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, 65 insertions, 44 deletions

diff --git a/intern/cycles/kernel/closure/volume.h b/intern/cycles/kernel/closure/volume.h
index 1430f712701..69959a3f21b 100644
--- a/intern/cycles/kernel/closure/volume.h
+++ b/intern/cycles/kernel/closure/volume.h
@@ -14,8 +14,7 @@
  * limitations under the License.
  */
 
-#ifndef __VOLUME_H__
-#define __VOLUME_H__
+#pragma once
 
 CCL_NAMESPACE_BEGIN
 
@@ -62,21 +61,12 @@ ccl_device int volume_henyey_greenstein_setup(HenyeyGreensteinVolume *volume)
   return SD_SCATTER;
 }
 
-ccl_device bool volume_henyey_greenstein_merge(const ShaderClosure *a, const ShaderClosure *b)
-{
-  const HenyeyGreensteinVolume *volume_a = (const HenyeyGreensteinVolume *)a;
-  const HenyeyGreensteinVolume *volume_b = (const HenyeyGreensteinVolume *)b;
-
-  return (volume_a->g == volume_b->g);
-}
-
-ccl_device float3 volume_henyey_greenstein_eval_phase(const ShaderClosure *sc,
+ccl_device float3 volume_henyey_greenstein_eval_phase(const ShaderVolumeClosure *svc,
                                                       const float3 I,
                                                       float3 omega_in,
                                                       float *pdf)
 {
-  const HenyeyGreensteinVolume *volume = (const HenyeyGreensteinVolume *)sc;
-  float g = volume->g;
+  float g = svc->g;
 
   /* note that I points towards the viewer */
   if (fabsf(g) < 1e-3f) {
@@ -122,7 +112,7 @@ henyey_greenstrein_sample(float3 D, float g, float randu, float randv, float *pd
   return dir;
 }
 
-ccl_device int volume_henyey_greenstein_sample(const ShaderClosure *sc,
+ccl_device int volume_henyey_greenstein_sample(const ShaderVolumeClosure *svc,
                                                float3 I,
                                                float3 dIdx,
                                                float3 dIdy,
@@ -134,8 +124,7 @@ ccl_device int volume_henyey_greenstein_sample(const ShaderClosure *sc,
                                                float3 *domega_in_dy,
                                                float *pdf)
 {
-  const HenyeyGreensteinVolume *volume = (const HenyeyGreensteinVolume *)sc;
-  float g = volume->g;
+  float g = svc->g;
 
   /* note that I points towards the viewer and so is used negated */
   *omega_in = henyey_greenstrein_sample(-I, g, randu, randv, pdf);
@@ -153,17 +142,15 @@ ccl_device int volume_henyey_greenstein_sample(const ShaderClosure *sc,
 /* VOLUME CLOSURE */
 
 ccl_device float3 volume_phase_eval(const ShaderData *sd,
-                                    const ShaderClosure *sc,
+                                    const ShaderVolumeClosure *svc,
                                     float3 omega_in,
                                     float *pdf)
 {
-  kernel_assert(sc->type == CLOSURE_VOLUME_HENYEY_GREENSTEIN_ID);
-
-  return volume_henyey_greenstein_eval_phase(sc, sd->I, omega_in, pdf);
+  return volume_henyey_greenstein_eval_phase(svc, sd->I, omega_in, pdf);
 }
 
 ccl_device int volume_phase_sample(const ShaderData *sd,
-                                   const ShaderClosure *sc,
+                                   const ShaderVolumeClosure *svc,
                                    float randu,
                                    float randv,
                                    float3 *eval,
@@ -171,31 +158,65 @@ ccl_device int volume_phase_sample(const ShaderData *sd,
                                    differential3 *domega_in,
                                    float *pdf)
 {
-  int label;
-
-  switch (sc->type) {
-    case CLOSURE_VOLUME_HENYEY_GREENSTEIN_ID:
-      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;
-    default:
-      *eval = make_float3(0.0f, 0.0f, 0.0f);
-      label = LABEL_NONE;
-      break;
+  return volume_henyey_greenstein_sample(svc,
+                                         sd->I,
+                                         sd->dI.dx,
+                                         sd->dI.dy,
+                                         randu,
+                                         randv,
+                                         eval,
+                                         omega_in,
+                                         &domega_in->dx,
+                                         &domega_in->dy,
+                                         pdf);
+}
+
+/* Volume sampling utilities. */
+
+/* todo: this value could be tweaked or turned into a probability to avoid
+ * unnecessary work in volumes and subsurface scattering. */
+#define VOLUME_THROUGHPUT_EPSILON 1e-6f
+
+ccl_device float3 volume_color_transmittance(float3 sigma, float t)
+{
+  return exp3(-sigma * t);
+}
+
+ccl_device float volume_channel_get(float3 value, int channel)
+{
+  return (channel == 0) ? value.x : ((channel == 1) ? value.y : value.z);
+}
+
+ccl_device int volume_sample_channel(float3 albedo, float3 throughput, float rand, float3 *pdf)
+{
+  /* Sample color channel proportional to throughput and single scattering
+   * albedo, to significantly reduce noise with many bounce, following:
+   *
+   * "Practical and Controllable Subsurface Scattering for Production Path
+   *  Tracing". Matt Jen-Yuan Chiang, Peter Kutz, Brent Burley. SIGGRAPH 2016. */
+  float3 weights = fabs(throughput * albedo);
+  float sum_weights = weights.x + weights.y + weights.z;
+  float3 weights_pdf;
+
+  if (sum_weights > 0.0f) {
+    weights_pdf = weights / sum_weights;
   }
+  else {
+    weights_pdf = make_float3(1.0f / 3.0f, 1.0f / 3.0f, 1.0f / 3.0f);
+  }
+
+  *pdf = weights_pdf;
 
-  return label;
+  /* OpenCL does not support -> on float3, so don't use pdf->x. */
+  if (rand < weights_pdf.x) {
+    return 0;
+  }
+  else if (rand < weights_pdf.x + weights_pdf.y) {
+    return 1;
+  }
+  else {
+    return 2;
+  }
 }
 
 CCL_NAMESPACE_END
-
-#endif