Cycles: Implement denoising option for reducing noise in the rendered image

This commit contains the first part of the new Cycles denoising option,
which filters the resulting image using information gathered during rendering
to get rid of noise while preserving visual features as well as possible.

To use the option, enable it in the render layer options. The default settings
fit a wide range of scenes, but the user can tweak individual settings to
control the tradeoff between a noise-free image, image details, and calculation
time.

Note that the denoiser may still change in the future and that some features
are not implemented yet. The most important missing feature is animation
denoising, which uses information from multiple frames at once to produce a
flicker-free and smoother result. These features will be added in the future.

Finally, thanks to all the people who supported this project:

- Google (through the GSoC) and Theory Studios for sponsoring the development
- The authors of the papers I used for implementing the denoiser (more details
  on them will be included in the technical docs)
- The other Cycles devs for feedback on the code, especially Sergey for
  mentoring the GSoC project and Brecht for the code review!
- And of course the users who helped with testing, reported bugs and things
  that could and/or should work better!

4 files changed, 30 insertions, 17 deletions

diff --git a/intern/cycles/kernel/split/kernel_branched.h b/intern/cycles/kernel/split/kernel_branched.h
index c7bc1b4df0a..dc74a2ada53 100644
--- a/intern/cycles/kernel/split/kernel_branched.h
+++ b/intern/cycles/kernel/split/kernel_branched.h
@@ -76,6 +76,26 @@ ccl_device_noinline bool kernel_split_branched_path_surface_indirect_light_iter(
 	RNG rng = kernel_split_state.rng[ray_index];
 	PathRadiance *L = &kernel_split_state.path_radiance[ray_index];
 	float3 throughput = branched_state->throughput;
+	ccl_global PathState *ps = &kernel_split_state.path_state[ray_index];
+
+	float sum_sample_weight = 0.0f;
+#ifdef __DENOISING_FEATURES__
+	if(ps->denoising_feature_weight > 0.0f) {
+		for(int i = 0; i < sd->num_closure; i++) {
+			const ShaderClosure *sc = &sd->closure[i];
+
+			/* transparency is not handled here, but in outer loop */
+			if(!CLOSURE_IS_BSDF(sc->type) || CLOSURE_IS_BSDF_TRANSPARENT(sc->type)) {
+				continue;
+			}
+
+			sum_sample_weight += sc->sample_weight;
+		}
+	}
+	else {
+		sum_sample_weight = 1.0f;
+	}
+#endif  /* __DENOISING_FEATURES__ */
 
 	for(int i = branched_state->next_closure; i < sd->num_closure; i++) {
 		const ShaderClosure *sc = &sd->closure[i];
@@ -103,7 +123,6 @@ ccl_device_noinline bool kernel_split_branched_path_surface_indirect_light_iter(
 		RNG bsdf_rng = cmj_hash(rng, i);
 
 		for(int j = branched_state->next_sample; j < num_samples; j++) {
-			ccl_global PathState *ps = &kernel_split_state.path_state[ray_index];
 			if(reset_path_state) {
 				*ps = branched_state->path_state;
 			}
@@ -122,7 +141,8 @@ ccl_device_noinline bool kernel_split_branched_path_surface_indirect_light_iter(
 			                                        tp,
 			                                        ps,
 			                                        L,
-			                                        bsdf_ray))
+			                                        bsdf_ray,
+			                                        sum_sample_weight))
 			{
 				continue;
 			}
diff --git a/intern/cycles/kernel/split/kernel_buffer_update.h b/intern/cycles/kernel/split/kernel_buffer_update.h
index 859c221d976..1f6dce0253c 100644
--- a/intern/cycles/kernel/split/kernel_buffer_update.h
+++ b/intern/cycles/kernel/split/kernel_buffer_update.h
@@ -111,24 +111,15 @@ ccl_device void kernel_buffer_update(KernelGlobals *kg,
 	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);
+		bool is_shadow_catcher = (state->flag & PATH_RAY_SHADOW_CATCHER);
+		kernel_write_result(kg, buffer, sample, L, 1.0f - (*L_transparent), is_shadow_catcher);
+
 		path_rng_end(kg, rng_state, rng);
 
 		ASSIGN_RAY_STATE(ray_state, ray_index, RAY_TO_REGENERATE);
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 87498910d38..670a557f084 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
@@ -125,7 +125,7 @@ ccl_device void kernel_holdout_emission_blurring_pathtermination_ao(
 #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->flag |= (PATH_RAY_SHADOW_CATCHER | PATH_RAY_SHADOW_CATCHER_ONLY | PATH_RAY_STORE_SHADOW_INFO);
 				state->catcher_object = sd->object;
 				if(!kernel_data.background.transparent) {
 					PathRadiance *L = &kernel_split_state.path_radiance[ray_index];
@@ -246,6 +246,8 @@ ccl_device void kernel_holdout_emission_blurring_pathtermination_ao(
 					kernel_split_state.throughput[ray_index] = throughput/probability;
 				}
 			}
+
+			kernel_update_denoising_features(kg, sd, state, L);
 		}
 	}
 
diff --git a/intern/cycles/kernel/split/kernel_shadow_blocked_dl.h b/intern/cycles/kernel/split/kernel_shadow_blocked_dl.h
index 452b6e45a36..386fbbc4d09 100644
--- a/intern/cycles/kernel/split/kernel_shadow_blocked_dl.h
+++ b/intern/cycles/kernel/split/kernel_shadow_blocked_dl.h
@@ -89,10 +89,10 @@ ccl_device void kernel_shadow_blocked_dl(KernelGlobals *kg)
 			               &shadow))
 		{
 			/* accumulate */
-			path_radiance_accum_light(L, throughput, &L_light, shadow, 1.0f, state->bounce, is_lamp);
+			path_radiance_accum_light(L, state, throughput, &L_light, shadow, 1.0f, is_lamp);
 		}
 		else {
-			path_radiance_accum_total_light(L, throughput, &L_light);
+			path_radiance_accum_total_light(L, state, throughput, &L_light);
 		}
 	}