provide buffer for results */ -void GPU_select_begin(unsigned int *buffer, unsigned int bufsize, rctf *input, char mode, int oldhits) +void GPU_select_begin(unsigned int *buffer, unsigned int bufsize, const rcti *input, char mode, int oldhits) { - g_query_state.select_is_active = true; - g_query_state.query_issued = false; - g_query_state.active_query = 0; - g_query_state.use_gpu_select = GPU_select_query_check_active(); - g_query_state.num_of_queries = 0; - g_query_state.bufsize = bufsize; - g_query_state.buffer = buffer; - g_query_state.mode = mode; - g_query_state.index = 0; - g_query_state.oldhits = oldhits; + g_select_state.select_is_active = true; + g_select_state.use_gpu_select = GPU_select_query_check_active(); + g_select_state.mode = mode; - if (!g_query_state.use_gpu_select) { - glSelectBuffer(bufsize, (GLuint *)buffer); - glRenderMode(GL_SELECT); - glInitNames(); - glPushName(-1); + if (ELEM(g_select_state.mode, GPU_SELECT_PICK_ALL, GPU_SELECT_PICK_NEAREST)) { + g_select_state.algorithm = ALGO_GL_PICK; + } + else if (!g_select_state.use_gpu_select) { + g_select_state.algorithm = ALGO_GL_LEGACY; } else { - float viewport[4]; - - g_query_state.num_of_queries = ALLOC_QUERIES; - - g_query_state.queries = MEM_mallocN(g_query_state.num_of_queries * sizeof(*g_query_state.queries), "gpu selection queries"); - g_query_state.id = MEM_mallocN(g_query_state.num_of_queries * sizeof(*g_query_state.id), "gpu selection ids"); - glGenQueries(g_query_state.num_of_queries, g_query_state.queries); - - glPushAttrib(GL_DEPTH_BUFFER_BIT | GL_VIEWPORT_BIT); - /* disable writing to the framebuffer */ - glColorMask(GL_FALSE, GL_FALSE, GL_FALSE, GL_FALSE); - - /* In order to save some fill rate we minimize the viewport using rect. - * We need to get the region of the scissor so that our geometry doesn't - * get rejected before the depth test. Should probably cull rect against - * scissor for viewport but this is a rare case I think */ - glGetFloatv(GL_SCISSOR_BOX, viewport); - if (!input || input->xmin == input->xmax) { - glViewport(viewport[0], viewport[1], 24, 24); - } - else { - glViewport(viewport[0], viewport[1], (int)(input->xmax - input->xmin), (int)(input->ymax - input->ymin)); - } + g_select_state.algorithm = ALGO_GL_QUERY; + } - /* occlusion queries operates on fragments that pass tests and since we are interested on all - * objects in the view frustum independently of their order, we need to disable the depth test */ - if (mode == GPU_SELECT_ALL) { - glDisable(GL_DEPTH_TEST); - glDepthMask(GL_FALSE); + switch (g_select_state.algorithm) { + case ALGO_GL_LEGACY: + { + g_select_state.use_cache = false; + glSelectBuffer(bufsize, (GLuint *)buffer); + glRenderMode(GL_SELECT); + glInitNames(); + glPushName(-1); + break; } - else if (mode == GPU_SELECT_NEAREST_FIRST_PASS) { - glClear(GL_DEPTH_BUFFER_BIT); - glEnable(GL_DEPTH_TEST); - glDepthMask(GL_TRUE); - glDepthFunc(GL_LEQUAL); + case ALGO_GL_QUERY: + { + g_select_state.use_cache = false; + gpu_select_query_begin((unsigned int (*)[4])buffer, bufsize / 4, input, mode, oldhits); + break; } - else if (mode == GPU_SELECT_NEAREST_SECOND_PASS) { - glEnable(GL_DEPTH_TEST); - glDepthMask(GL_FALSE); - glDepthFunc(GL_EQUAL); + default: /* ALGO_GL_PICK */ + { + gpu_select_pick_begin((unsigned int (*)[4])buffer, bufsize / 4, input, mode); + break; } } } @@ -146,41 +123,24 @@ void GPU_select_begin(unsigned int *buffer, unsigned int bufsize, rctf *input, c bool GPU_select_load_id(unsigned int id) { /* if no selection mode active, ignore */ - if (!g_query_state.select_is_active) + if (!g_select_state.select_is_active) return true; - if (!g_query_state.use_gpu_select) { - glLoadName(id); - } - else { - if (g_query_state.query_issued) { - glEndQuery(GL_SAMPLES_PASSED); + switch (g_select_state.algorithm) { + case ALGO_GL_LEGACY: + { + glLoadName(id); + return true; } - /* if required, allocate extra queries */ - if (g_query_state.active_query == g_query_state.num_of_queries) { - g_query_state.num_of_queries += ALLOC_QUERIES; - g_query_state.queries = MEM_reallocN(g_query_state.queries, g_query_state.num_of_queries * sizeof(*g_query_state.queries)); - g_query_state.id = MEM_reallocN(g_query_state.id, g_query_state.num_of_queries * sizeof(*g_query_state.id)); - glGenQueries(ALLOC_QUERIES, &g_query_state.queries[g_query_state.active_query]); + case ALGO_GL_QUERY: + { + return gpu_select_query_load_id(id); } - - glBeginQuery(GL_SAMPLES_PASSED, g_query_state.queries[g_query_state.active_query]); - g_query_state.id[g_query_state.active_query] = id; - g_query_state.active_query++; - g_query_state.query_issued = true; - - if (g_query_state.mode == GPU_SELECT_NEAREST_SECOND_PASS && g_query_state.index < g_query_state.oldhits) { - if (g_query_state.buffer[g_query_state.index * 4 + 3] == id) { - g_query_state.index++; - return true; - } - else { - return false; - } + default: /* ALGO_GL_PICK */ + { + return gpu_select_pick_load_id(id); } } - - return true; } /** @@ -191,59 +151,27 @@ bool GPU_select_load_id(unsigned int id) unsigned int GPU_select_end(void) { unsigned int hits = 0; - if (!g_query_state.use_gpu_select) { - glPopName(); - hits = glRenderMode(GL_RENDER); - } - else { - int i; - if (g_query_state.query_issued) { - glEndQuery(GL_SAMPLES_PASSED); + switch (g_select_state.algorithm) { + case ALGO_GL_LEGACY: + { + glPopName(); + hits = glRenderMode(GL_RENDER); + break; } - - for (i = 0; i < g_query_state.active_query; i++) { - unsigned int result; - glGetQueryObjectuiv(g_query_state.queries[i], GL_QUERY_RESULT, &result); - if (result > 0) { - if (g_query_state.mode != GPU_SELECT_NEAREST_SECOND_PASS) { - int maxhits = g_query_state.bufsize / 4; - - if (hits < maxhits) { - g_query_state.buffer[hits * 4] = 1; - g_query_state.buffer[hits * 4 + 1] = 0xFFFF; - g_query_state.buffer[hits * 4 + 2] = 0xFFFF; - g_query_state.buffer[hits * 4 + 3] = g_query_state.id[i]; - - hits++; - } - else { - hits = -1; - break; - } - } - else { - int j; - /* search in buffer and make selected object first */ - for (j = 0; j < g_query_state.oldhits; j++) { - if (g_query_state.buffer[j * 4 + 3] == g_query_state.id[i]) { - g_query_state.buffer[j * 4 + 1] = 0; - g_query_state.buffer[j * 4 + 2] = 0; - } - } - break; - } - } + case ALGO_GL_QUERY: + { + hits = gpu_select_query_end(); + break; + } + default: /* ALGO_GL_PICK */ + { + hits = gpu_select_pick_end(); + break; } - - glDeleteQueries(g_query_state.num_of_queries, g_query_state.queries); - MEM_freeN(g_query_state.queries); - MEM_freeN(g_query_state.id); - glPopAttrib(); - glColorMask(GL_TRUE, GL_TRUE, GL_TRUE, GL_TRUE); } - g_query_state.select_is_active = false; + g_select_state.select_is_active = false; return hits; } @@ -260,3 +188,41 @@ bool GPU_select_query_check_active(void) GPU_type_matches(GPU_DEVICE_NVIDIA, GPU_OS_UNIX, GPU_DRIVER_OPENSOURCE)))); } + +/* ---------------------------------------------------------------------------- + * Caching + * + * Support multiple begin/end's as long as they are within the initial region. + * Currently only used by ALGO_GL_PICK. + */ + +void GPU_select_cache_begin(void) +{ + /* validate on GPU_select_begin, clear if not supported */ + BLI_assert(g_select_state.use_cache == false); + g_select_state.use_cache = true; + if (g_select_state.algorithm == ALGO_GL_PICK) { + gpu_select_pick_cache_begin(); + } +} + +void GPU_select_cache_load_id(void) +{ + BLI_assert(g_select_state.use_cache == true); + if (g_select_state.algorithm == ALGO_GL_PICK) { + gpu_select_pick_cache_load_id(); + } +} + +void GPU_select_cache_end(void) +{ + if (g_select_state.algorithm == ALGO_GL_PICK) { + gpu_select_pick_cache_end(); + } + g_select_state.use_cache = false; +} + +bool GPU_select_is_cached(void) +{ + return g_select_state.use_cache && gpu_select_pick_is_cached(); +} diff --git a/source/blender/gpu/intern/gpu_select_pick.c b/source/blender/gpu/intern/gpu_select_pick.c new file mode 100644 index 00000000000..0a77420fa25 --- /dev/null +++ b/source/blender/gpu/intern/gpu_select_pick.c @@ -0,0 +1,744 @@ +/* + * ***** BEGIN GPL LICENSE BLOCK ***** + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License + * as published by the Free Software Foundation; either version 2 + * of the License, or (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software Foundation, + * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. + * + * The Original Code is Copyright (C) 2017 Blender Foundation. + * All rights reserved. + * + * ***** END GPL LICENSE BLOCK ***** + */ + +/** \file blender/gpu/intern/gpu_select_pick.c + * \ingroup gpu + * + * Custom select code for picking small regions (not efficient for large regions). + * `gpu_select_pick_*` API. + */ +#include <string.h> +#include <stdlib.h> +#include <float.h> + +#include "GPU_select.h" +#include "GPU_extensions.h" +#include "GPU_glew.h" + +#include "MEM_guardedalloc.h" + +#include "BLI_rect.h" +#include "BLI_listbase.h" +#include "BLI_math_vector.h" +#include "BLI_utildefines.h" + +#include "gpu_select_private.h" + +#include "BLI_strict_flags.h" + +/* #define DEBUG_PRINT */ + +/* Alloc number for depths */ +#define ALLOC_DEPTHS 200 + +/* Z-depth of cleared depth buffer */ +#define DEPTH_MAX 0xffffffff + +/* ---------------------------------------------------------------------------- + * SubRectStride + */ + +/* For looping over a sub-region of a rect, could be moved into 'rct.c'*/ +typedef struct SubRectStride { + unsigned int start; /* start here */ + unsigned int span; /* read these */ + unsigned int span_len; /* len times (read span 'len' times). */ + unsigned int skip; /* skip those */ +} SubRectStride; + +/* we may want to change back to float if uint isn't well supported */ +typedef unsigned int depth_t; + +/** + * Calculate values needed for looping over a sub-region (smaller buffer within a larger buffer). + * + * 'src' must be bigger than 'dst'. + */ +static void rect_subregion_stride_calc(const rcti *src, const rcti *dst, SubRectStride *r_sub) +{ + const int src_x = BLI_rcti_size_x(src); + // const int src_y = BLI_rcti_size_y(src); + const int dst_x = BLI_rcti_size_x(dst); + const int dst_y = BLI_rcti_size_y(dst); + const int x = dst->xmin - src->xmin; + const int y = dst->ymin - src->ymin; + + BLI_assert(src->xmin <= dst->xmin && src->ymin <= dst->ymin && + src->ymax >= dst->ymax && src->ymax >= dst->ymax); + BLI_assert(x >= 0 && y >= 0); + + r_sub->start = (unsigned int)((src_x * y) + x); + r_sub->span = (unsigned int)dst_x; + r_sub->span_len = (unsigned int)dst_y; + r_sub->skip = (unsigned int)(src_x - dst_x); +} + +/** + * Ignore depth clearing as a change, + * only check if its been changed _and_ filled in (ignore clearing since XRAY does this). + */ +BLI_INLINE bool depth_is_filled(const depth_t *prev, const depth_t *curr) +{ + return (*prev != *curr) && (*curr != DEPTH_MAX); +} + +/* ---------------------------------------------------------------------------- + * DepthBufCache + * + * Result of reading glReadPixels, + * use for both cache and non-cached storage. + */ + +/* store result of glReadPixels */ +typedef struct DepthBufCache { + struct DepthBufCache *next, *prev; + unsigned int id; + depth_t buf[0]; +} DepthBufCache; + +static DepthBufCache *depth_buf_malloc(unsigned int rect_len) +{ + DepthBufCache *rect = MEM_mallocN(sizeof(DepthBufCache) + sizeof(depth_t) * rect_len, __func__); + rect->id = SELECT_ID_NONE; + return rect; +} + +static bool depth_buf_rect_depth_any( + const DepthBufCache *rect_depth, + unsigned int rect_len) +{ + const depth_t *curr = rect_depth->buf; + for (unsigned int i = 0; i < rect_len; i++, curr++) { + if (*curr != DEPTH_MAX) { + return true; + } + } + return false; +} + +static bool depth_buf_subrect_depth_any( + const DepthBufCache *rect_depth, + const SubRectStride *sub_rect) +{ + const depth_t *curr = rect_depth->buf + sub_rect->start; + for (unsigned int i = 0; i < sub_rect->span_len; i++) { + const depth_t *curr_end = curr + sub_rect->span; + for (; curr < curr_end; curr++, curr++) { + if (*curr != DEPTH_MAX) { + return true; + } + } + curr += sub_rect->skip; + } + return false; +} + +static bool depth_buf_rect_depth_any_filled( + const DepthBufCache *rect_prev, const DepthBufCache *rect_curr, + unsigned int rect_len) +{ +#if 0 + return memcmp(rect_depth_a->buf, rect_depth_b->buf, rect_len * sizeof(depth_t)) != 0; +#else + const depth_t *prev = rect_prev->buf; + const depth_t *curr = rect_curr->buf; + for (unsigned int i = 0; i < rect_len; i++, curr++, prev++) { + if (depth_is_filled(prev, curr)) { + return true; + } + } + return false; +#endif +} + +/** + * Both buffers are the same size, just check if the sub-rect contains any differences. + */ +static bool depth_buf_subrect_depth_any_filled( + const DepthBufCache *rect_src, const DepthBufCache *rect_dst, + const SubRectStride *sub_rect) +{ + /* same as above but different rect sizes */ + const depth_t *prev = rect_src->buf + sub_rect->start; + const depth_t *curr = rect_dst->buf + sub_rect->start; + for (unsigned int i = 0; i < sub_rect->span_len; i++) { + const depth_t *curr_end = curr + sub_rect->span; + for (; curr < curr_end; prev++, curr++) { + if (depth_is_filled(prev, curr)) { + return true; + } + } + prev += sub_rect->skip; + curr += sub_rect->skip; + } + return false; +} + +/* ---------------------------------------------------------------------------- + * DepthID + * + * Internal structure for storing hits. + */ + +typedef struct DepthID { + unsigned int id; + depth_t depth; +} DepthID; + +static int depth_id_cmp(const void *v1, const void *v2) +{ + const DepthID *d1 = v1, *d2 = v2; + if (d1->id < d2->id) { + return -1; + } + else if (d1->id > d2->id) { + return 1; + } + else { + return 0; + } +} + +static int depth_cmp(const void *v1, const void *v2) +{ + const DepthID *d1 = v1, *d2 = v2; + if (d1->depth < d2->depth) { + return -1; + } + else if (d1->depth > d2->depth) { + return 1; + } + else { + return 0; + } +} + +/* depth sorting */ +typedef struct GPUPickState { + /* cache on initialization */ + unsigned int (*buffer)[4]; + + /* buffer size (stores number of integers, for actual size multiply by sizeof integer)*/ + unsigned int bufsize; + /* mode of operation */ + char mode; + + /* OpenGL drawing, never use when (is_cached == true). */ + struct { + /* The current depth, accumulated as we draw */ + DepthBufCache *rect_depth; + /* Scratch buffer, avoid allocs every time (when not caching) */ + DepthBufCache *rect_depth_test; + + /* Pass to glReadPixels (x, y, w, h) */ + int clip_readpixels[4]; + + /* Set after first draw */ + bool is_init; + unsigned int prev_id; + } gl; + + /* src: data stored in 'cache' and 'gl', + * dst: use when cached region is smaller (where src -> dst isn't 1:1) */ + struct { + rcti clip_rect; + unsigned int rect_len; + } src, dst; + + /* Store cache between `GPU_select_cache_begin/end` */ + bool use_cache; + bool is_cached; + struct { + /* Cleanup used for iterating over both source and destination buffers: + * src.clip_rect -> dst.clip_rect */ + SubRectStride sub_rect; + + /* List of DepthBufCache, sized of 'src.clip_rect' */ + ListBase bufs; + } cache; + + /* Pickign methods */ + union { + /* GPU_SELECT_PICK_ALL */ + struct { + DepthID *hits; + unsigned int hits_len; + unsigned int hits_len_alloc; + } all; + + /* GPU_SELECT_PICK_NEAREST */ + struct { + unsigned int *rect_id; + } nearest; + }; +} GPUPickState; + + +static GPUPickState g_pick_state = {0}; + +void gpu_select_pick_begin( + unsigned int (*buffer)[4], unsigned int bufsize, + const rcti *input, char mode) +{ + GPUPickState *ps = &g_pick_state; + +#ifdef DEBUG_PRINT + printf("%s: mode=%d, use_cache=%d, is_cache=%d\n", __func__, mode, ps->use_cache, ps->is_cached); +#endif + + ps->bufsize = bufsize; + ps->buffer = buffer; + ps->mode = mode; + + const unsigned int rect_len = (unsigned int)(BLI_rcti_size_x(input) * BLI_rcti_size_y(input)); + ps->dst.clip_rect = *input; + ps->dst.rect_len = rect_len; + + /* Restrict OpenGL operations for when we don't have cache */ + if (ps->is_cached == false) { + + glPushAttrib(GL_DEPTH_BUFFER_BIT | GL_VIEWPORT_BIT); + /* disable writing to the framebuffer */ + glColorMask(GL_FALSE, GL_FALSE, GL_FALSE, GL_FALSE); + + glEnable(GL_DEPTH_TEST); + glDepthMask(GL_TRUE); + + if (mode == GPU_SELECT_PICK_ALL) { + glDepthFunc(GL_ALWAYS); + } + else { + glDepthFunc(GL_LEQUAL); + } + + /* set just in case */ + glPixelTransferf(GL_DEPTH_BIAS, 0.0); + glPixelTransferf(GL_DEPTH_SCALE, 1.0); + + float viewport[4]; + glGetFloatv(GL_SCISSOR_BOX, viewport); + + ps->src.clip_rect = *input; + ps->src.rect_len = rect_len; + + ps->gl.clip_readpixels[0] = (int)viewport[0]; + ps->gl.clip_readpixels[1] = (int)viewport[1]; + ps->gl.clip_readpixels[2] = BLI_rcti_size_x(&ps->src.clip_rect); + ps->gl.clip_readpixels[3] = BLI_rcti_size_y(&ps->src.clip_rect); + + glViewport(UNPACK4(ps->gl.clip_readpixels)); + + /* It's possible we don't want to clear depth buffer, + * so existing elements are masked by current z-buffer. */ + glClear(GL_DEPTH_BUFFER_BIT); + + /* scratch buffer (read new values here) */ + ps->gl.rect_depth_test = depth_buf_malloc(rect_len); + ps->gl.rect_depth = depth_buf_malloc(rect_len); + + /* set initial 'far' value */ +#if 0 + glReadPixels(UNPACK4(ps->gl.clip_readpixels), GL_DEPTH_COMPONENT, GL_UNSIGNED_INT, ps->gl.rect_depth->buf); +#else + for (unsigned int i = 0; i < rect_len; i++) { + ps->gl.rect_depth->buf[i] = DEPTH_MAX; + } +#endif + + ps->gl.is_init = false; + ps->gl.prev_id = 0; + } + else { + /* Using cache (ps->is_cached == true) */ + /* src.clip_rect -> dst.clip_rect */ + rect_subregion_stride_calc(&ps->src.clip_rect, &ps->dst.clip_rect, &ps->cache.sub_rect); + BLI_assert(ps->gl.rect_depth == NULL); + BLI_assert(ps->gl.rect_depth_test == NULL); + } + + if (mode == GPU_SELECT_PICK_ALL) { + ps->all.hits = MEM_mallocN(sizeof(*ps->all.hits) * ALLOC_DEPTHS, __func__); + ps->all.hits_len = 0; + ps->all.hits_len_alloc = ALLOC_DEPTHS; + } + else { + /* Set to 0xff for SELECT_ID_NONE */ + ps->nearest.rect_id = MEM_mallocN(sizeof(unsigned int) * ps->dst.rect_len, __func__); + memset(ps->nearest.rect_id, 0xff, sizeof(unsigned int) * ps->dst.rect_len); + } +} + +/** + * Given 2x depths, we know are different - update the depth information + * use for both cached/uncached depth buffers. + */ +static void gpu_select_load_id_pass_all(const DepthBufCache *rect_curr) +{ + GPUPickState *ps = &g_pick_state; + const unsigned int id = rect_curr->id; + /* find the best depth for this pass and store in 'all.hits' */ + depth_t depth_best = DEPTH_MAX; + +#define EVAL_TEST() \ + if (depth_best > *curr) { \ + depth_best = *curr; \ + } ((void)0) + + if (ps->is_cached == false) { + const depth_t *curr = rect_curr->buf; + BLI_assert(ps->src.rect_len == ps->dst.rect_len); + const unsigned int rect_len = ps->src.rect_len; + for (unsigned int i = 0; i < rect_len; i++, curr++) { + EVAL_TEST(); + } + } + else { + /* same as above but different rect sizes */ + const depth_t *curr = rect_curr->buf + ps->cache.sub_rect.start; + for (unsigned int i = 0; i < ps->cache.sub_rect.span_len; i++) { + const depth_t *curr_end = curr + ps->cache.sub_rect.span; + for (; curr < curr_end; curr++) { + EVAL_TEST(); + } + curr += ps->cache.sub_rect.skip; + } + } + +#undef EVAL_TEST + + /* ensure enough space */ + if (UNLIKELY(ps->all.hits_len == ps->all.hits_len_alloc)) { + ps->all.hits_len_alloc += ALLOC_DEPTHS; + ps->all.hits = MEM_reallocN(ps->all.hits, ps->all.hits_len_alloc * sizeof(*ps->all.hits)); + } + DepthID *d = &ps->all.hits[ps->all.hits_len++]; + d->id = id; + d->depth = depth_best; +} + +static void gpu_select_load_id_pass_nearest(const DepthBufCache *rect_prev, const DepthBufCache *rect_curr) +{ + GPUPickState *ps = &g_pick_state; + const unsigned int id = rect_curr->id; + /* keep track each pixels ID in 'nearest.rect_id' */ + if (id != SELECT_ID_NONE) { + unsigned int *id_ptr = ps->nearest.rect_id; + + /* Check against DEPTH_MAX because XRAY will clear the buffer, + * so previously set values will become unset. + * In this case just leave those id's left as-is. */ +#define EVAL_TEST() \ + if (depth_is_filled(prev, curr)) { \ + *id_ptr = id; \ + } ((void)0) + + if (ps->is_cached == false) { + const depth_t *prev = rect_prev->buf; + const depth_t *curr = rect_curr->buf; + BLI_assert(ps->src.rect_len == ps->dst.rect_len); + const unsigned int rect_len = ps->src.rect_len; + for (unsigned int i = 0; i < rect_len; i++, curr++, prev++, id_ptr++) { + EVAL_TEST(); + } + } + else { + /* same as above but different rect sizes */ + const depth_t *prev = rect_prev->buf + ps->cache.sub_rect.start; + const depth_t *curr = rect_curr->buf + ps->cache.sub_rect.start; + for (unsigned int i = 0; i < ps->cache.sub_rect.span_len; i++) { + const depth_t *curr_end = curr + ps->cache.sub_rect.span; + for (; curr < curr_end; prev++, curr++, id_ptr++) { + EVAL_TEST(); + } + prev += ps->cache.sub_rect.skip; + curr += ps->cache.sub_rect.skip; + } + } + +#undef EVAL_TEST + } +} + + +bool gpu_select_pick_load_id(unsigned int id) +{ + GPUPickState *ps = &g_pick_state; + if (ps->gl.is_init) { + const unsigned int rect_len = ps->src.rect_len; + glReadPixels(UNPACK4(ps->gl.clip_readpixels), GL_DEPTH_COMPONENT, GL_UNSIGNED_INT, ps->gl.rect_depth_test->buf); + /* perform initial check since most cases the array remains unchanged */ + + bool do_pass = false; + if (g_pick_state.mode == GPU_SELECT_PICK_ALL) { + if (depth_buf_rect_depth_any(ps->gl.rect_depth_test, rect_len)) { + ps->gl.rect_depth_test->id = ps->gl.prev_id; + gpu_select_load_id_pass_all(ps->gl.rect_depth_test); + do_pass = true; + } + } + else { + if (depth_buf_rect_depth_any_filled(ps->gl.rect_depth, ps->gl.rect_depth_test, rect_len)) { + ps->gl.rect_depth_test->id = ps->gl.prev_id; + gpu_select_load_id_pass_nearest(ps->gl.rect_depth, ps->gl.rect_depth_test); + do_pass = true; + } + } + + if (do_pass) { + /* Store depth in cache */ + if (ps->use_cache) { + BLI_addtail(&ps->cache.bufs, ps->gl.rect_depth); + ps->gl.rect_depth = depth_buf_malloc(ps->src.rect_len); + } + + SWAP(DepthBufCache *, ps->gl.rect_depth, ps->gl.rect_depth_test); + + if (g_pick_state.mode == GPU_SELECT_PICK_ALL) { + /* we want new depths every time */ + glClear(GL_DEPTH_BUFFER_BIT); + } + } + } + + ps->gl.is_init = true; + ps->gl.prev_id = id; + + return true; +} + +unsigned int gpu_select_pick_end(void) +{ + GPUPickState *ps = &g_pick_state; + +#ifdef DEBUG_PRINT + printf("%s\n", __func__); +#endif + + if (ps->is_cached == false) { + if (ps->gl.is_init) { + /* force finishing last pass */ + gpu_select_pick_load_id(ps->gl.prev_id); + } + + glPopAttrib(); + glColorMask(GL_TRUE, GL_TRUE, GL_TRUE, GL_TRUE); + } + + /* assign but never free directly since it may be in cache */ + DepthBufCache *rect_depth_final; + + /* Store depth in cache */ + if (ps->use_cache && !ps->is_cached) { + BLI_addtail(&ps->cache.bufs, ps->gl.rect_depth); + ps->gl.rect_depth = NULL; + rect_depth_final = ps->cache.bufs.last; + } + else if (ps->is_cached) { + rect_depth_final = ps->cache.bufs.last; + } + else { + /* common case, no cache */ + rect_depth_final = ps->gl.rect_depth; + } + + unsigned int maxhits = g_pick_state.bufsize; + DepthID *depth_data; + unsigned int depth_data_len = 0; + + if (g_pick_state.mode == GPU_SELECT_PICK_ALL) { + depth_data = ps->all.hits; + depth_data_len = ps->all.hits_len; + /* move ownership */ + ps->all.hits = NULL; + ps->all.hits_len = 0; + ps->all.hits_len_alloc = 0; + } + else { + /* GPU_SELECT_PICK_NEAREST */ + + /* Over alloc (unlikely we have as many depths as pixels) */ + unsigned int depth_data_len_first_pass = 0; + depth_data = MEM_mallocN(ps->dst.rect_len * sizeof(*depth_data), __func__); + + /* Partially de-duplicating copy, + * when contiguous ID's are found - update their closest depth. + * This isn't essential but means there is less data to sort. */ + +#define EVAL_TEST(i_src, i_dst) \ + { \ + const unsigned int id = ps->nearest.rect_id[i_dst]; \ + if (id != SELECT_ID_NONE) { \ + const depth_t depth = rect_depth_final->buf[i_src]; \ + if (depth_last == NULL || depth_last->id != id) { \ + DepthID *d = &depth_data[depth_data_len_first_pass++]; \ + d->id = id; \ + d->depth = depth; \ + } \ + else if (depth_last->depth > depth) { \ + depth_last->depth = depth; \ + } \ + } \ + } ((void)0) + + { + DepthID *depth_last = NULL; + if (ps->is_cached == false) { + for (unsigned int i = 0; i < ps->src.rect_len; i++) { + EVAL_TEST(i, i); + } + } + else { + /* same as above but different rect sizes */ + unsigned int i_src = ps->cache.sub_rect.start, i_dst = 0; + for (unsigned int j = 0; j < ps->cache.sub_rect.span_len; j++) { + const unsigned int i_src_end = i_src + ps->cache.sub_rect.span; + for (; i_src < i_src_end; i_src++, i_dst++) { + EVAL_TEST(i_src, i_dst); + } + i_src += ps->cache.sub_rect.skip; + } + } + } + +#undef EVAL_TEST + + qsort(depth_data, depth_data_len_first_pass, sizeof(DepthID), depth_id_cmp); + + /* Sort by ID's then keep the best depth for each ID */ + depth_data_len = 0; + { + DepthID *depth_last = NULL; + for (unsigned int i = 0; i < depth_data_len_first_pass; i++) { + if (depth_last == NULL || depth_last->id != depth_data[i].id) { + depth_last = &depth_data[depth_data_len++]; + *depth_last = depth_data[i]; + } + else if (depth_last->depth > depth_data[i].depth) { + depth_last->depth = depth_data[i].depth; + } + } + } + } + + /* Finally sort each unique (id, depth) pair by depth + * so the final hit-list is sorted by depth (nearest first) */ + unsigned int hits = 0; + + if (depth_data_len > maxhits) { + hits = (unsigned int)-1; + } + else { + /* leave sorting up to the caller */ + qsort(depth_data, depth_data_len, sizeof(DepthID), depth_cmp); + + for (unsigned int i = 0; i < depth_data_len; i++) { +#ifdef DEBUG_PRINT + printf(" hit: %u: depth %u\n", depth_data[i].id, depth_data[i].depth); +#endif + /* first 3 are dummy values */ + g_pick_state.buffer[hits][0] = 1; + g_pick_state.buffer[hits][1] = 0x0; /* depth_data[i].depth; */ /* unused */ + g_pick_state.buffer[hits][2] = 0x0; /* z-far is currently never used. */ + g_pick_state.buffer[hits][3] = depth_data[i].id; + hits++; + } + BLI_assert(hits < maxhits); + } + + MEM_freeN(depth_data); + + MEM_SAFE_FREE(ps->gl.rect_depth); + MEM_SAFE_FREE(ps->gl.rect_depth_test); + + if (g_pick_state.mode == GPU_SELECT_PICK_ALL) { + /* 'hits' already freed as 'depth_data' */ + } + else { + MEM_freeN(ps->nearest.rect_id); + ps->nearest.rect_id = NULL; + } + + if (ps->use_cache) { + ps->is_cached = true; + } + + return hits; +} + +/* ---------------------------------------------------------------------------- + * Caching + * + * Support multiple begin/end's reusing depth buffers. + */ + +void gpu_select_pick_cache_begin(void) +{ + BLI_assert(g_pick_state.use_cache == false); +#ifdef DEBUG_PRINT + printf("%s\n", __func__); +#endif + g_pick_state.use_cache = true; + g_pick_state.is_cached = false; +} + +void gpu_select_pick_cache_end(void) +{ +#ifdef DEBUG_PRINT + printf("%s: with %d buffers\n", __func__, BLI_listbase_count(&g_pick_state.cache.bufs)); +#endif + g_pick_state.use_cache = false; + g_pick_state.is_cached = false; + + BLI_freelistN(&g_pick_state.cache.bufs); +} + +/* is drawing needed? */ +bool gpu_select_pick_is_cached(void) +{ + return g_pick_state.is_cached; +} + +void gpu_select_pick_cache_load_id(void) +{ + BLI_assert(g_pick_state.is_cached == true); + GPUPickState *ps = &g_pick_state; +#ifdef DEBUG_PRINT + printf("%s (building depth from cache)\n", __func__); +#endif + for (DepthBufCache *rect_depth = ps->cache.bufs.first; rect_depth; rect_depth = rect_depth->next) { + if (rect_depth->next != NULL) { + /* we know the buffers differ, but this sub-region may not. + * double check before adding an id-pass */ + if (g_pick_state.mode == GPU_SELECT_PICK_ALL) { + if (depth_buf_subrect_depth_any(rect_depth->next, &ps->cache.sub_rect)) { + gpu_select_load_id_pass_all(rect_depth->next); + } + } + else { + if (depth_buf_subrect_depth_any_filled(rect_depth, rect_depth->next, &ps->cache.sub_rect)) { + gpu_select_load_id_pass_nearest(rect_depth, rect_depth->next); + } + } + } + } +} diff --git a/source/blender/gpu/intern/gpu_select_private.h b/source/blender/gpu/intern/gpu_select_private.h new file mode 100644 index 00000000000..8935bd7b253 --- /dev/null +++ b/source/blender/gpu/intern/gpu_select_private.h @@ -0,0 +1,53 @@ +/* + * ***** BEGIN GPL LICENSE BLOCK ***** + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License + * as published by the Free Software Foundation; either version 2 + * of the License, or (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software Foundation, + * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. + * + * The Original Code is Copyright (C) 2014 Blender Foundation. + * All rights reserved. + * + * Contributor(s): Antony Riakiotakis. + * + * ***** END GPL LICENSE BLOCK ***** + */ + +/** \file blender/gpu/intern/gpu_select_private.h + * \ingroup gpu + * + * Selection implementations. + */ + +#ifndef __GPU_SELECT_PRIVATE_H__ +#define __GPU_SELECT_PRIVATE_H__ + +/* gpu_select_pick */ +void gpu_select_pick_begin(unsigned int (*buffer)[4], unsigned int bufsize, const rcti *input, char mode); +bool gpu_select_pick_load_id(unsigned int id); +unsigned int gpu_select_pick_end(void); + +void gpu_select_pick_cache_begin(void); +void gpu_select_pick_cache_end(void); +bool gpu_select_pick_is_cached(void); +void gpu_select_pick_cache_load_id(void); + +/* gpu_select_sample_query */ +void gpu_select_query_begin(unsigned int (*buffer)[4], unsigned int bufsize, const rcti *input, char mode, int oldhits); +bool gpu_select_query_load_id(unsigned int id); +unsigned int gpu_select_query_end(void); + + +#define SELECT_ID_NONE ((unsigned int)0xffffffff) + +#endif /* __GPU_SELECT_PRIVATE_H__ */ diff --git a/source/blender/gpu/intern/gpu_select_sample_query.c b/source/blender/gpu/intern/gpu_select_sample_query.c new file mode 100644 index 00000000000..ba5fefc5227 --- /dev/null +++ b/source/blender/gpu/intern/gpu_select_sample_query.c @@ -0,0 +1,209 @@ +/* + * ***** BEGIN GPL LICENSE BLOCK ***** + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License + * as published by the Free Software Foundation; either version 2 + * of the License, or (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software Foundation, + * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. + * + * The Original Code is Copyright (C) 2014 Blender Foundation. + * All rights reserved. + * + * Contributor(s): Antony Riakiotakis. + * + * ***** END GPL LICENSE BLOCK ***** + */ + +/** \file blender/gpu/intern/gpu_select_sample_query.c + * \ingroup gpu + * + * Interface for accessing gpu-related methods for selection. The semantics will be + * similar to glRenderMode(GL_SELECT) since the goal is to maintain compatibility. + */ + +#include <stdlib.h> + +#include "GPU_select.h" +#include "GPU_extensions.h" +#include "GPU_glew.h" + +#include "MEM_guardedalloc.h" + +#include "BLI_rect.h" + +#include "BLI_utildefines.h" + +#include "gpu_select_private.h" + + +/* Ad hoc number of queries to allocate to skip doing many glGenQueries */ +#define ALLOC_QUERIES 200 + +typedef struct GPUQueryState { + /* Tracks whether a query has been issued so that gpu_load_id can end the previous one */ + bool query_issued; + /* array holding the OpenGL query identifiers */ + unsigned int *queries; + /* array holding the id corresponding to each query */ + unsigned int *id; + /* number of queries in *queries and *id */ + unsigned int num_of_queries; + /* index to the next query to start */ + unsigned int active_query; + /* cache on initialization */ + unsigned int (*buffer)[4]; + /* buffer size (stores number of integers, for actual size multiply by sizeof integer)*/ + unsigned int bufsize; + /* mode of operation */ + char mode; + unsigned int index; + int oldhits; +} GPUQueryState; + +static GPUQueryState g_query_state = {0}; + + +void gpu_select_query_begin( + unsigned int (*buffer)[4], unsigned int bufsize, + const rcti *input, char mode, + int oldhits) +{ + float viewport[4]; + + g_query_state.query_issued = false; + g_query_state.active_query = 0; + g_query_state.num_of_queries = 0; + g_query_state.bufsize = bufsize; + g_query_state.buffer = buffer; + g_query_state.mode = mode; + g_query_state.index = 0; + g_query_state.oldhits = oldhits; + + g_query_state.num_of_queries = ALLOC_QUERIES; + + g_query_state.queries = MEM_mallocN(g_query_state.num_of_queries * sizeof(*g_query_state.queries), "gpu selection queries"); + g_query_state.id = MEM_mallocN(g_query_state.num_of_queries * sizeof(*g_query_state.id), "gpu selection ids"); + glGenQueries(g_query_state.num_of_queries, g_query_state.queries); + + glPushAttrib(GL_DEPTH_BUFFER_BIT | GL_VIEWPORT_BIT); + /* disable writing to the framebuffer */ + glColorMask(GL_FALSE, GL_FALSE, GL_FALSE, GL_FALSE); + + /* In order to save some fill rate we minimize the viewport using rect. + * We need to get the region of the scissor so that our geometry doesn't + * get rejected before the depth test. Should probably cull rect against + * scissor for viewport but this is a rare case I think */ + glGetFloatv(GL_SCISSOR_BOX, viewport); + glViewport(viewport[0], viewport[1], BLI_rcti_size_x(input), BLI_rcti_size_y(input)); + + /* occlusion queries operates on fragments that pass tests and since we are interested on all + * objects in the view frustum independently of their order, we need to disable the depth test */ + if (mode == GPU_SELECT_ALL) { + glDisable(GL_DEPTH_TEST); + glDepthMask(GL_FALSE); + } + else if (mode == GPU_SELECT_NEAREST_FIRST_PASS) { + glClear(GL_DEPTH_BUFFER_BIT); + glEnable(GL_DEPTH_TEST); + glDepthMask(GL_TRUE); + glDepthFunc(GL_LEQUAL); + } + else if (mode == GPU_SELECT_NEAREST_SECOND_PASS) { + glEnable(GL_DEPTH_TEST); + glDepthMask(GL_FALSE); + glDepthFunc(GL_EQUAL); + } +} + +bool gpu_select_query_load_id(unsigned int id) +{ + if (g_query_state.query_issued) { + glEndQuery(GL_SAMPLES_PASSED); + } + /* if required, allocate extra queries */ + if (g_query_state.active_query == g_query_state.num_of_queries) { + g_query_state.num_of_queries += ALLOC_QUERIES; + g_query_state.queries = MEM_reallocN(g_query_state.queries, g_query_state.num_of_queries * sizeof(*g_query_state.queries)); + g_query_state.id = MEM_reallocN(g_query_state.id, g_query_state.num_of_queries * sizeof(*g_query_state.id)); + glGenQueries(ALLOC_QUERIES, &g_query_state.queries[g_query_state.active_query]); + } + + glBeginQuery(GL_SAMPLES_PASSED, g_query_state.queries[g_query_state.active_query]); + g_query_state.id[g_query_state.active_query] = id; + g_query_state.active_query++; + g_query_state.query_issued = true; + + if (g_query_state.mode == GPU_SELECT_NEAREST_SECOND_PASS && g_query_state.index < g_query_state.oldhits) { + if (g_query_state.buffer[g_query_state.index][3] == id) { + g_query_state.index++; + return true; + } + else { + return false; + } + } + + return true; +} + +unsigned int gpu_select_query_end(void) +{ + int i; + + unsigned int hits = 0; + const unsigned int maxhits = g_query_state.bufsize; + + if (g_query_state.query_issued) { + glEndQuery(GL_SAMPLES_PASSED); + } + + for (i = 0; i < g_query_state.active_query; i++) { + unsigned int result; + glGetQueryObjectuiv(g_query_state.queries[i], GL_QUERY_RESULT, &result); + if (result > 0) { + if (g_query_state.mode != GPU_SELECT_NEAREST_SECOND_PASS) { + + if (hits < maxhits) { + g_query_state.buffer[hits][0] = 1; + g_query_state.buffer[hits][1] = 0xFFFF; + g_query_state.buffer[hits][2] = 0xFFFF; + g_query_state.buffer[hits][3] = g_query_state.id[i]; + + hits++; + } + else { + hits = -1; + break; + } + } + else { + int j; + /* search in buffer and make selected object first */ + for (j = 0; j < g_query_state.oldhits; j++) { + if (g_query_state.buffer[j][3] == g_query_state.id[i]) { + g_query_state.buffer[j][1] = 0; + g_query_state.buffer[j][2] = 0; + } + } + break; + } + } + } + + glDeleteQueries(g_query_state.num_of_queries, g_query_state.queries); + MEM_freeN(g_query_state.queries); + MEM_freeN(g_query_state.id); + glPopAttrib(); + glColorMask(GL_TRUE, GL_TRUE, GL_TRUE, GL_TRUE); + + return hits; +} diff --git a/source/blender/imbuf/IMB_colormanagement.h b/source/blender/imbuf/IMB_colormanagement.h index 52febe642a0..4164f5fa75d 100644 --- a/source/blender/imbuf/IMB_colormanagement.h +++ b/source/blender/imbuf/IMB_colormanagement.h @@ -77,6 +77,16 @@ void IMB_colormanagement_transform(float *buffer, int width, int height, int cha const char *from_colorspace, const char *to_colorspace, bool predivide); void IMB_colormanagement_transform_threaded(float *buffer, int width, int height, int channels, const char *from_colorspace, const char *to_colorspace, bool predivide); +void IMB_colormanagement_transform_byte(unsigned char *buffer, int width, int height, int channels, + const char *from_colorspace, const char *to_colorspace); +void IMB_colormanagement_transform_byte_threaded(unsigned char *buffer, int width, int height, int channels, + const char *from_colorspace, const char *to_colorspace); +void IMB_colormanagement_transform_from_byte(float *float_buffer, unsigned char *byte_buffer, + int width, int height, int channels, + const char *from_colorspace, const char *to_colorspace); +void IMB_colormanagement_transform_from_byte_threaded(float *float_buffer, unsigned char *byte_buffer, + int width, int height, int channels, + const char *from_colorspace, const char *to_colorspace); void IMB_colormanagement_transform_v4(float pixel[4], const char *from_colorspace, const char *to_colorspace); void IMB_colormanagement_colorspace_to_scene_linear_v3(float pixel[3], struct ColorSpace *colorspace); @@ -185,6 +195,8 @@ void IMB_colormanagement_processor_apply_v3(struct ColormanageProcessor *cm_proc void IMB_colormanagement_processor_apply_pixel(struct ColormanageProcessor *cm_processor, float *pixel, int channels); void IMB_colormanagement_processor_apply(struct ColormanageProcessor *cm_processor, float *buffer, int width, int height, int channels, bool predivide); +void IMB_colormanagement_processor_apply_byte(struct ColormanageProcessor *cm_processor, + unsigned char *buffer, int width, int height, int channels); void IMB_colormanagement_processor_free(struct ColormanageProcessor *cm_processor); /* ** OpenGL drawing routines using GLSL for color space transform ** */ diff --git a/source/blender/imbuf/intern/colormanagement.c b/source/blender/imbuf/intern/colormanagement.c index a45346279d9..48cba3e0800 100644 --- a/source/blender/imbuf/intern/colormanagement.c +++ b/source/blender/imbuf/intern/colormanagement.c @@ -1555,21 +1555,25 @@ static void colormanage_display_buffer_process(ImBuf *ibuf, unsigned char *displ typedef struct ProcessorTransformThread { ColormanageProcessor *cm_processor; - float *buffer; + unsigned char *byte_buffer; + float *float_buffer; int width; int start_line; int tot_line; int channels; bool predivide; + bool float_from_byte; } ProcessorTransformThread; typedef struct ProcessorTransformInit { ColormanageProcessor *cm_processor; - float *buffer; + unsigned char *byte_buffer; + float *float_buffer; int width; int height; int channels; bool predivide; + bool float_from_byte; } ProcessorTransformInitData; static void processor_transform_init_handle(void *handle_v, int start_line, int tot_line, void *init_data_v) @@ -1577,17 +1581,24 @@ static void processor_transform_init_handle(void *handle_v, int start_line, int ProcessorTransformThread *handle = (ProcessorTransformThread *) handle_v; ProcessorTransformInitData *init_data = (ProcessorTransformInitData *) init_data_v; - int channels = init_data->channels; - int width = init_data->width; - bool predivide = init_data->predivide; + const int channels = init_data->channels; + const int width = init_data->width; + const bool predivide = init_data->predivide; + const bool float_from_byte = init_data->float_from_byte; - size_t offset = ((size_t)channels) * start_line * width; + const size_t offset = ((size_t)channels) * start_line * width; memset(handle, 0, sizeof(ProcessorTransformThread)); handle->cm_processor = init_data->cm_processor; - handle->buffer = init_data->buffer + offset; + if (init_data->byte_buffer != NULL) { + /* TODO(serge): Offset might be different for byte and float buffers. */ + handle->byte_buffer = init_data->byte_buffer + offset; + } + if (init_data->float_buffer != NULL) { + handle->float_buffer = init_data->float_buffer + offset; + } handle->width = width; @@ -1596,33 +1607,62 @@ static void processor_transform_init_handle(void *handle_v, int start_line, int handle->channels = channels; handle->predivide = predivide; + handle->float_from_byte = float_from_byte; } static void *do_processor_transform_thread(void *handle_v) { ProcessorTransformThread *handle = (ProcessorTransformThread *) handle_v; - float *buffer = handle->buffer; - int channels = handle->channels; - int width = handle->width; - int height = handle->tot_line; - bool predivide = handle->predivide; - - IMB_colormanagement_processor_apply(handle->cm_processor, buffer, width, height, channels, predivide); + unsigned char *byte_buffer = handle->byte_buffer; + float *float_buffer = handle->float_buffer; + const int channels = handle->channels; + const int width = handle->width; + const int height = handle->tot_line; + const bool predivide = handle->predivide; + const bool float_from_byte = handle->float_from_byte; + + if (float_from_byte) { + IMB_buffer_float_from_byte(float_buffer, byte_buffer, + IB_PROFILE_SRGB, IB_PROFILE_SRGB, + true, + width, height, width, width); + IMB_colormanagement_processor_apply(handle->cm_processor, + float_buffer, + width, height, channels, + predivide); + } + else { + if (byte_buffer != NULL) { + IMB_colormanagement_processor_apply_byte(handle->cm_processor, + byte_buffer, + width, height, channels); + } + if (float_buffer != NULL) { + IMB_colormanagement_processor_apply(handle->cm_processor, + float_buffer, + width, height, channels, + predivide); + } + } return NULL; } -static void processor_transform_apply_threaded(float *buffer, int width, int height, int channels, - ColormanageProcessor *cm_processor, bool predivide) +static void processor_transform_apply_threaded(unsigned char *byte_buffer, float *float_buffer, + const int width, const int height, const int channels, + ColormanageProcessor *cm_processor, + const bool predivide, const bool float_from_byte) { ProcessorTransformInitData init_data; init_data.cm_processor = cm_processor; - init_data.buffer = buffer; + init_data.byte_buffer = byte_buffer; + init_data.float_buffer = float_buffer; init_data.width = width; init_data.height = height; init_data.channels = channels; init_data.predivide = predivide; + init_data.float_from_byte = float_from_byte; IMB_processor_apply_threaded(height, sizeof(ProcessorTransformThread), &init_data, processor_transform_init_handle, do_processor_transform_thread); @@ -1631,8 +1671,10 @@ static void processor_transform_apply_threaded(float *buffer, int width, int hei /*********************** Color space transformation functions *************************/ /* convert the whole buffer from specified by name color space to another - internal implementation */ -static void colormanagement_transform_ex(float *buffer, int width, int height, int channels, const char *from_colorspace, - const char *to_colorspace, bool predivide, bool do_threaded) +static void colormanagement_transform_ex(unsigned char *byte_buffer, float *float_buffer, + int width, int height, int channels, + const char *from_colorspace, const char *to_colorspace, + bool predivide, bool do_threaded) { ColormanageProcessor *cm_processor; @@ -1649,10 +1691,19 @@ static void colormanagement_transform_ex(float *buffer, int width, int height, i cm_processor = IMB_colormanagement_colorspace_processor_new(from_colorspace, to_colorspace); - if (do_threaded) - processor_transform_apply_threaded(buffer, width, height, channels, cm_processor, predivide); - else - IMB_colormanagement_processor_apply(cm_processor, buffer, width, height, channels, predivide); + if (do_threaded) { + processor_transform_apply_threaded(byte_buffer, float_buffer, + width, height, channels, + cm_processor, predivide, false); + } + else { + if (byte_buffer != NULL) { + IMB_colormanagement_processor_apply_byte(cm_processor, byte_buffer, width, height, channels); + } + if (float_buffer != NULL) { + IMB_colormanagement_processor_apply(cm_processor, float_buffer, width, height, channels, predivide); + } + } IMB_colormanagement_processor_free(cm_processor); } @@ -1661,7 +1712,7 @@ static void colormanagement_transform_ex(float *buffer, int width, int height, i void IMB_colormanagement_transform(float *buffer, int width, int height, int channels, const char *from_colorspace, const char *to_colorspace, bool predivide) { - colormanagement_transform_ex(buffer, width, height, channels, from_colorspace, to_colorspace, predivide, false); + colormanagement_transform_ex(NULL, buffer, width, height, channels, from_colorspace, to_colorspace, predivide, false); } /* convert the whole buffer from specified by name color space to another @@ -1670,7 +1721,54 @@ void IMB_colormanagement_transform(float *buffer, int width, int height, int cha void IMB_colormanagement_transform_threaded(float *buffer, int width, int height, int channels, const char *from_colorspace, const char *to_colorspace, bool predivide) { - colormanagement_transform_ex(buffer, width, height, channels, from_colorspace, to_colorspace, predivide, true); + colormanagement_transform_ex(NULL, buffer, width, height, channels, from_colorspace, to_colorspace, predivide, true); +} + +/* Similar to functions above, but operates on byte buffer. */ +void IMB_colormanagement_transform_byte(unsigned char *buffer, int width, int height, int channels, + const char *from_colorspace, const char *to_colorspace) +{ + colormanagement_transform_ex(buffer, NULL, width, height, channels, from_colorspace, to_colorspace, false, false); +} +void IMB_colormanagement_transform_byte_threaded(unsigned char *buffer, int width, int height, int channels, + const char *from_colorspace, const char *to_colorspace) +{ + colormanagement_transform_ex(buffer, NULL, width, height, channels, from_colorspace, to_colorspace, false, true); +} + +/* Similar to above, but gets float buffer from display one. */ +void IMB_colormanagement_transform_from_byte(float *float_buffer, unsigned char *byte_buffer, + int width, int height, int channels, + const char *from_colorspace, const char *to_colorspace) +{ + IMB_buffer_float_from_byte(float_buffer, byte_buffer, + IB_PROFILE_SRGB, IB_PROFILE_SRGB, + true, + width, height, width, width); + IMB_colormanagement_transform(float_buffer, + width, height, channels, + from_colorspace, to_colorspace, + true); +} +void IMB_colormanagement_transform_from_byte_threaded(float *float_buffer, unsigned char *byte_buffer, + int width, int height, int channels, + const char *from_colorspace, const char *to_colorspace) +{ + ColormanageProcessor *cm_processor; + if (from_colorspace == NULL || from_colorspace[0] == '\0') { + return; + } + if (STREQ(from_colorspace, to_colorspace)) { + /* If source and destination color spaces are identical, skip + * threading overhead and simply do nothing + */ + return; + } + cm_processor = IMB_colormanagement_colorspace_processor_new(from_colorspace, to_colorspace); + processor_transform_apply_threaded(byte_buffer, float_buffer, + width, height, channels, + cm_processor, true, true); + IMB_colormanagement_processor_free(cm_processor); } void IMB_colormanagement_transform_v4(float pixel[4], const char *from_colorspace, const char *to_colorspace) @@ -1974,12 +2072,14 @@ void IMB_colormanagement_buffer_make_display_space(float *buffer, unsigned char size_t float_buffer_size = ((size_t)width) * height * channels * sizeof(float); float *display_buffer_float = MEM_mallocN(float_buffer_size, "byte_buffer_make_display_space"); + /* TODO(sergey): Convert float directly to byte buffer. */ + memcpy(display_buffer_float, buffer, float_buffer_size); cm_processor = IMB_colormanagement_display_processor_new(view_settings, display_settings); - processor_transform_apply_threaded(display_buffer_float, width, height, channels, - cm_processor, true); + processor_transform_apply_threaded(NULL, display_buffer_float, width, height, channels, + cm_processor, true, false); IMB_buffer_byte_from_float(display_buffer, display_buffer_float, channels, dither, IB_PROFILE_SRGB, IB_PROFILE_SRGB, @@ -3100,6 +3200,25 @@ void IMB_colormanagement_processor_apply(ColormanageProcessor *cm_processor, flo } } +void IMB_colormanagement_processor_apply_byte(ColormanageProcessor *cm_processor, + unsigned char *buffer, + int width, int height, int channels) +{ + /* TODO(sergey): Would be nice to support arbitrary channels configurations, + * but for now it's not so important. + */ + BLI_assert(channels == 4); + float pixel[4]; + for (int y = 0; y < height; y++) { + for (int x = 0; x < width; x++) { + size_t offset = channels * (((size_t)y) * width + x); + rgba_uchar_to_float(pixel, buffer + offset); + IMB_colormanagement_processor_apply_v4(cm_processor, pixel); + rgba_float_to_uchar(buffer + offset, pixel); + } + } +} + void IMB_colormanagement_processor_free(ColormanageProcessor *cm_processor) { if (cm_processor->curve_mapping) diff --git a/source/blender/imbuf/intern/imbuf.h b/source/blender/imbuf/intern/imbuf.h index 897a149a45c..90dad70fa61 100644 --- a/source/blender/imbuf/intern/imbuf.h +++ b/source/blender/imbuf/intern/imbuf.h @@ -67,8 +67,6 @@ # define BIG_LONG SWAP_LONG #endif -typedef unsigned char uchar; - #define IMB_DPI_DEFAULT 72.0f #endif /* __IMBUF_H__ */ diff --git a/source/blender/makesdna/DNA_modifier_types.h b/source/blender/makesdna/DNA_modifier_types.h index f95533a88f9..823a7f0812f 100644 --- a/source/blender/makesdna/DNA_modifier_types.h +++ b/source/blender/makesdna/DNA_modifier_types.h @@ -86,6 +86,7 @@ typedef enum ModifierType { eModifierType_NormalEdit = 50, eModifierType_CorrectiveSmooth = 51, eModifierType_MeshSequenceCache = 52, + eModifierType_SurfaceDeform = 53, NUM_MODIFIER_TYPES } ModifierType; @@ -276,6 +277,7 @@ typedef struct MirrorModifierData { short axis DNA_DEPRECATED; /* deprecated, use flag instead */ short flag; float tolerance; + float uv_offset[2]; struct Object *mirror_ob; } MirrorModifierData; @@ -1570,6 +1572,46 @@ enum { MOD_MESHSEQ_READ_COLOR = (1 << 3), }; +typedef struct SDefBind { + unsigned int *vert_inds; + unsigned int numverts; + int mode; + float *vert_weights; + float normal_dist; + float influence; +} SDefBind; + +typedef struct SDefVert { + SDefBind *binds; + unsigned int numbinds; + char pad[4]; +} SDefVert; + +typedef struct SurfaceDeformModifierData { + ModifierData modifier; + + struct Object *target; /* bind target object */ + SDefVert *verts; /* vertex bind data */ + float falloff; + unsigned int numverts, numpoly; + int flags; + float mat[4][4]; +} SurfaceDeformModifierData; + +/* Surface Deform modifier flags */ +enum { + MOD_SDEF_BIND = (1 << 0), + MOD_SDEF_USES_LOOPTRI = (1 << 1), + MOD_SDEF_HAS_CONCAVE = (1 << 2), +}; + +/* Surface Deform vertex bind modes */ +enum { + MOD_SDEF_MODE_LOOPTRI = 0, + MOD_SDEF_MODE_NGON = 1, + MOD_SDEF_MODE_CENTROID = 2, +}; + #define MOD_MESHSEQ_READ_ALL \ (MOD_MESHSEQ_READ_VERT | MOD_MESHSEQ_READ_POLY | MOD_MESHSEQ_READ_UV | MOD_MESHSEQ_READ_COLOR) diff --git a/source/blender/makesdna/DNA_node_types.h b/source/blender/makesdna/DNA_node_types.h index fd601e55550..47677e50451 100644 --- a/source/blender/makesdna/DNA_node_types.h +++ b/source/blender/makesdna/DNA_node_types.h @@ -668,7 +668,8 @@ typedef struct NodeScriptDict { /* qdn: glare node */ typedef struct NodeGlare { char quality, type, iter; - char angle, pad_c1, size, pad[2]; + /* XXX angle is only kept for backward/forward compatibility, was used for two different things, see T50736. */ + char angle DNA_DEPRECATED, pad_c1, size, star_45, streaks; float colmod, mix, threshold, fade; float angle_ofs, pad_f1; } NodeGlare; diff --git a/source/blender/makesdna/DNA_object_force.h b/source/blender/makesdna/DNA_object_force.h index 59acefeffe4..ed14c4b9311 100644 --- a/source/blender/makesdna/DNA_object_force.h +++ b/source/blender/makesdna/DNA_object_force.h @@ -372,6 +372,7 @@ typedef struct SoftBody { #define PFIELD_DO_ROTATION (1<<15) #define PFIELD_GUIDE_PATH_WEIGHT (1<<16) /* apply curve weights */ #define PFIELD_SMOKE_DENSITY (1<<17) /* multiply smoke force by density */ +#define PFIELD_GRAVITATION (1<<18) /* used for (simple) force */ /* pd->falloff */ #define PFIELD_FALL_SPHERE 0 diff --git a/source/blender/makesdna/DNA_scene_types.h b/source/blender/makesdna/DNA_scene_types.h index 8ee15ef21a3..918d0f00040 100644 --- a/source/blender/makesdna/DNA_scene_types.h +++ b/source/blender/makesdna/DNA_scene_types.h @@ -1716,6 +1716,7 @@ typedef struct Scene { #define SCER_LOCK_FRAME_SELECTION (1<<1) /* timeline/keyframe jumping - only selected items (on by default) */ #define SCE_KEYS_NO_SELONLY (1<<2) +#define SCER_SHOW_SUBFRAME (1<<3) /* mode (int now) */ #define R_OSA 0x0001 diff --git a/source/blender/makesdna/DNA_userdef_types.h b/source/blender/makesdna/DNA_userdef_types.h index 0ad4482708f..5e7e7366e35 100644 --- a/source/blender/makesdna/DNA_userdef_types.h +++ b/source/blender/makesdna/DNA_userdef_types.h @@ -467,13 +467,18 @@ typedef struct UserDef { int audioformat; int audiochannels; - int scrollback; /* console scrollback limit */ - int dpi; /* range 48-128? */ - char node_margin; /* node insert offset (aka auto-offset) margin, but might be useful for later stuff as well */ + int scrollback; /* console scrollback limit */ + int dpi; /* range 48-128? */ + float ui_scale; /* interface scale */ + int pad1; + char node_margin; /* node insert offset (aka auto-offset) margin, but might be useful for later stuff as well */ char pad2; short transopts; short menuthreshold1, menuthreshold2; - + + /* startup template */ + char app_template[64]; + struct ListBase themes; struct ListBase uifonts; struct ListBase uistyles; @@ -497,7 +502,6 @@ typedef struct UserDef { int prefetchframes; float pad_rot_angle; /* control the rotation step of the view when PAD2, PAD4, PAD6&PAD8 is use */ short frameserverport; - short pad4; short obcenter_dia; short rvisize; /* rotating view icon size */ short rvibright; /* rotating view icon brightness */ @@ -509,6 +513,8 @@ typedef struct UserDef { char ipo_new; /* interpolation mode for newly added F-Curves */ char keyhandles_new; /* handle types for newly added keyframes */ char gpu_select_method; + char gpu_select_pick_deph; + char pad4; char view_frame_type; int view_frame_keyframes; /* number of keyframes to zoom around current frame */ diff --git a/source/blender/makesrna/RNA_access.h b/source/blender/makesrna/RNA_access.h index 66e6f30feeb..f9aaec69ce7 100644 --- a/source/blender/makesrna/RNA_access.h +++ b/source/blender/makesrna/RNA_access.h @@ -598,6 +598,7 @@ extern StructRNA RNA_StucciTexture; extern StructRNA RNA_SubsurfModifier; extern StructRNA RNA_SunLamp; extern StructRNA RNA_SurfaceCurve; +extern StructRNA RNA_SurfaceDeformModifier; extern StructRNA RNA_SurfaceModifier; extern StructRNA RNA_TexMapping; extern StructRNA RNA_Text; diff --git a/source/blender/makesrna/intern/rna_action.c b/source/blender/makesrna/intern/rna_action.c index 0c4c7ddac81..d398ce95a52 100644 --- a/source/blender/makesrna/intern/rna_action.c +++ b/source/blender/makesrna/intern/rna_action.c @@ -34,6 +34,8 @@ #include "BLI_utildefines.h" +#include "BLT_translation.h" + #include "BKE_action.h" #include "RNA_access.h" @@ -731,6 +733,7 @@ static void rna_def_action(BlenderRNA *brna) RNA_def_property_ui_text(prop, "ID Root Type", "Type of ID block that action can be used on - " "DO NOT CHANGE UNLESS YOU KNOW WHAT YOU ARE DOING"); + RNA_def_property_translation_context(prop, BLT_I18NCONTEXT_ID_ID); /* API calls */ RNA_api_action(srna); diff --git a/source/blender/makesrna/intern/rna_animation.c b/source/blender/makesrna/intern/rna_animation.c index 9adbf5f6b2e..f271bccd326 100644 --- a/source/blender/makesrna/intern/rna_animation.c +++ b/source/blender/makesrna/intern/rna_animation.c @@ -32,6 +32,8 @@ #include "BLI_utildefines.h" +#include "BLT_translation.h" + #include "MEM_guardedalloc.h" #include "RNA_access.h" @@ -742,6 +744,7 @@ static void rna_def_keyingset_path(BlenderRNA *brna) RNA_def_property_enum_default(prop, ID_OB); RNA_def_property_enum_funcs(prop, NULL, "rna_ksPath_id_type_set", NULL); RNA_def_property_ui_text(prop, "ID Type", "Type of ID-block that can be used"); + RNA_def_property_translation_context(prop, BLT_I18NCONTEXT_ID_ID); RNA_def_property_update(prop, NC_SCENE | ND_KEYINGSET | NA_EDITED, NULL); /* XXX: maybe a bit too noisy */ /* Group */ diff --git a/source/blender/makesrna/intern/rna_armature.c b/source/blender/makesrna/intern/rna_armature.c index 07d295c8bbc..891f5c43ca6 100644 --- a/source/blender/makesrna/intern/rna_armature.c +++ b/source/blender/makesrna/intern/rna_armature.c @@ -187,6 +187,9 @@ static void rna_Bone_select_update(Main *UNUSED(bmain), Scene *UNUSED(scene), Po } WM_main_add_notifier(NC_GEOM | ND_DATA, id); + + /* spaces that show animation data of the selected bone need updating */ + WM_main_add_notifier(NC_ANIMATION | ND_ANIMCHAN, id); } static char *rna_Bone_path(PointerRNA *ptr) diff --git a/source/blender/makesrna/intern/rna_define.c b/source/blender/makesrna/intern/rna_define.c index dc97d39052b..1d232d2df39 100644 --- a/source/blender/makesrna/intern/rna_define.c +++ b/source/blender/makesrna/intern/rna_define.c @@ -3157,8 +3157,9 @@ int rna_parameter_size(PropertyRNA *parm) StringPropertyRNA *sparm = (StringPropertyRNA *)parm; return sizeof(char) * sparm->maxlength; } - else + else { return sizeof(char *); + } case PROP_POINTER: { #ifdef RNA_RUNTIME diff --git a/source/blender/makesrna/intern/rna_fcurve.c b/source/blender/makesrna/intern/rna_fcurve.c index 4acdee490b8..bccc47aa95d 100644 --- a/source/blender/makesrna/intern/rna_fcurve.c +++ b/source/blender/makesrna/intern/rna_fcurve.c @@ -35,6 +35,8 @@ #include "BLI_math.h" +#include "BLT_translation.h" + #include "BKE_action.h" #include "RNA_access.h" @@ -1448,6 +1450,7 @@ static void rna_def_drivertarget(BlenderRNA *brna) RNA_def_property_enum_funcs(prop, NULL, "rna_DriverTarget_id_type_set", NULL); RNA_def_property_editable_func(prop, "rna_DriverTarget_id_type_editable"); RNA_def_property_ui_text(prop, "ID Type", "Type of ID-block that can be used"); + RNA_def_property_translation_context(prop, BLT_I18NCONTEXT_ID_ID); RNA_def_property_update(prop, 0, "rna_DriverTarget_update_data"); /* Target Properties - Property to Drive */ diff --git a/source/blender/makesrna/intern/rna_mesh_api.c b/source/blender/makesrna/intern/rna_mesh_api.c index ff9873fb3d1..9b0a25560f9 100644 --- a/source/blender/makesrna/intern/rna_mesh_api.c +++ b/source/blender/makesrna/intern/rna_mesh_api.c @@ -209,6 +209,11 @@ static void rna_Mesh_flip_normals(Mesh *mesh) DAG_id_tag_update(&mesh->id, 0); } +static void rna_Mesh_split_faces(Mesh *mesh, int free_loop_normals) +{ + BKE_mesh_split_faces(mesh, free_loop_normals != 0); +} + #else void RNA_api_mesh(StructRNA *srna) @@ -240,8 +245,10 @@ void RNA_api_mesh(StructRNA *srna) func = RNA_def_function(srna, "free_normals_split", "rna_Mesh_free_normals_split"); RNA_def_function_ui_description(func, "Free split vertex normals"); - func = RNA_def_function(srna, "split_faces", "BKE_mesh_split_faces"); + func = RNA_def_function(srna, "split_faces", "rna_Mesh_split_faces"); RNA_def_function_ui_description(func, "Split faces based on the edge angle"); + RNA_def_boolean(func, "free_loop_normals", 1, "Free Loop Notmals", + "Free loop normals custom data layer"); func = RNA_def_function(srna, "calc_tangents", "rna_Mesh_calc_tangents"); RNA_def_function_flag(func, FUNC_USE_REPORTS); diff --git a/source/blender/makesrna/intern/rna_modifier.c b/source/blender/makesrna/intern/rna_modifier.c index c4f0db38a16..1a8dd05a7b5 100644 --- a/source/blender/makesrna/intern/rna_modifier.c +++ b/source/blender/makesrna/intern/rna_modifier.c @@ -105,6 +105,7 @@ EnumPropertyItem rna_enum_object_modifier_type_items[] = { {eModifierType_Shrinkwrap, "SHRINKWRAP", ICON_MOD_SHRINKWRAP, "Shrinkwrap", ""}, {eModifierType_SimpleDeform, "SIMPLE_DEFORM", ICON_MOD_SIMPLEDEFORM, "Simple Deform", ""}, {eModifierType_Smooth, "SMOOTH", ICON_MOD_SMOOTH, "Smooth", ""}, + {eModifierType_SurfaceDeform, "SURFACE_DEFORM", ICON_MOD_MESHDEFORM, "Surface Deform", ""}, {eModifierType_Warp, "WARP", ICON_MOD_WARP, "Warp", ""}, {eModifierType_Wave, "WAVE", ICON_MOD_WAVE, "Wave", ""}, {0, "", 0, N_("Simulate"), ""}, @@ -408,6 +409,8 @@ static StructRNA *rna_Modifier_refine(struct PointerRNA *ptr) return &RNA_CorrectiveSmoothModifier; case eModifierType_MeshSequenceCache: return &RNA_MeshSequenceCacheModifier; + case eModifierType_SurfaceDeform: + return &RNA_SurfaceDeformModifier; /* Default */ case eModifierType_None: case eModifierType_ShapeKey: @@ -573,6 +576,7 @@ RNA_MOD_OBJECT_SET(MeshDeform, object, OB_MESH); RNA_MOD_OBJECT_SET(NormalEdit, target, OB_EMPTY); RNA_MOD_OBJECT_SET(Shrinkwrap, target, OB_MESH); RNA_MOD_OBJECT_SET(Shrinkwrap, auxTarget, OB_MESH); +RNA_MOD_OBJECT_SET(SurfaceDeform, target, OB_MESH); static void rna_HookModifier_object_set(PointerRNA *ptr, PointerRNA value) { @@ -1131,6 +1135,11 @@ static int rna_CorrectiveSmoothModifier_is_bind_get(PointerRNA *ptr) return (csmd->bind_coords != NULL); } +static int rna_SurfaceDeformModifier_is_bound_get(PointerRNA *ptr) +{ + return (((SurfaceDeformModifierData *)ptr->data)->verts != NULL); +} + static void rna_MeshSequenceCache_object_path_update(Main *bmain, Scene *scene, PointerRNA *ptr) { #ifdef WITH_ALEMBIC @@ -1517,6 +1526,20 @@ static void rna_def_modifier_mirror(BlenderRNA *brna) RNA_def_property_ui_text(prop, "Mirror V", "Mirror the V texture coordinate around the 0.5 point"); RNA_def_property_update(prop, 0, "rna_Modifier_update"); + prop = RNA_def_property(srna, "mirror_offset_u", PROP_FLOAT, PROP_FACTOR); + RNA_def_property_float_sdna(prop, NULL, "uv_offset[0]"); + RNA_def_property_range(prop, -1, 1); + RNA_def_property_ui_range(prop, -1, 1, 2, 4); + RNA_def_property_ui_text(prop, "U Offset", "Amount to offset mirrored UVs from the 0.5 point on the U axis"); + RNA_def_property_update(prop, 0, "rna_Modifier_update"); + + prop = RNA_def_property(srna, "mirror_offset_v", PROP_FLOAT, PROP_FACTOR); + RNA_def_property_float_sdna(prop, NULL, "uv_offset[1]"); + RNA_def_property_range(prop, -1, 1); + RNA_def_property_ui_range(prop, -1, 1, 2, 4); + RNA_def_property_ui_text(prop, "V Offset", "Amount to offset mirrored UVs from the 0.5 point on the V axis"); + RNA_def_property_update(prop, 0, "rna_Modifier_update"); + prop = RNA_def_property(srna, "merge_threshold", PROP_FLOAT, PROP_DISTANCE); RNA_def_property_float_sdna(prop, NULL, "tolerance"); RNA_def_property_range(prop, 0, FLT_MAX); @@ -4702,6 +4725,33 @@ static void rna_def_modifier_normaledit(BlenderRNA *brna) RNA_def_property_update(prop, 0, "rna_Modifier_update"); } +static void rna_def_modifier_surfacedeform(BlenderRNA *brna) +{ + StructRNA *srna; + PropertyRNA *prop; + + srna = RNA_def_struct(brna, "SurfaceDeformModifier", "Modifier"); + RNA_def_struct_ui_text(srna, "SurfaceDeform Modifier", ""); + RNA_def_struct_sdna(srna, "SurfaceDeformModifierData"); + RNA_def_struct_ui_icon(srna, ICON_MOD_MESHDEFORM); + + prop = RNA_def_property(srna, "target", PROP_POINTER, PROP_NONE); + RNA_def_property_ui_text(prop, "Target", "Mesh object to deform with"); + RNA_def_property_pointer_funcs(prop, NULL, "rna_SurfaceDeformModifier_target_set", NULL, "rna_Mesh_object_poll"); + RNA_def_property_flag(prop, PROP_EDITABLE | PROP_ID_SELF_CHECK); + RNA_def_property_update(prop, 0, "rna_Modifier_dependency_update"); + + prop = RNA_def_property(srna, "falloff", PROP_FLOAT, PROP_NONE); + RNA_def_property_range(prop, 2.0f, 16.0f); + RNA_def_property_ui_text(prop, "Interpolation falloff", "Controls how much nearby polygons influence deformation"); + RNA_def_property_update(prop, 0, "rna_Modifier_update"); + + prop = RNA_def_property(srna, "is_bound", PROP_BOOLEAN, PROP_NONE); + RNA_def_property_boolean_funcs(prop, "rna_SurfaceDeformModifier_is_bound_get", NULL); + RNA_def_property_ui_text(prop, "Bound", "Whether geometry has been bound to target mesh"); + RNA_def_property_clear_flag(prop, PROP_EDITABLE); +} + void RNA_def_modifier(BlenderRNA *brna) { StructRNA *srna; @@ -4819,6 +4869,7 @@ void RNA_def_modifier(BlenderRNA *brna) rna_def_modifier_datatransfer(brna); rna_def_modifier_normaledit(brna); rna_def_modifier_meshseqcache(brna); + rna_def_modifier_surfacedeform(brna); } #endif diff --git a/source/blender/makesrna/intern/rna_nodetree.c b/source/blender/makesrna/intern/rna_nodetree.c index 8afaf1f8fcb..c3477644979 100644 --- a/source/blender/makesrna/intern/rna_nodetree.c +++ b/source/blender/makesrna/intern/rna_nodetree.c @@ -3263,6 +3263,12 @@ static EnumPropertyItem node_script_mode_items[] = { {0, NULL, 0, NULL, NULL} }; +static EnumPropertyItem node_principled_distribution_items[] = { + { SHD_GLOSSY_GGX, "GGX", 0, "GGX", "" }, + { SHD_GLOSSY_MULTI_GGX, "MULTI_GGX", 0, "Multiscatter GGX", "" }, + { 0, NULL, 0, NULL, NULL } +}; + /* -- Common nodes ---------------------------------------------------------- */ static void def_group_input(StructRNA *srna) @@ -3336,6 +3342,7 @@ static void def_frame(StructRNA *srna) RNA_def_property_update(prop, NC_NODE | NA_EDITED, "rna_Node_update"); RNA_def_struct_sdna_from(srna, "NodeFrame", "storage"); + RNA_def_struct_translation_context(srna, BLT_I18NCONTEXT_ID_NODETREE); prop = RNA_def_property(srna, "shrink", PROP_BOOLEAN, PROP_NONE); RNA_def_property_boolean_sdna(prop, NULL, "flag", NODE_FRAME_SHRINK); @@ -4198,17 +4205,11 @@ static void def_glass(StructRNA *srna) static void def_principled(StructRNA *srna) { - static EnumPropertyItem prop_principled_distribution_items[] = { - { SHD_GLOSSY_GGX, "GGX", 0, "GGX", "" }, - { SHD_GLOSSY_MULTI_GGX, "MULTI_GGX", 0, "Multiscatter GGX", "" }, - { 0, NULL, 0, NULL, NULL } - }; - PropertyRNA *prop; prop = RNA_def_property(srna, "distribution", PROP_ENUM, PROP_NONE); RNA_def_property_enum_sdna(prop, NULL, "custom1"); - RNA_def_property_enum_items(prop, prop_principled_distribution_items); + RNA_def_property_enum_items(prop, node_principled_distribution_items); RNA_def_property_ui_text(prop, "Distribution", ""); RNA_def_property_update(prop, NC_NODE | NA_EDITED, "rna_ShaderNodePrincipled_update"); } @@ -5747,8 +5748,8 @@ static void def_cmp_glare(StructRNA *srna) RNA_def_property_update(prop, NC_NODE | NA_EDITED, "rna_Node_update"); prop = RNA_def_property(srna, "streaks", PROP_INT, PROP_NONE); - RNA_def_property_int_sdna(prop, NULL, "angle"); - RNA_def_property_range(prop, 2, 16); + RNA_def_property_int_sdna(prop, NULL, "streaks"); + RNA_def_property_range(prop, 1, 16); RNA_def_property_ui_text(prop, "Streaks", "Total number of streaks"); RNA_def_property_update(prop, NC_NODE | NA_EDITED, "rna_Node_update"); @@ -5765,7 +5766,7 @@ static void def_cmp_glare(StructRNA *srna) RNA_def_property_update(prop, NC_NODE | NA_EDITED, "rna_Node_update"); prop = RNA_def_property(srna, "use_rotate_45", PROP_BOOLEAN, PROP_NONE); - RNA_def_property_boolean_sdna(prop, NULL, "angle", 0); + RNA_def_property_boolean_sdna(prop, NULL, "star_45", 0); RNA_def_property_ui_text(prop, "Rotate 45", "Simple star filter: add 45 degree rotation offset"); RNA_def_property_update(prop, NC_NODE | NA_EDITED, "rna_Node_update"); diff --git a/source/blender/makesrna/intern/rna_object.c b/source/blender/makesrna/intern/rna_object.c index 0cffba47f16..b3c166a6810 100644 --- a/source/blender/makesrna/intern/rna_object.c +++ b/source/blender/makesrna/intern/rna_object.c @@ -1321,8 +1321,12 @@ static void rna_Object_active_constraint_set(PointerRNA *ptr, PointerRNA value) static bConstraint *rna_Object_constraints_new(Object *object, int type) { + bConstraint *new_con = BKE_constraint_add_for_object(object, NULL, type); + + ED_object_constraint_tag_update(object, new_con); WM_main_add_notifier(NC_OBJECT | ND_CONSTRAINT | NA_ADDED, object); - return BKE_constraint_add_for_object(object, NULL, type); + + return new_con; } static void rna_Object_constraints_remove(Object *object, ReportList *reports, PointerRNA *con_ptr) diff --git a/source/blender/makesrna/intern/rna_object_force.c b/source/blender/makesrna/intern/rna_object_force.c index 1d89f7535c4..514fca1b011 100644 --- a/source/blender/makesrna/intern/rna_object_force.c +++ b/source/blender/makesrna/intern/rna_object_force.c @@ -1275,7 +1275,7 @@ static void rna_def_field(BlenderRNA *brna) prop = RNA_def_property(srna, "falloff_power", PROP_FLOAT, PROP_NONE); RNA_def_property_float_sdna(prop, NULL, "f_power"); RNA_def_property_range(prop, 0.0f, 10.0f); - RNA_def_property_ui_text(prop, "Falloff Power", "Falloff power (real gravitational falloff = 2)"); + RNA_def_property_ui_text(prop, "Falloff Power", ""); RNA_def_property_update(prop, 0, "rna_FieldSettings_update"); prop = RNA_def_property(srna, "distance_min", PROP_FLOAT, PROP_NONE); @@ -1394,6 +1394,11 @@ static void rna_def_field(BlenderRNA *brna) RNA_def_property_boolean_sdna(prop, NULL, "flag", PFIELD_SMOKE_DENSITY); RNA_def_property_ui_text(prop, "Apply Density", "Adjust force strength based on smoke density"); RNA_def_property_update(prop, 0, "rna_FieldSettings_update"); + prop = RNA_def_property(srna, "use_gravity_falloff", PROP_BOOLEAN, PROP_NONE); + RNA_def_property_boolean_sdna(prop, NULL, "flag", PFIELD_GRAVITATION); + RNA_def_property_ui_text(prop, "Gravity Falloff", "Multiply force by 1/distance²"); + RNA_def_property_update(prop, 0, "rna_FieldSettings_update"); + /* Pointer */ diff --git a/source/blender/makesrna/intern/rna_pose.c b/source/blender/makesrna/intern/rna_pose.c index 28ce63a61bd..8d161466d56 100644 --- a/source/blender/makesrna/intern/rna_pose.c +++ b/source/blender/makesrna/intern/rna_pose.c @@ -524,12 +524,15 @@ static void rna_PoseChannel_active_constraint_set(PointerRNA *ptr, PointerRNA va BKE_constraints_active_set(&pchan->constraints, (bConstraint *)value.data); } -static bConstraint *rna_PoseChannel_constraints_new(bPoseChannel *pchan, int type) +static bConstraint *rna_PoseChannel_constraints_new(ID *id, bPoseChannel *pchan, Main *main, int type) { - /*WM_main_add_notifier(NC_OBJECT|ND_CONSTRAINT|NA_ADDED, object); */ - /* TODO, pass object also */ - /* TODO, new pose bones don't have updated draw flags */ - return BKE_constraint_add_for_pose(NULL, pchan, NULL, type); + Object *ob = (Object *)id; + bConstraint *new_con = BKE_constraint_add_for_pose(ob, pchan, NULL, type); + + ED_object_constraint_dependency_tag_update(main, ob, new_con); + WM_main_add_notifier(NC_OBJECT | ND_CONSTRAINT | NA_ADDED, id); + + return new_con; } static void rna_PoseChannel_constraints_remove(ID *id, bPoseChannel *pchan, ReportList *reports, PointerRNA *con_ptr) @@ -764,6 +767,7 @@ static void rna_def_pose_channel_constraints(BlenderRNA *brna, PropertyRNA *cpro /* Constraint collection */ func = RNA_def_function(srna, "new", "rna_PoseChannel_constraints_new"); RNA_def_function_ui_description(func, "Add a constraint to this object"); + RNA_def_function_flag(func, FUNC_USE_MAIN | FUNC_USE_SELF_ID); /* ID and Main needed for refresh */ /* return type */ parm = RNA_def_pointer(func, "constraint", "Constraint", "", "New constraint"); RNA_def_function_return(func, parm); diff --git a/source/blender/makesrna/intern/rna_scene.c b/source/blender/makesrna/intern/rna_scene.c index 1166fb89a0a..899466f05df 100644 --- a/source/blender/makesrna/intern/rna_scene.c +++ b/source/blender/makesrna/intern/rna_scene.c @@ -411,7 +411,7 @@ EnumPropertyItem rna_enum_gpencil_interpolation_mode_items[] = { /* interpolation */ {0, "", 0, N_("Interpolation"), "Standard transitions between keyframes"}, {GP_IPO_LINEAR, "LINEAR", ICON_IPO_LINEAR, "Linear", "Straight-line interpolation between A and B (i.e. no ease in/out)"}, - {GP_IPO_CURVEMAP, "CUSTOM", ICON_IPO_BEZIER, "Custom", "Custom interpolation defined using a curvemap"}, + {GP_IPO_CURVEMAP, "CUSTOM", ICON_IPO_BEZIER, "Custom", "Custom interpolation defined using a curve map"}, /* easing */ {0, "", 0, N_("Easing (by strength)"), "Predefined inertial transitions, useful for motion graphics (from least to most ''dramatic'')"}, @@ -792,6 +792,21 @@ static void rna_Scene_frame_current_set(PointerRNA *ptr, int value) data->r.cfra = value; } +static float rna_Scene_frame_float_get(PointerRNA *ptr) +{ + Scene *data = (Scene *)ptr->data; + return (float)data->r.cfra + data->r.subframe; +} + +static void rna_Scene_frame_float_set(PointerRNA *ptr, float value) +{ + Scene *data = (Scene *)ptr->data; + /* if negative frames aren't allowed, then we can't use them */ + FRAMENUMBER_MIN_CLAMP(value); + data->r.cfra = (int)value; + data->r.subframe = value - data->r.cfra; +} + static float rna_Scene_frame_current_final_get(PointerRNA *ptr) { Scene *scene = (Scene *)ptr->data; @@ -872,6 +887,12 @@ static void rna_Scene_preview_range_end_frame_set(PointerRNA *ptr, int value) data->r.pefra = value; } +static void rna_Scene_show_subframe_update(Main *UNUSED(bmain), Scene *UNUSED(current_scene), PointerRNA *ptr) +{ + Scene *scene = (Scene *)ptr->id.data; + scene->r.subframe = 0.0f; +} + static void rna_Scene_frame_update(Main *bmain, Scene *UNUSED(current_scene), PointerRNA *ptr) { Scene *scene = (Scene *)ptr->id.data; @@ -5369,7 +5390,7 @@ static void rna_def_scene_image_format_data(BlenderRNA *brna) prop = RNA_def_property(srna, "jpeg2k_codec", PROP_ENUM, PROP_NONE); RNA_def_property_enum_sdna(prop, NULL, "jp2_codec"); RNA_def_property_enum_items(prop, jp2_codec_items); - RNA_def_property_ui_text(prop, "Codec", "Codec settings for Jpek2000"); + RNA_def_property_ui_text(prop, "Codec", "Codec settings for Jpeg2000"); RNA_def_property_update(prop, NC_SCENE | ND_RENDER_OPTIONS, NULL); #endif @@ -7081,8 +7102,19 @@ void RNA_def_scene(BlenderRNA *brna) prop = RNA_def_property(srna, "frame_subframe", PROP_FLOAT, PROP_TIME); RNA_def_property_float_sdna(prop, NULL, "r.subframe"); RNA_def_property_ui_text(prop, "Current Sub-Frame", ""); - RNA_def_property_clear_flag(prop, PROP_ANIMATABLE | PROP_EDITABLE); - + RNA_def_property_clear_flag(prop, PROP_ANIMATABLE); + RNA_def_property_range(prop, 0.0f, 1.0f); + RNA_def_property_ui_range(prop, 0.0f, 1.0f, 0.01, 2); + RNA_def_property_update(prop, NC_SCENE | ND_FRAME, "rna_Scene_frame_update"); + + prop = RNA_def_property(srna, "frame_float", PROP_FLOAT, PROP_TIME); + RNA_def_property_ui_text(prop, "Current Sub-Frame", ""); + RNA_def_property_clear_flag(prop, PROP_ANIMATABLE); + RNA_def_property_range(prop, MINAFRAME, MAXFRAME); + RNA_def_property_ui_range(prop, MINAFRAME, MAXFRAME, 0.1, 2); + RNA_def_property_float_funcs(prop, "rna_Scene_frame_float_get", "rna_Scene_frame_float_set", NULL); + RNA_def_property_update(prop, NC_SCENE | ND_FRAME, "rna_Scene_frame_update"); + prop = RNA_def_property(srna, "frame_start", PROP_INT, PROP_TIME); RNA_def_property_clear_flag(prop, PROP_ANIMATABLE); RNA_def_property_int_sdna(prop, NULL, "r.sfra"); @@ -7147,7 +7179,15 @@ void RNA_def_scene(BlenderRNA *brna) RNA_def_property_int_funcs(prop, NULL, "rna_Scene_preview_range_end_frame_set", NULL); RNA_def_property_ui_text(prop, "Preview Range End Frame", "Alternative end frame for UI playback"); RNA_def_property_update(prop, NC_SCENE | ND_FRAME, NULL); - + + /* Subframe for moblur debug. */ + prop = RNA_def_property(srna, "show_subframe", PROP_BOOLEAN, PROP_NONE); + RNA_def_property_clear_flag(prop, PROP_ANIMATABLE); + RNA_def_property_boolean_sdna(prop, NULL, "r.flag", SCER_SHOW_SUBFRAME); + RNA_def_property_ui_text(prop, "Show Subframe", + "Show current scene subframe and allow set it using interface tools"); + RNA_def_property_update(prop, NC_SCENE | ND_FRAME, "rna_Scene_show_subframe_update"); + /* Timeline / Time Navigation settings */ prop = RNA_def_property(srna, "show_keys_from_selected_only", PROP_BOOLEAN, PROP_NONE); RNA_def_property_boolean_negative_sdna(prop, NULL, "flag", SCE_KEYS_NO_SELONLY); diff --git a/source/blender/makesrna/intern/rna_scene_api.c b/source/blender/makesrna/intern/rna_scene_api.c index fe781a309a4..f3146c9e6a8 100644 --- a/source/blender/makesrna/intern/rna_scene_api.c +++ b/source/blender/makesrna/intern/rna_scene_api.c @@ -243,9 +243,9 @@ static void rna_Scene_alembic_export( .use_subdiv_schema = use_subdiv_schema, .compression_type = compression_type, .packuv = packuv, - .triangulate = triangulate, - .quad_method = quad_method, - .ngon_method = ngon_method, + .triangulate = triangulate, + .quad_method = quad_method, + .ngon_method = ngon_method, .global_scale = scale, }; @@ -263,34 +263,55 @@ static void rna_Scene_alembic_export( /* don't remove this, as COLLADA exporting cannot be done through operators in render() callback. */ #include "../../collada/collada.h" +/* Note: This definition must match to the generated function call */ static void rna_Scene_collada_export( Scene *scene, - const char *filepath, + const char *filepath, int apply_modifiers, - int export_mesh_type, + int export_mesh_type, int selected, int include_children, int include_armatures, int include_shapekeys, int deform_bones_only, - int active_uv_only, int include_uv_textures, int include_material_textures, int use_texture_copies, - - int use_ngons, + int triangulate, int use_object_instantiation, int use_blender_profile, int sort_by_name, + int export_transformation_type, int open_sim, - int export_transformation_type) + int keep_bind_info) { - collada_export(scene, filepath, apply_modifiers, export_mesh_type, selected, - include_children, include_armatures, include_shapekeys, deform_bones_only, - active_uv_only, include_uv_textures, include_material_textures, - use_texture_copies, use_ngons, use_object_instantiation, use_blender_profile, sort_by_name, export_transformation_type, open_sim); + collada_export(scene, + filepath, + + apply_modifiers, + export_mesh_type, + + selected, + include_children, + include_armatures, + include_shapekeys, + deform_bones_only, + + active_uv_only, + include_uv_textures, + include_material_textures, + use_texture_copies, + + triangulate, + use_object_instantiation, + use_blender_profile, + sort_by_name, + + export_transformation_type, + open_sim, + keep_bind_info); } #endif @@ -354,30 +375,56 @@ void RNA_api_scene(StructRNA *srna) parm = RNA_def_string(func, "filepath", NULL, FILE_MAX, "File Path", "File path to write Collada file"); RNA_def_parameter_flags(parm, 0, PARM_REQUIRED); RNA_def_property_subtype(parm, PROP_FILEPATH); /* allow non utf8 */ - RNA_def_boolean(func, "apply_modifiers", 0, "Apply Modifiers", "Apply modifiers"); + + RNA_def_boolean(func, "apply_modifiers", false, + "Apply Modifiers", "Apply modifiers to exported mesh (non destructive))"); + RNA_def_int(func, "export_mesh_type", 0, INT_MIN, INT_MAX, "Resolution", "Modifier resolution for export", INT_MIN, INT_MAX); - RNA_def_boolean(func, "selected", 0, "Selection Only", "Export only selected elements"); - RNA_def_boolean(func, "include_children", 0, "Include Children", "Export all children of selected objects (even if not selected)"); - RNA_def_boolean(func, "include_armatures", 0, "Include Armatures", "Export related armatures (even if not selected)"); - RNA_def_boolean(func, "include_shapekeys", 0, "Include Shape Keys", "Export all Shape Keys from Mesh Objects"); - RNA_def_boolean(func, "deform_bones_only", 0, "Deform Bones only", "Only export deforming bones with armatures"); - - RNA_def_boolean(func, "active_uv_only", 0, "Active UV Layer only", "Export only the active UV Layer"); - RNA_def_boolean(func, "include_uv_textures", 0, "Include UV Textures", "Export textures assigned to the object UV maps"); - RNA_def_boolean(func, "include_material_textures", 0, "Include Material Textures", "Export textures assigned to the object Materials"); - RNA_def_boolean(func, "use_texture_copies", 0, "copy", "Copy textures to same folder where the .dae file is exported"); - - RNA_def_boolean(func, "use_ngons", 1, "Use NGons", "Keep NGons in Export"); - RNA_def_boolean(func, "use_object_instantiation", 1, "Use Object Instances", "Instantiate multiple Objects from same Data"); - RNA_def_boolean(func, "use_blender_profile", 1, "Use Blender Profile", "Export additional Blender specific information (for material, shaders, bones, etc.)"); - RNA_def_boolean(func, "sort_by_name", 0, "Sort by Object name", "Sort exported data by Object name"); - RNA_def_boolean(func, "open_sim", 0, "Export for SL/OpenSim", "Compatibility mode for SL, OpenSim and similar online worlds"); + + RNA_def_boolean(func, "selected", false, "Selection Only", "Export only selected elements"); + + RNA_def_boolean(func, "include_children", false, + "Include Children", "Export all children of selected objects (even if not selected)"); + + RNA_def_boolean(func, "include_armatures", false, + "Include Armatures", "Export related armatures (even if not selected)"); + + RNA_def_boolean(func, "include_shapekeys", true, "Include Shape Keys", "Export all Shape Keys from Mesh Objects"); + + RNA_def_boolean(func, "deform_bones_only", false, + "Deform Bones only", "Only export deforming bones with armatures"); + + RNA_def_boolean(func, "active_uv_only", false, "Only Selected UV Map", "Export only the selected UV Map"); + + RNA_def_boolean(func, "include_uv_textures", false, + "Include UV Textures", "Export textures assigned to the object UV Maps"); + + RNA_def_boolean(func, "include_material_textures", false, + "Include Material Textures", "Export textures assigned to the object Materials"); + + RNA_def_boolean(func, "use_texture_copies", true, + "Copy", "Copy textures to same folder where the .dae file is exported"); + + RNA_def_boolean(func, "triangulate", true, "Triangulate", "Export Polygons (Quads & NGons) as Triangles"); + + RNA_def_boolean(func, "use_object_instantiation", true, + "Use Object Instances", "Instantiate multiple Objects from same Data"); + + RNA_def_boolean(func, "use_blender_profile", true, "Use Blender Profile", + "Export additional Blender specific information (for material, shaders, bones, etc.)"); + + RNA_def_boolean(func, "sort_by_name", false, "Sort by Object name", "Sort exported data by Object name"); RNA_def_int(func, "export_transformation_type", 0, INT_MIN, INT_MAX, - "Transformation", "Transformation type for translation, scale and rotation", INT_MIN, INT_MAX); + "Transform", "Transformation type for translation, scale and rotation", INT_MIN, INT_MAX); + + RNA_def_boolean(func, "open_sim", false, + "Export to SL/OpenSim", "Compatibility mode for SL, OpenSim and other compatible online worlds"); + + RNA_def_boolean(func, "keep_bind_info", false, "Keep Bind Info", + "Store bind pose information in custom bone properties for later use during Collada export"); - RNA_def_function_ui_description(func, "Export to collada file"); #endif #ifdef WITH_ALEMBIC diff --git a/source/blender/makesrna/intern/rna_smoke.c b/source/blender/makesrna/intern/rna_smoke.c index 6db370fc152..c12937bd2bf 100644 --- a/source/blender/makesrna/intern/rna_smoke.c +++ b/source/blender/makesrna/intern/rna_smoke.c @@ -832,14 +832,14 @@ static void rna_def_smoke_domain_settings(BlenderRNA *brna) {FLUID_FIELD_COLOR_R, "COLOR_R", 0, "Red", "Red component of the color field"}, {FLUID_FIELD_COLOR_G, "COLOR_G", 0, "Green", "Green component of the color field"}, {FLUID_FIELD_COLOR_B, "COLOR_B", 0, "Blue", "Blue component of the color field"}, - {FLUID_FIELD_DENSITY, "DENSITY", 0, "Density", "Quantity of soot in the fluid"}, + {FLUID_FIELD_DENSITY, "DENSITY", 0, "Density", "Quantity of soot in the fluid"}, {FLUID_FIELD_FLAME, "FLAME", 0, "Flame", "Flame field"}, {FLUID_FIELD_FUEL, "FUEL", 0, "Fuel", "Fuel field"}, {FLUID_FIELD_HEAT, "HEAT", 0, "Heat", "Temperature of the fluid"}, {FLUID_FIELD_VELOCITY_X, "VELOCITY_X", 0, "X Velocity", "X component of the velocity field"}, {FLUID_FIELD_VELOCITY_Y, "VELOCITY_Y", 0, "Y Velocity", "Y component of the velocity field"}, {FLUID_FIELD_VELOCITY_Z, "VELOCITY_Z", 0, "Z Velocity", "Z component of the velocity field"}, - {0, NULL, 0, NULL, NULL} + {0, NULL, 0, NULL, NULL} }; prop = RNA_def_property(srna, "coba_field", PROP_ENUM, PROP_NONE); diff --git a/source/blender/makesrna/intern/rna_userdef.c b/source/blender/makesrna/intern/rna_userdef.c index e68e67586e9..4d70b87843a 100644 --- a/source/blender/makesrna/intern/rna_userdef.c +++ b/source/blender/makesrna/intern/rna_userdef.c @@ -138,23 +138,11 @@ static void rna_userdef_update(Main *UNUSED(bmain), Scene *UNUSED(scene), Pointe } /* also used by buffer swap switching */ -static void rna_userdef_dpi_update(Main *UNUSED(bmain), Scene *UNUSED(scene), PointerRNA *UNUSED(ptr)) +static void rna_userdef_dpi_update(Main *bmain, Scene *UNUSED(scene), PointerRNA *UNUSED(ptr)) { /* font's are stored at each DPI level, without this we can easy load 100's of fonts */ BLF_cache_clear(); - BKE_blender_userdef_refresh(); - WM_main_add_notifier(NC_WINDOW, NULL); /* full redraw */ - WM_main_add_notifier(NC_SCREEN | NA_EDITED, NULL); /* refresh region sizes */ -} - -static void rna_userdef_virtual_pixel_update(Main *bmain, Scene *UNUSED(scene), PointerRNA *UNUSED(ptr)) -{ - /* font's are stored at each DPI level, without this we can easy load 100's of fonts */ - BLF_cache_clear(); - - BKE_blender_userdef_refresh(); - /* force setting drawable again */ wmWindowManager *wm = bmain->wm.first; if (wm) { @@ -3325,6 +3313,12 @@ static void rna_def_userdef_view(BlenderRNA *brna) RNA_def_struct_ui_text(srna, "View & Controls", "Preferences related to viewing data"); /* View */ + prop = RNA_def_property(srna, "ui_scale", PROP_FLOAT, PROP_FACTOR); + RNA_def_property_ui_text(prop, "UI Scale", "Changes the size of the fonts and buttons in the interface"); + RNA_def_property_range(prop, 0.25f, 4.0f); + RNA_def_property_ui_range(prop, 0.5f, 2.0f, 1, 1); + RNA_def_property_float_default(prop, 1.0f); + RNA_def_property_update(prop, 0, "rna_userdef_dpi_update"); /* display */ prop = RNA_def_property(srna, "show_tooltips", PROP_BOOLEAN, PROP_NONE); @@ -3916,12 +3910,6 @@ static void rna_def_userdef_system(BlenderRNA *brna) {0, NULL, 0, NULL, NULL} }; - static EnumPropertyItem virtual_pixel_mode_items[] = { - {VIRTUAL_PIXEL_NATIVE, "NATIVE", 0, "Native", "Use native pixel size of the display"}, - {VIRTUAL_PIXEL_DOUBLE, "DOUBLE", 0, "Double", "Use double the native pixel size of the display"}, - {0, NULL, 0, NULL, NULL} - }; - srna = RNA_def_struct(brna, "UserPreferencesSystem", NULL); RNA_def_struct_sdna(srna, "UserDef"); RNA_def_struct_nested(brna, srna, "UserPreferences"); @@ -3936,16 +3924,8 @@ static void rna_def_userdef_system(BlenderRNA *brna) RNA_def_property_update(prop, NC_WINDOW, "rna_userdef_language_update"); prop = RNA_def_property(srna, "dpi", PROP_INT, PROP_NONE); - RNA_def_property_int_sdna(prop, NULL, "dpi"); - RNA_def_property_range(prop, 48, 144); + RNA_def_property_clear_flag(prop, PROP_EDITABLE); RNA_def_property_ui_text(prop, "DPI", "Font size and resolution for display"); - RNA_def_property_update(prop, 0, "rna_userdef_dpi_update"); - - prop = RNA_def_property(srna, "virtual_pixel_mode", PROP_ENUM, PROP_NONE); - RNA_def_property_enum_sdna(prop, NULL, "virtual_pixel"); - RNA_def_property_enum_items(prop, virtual_pixel_mode_items); - RNA_def_property_ui_text(prop, "Virtual Pixel Mode", "Modify the pixel size for hi-res devices"); - RNA_def_property_update(prop, 0, "rna_userdef_virtual_pixel_update"); prop = RNA_def_property(srna, "pixel_size", PROP_FLOAT, PROP_NONE); RNA_def_property_clear_flag(prop, PROP_EDITABLE); @@ -4180,6 +4160,10 @@ static void rna_def_userdef_system(BlenderRNA *brna) RNA_def_property_ui_text(prop, "Selection Method", "Use OpenGL occlusion queries or selection render mode to accelerate selection"); + prop = RNA_def_property(srna, "use_select_pick_depth", PROP_BOOLEAN, PROP_NONE); + RNA_def_property_boolean_sdna(prop, NULL, "gpu_select_pick_deph", 1); + RNA_def_property_ui_text(prop, "OpenGL Depth Picking", "Use the depth buffer for picking 3D View selection"); + /* Full scene anti-aliasing */ prop = RNA_def_property(srna, "multi_sample", PROP_ENUM, PROP_NONE); RNA_def_property_enum_bitflag_sdna(prop, NULL, "ogl_multisamples"); @@ -4674,6 +4658,11 @@ void RNA_def_userdef(BlenderRNA *brna) "Active section of the user preferences shown in the user interface"); RNA_def_property_update(prop, 0, "rna_userdef_update"); + /* don't expose this directly via the UI, modify via an operator */ + prop = RNA_def_property(srna, "app_template", PROP_STRING, PROP_NONE); + RNA_def_property_string_sdna(prop, NULL, "app_template"); + RNA_def_property_ui_text(prop, "Application Template", ""); + prop = RNA_def_property(srna, "themes", PROP_COLLECTION, PROP_NONE); RNA_def_property_collection_sdna(prop, NULL, "themes", NULL); RNA_def_property_struct_type(prop, "Theme"); diff --git a/source/blender/makesrna/intern/rna_wm_api.c b/source/blender/makesrna/intern/rna_wm_api.c index 1f23ab938fb..677ea92aea1 100644 --- a/source/blender/makesrna/intern/rna_wm_api.c +++ b/source/blender/makesrna/intern/rna_wm_api.c @@ -87,6 +87,11 @@ static void rna_Operator_report(wmOperator *op, int type, const char *msg) BKE_report(op->reports, type, msg); } +static int rna_Operator_is_repeat(wmOperator *op, bContext *C) +{ + return WM_operator_is_repeat(C, op); +} + /* since event isn't needed... */ static void rna_Operator_enum_search_invoke(bContext *C, wmOperator *op) { @@ -521,6 +526,12 @@ void RNA_api_operator(StructRNA *srna) parm = RNA_def_string(func, "message", NULL, 0, "Report Message", ""); RNA_def_parameter_flags(parm, 0, PARM_REQUIRED); + /* utility, not for registering */ + func = RNA_def_function(srna, "is_repeat", "rna_Operator_is_repeat"); + RNA_def_function_flag(func, FUNC_USE_CONTEXT); + /* return */ + parm = RNA_def_boolean(func, "result", 0, "result", ""); + RNA_def_function_return(func, parm); /* Registration */ diff --git a/source/blender/modifiers/CMakeLists.txt b/source/blender/modifiers/CMakeLists.txt index bacfc177432..ad2b862141c 100644 --- a/source/blender/modifiers/CMakeLists.txt +++ b/source/blender/modifiers/CMakeLists.txt @@ -93,6 +93,7 @@ set(SRC intern/MOD_solidify.c intern/MOD_subsurf.c intern/MOD_surface.c + intern/MOD_surfacedeform.c intern/MOD_triangulate.c intern/MOD_util.c intern/MOD_uvwarp.c diff --git a/source/blender/modifiers/MOD_modifiertypes.h b/source/blender/modifiers/MOD_modifiertypes.h index 4c881445893..bf121af2bd1 100644 --- a/source/blender/modifiers/MOD_modifiertypes.h +++ b/source/blender/modifiers/MOD_modifiertypes.h @@ -85,6 +85,7 @@ extern ModifierTypeInfo modifierType_DataTransfer; extern ModifierTypeInfo modifierType_NormalEdit; extern ModifierTypeInfo modifierType_CorrectiveSmooth; extern ModifierTypeInfo modifierType_MeshSequenceCache; +extern ModifierTypeInfo modifierType_SurfaceDeform; /* MOD_util.c */ void modifier_type_init(ModifierTypeInfo *types[]); diff --git a/source/blender/modifiers/intern/MOD_boolean.c b/source/blender/modifiers/intern/MOD_boolean.c index f828bc68857..f86d8b99f3c 100644 --- a/source/blender/modifiers/intern/MOD_boolean.c +++ b/source/blender/modifiers/intern/MOD_boolean.c @@ -319,6 +319,7 @@ static DerivedMesh *applyModifier_bmesh( use_separate, use_dissolve, use_island_connect, + false, bmd->operation, bmd->double_threshold); diff --git a/source/blender/modifiers/intern/MOD_displace.c b/source/blender/modifiers/intern/MOD_displace.c index 08d118a2026..18f60bab490 100644 --- a/source/blender/modifiers/intern/MOD_displace.c +++ b/source/blender/modifiers/intern/MOD_displace.c @@ -38,10 +38,12 @@ #include "BLI_utildefines.h" #include "BLI_math.h" +#include "BLI_task.h" #include "BKE_cdderivedmesh.h" #include "BKE_library.h" #include "BKE_library_query.h" +#include "BKE_image.h" #include "BKE_mesh.h" #include "BKE_modifier.h" #include "BKE_texture.h" @@ -200,21 +202,130 @@ static void updateDepsgraph(ModifierData *md, } } +typedef struct DisplaceUserdata { + /*const*/ DisplaceModifierData *dmd; + struct ImagePool *pool; + MDeformVert *dvert; + float weight; + int defgrp_index; + int direction; + bool use_global_direction; + float (*tex_co)[3]; + float (*vertexCos)[3]; + float local_mat[4][4]; + MVert *mvert; + float (*vert_clnors)[3]; +} DisplaceUserdata; + +static void displaceModifier_do_task(void *userdata, const int iter) +{ + DisplaceUserdata *data = (DisplaceUserdata *)userdata; + DisplaceModifierData *dmd = data->dmd; + MDeformVert *dvert = data->dvert; + float weight = data->weight; + int defgrp_index = data->defgrp_index; + int direction = data->direction; + bool use_global_direction = data->use_global_direction; + float (*tex_co)[3] = data->tex_co; + float (*vertexCos)[3] = data->vertexCos; + MVert *mvert = data->mvert; + float (*vert_clnors)[3] = data->vert_clnors; + + const float delta_fixed = 1.0f - dmd->midlevel; /* when no texture is used, we fallback to white */ + + TexResult texres; + float strength = dmd->strength; + float delta; + float local_vec[3]; + + if (dvert) { + weight = defvert_find_weight(dvert + iter, defgrp_index); + if (weight == 0.0f) { + return; + } + } + + if (dmd->texture) { + texres.nor = NULL; + BKE_texture_get_value_ex(dmd->modifier.scene, dmd->texture, tex_co[iter], &texres, data->pool, false); + delta = texres.tin - dmd->midlevel; + } + else { + delta = delta_fixed; /* (1.0f - dmd->midlevel) */ /* never changes */ + } + + if (dvert) { + strength *= weight; + } + + delta *= strength; + CLAMP(delta, -10000, 10000); + + switch (direction) { + case MOD_DISP_DIR_X: + if (use_global_direction) { + vertexCos[iter][0] += delta * data->local_mat[0][0]; + vertexCos[iter][1] += delta * data->local_mat[1][0]; + vertexCos[iter][2] += delta * data->local_mat[2][0]; + } + else { + vertexCos[iter][0] += delta; + } + break; + case MOD_DISP_DIR_Y: + if (use_global_direction) { + vertexCos[iter][0] += delta * data->local_mat[0][1]; + vertexCos[iter][1] += delta * data->local_mat[1][1]; + vertexCos[iter][2] += delta * data->local_mat[2][1]; + } + else { + vertexCos[iter][1] += delta; + } + break; + case MOD_DISP_DIR_Z: + if (use_global_direction) { + vertexCos[iter][0] += delta * data->local_mat[0][2]; + vertexCos[iter][1] += delta * data->local_mat[1][2]; + vertexCos[iter][2] += delta * data->local_mat[2][2]; + } + else { + vertexCos[iter][2] += delta; + } + break; + case MOD_DISP_DIR_RGB_XYZ: + local_vec[0] = texres.tr - dmd->midlevel; + local_vec[1] = texres.tg - dmd->midlevel; + local_vec[2] = texres.tb - dmd->midlevel; + if (use_global_direction) { + mul_transposed_mat3_m4_v3(data->local_mat, local_vec); + } + mul_v3_fl(local_vec, strength); + add_v3_v3(vertexCos[iter], local_vec); + break; + case MOD_DISP_DIR_NOR: + vertexCos[iter][0] += delta * (mvert[iter].no[0] / 32767.0f); + vertexCos[iter][1] += delta * (mvert[iter].no[1] / 32767.0f); + vertexCos[iter][2] += delta * (mvert[iter].no[2] / 32767.0f); + break; + case MOD_DISP_DIR_CLNOR: + madd_v3_v3fl(vertexCos[iter], vert_clnors[iter], delta); + break; + } +} + /* dm must be a CDDerivedMesh */ static void displaceModifier_do( DisplaceModifierData *dmd, Object *ob, DerivedMesh *dm, float (*vertexCos)[3], int numVerts) { - int i; MVert *mvert; MDeformVert *dvert; int direction = dmd->direction; int defgrp_index; float (*tex_co)[3]; float weight = 1.0f; /* init value unused but some compilers may complain */ - const float delta_fixed = 1.0f - dmd->midlevel; /* when no texture is used, we fallback to white */ float (*vert_clnors)[3] = NULL; - float local_mat[4][4]; + float local_mat[4][4] = {{0}}; const bool use_global_direction = dmd->space == MOD_DISP_SPACE_GLOBAL; if (!dmd->texture && dmd->direction == MOD_DISP_DIR_RGB_XYZ) return; @@ -259,81 +370,25 @@ static void displaceModifier_do( copy_m4_m4(local_mat, ob->obmat); } - for (i = 0; i < numVerts; i++) { - TexResult texres; - float strength = dmd->strength; - float delta; - float local_vec[3]; - - if (dvert) { - weight = defvert_find_weight(dvert + i, defgrp_index); - if (weight == 0.0f) continue; - } - - if (dmd->texture) { - texres.nor = NULL; - BKE_texture_get_value(dmd->modifier.scene, dmd->texture, tex_co[i], &texres, false); - delta = texres.tin - dmd->midlevel; - } - else { - delta = delta_fixed; /* (1.0f - dmd->midlevel) */ /* never changes */ - } + DisplaceUserdata data = {NULL}; + data.dmd = dmd; + data.dvert = dvert; + data.weight = weight; + data.defgrp_index = defgrp_index; + data.direction = direction; + data.use_global_direction = use_global_direction; + data.tex_co = tex_co; + data.vertexCos = vertexCos; + copy_m4_m4(data.local_mat, local_mat); + data.mvert = mvert; + data.vert_clnors = vert_clnors; + if (dmd->texture != NULL) { + data.pool = BKE_image_pool_new(); + } + BLI_task_parallel_range(0, numVerts, &data, displaceModifier_do_task, numVerts > 512); - if (dvert) strength *= weight; - - delta *= strength; - CLAMP(delta, -10000, 10000); - - switch (direction) { - case MOD_DISP_DIR_X: - if (use_global_direction) { - vertexCos[i][0] += delta * local_mat[0][0]; - vertexCos[i][1] += delta * local_mat[1][0]; - vertexCos[i][2] += delta * local_mat[2][0]; - } - else { - vertexCos[i][0] += delta; - } - break; - case MOD_DISP_DIR_Y: - if (use_global_direction) { - vertexCos[i][0] += delta * local_mat[0][1]; - vertexCos[i][1] += delta * local_mat[1][1]; - vertexCos[i][2] += delta * local_mat[2][1]; - } - else { - vertexCos[i][1] += delta; - } - break; - case MOD_DISP_DIR_Z: - if (use_global_direction) { - vertexCos[i][0] += delta * local_mat[0][2]; - vertexCos[i][1] += delta * local_mat[1][2]; - vertexCos[i][2] += delta * local_mat[2][2]; - } - else { - vertexCos[i][2] += delta; - } - break; - case MOD_DISP_DIR_RGB_XYZ: - local_vec[0] = texres.tr - dmd->midlevel; - local_vec[1] = texres.tg - dmd->midlevel; - local_vec[2] = texres.tb - dmd->midlevel; - if (use_global_direction) { - mul_transposed_mat3_m4_v3(local_mat, local_vec); - } - mul_v3_fl(local_vec, strength); - add_v3_v3(vertexCos[i], local_vec); - break; - case MOD_DISP_DIR_NOR: - vertexCos[i][0] += delta * (mvert[i].no[0] / 32767.0f); - vertexCos[i][1] += delta * (mvert[i].no[1] / 32767.0f); - vertexCos[i][2] += delta * (mvert[i].no[2] / 32767.0f); - break; - case MOD_DISP_DIR_CLNOR: - madd_v3_v3fl(vertexCos[i], vert_clnors[i], delta); - break; - } + if (data.pool != NULL) { + BKE_image_pool_free(data.pool); } if (tex_co) { diff --git a/source/blender/modifiers/intern/MOD_dynamicpaint.c b/source/blender/modifiers/intern/MOD_dynamicpaint.c index 05068b9b597..bb75d655802 100644 --- a/source/blender/modifiers/intern/MOD_dynamicpaint.c +++ b/source/blender/modifiers/intern/MOD_dynamicpaint.c @@ -116,7 +116,7 @@ static DerivedMesh *applyModifier(ModifierData *md, Object *ob, static bool is_brush_cb(Object *UNUSED(ob), ModifierData *pmd) { - return ((DynamicPaintModifierData*)pmd)->brush != NULL; + return ((DynamicPaintModifierData *)pmd)->brush != NULL; } static void updateDepgraph(ModifierData *md, DagForest *forest, diff --git a/source/blender/modifiers/intern/MOD_mirror.c b/source/blender/modifiers/intern/MOD_mirror.c index dd127cc255c..9705edc580c 100644 --- a/source/blender/modifiers/intern/MOD_mirror.c +++ b/source/blender/modifiers/intern/MOD_mirror.c @@ -274,8 +274,8 @@ static DerivedMesh *doMirrorOnAxis(MirrorModifierData *mmd, int j = maxLoops; dmloopuv += j; /* second set of loops only */ for (; j-- > 0; dmloopuv++) { - if (do_mirr_u) dmloopuv->uv[0] = 1.0f - dmloopuv->uv[0]; - if (do_mirr_v) dmloopuv->uv[1] = 1.0f - dmloopuv->uv[1]; + if (do_mirr_u) dmloopuv->uv[0] = 1.0f - dmloopuv->uv[0] + mmd->uv_offset[0]; + if (do_mirr_v) dmloopuv->uv[1] = 1.0f - dmloopuv->uv[1] + mmd->uv_offset[1]; } } } diff --git a/source/blender/modifiers/intern/MOD_surface.c b/source/blender/modifiers/intern/MOD_surface.c index 68987a1d28e..e5b0f9c0001 100644 --- a/source/blender/modifiers/intern/MOD_surface.c +++ b/source/blender/modifiers/intern/MOD_surface.c @@ -172,6 +172,7 @@ ModifierTypeInfo modifierType_Surface = { /* structSize */ sizeof(SurfaceModifierData), /* type */ eModifierTypeType_OnlyDeform, /* flags */ eModifierTypeFlag_AcceptsMesh | + eModifierTypeFlag_AcceptsCVs | eModifierTypeFlag_NoUserAdd, /* copyData */ NULL, diff --git a/source/blender/modifiers/intern/MOD_surfacedeform.c b/source/blender/modifiers/intern/MOD_surfacedeform.c new file mode 100644 index 00000000000..776cf02754e --- /dev/null +++ b/source/blender/modifiers/intern/MOD_surfacedeform.c @@ -0,0 +1,1226 @@ +#include "DNA_object_types.h" +#include "DNA_scene_types.h" + +#include "BLI_alloca.h" +#include "BLI_math.h" +#include "BLI_math_geom.h" +#include "BLI_task.h" + +#include "BKE_cdderivedmesh.h" +#include "BKE_editmesh.h" +#include "BKE_library_query.h" +#include "BKE_modifier.h" + +#include "depsgraph_private.h" + +#include "MEM_guardedalloc.h" + +#include "MOD_util.h" + +typedef struct SDefAdjacency { + struct SDefAdjacency *next; + unsigned int index; +} SDefAdjacency; + +typedef struct SDefAdjacencyArray { + SDefAdjacency *first; + unsigned int num; /* Careful, this is twice the number of polygons (avoids an extra loop) */ +} SDefAdjacencyArray; + +typedef struct SDefEdgePolys { + unsigned int polys[2], num; +} SDefEdgePolys; + +typedef struct SDefBindCalcData { + BVHTreeFromMesh * const treeData; + const SDefAdjacencyArray * const vert_edges; + const SDefEdgePolys * const edge_polys; + SDefVert * const bind_verts; + const MLoopTri * const looptri; + const MPoly * const mpoly; + const MEdge * const medge; + const MLoop * const mloop; + float (* const targetCos)[3]; + float (* const vertexCos)[3]; + float imat[4][4]; + const float falloff; + int success; +} SDefBindCalcData; + +typedef struct SDefBindPoly { + float (*coords)[3]; + float (*coords_v2)[2]; + float point_v2[2]; + float weight_angular; + float weight_dist_proj; + float weight_dist; + float weight; + float scales[2]; + float centroid[3]; + float centroid_v2[2]; + float normal[3]; + float cent_edgemid_vecs_v2[2][2]; + float edgemid_angle; + float point_edgemid_angles[2]; + float corner_edgemid_angles[2]; + float dominant_angle_weight; + unsigned int index; + unsigned int numverts; + unsigned int loopstart; + unsigned int edge_inds[2]; + unsigned int edge_vert_inds[2]; + unsigned int corner_ind; + unsigned int dominant_edge; + bool inside; +} SDefBindPoly; + +typedef struct SDefBindWeightData { + SDefBindPoly *bind_polys; + unsigned int numpoly; + unsigned int numbinds; +} SDefBindWeightData; + +typedef struct SDefDeformData { + const SDefVert * const bind_verts; + float (* const targetCos)[3]; + float (* const vertexCos)[3]; +} SDefDeformData; + +/* Bind result values */ +enum { + MOD_SDEF_BIND_RESULT_SUCCESS = 1, + MOD_SDEF_BIND_RESULT_GENERIC_ERR = 0, + MOD_SDEF_BIND_RESULT_MEM_ERR = -1, + MOD_SDEF_BIND_RESULT_NONMANY_ERR = -2, + MOD_SDEF_BIND_RESULT_CONCAVE_ERR = -3, + MOD_SDEF_BIND_RESULT_OVERLAP_ERR = -4, +}; + +/* Infinite weight flags */ +enum { + MOD_SDEF_INFINITE_WEIGHT_ANGULAR = (1 << 0), + MOD_SDEF_INFINITE_WEIGHT_DIST_PROJ = (1 << 1), + MOD_SDEF_INFINITE_WEIGHT_DIST = (1 << 2), +}; + +static void initData(ModifierData *md) +{ + SurfaceDeformModifierData *smd = (SurfaceDeformModifierData *)md; + smd->target = NULL; + smd->verts = NULL; + smd->flags = 0; + smd->falloff = 4.0f; +} + +static void freeData(ModifierData *md) +{ + SurfaceDeformModifierData *smd = (SurfaceDeformModifierData *)md; + + if (smd->verts) { + for (int i = 0; i < smd->numverts; i++) { + if (smd->verts[i].binds) { + for (int j = 0; j < smd->verts[i].numbinds; j++) { + MEM_SAFE_FREE(smd->verts[i].binds[j].vert_inds); + MEM_SAFE_FREE(smd->verts[i].binds[j].vert_weights); + } + + MEM_freeN(smd->verts[i].binds); + } + } + + MEM_freeN(smd->verts); + smd->verts = NULL; + } +} + +static void copyData(ModifierData *md, ModifierData *target) +{ + SurfaceDeformModifierData *smd = (SurfaceDeformModifierData *)md; + SurfaceDeformModifierData *tsmd = (SurfaceDeformModifierData *)target; + + *tsmd = *smd; + + if (smd->verts) { + tsmd->verts = MEM_dupallocN(smd->verts); + + for (int i = 0; i < smd->numverts; i++) { + if (smd->verts[i].binds) { + tsmd->verts[i].binds = MEM_dupallocN(smd->verts[i].binds); + + for (int j = 0; j < smd->verts[i].numbinds; j++) { + if (smd->verts[i].binds[j].vert_inds) { + tsmd->verts[i].binds[j].vert_inds = MEM_dupallocN(smd->verts[i].binds[j].vert_inds); + } + + if (smd->verts[i].binds[j].vert_weights) { + tsmd->verts[i].binds[j].vert_weights = MEM_dupallocN(smd->verts[i].binds[j].vert_weights); + } + } + } + } + } +} + +static void foreachObjectLink(ModifierData *md, Object *ob, ObjectWalkFunc walk, void *userData) +{ + SurfaceDeformModifierData *smd = (SurfaceDeformModifierData *)md; + + walk(userData, ob, &smd->target, IDWALK_NOP); +} + +static void updateDepgraph(ModifierData *md, DagForest *forest, + struct Main *UNUSED(bmain), + struct Scene *UNUSED(scene), + Object *UNUSED(ob), + DagNode *obNode) +{ + SurfaceDeformModifierData *smd = (SurfaceDeformModifierData *)md; + + if (smd->target) { + DagNode *curNode = dag_get_node(forest, smd->target); + + dag_add_relation(forest, curNode, obNode, DAG_RL_DATA_DATA, "Surface Deform Modifier"); + } +} + +static void updateDepsgraph(ModifierData *md, + struct Main *UNUSED(bmain), + struct Scene *UNUSED(scene), + Object *UNUSED(ob), + struct DepsNodeHandle *node) +{ + SurfaceDeformModifierData *smd = (SurfaceDeformModifierData *)md; + if (smd->target != NULL) { + DEG_add_object_relation(node, smd->target, DEG_OB_COMP_GEOMETRY, "Surface Deform Modifier"); + } +} + +static void freeAdjacencyMap(SDefAdjacencyArray * const vert_edges, SDefAdjacency * const adj_ref, SDefEdgePolys * const edge_polys) +{ + MEM_freeN(edge_polys); + + MEM_freeN(adj_ref); + + MEM_freeN(vert_edges); +} + +static int buildAdjacencyMap(const MPoly *poly, const MEdge *edge, const MLoop * const mloop, const unsigned int numpoly, const unsigned int numedges, + SDefAdjacencyArray * const vert_edges, SDefAdjacency *adj, SDefEdgePolys * const edge_polys) +{ + const MLoop *loop; + + /* Fing polygons adjacent to edges */ + for (int i = 0; i < numpoly; i++, poly++) { + loop = &mloop[poly->loopstart]; + + for (int j = 0; j < poly->totloop; j++, loop++) { + if (edge_polys[loop->e].num == 0) { + edge_polys[loop->e].polys[0] = i; + edge_polys[loop->e].polys[1] = -1; + edge_polys[loop->e].num++; + } + else if (edge_polys[loop->e].num == 1) { + edge_polys[loop->e].polys[1] = i; + edge_polys[loop->e].num++; + } + else { + return MOD_SDEF_BIND_RESULT_NONMANY_ERR; + } + } + } + + /* Find edges adjacent to vertices */ + for (int i = 0; i < numedges; i++, edge++) { + adj->next = vert_edges[edge->v1].first; + adj->index = i; + vert_edges[edge->v1].first = adj; + vert_edges[edge->v1].num += edge_polys[i].num; + adj++; + + adj->next = vert_edges[edge->v2].first; + adj->index = i; + vert_edges[edge->v2].first = adj; + vert_edges[edge->v2].num += edge_polys[i].num; + adj++; + } + + return MOD_SDEF_BIND_RESULT_SUCCESS; +} + +BLI_INLINE void sortPolyVertsEdge(unsigned int *indices, const MLoop * const mloop, const unsigned int edge, const unsigned int num) +{ + bool found = false; + + for (int i = 0; i < num; i++) { + if (mloop[i].e == edge) { + found = true; + } + if (found) { + *indices = mloop[i].v; + indices++; + } + } + + /* Fill in remaining vertex indices that occur before the edge */ + for (int i = 0; mloop[i].e != edge; i++) { + *indices = mloop[i].v; + indices++; + } +} + +BLI_INLINE void sortPolyVertsTri(unsigned int *indices, const MLoop * const mloop, const unsigned int loopstart, const unsigned int num) +{ + for (int i = loopstart; i < num; i++) { + *indices = mloop[i].v; + indices++; + } + + for (int i = 0; i < loopstart; i++) { + *indices = mloop[i].v; + indices++; + } +} + +BLI_INLINE unsigned int nearestVert(SDefBindCalcData * const data, const float point_co[3]) +{ + BVHTreeNearest nearest = {.dist_sq = FLT_MAX, .index = -1}; + const MPoly *poly; + const MEdge *edge; + const MLoop *loop; + float t_point[3]; + float max_dist = FLT_MAX; + float dist; + unsigned int index = 0; + + mul_v3_m4v3(t_point, data->imat, point_co); + + BLI_bvhtree_find_nearest(data->treeData->tree, t_point, &nearest, data->treeData->nearest_callback, data->treeData); + + poly = &data->mpoly[data->looptri[nearest.index].poly]; + loop = &data->mloop[poly->loopstart]; + + for (int i = 0; i < poly->totloop; i++, loop++) { + edge = &data->medge[loop->e]; + dist = dist_squared_to_line_segment_v3(point_co, data->targetCos[edge->v1], data->targetCos[edge->v2]); + + if (dist < max_dist) { + max_dist = dist; + index = loop->e; + } + } + + edge = &data->medge[index]; + if (len_squared_v3v3(point_co, data->targetCos[edge->v1]) < len_squared_v3v3(point_co, data->targetCos[edge->v2])) { + return edge->v1; + } + else { + return edge->v2; + } +} + +BLI_INLINE int isPolyValid(const float coords[][2], const unsigned int nr) +{ + float prev_co[2]; + float curr_vec[2], prev_vec[2]; + + if (!is_poly_convex_v2(coords, nr)) { + return MOD_SDEF_BIND_RESULT_CONCAVE_ERR; + } + + copy_v2_v2(prev_co, coords[nr - 1]); + sub_v2_v2v2(prev_vec, prev_co, coords[nr - 2]); + + for (int i = 0; i < nr; i++) { + sub_v2_v2v2(curr_vec, coords[i], prev_co); + + if (len_squared_v2(curr_vec) < FLT_EPSILON) { + return MOD_SDEF_BIND_RESULT_OVERLAP_ERR; + } + + if (1.0f - dot_v2v2(prev_vec, curr_vec) < FLT_EPSILON) { + return MOD_SDEF_BIND_RESULT_CONCAVE_ERR; + } + + copy_v2_v2(prev_co, coords[i]); + copy_v2_v2(prev_vec, curr_vec); + } + + return MOD_SDEF_BIND_RESULT_SUCCESS; +} + +static void freeBindData(SDefBindWeightData * const bwdata) +{ + SDefBindPoly *bpoly = bwdata->bind_polys; + + if (bwdata->bind_polys) { + for (int i = 0; i < bwdata->numpoly; bpoly++, i++) { + MEM_SAFE_FREE(bpoly->coords); + MEM_SAFE_FREE(bpoly->coords_v2); + } + + MEM_freeN(bwdata->bind_polys); + } + + MEM_freeN(bwdata); +} + +BLI_INLINE float computeAngularWeight(const float point_angle, const float edgemid_angle) +{ + float weight; + + weight = point_angle; + weight /= edgemid_angle; + weight *= M_PI_2; + + return sinf(weight); +} + +BLI_INLINE SDefBindWeightData *computeBindWeights(SDefBindCalcData * const data, const float point_co[3]) +{ + const unsigned int nearest = nearestVert(data, point_co); + const SDefAdjacency * const vert_edges = data->vert_edges[nearest].first; + const SDefEdgePolys * const edge_polys = data->edge_polys; + + const SDefAdjacency *vedge; + const MPoly *poly; + const MLoop *loop; + + SDefBindWeightData *bwdata; + SDefBindPoly *bpoly; + + float world[3] = {0.0f, 0.0f, 1.0f}; + float avg_point_dist = 0.0f; + float tot_weight = 0.0f; + int inf_weight_flags = 0; + + bwdata = MEM_callocN(sizeof(*bwdata), "SDefBindWeightData"); + if (bwdata == NULL) { + data->success = MOD_SDEF_BIND_RESULT_MEM_ERR; + return NULL; + } + + bwdata->numpoly = data->vert_edges[nearest].num / 2; + + bpoly = MEM_callocN(sizeof(*bpoly) * bwdata->numpoly, "SDefBindPoly"); + if (bpoly == NULL) { + freeBindData(bwdata); + data->success = MOD_SDEF_BIND_RESULT_MEM_ERR; + return NULL; + } + + bwdata->bind_polys = bpoly; + + /* Loop over all adjacent edges, and build the SDefBindPoly data for each poly adjacent to those */ + for (vedge = vert_edges; vedge; vedge = vedge->next) { + unsigned int edge_ind = vedge->index; + + for (int i = 0; i < edge_polys[edge_ind].num; i++) { + { + bpoly = bwdata->bind_polys; + + for (int j = 0; j < bwdata->numpoly; bpoly++, j++) { + /* If coords isn't allocated, we have reached the first uninitialized bpoly */ + if ((bpoly->index == edge_polys[edge_ind].polys[i]) || (!bpoly->coords)) { + break; + } + } + } + + /* Check if poly was already created by another edge or still has to be initialized */ + if (!bpoly->coords) { + float angle; + float axis[3]; + float tmp_vec_v2[2]; + int is_poly_valid; + + bpoly->index = edge_polys[edge_ind].polys[i]; + bpoly->coords = NULL; + bpoly->coords_v2 = NULL; + + /* Copy poly data */ + poly = &data->mpoly[bpoly->index]; + loop = &data->mloop[poly->loopstart]; + + bpoly->numverts = poly->totloop; + bpoly->loopstart = poly->loopstart; + + bpoly->coords = MEM_mallocN(sizeof(*bpoly->coords) * poly->totloop, "SDefBindPolyCoords"); + if (bpoly->coords == NULL) { + freeBindData(bwdata); + data->success = MOD_SDEF_BIND_RESULT_MEM_ERR; + return NULL; + } + + bpoly->coords_v2 = MEM_mallocN(sizeof(*bpoly->coords_v2) * poly->totloop, "SDefBindPolyCoords_v2"); + if (bpoly->coords_v2 == NULL) { + freeBindData(bwdata); + data->success = MOD_SDEF_BIND_RESULT_MEM_ERR; + return NULL; + } + + for (int j = 0; j < poly->totloop; j++, loop++) { + copy_v3_v3(bpoly->coords[j], data->targetCos[loop->v]); + + /* Find corner and edge indices within poly loop array */ + if (loop->v == nearest) { + bpoly->corner_ind = j; + bpoly->edge_vert_inds[0] = (j == 0) ? (poly->totloop - 1) : (j - 1); + bpoly->edge_vert_inds[1] = (j == poly->totloop - 1) ? (0) : (j + 1); + + bpoly->edge_inds[0] = data->mloop[poly->loopstart + bpoly->edge_vert_inds[0]].e; + bpoly->edge_inds[1] = loop->e; + } + } + + /* Compute poly's parametric data */ + mid_v3_v3_array(bpoly->centroid, bpoly->coords, poly->totloop); + normal_poly_v3(bpoly->normal, bpoly->coords, poly->totloop); + + /* Compute poly skew angle and axis */ + angle = angle_normalized_v3v3(bpoly->normal, world); + + cross_v3_v3v3(axis, bpoly->normal, world); + normalize_v3(axis); + + /* Map coords onto 2d normal plane */ + map_to_plane_axis_angle_v2_v3v3fl(bpoly->point_v2, point_co, axis, angle); + + zero_v2(bpoly->centroid_v2); + for (int j = 0; j < poly->totloop; j++) { + map_to_plane_axis_angle_v2_v3v3fl(bpoly->coords_v2[j], bpoly->coords[j], axis, angle); + madd_v2_v2fl(bpoly->centroid_v2, bpoly->coords_v2[j], 1.0f / poly->totloop); + } + + is_poly_valid = isPolyValid(bpoly->coords_v2, poly->totloop); + + if (is_poly_valid != MOD_SDEF_BIND_RESULT_SUCCESS) { + freeBindData(bwdata); + data->success = is_poly_valid; + return NULL; + } + + bpoly->inside = isect_point_poly_v2(bpoly->point_v2, bpoly->coords_v2, poly->totloop, false); + + /* Initialize weight components */ + bpoly->weight_angular = 1.0f; + bpoly->weight_dist_proj = len_v2v2(bpoly->centroid_v2, bpoly->point_v2); + bpoly->weight_dist = len_v3v3(bpoly->centroid, point_co); + + avg_point_dist += bpoly->weight_dist; + + /* Compute centroid to mid-edge vectors */ + mid_v2_v2v2(bpoly->cent_edgemid_vecs_v2[0], + bpoly->coords_v2[bpoly->edge_vert_inds[0]], + bpoly->coords_v2[bpoly->corner_ind]); + + mid_v2_v2v2(bpoly->cent_edgemid_vecs_v2[1], + bpoly->coords_v2[bpoly->edge_vert_inds[1]], + bpoly->coords_v2[bpoly->corner_ind]); + + sub_v2_v2(bpoly->cent_edgemid_vecs_v2[0], bpoly->centroid_v2); + sub_v2_v2(bpoly->cent_edgemid_vecs_v2[1], bpoly->centroid_v2); + + /* Compute poly scales with respect to mid-edges, and normalize the vectors */ + bpoly->scales[0] = normalize_v2(bpoly->cent_edgemid_vecs_v2[0]); + bpoly->scales[1] = normalize_v2(bpoly->cent_edgemid_vecs_v2[1]); + + /* Compute the required polygon angles */ + bpoly->edgemid_angle = angle_normalized_v2v2(bpoly->cent_edgemid_vecs_v2[0], bpoly->cent_edgemid_vecs_v2[1]); + + sub_v2_v2v2(tmp_vec_v2, bpoly->coords_v2[bpoly->corner_ind], bpoly->centroid_v2); + normalize_v2(tmp_vec_v2); + + bpoly->corner_edgemid_angles[0] = angle_normalized_v2v2(tmp_vec_v2, bpoly->cent_edgemid_vecs_v2[0]); + bpoly->corner_edgemid_angles[1] = angle_normalized_v2v2(tmp_vec_v2, bpoly->cent_edgemid_vecs_v2[1]); + + /* Check for inifnite weights, and compute angular data otherwise */ + if (bpoly->weight_dist < FLT_EPSILON) { + inf_weight_flags |= MOD_SDEF_INFINITE_WEIGHT_DIST_PROJ; + inf_weight_flags |= MOD_SDEF_INFINITE_WEIGHT_DIST; + } + else if (bpoly->weight_dist_proj < FLT_EPSILON) { + inf_weight_flags |= MOD_SDEF_INFINITE_WEIGHT_DIST_PROJ; + } + else { + float cent_point_vec[2]; + + sub_v2_v2v2(cent_point_vec, bpoly->point_v2, bpoly->centroid_v2); + normalize_v2(cent_point_vec); + + bpoly->point_edgemid_angles[0] = angle_normalized_v2v2(cent_point_vec, bpoly->cent_edgemid_vecs_v2[0]); + bpoly->point_edgemid_angles[1] = angle_normalized_v2v2(cent_point_vec, bpoly->cent_edgemid_vecs_v2[1]); + } + } + } + } + + avg_point_dist /= bwdata->numpoly; + + /* If weights 1 and 2 are not infinite, loop over all adjacent edges again, + * and build adjacency dependent angle data (depends on all polygons having been computed) */ + if (!inf_weight_flags) { + for (vedge = vert_edges; vedge; vedge = vedge->next) { + SDefBindPoly *bpolys[2]; + const SDefEdgePolys *epolys; + float ang_weights[2]; + unsigned int edge_ind = vedge->index; + unsigned int edge_on_poly[2]; + + epolys = &edge_polys[edge_ind]; + + /* Find bind polys corresponding to the edge's adjacent polys */ + bpoly = bwdata->bind_polys; + + for (int i = 0, j = 0; (i < bwdata->numpoly) && (j < epolys->num); bpoly++, i++) { + if (ELEM(bpoly->index, epolys->polys[0], epolys->polys[1])) { + bpolys[j] = bpoly; + + if (bpoly->edge_inds[0] == edge_ind) { + edge_on_poly[j] = 0; + } + else { + edge_on_poly[j] = 1; + } + + j++; + } + } + + /* Compute angular weight component */ + if (epolys->num == 1) { + ang_weights[0] = computeAngularWeight(bpolys[0]->point_edgemid_angles[edge_on_poly[0]], bpolys[0]->edgemid_angle); + bpolys[0]->weight_angular *= ang_weights[0] * ang_weights[0]; + } + else if (epolys->num == 2) { + ang_weights[0] = computeAngularWeight(bpolys[0]->point_edgemid_angles[edge_on_poly[0]], bpolys[0]->edgemid_angle); + ang_weights[1] = computeAngularWeight(bpolys[1]->point_edgemid_angles[edge_on_poly[1]], bpolys[1]->edgemid_angle); + + bpolys[0]->weight_angular *= ang_weights[0] * ang_weights[1]; + bpolys[1]->weight_angular *= ang_weights[0] * ang_weights[1]; + } + } + } + + /* Compute scalings and falloff. + * Scale all weights if no infinite weight is found, + * scale only unprojected weight if projected weight is infinite, + * scale none if both are infinite. */ + if (!inf_weight_flags) { + bpoly = bwdata->bind_polys; + + for (int i = 0; i < bwdata->numpoly; bpoly++, i++) { + float corner_angle_weights[2]; + float scale_weight, sqr, inv_sqr; + + corner_angle_weights[0] = bpoly->point_edgemid_angles[0] / bpoly->corner_edgemid_angles[0]; + corner_angle_weights[1] = bpoly->point_edgemid_angles[1] / bpoly->corner_edgemid_angles[1]; + + if (isnan(corner_angle_weights[0]) || isnan(corner_angle_weights[1])) { + freeBindData(bwdata); + data->success = MOD_SDEF_BIND_RESULT_GENERIC_ERR; + return NULL; + } + + /* Find which edge the point is closer to */ + if (corner_angle_weights[0] < corner_angle_weights[1]) { + bpoly->dominant_edge = 0; + bpoly->dominant_angle_weight = corner_angle_weights[0]; + } + else { + bpoly->dominant_edge = 1; + bpoly->dominant_angle_weight = corner_angle_weights[1]; + } + + bpoly->dominant_angle_weight = sinf(bpoly->dominant_angle_weight * M_PI_2); + + /* Compute quadratic angular scale interpolation weight */ + scale_weight = bpoly->point_edgemid_angles[bpoly->dominant_edge] / bpoly->edgemid_angle; + scale_weight /= scale_weight + (bpoly->point_edgemid_angles[!bpoly->dominant_edge] / bpoly->edgemid_angle); + + sqr = scale_weight * scale_weight; + inv_sqr = 1.0f - scale_weight; + inv_sqr *= inv_sqr; + scale_weight = sqr / (sqr + inv_sqr); + + /* Compute interpolated scale (no longer need the individual scales, + * so simply storing the result over the scale in index zero) */ + bpoly->scales[0] = bpoly->scales[bpoly->dominant_edge] * (1.0f - scale_weight) + + bpoly->scales[!bpoly->dominant_edge] * scale_weight; + + /* Scale the point distance weights, and introduce falloff */ + bpoly->weight_dist_proj /= bpoly->scales[0]; + bpoly->weight_dist_proj = powf(bpoly->weight_dist_proj, data->falloff); + + bpoly->weight_dist /= avg_point_dist; + bpoly->weight_dist = powf(bpoly->weight_dist, data->falloff); + + /* Re-check for infinite weights, now that all scalings and interpolations are computed */ + if (bpoly->weight_dist < FLT_EPSILON) { + inf_weight_flags |= MOD_SDEF_INFINITE_WEIGHT_DIST_PROJ; + inf_weight_flags |= MOD_SDEF_INFINITE_WEIGHT_DIST; + } + else if (bpoly->weight_dist_proj < FLT_EPSILON) { + inf_weight_flags |= MOD_SDEF_INFINITE_WEIGHT_DIST_PROJ; + } + else if (bpoly->weight_angular < FLT_EPSILON) { + inf_weight_flags |= MOD_SDEF_INFINITE_WEIGHT_ANGULAR; + } + } + } + else if (!(inf_weight_flags & MOD_SDEF_INFINITE_WEIGHT_DIST)) { + bpoly = bwdata->bind_polys; + + for (int i = 0; i < bwdata->numpoly; bpoly++, i++) { + /* Scale the point distance weight by average point distance, and introduce falloff */ + bpoly->weight_dist /= avg_point_dist; + bpoly->weight_dist = powf(bpoly->weight_dist, data->falloff); + + /* Re-check for infinite weights, now that all scalings and interpolations are computed */ + if (bpoly->weight_dist < FLT_EPSILON) { + inf_weight_flags |= MOD_SDEF_INFINITE_WEIGHT_DIST; + } + } + } + + /* Final loop, to compute actual weights */ + bpoly = bwdata->bind_polys; + + for (int i = 0; i < bwdata->numpoly; bpoly++, i++) { + /* Weight computation from components */ + if (inf_weight_flags & MOD_SDEF_INFINITE_WEIGHT_DIST) { + bpoly->weight = bpoly->weight_dist < FLT_EPSILON ? 1.0f : 0.0f; + } + else if (inf_weight_flags & MOD_SDEF_INFINITE_WEIGHT_DIST_PROJ) { + bpoly->weight = bpoly->weight_dist_proj < FLT_EPSILON ? + 1.0f / bpoly->weight_dist : 0.0f; + } + else if (inf_weight_flags & MOD_SDEF_INFINITE_WEIGHT_ANGULAR) { + bpoly->weight = bpoly->weight_angular < FLT_EPSILON ? + 1.0f / bpoly->weight_dist_proj / bpoly->weight_dist : 0.0f; + } + else { + bpoly->weight = 1.0f / bpoly->weight_angular / + bpoly->weight_dist_proj / + bpoly->weight_dist; + } + + tot_weight += bpoly->weight; + } + + bpoly = bwdata->bind_polys; + + for (int i = 0; i < bwdata->numpoly; bpoly++, i++) { + bpoly->weight /= tot_weight; + + /* Evaluate if this poly is relevant to bind */ + /* Even though the weights should add up to 1.0, + * the losses of weights smaller than epsilon here + * should be negligible... */ + if (bpoly->weight >= FLT_EPSILON) { + if (bpoly->inside) { + bwdata->numbinds += 1; + } + else { + if (bpoly->dominant_angle_weight < FLT_EPSILON || 1.0f - bpoly->dominant_angle_weight < FLT_EPSILON) { + bwdata->numbinds += 1; + } + else { + bwdata->numbinds += 2; + } + } + } + } + + return bwdata; +} + +BLI_INLINE float computeNormalDisplacement(const float point_co[3], const float point_co_proj[3], const float normal[3]) +{ + float disp_vec[3]; + float normal_dist; + + sub_v3_v3v3(disp_vec, point_co, point_co_proj); + normal_dist = len_v3(disp_vec); + + if (dot_v3v3(disp_vec, normal) < 0) { + normal_dist *= -1; + } + + return normal_dist; +} + +static void bindVert(void *userdata, void *UNUSED(userdata_chunk), const int index, const int UNUSED(threadid)) +{ + SDefBindCalcData * const data = (SDefBindCalcData *)userdata; + float point_co[3]; + float point_co_proj[3]; + + SDefBindWeightData *bwdata; + SDefVert *sdvert = data->bind_verts + index; + SDefBindPoly *bpoly; + SDefBind *sdbind; + + if (data->success != MOD_SDEF_BIND_RESULT_SUCCESS) { + sdvert->binds = NULL; + sdvert->numbinds = 0; + return; + } + + copy_v3_v3(point_co, data->vertexCos[index]); + bwdata = computeBindWeights(data, point_co); + + if (bwdata == NULL) { + sdvert->binds = NULL; + sdvert->numbinds = 0; + return; + } + + sdvert->binds = MEM_callocN(sizeof(*sdvert->binds) * bwdata->numbinds, "SDefVertBindData"); + if (sdvert->binds == NULL) { + data->success = MOD_SDEF_BIND_RESULT_MEM_ERR; + sdvert->numbinds = 0; + return; + } + + sdvert->numbinds = bwdata->numbinds; + + sdbind = sdvert->binds; + + bpoly = bwdata->bind_polys; + + for (int i = 0; i < bwdata->numbinds; bpoly++) { + if (bpoly->weight >= FLT_EPSILON) { + if (bpoly->inside) { + const MLoop *loop = &data->mloop[bpoly->loopstart]; + + sdbind->influence = bpoly->weight; + sdbind->numverts = bpoly->numverts; + + sdbind->mode = MOD_SDEF_MODE_NGON; + sdbind->vert_weights = MEM_mallocN(sizeof(*sdbind->vert_weights) * bpoly->numverts, "SDefNgonVertWeights"); + if (sdbind->vert_weights == NULL) { + data->success = MOD_SDEF_BIND_RESULT_MEM_ERR; + return; + } + + sdbind->vert_inds = MEM_mallocN(sizeof(*sdbind->vert_inds) * bpoly->numverts, "SDefNgonVertInds"); + if (sdbind->vert_inds == NULL) { + data->success = MOD_SDEF_BIND_RESULT_MEM_ERR; + return; + } + + interp_weights_poly_v2(sdbind->vert_weights, bpoly->coords_v2, bpoly->numverts, bpoly->point_v2); + + /* Reproject vert based on weights and original poly verts, to reintroduce poly non-planarity */ + zero_v3(point_co_proj); + for (int j = 0; j < bpoly->numverts; j++, loop++) { + madd_v3_v3fl(point_co_proj, bpoly->coords[j], sdbind->vert_weights[j]); + sdbind->vert_inds[j] = loop->v; + } + + sdbind->normal_dist = computeNormalDisplacement(point_co, point_co_proj, bpoly->normal); + + sdbind++; + i++; + } + else { + float tmp_vec[3]; + float cent[3], norm[3]; + float v1[3], v2[3], v3[3]; + + if (1.0f - bpoly->dominant_angle_weight >= FLT_EPSILON) { + sdbind->influence = bpoly->weight * (1.0f - bpoly->dominant_angle_weight); + sdbind->numverts = bpoly->numverts; + + sdbind->mode = MOD_SDEF_MODE_CENTROID; + sdbind->vert_weights = MEM_mallocN(sizeof(*sdbind->vert_weights) * 3, "SDefCentVertWeights"); + if (sdbind->vert_weights == NULL) { + data->success = MOD_SDEF_BIND_RESULT_MEM_ERR; + return; + } + + sdbind->vert_inds = MEM_mallocN(sizeof(*sdbind->vert_inds) * bpoly->numverts, "SDefCentVertInds"); + if (sdbind->vert_inds == NULL) { + data->success = MOD_SDEF_BIND_RESULT_MEM_ERR; + return; + } + + sortPolyVertsEdge(sdbind->vert_inds, &data->mloop[bpoly->loopstart], + bpoly->edge_inds[bpoly->dominant_edge], bpoly->numverts); + + copy_v3_v3(v1, data->targetCos[sdbind->vert_inds[0]]); + copy_v3_v3(v2, data->targetCos[sdbind->vert_inds[1]]); + copy_v3_v3(v3, bpoly->centroid); + + mid_v3_v3v3v3(cent, v1, v2, v3); + normal_tri_v3(norm, v1, v2, v3); + + add_v3_v3v3(tmp_vec, point_co, bpoly->normal); + + /* We are sure the line is not parallel to the plane. + * Checking return value just to avoid warning... */ + if (!isect_line_plane_v3(point_co_proj, point_co, tmp_vec, cent, norm)) { + BLI_assert(false); + } + + interp_weights_tri_v3(sdbind->vert_weights, v1, v2, v3, point_co_proj); + + sdbind->normal_dist = computeNormalDisplacement(point_co, point_co_proj, bpoly->normal); + + sdbind++; + i++; + } + + if (bpoly->dominant_angle_weight >= FLT_EPSILON) { + sdbind->influence = bpoly->weight * bpoly->dominant_angle_weight; + sdbind->numverts = bpoly->numverts; + + sdbind->mode = MOD_SDEF_MODE_LOOPTRI; + sdbind->vert_weights = MEM_mallocN(sizeof(*sdbind->vert_weights) * 3, "SDefTriVertWeights"); + if (sdbind->vert_weights == NULL) { + data->success = MOD_SDEF_BIND_RESULT_MEM_ERR; + return; + } + + sdbind->vert_inds = MEM_mallocN(sizeof(*sdbind->vert_inds) * bpoly->numverts, "SDefTriVertInds"); + if (sdbind->vert_inds == NULL) { + data->success = MOD_SDEF_BIND_RESULT_MEM_ERR; + return; + } + + sortPolyVertsTri(sdbind->vert_inds, &data->mloop[bpoly->loopstart], bpoly->edge_vert_inds[0], bpoly->numverts); + + copy_v3_v3(v1, data->targetCos[sdbind->vert_inds[0]]); + copy_v3_v3(v2, data->targetCos[sdbind->vert_inds[1]]); + copy_v3_v3(v3, data->targetCos[sdbind->vert_inds[2]]); + + mid_v3_v3v3v3(cent, v1, v2, v3); + normal_tri_v3(norm, v1, v2, v3); + + add_v3_v3v3(tmp_vec, point_co, bpoly->normal); + + /* We are sure the line is not parallel to the plane. + * Checking return value just to avoid warning... */ + if (!isect_line_plane_v3(point_co_proj, point_co, tmp_vec, cent, norm)) { + BLI_assert(false); + } + + interp_weights_tri_v3(sdbind->vert_weights, v1, v2, v3, point_co_proj); + + sdbind->normal_dist = computeNormalDisplacement(point_co, point_co_proj, bpoly->normal); + + sdbind++; + i++; + } + } + } + } + + freeBindData(bwdata); +} + +static bool surfacedeformBind(SurfaceDeformModifierData *smd, float (*vertexCos)[3], + unsigned int numverts, unsigned int tnumpoly, unsigned int tnumverts, DerivedMesh *tdm) +{ + BVHTreeFromMesh treeData = {NULL}; + const MVert *mvert = tdm->getVertArray(tdm); + const MPoly *mpoly = tdm->getPolyArray(tdm); + const MEdge *medge = tdm->getEdgeArray(tdm); + const MLoop *mloop = tdm->getLoopArray(tdm); + unsigned int tnumedges = tdm->getNumEdges(tdm); + int adj_result; + SDefAdjacencyArray *vert_edges; + SDefAdjacency *adj_array; + SDefEdgePolys *edge_polys; + + vert_edges = MEM_callocN(sizeof(*vert_edges) * tnumverts, "SDefVertEdgeMap"); + if (vert_edges == NULL) { + modifier_setError((ModifierData *)smd, "Out of memory"); + return false; + } + + adj_array = MEM_mallocN(sizeof(*adj_array) * tnumedges * 2, "SDefVertEdge"); + if (adj_array == NULL) { + modifier_setError((ModifierData *)smd, "Out of memory"); + MEM_freeN(vert_edges); + return false; + } + + edge_polys = MEM_callocN(sizeof(*edge_polys) * tnumedges, "SDefEdgeFaceMap"); + if (edge_polys == NULL) { + modifier_setError((ModifierData *)smd, "Out of memory"); + MEM_freeN(vert_edges); + MEM_freeN(adj_array); + return false; + } + + smd->verts = MEM_mallocN(sizeof(*smd->verts) * numverts, "SDefBindVerts"); + if (smd->verts == NULL) { + modifier_setError((ModifierData *)smd, "Out of memory"); + freeAdjacencyMap(vert_edges, adj_array, edge_polys); + return false; + } + + bvhtree_from_mesh_looptri(&treeData, tdm, 0.0, 2, 6); + if (treeData.tree == NULL) { + modifier_setError((ModifierData *)smd, "Out of memory"); + freeAdjacencyMap(vert_edges, adj_array, edge_polys); + MEM_freeN(smd->verts); + smd->verts = NULL; + return false; + } + + adj_result = buildAdjacencyMap(mpoly, medge, mloop, tnumpoly, tnumedges, vert_edges, adj_array, edge_polys); + + if (adj_result == MOD_SDEF_BIND_RESULT_NONMANY_ERR) { + modifier_setError((ModifierData *)smd, "Target has edges with more than two polygons"); + freeAdjacencyMap(vert_edges, adj_array, edge_polys); + free_bvhtree_from_mesh(&treeData); + MEM_freeN(smd->verts); + smd->verts = NULL; + return false; + } + + smd->numverts = numverts; + smd->numpoly = tnumpoly; + + SDefBindCalcData data = {.treeData = &treeData, + .vert_edges = vert_edges, + .edge_polys = edge_polys, + .mpoly = mpoly, + .medge = medge, + .mloop = mloop, + .looptri = tdm->getLoopTriArray(tdm), + .targetCos = MEM_mallocN(sizeof(float[3]) * tnumverts, "SDefTargetBindVertArray"), + .bind_verts = smd->verts, + .vertexCos = vertexCos, + .falloff = smd->falloff, + .success = MOD_SDEF_BIND_RESULT_SUCCESS}; + + if (data.targetCos == NULL) { + modifier_setError((ModifierData *)smd, "Out of memory"); + freeData((ModifierData *)smd); + return false; + } + + invert_m4_m4(data.imat, smd->mat); + + for (int i = 0; i < tnumverts; i++) { + mul_v3_m4v3(data.targetCos[i], smd->mat, mvert[i].co); + } + + BLI_task_parallel_range_ex(0, numverts, &data, NULL, 0, bindVert, + numverts > 10000, false); + + MEM_freeN(data.targetCos); + + if (data.success == MOD_SDEF_BIND_RESULT_MEM_ERR) { + modifier_setError((ModifierData *)smd, "Out of memory"); + freeData((ModifierData *)smd); + } + else if (data.success == MOD_SDEF_BIND_RESULT_NONMANY_ERR) { + modifier_setError((ModifierData *)smd, "Target has edges with more than two polygons"); + freeData((ModifierData *)smd); + } + else if (data.success == MOD_SDEF_BIND_RESULT_CONCAVE_ERR) { + modifier_setError((ModifierData *)smd, "Target contains concave polygons"); + freeData((ModifierData *)smd); + } + else if (data.success == MOD_SDEF_BIND_RESULT_OVERLAP_ERR) { + modifier_setError((ModifierData *)smd, "Target contains overlapping verts"); + freeData((ModifierData *)smd); + } + else if (data.success == MOD_SDEF_BIND_RESULT_GENERIC_ERR) { + /* I know this message is vague, but I could not think of a way + * to explain this whith a reasonably sized message. + * Though it shouldn't really matter all that much, + * because this is very unlikely to occur */ + modifier_setError((ModifierData *)smd, "Target contains invalid polygons"); + freeData((ModifierData *)smd); + } + + freeAdjacencyMap(vert_edges, adj_array, edge_polys); + free_bvhtree_from_mesh(&treeData); + + return data.success == 1; +} + +static void deformVert(void *userdata, void *UNUSED(userdata_chunk), const int index, const int UNUSED(threadid)) +{ + const SDefDeformData * const data = (SDefDeformData *)userdata; + const SDefBind *sdbind = data->bind_verts[index].binds; + float * const vertexCos = data->vertexCos[index]; + float norm[3], temp[3]; + + zero_v3(vertexCos); + + for (int j = 0; j < data->bind_verts[index].numbinds; j++, sdbind++) { + /* Mode-generic operations (allocate poly coordinates) */ + float (*coords)[3] = MEM_mallocN(sizeof(*coords) * sdbind->numverts, "SDefDoPolyCoords"); + + for (int k = 0; k < sdbind->numverts; k++) { + copy_v3_v3(coords[k], data->targetCos[sdbind->vert_inds[k]]); + } + + normal_poly_v3(norm, coords, sdbind->numverts); + zero_v3(temp); + + /* ---------- looptri mode ---------- */ + if (sdbind->mode == MOD_SDEF_MODE_LOOPTRI) { + madd_v3_v3fl(temp, data->targetCos[sdbind->vert_inds[0]], sdbind->vert_weights[0]); + madd_v3_v3fl(temp, data->targetCos[sdbind->vert_inds[1]], sdbind->vert_weights[1]); + madd_v3_v3fl(temp, data->targetCos[sdbind->vert_inds[2]], sdbind->vert_weights[2]); + } + else { + /* ---------- ngon mode ---------- */ + if (sdbind->mode == MOD_SDEF_MODE_NGON) { + for (int k = 0; k < sdbind->numverts; k++) { + madd_v3_v3fl(temp, coords[k], sdbind->vert_weights[k]); + } + } + + /* ---------- centroid mode ---------- */ + else if (sdbind->mode == MOD_SDEF_MODE_CENTROID) { + float cent[3]; + mid_v3_v3_array(cent, coords, sdbind->numverts); + + madd_v3_v3fl(temp, data->targetCos[sdbind->vert_inds[0]], sdbind->vert_weights[0]); + madd_v3_v3fl(temp, data->targetCos[sdbind->vert_inds[1]], sdbind->vert_weights[1]); + madd_v3_v3fl(temp, cent, sdbind->vert_weights[2]); + } + } + + MEM_freeN(coords); + + /* Apply normal offset (generic for all modes) */ + madd_v3_v3fl(temp, norm, sdbind->normal_dist); + + madd_v3_v3fl(vertexCos, temp, sdbind->influence); + } +} + +static void surfacedeformModifier_do(ModifierData *md, float (*vertexCos)[3], unsigned int numverts, Object *ob) +{ + SurfaceDeformModifierData *smd = (SurfaceDeformModifierData *)md; + DerivedMesh *tdm; + unsigned int tnumverts, tnumpoly; + + /* Exit function if bind flag is not set (free bind data if any) */ + if (!(smd->flags & MOD_SDEF_BIND)) { + freeData(md); + return; + } + + /* Handle target mesh both in and out of edit mode */ + if (smd->target == md->scene->obedit) { + BMEditMesh *em = BKE_editmesh_from_object(smd->target); + tdm = em->derivedFinal; + } + else { + tdm = smd->target->derivedFinal; + } + + tnumverts = tdm->getNumVerts(tdm); + tnumpoly = tdm->getNumPolys(tdm); + + /* If not bound, execute bind */ + if (!(smd->verts)) { + float tmp_mat[4][4]; + + invert_m4_m4(tmp_mat, ob->obmat); + mul_m4_m4m4(smd->mat, tmp_mat, smd->target->obmat); + + if (!surfacedeformBind(smd, vertexCos, numverts, tnumpoly, tnumverts, tdm)) { + smd->flags &= ~MOD_SDEF_BIND; + return; + } + } + + /* Poly count checks */ + if (smd->numverts != numverts) { + modifier_setError(md, "Verts changed from %u to %u", smd->numverts, numverts); + tdm->release(tdm); + return; + } + else if (smd->numpoly != tnumpoly) { + modifier_setError(md, "Target polygons changed from %u to %u", smd->numpoly, tnumpoly); + tdm->release(tdm); + return; + } + + /* Actual vertex location update starts here */ + SDefDeformData data = {.bind_verts = smd->verts, + .targetCos = MEM_mallocN(sizeof(float[3]) * tnumverts, "SDefTargetVertArray"), + .vertexCos = vertexCos}; + + if (data.targetCos != NULL) { + bool tdm_vert_alloc; + const MVert * const mvert = DM_get_vert_array(tdm, &tdm_vert_alloc); + + for (int i = 0; i < tnumverts; i++) { + mul_v3_m4v3(data.targetCos[i], smd->mat, mvert[i].co); + } + + BLI_task_parallel_range_ex(0, numverts, &data, NULL, 0, deformVert, + numverts > 10000, false); + + if (tdm_vert_alloc) { + MEM_freeN((void *)mvert); + } + + MEM_freeN(data.targetCos); + } + + tdm->release(tdm); +} + +static void deformVerts(ModifierData *md, Object *ob, + DerivedMesh *UNUSED(derivedData), + float (*vertexCos)[3], int numVerts, + ModifierApplyFlag UNUSED(flag)) +{ + surfacedeformModifier_do(md, vertexCos, numVerts, ob); +} + +static void deformVertsEM(ModifierData *md, Object *ob, + struct BMEditMesh *UNUSED(editData), + DerivedMesh *UNUSED(derivedData), + float (*vertexCos)[3], int numVerts) +{ + surfacedeformModifier_do(md, vertexCos, numVerts, ob); +} + +static bool isDisabled(ModifierData *md, int UNUSED(useRenderParams)) +{ + SurfaceDeformModifierData *smd = (SurfaceDeformModifierData *)md; + + return !smd->target; +} + +ModifierTypeInfo modifierType_SurfaceDeform = { + /* name */ "Surface Deform", + /* structName */ "SurfaceDeformModifierData", + /* structSize */ sizeof(SurfaceDeformModifierData), + /* type */ eModifierTypeType_OnlyDeform, + /* flags */ eModifierTypeFlag_AcceptsMesh | + eModifierTypeFlag_SupportsEditmode, + + /* copyData */ copyData, + /* deformVerts */ deformVerts, + /* deformMatrices */ NULL, + /* deformVertsEM */ deformVertsEM, + /* deformMatricesEM */ NULL, + /* applyModifier */ NULL, + /* applyModifierEM */ NULL, + /* initData */ initData, + /* requiredDataMask */ NULL, + /* freeData */ freeData, + /* isDisabled */ isDisabled, + /* updateDepgraph */ updateDepgraph, + /* updateDepsgraph */ updateDepsgraph, + /* dependsOnTime */ NULL, + /* dependsOnNormals */ NULL, + /* foreachObjectLink */ foreachObjectLink, + /* foreachIDLink */ NULL, + /* foreachTexLink */ NULL, +}; diff --git a/source/blender/modifiers/intern/MOD_util.c b/source/blender/modifiers/intern/MOD_util.c index 93414562ccf..ded1f0b77e6 100644 --- a/source/blender/modifiers/intern/MOD_util.c +++ b/source/blender/modifiers/intern/MOD_util.c @@ -287,5 +287,6 @@ void modifier_type_init(ModifierTypeInfo *types[]) INIT_TYPE(NormalEdit); INIT_TYPE(CorrectiveSmooth); INIT_TYPE(MeshSequenceCache); + INIT_TYPE(SurfaceDeform); #undef INIT_TYPE } diff --git a/source/blender/nodes/composite/nodes/node_composite_glare.c b/source/blender/nodes/composite/nodes/node_composite_glare.c index c512ea49586..76020e55463 100644 --- a/source/blender/nodes/composite/nodes/node_composite_glare.c +++ b/source/blender/nodes/composite/nodes/node_composite_glare.c @@ -50,7 +50,8 @@ static void node_composit_init_glare(bNodeTree *UNUSED(ntree), bNode *node) ndg->colmod = 0.25; ndg->mix = 0; ndg->threshold = 1; - ndg->angle = 4; + ndg->star_45 = true; + ndg->streaks = 4; ndg->angle_ofs = 0.0f; ndg->fade = 0.9; ndg->size = 8; diff --git a/source/blender/nodes/shader/nodes/node_shader_fresnel.c b/source/blender/nodes/shader/nodes/node_shader_fresnel.c index d5e11795fc0..5a9e33a4053 100644 --- a/source/blender/nodes/shader/nodes/node_shader_fresnel.c +++ b/source/blender/nodes/shader/nodes/node_shader_fresnel.c @@ -64,10 +64,11 @@ static void node_shader_exec_fresnel(void *data, int UNUSED(thread), bNode *UNUS copy_v3_v3(n, shi->vn); } - if(shi->use_world_space_shading) + if (shi->use_world_space_shading) { mul_mat3_m4_v3((float (*)[4])RE_render_current_get_matrix(RE_VIEW_MATRIX), n); + } - out[0]->vec[0] = RE_fresnel_dielectric(shi->view, n, shi->flippednor ? 1/eta : eta); + out[0]->vec[0] = RE_fresnel_dielectric(shi->view, n, shi->flippednor ? 1 / eta : eta); } /* node type definition */ diff --git a/source/blender/nodes/shader/nodes/node_shader_layer_weight.c b/source/blender/nodes/shader/nodes/node_shader_layer_weight.c index 90e2625b961..a0b2408a7bb 100644 --- a/source/blender/nodes/shader/nodes/node_shader_layer_weight.c +++ b/source/blender/nodes/shader/nodes/node_shader_layer_weight.c @@ -69,7 +69,7 @@ static void node_shader_exec_layer_weight(void *data, int UNUSED(thread), bNode if (shi->use_world_space_shading) mul_mat3_m4_v3((float (*)[4])RE_render_current_get_matrix(RE_VIEW_MATRIX), n); - out[0]->vec[0] = RE_fresnel_dielectric(shi->view, n, shi->flippednor ? eta : 1/eta); + out[0]->vec[0] = RE_fresnel_dielectric(shi->view, n, shi->flippednor ? eta : 1 / eta); float facing = fabs(dot_v3v3(shi->view, n)); if (blend != 0.5) { diff --git a/source/blender/nodes/shader/nodes/node_shader_normal_map.c b/source/blender/nodes/shader/nodes/node_shader_normal_map.c index 48d1688c386..e0bf34f42e4 100644 --- a/source/blender/nodes/shader/nodes/node_shader_normal_map.c +++ b/source/blender/nodes/shader/nodes/node_shader_normal_map.c @@ -63,7 +63,7 @@ static void node_shader_exec_normal_map(void *data, int UNUSED(thread), bNode *n CLAMP_MIN(strength, 0.0f); - float *N = shi->vno; + float *N = shi->nmapnorm; int uv_index = 0; switch (nm->space) { case SHD_NORMAL_MAP_TANGENT: diff --git a/source/blender/nodes/shader/nodes/node_shader_tex_brick.c b/source/blender/nodes/shader/nodes/node_shader_tex_brick.c index 0be47c4f751..1dfebc45d60 100644 --- a/source/blender/nodes/shader/nodes/node_shader_tex_brick.c +++ b/source/blender/nodes/shader/nodes/node_shader_tex_brick.c @@ -64,7 +64,7 @@ static void node_shader_init_tex_brick(bNodeTree *UNUSED(ntree), bNode *node) for (bNodeSocket *sock = node->inputs.first; sock; sock = sock->next) { if (STREQ(sock->name, "Mortar Smooth")) { - ((bNodeSocketValueFloat*)sock->default_value)->value = 0.1f; + ((bNodeSocketValueFloat *)sock->default_value)->value = 0.1f; } } } diff --git a/source/blender/physics/intern/BPH_mass_spring.cpp b/source/blender/physics/intern/BPH_mass_spring.cpp index 359395b63c4..b694b6e994d 100644 --- a/source/blender/physics/intern/BPH_mass_spring.cpp +++ b/source/blender/physics/intern/BPH_mass_spring.cpp @@ -333,19 +333,14 @@ static int UNUSED_FUNCTION(cloth_calc_helper_forces)(Object *UNUSED(ob), ClothMo return 1; } -BLI_INLINE void cloth_calc_spring_force(ClothModifierData *clmd, ClothSpring *s, float time) +BLI_INLINE void cloth_calc_spring_force(ClothModifierData *clmd, ClothSpring *s) { Cloth *cloth = clmd->clothObject; ClothSimSettings *parms = clmd->sim_parms; Implicit_Data *data = cloth->implicit; - ClothVertex *verts = cloth->verts; bool no_compress = parms->flags & CLOTH_SIMSETTINGS_FLAG_NO_SPRING_COMPRESS; - zero_v3(s->f); - zero_m3(s->dfdx); - zero_m3(s->dfdv); - s->flags &= ~CLOTH_SPRING_FLAG_NEEDED; // calculate force of structural + shear springs @@ -361,31 +356,13 @@ BLI_INLINE void cloth_calc_spring_force(ClothModifierData *clmd, ClothSpring *s, if (s->type & CLOTH_SPRING_TYPE_SEWING) { // TODO: verify, half verified (couldn't see error) // sewing springs usually have a large distance at first so clamp the force so we don't get tunnelling through colission objects - BPH_mass_spring_force_spring_linear(data, s->ij, s->kl, s->restlen, k, parms->Cdis, no_compress, parms->max_sewing, s->f, s->dfdx, s->dfdv); + BPH_mass_spring_force_spring_linear(data, s->ij, s->kl, s->restlen, k, parms->Cdis, no_compress, parms->max_sewing); } else { - BPH_mass_spring_force_spring_linear(data, s->ij, s->kl, s->restlen, k, parms->Cdis, no_compress, 0.0f, s->f, s->dfdx, s->dfdv); + BPH_mass_spring_force_spring_linear(data, s->ij, s->kl, s->restlen, k, parms->Cdis, no_compress, 0.0f); } #endif } - else if (s->type & CLOTH_SPRING_TYPE_GOAL) { -#ifdef CLOTH_FORCE_SPRING_GOAL - float goal_x[3], goal_v[3]; - float k, scaling; - - s->flags |= CLOTH_SPRING_FLAG_NEEDED; - - // current_position = xold + t * (newposition - xold) - /* divide by time_scale to prevent goal vertices' delta locations from being multiplied */ - interp_v3_v3v3(goal_x, verts[s->ij].xold, verts[s->ij].xconst, time / parms->time_scale); - sub_v3_v3v3(goal_v, verts[s->ij].xconst, verts[s->ij].xold); // distance covered over dt==1 - - scaling = parms->goalspring + s->stiffness * fabsf(parms->max_struct - parms->goalspring); - k = verts[s->ij].goal * scaling / (parms->avg_spring_len + FLT_EPSILON); - - BPH_mass_spring_force_spring_goal(data, s->ij, goal_x, goal_v, k, parms->goalfrict * 0.01f, s->f, s->dfdx, s->dfdv); -#endif - } else if (s->type & CLOTH_SPRING_TYPE_BENDING) { /* calculate force of bending springs */ #ifdef CLOTH_FORCE_SPRING_BEND float kb, cb, scaling; @@ -398,7 +375,7 @@ BLI_INLINE void cloth_calc_spring_force(ClothModifierData *clmd, ClothSpring *s, // Fix for [#45084] for cloth stiffness must have cb proportional to kb cb = kb * parms->bending_damping; - BPH_mass_spring_force_spring_bending(data, s->ij, s->kl, s->restlen, kb, cb, s->f, s->dfdx, s->dfdv); + BPH_mass_spring_force_spring_bending(data, s->ij, s->kl, s->restlen, kb, cb); #endif } else if (s->type & CLOTH_SPRING_TYPE_BENDING_ANG) { @@ -474,9 +451,24 @@ static void cloth_calc_force(ClothModifierData *clmd, float UNUSED(frame), ListB /* scale gravity force */ mul_v3_v3fl(gravity, clmd->scene->physics_settings.gravity, 0.001f * clmd->sim_parms->effector_weights->global_gravity); } + vert = cloth->verts; for (i = 0; i < cloth->mvert_num; i++, vert++) { BPH_mass_spring_force_gravity(data, i, vert->mass, gravity); + + /* Vertex goal springs */ + if ((!(vert->flags & CLOTH_VERT_FLAG_PINNED)) && (vert->goal > FLT_EPSILON)) { + float goal_x[3], goal_v[3]; + float k; + + /* divide by time_scale to prevent goal vertices' delta locations from being multiplied */ + interp_v3_v3v3(goal_x, vert->xold, vert->xconst, time / clmd->sim_parms->time_scale); + sub_v3_v3v3(goal_v, vert->xconst, vert->xold); /* distance covered over dt==1 */ + + k = vert->goal * clmd->sim_parms->goalspring / (clmd->sim_parms->avg_spring_len + FLT_EPSILON); + + BPH_mass_spring_force_spring_goal(data, i, goal_x, goal_v, k, clmd->sim_parms->goalfrict * 0.01f); + } } #endif @@ -544,8 +536,9 @@ static void cloth_calc_force(ClothModifierData *clmd, float UNUSED(frame), ListB for (LinkNode *link = cloth->springs; link; link = link->next) { ClothSpring *spring = (ClothSpring *)link->link; // only handle active springs - if (!(spring->flags & CLOTH_SPRING_FLAG_DEACTIVATE)) - cloth_calc_spring_force(clmd, spring, time); + if (!(spring->flags & CLOTH_SPRING_FLAG_DEACTIVATE)) { + cloth_calc_spring_force(clmd, spring); + } } } diff --git a/source/blender/physics/intern/implicit.h b/source/blender/physics/intern/implicit.h index d1a75ca5297..2f62ab98e12 100644 --- a/source/blender/physics/intern/implicit.h +++ b/source/blender/physics/intern/implicit.h @@ -114,19 +114,15 @@ void BPH_mass_spring_force_edge_wind(struct Implicit_Data *data, int v1, int v2, void BPH_mass_spring_force_vertex_wind(struct Implicit_Data *data, int v, float radius, const float (*winvec)[3]); /* Linear spring force between two points */ bool BPH_mass_spring_force_spring_linear(struct Implicit_Data *data, int i, int j, float restlen, - float stiffness, float damping, bool no_compress, float clamp_force, - float r_f[3], float r_dfdx[3][3], float r_dfdv[3][3]); + float stiffness, float damping, bool no_compress, float clamp_force); /* Bending force, forming a triangle at the base of two structural springs */ -bool BPH_mass_spring_force_spring_bending(struct Implicit_Data *data, int i, int j, float restlen, - float kb, float cb, - float r_f[3], float r_dfdx[3][3], float r_dfdv[3][3]); +bool BPH_mass_spring_force_spring_bending(struct Implicit_Data *data, int i, int j, float restlen, float kb, float cb); /* Angular bending force based on local target vectors */ bool BPH_mass_spring_force_spring_bending_angular(struct Implicit_Data *data, int i, int j, int k, const float target[3], float stiffness, float damping); /* Global goal spring */ bool BPH_mass_spring_force_spring_goal(struct Implicit_Data *data, int i, const float goal_x[3], const float goal_v[3], - float stiffness, float damping, - float r_f[3], float r_dfdx[3][3], float r_dfdv[3][3]); + float stiffness, float damping); /* ======== Hair Volumetric Forces ======== */ diff --git a/source/blender/physics/intern/implicit_blender.c b/source/blender/physics/intern/implicit_blender.c index 2ad8ee0547f..16cd335dc0c 100644 --- a/source/blender/physics/intern/implicit_blender.c +++ b/source/blender/physics/intern/implicit_blender.c @@ -1579,8 +1579,7 @@ BLI_INLINE void apply_spring(Implicit_Data *data, int i, int j, const float f[3] } bool BPH_mass_spring_force_spring_linear(Implicit_Data *data, int i, int j, float restlen, - float stiffness, float damping, bool no_compress, float clamp_force, - float r_f[3], float r_dfdx[3][3], float r_dfdv[3][3]) + float stiffness, float damping, bool no_compress, float clamp_force) { float extent[3], length, dir[3], vel[3]; @@ -1608,25 +1607,15 @@ bool BPH_mass_spring_force_spring_linear(Implicit_Data *data, int i, int j, floa apply_spring(data, i, j, f, dfdx, dfdv); - if (r_f) copy_v3_v3(r_f, f); - if (r_dfdx) copy_m3_m3(r_dfdx, dfdx); - if (r_dfdv) copy_m3_m3(r_dfdv, dfdv); - return true; } else { - if (r_f) zero_v3(r_f); - if (r_dfdx) zero_m3(r_dfdx); - if (r_dfdv) zero_m3(r_dfdv); - return false; } } /* See "Stable but Responsive Cloth" (Choi, Ko 2005) */ -bool BPH_mass_spring_force_spring_bending(Implicit_Data *data, int i, int j, float restlen, - float kb, float cb, - float r_f[3], float r_dfdx[3][3], float r_dfdv[3][3]) +bool BPH_mass_spring_force_spring_bending(Implicit_Data *data, int i, int j, float restlen, float kb, float cb) { float extent[3], length, dir[3], vel[3]; @@ -1646,17 +1635,9 @@ bool BPH_mass_spring_force_spring_bending(Implicit_Data *data, int i, int j, flo apply_spring(data, i, j, f, dfdx, dfdv); - if (r_f) copy_v3_v3(r_f, f); - if (r_dfdx) copy_m3_m3(r_dfdx, dfdx); - if (r_dfdv) copy_m3_m3(r_dfdv, dfdv); - return true; } else { - if (r_f) zero_v3(r_f); - if (r_dfdx) zero_m3(r_dfdx); - if (r_dfdv) zero_m3(r_dfdv); - return false; } } @@ -1945,8 +1926,7 @@ bool BPH_mass_spring_force_spring_bending_angular(Implicit_Data *data, int i, in } bool BPH_mass_spring_force_spring_goal(Implicit_Data *data, int i, const float goal_x[3], const float goal_v[3], - float stiffness, float damping, - float r_f[3], float r_dfdx[3][3], float r_dfdv[3][3]) + float stiffness, float damping) { float root_goal_x[3], root_goal_v[3], extent[3], length, dir[3], vel[3]; float f[3], dfdx[3][3], dfdv[3][3]; @@ -1973,17 +1953,9 @@ bool BPH_mass_spring_force_spring_goal(Implicit_Data *data, int i, const float g add_m3_m3m3(data->dFdX[i].m, data->dFdX[i].m, dfdx); add_m3_m3m3(data->dFdV[i].m, data->dFdV[i].m, dfdv); - if (r_f) copy_v3_v3(r_f, f); - if (r_dfdx) copy_m3_m3(r_dfdx, dfdx); - if (r_dfdv) copy_m3_m3(r_dfdv, dfdv); - return true; } else { - if (r_f) zero_v3(r_f); - if (r_dfdx) zero_m3(r_dfdx); - if (r_dfdv) zero_m3(r_dfdv); - return false; } } diff --git a/source/blender/python/BPY_extern.h b/source/blender/python/BPY_extern.h index 3148dab3c50..b4c36a7c516 100644 --- a/source/blender/python/BPY_extern.h +++ b/source/blender/python/BPY_extern.h @@ -75,7 +75,8 @@ void BPY_thread_restore(BPy_ThreadStatePtr tstate); bool BPY_execute_filepath(struct bContext *C, const char *filepath, struct ReportList *reports); bool BPY_execute_text(struct bContext *C, struct Text *text, struct ReportList *reports, const bool do_jump); -bool BPY_execute_string_as_number(struct bContext *C, const char *expr, double *value, const bool verbose); +bool BPY_execute_string_as_number(struct bContext *C, const char *expr, const bool verbose, double *r_value); +bool BPY_execute_string_as_string(struct bContext *C, const char *expr, const bool verbose, char **r_value); bool BPY_execute_string_ex(struct bContext *C, const char *expr, bool use_eval); bool BPY_execute_string(struct bContext *C, const char *expr); diff --git a/source/blender/python/generic/py_capi_utils.c b/source/blender/python/generic/py_capi_utils.c index 7b2d58a1268..2e789d6d4b3 100644 --- a/source/blender/python/generic/py_capi_utils.c +++ b/source/blender/python/generic/py_capi_utils.c @@ -918,11 +918,11 @@ char *PyC_FlagSet_AsString(PyC_FlagSet *item) return cstring; } -int PyC_FlagSet_ValueFromID_int(PyC_FlagSet *item, const char *identifier, int *value) +int PyC_FlagSet_ValueFromID_int(PyC_FlagSet *item, const char *identifier, int *r_value) { for ( ; item->identifier; item++) { if (STREQ(item->identifier, identifier)) { - *value = item->value; + *r_value = item->value; return 1; } } @@ -930,9 +930,9 @@ int PyC_FlagSet_ValueFromID_int(PyC_FlagSet *item, const char *identifier, int * return 0; } -int PyC_FlagSet_ValueFromID(PyC_FlagSet *item, const char *identifier, int *value, const char *error_prefix) +int PyC_FlagSet_ValueFromID(PyC_FlagSet *item, const char *identifier, int *r_value, const char *error_prefix) { - if (PyC_FlagSet_ValueFromID_int(item, identifier, value) == 0) { + if (PyC_FlagSet_ValueFromID_int(item, identifier, r_value) == 0) { const char *enum_str = PyC_FlagSet_AsString(item); PyErr_Format(PyExc_ValueError, "%s: '%.200s' not found in (%s)", @@ -1006,7 +1006,7 @@ PyObject *PyC_FlagSet_FromBitfield(PyC_FlagSet *items, int flag) * * \note it is caller's responsibility to acquire & release GIL! */ -bool PyC_RunString_AsNumber(const char *expr, double *value, const char *filename) +bool PyC_RunString_AsNumber(const char *expr, const char *filename, double *r_value) { PyObject *py_dict, *mod, *retval; bool ok = true; @@ -1058,10 +1058,10 @@ bool PyC_RunString_AsNumber(const char *expr, double *value, const char *filenam ok = false; } else if (!isfinite(val)) { - *value = 0.0; + *r_value = 0.0; } else { - *value = val; + *r_value = val; } } @@ -1070,4 +1070,41 @@ bool PyC_RunString_AsNumber(const char *expr, double *value, const char *filenam return ok; } +bool PyC_RunString_AsString(const char *expr, const char *filename, char **r_value) +{ + PyObject *py_dict, *retval; + bool ok = true; + PyObject *main_mod = NULL; + + PyC_MainModule_Backup(&main_mod); + + py_dict = PyC_DefaultNameSpace(filename); + + retval = PyRun_String(expr, Py_eval_input, py_dict, py_dict); + + if (retval == NULL) { + ok = false; + } + else { + const char *val; + Py_ssize_t val_len; + + val = _PyUnicode_AsStringAndSize(retval, &val_len); + if (val == NULL && PyErr_Occurred()) { + ok = false; + } + else { + char *val_alloc = MEM_mallocN(val_len + 1, __func__); + memcpy(val_alloc, val, val_len + 1); + *r_value = val_alloc; + } + + Py_DECREF(retval); + } + + PyC_MainModule_Restore(main_mod); + + return ok; +} + #endif /* #ifndef MATH_STANDALONE */ diff --git a/source/blender/python/generic/py_capi_utils.h b/source/blender/python/generic/py_capi_utils.h index 04cfc8801eb..3f89e1d82a0 100644 --- a/source/blender/python/generic/py_capi_utils.h +++ b/source/blender/python/generic/py_capi_utils.h @@ -75,12 +75,13 @@ typedef struct PyC_FlagSet { } PyC_FlagSet; char *PyC_FlagSet_AsString(PyC_FlagSet *item); -int PyC_FlagSet_ValueFromID_int(PyC_FlagSet *item, const char *identifier, int *value); -int PyC_FlagSet_ValueFromID(PyC_FlagSet *item, const char *identifier, int *value, const char *error_prefix); +int PyC_FlagSet_ValueFromID_int(PyC_FlagSet *item, const char *identifier, int *r_value); +int PyC_FlagSet_ValueFromID(PyC_FlagSet *item, const char *identifier, int *r_value, const char *error_prefix); int PyC_FlagSet_ToBitfield(PyC_FlagSet *items, PyObject *value, int *r_value, const char *error_prefix); PyObject *PyC_FlagSet_FromBitfield(PyC_FlagSet *items, int flag); -bool PyC_RunString_AsNumber(const char *expr, double *value, const char *filename); +bool PyC_RunString_AsNumber(const char *expr, const char *filename, double *r_value); +bool PyC_RunString_AsString(const char *expr, const char *filename, char **r_value); int PyC_ParseBool(PyObject *o, void *p); diff --git a/source/blender/python/intern/bpy_interface.c b/source/blender/python/intern/bpy_interface.c index 311f621e13b..55e477b0214 100644 --- a/source/blender/python/intern/bpy_interface.c +++ b/source/blender/python/intern/bpy_interface.c @@ -572,21 +572,58 @@ void BPY_DECREF_RNA_INVALIDATE(void *pyob_ptr) /** * \return success */ -bool BPY_execute_string_as_number(bContext *C, const char *expr, double *value, const bool verbose) +bool BPY_execute_string_as_number(bContext *C, const char *expr, const bool verbose, double *r_value) { PyGILState_STATE gilstate; bool ok = true; - if (!value || !expr) return -1; + if (!r_value || !expr) { + return -1; + } + + if (expr[0] == '\0') { + *r_value = 0.0; + return ok; + } + + bpy_context_set(C, &gilstate); + + ok = PyC_RunString_AsNumber(expr, "<blender button>", r_value); + + if (ok == false) { + if (verbose) { + BPy_errors_to_report_ex(CTX_wm_reports(C), false, false); + } + else { + PyErr_Clear(); + } + } + + bpy_context_clear(C, &gilstate); + + return ok; +} + +/** + * \return success + */ +bool BPY_execute_string_as_string(bContext *C, const char *expr, const bool verbose, char **r_value) +{ + PyGILState_STATE gilstate; + bool ok = true; + + if (!r_value || !expr) { + return -1; + } if (expr[0] == '\0') { - *value = 0.0; + *r_value = NULL; return ok; } bpy_context_set(C, &gilstate); - ok = PyC_RunString_AsNumber(expr, value, "<blender button>"); + ok = PyC_RunString_AsString(expr, "<blender button>", r_value); if (ok == false) { if (verbose) { @@ -602,6 +639,7 @@ bool BPY_execute_string_as_number(bContext *C, const char *expr, double *value, return ok; } + bool BPY_execute_string_ex(bContext *C, const char *expr, bool use_eval) { PyGILState_STATE gilstate; diff --git a/source/blender/python/intern/bpy_utils_units.c b/source/blender/python/intern/bpy_utils_units.c index 974d7c5549c..e0c70483e3c 100644 --- a/source/blender/python/intern/bpy_utils_units.c +++ b/source/blender/python/intern/bpy_utils_units.c @@ -201,7 +201,7 @@ static PyObject *bpyunits_to_value(PyObject *UNUSED(self), PyObject *args, PyObj bUnit_ReplaceString(str, (int)str_len, uref, scale, usys, ucat); - if (!PyC_RunString_AsNumber(str, &result, "<bpy_units_api>")) { + if (!PyC_RunString_AsNumber(str, "<bpy_units_api>", &result)) { if (PyErr_Occurred()) { PyErr_Print(); PyErr_Clear(); diff --git a/source/blender/python/intern/gpu_offscreen.c b/source/blender/python/intern/gpu_offscreen.c index c4863b2a92f..7711ce18bd0 100644 --- a/source/blender/python/intern/gpu_offscreen.c +++ b/source/blender/python/intern/gpu_offscreen.c @@ -202,7 +202,7 @@ static PyObject *pygpu_offscreen_draw_view3d(BPy_GPUOffScreen *self, PyObject *a ARegion *ar; GPUFX *fx; GPUFXSettings fx_settings; - void *rv3d_mats; + struct RV3DMatrixStore *rv3d_mats; BPY_GPU_OFFSCREEN_CHECK_OBJ(self); diff --git a/source/blender/python/mathutils/mathutils_Matrix.c b/source/blender/python/mathutils/mathutils_Matrix.c index 4e980e4c0e6..bd44e77e7c6 100644 --- a/source/blender/python/mathutils/mathutils_Matrix.c +++ b/source/blender/python/mathutils/mathutils_Matrix.c @@ -1301,7 +1301,7 @@ PyDoc_STRVAR(Matrix_to_scale_doc, " :return: Return the scale of a matrix.\n" " :rtype: :class:`Vector`\n" "\n" -" .. note:: This method does not return negative a scale on any axis because it is not possible to obtain this data from the matrix alone.\n" +" .. note:: This method does not return a negative scale on any axis because it is not possible to obtain this data from the matrix alone.\n" ); static PyObject *Matrix_to_scale(MatrixObject *self) { @@ -1390,11 +1390,11 @@ PyDoc_STRVAR(Matrix_invert_doc, "\n" " Set the matrix to its inverse.\n" "\n" -" :arg fallback: Set the matrix to this value when the inverse can't be calculated\n" +" :arg fallback: Set the matrix to this value when the inverse cannot be calculated\n" " (instead of raising a :exc:`ValueError` exception).\n" " :type fallback: :class:`Matrix`\n" "\n" -" .. seealso:: <https://en.wikipedia.org/wiki/Inverse_matrix>\n" +" .. seealso:: `Inverse matrix <https://en.wikipedia.org/wiki/Inverse_matrix>` on Wikipedia.\n" ); static PyObject *Matrix_invert(MatrixObject *self, PyObject *args) { @@ -1505,7 +1505,7 @@ PyDoc_STRVAR(Matrix_invert_safe_doc, " If degenerated (e.g. zero scale on an axis), add some epsilon to its diagonal, to get an invertible one.\n" " If tweaked matrix is still degenerated, set to the identity matrix instead.\n" "\n" -" .. seealso:: <https://en.wikipedia.org/wiki/Inverse_matrix>\n" +" .. seealso:: `Inverse Matrix <https://en.wikipedia.org/wiki/Inverse_matrix>` on Wikipedia.\n" ); static PyObject *Matrix_invert_safe(MatrixObject *self) { @@ -1554,9 +1554,9 @@ PyDoc_STRVAR(Matrix_adjugate_doc, "\n" " Set the matrix to its adjugate.\n" "\n" -" .. note:: When the matrix cant be adjugated a :exc:`ValueError` exception is raised.\n" +" .. note:: When the matrix cannot be adjugated a :exc:`ValueError` exception is raised.\n" "\n" -" .. seealso:: <https://en.wikipedia.org/wiki/Adjugate_matrix>\n" +" .. seealso:: `Adjugate matrix <https://en.wikipedia.org/wiki/Adjugate_matrix>` on Wikipedia.\n" ); static PyObject *Matrix_adjugate(MatrixObject *self) { @@ -1733,7 +1733,7 @@ PyDoc_STRVAR(Matrix_determinant_doc, " :return: Return the determinant of a matrix.\n" " :rtype: float\n" "\n" -" .. seealso:: <https://en.wikipedia.org/wiki/Determinant>\n" +" .. seealso:: `Determinant <https://en.wikipedia.org/wiki/Determinant>` on Wikipedia.\n" ); static PyObject *Matrix_determinant(MatrixObject *self) { @@ -1755,7 +1755,7 @@ PyDoc_STRVAR(Matrix_transpose_doc, "\n" " Set the matrix to its transpose.\n" "\n" -" .. seealso:: <https://en.wikipedia.org/wiki/Transpose>\n" +" .. seealso:: `Transpose <https://en.wikipedia.org/wiki/Transpose>` on Wikipedia.\n" ); static PyObject *Matrix_transpose(MatrixObject *self) { @@ -1887,10 +1887,10 @@ PyDoc_STRVAR(Matrix_identity_doc, "\n" " Set the matrix to the identity matrix.\n" "\n" -" .. note:: An object with zero location and rotation, a scale of one,\n" +" .. note:: An object with a location and rotation of zero, and a scale of one\n" " will have an identity matrix.\n" "\n" -" .. seealso:: <https://en.wikipedia.org/wiki/Identity_matrix>\n" +" .. seealso:: `Identity matrix <https://en.wikipedia.org/wiki/Identity_matrix>` on Wikipedia.\n" ); static PyObject *Matrix_identity(MatrixObject *self) { diff --git a/source/blender/render/CMakeLists.txt b/source/blender/render/CMakeLists.txt index 9e40ab02ee4..569b207c966 100644 --- a/source/blender/render/CMakeLists.txt +++ b/source/blender/render/CMakeLists.txt @@ -35,6 +35,7 @@ set(INC ../makesdna ../makesrna ../physics + ../../../intern/atomic ../../../intern/guardedalloc ../../../intern/mikktspace ../../../intern/smoke/extern diff --git a/source/blender/render/intern/source/pointdensity.c b/source/blender/render/intern/source/pointdensity.c index a03ea9cb896..fb047aad897 100644 --- a/source/blender/render/intern/source/pointdensity.c +++ b/source/blender/render/intern/source/pointdensity.c @@ -983,11 +983,12 @@ void RE_point_density_minmax( } else { float radius[3] = {pd->radius, pd->radius, pd->radius}; - float *loc, *size; + BoundBox *bb = BKE_object_boundbox_get(object); - if (BKE_object_obdata_texspace_get(pd->object, NULL, &loc, &size, NULL)) { - sub_v3_v3v3(r_min, loc, size); - add_v3_v3v3(r_max, loc, size); + if (bb != NULL) { + BLI_assert((bb->flag & BOUNDBOX_DIRTY) == 0); + copy_v3_v3(r_min, bb->vec[0]); + copy_v3_v3(r_max, bb->vec[6]); /* Adjust texture space to include density points on the boundaries. */ sub_v3_v3(r_min, radius); add_v3_v3(r_max, radius); diff --git a/source/blender/render/intern/source/render_texture.c b/source/blender/render/intern/source/render_texture.c index b4a14f5337d..42e4d4f27f6 100644 --- a/source/blender/render/intern/source/render_texture.c +++ b/source/blender/render/intern/source/render_texture.c @@ -1112,14 +1112,15 @@ static int multitex(Tex *tex, const short which_output, struct ImagePool *pool, const bool skip_load_image, - const bool texnode_preview) + const bool texnode_preview, + const bool use_nodes) { float tmpvec[3]; int retval = 0; /* return value, int:0, col:1, nor:2, everything:3 */ texres->talpha = false; /* is set when image texture returns alpha (considered premul) */ - if (tex->use_nodes && tex->nodetree) { + if (use_nodes && tex->use_nodes && tex->nodetree) { retval = ntreeTexExecTree(tex->nodetree, texres, texvec, dxt, dyt, osatex, thread, tex, which_output, R.r.cfra, texnode_preview, NULL, NULL); } @@ -1239,7 +1240,8 @@ static int multitex_nodes_intern(Tex *tex, ImagePool *pool, const bool scene_color_manage, const bool skip_load_image, - const bool texnode_preview) + const bool texnode_preview, + const bool use_nodes) { if (tex==NULL) { memset(texres, 0, sizeof(TexResult)); @@ -1264,7 +1266,8 @@ static int multitex_nodes_intern(Tex *tex, which_output, pool, skip_load_image, - texnode_preview); + texnode_preview, + use_nodes); if (mtex->mapto & (MAP_COL+MAP_COLSPEC+MAP_COLMIR)) { ImBuf *ibuf = BKE_image_pool_acquire_ibuf(tex->ima, &tex->iuser, pool); @@ -1311,7 +1314,8 @@ static int multitex_nodes_intern(Tex *tex, which_output, pool, skip_load_image, - texnode_preview); + texnode_preview, + use_nodes); { ImBuf *ibuf = BKE_image_pool_acquire_ibuf(tex->ima, &tex->iuser, pool); @@ -1341,7 +1345,8 @@ static int multitex_nodes_intern(Tex *tex, which_output, pool, skip_load_image, - texnode_preview); + texnode_preview, + use_nodes); } } @@ -1354,7 +1359,8 @@ int multitex_nodes(Tex *tex, float texvec[3], float dxt[3], float dyt[3], int os return multitex_nodes_intern(tex, texvec, dxt, dyt, osatex, texres, thread, which_output, shi, mtex, pool, R.scene_color_manage, (R.r.scemode & R_NO_IMAGE_LOAD) != 0, - (R.r.scemode & R_TEXNODE_PREVIEW) != 0); + (R.r.scemode & R_TEXNODE_PREVIEW) != 0, + true); } /* this is called for surface shading */ @@ -1378,7 +1384,8 @@ static int multitex_mtex(ShadeInput *shi, MTex *mtex, float texvec[3], float dxt mtex->which_output, pool, skip_load_image, - (R.r.scemode & R_TEXNODE_PREVIEW) != 0); + (R.r.scemode & R_TEXNODE_PREVIEW) != 0, + true); } } @@ -1408,7 +1415,8 @@ int multitex_ext(Tex *tex, pool, scene_color_manage, skip_load_image, - false); + false, + true); } /* extern-tex doesn't support nodes (ntreeBeginExec() can't be called when rendering is going on)\ @@ -1417,13 +1425,19 @@ int multitex_ext(Tex *tex, */ int multitex_ext_safe(Tex *tex, float texvec[3], TexResult *texres, struct ImagePool *pool, bool scene_color_manage, const bool skip_load_image) { - int use_nodes= tex->use_nodes, retval; - - tex->use_nodes = false; - retval= multitex_nodes_intern(tex, texvec, NULL, NULL, 0, texres, 0, 0, NULL, NULL, pool, scene_color_manage, skip_load_image, false); - tex->use_nodes= use_nodes; - - return retval; + return multitex_nodes_intern(tex, + texvec, + NULL, NULL, + 0, + texres, + 0, + 0, + NULL, NULL, + pool, + scene_color_manage, + skip_load_image, + false, + false); } @@ -2873,7 +2887,8 @@ void do_volume_tex(ShadeInput *shi, const float *xyz, int mapto_flag, float col_ mtex->which_output, re->pool, skip_load_image, - texnode_preview); /* NULL = dxt/dyt, 0 = shi->osatex - not supported */ + texnode_preview, + true); /* NULL = dxt/dyt, 0 = shi->osatex - not supported */ /* texture output */ @@ -3051,7 +3066,8 @@ void do_halo_tex(HaloRen *har, float xn, float yn, float col_r[4]) mtex->which_output, har->pool, skip_load_image, - texnode_preview); + texnode_preview, + true); /* texture output */ if (rgb && (mtex->texflag & MTEX_RGBTOINT)) { @@ -3274,7 +3290,8 @@ void do_sky_tex( mtex->which_output, R.pool, skip_load_image, - texnode_preview); + texnode_preview, + true); /* texture output */ if (rgb && (mtex->texflag & MTEX_RGBTOINT)) { @@ -3500,7 +3517,8 @@ void do_lamp_tex(LampRen *la, const float lavec[3], ShadeInput *shi, float col_r mtex->which_output, R.pool, skip_load_image, - texnode_preview); + texnode_preview, + true); /* texture output */ if (rgb && (mtex->texflag & MTEX_RGBTOINT)) { @@ -3614,7 +3632,8 @@ int externtex(MTex *mtex, mtex->which_output, pool, skip_load_image, - texnode_preview); + texnode_preview, + true); if (rgb) { texr.tin = IMB_colormanagement_get_luminance(&texr.tr); diff --git a/source/blender/render/intern/source/shadeoutput.c b/source/blender/render/intern/source/shadeoutput.c index c4de95e863e..a8fb72fb7f8 100644 --- a/source/blender/render/intern/source/shadeoutput.c +++ b/source/blender/render/intern/source/shadeoutput.c @@ -2143,7 +2143,7 @@ const float (*RE_object_instance_get_matrix(struct ObjectInstanceRen *obi, int m const float (*RE_render_current_get_matrix(int matrix_id))[4] { - switch(matrix_id) { + switch (matrix_id) { case RE_VIEW_MATRIX: return (const float(*)[4])R.viewmat; case RE_VIEWINV_MATRIX: @@ -2171,4 +2171,4 @@ float RE_fresnel_dielectric(float incoming[3], float normal[3], float eta) } return result; -}
\ No newline at end of file +} diff --git a/source/blender/render/intern/source/volume_precache.c b/source/blender/render/intern/source/volume_precache.c index 5377d0eba00..752a9df0b79 100644 --- a/source/blender/render/intern/source/volume_precache.c +++ b/source/blender/render/intern/source/volume_precache.c @@ -60,6 +60,8 @@ #include "volumetric.h" #include "volume_precache.h" +#include "atomic_ops.h" + /* ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ */ /* defined in pipeline.c, is hardcopy of active dynamic allocated Render */ @@ -509,7 +511,8 @@ static void *vol_precache_part_test(void *data) */ typedef struct VolPrecacheState { double lasttime; - int totparts; + unsigned int doneparts; + unsigned int totparts; } VolPrecacheState; static void vol_precache_part(TaskPool * __restrict pool, void *taskdata, int UNUSED(threadid)) @@ -574,13 +577,15 @@ static void vol_precache_part(TaskPool * __restrict pool, void *taskdata, int UN } } + unsigned int doneparts = atomic_add_and_fetch_u(&state->doneparts, 1); + time = PIL_check_seconds_timer(); if (time - state->lasttime > 1.0) { ThreadMutex *mutex = BLI_task_pool_user_mutex(pool); if (BLI_mutex_trylock(mutex)) { char str[64]; - float ratio = (float)BLI_task_pool_tasks_done(pool)/(float)state->totparts; + float ratio = (float)doneparts/(float)state->totparts; BLI_snprintf(str, sizeof(str), IFACE_("Precaching volume: %d%%"), (int)(100.0f * ratio)); re->i.infostr = str; re->stats_draw(re->sdh, &re->i); @@ -631,6 +636,7 @@ static void precache_launch_parts(Render *re, RayObject *tree, ShadeInput *shi, /* setup task scheduler */ memset(&state, 0, sizeof(state)); + state.doneparts = 0; state.totparts = parts[0]*parts[1]*parts[2]; state.lasttime = PIL_check_seconds_timer(); diff --git a/source/blender/windowmanager/CMakeLists.txt b/source/blender/windowmanager/CMakeLists.txt index b6245a8c0d1..925563a0777 100644 --- a/source/blender/windowmanager/CMakeLists.txt +++ b/source/blender/windowmanager/CMakeLists.txt @@ -128,10 +128,6 @@ if(WITH_PYTHON) ../python ) add_definitions(-DWITH_PYTHON) - - if(WITH_PYTHON_SECURITY) - add_definitions(-DWITH_PYTHON_SECURITY) - endif() endif() if(WITH_GAMEENGINE) diff --git a/source/blender/windowmanager/WM_api.h b/source/blender/windowmanager/WM_api.h index 2b82f1becb3..1cfd20defe0 100644 --- a/source/blender/windowmanager/WM_api.h +++ b/source/blender/windowmanager/WM_api.h @@ -252,6 +252,7 @@ int WM_operator_confirm_message(struct bContext *C, struct wmOperator *o /* operator api */ void WM_operator_free (struct wmOperator *op); +void WM_operator_free_all_after(wmWindowManager *wm, struct wmOperator *op); void WM_operator_type_set(struct wmOperator *op, struct wmOperatorType *ot); void WM_operator_stack_clear(struct wmWindowManager *wm); void WM_operator_handlers_clear(wmWindowManager *wm, struct wmOperatorType *ot); @@ -276,6 +277,7 @@ int WM_operator_call (struct bContext *C, struct wmOperator *op); int WM_operator_call_notest(struct bContext *C, struct wmOperator *op); int WM_operator_repeat (struct bContext *C, struct wmOperator *op); bool WM_operator_repeat_check(const struct bContext *C, struct wmOperator *op); +bool WM_operator_is_repeat(const struct bContext *C, const struct wmOperator *op); int WM_operator_name_call_ptr(struct bContext *C, struct wmOperatorType *ot, short context, struct PointerRNA *properties); int WM_operator_name_call(struct bContext *C, const char *opstring, short context, struct PointerRNA *properties); int WM_operator_call_py(struct bContext *C, struct wmOperatorType *ot, short context, struct PointerRNA *properties, struct ReportList *reports, const bool is_undo); diff --git a/source/blender/windowmanager/WM_types.h b/source/blender/windowmanager/WM_types.h index cd46e24264d..97f9257b4cb 100644 --- a/source/blender/windowmanager/WM_types.h +++ b/source/blender/windowmanager/WM_types.h @@ -542,7 +542,15 @@ typedef struct wmOperatorType { * canceled due to some external reason, cancel is called * - see defines below for return values */ int (*invoke)(struct bContext *, struct wmOperator *, const struct wmEvent *) ATTR_WARN_UNUSED_RESULT; + + /* Called when a modal operator is canceled (not used often). + * Internal cleanup can be done here if needed. */ void (*cancel)(struct bContext *, struct wmOperator *); + + /* Modal is used for operators which continuously run, eg: + * fly mode, knife tool, circle select are all examples of modal operators. + * Modal operators can handle events which would normally access other operators, + * they keep running until they don't return `OPERATOR_RUNNING_MODAL`. */ int (*modal)(struct bContext *, struct wmOperator *, const struct wmEvent *) ATTR_WARN_UNUSED_RESULT; /* verify if the operator can be executed in the current context, note diff --git a/source/blender/windowmanager/intern/wm.c b/source/blender/windowmanager/intern/wm.c index b76a1f1d422..4351cd22b18 100644 --- a/source/blender/windowmanager/intern/wm.c +++ b/source/blender/windowmanager/intern/wm.c @@ -107,6 +107,17 @@ void WM_operator_free(wmOperator *op) MEM_freeN(op); } +void WM_operator_free_all_after(wmWindowManager *wm, struct wmOperator *op) +{ + op = op->next; + while (op != NULL) { + wmOperator *op_next = op->next; + BLI_remlink(&wm->operators, op); + WM_operator_free(op); + op = op_next; + } +} + /** * Use with extreme care!, * properties, customdata etc - must be compatible. @@ -149,18 +160,23 @@ static void wm_reports_free(wmWindowManager *wm) void wm_operator_register(bContext *C, wmOperator *op) { wmWindowManager *wm = CTX_wm_manager(C); - int tot; + int tot = 0; BLI_addtail(&wm->operators, op); - tot = BLI_listbase_count(&wm->operators); - - while (tot > MAX_OP_REGISTERED) { - wmOperator *opt = wm->operators.first; - BLI_remlink(&wm->operators, opt); - WM_operator_free(opt); - tot--; + + /* only count registered operators */ + while (op) { + wmOperator *op_prev = op->prev; + if (op->type->flag & OPTYPE_REGISTER) { + tot += 1; + } + if (tot > MAX_OP_REGISTERED) { + BLI_remlink(&wm->operators, op); + WM_operator_free(op); + } + op = op_prev; } - + /* so the console is redrawn */ WM_event_add_notifier(C, NC_SPACE | ND_SPACE_INFO_REPORT, NULL); WM_event_add_notifier(C, NC_WM | ND_HISTORY, NULL); diff --git a/source/blender/windowmanager/intern/wm_event_system.c b/source/blender/windowmanager/intern/wm_event_system.c index d2b0acd836b..f26ee0e73ec 100644 --- a/source/blender/windowmanager/intern/wm_event_system.c +++ b/source/blender/windowmanager/intern/wm_event_system.c @@ -715,7 +715,9 @@ static void wm_operator_reports(bContext *C, wmOperator *op, int retval, bool ca */ static bool wm_operator_register_check(wmWindowManager *wm, wmOperatorType *ot) { - return wm && (wm->op_undo_depth == 0) && (ot->flag & OPTYPE_REGISTER); + /* Check undo flag here since undo operators are also added to the list, + * to support checking if the same operator is run twice. */ + return wm && (wm->op_undo_depth == 0) && (ot->flag & (OPTYPE_REGISTER | OPTYPE_UNDO)); } static void wm_operator_finished(bContext *C, wmOperator *op, const bool repeat) @@ -878,6 +880,20 @@ bool WM_operator_repeat_check(const bContext *UNUSED(C), wmOperator *op) return false; } +bool WM_operator_is_repeat(const bContext *C, const wmOperator *op) +{ + /* may be in the operators list or not */ + wmOperator *op_prev; + if (op->prev == NULL && op->next == NULL) { + wmWindowManager *wm = CTX_wm_manager(C); + op_prev = wm->operators.last; + } + else { + op_prev = op->prev; + } + return (op_prev && (op->type == op_prev->type)); +} + static wmOperator *wm_operator_create(wmWindowManager *wm, wmOperatorType *ot, PointerRNA *properties, ReportList *reports) { @@ -3179,6 +3195,8 @@ void wm_event_add_ghostevent(wmWindowManager *wm, wmWindow *win, int type, int U GHOST_TEventCursorData *cd = customdata; copy_v2_v2_int(&event.x, &cd->x); + wm_stereo3d_mouse_offset_apply(win, &event.x); + event.type = MOUSEMOVE; wm_event_add_mousemove(win, &event); copy_v2_v2_int(&evt->x, &event.x); diff --git a/source/blender/windowmanager/intern/wm_files.c b/source/blender/windowmanager/intern/wm_files.c index 05d63869074..028355f9af3 100644 --- a/source/blender/windowmanager/intern/wm_files.c +++ b/source/blender/windowmanager/intern/wm_files.c @@ -316,7 +316,7 @@ static void wm_window_match_do(bContext *C, ListBase *oldwmlist) } /* in case UserDef was read, we re-initialize all, and do versioning */ -static void wm_init_userdef(bContext *C, const bool from_memory) +static void wm_init_userdef(bContext *C, const bool use_factory_settings) { Main *bmain = CTX_data_main(C); @@ -336,7 +336,7 @@ static void wm_init_userdef(bContext *C, const bool from_memory) } /* avoid re-saving for every small change to our prefs, allow overrides */ - if (from_memory) { + if (use_factory_settings) { BLO_update_defaults_userpref_blend(); } @@ -470,6 +470,10 @@ static void wm_file_read_post(bContext *C, bool is_startup_file) if (is_startup_file) { /* possible python hasn't been initialized */ if (CTX_py_init_get(C)) { + /* Only run when we have a template path found. */ + if (BKE_appdir_app_template_any()) { + BPY_execute_string(C, "__import__('bl_app_template_utils').reset()"); + } /* sync addons, these may have changed from the defaults */ BPY_execute_string(C, "__import__('addon_utils').reset_all()"); @@ -554,7 +558,7 @@ bool WM_file_read(bContext *C, const char *filepath, ReportList *reports) /* confusing this global... */ G.relbase_valid = 1; - retval = BKE_blendfile_read(C, filepath, reports); + retval = BKE_blendfile_read(C, filepath, reports, 0); /* when loading startup.blend's, we can be left with a blank path */ if (G.main->name[0]) { G.save_over = 1; @@ -629,22 +633,34 @@ bool WM_file_read(bContext *C, const char *filepath, ReportList *reports) /** - * called on startup, (context entirely filled with NULLs) - * or called for 'New File' - * both startup.blend and userpref.blend are checked - * the optional parameter custom_file points to an alternative startup page - * custom_file can be NULL + * Called on startup, (context entirely filled with NULLs) + * or called for 'New File' both startup.blend and userpref.blend are checked. + * + * \param use_factory_settings: Ignore on-disk startup file, use bundled ``datatoc_startup_blend`` instead. + * Used for "Restore Factory Settings". + * \param filepath_startup_override: Optional path pointing to an alternative blend file (may be NULL). + * \param app_template_override: Template to use instead of the template defined in user-preferences. + * When not-null, this is written into the user preferences. */ -int wm_homefile_read(bContext *C, ReportList *reports, bool from_memory, const char *custom_file) +int wm_homefile_read( + bContext *C, ReportList *reports, + bool use_factory_settings, bool use_empty_data, + const char *filepath_startup_override, const char *app_template_override) { ListBase wmbase; - char startstr[FILE_MAX]; - char prefstr[FILE_MAX]; - int success = 0; + bool success = false; + + char filepath_startup[FILE_MAX]; + char filepath_userdef[FILE_MAX]; + + /* When 'app_template' is set: '{BLENDER_USER_CONFIG}/{app_template}' */ + char app_template_system[FILE_MAX]; + /* When 'app_template' is set: '{BLENDER_SYSTEM_SCRIPTS}/startup/bl_app_templates_system/{app_template}' */ + char app_template_config[FILE_MAX]; /* Indicates whether user preferences were really load from memory. * - * This is used for versioning code, and for this we can not rely on from_memory + * This is used for versioning code, and for this we can not rely on use_factory_settings * passed via argument. This is because there might be configuration folder * exists but it might not have userpref.blend and in this case we fallback to * reading home file from memory. @@ -652,9 +668,10 @@ int wm_homefile_read(bContext *C, ReportList *reports, bool from_memory, const c * And in this case versioning code is to be run. */ bool read_userdef_from_memory = true; + eBLOReadSkip skip_flags = 0; /* options exclude eachother */ - BLI_assert((from_memory && custom_file) == 0); + BLI_assert((use_factory_settings && filepath_startup_override) == 0); if ((G.f & G_SCRIPT_OVERRIDE_PREF) == 0) { BKE_BIT_TEST_SET(G.f, (U.flag & USER_SCRIPT_AUTOEXEC_DISABLE) == 0, G_SCRIPT_AUTOEXEC); @@ -665,71 +682,146 @@ int wm_homefile_read(bContext *C, ReportList *reports, bool from_memory, const c UI_view2d_zoom_cache_reset(); G.relbase_valid = 0; - if (!from_memory) { - const char * const cfgdir = BKE_appdir_folder_id(BLENDER_USER_CONFIG, NULL); - if (custom_file) { - BLI_strncpy(startstr, custom_file, FILE_MAX); - if (cfgdir) { - BLI_make_file_string(G.main->name, prefstr, cfgdir, BLENDER_USERPREF_FILE); - } - else { - prefstr[0] = '\0'; - } + /* put aside screens to match with persistent windows later */ + wm_window_match_init(C, &wmbase); + + filepath_startup[0] = '\0'; + filepath_userdef[0] = '\0'; + app_template_system[0] = '\0'; + app_template_config[0] = '\0'; + + const char * const cfgdir = BKE_appdir_folder_id(BLENDER_USER_CONFIG, NULL); + if (!use_factory_settings) { + if (cfgdir) { + BLI_path_join(filepath_startup, sizeof(filepath_startup), cfgdir, BLENDER_STARTUP_FILE, NULL); + BLI_path_join(filepath_userdef, sizeof(filepath_startup), cfgdir, BLENDER_USERPREF_FILE, NULL); } - else if (cfgdir) { - BLI_make_file_string(G.main->name, startstr, cfgdir, BLENDER_STARTUP_FILE); - BLI_make_file_string(G.main->name, prefstr, cfgdir, BLENDER_USERPREF_FILE); + else { + use_factory_settings = true; + } + + if (filepath_startup_override) { + BLI_strncpy(filepath_startup, filepath_startup_override, FILE_MAX); + } + } + + /* load preferences before startup.blend */ + if (!use_factory_settings && BLI_exists(filepath_userdef)) { + UserDef *userdef = BKE_blendfile_userdef_read(filepath_userdef, NULL); + if (userdef != NULL) { + BKE_blender_userdef_set_data(userdef); + MEM_freeN(userdef); + + read_userdef_from_memory = false; + skip_flags |= BLO_READ_SKIP_USERDEF; + printf("Read prefs: %s\n", filepath_userdef); + } + } + + const char *app_template = NULL; + + if (filepath_startup_override != NULL) { + /* pass */ + } + else if (app_template_override) { + app_template = app_template_override; + } + else if (!use_factory_settings && U.app_template[0]) { + app_template = U.app_template; + } + + if (app_template != NULL) { + BKE_appdir_app_template_id_search(app_template, app_template_system, sizeof(app_template_system)); + BLI_path_join(app_template_config, sizeof(app_template_config), cfgdir, app_template, NULL); + } + + /* insert template name into startup file */ + if (app_template != NULL) { + /* note that the path is being set even when 'use_factory_settings == true' + * this is done so we can load a templates factory-settings */ + if (!use_factory_settings) { + BLI_path_join(filepath_startup, sizeof(filepath_startup), app_template_config, BLENDER_STARTUP_FILE, NULL); + if (BLI_access(filepath_startup, R_OK) != 0) { + filepath_startup[0] = '\0'; + } } else { - startstr[0] = '\0'; - prefstr[0] = '\0'; - from_memory = 1; + filepath_startup[0] = '\0'; + } + + if (filepath_startup[0] == '\0') { + BLI_path_join(filepath_startup, sizeof(filepath_startup), app_template_system, BLENDER_STARTUP_FILE, NULL); } } - - /* put aside screens to match with persistent windows later */ - wm_window_match_init(C, &wmbase); - - if (!from_memory) { - if (BLI_access(startstr, R_OK) == 0) { - success = (BKE_blendfile_read(C, startstr, NULL) != BKE_BLENDFILE_READ_FAIL); + + if (!use_factory_settings || (filepath_startup[0] != '\0')) { + if (BLI_access(filepath_startup, R_OK) == 0) { + success = (BKE_blendfile_read(C, filepath_startup, NULL, skip_flags) != BKE_BLENDFILE_READ_FAIL); } if (BLI_listbase_is_empty(&U.themes)) { if (G.debug & G_DEBUG) - printf("\nNote: No (valid) '%s' found, fall back to built-in default.\n\n", startstr); - success = 0; + printf("\nNote: No (valid) '%s' found, fall back to built-in default.\n\n", filepath_startup); + success = false; } } - if (success == 0 && custom_file && reports) { - BKE_reportf(reports, RPT_ERROR, "Could not read '%s'", custom_file); - /*We can not return from here because wm is already reset*/ + if (success == false && filepath_startup_override && reports) { + /* We can not return from here because wm is already reset */ + BKE_reportf(reports, RPT_ERROR, "Could not read '%s'", filepath_startup_override); } - if (success == 0) { - success = BKE_blendfile_read_from_memory(C, datatoc_startup_blend, datatoc_startup_blend_size, NULL, true); + if (success == false) { + success = BKE_blendfile_read_from_memory( + C, datatoc_startup_blend, datatoc_startup_blend_size, + NULL, skip_flags, true); if (BLI_listbase_is_empty(&wmbase)) { wm_clear_default_size(C); } - BKE_tempdir_init(U.tempdir); + } -#ifdef WITH_PYTHON_SECURITY - /* use alternative setting for security nuts - * otherwise we'd need to patch the binary blob - startup.blend.c */ - U.flag |= USER_SCRIPT_AUTOEXEC_DISABLE; -#endif + if (use_empty_data) { + BKE_blendfile_read_make_empty(C); } - - /* check new prefs only after startup.blend was finished */ - if (!from_memory && BLI_exists(prefstr)) { - int done = BKE_blendfile_read_userdef(prefstr, NULL); - if (done != BKE_BLENDFILE_READ_FAIL) { - read_userdef_from_memory = false; - printf("Read new prefs: %s\n", prefstr); + + /* Load template preferences, + * unlike regular preferences we only use some of the settings, + * see: BKE_blender_userdef_set_app_template */ + if (app_template_system[0] != '\0') { + char temp_path[FILE_MAX]; + temp_path[0] = '\0'; + if (!use_factory_settings) { + BLI_path_join(temp_path, sizeof(temp_path), app_template_config, BLENDER_USERPREF_FILE, NULL); + if (BLI_access(temp_path, R_OK) != 0) { + temp_path[0] = '\0'; + } + } + + if (temp_path[0] == '\0') { + BLI_path_join(temp_path, sizeof(temp_path), app_template_system, BLENDER_USERPREF_FILE, NULL); + } + + UserDef *userdef_template = NULL; + /* just avoids missing file warning */ + if (BLI_exists(temp_path)) { + userdef_template = BKE_blendfile_userdef_read(temp_path, NULL); + } + if (userdef_template == NULL) { + /* we need to have preferences load to overwrite preferences from previous template */ + userdef_template = BKE_blendfile_userdef_read_from_memory( + datatoc_startup_blend, datatoc_startup_blend_size, NULL); + } + if (userdef_template) { + BKE_blender_userdef_set_app_template(userdef_template); + BKE_blender_userdef_free_data(userdef_template); + MEM_freeN(userdef_template); } } - + + if (app_template_override) { + BLI_strncpy(U.app_template, app_template_override, sizeof(U.app_template)); + } + /* prevent buggy files that had G_FILE_RELATIVE_REMAP written out by mistake. Screws up autosaves otherwise * can remove this eventually, only in a 2.53 and older, now its not written */ G.fileflags &= ~G_FILE_RELATIVE_REMAP; @@ -744,11 +836,14 @@ int wm_homefile_read(bContext *C, ReportList *reports, bool from_memory, const c G.main->name[0] = '\0'; /* When loading factory settings, the reset solid OpenGL lights need to be applied. */ - if (!G.background) GPU_default_lights(); - - /* XXX */ - G.save_over = 0; // start with save preference untitled.blend - G.fileflags &= ~G_FILE_AUTOPLAY; /* disable autoplay in startup.blend... */ + if (!G.background) { + GPU_default_lights(); + } + + /* start with save preference untitled.blend */ + G.save_over = 0; + /* disable auto-play in startup.blend... */ + G.fileflags &= ~G_FILE_AUTOPLAY; wm_file_read_post(C, true); @@ -1264,6 +1359,13 @@ static int wm_homefile_write_exec(bContext *C, wmOperator *op) char filepath[FILE_MAX]; int fileflags; + const char *app_template = U.app_template[0] ? U.app_template : NULL; + const char * const cfgdir = BKE_appdir_folder_id_create(BLENDER_USER_CONFIG, app_template); + if (cfgdir == NULL) { + BKE_report(op->reports, RPT_ERROR, "Unable to create user config path"); + return OPERATOR_CANCELLED; + } + BLI_callback_exec(G.main, NULL, BLI_CB_EVT_SAVE_PRE); /* check current window and close it if temp */ @@ -1273,7 +1375,8 @@ static int wm_homefile_write_exec(bContext *C, wmOperator *op) /* update keymaps in user preferences */ WM_keyconfig_update(wm); - BLI_make_file_string("/", filepath, BKE_appdir_folder_id_create(BLENDER_USER_CONFIG, NULL), BLENDER_STARTUP_FILE); + BLI_path_join(filepath, sizeof(filepath), cfgdir, BLENDER_STARTUP_FILE, NULL); + printf("trying to save homefile at %s ", filepath); ED_editors_flush_edits(C, false); @@ -1351,21 +1454,44 @@ static int wm_userpref_write_exec(bContext *C, wmOperator *op) { wmWindowManager *wm = CTX_wm_manager(C); char filepath[FILE_MAX]; + const char *cfgdir; + bool ok = false; /* update keymaps in user preferences */ WM_keyconfig_update(wm); - BLI_make_file_string("/", filepath, BKE_appdir_folder_id_create(BLENDER_USER_CONFIG, NULL), BLENDER_USERPREF_FILE); - printf("trying to save userpref at %s ", filepath); - - if (BKE_blendfile_write_userdef(filepath, op->reports) == 0) { - printf("fail\n"); - return OPERATOR_CANCELLED; + if ((cfgdir = BKE_appdir_folder_id_create(BLENDER_USER_CONFIG, NULL))) { + BLI_path_join(filepath, sizeof(filepath), cfgdir, BLENDER_USERPREF_FILE, NULL); + printf("trying to save userpref at %s ", filepath); + if (BKE_blendfile_userdef_write(filepath, op->reports) != 0) { + printf("ok\n"); + ok = true; + } + else { + printf("fail\n"); + } + } + else { + BKE_report(op->reports, RPT_ERROR, "Unable to create userpref path"); } - printf("ok\n"); + if (U.app_template[0] && (cfgdir = BKE_appdir_folder_id_create(BLENDER_USER_CONFIG, U.app_template))) { + /* Also save app-template prefs */ + BLI_path_join(filepath, sizeof(filepath), cfgdir, BLENDER_USERPREF_FILE, NULL); + printf("trying to save app-template userpref at %s ", filepath); + if (BKE_blendfile_userdef_write(filepath, op->reports) == 0) { + printf("fail\n"); + ok = true; + } + else { + printf("ok\n"); + } + } + else if (U.app_template[0]) { + BKE_report(op->reports, RPT_ERROR, "Unable to create app-template userpref path"); + } - return OPERATOR_FINISHED; + return ok ? OPERATOR_FINISHED : OPERATOR_CANCELLED; } void WM_OT_save_userpref(wmOperatorType *ot) @@ -1400,11 +1526,11 @@ void WM_OT_read_history(wmOperatorType *ot) static int wm_homefile_read_exec(bContext *C, wmOperator *op) { - const bool from_memory = (STREQ(op->type->idname, "WM_OT_read_factory_settings")); + const bool use_factory_settings = (STREQ(op->type->idname, "WM_OT_read_factory_settings")); char filepath_buf[FILE_MAX]; const char *filepath = NULL; - if (!from_memory) { + if (!use_factory_settings) { PropertyRNA *prop = RNA_struct_find_property(op->ptr, "filepath"); /* This can be used when loading of a start-up file should only change @@ -1426,7 +1552,34 @@ static int wm_homefile_read_exec(bContext *C, wmOperator *op) G.fileflags &= ~G_FILE_NO_UI; } - return wm_homefile_read(C, op->reports, from_memory, filepath) ? OPERATOR_FINISHED : OPERATOR_CANCELLED; + char app_template_buf[sizeof(U.app_template)]; + const char *app_template; + PropertyRNA *prop_app_template = RNA_struct_find_property(op->ptr, "app_template"); + const bool use_splash = !use_factory_settings && RNA_boolean_get(op->ptr, "use_splash"); + const bool use_empty_data = RNA_boolean_get(op->ptr, "use_empty"); + + if (prop_app_template && RNA_property_is_set(op->ptr, prop_app_template)) { + RNA_property_string_get(op->ptr, prop_app_template, app_template_buf); + app_template = app_template_buf; + } + else if (!use_factory_settings) { + /* TODO: dont reset prefs on 'New File' */ + BLI_strncpy(app_template_buf, U.app_template, sizeof(app_template_buf)); + app_template = app_template_buf; + } + else { + app_template = NULL; + } + + if (wm_homefile_read(C, op->reports, use_factory_settings, use_empty_data, filepath, app_template)) { + if (use_splash) { + WM_init_splash(C); + } + return OPERATOR_FINISHED; + } + else { + return OPERATOR_CANCELLED; + } } void WM_OT_read_homefile(wmOperatorType *ot) @@ -1449,17 +1602,36 @@ void WM_OT_read_homefile(wmOperatorType *ot) "Load user interface setup from the .blend file"); RNA_def_property_flag(prop, PROP_HIDDEN | PROP_SKIP_SAVE); + prop = RNA_def_boolean(ot->srna, "use_empty", false, "Empty", ""); + RNA_def_property_flag(prop, PROP_HIDDEN | PROP_SKIP_SAVE); + + /* So the splash can be kept open after loading a file (for templates). */ + prop = RNA_def_boolean(ot->srna, "use_splash", false, "Splash", ""); + RNA_def_property_flag(prop, PROP_HIDDEN | PROP_SKIP_SAVE); + + prop = RNA_def_string(ot->srna, "app_template", "Template", sizeof(U.app_template), "", ""); + RNA_def_property_flag(prop, PROP_HIDDEN | PROP_SKIP_SAVE); + /* omit poll to run in background mode */ } void WM_OT_read_factory_settings(wmOperatorType *ot) { + PropertyRNA *prop; + ot->name = "Load Factory Settings"; ot->idname = "WM_OT_read_factory_settings"; ot->description = "Load default file and user preferences"; ot->invoke = WM_operator_confirm; ot->exec = wm_homefile_read_exec; + + prop = RNA_def_string(ot->srna, "app_template", "Template", sizeof(U.app_template), "", ""); + RNA_def_property_flag(prop, PROP_HIDDEN | PROP_SKIP_SAVE); + + prop = RNA_def_boolean(ot->srna, "use_empty", false, "Empty", ""); + RNA_def_property_flag(prop, PROP_HIDDEN | PROP_SKIP_SAVE); + /* omit poll to run in background mode */ } diff --git a/source/blender/windowmanager/intern/wm_init_exit.c b/source/blender/windowmanager/intern/wm_init_exit.c index c11c398c616..9bafe72d805 100644 --- a/source/blender/windowmanager/intern/wm_init_exit.c +++ b/source/blender/windowmanager/intern/wm_init_exit.c @@ -192,7 +192,7 @@ void WM_init(bContext *C, int argc, const char **argv) wm_init_reports(C); /* get the default database, plus a wm */ - wm_homefile_read(C, NULL, G.factory_startup, NULL); + wm_homefile_read(C, NULL, G.factory_startup, false, NULL, NULL); BLT_lang_set(NULL); @@ -444,8 +444,6 @@ void WM_exit_ext(bContext *C, const bool do_python) { wmWindowManager *wm = C ? CTX_wm_manager(C) : NULL; - BKE_sound_exit(); - /* first wrap up running stuff, we assume only the active WM is running */ /* modal handlers are on window level freed, others too? */ /* note; same code copied in wm_files.c */ @@ -574,7 +572,7 @@ void WM_exit_ext(bContext *C, const bool do_python) ED_file_exit(); /* for fsmenu */ UI_exit(); - BKE_blender_userdef_free(); + BKE_blender_userdef_free_data(&U); RNA_exit(); /* should be after BPY_python_end so struct python slots are cleared */ @@ -591,6 +589,10 @@ void WM_exit_ext(bContext *C, const bool do_python) BLI_threadapi_exit(); + /* No need to call this early, rather do it late so that other pieces of Blender using sound may exit cleanly, + * see also T50676. */ + BKE_sound_exit(); + BKE_blender_atexit(); if (MEM_get_memory_blocks_in_use() != 0) { diff --git a/source/blender/windowmanager/intern/wm_keymap.c b/source/blender/windowmanager/intern/wm_keymap.c index e201fa433d4..db5fc23146f 100644 --- a/source/blender/windowmanager/intern/wm_keymap.c +++ b/source/blender/windowmanager/intern/wm_keymap.c @@ -121,6 +121,13 @@ static void wm_keymap_item_properties_update_ot(wmKeyMapItem *kmi) if (ot->srna != kmi->ptr->type) { /* matches wm_keymap_item_properties_set but doesnt alloc new ptr */ WM_operator_properties_create_ptr(kmi->ptr, ot); + /* 'kmi->ptr->data' NULL'd above, keep using existing properties. + * Note: the operators property types may have changed, + * we will need a more comprehensive sanitize function to support this properly. + */ + if (kmi->properties) { + kmi->ptr->data = kmi->properties; + } WM_operator_properties_sanitize(kmi->ptr, 1); } } diff --git a/source/blender/windowmanager/intern/wm_operators.c b/source/blender/windowmanager/intern/wm_operators.c index cef5e7e4a8e..d827ccafda8 100644 --- a/source/blender/windowmanager/intern/wm_operators.c +++ b/source/blender/windowmanager/intern/wm_operators.c @@ -900,7 +900,7 @@ void WM_operator_properties_create_ptr(PointerRNA *ptr, wmOperatorType *ot) void WM_operator_properties_create(PointerRNA *ptr, const char *opstring) { - wmOperatorType *ot = WM_operatortype_find(opstring, 0); + wmOperatorType *ot = WM_operatortype_find(opstring, false); if (ot) WM_operator_properties_create_ptr(ptr, ot); @@ -1118,6 +1118,7 @@ static uiBlock *wm_enum_search_menu(bContext *C, ARegion *ar, void *arg_op) block = UI_block_begin(C, ar, "_popup", UI_EMBOSS); UI_block_flag_enable(block, UI_BLOCK_LOOP | UI_BLOCK_MOVEMOUSE_QUIT | UI_BLOCK_SEARCH_MENU); + search[0] = '\0'; #if 0 /* ok, this isn't so easy... */ uiDefBut(block, UI_BTYPE_LABEL, 0, RNA_struct_ui_name(op->type->srna), 10, 10, UI_searchbox_size_x(), UI_UNIT_Y, NULL, 0.0, 0.0, 0, 0, ""); #endif @@ -1402,20 +1403,6 @@ static void dialog_exec_cb(bContext *C, void *arg1, void *arg2) } } -static void popup_check_cb(bContext *C, void *op_ptr, void *UNUSED(arg)) -{ - wmOperator *op = op_ptr; - if (op->type->check) { - if (op->type->check(C, op)) { - /* check for popup and re-layout buttons */ - ARegion *ar_menu = CTX_wm_menu(C); - if (ar_menu) { - ED_region_tag_refresh_ui(ar_menu); - } - } - } -} - /* Dialogs are popups that require user verification (click OK) before exec */ static uiBlock *wm_block_dialog_create(bContext *C, ARegion *ar, void *userData) { @@ -1434,8 +1421,6 @@ static uiBlock *wm_block_dialog_create(bContext *C, ARegion *ar, void *userData) layout = UI_block_layout(block, UI_LAYOUT_VERTICAL, UI_LAYOUT_PANEL, 0, 0, data->width, data->height, 0, style); - UI_block_func_set(block, popup_check_cb, op, NULL); - uiLayoutOperatorButs(C, layout, op, NULL, 'H', UI_LAYOUT_OP_SHOW_TITLE); /* clear so the OK button is left alone */ @@ -1474,8 +1459,6 @@ static uiBlock *wm_operator_ui_create(bContext *C, ARegion *ar, void *userData) layout = UI_block_layout(block, UI_LAYOUT_VERTICAL, UI_LAYOUT_PANEL, 0, 0, data->width, data->height, 0, style); - UI_block_func_set(block, popup_check_cb, op, NULL); - /* since ui is defined the auto-layout args are not used */ uiLayoutOperatorButs(C, layout, op, NULL, 'V', 0); @@ -1522,7 +1505,7 @@ int WM_operator_ui_popup(bContext *C, wmOperator *op, int width, int height) data->width = width; data->height = height; data->free_op = true; /* if this runs and gets registered we may want not to free it */ - UI_popup_block_ex(C, wm_operator_ui_create, NULL, wm_operator_ui_popup_cancel, data); + UI_popup_block_ex(C, wm_operator_ui_create, NULL, wm_operator_ui_popup_cancel, data, op); return OPERATOR_RUNNING_MODAL; } @@ -1552,7 +1535,7 @@ static int wm_operator_props_popup_ex(bContext *C, wmOperator *op, if (!do_redo || !(U.uiflag & USER_GLOBALUNDO)) return WM_operator_props_dialog_popup(C, op, 15 * UI_UNIT_X, UI_UNIT_Y); - UI_popup_block_ex(C, wm_block_create_redo, NULL, wm_block_redo_cancel_cb, op); + UI_popup_block_ex(C, wm_block_create_redo, NULL, wm_block_redo_cancel_cb, op, op); if (do_call) wm_block_redo_cb(C, op, 0); @@ -1594,7 +1577,7 @@ int WM_operator_props_dialog_popup(bContext *C, wmOperator *op, int width, int h data->free_op = true; /* if this runs and gets registered we may want not to free it */ /* op is not executed until popup OK but is clicked */ - UI_popup_block_ex(C, wm_block_dialog_create, wm_operator_ui_popup_ok, wm_operator_ui_popup_cancel, data); + UI_popup_block_ex(C, wm_block_dialog_create, wm_operator_ui_popup_ok, wm_operator_ui_popup_cancel, data, op); return OPERATOR_RUNNING_MODAL; } @@ -1761,6 +1744,36 @@ static uiBlock *wm_block_create_splash(bContext *C, ARegion *ar, void *UNUSED(ar ibuf = IMB_ibImageFromMemory((unsigned char *)datatoc_splash_png, datatoc_splash_png_size, IB_rect, NULL, "<splash screen>"); } + + /* overwrite splash with template image */ + if (U.app_template[0] != '\0') { + ImBuf *ibuf_template = NULL; + char splash_filepath[FILE_MAX]; + char template_directory[FILE_MAX]; + + if (BKE_appdir_app_template_id_search( + U.app_template, + template_directory, sizeof(template_directory))) + { + BLI_join_dirfile( + splash_filepath, sizeof(splash_filepath), template_directory, + (U.pixelsize == 2) ? "splash_2x.png" : "splash.png"); + ibuf_template = IMB_loadiffname(splash_filepath, IB_rect, NULL); + if (ibuf_template) { + const int x_expect = ibuf->x; + const int y_expect = 230 * (int)U.pixelsize; + /* don't cover the header text */ + if (ibuf_template->x == x_expect && ibuf_template->y == y_expect) { + memcpy(ibuf->rect, ibuf_template->rect, ibuf_template->x * ibuf_template->y * sizeof(char[4])); + } + else { + printf("Splash expected %dx%d found %dx%d, ignoring: %s\n", + x_expect, y_expect, ibuf_template->x, ibuf_template->y, splash_filepath); + } + IMB_freeImBuf(ibuf_template); + } + } + } #endif block = UI_block_begin(C, ar, "_popup", UI_EMBOSS); diff --git a/source/blender/windowmanager/intern/wm_stereo.c b/source/blender/windowmanager/intern/wm_stereo.c index 46cee907991..66ebf18c9e1 100644 --- a/source/blender/windowmanager/intern/wm_stereo.c +++ b/source/blender/windowmanager/intern/wm_stereo.c @@ -345,6 +345,32 @@ bool WM_stereo3d_enabled(wmWindow *win, bool skip_stereo3d_check) return true; } +/** + * If needed, this adjusts \a r_mouse_xy so that drawn cursor and handled mouse position are matching visually. +*/ +void wm_stereo3d_mouse_offset_apply(wmWindow *win, int *r_mouse_xy) +{ + if (!WM_stereo3d_enabled(win, false)) + return; + + if (win->stereo3d_format->display_mode == S3D_DISPLAY_SIDEBYSIDE) { + const int half_x = win->sizex / 2; + /* right half of the screen */ + if (r_mouse_xy[0] > half_x) { + r_mouse_xy[0] -= half_x; + } + r_mouse_xy[0] *= 2; + } + else if (win->stereo3d_format->display_mode == S3D_DISPLAY_TOPBOTTOM) { + const int half_y = win->sizey / 2; + /* upper half of the screen */ + if (r_mouse_xy[1] > half_y) { + r_mouse_xy[1] -= half_y; + } + r_mouse_xy[1] *= 2; + } +} + /************************** Stereo 3D operator **********************************/ typedef struct Stereo3dData { Stereo3dFormat stereo3d_format; diff --git a/source/blender/windowmanager/intern/wm_window.c b/source/blender/windowmanager/intern/wm_window.c index 2d43c47679d..aaf77946412 100644 --- a/source/blender/windowmanager/intern/wm_window.c +++ b/source/blender/windowmanager/intern/wm_window.c @@ -77,6 +77,7 @@ #include "GPU_extensions.h" #include "GPU_init_exit.h" #include "GPU_glew.h" +#include "BLF_api.h" /* for assert */ #ifndef NDEBUG @@ -311,7 +312,7 @@ void wm_window_close(bContext *C, wmWindowManager *wm, wmWindow *win) if (tmpwin == NULL) do_exit = 1; - if ((U.uiflag & USER_QUIT_PROMPT) && !wm->file_saved) { + if ((U.uiflag & USER_QUIT_PROMPT) && !wm->file_saved && !G.background) { if (do_exit) { if (!GHOST_confirmQuit(win->ghostwin)) return; @@ -374,14 +375,39 @@ void wm_window_title(wmWindowManager *wm, wmWindow *win) } } -static float wm_window_get_virtual_pixelsize(void) +static void wm_window_set_dpi(wmWindow *win) { - return ((U.virtual_pixel == VIRTUAL_PIXEL_NATIVE) ? 1.0f : 2.0f); -} + int auto_dpi = GHOST_GetDPIHint(win->ghostwin); -float wm_window_pixelsize(wmWindow *win) -{ - return (GHOST_GetNativePixelSize(win->ghostwin) * wm_window_get_virtual_pixelsize()); + /* Lazily init UI scale size, preserving backwards compatibility by + * computing UI scale from ratio of previous DPI and auto DPI */ + if (U.ui_scale == 0) { + int virtual_pixel = (U.virtual_pixel == VIRTUAL_PIXEL_NATIVE) ? 1 : 2; + + if (U.dpi == 0) { + U.ui_scale = virtual_pixel; + } + else { + U.ui_scale = (virtual_pixel * U.dpi * 96.0f) / (auto_dpi * 72.0f); + } + + CLAMP(U.ui_scale, 0.25f, 4.0f); + } + + /* Blender's UI drawing assumes DPI 72 as a good default following macOS + * while Windows and Linux use DPI 96. GHOST assumes a default 96 so we + * remap the DPI to Blender's convention. */ + int dpi = auto_dpi * U.ui_scale * (72.0 / 96.0f); + + /* Automatically set larger pixel size for high DPI. */ + int pixelsize = MAX2(1, dpi / 54); + + /* Set user preferences globals for drawing, and for forward compatibility. */ + U.pixelsize = GHOST_GetNativePixelSize(win->ghostwin) * pixelsize; + U.dpi = dpi / pixelsize; + U.virtual_pixel = (pixelsize == 1) ? VIRTUAL_PIXEL_NATIVE : VIRTUAL_PIXEL_DOUBLE; + + BKE_blender_userdef_refresh(); } /* belongs to below */ @@ -456,10 +482,8 @@ static void wm_window_ghostwindow_add(wmWindowManager *wm, const char *title, wm glClear(GL_COLOR_BUFFER_BIT); } - /* displays with larger native pixels, like Macbook. Used to scale dpi with */ /* needed here, because it's used before it reads userdef */ - U.pixelsize = wm_window_pixelsize(win); - BKE_blender_userdef_refresh(); + wm_window_set_dpi(win); wm_window_swap_buffers(win); @@ -626,7 +650,6 @@ wmWindow *WM_window_open_temp(bContext *C, const rcti *rect_init, int type) Scene *scene = CTX_data_scene(C); const char *title; rcti rect = *rect_init; - const short px_virtual = (short)wm_window_get_virtual_pixelsize(); /* changes rect to fit within desktop */ wm_window_check_position(&rect); @@ -644,9 +667,8 @@ wmWindow *WM_window_open_temp(bContext *C, const rcti *rect_init, int type) win->posy = rect.ymin; } - /* multiply with virtual pixelsize, ghost handles native one (e.g. for retina) */ - win->sizex = BLI_rcti_size_x(&rect) * px_virtual; - win->sizey = BLI_rcti_size_y(&rect) * px_virtual; + win->sizex = BLI_rcti_size_x(&rect); + win->sizey = BLI_rcti_size_y(&rect); if (win->ghostwin) { wm_window_set_size(win, win->sizex, win->sizey); @@ -835,8 +857,7 @@ void wm_window_make_drawable(wmWindowManager *wm, wmWindow *win) GHOST_ActivateWindowDrawingContext(win->ghostwin); /* this can change per window */ - U.pixelsize = wm_window_pixelsize(win); - BKE_blender_userdef_refresh(); + wm_window_set_dpi(win); } } @@ -1035,6 +1056,8 @@ static int ghost_event_proc(GHOST_EventHandle evt, GHOST_TUserDataPtr C_void_ptr if (type == GHOST_kEventWindowSize) { WM_jobs_stop(wm, win->screen, NULL); } + + wm_window_set_dpi(win); /* win32: gives undefined window size when minimized */ if (state != GHOST_kWindowStateMinimized) { @@ -1118,7 +1141,19 @@ static int ghost_event_proc(GHOST_EventHandle evt, GHOST_TUserDataPtr C_void_ptr } break; } - + + case GHOST_kEventWindowDPIHintChanged: + { + wm_window_set_dpi(win); + /* font's are stored at each DPI level, without this we can easy load 100's of fonts */ + BLF_cache_clear(); + + BKE_blender_userdef_refresh(); + WM_main_add_notifier(NC_WINDOW, NULL); /* full redraw */ + WM_main_add_notifier(NC_SCREEN | NA_EDITED, NULL); /* refresh region sizes */ + break; + } + case GHOST_kEventOpenMainFile: { PointerRNA props_ptr; @@ -1199,11 +1234,9 @@ static int ghost_event_proc(GHOST_EventHandle evt, GHOST_TUserDataPtr C_void_ptr { // only update if the actual pixel size changes float prev_pixelsize = U.pixelsize; - U.pixelsize = wm_window_pixelsize(win); + wm_window_set_dpi(win); if (U.pixelsize != prev_pixelsize) { - BKE_blender_userdef_refresh(); - // close all popups since they are positioned with the pixel // size baked in and it's difficult to correct them wmWindow *oldWindow = CTX_wm_window(C); diff --git a/source/blender/windowmanager/wm.h b/source/blender/windowmanager/wm.h index 2f06ddab1e8..e8485359490 100644 --- a/source/blender/windowmanager/wm.h +++ b/source/blender/windowmanager/wm.h @@ -78,6 +78,7 @@ void wm_autosave_location(char *filepath); /* wm_stereo.c */ void wm_method_draw_stereo3d(const bContext *C, wmWindow *win); +void wm_stereo3d_mouse_offset_apply(wmWindow *win, int *r_mouse_xy); int wm_stereo3d_set_exec(bContext *C, wmOperator *op); int wm_stereo3d_set_invoke(bContext *C, wmOperator *op, const wmEvent *event); void wm_stereo3d_set_draw(bContext *C, wmOperator *op); diff --git a/source/blender/windowmanager/wm_files.h b/source/blender/windowmanager/wm_files.h index 396907a3f6d..9a1518e15b0 100644 --- a/source/blender/windowmanager/wm_files.h +++ b/source/blender/windowmanager/wm_files.h @@ -35,7 +35,10 @@ struct wmOperatorType; /* wm_files.c */ void wm_history_file_read(void); -int wm_homefile_read(struct bContext *C, struct ReportList *reports, bool from_memory, const char *filepath); +int wm_homefile_read( + struct bContext *C, struct ReportList *reports, + bool use_factory_settings, bool use_empty_data, + const char *filepath_startup_override, const char *app_template_override); void wm_file_read_report(bContext *C); void WM_OT_save_homefile(struct wmOperatorType *ot); diff --git a/source/blender/windowmanager/wm_window.h b/source/blender/windowmanager/wm_window.h index c106f9d7851..f70ec6b47f6 100644 --- a/source/blender/windowmanager/wm_window.h +++ b/source/blender/windowmanager/wm_window.h @@ -63,8 +63,6 @@ void wm_window_swap_buffers (wmWindow *win); void wm_window_set_swap_interval(wmWindow *win, int interval); bool wm_window_get_swap_interval(wmWindow *win, int *intervalOut); -float wm_window_pixelsize(wmWindow *win); - void wm_get_cursor_position (wmWindow *win, int *x, int *y); void wm_cursor_position_from_ghost (wmWindow *win, int *x, int *y); void wm_cursor_position_to_ghost (wmWindow *win, int *x, int *y); diff --git a/source/blenderplayer/bad_level_call_stubs/stubs.c b/source/blenderplayer/bad_level_call_stubs/stubs.c index 08d00e9148a..d659f5c1a8a 100644 --- a/source/blenderplayer/bad_level_call_stubs/stubs.c +++ b/source/blenderplayer/bad_level_call_stubs/stubs.c @@ -306,6 +306,7 @@ int WM_operator_confirm(struct bContext *C, struct wmOperator *op, const struct struct MenuType *WM_menutype_find(const char *idname, bool quiet) RET_NULL void WM_operator_stack_clear(struct wmWindowManager *wm) RET_NONE void WM_operator_handlers_clear(wmWindowManager *wm, struct wmOperatorType *ot) RET_NONE +bool WM_operator_is_repeat(const struct bContext *C, const struct wmOperator *op) RET_ZERO; void WM_autosave_init(wmWindowManager *wm) RET_NONE void WM_jobs_kill_all_except(struct wmWindowManager *wm, void *owner) RET_NONE @@ -736,7 +737,8 @@ int collada_export(struct Scene *sce, int use_blender_profile, int sort_by_name, BC_export_transformation_type export_transformation_type, - int open_sim) RET_ZERO + int open_sim, + int keep_bind_info) RET_ZERO void ED_mesh_calc_tessface(struct Mesh *mesh, bool free_mpoly) RET_NONE diff --git a/source/creator/creator_args.c b/source/creator/creator_args.c index 27579e58dba..658a0b2db08 100644 --- a/source/creator/creator_args.c +++ b/source/creator/creator_args.c @@ -946,7 +946,7 @@ static int arg_handle_native_pixels_set(int UNUSED(argc), const char **UNUSED(ar } static const char arg_handle_with_borders_doc[] = -"\n\tForce opening without borders" +"\n\tForce opening with borders" ; static int arg_handle_with_borders(int UNUSED(argc), const char **UNUSED(argv), void *UNUSED(data)) { @@ -1364,7 +1364,7 @@ static int arg_handle_render_frame(int argc, const char **argv, void *data) re = RE_NewRender(scene->id.name); BLI_begin_threaded_malloc(); - BKE_reports_init(&reports, RPT_PRINT); + BKE_reports_init(&reports, RPT_STORE); RE_SetReports(re, &reports); for (int i = 0; i < frames_range_len; i++) { @@ -1379,6 +1379,7 @@ static int arg_handle_render_frame(int argc, const char **argv, void *data) } } RE_SetReports(re, NULL); + BKE_reports_clear(&reports); BLI_end_threaded_malloc(); MEM_freeN(frame_range_arr); return 1; @@ -1406,10 +1407,11 @@ static int arg_handle_render_animation(int UNUSED(argc), const char **UNUSED(arg Render *re = RE_NewRender(scene->id.name); ReportList reports; BLI_begin_threaded_malloc(); - BKE_reports_init(&reports, RPT_PRINT); + BKE_reports_init(&reports, RPT_STORE); RE_SetReports(re, &reports); RE_BlenderAnim(re, bmain, scene, NULL, scene->lay, scene->r.sfra, scene->r.efra, scene->r.frame_step); RE_SetReports(re, NULL); + BKE_reports_clear(&reports); BLI_end_threaded_malloc(); } else { diff --git a/source/gameengine/BlenderRoutines/BL_KetsjiEmbedStart.cpp b/source/gameengine/BlenderRoutines/BL_KetsjiEmbedStart.cpp index d7fe8d8ce59..6c3751ae34d 100644 --- a/source/gameengine/BlenderRoutines/BL_KetsjiEmbedStart.cpp +++ b/source/gameengine/BlenderRoutines/BL_KetsjiEmbedStart.cpp @@ -104,13 +104,13 @@ typedef void * wmUIHandlerRemoveFunc; # include AUD_DEVICE_H #endif -static BlendFileData *load_game_data(char *filename) +static BlendFileData *load_game_data(const char *filename) { ReportList reports; BlendFileData *bfd; BKE_reports_init(&reports, RPT_STORE); - bfd= BLO_read_from_file(filename, &reports); + bfd= BLO_read_from_file(filename, &reports, BLO_READ_SKIP_USERDEF); if (!bfd) { printf("Loading %s failed: ", filename); diff --git a/source/gameengine/GamePlayer/ghost/GPG_ghost.cpp b/source/gameengine/GamePlayer/ghost/GPG_ghost.cpp index dd93a6ff424..0c206dfce3d 100644 --- a/source/gameengine/GamePlayer/ghost/GPG_ghost.cpp +++ b/source/gameengine/GamePlayer/ghost/GPG_ghost.cpp @@ -355,7 +355,7 @@ static BlendFileData *load_game_data(const char *progname, char *filename = NULL BLI_strncpy(bfd->main->name, progname, sizeof(bfd->main->name)); } } else { - bfd= BLO_read_from_file(progname, &reports); + bfd= BLO_read_from_file(progname, &reports, BLO_READ_SKIP_NONE); } if (!bfd && filename) { diff --git a/source/gameengine/VideoTexture/FilterColor.cpp b/source/gameengine/VideoTexture/FilterColor.cpp index eed84a8580c..15a7e9e4cd1 100644 --- a/source/gameengine/VideoTexture/FilterColor.cpp +++ b/source/gameengine/VideoTexture/FilterColor.cpp @@ -68,7 +68,7 @@ PyTypeObject FilterGrayType = 0, /*tp_setattro*/ 0, /*tp_as_buffer*/ Py_TPFLAGS_DEFAULT, /*tp_flags*/ - "Filter for gray scale effect", /* tp_doc */ + "Filter for grayscale effect", /* tp_doc */ 0, /* tp_traverse */ 0, /* tp_clear */ 0, /* tp_richcompare */ diff --git a/source/gameengine/VideoTexture/FilterColor.h b/source/gameengine/VideoTexture/FilterColor.h index 350f7270874..d042863d7e8 100644 --- a/source/gameengine/VideoTexture/FilterColor.h +++ b/source/gameengine/VideoTexture/FilterColor.h @@ -36,7 +36,7 @@ #include "FilterBase.h" -/// pixel filter for gray scale +/// pixel filter for grayscale class FilterGray : public FilterBase { public: @@ -53,7 +53,7 @@ protected: // calculate gray value unsigned int gray = (28 * (VT_B(val)) + 151 * (VT_G(val)) + 77 * (VT_R(val))) >> 8; - // return gray scale value + // return grayscale value VT_R(val) = gray; VT_G(val) = gray; VT_B(val) = gray; diff --git a/tests/gtests/blenlib/BLI_array_store_test.cc b/tests/gtests/blenlib/BLI_array_store_test.cc index 5af6e639e64..370a4111bae 100644 --- a/tests/gtests/blenlib/BLI_array_store_test.cc +++ b/tests/gtests/blenlib/BLI_array_store_test.cc @@ -36,15 +36,15 @@ static void print_mem_saved(const char *id, const BArrayStore *bs) /* -------------------------------------------------------------------- */ /* Test Chunks (building data from list of chunks) */ -typedef struct TestChunnk { - struct TestChunnk *next, *prev; +typedef struct TestChunk { + struct TestChunk *next, *prev; const void *data; size_t data_len; -} TestChunnk; +} TestChunk; -static TestChunnk *testchunk_list_add(ListBase *lb, const void *data, size_t data_len) +static TestChunk *testchunk_list_add(ListBase *lb, const void *data, size_t data_len) { - TestChunnk *tc = (TestChunnk *)MEM_mallocN(sizeof(*tc), __func__); + TestChunk *tc = (TestChunk *)MEM_mallocN(sizeof(*tc), __func__); tc->data = data; tc->data_len = data_len; BLI_addtail(lb, tc); @@ -53,7 +53,7 @@ static TestChunnk *testchunk_list_add(ListBase *lb, const void *data, size_t dat } #if 0 -static TestChunnk *testchunk_list_add_copydata(ListBase *lb, const void *data, size_t data_len) +static TestChunk *testchunk_list_add_copydata(ListBase *lb, const void *data, size_t data_len) { void *data_copy = MEM_mallocN(data_len, __func__); memcpy(data_copy, data, data_len); @@ -63,7 +63,7 @@ static TestChunnk *testchunk_list_add_copydata(ListBase *lb, const void *data, s static void testchunk_list_free(ListBase *lb) { - for (TestChunnk *tc = (TestChunnk *)lb->first, *tb_next; tc; tc = tb_next) { + for (TestChunk *tc = (TestChunk *)lb->first, *tb_next; tc; tc = tb_next) { tb_next = tc->next; MEM_freeN((void *)tc->data); MEM_freeN(tc); @@ -77,12 +77,12 @@ static char *testchunk_as_data( size_t *r_data_len) { size_t data_len = 0; - for (TestChunnk *tc = (TestChunnk *)lb->first; tc; tc = tc->next) { + for (TestChunk *tc = (TestChunk *)lb->first; tc; tc = tc->next) { data_len += tc->data_len; } char *data = (char *)MEM_mallocN(data_len, __func__); size_t i = 0; - for (TestChunnk *tc = (TestChunnk *)lb->first; tc; tc = tc->next) { + for (TestChunk *tc = (TestChunk *)lb->first; tc; tc = tc->next) { memcpy(&data[i], tc->data, tc->data_len); data_len += tc->data_len; i += tc->data_len; @@ -95,7 +95,7 @@ static char *testchunk_as_data( #endif static char *testchunk_as_data_array( - TestChunnk **tc_array, int tc_array_len, + TestChunk **tc_array, int tc_array_len, size_t *r_data_len) { size_t data_len = 0; @@ -105,7 +105,7 @@ static char *testchunk_as_data_array( char *data = (char *)MEM_mallocN(data_len, __func__); size_t i = 0; for (int tc_index = 0; tc_index < tc_array_len; tc_index++) { - TestChunnk *tc = tc_array[tc_index]; + TestChunk *tc = tc_array[tc_index]; memcpy(&data[i], tc->data, tc->data_len); i += tc->data_len; } @@ -677,9 +677,9 @@ static void random_chunk_mutate_helper( ListBase random_chunks; BLI_listbase_clear(&random_chunks); random_chunk_generate(&random_chunks, chunks_per_buffer, stride, chunk_count, random_seed); - TestChunnk **chunks_array = (TestChunnk **)MEM_mallocN(chunks_per_buffer * sizeof(TestChunnk *), __func__); + TestChunk **chunks_array = (TestChunk **)MEM_mallocN(chunks_per_buffer * sizeof(TestChunk *), __func__); { - TestChunnk *tc = (TestChunnk *)random_chunks.first; + TestChunk *tc = (TestChunk *)random_chunks.first; for (int i = 0; i < chunks_per_buffer; i++, tc = tc->next) { chunks_array[i] = tc; } @@ -692,7 +692,7 @@ static void random_chunk_mutate_helper( { RNG *rng = BLI_rng_new(random_seed); for (int i = 0; i < items_total; i++) { - BLI_rng_shuffle_array(rng, chunks_array, sizeof(TestChunnk *), chunks_per_buffer); + BLI_rng_shuffle_array(rng, chunks_array, sizeof(TestChunk *), chunks_per_buffer); size_t data_len; char *data = testchunk_as_data_array(chunks_array, chunks_per_buffer, &data_len); BLI_assert(data_len == chunks_per_buffer * chunk_count * stride); diff --git a/tests/gtests/blenlib/BLI_path_util_test.cc b/tests/gtests/blenlib/BLI_path_util_test.cc index d017ab18b4d..ef469da50b2 100644 --- a/tests/gtests/blenlib/BLI_path_util_test.cc +++ b/tests/gtests/blenlib/BLI_path_util_test.cc @@ -5,6 +5,7 @@ extern "C" { #include "BLI_fileops.h" #include "BLI_path_util.h" +#include "BLI_string.h" #include "../../../source/blender/imbuf/IMB_imbuf.h" #ifdef _WIN32 @@ -57,7 +58,7 @@ char *zLhm65070058860608_br_find_exe(const char *default_exe) /* BLI_cleanup_path */ #ifndef _WIN32 -TEST(path_util, PathUtilClean) +TEST(path_util, Clean) { /* "/./" -> "/" */ { @@ -113,8 +114,238 @@ TEST(path_util, PathUtilClean) } #endif + +#define AT_INDEX(str_input, index_input, str_expect) \ + { \ + char path[] = str_input; \ + const char *expect = str_expect; \ + int index_output, len_output; \ + const bool ret = BLI_path_name_at_index(path, index_input, &index_output, &len_output); \ + if (expect == NULL) { \ + EXPECT_EQ(ret, false); \ + } \ + else { \ + EXPECT_EQ(ret, true); \ + EXPECT_EQ(strlen(expect), len_output); \ + path[index_output + len_output] = '\0'; \ + EXPECT_STREQ(&path[index_output], expect); \ + } \ + }((void)0) + +/* BLI_path_name_at_index */ +TEST(path_util, NameAtIndex_Single) +{ + AT_INDEX("/a", 0, "a"); + AT_INDEX("/a/", 0, "a"); + AT_INDEX("a/", 0, "a"); + AT_INDEX("//a//", 0, "a"); + AT_INDEX("a/b", 0, "a"); + + AT_INDEX("/a", 1, NULL); + AT_INDEX("/a/", 1, NULL); + AT_INDEX("a/", 1, NULL); + AT_INDEX("//a//", 1, NULL); +} +TEST(path_util, NameAtIndex_SingleNeg) +{ + AT_INDEX("/a", -1, "a"); + AT_INDEX("/a/", -1, "a"); + AT_INDEX("a/", -1, "a"); + AT_INDEX("//a//", -1, "a"); + AT_INDEX("a/b", -1, "b"); + + AT_INDEX("/a", -2, NULL); + AT_INDEX("/a/", -2, NULL); + AT_INDEX("a/", -2, NULL); + AT_INDEX("//a//", -2, NULL); +} + +TEST(path_util, NameAtIndex_Double) +{ + AT_INDEX("/ab", 0, "ab"); + AT_INDEX("/ab/", 0, "ab"); + AT_INDEX("ab/", 0, "ab"); + AT_INDEX("//ab//", 0, "ab"); + AT_INDEX("ab/c", 0, "ab"); + + AT_INDEX("/ab", 1, NULL); + AT_INDEX("/ab/", 1, NULL); + AT_INDEX("ab/", 1, NULL); + AT_INDEX("//ab//", 1, NULL); +} + +TEST(path_util, NameAtIndex_DoublNeg) +{ + AT_INDEX("/ab", -1, "ab"); + AT_INDEX("/ab/", -1, "ab"); + AT_INDEX("ab/", -1, "ab"); + AT_INDEX("//ab//", -1, "ab"); + AT_INDEX("ab/c", -1, "c"); + + AT_INDEX("/ab", -2, NULL); + AT_INDEX("/ab/", -2, NULL); + AT_INDEX("ab/", -2, NULL); + AT_INDEX("//ab//", -2, NULL); +} + +TEST(path_util, NameAtIndex_Misc) +{ + AT_INDEX("/how/now/brown/cow", 0, "how"); + AT_INDEX("/how/now/brown/cow", 1, "now"); + AT_INDEX("/how/now/brown/cow", 2, "brown"); + AT_INDEX("/how/now/brown/cow", 3, "cow"); + AT_INDEX("/how/now/brown/cow", 4, NULL); + AT_INDEX("/how/now/brown/cow/", 4, NULL); +} + +TEST(path_util, NameAtIndex_MiscNeg) +{ + AT_INDEX("/how/now/brown/cow", 0, "how"); + AT_INDEX("/how/now/brown/cow", 1, "now"); + AT_INDEX("/how/now/brown/cow", 2, "brown"); + AT_INDEX("/how/now/brown/cow", 3, "cow"); + AT_INDEX("/how/now/brown/cow", 4, NULL); + AT_INDEX("/how/now/brown/cow/", 4, NULL); +} + +TEST(path_util, NameAtIndex_MiscComplex) +{ + AT_INDEX("how//now/brown/cow", 0, "how"); + AT_INDEX("//how///now\\/brown/cow", 1, "now"); + AT_INDEX("/how/now\\//brown\\/cow", 2, "brown"); + AT_INDEX("/how/now/brown/cow//\\", 3, "cow"); + AT_INDEX("/how/now/brown/\\cow", 4, NULL); + AT_INDEX("how/now/brown/\\cow\\", 4, NULL); +} + +TEST(path_util, NameAtIndex_MiscComplexNeg) +{ + AT_INDEX("how//now/brown/cow", -4, "how"); + AT_INDEX("//how///now\\/brown/cow", -3, "now"); + AT_INDEX("/how/now\\//brown\\/cow", -2, "brown"); + AT_INDEX("/how/now/brown/cow//\\", -1, "cow"); + AT_INDEX("/how/now/brown/\\cow", -5, NULL); + AT_INDEX("how/now/brown/\\cow\\", -5, NULL); +} + +TEST(path_util, NameAtIndex_NoneComplex) +{ + AT_INDEX("", 0, NULL); + AT_INDEX("/", 0, NULL); + AT_INDEX("//", 0, NULL); + AT_INDEX("///", 0, NULL); +} + +TEST(path_util, NameAtIndex_NoneComplexNeg) +{ + AT_INDEX("", -1, NULL); + AT_INDEX("/", -1, NULL); + AT_INDEX("//", -1, NULL); + AT_INDEX("///", -1, NULL); +} + +#undef AT_INDEX + +#define JOIN(str_expect, out_size, ...) \ + { \ + const char *expect = str_expect; \ + char result[(out_size) + 1024]; \ + /* check we don't write past the last byte */ \ + result[out_size] = '\0'; \ + BLI_path_join(result, out_size, __VA_ARGS__, NULL); \ + /* simplify expected string */ \ + BLI_str_replace_char(result, '\\', '/'); \ + EXPECT_STREQ(result, expect); \ + EXPECT_EQ(result[out_size], '\0'); \ + } ((void)0) + +/* BLI_path_join */ +TEST(path_util, JoinNop) +{ + JOIN("", 100, ""); + JOIN("", 100, "", ""); + JOIN("", 100, "", "", ""); + JOIN("/", 100, "/", "", ""); + JOIN("/", 100, "/", "/"); + JOIN("/", 100, "/", "", "/"); + JOIN("/", 100, "/", "", "/", ""); +} + +TEST(path_util, JoinSingle) +{ + JOIN("test", 100, "test"); + JOIN("", 100, ""); + JOIN("a", 100, "a"); + JOIN("/a", 100, "/a"); + JOIN("a/", 100, "a/"); + JOIN("/a/", 100, "/a/"); + JOIN("/a/", 100, "/a//"); + JOIN("//a/", 100, "//a//"); +} + +TEST(path_util, JoinTriple) +{ + JOIN("/a/b/c", 100, "/a", "b", "c"); + JOIN("/a/b/c", 100, "/a/", "/b/", "/c"); + JOIN("/a/b/c", 100, "/a/b/", "/c"); + JOIN("/a/b/c", 100, "/a/b/c"); + JOIN("/a/b/c", 100, "/", "a/b/c"); + + JOIN("/a/b/c/", 100, "/a/", "/b/", "/c/"); + JOIN("/a/b/c/", 100, "/a/b/c/"); + JOIN("/a/b/c/", 100, "/a/b/", "/c/"); + JOIN("/a/b/c/", 100, "/a/b/c", "/"); + JOIN("/a/b/c/", 100, "/", "a/b/c", "/"); +} + +TEST(path_util, JoinTruncateShort) +{ + JOIN("", 1, "/"); + JOIN("/", 2, "/"); + JOIN("a", 2, "", "aa"); + JOIN("a", 2, "", "a/"); + JOIN("a/b", 4, "a", "bc"); + JOIN("ab/", 4, "ab", "c"); + JOIN("/a/", 4, "/a", "b"); + JOIN("/a/", 4, "/a/", "b/"); + JOIN("/a/", 4, "/a", "/b/"); + JOIN("/a/", 4, "/", "a/b/"); + JOIN("//a", 4, "//", "a/b/"); + + JOIN("/a/b", 5, "/a", "b", "c"); +} + +TEST(path_util, JoinTruncateLong) +{ + JOIN("", 1, "//", "//longer", "path"); + JOIN("/", 2, "//", "//longer", "path"); + JOIN("//", 3, "//", "//longer", "path"); + JOIN("//l", 4, "//", "//longer", "path"); + /* snip */ + JOIN("//longe", 8, "//", "//longer", "path"); + JOIN("//longer", 9, "//", "//longer", "path"); + JOIN("//longer/", 10, "//", "//longer", "path"); + JOIN("//longer/p", 11, "//", "//longer", "path"); + JOIN("//longer/pa", 12, "//", "//longer", "path"); + JOIN("//longer/pat", 13, "//", "//longer", "path"); + JOIN("//longer/path", 14, "//", "//longer", "path"); // not truncated + JOIN("//longer/path", 14, "//", "//longer", "path/"); + JOIN("//longer/path/", 15, "//", "//longer", "path/"); // not truncated + JOIN("//longer/path/", 15, "//", "//longer", "path/", "trunc"); + JOIN("//longer/path/t", 16, "//", "//longer", "path/", "trunc"); +} + +TEST(path_util, JoinComplex) +{ + JOIN("/a/b/c/d/e/f/g/", 100, "/", "\\a/b", "//////c/d", "", "e\\\\", "f", "g//"); + JOIN("/aa/bb/cc/dd/ee/ff/gg/", 100, "/", "\\aa/bb", "//////cc/dd", "", "ee\\\\", "ff", "gg//"); + JOIN("1/2/3/", 100, "1", "////////", "", "2", "3\\"); +} + +#undef JOIN + /* BLI_path_frame */ -TEST(path_util, PathUtilFrame) +TEST(path_util, Frame) { bool ret; @@ -177,7 +408,7 @@ TEST(path_util, PathUtilFrame) } /* BLI_split_dirfile */ -TEST(path_util, PathUtilSplitDirfile) +TEST(path_util, SplitDirfile) { { const char *path = ""; diff --git a/tests/python/CMakeLists.txt b/tests/python/CMakeLists.txt index 4b5666e9b5a..f7ca9b02137 100644 --- a/tests/python/CMakeLists.txt +++ b/tests/python/CMakeLists.txt @@ -83,11 +83,14 @@ add_test(script_pyapi_bpy_utils_units ${TEST_BLENDER_EXE} --python ${CMAKE_CURRENT_LIST_DIR}/bl_pyapi_bpy_utils_units.py ) -# test running mathutils testing script add_test(script_pyapi_mathutils ${TEST_BLENDER_EXE} --python ${CMAKE_CURRENT_LIST_DIR}/bl_pyapi_mathutils.py ) +add_test(script_pyapi_idprop ${TEST_BLENDER_EXE} + --python ${CMAKE_CURRENT_LIST_DIR}/bl_pyapi_idprop.py +) + # ------------------------------------------------------------------------------ # MODELING TESTS add_test(bevel ${TEST_BLENDER_EXE} @@ -95,6 +98,11 @@ add_test(bevel ${TEST_BLENDER_EXE} --python-text run_tests ) +add_test(split_faces ${TEST_BLENDER_EXE} + ${TEST_SRC_DIR}/modeling/split_faces_test.blend + --python-text run_tests +) + # ------------------------------------------------------------------------------ # IO TESTS @@ -417,7 +425,12 @@ if(WITH_CYCLES) -idiff "${OPENIMAGEIO_IDIFF}" ) endif() - endmacro() + endmacro() + if(WITH_OPENGL_TESTS) + add_cycles_render_test(opengl) + endif() + add_cycles_render_test(image) + add_cycles_render_test(mblur) add_cycles_render_test(reports) add_cycles_render_test(render) add_cycles_render_test(shader) diff --git a/tests/python/batch_import.py b/tests/python/batch_import.py index 8fc679a7c15..bbe3a70327f 100644 --- a/tests/python/batch_import.py +++ b/tests/python/batch_import.py @@ -48,29 +48,17 @@ import os import sys -def clear_scene(): - import bpy - unique_obs = set() - for scene in bpy.data.scenes: - for obj in scene.objects[:]: - scene.objects.unlink(obj) - unique_obs.add(obj) - - # remove obdata, for now only worry about the startup scene - for bpy_data_iter in (bpy.data.objects, bpy.data.meshes, bpy.data.lamps, bpy.data.cameras): - for id_data in bpy_data_iter: - bpy_data_iter.remove(id_data) - - -def batch_import(operator="", - path="", - save_path="", - match="", - start=0, - end=sys.maxsize, - ): +def batch_import( + operator="", + path="", + save_path="", + match="", + start=0, + end=sys.maxsize, +): import addon_utils _reset_all = addon_utils.reset_all # XXX, hack + _disable_all = addon_utils.disable_all # XXX, hack import fnmatch @@ -116,11 +104,12 @@ def batch_import(operator="", # hack so loading the new file doesn't undo our loaded addons addon_utils.reset_all = lambda: None # XXX, hack + addon_utils.disable_all = lambda: None # XXX, hack - bpy.ops.wm.read_factory_settings() + bpy.ops.wm.read_factory_settings(use_empty=True) addon_utils.reset_all = _reset_all # XXX, hack - clear_scene() + addon_utils.disable_all = _disable_all # XXX, hack result = op(filepath=f) diff --git a/tests/python/bl_mesh_modifiers.py b/tests/python/bl_mesh_modifiers.py index 526a54a49a2..bff2c31984c 100644 --- a/tests/python/bl_mesh_modifiers.py +++ b/tests/python/bl_mesh_modifiers.py @@ -31,7 +31,6 @@ import math USE_QUICK_RENDER = False -IS_BMESH = hasattr(__import__("bpy").types, "LoopColors") # ----------------------------------------------------------------------------- # utility functions @@ -203,13 +202,8 @@ def defaults_object(obj): mesh.show_normal_vertex = True - # lame! - if IS_BMESH: - for poly in mesh.polygons: - poly.use_smooth = True - else: - for face in mesh.faces: - face.use_smooth = True + for poly in mesh.polygons: + poly.use_smooth = True def defaults_modifier(mod): @@ -220,16 +214,14 @@ def defaults_modifier(mod): # ----------------------------------------------------------------------------- # models (utils) +def mesh_bmesh_poly_elems(poly, elems): + vert_start = poly.loop_start + vert_total = poly.loop_total + return elems[vert_start:vert_start + vert_total] -if IS_BMESH: - def mesh_bmesh_poly_elems(poly, elems): - vert_start = poly.loop_start - vert_total = poly.loop_total - return elems[vert_start:vert_start + vert_total] - - def mesh_bmesh_poly_vertices(poly): - return [loop.vertex_index - for loop in mesh_bmesh_poly_elems(poly, poly.id_data.loops)] +def mesh_bmesh_poly_vertices(poly): + return [loop.vertex_index + for loop in mesh_bmesh_poly_elems(poly, poly.id_data.loops)] def mesh_bounds(mesh): @@ -258,21 +250,14 @@ def mesh_uv_add(obj): uv_lay = obj.data.uv_textures.new() - if IS_BMESH: - # XXX, odd that we need to do this. until UV's and texface - # are separated we will need to keep it - uv_loops = obj.data.uv_layers[-1] - uv_list = uv_loops.data[:] - for poly in obj.data.polygons: - poly_uvs = mesh_bmesh_poly_elems(poly, uv_list) - for i, c in enumerate(poly_uvs): - c.uv = uvs[i % 4] - else: - for uv in uv_lay.data: - uv.uv1 = uvs[0] - uv.uv2 = uvs[1] - uv.uv3 = uvs[2] - uv.uv4 = uvs[3] + # XXX, odd that we need to do this. until UV's and texface + # are separated we will need to keep it + uv_loops = obj.data.uv_layers[-1] + uv_list = uv_loops.data[:] + for poly in obj.data.polygons: + poly_uvs = mesh_bmesh_poly_elems(poly, uv_list) + for i, c in enumerate(poly_uvs): + c.uv = uvs[i % 4] return uv_lay @@ -296,21 +281,12 @@ def mesh_vcol_add(obj, mode=0): mesh = obj.data - if IS_BMESH: - col_list = vcol_lay.data[:] - for poly in mesh.polygons: - face_verts = mesh_bmesh_poly_vertices(poly) - poly_cols = mesh_bmesh_poly_elems(poly, col_list) - for i, c in enumerate(poly_cols): - c.color = colors_get(face_verts[i]) - else: - for i, col in enumerate(vcol_lay.data): - face_verts = mesh.faces[i].vertices - col.color1 = colors_get(face_verts[0]) - col.color2 = colors_get(face_verts[1]) - col.color3 = colors_get(face_verts[2]) - if len(face_verts) == 4: - col.color4 = colors_get(face_verts[3]) + col_list = vcol_lay.data[:] + for poly in mesh.polygons: + face_verts = mesh_bmesh_poly_vertices(poly) + poly_cols = mesh_bmesh_poly_elems(poly, col_list) + for i, c in enumerate(poly_cols): + c.color = colors_get(face_verts[i]) return vcol_lay @@ -470,10 +446,7 @@ def modifier_build_add(scene, obj): defaults_modifier(mod) # ensure we display some faces - if IS_BMESH: - totface = len(obj.data.polygons) - else: - totface = len(obj.data.faces) + totface = len(obj.data.polygons) mod.frame_start = totface // 2 mod.frame_duration = totface diff --git a/tests/python/bl_pyapi_idprop.py b/tests/python/bl_pyapi_idprop.py new file mode 100644 index 00000000000..0a9cb044571 --- /dev/null +++ b/tests/python/bl_pyapi_idprop.py @@ -0,0 +1,144 @@ +# Apache License, Version 2.0 + +# ./blender.bin --background -noaudio --python tests/python/bl_pyapi_idprop.py -- --verbose +import bpy +import unittest +from array import array + + +class TestHelper: + + @property + def id(self): + return self._id + + def setUp(self): + self._id = bpy.context.scene + assert(len(self._id.keys()) == 0) + + def tearDown(self): + for key in list(self._id.keys()): + del self._id[key] + + def assertAlmostEqualSeq(self, list1, list2): + self.assertEqual(len(list1), len(list2)) + for v1, v2 in zip(list1, list2): + self.assertAlmostEqual(v1, v2, places=5) + + +class TestIdPropertyCreation(TestHelper, unittest.TestCase): + + def test_name_empty(self): + self.id[""] = 4 + self.assertEqual(self.id[""], 4) + + def test_name_too_long(self): + with self.assertRaises(KeyError): + self.id["name" * 30] = 4 + + def test_int(self): + self.id["a"] = 2 + self.assertEqual(self.id["a"], 2) + self.assertTrue(isinstance(self.id["a"], int)) + + with self.assertRaises(OverflowError): + self.id["a"] = 2 ** 31 # integer <= 2 ** 31-1 + + def test_double(self): + self.id["a"] = 2.5 + self.assertEqual(self.id["a"], 2.5) + self.assertTrue(isinstance(self.id["a"], float)) + + def test_unicode(self): + self.id["a"] = "Hello World" + self.assertEqual(self.id["a"], "Hello World") + self.assertTrue(isinstance(self.id["a"], str)) + + def test_bytes(self): + self.id["a"] = b"Hello World" + self.assertEqual(self.id["a"], b"Hello World") + self.assertTrue(isinstance(self.id["a"], bytes)) + + def test_sequence_double_list(self): + mylist = [1.2, 3.4, 5.6] + self.id["a"] = mylist + self.assertEqual(self.id["a"].to_list(), mylist) + self.assertEqual(self.id["a"].typecode, "d") + + def test_sequence_int_list(self): + mylist = [1, 2, 3] + self.id["a"] = mylist + self.assertEqual(self.id["a"].to_list(), mylist) + self.assertEqual(self.id["a"].typecode, "i") + + def test_sequence_float_array(self): + mylist = [1.2, 3.4, 5.6] + self.id["a"] = array("f", mylist) + self.assertAlmostEqualSeq(self.id["a"].to_list(), mylist) + self.assertEqual(self.id["a"].typecode, "d") + + def test_sequence_double_array(self): + mylist = [1.2, 3.4, 5.6] + self.id["a"] = array("d", mylist) + self.assertAlmostEqualSeq(self.id["a"].to_list(), mylist) + self.assertEqual(self.id["a"].typecode, "d") + + def test_sequence_int_array(self): + mylist = [1, 2, 3] + self.id["a"] = array("i", mylist) + self.assertAlmostEqualSeq(self.id["a"].to_list(), mylist) + self.assertEqual(self.id["a"].typecode, "i") + + def test_sequence_other_array(self): + mylist = [1, 2, 3] + self.id["a"] = array("Q", mylist) + self.assertEqual(self.id["a"].to_list(), mylist) + + def test_sequence_mixed_numerical_type(self): + self.id["a"] = [1, 2, 3.4, 5] + self.assertAlmostEqualSeq(self.id["a"].to_list(), [1.0, 2.0, 3.4, 5.0]) + self.assertEqual(self.id["a"].typecode, "d") + + def test_sequence_str_list(self): + # I'm a bit surprised that this works + mylist = ["abc", "qwe"] + self.id["a"] = mylist + self.assertEqual(self.id["a"], mylist) + + def test_sequence_mixed_type(self): + with self.assertRaises(TypeError): + mylist = ["abc", 3, "qwe", 3.4] + self.id["a"] = mylist + + def test_mapping_simple(self): + mydict = {"1": 10, "2": "20", "3": 30.5} + self.id["a"] = mydict + self.assertEqual(self.id["a"]["1"], mydict["1"]) + self.assertEqual(self.id["a"]["2"], mydict["2"]) + self.assertEqual(self.id["a"]["3"], mydict["3"]) + + def test_mapping_complex(self): + mydict = { + "1": [1, 2, 3], + "2": {"1": "abc", "2": array("i", [4, 5, 6])}, + "3": {"1": {"1": 10}, "2": b"qwe"}, + } + self.id["a"] = mydict + self.assertEqual(self.id["a"]["1"].to_list(), [1, 2, 3]) + self.assertEqual(self.id["a"]["2"]["1"], "abc") + self.assertEqual(self.id["a"]["2"]["2"].to_list(), [4, 5, 6]) + self.assertEqual(self.id["a"]["3"]["1"]["1"], 10) + self.assertEqual(self.id["a"]["3"]["2"], b"qwe") + + with self.assertRaises(KeyError): + a = self.id["a"]["2"]["a"] + + def test_invalid_type(self): + with self.assertRaises(TypeError): + self.id["a"] = self + + +if __name__ == '__main__': + import sys + sys.argv = [__file__] + (sys.argv[sys.argv.index("--") + 1:] if "--" in sys.argv else []) + unittest.main() diff --git a/tests/python/bl_run_operators.py b/tests/python/bl_run_operators.py index 7e92b424faa..7d5f4127378 100644 --- a/tests/python/bl_run_operators.py +++ b/tests/python/bl_run_operators.py @@ -65,6 +65,7 @@ op_blacklist = ( "wm.blenderplayer_start", "wm.recover_auto_save", "wm.quit_blender", + "wm.window_close", "wm.url_open", "wm.doc_view", "wm.doc_edit", @@ -308,16 +309,7 @@ def run_ops(operators, setup_func=None, reset=True): # contexts def ctx_clear_scene(): # copied from batch_import.py - unique_obs = set() - for scene in bpy.data.scenes: - for obj in scene.objects[:]: - scene.objects.unlink(obj) - unique_obs.add(obj) - - # remove obdata, for now only worry about the startup scene - for bpy_data_iter in (bpy.data.objects, bpy.data.meshes, bpy.data.lamps, bpy.data.cameras): - for id_data in bpy_data_iter: - bpy_data_iter.remove(id_data) + bpy.ops.wm.read_factory_settings(use_empty=True) def ctx_editmode_mesh(): diff --git a/tests/python/cycles_render_tests.py b/tests/python/cycles_render_tests.py index 64a71da301a..a030cc5e0de 100755 --- a/tests/python/cycles_render_tests.py +++ b/tests/python/cycles_render_tests.py @@ -47,20 +47,41 @@ def printMessage(type, status, message): def render_file(filepath): - command = ( - BLENDER, - "--background", - "-noaudio", - "--factory-startup", - "--enable-autoexec", - filepath, - "-E", "CYCLES", - # Run with OSL enabled - # "--python-expr", "import bpy; bpy.context.scene.cycles.shading_system = True", - "-o", TEMP_FILE_MASK, - "-F", "PNG", - "-f", "1", + dirname = os.path.dirname(filepath) + basedir = os.path.dirname(dirname) + subject = os.path.basename(dirname) + if subject == 'opengl': + command = ( + BLENDER, + "--window-geometry", "0", "0", "1", "1", + "-noaudio", + "--factory-startup", + "--enable-autoexec", + filepath, + "-E", "CYCLES", + # Run with OSL enabled + # "--python-expr", "import bpy; bpy.context.scene.cycles.shading_system = True", + "-o", TEMP_FILE_MASK, + "-F", "PNG", + '--python', os.path.join(basedir, + "util", + "render_opengl.py") ) + else: + command = ( + BLENDER, + "--background", + "-noaudio", + "--factory-startup", + "--enable-autoexec", + filepath, + "-E", "CYCLES", + # Run with OSL enabled + # "--python-expr", "import bpy; bpy.context.scene.cycles.shading_system = True", + "-o", TEMP_FILE_MASK, + "-F", "PNG", + "-f", "1", + ) try: output = subprocess.check_output(command) if VERBOSE: |