diff options
Diffstat (limited to 'intern/cycles/kernel')
-rw-r--r-- | intern/cycles/kernel/CMakeLists.txt | 1 | ||||
-rw-r--r-- | intern/cycles/kernel/kernel.cpp | 110 | ||||
-rw-r--r-- | intern/cycles/kernel/kernel_compat_cpu.h | 2 | ||||
-rw-r--r-- | intern/cycles/kernel/kernel_globals.h | 7 | ||||
-rw-r--r-- | intern/cycles/kernel/kernel_path.h | 2 | ||||
-rw-r--r-- | intern/cycles/kernel/kernel_textures.h | 24 | ||||
-rw-r--r-- | intern/cycles/kernel/svm/svm.h | 7 | ||||
-rw-r--r-- | intern/cycles/kernel/svm/svm_brick.h | 114 | ||||
-rw-r--r-- | intern/cycles/kernel/svm/svm_image.h | 146 | ||||
-rw-r--r-- | intern/cycles/kernel/svm/svm_types.h | 4 |
10 files changed, 269 insertions, 148 deletions
diff --git a/intern/cycles/kernel/CMakeLists.txt b/intern/cycles/kernel/CMakeLists.txt index 98cb16d5dfc..c26954e23b6 100644 --- a/intern/cycles/kernel/CMakeLists.txt +++ b/intern/cycles/kernel/CMakeLists.txt @@ -61,6 +61,7 @@ set(SRC_SVM_HEADERS svm/svm_closure.h svm/svm_convert.h svm/svm_checker.h + svm/svm_brick.h svm/svm_displace.h svm/svm_fresnel.h svm/svm_gamma.h diff --git a/intern/cycles/kernel/kernel.cpp b/intern/cycles/kernel/kernel.cpp index 667db1e5f03..62d79bdd946 100644 --- a/intern/cycles/kernel/kernel.cpp +++ b/intern/cycles/kernel/kernel.cpp @@ -87,14 +87,10 @@ void kernel_tex_copy(KernelGlobals *kg, const char *name, device_ptr mem, size_t else if(strstr(name, "__tex_image_float")) { texture_image_float4 *tex = NULL; int id = atoi(name + strlen("__tex_image_float_")); + int array_index = id; - switch(id) { - case 95: tex = &kg->__tex_image_float_095; break; - case 96: tex = &kg->__tex_image_float_096; break; - case 97: tex = &kg->__tex_image_float_097; break; - case 98: tex = &kg->__tex_image_float_098; break; - case 99: tex = &kg->__tex_image_float_099; break; - default: break; + if (array_index >= 0 && array_index < MAX_FLOAT_IMAGES) { + tex = &kg->texture_float_images[array_index]; } if(tex) { @@ -106,104 +102,10 @@ void kernel_tex_copy(KernelGlobals *kg, const char *name, device_ptr mem, size_t else if(strstr(name, "__tex_image")) { texture_image_uchar4 *tex = NULL; int id = atoi(name + strlen("__tex_image_")); + int array_index = id - MAX_FLOAT_IMAGES; - switch(id) { - case 0: tex = &kg->__tex_image_000; break; - case 1: tex = &kg->__tex_image_001; break; - case 2: tex = &kg->__tex_image_002; break; - case 3: tex = &kg->__tex_image_003; break; - case 4: tex = &kg->__tex_image_004; break; - case 5: tex = &kg->__tex_image_005; break; - case 6: tex = &kg->__tex_image_006; break; - case 7: tex = &kg->__tex_image_007; break; - case 8: tex = &kg->__tex_image_008; break; - case 9: tex = &kg->__tex_image_009; break; - case 10: tex = &kg->__tex_image_010; break; - case 11: tex = &kg->__tex_image_011; break; - case 12: tex = &kg->__tex_image_012; break; - case 13: tex = &kg->__tex_image_013; break; - case 14: tex = &kg->__tex_image_014; break; - case 15: tex = &kg->__tex_image_015; break; - case 16: tex = &kg->__tex_image_016; break; - case 17: tex = &kg->__tex_image_017; break; - case 18: tex = &kg->__tex_image_018; break; - case 19: tex = &kg->__tex_image_019; break; - case 20: tex = &kg->__tex_image_020; break; - case 21: tex = &kg->__tex_image_021; break; - case 22: tex = &kg->__tex_image_022; break; - case 23: tex = &kg->__tex_image_023; break; - case 24: tex = &kg->__tex_image_024; break; - case 25: tex = &kg->__tex_image_025; break; - case 26: tex = &kg->__tex_image_026; break; - case 27: tex = &kg->__tex_image_027; break; - case 28: tex = &kg->__tex_image_028; break; - case 29: tex = &kg->__tex_image_029; break; - case 30: tex = &kg->__tex_image_030; break; - case 31: tex = &kg->__tex_image_031; break; - case 32: tex = &kg->__tex_image_032; break; - case 33: tex = &kg->__tex_image_033; break; - case 34: tex = &kg->__tex_image_034; break; - case 35: tex = &kg->__tex_image_035; break; - case 36: tex = &kg->__tex_image_036; break; - case 37: tex = &kg->__tex_image_037; break; - case 38: tex = &kg->__tex_image_038; break; - case 39: tex = &kg->__tex_image_039; break; - case 40: tex = &kg->__tex_image_040; break; - case 41: tex = &kg->__tex_image_041; break; - case 42: tex = &kg->__tex_image_042; break; - case 43: tex = &kg->__tex_image_043; break; - case 44: tex = &kg->__tex_image_044; break; - case 45: tex = &kg->__tex_image_045; break; - case 46: tex = &kg->__tex_image_046; break; - case 47: tex = &kg->__tex_image_047; break; - case 48: tex = &kg->__tex_image_048; break; - case 49: tex = &kg->__tex_image_049; break; - case 50: tex = &kg->__tex_image_050; break; - case 51: tex = &kg->__tex_image_051; break; - case 52: tex = &kg->__tex_image_052; break; - case 53: tex = &kg->__tex_image_053; break; - case 54: tex = &kg->__tex_image_054; break; - case 55: tex = &kg->__tex_image_055; break; - case 56: tex = &kg->__tex_image_056; break; - case 57: tex = &kg->__tex_image_057; break; - case 58: tex = &kg->__tex_image_058; break; - case 59: tex = &kg->__tex_image_059; break; - case 60: tex = &kg->__tex_image_060; break; - case 61: tex = &kg->__tex_image_061; break; - case 62: tex = &kg->__tex_image_062; break; - case 63: tex = &kg->__tex_image_063; break; - case 64: tex = &kg->__tex_image_064; break; - case 65: tex = &kg->__tex_image_065; break; - case 66: tex = &kg->__tex_image_066; break; - case 67: tex = &kg->__tex_image_067; break; - case 68: tex = &kg->__tex_image_068; break; - case 69: tex = &kg->__tex_image_069; break; - case 70: tex = &kg->__tex_image_070; break; - case 71: tex = &kg->__tex_image_071; break; - case 72: tex = &kg->__tex_image_072; break; - case 73: tex = &kg->__tex_image_073; break; - case 74: tex = &kg->__tex_image_074; break; - case 75: tex = &kg->__tex_image_075; break; - case 76: tex = &kg->__tex_image_076; break; - case 77: tex = &kg->__tex_image_077; break; - case 78: tex = &kg->__tex_image_078; break; - case 79: tex = &kg->__tex_image_079; break; - case 80: tex = &kg->__tex_image_080; break; - case 81: tex = &kg->__tex_image_081; break; - case 82: tex = &kg->__tex_image_082; break; - case 83: tex = &kg->__tex_image_083; break; - case 84: tex = &kg->__tex_image_084; break; - case 85: tex = &kg->__tex_image_085; break; - case 86: tex = &kg->__tex_image_086; break; - case 87: tex = &kg->__tex_image_087; break; - case 88: tex = &kg->__tex_image_088; break; - case 89: tex = &kg->__tex_image_089; break; - case 90: tex = &kg->__tex_image_090; break; - case 91: tex = &kg->__tex_image_091; break; - case 92: tex = &kg->__tex_image_092; break; - case 93: tex = &kg->__tex_image_093; break; - case 94: tex = &kg->__tex_image_094; break; - default: break; + if (array_index >= 0 && array_index < MAX_BYTE_IMAGES) { + tex = &kg->texture_byte_images[array_index]; } if(tex) { diff --git a/intern/cycles/kernel/kernel_compat_cpu.h b/intern/cycles/kernel/kernel_compat_cpu.h index cc8f1f3323b..45f653a686c 100644 --- a/intern/cycles/kernel/kernel_compat_cpu.h +++ b/intern/cycles/kernel/kernel_compat_cpu.h @@ -158,7 +158,7 @@ typedef texture_image<uchar4> texture_image_uchar4; #define kernel_tex_fetch_m128(tex, index) (kg->tex.fetch_m128(index)) #define kernel_tex_fetch_m128i(tex, index) (kg->tex.fetch_m128i(index)) #define kernel_tex_interp(tex, t, size) (kg->tex.interp(t, size)) -#define kernel_tex_image_interp(tex, x, y) (kg->tex.interp(x, y)) +#define kernel_tex_image_interp(tex, x, y) ((tex < MAX_FLOAT_IMAGES) ? kg->texture_float_images[tex].interp(x, y) : kg->texture_byte_images[tex - MAX_FLOAT_IMAGES].interp(x, y)) #define kernel_data (kg->__data) diff --git a/intern/cycles/kernel/kernel_globals.h b/intern/cycles/kernel/kernel_globals.h index a99fffbc519..1e56c11ab90 100644 --- a/intern/cycles/kernel/kernel_globals.h +++ b/intern/cycles/kernel/kernel_globals.h @@ -35,10 +35,15 @@ CCL_NAMESPACE_BEGIN #ifdef __KERNEL_CPU__ +#define MAX_BYTE_IMAGES 512 +#define MAX_FLOAT_IMAGES 5 + typedef struct KernelGlobals { + texture_image_uchar4 texture_byte_images[MAX_BYTE_IMAGES]; + texture_image_float4 texture_float_images[MAX_FLOAT_IMAGES]; #define KERNEL_TEX(type, ttype, name) ttype name; -#define KERNEL_IMAGE_TEX(type, ttype, name) ttype name; +#define KERNEL_IMAGE_TEX(type, ttype, name) #include "kernel_textures.h" KernelData __data; diff --git a/intern/cycles/kernel/kernel_path.h b/intern/cycles/kernel/kernel_path.h index fc67ca98039..8e3a0c6e628 100644 --- a/intern/cycles/kernel/kernel_path.h +++ b/intern/cycles/kernel/kernel_path.h @@ -689,7 +689,7 @@ __device float4 kernel_path_non_progressive(KernelGlobals *kg, RNG *rng, int sam if(kernel_data.integrator.use_ambient_occlusion) { int num_samples = kernel_data.integrator.ao_samples; float num_samples_inv = 1.0f/num_samples; - float ao_factor = kernel_data.background.ao_factor/num_samples; + float ao_factor = kernel_data.background.ao_factor; for(int j = 0; j < num_samples; j++) { /* todo: solve correlation */ diff --git a/intern/cycles/kernel/kernel_textures.h b/intern/cycles/kernel/kernel_textures.h index c1b8eed3dff..4855a948c6e 100644 --- a/intern/cycles/kernel/kernel_textures.h +++ b/intern/cycles/kernel/kernel_textures.h @@ -66,12 +66,14 @@ KERNEL_TEX(float, texture_float, __filter_table) /* sobol */ KERNEL_TEX(uint, texture_uint, __sobol_directions) +/* full-float image */ +KERNEL_IMAGE_TEX(float4, texture_image_float4, __tex_image_float_000) +KERNEL_IMAGE_TEX(float4, texture_image_float4, __tex_image_float_001) +KERNEL_IMAGE_TEX(float4, texture_image_float4, __tex_image_float_002) +KERNEL_IMAGE_TEX(float4, texture_image_float4, __tex_image_float_003) +KERNEL_IMAGE_TEX(float4, texture_image_float4, __tex_image_float_004) + /* image */ -KERNEL_IMAGE_TEX(uchar4, texture_image_uchar4, __tex_image_000) -KERNEL_IMAGE_TEX(uchar4, texture_image_uchar4, __tex_image_001) -KERNEL_IMAGE_TEX(uchar4, texture_image_uchar4, __tex_image_002) -KERNEL_IMAGE_TEX(uchar4, texture_image_uchar4, __tex_image_003) -KERNEL_IMAGE_TEX(uchar4, texture_image_uchar4, __tex_image_004) KERNEL_IMAGE_TEX(uchar4, texture_image_uchar4, __tex_image_005) KERNEL_IMAGE_TEX(uchar4, texture_image_uchar4, __tex_image_006) KERNEL_IMAGE_TEX(uchar4, texture_image_uchar4, __tex_image_007) @@ -162,13 +164,11 @@ KERNEL_IMAGE_TEX(uchar4, texture_image_uchar4, __tex_image_091) KERNEL_IMAGE_TEX(uchar4, texture_image_uchar4, __tex_image_092) KERNEL_IMAGE_TEX(uchar4, texture_image_uchar4, __tex_image_093) KERNEL_IMAGE_TEX(uchar4, texture_image_uchar4, __tex_image_094) - -/* full-float image */ -KERNEL_IMAGE_TEX(float4, texture_image_float4, __tex_image_float_095) -KERNEL_IMAGE_TEX(float4, texture_image_float4, __tex_image_float_096) -KERNEL_IMAGE_TEX(float4, texture_image_float4, __tex_image_float_097) -KERNEL_IMAGE_TEX(float4, texture_image_float4, __tex_image_float_098) -KERNEL_IMAGE_TEX(float4, texture_image_float4, __tex_image_float_099) +KERNEL_IMAGE_TEX(uchar4, texture_image_uchar4, __tex_image_095) +KERNEL_IMAGE_TEX(uchar4, texture_image_uchar4, __tex_image_096) +KERNEL_IMAGE_TEX(uchar4, texture_image_uchar4, __tex_image_097) +KERNEL_IMAGE_TEX(uchar4, texture_image_uchar4, __tex_image_098) +KERNEL_IMAGE_TEX(uchar4, texture_image_uchar4, __tex_image_099) /* packed image (opencl) */ KERNEL_TEX(uchar4, texture_uchar4, __tex_image_packed) diff --git a/intern/cycles/kernel/svm/svm.h b/intern/cycles/kernel/svm/svm.h index 8901e5e9628..5b0f192ea47 100644 --- a/intern/cycles/kernel/svm/svm.h +++ b/intern/cycles/kernel/svm/svm.h @@ -154,6 +154,7 @@ CCL_NAMESPACE_END #include "svm_value.h" #include "svm_voronoi.h" #include "svm_checker.h" +#include "svm_brick.h" CCL_NAMESPACE_BEGIN @@ -220,6 +221,9 @@ __device_noinline void svm_eval_nodes(KernelGlobals *kg, ShaderData *sd, ShaderT case NODE_TEX_IMAGE: svm_node_tex_image(kg, sd, stack, node); break; + case NODE_TEX_IMAGE_BOX: + svm_node_tex_image_box(kg, sd, stack, node); + break; case NODE_TEX_ENVIRONMENT: svm_node_tex_environment(kg, sd, stack, node); break; @@ -249,6 +253,9 @@ __device_noinline void svm_eval_nodes(KernelGlobals *kg, ShaderData *sd, ShaderT case NODE_TEX_CHECKER: svm_node_tex_checker(kg, sd, stack, node, &offset); break; + case NODE_TEX_BRICK: + svm_node_tex_brick(kg, sd, stack, node, &offset); + break; #endif case NODE_CAMERA: svm_node_camera(kg, sd, stack, node.y, node.z, node.w); diff --git a/intern/cycles/kernel/svm/svm_brick.h b/intern/cycles/kernel/svm/svm_brick.h new file mode 100644 index 00000000000..50de19b825d --- /dev/null +++ b/intern/cycles/kernel/svm/svm_brick.h @@ -0,0 +1,114 @@ +/* + * Copyright 2012, Blender Foundation. + * + * 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. + */ + +CCL_NAMESPACE_BEGIN + +/* Brick */ + +__device_noinline float brick_noise(int n) /* fast integer noise */ +{ + int nn; + n = (n >> 13) ^ n; + nn = (n * (n * n * 60493 + 19990303) + 1376312589) & 0x7fffffff; + return 0.5f * ((float)nn / 1073741824.0f); +} + +__device_noinline float svm_brick(float3 p, float scale, float mortar_size, float bias, + float brick_width, float row_height, float offset_amount, int offset_frequency, + float squash_amount, int squash_frequency, float *tint) +{ + p *= scale; + + int bricknum, rownum; + float offset = 0.0f; + float x, y; + + rownum = (int)floor(p.y / row_height); + + if(offset_frequency && squash_frequency) { + brick_width *= ((int)(rownum) % squash_frequency ) ? 1.0f : squash_amount; /* squash */ + offset = ((int)(rownum) % offset_frequency ) ? 0 : (brick_width*offset_amount); /* offset */ + } + + bricknum = (int)floor((p.x+offset) / brick_width); + + x = (p.x+offset) - brick_width*bricknum; + y = p.y - row_height*rownum; + + *tint = clamp((brick_noise((rownum << 16) + (bricknum & 0xFFFF)) + bias), 0.0f, 1.0f); + + return (x < mortar_size || y < mortar_size || + x > (brick_width - mortar_size) || + y > (row_height - mortar_size)) ? 1.0f : 0.0f; +} + +__device void svm_node_tex_brick(KernelGlobals *kg, ShaderData *sd, float *stack, uint4 node, int *offset) +{ + uint4 node2 = read_node(kg, offset); + uint4 node3 = read_node(kg, offset); + + /* Input and Output Sockets */ + uint co_offset, color1_offset, color2_offset, mortar_offset, scale_offset; + uint mortar_size_offset, bias_offset, brick_width_offset, row_height_offset; + uint color_offset, fac_offset; + + /* RNA properties */ + uint offset_frequency, squash_frequency; + + float tint = 0; + + decode_node_uchar4(node.y, &co_offset, &color1_offset, &color2_offset, &mortar_offset); + decode_node_uchar4(node.z, &scale_offset, &mortar_size_offset, &bias_offset, &brick_width_offset); + decode_node_uchar4(node.w, &row_height_offset, &color_offset, &fac_offset, NULL); + + decode_node_uchar4(node2.x, &offset_frequency, &squash_frequency, NULL, NULL); + + float3 co = stack_load_float3(stack, co_offset); + + float3 color1 = stack_load_float3(stack, color1_offset); + float3 color2 = stack_load_float3(stack, color2_offset); + float3 mortar = stack_load_float3(stack, mortar_offset); + + float scale = stack_load_float_default(stack, scale_offset, node2.y); + float mortar_size = stack_load_float_default(stack, mortar_size_offset, node2.z); + float bias = stack_load_float_default(stack, bias_offset, node2.w); + float brick_width = stack_load_float_default(stack, brick_width_offset, node3.x); + float row_height = stack_load_float_default(stack, row_height_offset, node3.y); + float offset_amount = __int_as_float(node3.z); + float squash_amount = __int_as_float(node3.w); + + float f = svm_brick(co, scale, mortar_size, bias, brick_width, row_height, + offset_amount, offset_frequency, squash_amount, squash_frequency, + &tint); + + if(f != 1.0f) { + float facm = 1.0f - tint; + + color1.x = facm * (color1.x) + tint * color2.x; + color1.y = facm * (color1.y) + tint * color2.y; + color1.z = facm * (color1.z) + tint * color2.z; + } + + if(stack_valid(color_offset)) + stack_store_float3(stack, color_offset, (f == 1.0f)? mortar: color1); + if(stack_valid(fac_offset)) + stack_store_float(stack, fac_offset, f); +} + +CCL_NAMESPACE_END + diff --git a/intern/cycles/kernel/svm/svm_image.h b/intern/cycles/kernel/svm/svm_image.h index 3b2b9204d86..662419418e3 100644 --- a/intern/cycles/kernel/svm/svm_image.h +++ b/intern/cycles/kernel/svm/svm_image.h @@ -50,7 +50,7 @@ __device_inline float svm_image_texture_frac(float x, int *ix) return x - (float)i; } -__device float4 svm_image_texture(KernelGlobals *kg, int id, float x, float y) +__device float4 svm_image_texture(KernelGlobals *kg, int id, float x, float y, uint srgb) { uint4 info = kernel_tex_fetch(__tex_image_packed_info, id); uint width = info.x; @@ -82,15 +82,24 @@ __device float4 svm_image_texture(KernelGlobals *kg, int id, float x, float y) r += ty*(1.0f - tx)*svm_image_texture_read(kg, offset + ix + niy*width); r += ty*tx*svm_image_texture_read(kg, offset + nix + niy*width); + if(srgb) { + r.x = color_srgb_to_scene_linear(r.x); + r.y = color_srgb_to_scene_linear(r.y); + r.z = color_srgb_to_scene_linear(r.z); + } + return r; } #else -__device float4 svm_image_texture(KernelGlobals *kg, int id, float x, float y) +__device float4 svm_image_texture(KernelGlobals *kg, int id, float x, float y, uint srgb) { float4 r; +#ifdef __KERNEL_CPU__ + r = kernel_tex_image_interp(id, x, y); +#else /* not particularly proud of this massive switch, what are the * alternatives? * - use a single big 1D texture, and do our own lookup/filtering @@ -101,11 +110,11 @@ __device float4 svm_image_texture(KernelGlobals *kg, int id, float x, float y) * we still need some for other storage */ switch(id) { - case 0: r = kernel_tex_image_interp(__tex_image_000, x, y); break; - case 1: r = kernel_tex_image_interp(__tex_image_001, x, y); break; - case 2: r = kernel_tex_image_interp(__tex_image_002, x, y); break; - case 3: r = kernel_tex_image_interp(__tex_image_003, x, y); break; - case 4: r = kernel_tex_image_interp(__tex_image_004, x, y); break; + case 0: r = kernel_tex_image_interp(__tex_image_float_000, x, y); break; + case 1: r = kernel_tex_image_interp(__tex_image_float_001, x, y); break; + case 2: r = kernel_tex_image_interp(__tex_image_float_002, x, y); break; + case 3: r = kernel_tex_image_interp(__tex_image_float_003, x, y); break; + case 4: r = kernel_tex_image_interp(__tex_image_float_004, x, y); break; case 5: r = kernel_tex_image_interp(__tex_image_005, x, y); break; case 6: r = kernel_tex_image_interp(__tex_image_006, x, y); break; case 7: r = kernel_tex_image_interp(__tex_image_007, x, y); break; @@ -196,15 +205,22 @@ __device float4 svm_image_texture(KernelGlobals *kg, int id, float x, float y) case 92: r = kernel_tex_image_interp(__tex_image_092, x, y); break; case 93: r = kernel_tex_image_interp(__tex_image_093, x, y); break; case 94: r = kernel_tex_image_interp(__tex_image_094, x, y); break; - case 95: r = kernel_tex_image_interp(__tex_image_float_095, x, y); break; - case 96: r = kernel_tex_image_interp(__tex_image_float_096, x, y); break; - case 97: r = kernel_tex_image_interp(__tex_image_float_097, x, y); break; - case 98: r = kernel_tex_image_interp(__tex_image_float_098, x, y); break; - case 99: r = kernel_tex_image_interp(__tex_image_float_099, x, y); break; + case 95: r = kernel_tex_image_interp(__tex_image_095, x, y); break; + case 96: r = kernel_tex_image_interp(__tex_image_096, x, y); break; + case 97: r = kernel_tex_image_interp(__tex_image_097, x, y); break; + case 98: r = kernel_tex_image_interp(__tex_image_098, x, y); break; + case 99: r = kernel_tex_image_interp(__tex_image_099, x, y); break; default: kernel_assert(0); return make_float4(0.0f, 0.0f, 0.0f, 0.0f); } +#endif + + if(srgb) { + r.x = color_srgb_to_scene_linear(r.x); + r.y = color_srgb_to_scene_linear(r.y); + r.z = color_srgb_to_scene_linear(r.z); + } return r; } @@ -219,21 +235,102 @@ __device void svm_node_tex_image(KernelGlobals *kg, ShaderData *sd, float *stack decode_node_uchar4(node.z, &co_offset, &out_offset, &alpha_offset, &srgb); float3 co = stack_load_float3(stack, co_offset); - float4 f = svm_image_texture(kg, id, co.x, co.y); - float3 r = make_float3(f.x, f.y, f.z); + float4 f = svm_image_texture(kg, id, co.x, co.y, srgb); - if(srgb) { - r.x = color_srgb_to_scene_linear(r.x); - r.y = color_srgb_to_scene_linear(r.y); - r.z = color_srgb_to_scene_linear(r.z); + if(stack_valid(out_offset)) + stack_store_float3(stack, out_offset, make_float3(f.x, f.y, f.z)); + if(stack_valid(alpha_offset)) + stack_store_float(stack, alpha_offset, f.w); +} + +__device void svm_node_tex_image_box(KernelGlobals *kg, ShaderData *sd, float *stack, uint4 node) +{ + /* get object space normal */ + float3 N = sd->N; + + N = sd->N; + if(sd->object != ~0) + object_inverse_normal_transform(kg, sd, &N); + + /* project from direction vector to barycentric coordinates in triangles */ + N.x = fabsf(N.x); + N.y = fabsf(N.y); + N.z = fabsf(N.z); + + N /= (N.x + N.y + N.z); + + /* basic idea is to think of this as a triangle, each corner representing + * one of the 3 faces of the cube. in the corners we have single textures, + * in between we blend between two textures, and in the middle we a blend + * between three textures. + * + * the Nxyz values are the barycentric coordinates in an equilateral + * triangle, which in case of blending in the middle has a smaller + * equilateral triangle where 3 textures blend. this divides things into + * 7 zones, with an if() test for each zone */ + + float3 weight = make_float3(0.0f, 0.0f, 0.0f); + float blend = __int_as_float(node.w); + float limit = 0.5f*(1.0f + blend); + + /* first test for corners with single texture */ + if(N.x > limit*(N.x + N.y) && N.x > limit*(N.x + N.z)) { + weight.x = 1.0f; + } + else if(N.y > limit*(N.x + N.y) && N.y > limit*(N.y + N.z)) { + weight.y = 1.0f; } + else if(N.z > limit*(N.x + N.z) && N.z > limit*(N.y + N.z)) { + weight.z = 1.0f; + } + else if(blend > 0.0f) { + /* in case of blending, test for mixes between two textures */ + if(N.z < (1.0f - limit)*(N.y + N.x)) { + weight.x = N.x/(N.x + N.y); + weight.x = clamp((weight.x - 0.5f*(1.0f - blend))/blend, 0.0f, 1.0f); + weight.y = 1.0f - weight.x; + } + else if(N.x < (1.0f - limit)*(N.y + N.z)) { + weight.y = N.y/(N.y + N.z); + weight.y = clamp((weight.y - 0.5f*(1.0f - blend))/blend, 0.0f, 1.0f); + weight.z = 1.0f - weight.y; + } + else if(N.y < (1.0f - limit)*(N.x + N.z)) { + weight.x = N.x/(N.x + N.z); + weight.x = clamp((weight.x - 0.5f*(1.0f - blend))/blend, 0.0f, 1.0f); + weight.z = 1.0f - weight.x; + } + else { + /* last case, we have a mix between three */ + weight.x = ((2.0f - limit)*N.x + (limit - 1.0f))/(2.0f*limit - 1.0f); + weight.y = ((2.0f - limit)*N.y + (limit - 1.0f))/(2.0f*limit - 1.0f); + weight.z = ((2.0f - limit)*N.z + (limit - 1.0f))/(2.0f*limit - 1.0f); + } + } + + /* now fetch textures */ + uint co_offset, out_offset, alpha_offset, srgb; + decode_node_uchar4(node.z, &co_offset, &out_offset, &alpha_offset, &srgb); + + float3 co = stack_load_float3(stack, co_offset); + uint id = node.y; + + float4 f = make_float4(0.0f, 0.0f, 0.0f, 0.0f); + + if(weight.x > 0.0f) + f += weight.x*svm_image_texture(kg, id, co.y, co.z, srgb); + if(weight.y > 0.0f) + f += weight.y*svm_image_texture(kg, id, co.x, co.z, srgb); + if(weight.z > 0.0f) + f += weight.z*svm_image_texture(kg, id, co.y, co.x, srgb); if(stack_valid(out_offset)) - stack_store_float3(stack, out_offset, r); + stack_store_float3(stack, out_offset, make_float3(f.x, f.y, f.z)); if(stack_valid(alpha_offset)) stack_store_float(stack, alpha_offset, f.w); } + __device void svm_node_tex_environment(KernelGlobals *kg, ShaderData *sd, float *stack, uint4 node) { uint id = node.y; @@ -252,17 +349,10 @@ __device void svm_node_tex_environment(KernelGlobals *kg, ShaderData *sd, float else uv = direction_to_mirrorball(co); - float4 f = svm_image_texture(kg, id, uv.x, uv.y); - float3 r = make_float3(f.x, f.y, f.z); - - if(srgb) { - r.x = color_srgb_to_scene_linear(r.x); - r.y = color_srgb_to_scene_linear(r.y); - r.z = color_srgb_to_scene_linear(r.z); - } + float4 f = svm_image_texture(kg, id, uv.x, uv.y, srgb); if(stack_valid(out_offset)) - stack_store_float3(stack, out_offset, r); + stack_store_float3(stack, out_offset, make_float3(f.x, f.y, f.z)); if(stack_valid(alpha_offset)) stack_store_float(stack, alpha_offset, f.w); } diff --git a/intern/cycles/kernel/svm/svm_types.h b/intern/cycles/kernel/svm/svm_types.h index 16c726e7faa..c82eafc790a 100644 --- a/intern/cycles/kernel/svm/svm_types.h +++ b/intern/cycles/kernel/svm/svm_types.h @@ -40,6 +40,7 @@ typedef enum NodeType { NODE_MIX_CLOSURE, NODE_JUMP, NODE_TEX_IMAGE, + NODE_TEX_IMAGE_BOX, NODE_TEX_SKY, NODE_GEOMETRY, NODE_LIGHT_PATH, @@ -89,7 +90,8 @@ typedef enum NodeType { NODE_MIN_MAX, NODE_LIGHT_FALLOFF, NODE_OBJECT_INFO, - NODE_PARTICLE_INFO + NODE_PARTICLE_INFO, + NODE_TEX_BRICK } NodeType; typedef enum NodeAttributeType { |