Cycles: Scale denoising albedo contribution of Principled BSDFs according to average fresnel

The Principled BSDF uses Microfacet closures that include a fresnel term,
which are a special case since their weight tends to be near white even
if their average contribution is fairly low.

The sample weight is scaled by the average fresnel weight to account for
this, but the denoising albedo still used the unscaled weight.

This was fine for the original denoiser, but apparently OIDN can't handle
the resulting albedo pass well. Therefore, this commit adds the described
scaling to the albedo pass contribution as well.

This problem was described in T69770.

Reviewed By: brecht

Differential Revision: https://developer.blender.org/D6289

1 files changed, 11 insertions, 1 deletions

diff --git a/intern/cycles/kernel/kernel_passes.h b/intern/cycles/kernel/kernel_passes.h
index c1d74dddc2a..3e423e42573 100644
--- a/intern/cycles/kernel/kernel_passes.h
+++ b/intern/cycles/kernel/kernel_passes.h
@@ -145,7 +145,17 @@ ccl_device_inline void kernel_update_denoising_features(KernelGlobals *kg,
     normal += sc->N * sc->sample_weight;
     sum_weight += sc->sample_weight;
     if (bsdf_get_specular_roughness_squared(sc) > sqr(0.075f)) {
-      albedo += sc->weight;
+      float3 closure_albedo = sc->weight;
+      /* Closures that include a Fresnel term typically have weights close to 1 even though their
+       * actual contribution is significantly lower.
+       * To account for this, we scale their weight by the average fresnel factor (the same is also
+       * done for the sample weight in the BSDF setup, so we don't need to scale that here). */
+      if (CLOSURE_IS_BSDF_MICROFACET_FRESNEL(sc->type)) {
+        MicrofacetBsdf *bsdf = (MicrofacetBsdf *)sc;
+        closure_albedo *= bsdf->extra->fresnel_color;
+      }
+
+      albedo += closure_albedo;
       sum_nonspecular_weight += sc->sample_weight;
     }
   }