Cycles: merge of cycles-x branch, a major update to the renderer

This includes much improved GPU rendering performance, viewport interactivity,
new shadow catcher, revamped sampling settings, subsurface scattering anisotropy,
new GPU volume sampling, improved PMJ sampling pattern, and more.

Some features have also been removed or changed, breaking backwards compatibility.
Including the removal of the OpenCL backend, for which alternatives are under
development.

Release notes and code docs:
https://wiki.blender.org/wiki/Reference/Release_Notes/3.0/Cycles
https://wiki.blender.org/wiki/Source/Render/Cycles

Credits:
* Sergey Sharybin
* Brecht Van Lommel
* Patrick Mours (OptiX backend)
* Christophe Hery (subsurface scattering anisotropy)
* William Leeson (PMJ sampling pattern)
* Alaska (various fixes and tweaks)
* Thomas Dinges (various fixes)

For the full commit history, see the cycles-x branch. This squashes together
all the changes since intermediate changes would often fail building or tests.

Ref T87839, T87837, T87836
Fixes T90734, T89353, T80267, T80267, T77185, T69800

1 files changed, 181 insertions, 0 deletions

diff --git a/intern/cycles/kernel/integrator/integrator_init_from_bake.h b/intern/cycles/kernel/integrator/integrator_init_from_bake.h
new file mode 100644
index 00000000000..4898ff936c6
--- /dev/null
+++ b/intern/cycles/kernel/integrator/integrator_init_from_bake.h
@@ -0,0 +1,181 @@
+/*
+ * Copyright 2011-2021 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.
+ */
+
+#pragma once
+
+#include "kernel/kernel_accumulate.h"
+#include "kernel/kernel_adaptive_sampling.h"
+#include "kernel/kernel_camera.h"
+#include "kernel/kernel_path_state.h"
+#include "kernel/kernel_random.h"
+
+#include "kernel/geom/geom.h"
+
+CCL_NAMESPACE_BEGIN
+
+/* This helps with AA but it's not the real solution as it does not AA the geometry
+ * but it's better than nothing, thus committed. */
+ccl_device_inline float bake_clamp_mirror_repeat(float u, float max)
+{
+  /* use mirror repeat (like opengl texture) so that if the barycentric
+   * coordinate goes past the end of the triangle it is not always clamped
+   * to the same value, gives ugly patterns */
+  u /= max;
+  float fu = floorf(u);
+  u = u - fu;
+
+  return ((((int)fu) & 1) ? 1.0f - u : u) * max;
+}
+
+/* Return false to indicate that this pixel is finished.
+ * Used by CPU implementation to not attempt to sample pixel for multiple samples once its known
+ * that the pixel did converge. */
+ccl_device bool integrator_init_from_bake(INTEGRATOR_STATE_ARGS,
+                                          const ccl_global KernelWorkTile *ccl_restrict tile,
+                                          ccl_global float *render_buffer,
+                                          const int x,
+                                          const int y,
+                                          const int scheduled_sample)
+{
+  PROFILING_INIT(kg, PROFILING_RAY_SETUP);
+
+  /* Initialize path state to give basic buffer access and allow early outputs. */
+  path_state_init(INTEGRATOR_STATE_PASS, tile, x, y);
+
+  /* Check whether the pixel has converged and should not be sampled anymore. */
+  if (!kernel_need_sample_pixel(INTEGRATOR_STATE_PASS, render_buffer)) {
+    return false;
+  }
+
+  /* Always count the sample, even if the camera sample will reject the ray. */
+  const int sample = kernel_accum_sample(INTEGRATOR_STATE_PASS, render_buffer, scheduled_sample);
+
+  /* Setup render buffers. */
+  const int index = INTEGRATOR_STATE(path, render_pixel_index);
+  const int pass_stride = kernel_data.film.pass_stride;
+  render_buffer += index * pass_stride;
+
+  ccl_global float *primitive = render_buffer + kernel_data.film.pass_bake_primitive;
+  ccl_global float *differential = render_buffer + kernel_data.film.pass_bake_differential;
+
+  const int seed = __float_as_uint(primitive[0]);
+  int prim = __float_as_uint(primitive[1]);
+  if (prim == -1) {
+    return false;
+  }
+
+  prim += kernel_data.bake.tri_offset;
+
+  /* Random number generator. */
+  const uint rng_hash = hash_uint(seed) ^ kernel_data.integrator.seed;
+
+  float filter_x, filter_y;
+  if (sample == 0) {
+    filter_x = filter_y = 0.5f;
+  }
+  else {
+    path_rng_2D(kg, rng_hash, sample, PRNG_FILTER_U, &filter_x, &filter_y);
+  }
+
+  /* Initialize path state for path integration. */
+  path_state_init_integrator(INTEGRATOR_STATE_PASS, sample, rng_hash);
+
+  /* Barycentric UV with sub-pixel offset. */
+  float u = primitive[2];
+  float v = primitive[3];
+
+  float dudx = differential[0];
+  float dudy = differential[1];
+  float dvdx = differential[2];
+  float dvdy = differential[3];
+
+  if (sample > 0) {
+    u = bake_clamp_mirror_repeat(u + dudx * (filter_x - 0.5f) + dudy * (filter_y - 0.5f), 1.0f);
+    v = bake_clamp_mirror_repeat(v + dvdx * (filter_x - 0.5f) + dvdy * (filter_y - 0.5f),
+                                 1.0f - u);
+  }
+
+  /* Position and normal on triangle. */
+  float3 P, Ng;
+  int shader;
+  triangle_point_normal(kg, kernel_data.bake.object_index, prim, u, v, &P, &Ng, &shader);
+  shader &= SHADER_MASK;
+
+  if (kernel_data.film.pass_background != PASS_UNUSED) {
+    /* Environment baking. */
+
+    /* Setup and write ray. */
+    Ray ray ccl_optional_struct_init;
+    ray.P = zero_float3();
+    ray.D = normalize(P);
+    ray.t = FLT_MAX;
+    ray.time = 0.5f;
+    ray.dP = differential_zero_compact();
+    ray.dD = differential_zero_compact();
+    integrator_state_write_ray(INTEGRATOR_STATE_PASS, &ray);
+
+    /* Setup next kernel to execute. */
+    INTEGRATOR_PATH_INIT(DEVICE_KERNEL_INTEGRATOR_SHADE_BACKGROUND);
+  }
+  else {
+    /* Surface baking. */
+
+    /* Setup ray. */
+    Ray ray ccl_optional_struct_init;
+    ray.P = P + Ng;
+    ray.D = -Ng;
+    ray.t = FLT_MAX;
+    ray.time = 0.5f;
+
+    /* Setup differentials. */
+    float3 dPdu, dPdv;
+    triangle_dPdudv(kg, prim, &dPdu, &dPdv);
+    differential3 dP;
+    dP.dx = dPdu * dudx + dPdv * dvdx;
+    dP.dy = dPdu * dudy + dPdv * dvdy;
+    ray.dP = differential_make_compact(dP);
+    ray.dD = differential_zero_compact();
+
+    /* Write ray. */
+    integrator_state_write_ray(INTEGRATOR_STATE_PASS, &ray);
+
+    /* Setup and write intersection. */
+    Intersection isect ccl_optional_struct_init;
+    isect.object = kernel_data.bake.object_index;
+    isect.prim = prim;
+    isect.u = u;
+    isect.v = v;
+    isect.t = 1.0f;
+    isect.type = PRIMITIVE_TRIANGLE;
+#ifdef __EMBREE__
+    isect.Ng = Ng;
+#endif
+    integrator_state_write_isect(INTEGRATOR_STATE_PASS, &isect);
+
+    /* Setup next kernel to execute. */
+    const int shader_flags = kernel_tex_fetch(__shaders, shader).flags;
+    if ((shader_flags & SD_HAS_RAYTRACE) || (kernel_data.film.pass_ao != PASS_UNUSED)) {
+      INTEGRATOR_PATH_INIT_SORTED(DEVICE_KERNEL_INTEGRATOR_SHADE_SURFACE_RAYTRACE, shader);
+    }
+    else {
+      INTEGRATOR_PATH_INIT_SORTED(DEVICE_KERNEL_INTEGRATOR_SHADE_SURFACE, shader);
+    }
+  }
+
+  return true;
+}
+
+CCL_NAMESPACE_END