Merge branch 'master' into blender2.8
[blender.git] / intern / cycles / kernel / kernel_path_branched.h
index ae8a9c9106857b44d4f9d2b3ec8e00d68ef2cb9c..80fcf5b056557153e55a7b03891167cc1a5f4488 100644 (file)
@@ -18,7 +18,12 @@ CCL_NAMESPACE_BEGIN
 
 #ifdef __BRANCHED_PATH__
 
-ccl_device void kernel_branched_path_ao(KernelGlobals *kg, ShaderData *sd, PathRadiance *L, PathState *state, RNG *rng, float3 throughput)
+ccl_device_inline void kernel_branched_path_ao(KernelGlobals *kg,
+                                               ShaderData *sd,
+                                               ShaderData *emission_sd,
+                                               PathRadiance *L,
+                                               ccl_addr_space PathState *state,
+                                               float3 throughput)
 {
        int num_samples = kernel_data.integrator.ao_samples;
        float num_samples_inv = 1.0f/num_samples;
@@ -29,46 +34,225 @@ ccl_device void kernel_branched_path_ao(KernelGlobals *kg, ShaderData *sd, PathR
 
        for(int j = 0; j < num_samples; j++) {
                float bsdf_u, bsdf_v;
-               path_branched_rng_2D(kg, rng, state, j, num_samples, PRNG_BSDF_U, &bsdf_u, &bsdf_v);
+               path_branched_rng_2D(kg, state->rng_hash, state, j, num_samples, PRNG_BSDF_U, &bsdf_u, &bsdf_v);
 
                float3 ao_D;
                float ao_pdf;
 
                sample_cos_hemisphere(ao_N, bsdf_u, bsdf_v, &ao_D, &ao_pdf);
 
-               if(dot(ccl_fetch(sd, Ng), ao_D) > 0.0f && ao_pdf != 0.0f) {
+               if(dot(sd->Ng, ao_D) > 0.0f && ao_pdf != 0.0f) {
                        Ray light_ray;
                        float3 ao_shadow;
 
-                       light_ray.P = ray_offset(ccl_fetch(sd, P), ccl_fetch(sd, Ng));
+                       light_ray.P = ray_offset(sd->P, sd->Ng);
                        light_ray.D = ao_D;
                        light_ray.t = kernel_data.background.ao_distance;
-#ifdef __OBJECT_MOTION__
-                       light_ray.time = ccl_fetch(sd, time);
-#endif
-                       light_ray.dP = ccl_fetch(sd, dP);
+                       light_ray.time = sd->time;
+                       light_ray.dP = sd->dP;
                        light_ray.dD = differential3_zero();
 
-                       if(!shadow_blocked(kg, state, &light_ray, &ao_shadow))
-                               path_radiance_accum_ao(L, throughput*num_samples_inv, ao_alpha, ao_bsdf, ao_shadow, state->bounce);
+                       if(!shadow_blocked(kg, sd, emission_sd, state, &light_ray, &ao_shadow)) {
+                               path_radiance_accum_ao(L, state, throughput*num_samples_inv, ao_alpha, ao_bsdf, ao_shadow);
+                       }
+                       else {
+                               path_radiance_accum_total_ao(L, state, throughput*num_samples_inv, ao_bsdf);
+                       }
                }
        }
 }
 
+#ifndef __SPLIT_KERNEL__
+
+#ifdef __VOLUME__
+ccl_device_forceinline void kernel_branched_path_volume(
+       KernelGlobals *kg,
+       ShaderData *sd,
+       PathState *state,
+       Ray *ray,
+       float3 *throughput,
+       ccl_addr_space Intersection *isect,
+       bool hit,
+       ShaderData *indirect_sd,
+       ShaderData *emission_sd,
+       PathRadiance *L)
+{
+       /* Sanitize volume stack. */
+       if(!hit) {
+               kernel_volume_clean_stack(kg, state->volume_stack);
+       }
+
+       if(state->volume_stack[0].shader == SHADER_NONE) {
+               return;
+       }
+
+       /* volume attenuation, emission, scatter */
+       Ray volume_ray = *ray;
+       volume_ray.t = (hit)? isect->t: FLT_MAX;
+
+       bool heterogeneous = volume_stack_is_heterogeneous(kg, state->volume_stack);
+
+#  ifdef __VOLUME_DECOUPLED__
+       /* decoupled ray marching only supported on CPU */
+       if(kernel_data.integrator.volume_decoupled) {
+               /* cache steps along volume for repeated sampling */
+               VolumeSegment volume_segment;
+
+               shader_setup_from_volume(kg, sd, &volume_ray);
+               kernel_volume_decoupled_record(kg, state,
+                       &volume_ray, sd, &volume_segment, heterogeneous);
+
+               /* direct light sampling */
+               if(volume_segment.closure_flag & SD_SCATTER) {
+                       volume_segment.sampling_method = volume_stack_sampling_method(kg, state->volume_stack);
+
+                       int all = kernel_data.integrator.sample_all_lights_direct;
+
+                       kernel_branched_path_volume_connect_light(kg, sd,
+                               emission_sd, *throughput, state, L, all,
+                               &volume_ray, &volume_segment);
+
+                       /* indirect light sampling */
+                       int num_samples = kernel_data.integrator.volume_samples;
+                       float num_samples_inv = 1.0f/num_samples;
+
+                       for(int j = 0; j < num_samples; j++) {
+                               PathState ps = *state;
+                               Ray pray = *ray;
+                               float3 tp = *throughput;
+
+                               /* branch RNG state */
+                               path_state_branch(&ps, j, num_samples);
+
+                               /* scatter sample. if we use distance sampling and take just one
+                                * sample for direct and indirect light, we could share this
+                                * computation, but makes code a bit complex */
+                               float rphase = path_state_rng_1D(kg, &ps, PRNG_PHASE_CHANNEL);
+                               float rscatter = path_state_rng_1D(kg, &ps, PRNG_SCATTER_DISTANCE);
+
+                               VolumeIntegrateResult result = kernel_volume_decoupled_scatter(kg,
+                                       &ps, &pray, sd, &tp, rphase, rscatter, &volume_segment, NULL, false);
+
+                               if(result == VOLUME_PATH_SCATTERED &&
+                                  kernel_path_volume_bounce(kg,
+                                                            sd,
+                                                            &tp,
+                                                            &ps,
+                                                            &L->state,
+                                                            &pray))
+                               {
+                                       kernel_path_indirect(kg,
+                                                            indirect_sd,
+                                                            emission_sd,
+                                                            &pray,
+                                                            tp*num_samples_inv,
+                                                            &ps,
+                                                            L);
+
+                                       /* for render passes, sum and reset indirect light pass variables
+                                        * for the next samples */
+                                       path_radiance_sum_indirect(L);
+                                       path_radiance_reset_indirect(L);
+                               }
+                       }
+               }
+
+               /* emission and transmittance */
+               if(volume_segment.closure_flag & SD_EMISSION)
+                       path_radiance_accum_emission(L, state, *throughput, volume_segment.accum_emission);
+               *throughput *= volume_segment.accum_transmittance;
+
+               /* free cached steps */
+               kernel_volume_decoupled_free(kg, &volume_segment);
+       }
+       else
+#  endif  /* __VOLUME_DECOUPLED__ */
+       {
+               /* GPU: no decoupled ray marching, scatter probalistically */
+               int num_samples = kernel_data.integrator.volume_samples;
+               float num_samples_inv = 1.0f/num_samples;
+
+               /* todo: we should cache the shader evaluations from stepping
+                * through the volume, for now we redo them multiple times */
+
+               for(int j = 0; j < num_samples; j++) {
+                       PathState ps = *state;
+                       Ray pray = *ray;
+                       float3 tp = (*throughput) * num_samples_inv;
+
+                       /* branch RNG state */
+                       path_state_branch(&ps, j, num_samples);
+
+                       VolumeIntegrateResult result = kernel_volume_integrate(
+                               kg, &ps, sd, &volume_ray, L, &tp, heterogeneous);
+
+#  ifdef __VOLUME_SCATTER__
+                       if(result == VOLUME_PATH_SCATTERED) {
+                               /* todo: support equiangular, MIS and all light sampling.
+                                * alternatively get decoupled ray marching working on the GPU */
+                               kernel_path_volume_connect_light(kg, sd, emission_sd, tp, state, L);
+
+                               if(kernel_path_volume_bounce(kg,
+                                                            sd,
+                                                            &tp,
+                                                            &ps,
+                                                            &L->state,
+                                                            &pray))
+                               {
+                                       kernel_path_indirect(kg,
+                                                            indirect_sd,
+                                                            emission_sd,
+                                                            &pray,
+                                                            tp,
+                                                            &ps,
+                                                            L);
+
+                                       /* for render passes, sum and reset indirect light pass variables
+                                        * for the next samples */
+                                       path_radiance_sum_indirect(L);
+                                       path_radiance_reset_indirect(L);
+                               }
+                       }
+# endif  /* __VOLUME_SCATTER__ */
+               }
+
+               /* todo: avoid this calculation using decoupled ray marching */
+               kernel_volume_shadow(kg, emission_sd, state, &volume_ray, throughput);
+       }
+}
+#endif  /* __VOLUME__ */
 
 /* bounce off surface and integrate indirect light */
 ccl_device_noinline void kernel_branched_path_surface_indirect_light(KernelGlobals *kg,
-       RNG *rng, ShaderData *sd, float3 throughput, float num_samples_adjust,
-       PathState *state, PathRadiance *L)
+       ShaderData *sd, ShaderData *indirect_sd, ShaderData *emission_sd,
+       float3 throughput, float num_samples_adjust, PathState *state, PathRadiance *L)
 {
-       for(int i = 0; i< ccl_fetch(sd, num_closure); i++) {
-               const ShaderClosure *sc = &ccl_fetch(sd, closure)[i];
+       float sum_sample_weight = 0.0f;
+#ifdef __DENOISING_FEATURES__
+       if(state->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 = 0; i < sd->num_closure; i++) {
+               const ShaderClosure *sc = &sd->closure[i];
 
-               if(!CLOSURE_IS_BSDF(sc->type))
-                       continue;
                /* transparency is not handled here, but in outer loop */
-               if(sc->type == CLOSURE_BSDF_TRANSPARENT_ID)
+               if(!CLOSURE_IS_BSDF(sc->type) || CLOSURE_IS_BSDF_TRANSPARENT(sc->type)) {
                        continue;
+               }
 
                int num_samples;
 
@@ -84,22 +268,49 @@ ccl_device_noinline void kernel_branched_path_surface_indirect_light(KernelGloba
                num_samples = ceil_to_int(num_samples_adjust*num_samples);
 
                float num_samples_inv = num_samples_adjust/num_samples;
-               RNG bsdf_rng = cmj_hash(*rng, i);
 
                for(int j = 0; j < num_samples; j++) {
                        PathState ps = *state;
                        float3 tp = throughput;
                        Ray bsdf_ray;
+#ifdef __SHADOW_TRICKS__
+                       float shadow_transparency = L->shadow_transparency;
+#endif
 
-                       if(!kernel_branched_path_surface_bounce(kg, &bsdf_rng, sd, sc, j, num_samples, &tp, &ps, L, &bsdf_ray))
+                       ps.rng_hash = cmj_hash(state->rng_hash, i);
+
+                       if(!kernel_branched_path_surface_bounce(kg,
+                                                               sd,
+                                                               sc,
+                                                               j,
+                                                               num_samples,
+                                                               &tp,
+                                                               &ps,
+                                                               &L->state,
+                                                               &bsdf_ray,
+                                                               sum_sample_weight))
+                       {
                                continue;
+                       }
 
-                       kernel_path_indirect(kg, rng, bsdf_ray, tp*num_samples_inv, num_samples, ps, L);
+                       ps.rng_hash = state->rng_hash;
+
+                       kernel_path_indirect(kg,
+                                            indirect_sd,
+                                            emission_sd,
+                                            &bsdf_ray,
+                                            tp*num_samples_inv,
+                                            &ps,
+                                            L);
 
                        /* for render passes, sum and reset indirect light pass variables
                         * for the next samples */
                        path_radiance_sum_indirect(L);
                        path_radiance_reset_indirect(L);
+
+#ifdef __SHADOW_TRICKS__
+                       L->shadow_transparency = shadow_transparency;
+#endif
                }
        }
 }
@@ -107,319 +318,202 @@ ccl_device_noinline void kernel_branched_path_surface_indirect_light(KernelGloba
 #ifdef __SUBSURFACE__
 ccl_device void kernel_branched_path_subsurface_scatter(KernelGlobals *kg,
                                                         ShaderData *sd,
+                                                        ShaderData *indirect_sd,
+                                                        ShaderData *emission_sd,
                                                         PathRadiance *L,
                                                         PathState *state,
-                                                        RNG *rng,
                                                         Ray *ray,
                                                         float3 throughput)
 {
-       for(int i = 0; i< ccl_fetch(sd, num_closure); i++) {
-               ShaderClosure *sc = &ccl_fetch(sd, closure)[i];
+       for(int i = 0; i < sd->num_closure; i++) {
+               ShaderClosure *sc = &sd->closure[i];
 
                if(!CLOSURE_IS_BSSRDF(sc->type))
                        continue;
 
                /* set up random number generator */
-               uint lcg_state = lcg_state_init(rng, state, 0x68bc21eb);
-               int num_samples = kernel_data.integrator.subsurface_samples;
+               uint lcg_state = lcg_state_init(state, 0x68bc21eb);
+               int num_samples = kernel_data.integrator.subsurface_samples * 3;
                float num_samples_inv = 1.0f/num_samples;
-               RNG bssrdf_rng = cmj_hash(*rng, i);
+               uint bssrdf_rng_hash = cmj_hash(state->rng_hash, i);
 
                /* do subsurface scatter step with copy of shader data, this will
                 * replace the BSSRDF with a diffuse BSDF closure */
                for(int j = 0; j < num_samples; j++) {
-                       ShaderData bssrdf_sd[BSSRDF_MAX_HITS];
+                       PathState hit_state = *state;
+                       path_state_branch(&hit_state, j, num_samples);
+                       hit_state.rng_hash = bssrdf_rng_hash;
+
+                       LocalIntersection ss_isect;
                        float bssrdf_u, bssrdf_v;
-                       path_branched_rng_2D(kg, &bssrdf_rng, state, j, num_samples, PRNG_BSDF_U, &bssrdf_u, &bssrdf_v);
-                       int num_hits = subsurface_scatter_multi_step(kg, sd, bssrdf_sd, state->flag, sc, &lcg_state, bssrdf_u, bssrdf_v, true);
+                       path_state_rng_2D(kg, &hit_state, PRNG_BSDF_U, &bssrdf_u, &bssrdf_v);
+                       int num_hits = subsurface_scatter_multi_intersect(kg,
+                                                                         &ss_isect,
+                                                                         sd,
+                                                                         &hit_state,
+                                                                         sc,
+                                                                         &lcg_state,
+                                                                         bssrdf_u, bssrdf_v,
+                                                                         true);
+
+                       hit_state.rng_offset += PRNG_BOUNCE_NUM;
+
 #ifdef __VOLUME__
                        Ray volume_ray = *ray;
-                       bool need_update_volume_stack = kernel_data.integrator.use_volumes &&
-                                                       ccl_fetch(sd, flag) & SD_OBJECT_INTERSECTS_VOLUME;
-#endif
+                       bool need_update_volume_stack =
+                               kernel_data.integrator.use_volumes &&
+                               sd->object_flag & SD_OBJECT_INTERSECTS_VOLUME;
+#endif  /* __VOLUME__ */
 
                        /* compute lighting with the BSDF closure */
                        for(int hit = 0; hit < num_hits; hit++) {
-                               PathState hit_state = *state;
-
-                               path_state_branch(&hit_state, j, num_samples);
+                               ShaderData bssrdf_sd = *sd;
+                               subsurface_scatter_multi_setup(kg,
+                                                              &ss_isect,
+                                                              hit,
+                                                              &bssrdf_sd,
+                                                              &hit_state,
+                                                              sc);
 
 #ifdef __VOLUME__
                                if(need_update_volume_stack) {
                                        /* Setup ray from previous surface point to the new one. */
-                                       float3 P = ray_offset(bssrdf_sd[hit].P, -bssrdf_sd[hit].Ng);
+                                       float3 P = ray_offset(bssrdf_sd.P, -bssrdf_sd.Ng);
                                        volume_ray.D = normalize_len(P - volume_ray.P,
                                                                     &volume_ray.t);
 
+                                       for(int k = 0; k < VOLUME_STACK_SIZE; k++) {
+                                               hit_state.volume_stack[k] = state->volume_stack[k];
+                                       }
+
                                        kernel_volume_stack_update_for_subsurface(
                                            kg,
+                                           emission_sd,
                                            &volume_ray,
                                            hit_state.volume_stack);
-
-                                       /* Move volume ray forward. */
-                                       volume_ray.P = P;
                                }
-#endif
+#endif  /* __VOLUME__ */
 
 #ifdef __EMISSION__
                                /* direct light */
                                if(kernel_data.integrator.use_direct_light) {
-                                       bool all = kernel_data.integrator.sample_all_lights_direct;
-                                       kernel_branched_path_surface_connect_light(kg, rng,
-                                               &bssrdf_sd[hit], &hit_state, throughput, num_samples_inv, L, all);
+                                       int all = (kernel_data.integrator.sample_all_lights_direct) ||
+                                                 (hit_state.flag & PATH_RAY_SHADOW_CATCHER);
+                                       kernel_branched_path_surface_connect_light(
+                                               kg,
+                                               &bssrdf_sd,
+                                               emission_sd,
+                                               &hit_state,
+                                               throughput,
+                                               num_samples_inv,
+                                               L,
+                                               all);
                                }
-#endif
+#endif  /* __EMISSION__ */
 
                                /* indirect light */
-                               kernel_branched_path_surface_indirect_light(kg, rng,
-                                       &bssrdf_sd[hit], throughput, num_samples_inv,
-                                       &hit_state, L);
+                               kernel_branched_path_surface_indirect_light(
+                                       kg,
+                                       &bssrdf_sd,
+                                       indirect_sd,
+                                       emission_sd,
+                                       throughput,
+                                       num_samples_inv,
+                                       &hit_state,
+                                       L);
                        }
                }
        }
 }
-#endif
-
-ccl_device float4 kernel_branched_path_integrate(KernelGlobals *kg, RNG *rng, int sample, Ray ray, ccl_global float *buffer)
+#endif  /* __SUBSURFACE__ */
+
+ccl_device void kernel_branched_path_integrate(KernelGlobals *kg,
+                                               uint rng_hash,
+                                               int sample,
+                                               Ray ray,
+                                               ccl_global float *buffer,
+                                               PathRadiance *L)
 {
        /* initialize */
-       PathRadiance L;
        float3 throughput = make_float3(1.0f, 1.0f, 1.0f);
-       float L_transparent = 0.0f;
 
-       path_radiance_init(&L, kernel_data.film.use_light_pass);
+       path_radiance_init(L, kernel_data.film.use_light_pass);
 
-       PathState state;
-       path_state_init(kg, &state, rng, sample, &ray);
+       /* shader data memory used for both volumes and surfaces, saves stack space */
+       ShaderData sd;
+       /* shader data used by emission, shadows, volume stacks, indirect path */
+       ShaderDataTinyStorage emission_sd_storage;
+       ShaderData *emission_sd = AS_SHADER_DATA(&emission_sd_storage);
+       ShaderData indirect_sd;
 
-#ifdef __KERNEL_DEBUG__
-       DebugData debug_data;
-       debug_data_init(&debug_data);
-#endif
+       PathState state;
+       path_state_init(kg, emission_sd, &state, rng_hash, sample, &ray);
 
        /* Main Loop
         * Here we only handle transparency intersections from the camera ray.
         * Indirect bounces are handled in kernel_branched_path_surface_indirect_light().
         */
        for(;;) {
-               /* intersect scene */
+               /* Find intersection with objects in scene. */
                Intersection isect;
-               uint visibility = path_state_ray_visibility(kg, &state);
-
-#ifdef __HAIR__
-               float difl = 0.0f, extmax = 0.0f;
-               uint lcg_state = 0;
-
-               if(kernel_data.bvh.have_curves) {
-                       if(kernel_data.cam.resolution == 1) {
-                               float3 pixdiff = ray.dD.dx + ray.dD.dy;
-                               /*pixdiff = pixdiff - dot(pixdiff, ray.D)*ray.D;*/
-                               difl = kernel_data.curve.minimum_width * len(pixdiff) * 0.5f;
-                       }
-
-                       extmax = kernel_data.curve.maximum_width;
-                       lcg_state = lcg_state_init(rng, &state, 0x51633e2d);
-               }
-
-               bool hit = scene_intersect(kg, &ray, visibility, &isect, &lcg_state, difl, extmax);
-#else
-               bool hit = scene_intersect(kg, &ray, visibility, &isect, NULL, 0.0f, 0.0f);
-#endif
-
-#ifdef __KERNEL_DEBUG__
-               debug_data.num_bvh_traversal_steps += isect.num_traversal_steps;
-               debug_data.num_bvh_traversed_instances += isect.num_traversed_instances;
-               debug_data.num_ray_bounces++;
-#endif
+               bool hit = kernel_path_scene_intersect(kg, &state, &ray, &isect, L);
 
 #ifdef __VOLUME__
-               /* volume attenuation, emission, scatter */
-               if(state.volume_stack[0].shader != SHADER_NONE) {
-                       Ray volume_ray = ray;
-                       volume_ray.t = (hit)? isect.t: FLT_MAX;
-                       
-                       bool heterogeneous = volume_stack_is_heterogeneous(kg, state.volume_stack);
-
-#ifdef __VOLUME_DECOUPLED__
-                       /* decoupled ray marching only supported on CPU */
-
-                       /* cache steps along volume for repeated sampling */
-                       VolumeSegment volume_segment;
-                       ShaderData volume_sd;
-
-                       shader_setup_from_volume(kg, &volume_sd, &volume_ray, state.bounce, state.transparent_bounce);
-                       kernel_volume_decoupled_record(kg, &state,
-                               &volume_ray, &volume_sd, &volume_segment, heterogeneous);
-
-                       /* direct light sampling */
-                       if(volume_segment.closure_flag & SD_SCATTER) {
-                               volume_segment.sampling_method = volume_stack_sampling_method(kg, state.volume_stack);
-
-                               bool all = kernel_data.integrator.sample_all_lights_direct;
-
-                               kernel_branched_path_volume_connect_light(kg, rng, &volume_sd,
-                                       throughput, &state, &L, all, &volume_ray, &volume_segment);
-
-                               /* indirect light sampling */
-                               int num_samples = kernel_data.integrator.volume_samples;
-                               float num_samples_inv = 1.0f/num_samples;
-
-                               for(int j = 0; j < num_samples; j++) {
-                                       /* workaround to fix correlation bug in T38710, can find better solution
-                                        * in random number generator later, for now this is done here to not impact
-                                        * performance of rendering without volumes */
-                                       RNG tmp_rng = cmj_hash(*rng, state.rng_offset);
-
-                                       PathState ps = state;
-                                       Ray pray = ray;
-                                       float3 tp = throughput;
-
-                                       /* branch RNG state */
-                                       path_state_branch(&ps, j, num_samples);
-
-                                       /* scatter sample. if we use distance sampling and take just one
-                                        * sample for direct and indirect light, we could share this
-                                        * computation, but makes code a bit complex */
-                                       float rphase = path_state_rng_1D_for_decision(kg, &tmp_rng, &ps, PRNG_PHASE);
-                                       float rscatter = path_state_rng_1D_for_decision(kg, &tmp_rng, &ps, PRNG_SCATTER_DISTANCE);
-
-                                       VolumeIntegrateResult result = kernel_volume_decoupled_scatter(kg,
-                                               &ps, &pray, &volume_sd, &tp, rphase, rscatter, &volume_segment, NULL, false);
-                                               
-                                       (void)result;
-                                       kernel_assert(result == VOLUME_PATH_SCATTERED);
-
-                                       if(kernel_path_volume_bounce(kg, rng, &volume_sd, &tp, &ps, &L, &pray)) {
-                                               kernel_path_indirect(kg, rng, pray, tp*num_samples_inv, num_samples, ps, &L);
-
-                                               /* for render passes, sum and reset indirect light pass variables
-                                                * for the next samples */
-                                               path_radiance_sum_indirect(&L);
-                                               path_radiance_reset_indirect(&L);
-                                       }
-                               }
-                       }
-
-                       /* emission and transmittance */
-                       if(volume_segment.closure_flag & SD_EMISSION)
-                               path_radiance_accum_emission(&L, throughput, volume_segment.accum_emission, state.bounce);
-                       throughput *= volume_segment.accum_transmittance;
-
-                       /* free cached steps */
-                       kernel_volume_decoupled_free(kg, &volume_segment);
-#else
-                       /* GPU: no decoupled ray marching, scatter probalistically */
-                       int num_samples = kernel_data.integrator.volume_samples;
-                       float num_samples_inv = 1.0f/num_samples;
-
-                       /* todo: we should cache the shader evaluations from stepping
-                        * through the volume, for now we redo them multiple times */
-
-                       for(int j = 0; j < num_samples; j++) {
-                               PathState ps = state;
-                               Ray pray = ray;
-                               ShaderData volume_sd;
-                               float3 tp = throughput * num_samples_inv;
-
-                               /* branch RNG state */
-                               path_state_branch(&ps, j, num_samples);
-
-                               VolumeIntegrateResult result = kernel_volume_integrate(
-                                       kg, &ps, &volume_sd, &volume_ray, &L, &tp, rng, heterogeneous);
-                               
-#ifdef __VOLUME_SCATTER__
-                               if(result == VOLUME_PATH_SCATTERED) {
-                                       /* todo: support equiangular, MIS and all light sampling.
-                                        * alternatively get decoupled ray marching working on the GPU */
-                                       kernel_path_volume_connect_light(kg, rng, &volume_sd, tp, &state, &L);
-
-                                       if(kernel_path_volume_bounce(kg, rng, &volume_sd, &tp, &ps, &L, &pray)) {
-                                               kernel_path_indirect(kg, rng, pray, tp, num_samples, ps, &L);
-
-                                               /* for render passes, sum and reset indirect light pass variables
-                                                * for the next samples */
-                                               path_radiance_sum_indirect(&L);
-                                               path_radiance_reset_indirect(&L);
-                                       }
-                               }
-#endif
-                       }
-
-                       /* todo: avoid this calculation using decoupled ray marching */
-                       kernel_volume_shadow(kg, &state, &volume_ray, &throughput);
-#endif
-               }
-#endif
-
+               /* Volume integration. */
+               kernel_branched_path_volume(kg,
+                                           &sd,
+                                           &state,
+                                           &ray,
+                                           &throughput,
+                                           &isect,
+                                           hit,
+                                           &indirect_sd,
+                                           emission_sd,
+                                           L);
+#endif  /* __VOLUME__ */
+
+               /* Shade background. */
                if(!hit) {
-                       /* eval background shader if nothing hit */
-                       if(kernel_data.background.transparent) {
-                               L_transparent += average(throughput);
-
-#ifdef __PASSES__
-                               if(!(kernel_data.film.pass_flag & PASS_BACKGROUND))
-#endif
-                                       break;
-                       }
-
-#ifdef __BACKGROUND__
-                       /* sample background shader */
-                       float3 L_background = indirect_background(kg, &state, &ray);
-                       path_radiance_accum_background(&L, throughput, L_background, state.bounce);
-#endif
-
+                       kernel_path_background(kg, &state, &ray, throughput, &sd, L);
                        break;
                }
 
-               /* setup shading */
-               ShaderData sd;
-               shader_setup_from_ray(kg, &sd, &isect, &ray, state.bounce, state.transparent_bounce);
-               shader_eval_surface(kg, &sd, 0.0f, state.flag, SHADER_CONTEXT_MAIN);
-               shader_merge_closures(&sd);
+               /* Setup and evaluate shader. */
+               shader_setup_from_ray(kg, &sd, &isect, &ray);
 
-               /* holdout */
-#ifdef __HOLDOUT__
-               if(sd.flag & (SD_HOLDOUT|SD_HOLDOUT_MASK)) {
-                       if(kernel_data.background.transparent) {
-                               float3 holdout_weight;
-                               
-                               if(sd.flag & SD_HOLDOUT_MASK)
-                                       holdout_weight = make_float3(1.0f, 1.0f, 1.0f);
-                               else
-                                       holdout_weight = shader_holdout_eval(kg, &sd);
-
-                               /* any throughput is ok, should all be identical here */
-                               L_transparent += average(holdout_weight*throughput);
-                       }
-
-                       if(sd.flag & SD_HOLDOUT_MASK)
-                               break;
-               }
+               /* Skip most work for volume bounding surface. */
+#ifdef __VOLUME__
+               if(!(sd.flag & SD_HAS_ONLY_VOLUME)) {
 #endif
 
-               /* holdout mask objects do not write data passes */
-               kernel_write_data_passes(kg, buffer, &L, &sd, sample, &state, throughput);
+               shader_eval_surface(kg, &sd, &state, state.flag);
+               shader_merge_closures(&sd);
 
-#ifdef __EMISSION__
-               /* emission */
-               if(sd.flag & SD_EMISSION) {
-                       float3 emission = indirect_primitive_emission(kg, &sd, isect.t, state.flag, state.ray_pdf);
-                       path_radiance_accum_emission(&L, throughput, emission, state.bounce);
+               /* Apply shadow catcher, holdout, emission. */
+               if(!kernel_path_shader_apply(kg,
+                                            &sd,
+                                            &state,
+                                            &ray,
+                                            throughput,
+                                            emission_sd,
+                                            L,
+                                            buffer))
+               {
+                       break;
                }
-#endif
 
                /* transparency termination */
                if(state.flag & PATH_RAY_TRANSPARENT) {
                        /* path termination. this is a strange place to put the termination, it's
                         * mainly due to the mixed in MIS that we use. gives too many unneeded
                         * shader evaluations, only need emission if we are going to terminate */
-                       float probability = path_state_terminate_probability(kg, &state, throughput);
+                       float probability = path_state_continuation_probability(kg, &state, throughput);
 
                        if(probability == 0.0f) {
                                break;
                        }
                        else if(probability != 1.0f) {
-                               float terminate = path_state_rng_1D_for_decision(kg, rng, &state, PRNG_TERMINATE);
+                               float terminate = path_state_rng_1D(kg, &state, PRNG_TERMINATE);
 
                                if(terminate >= probability)
                                        break;
@@ -428,107 +522,100 @@ ccl_device float4 kernel_branched_path_integrate(KernelGlobals *kg, RNG *rng, in
                        }
                }
 
+               kernel_update_denoising_features(kg, &sd, &state, L);
+
 #ifdef __AO__
                /* ambient occlusion */
-               if(kernel_data.integrator.use_ambient_occlusion || (sd.flag & SD_AO)) {
-                       kernel_branched_path_ao(kg, &sd, &L, &state, rng, throughput);
+               if(kernel_data.integrator.use_ambient_occlusion) {
+                       kernel_branched_path_ao(kg, &sd, emission_sd, L, &state, throughput);
                }
-#endif
+#endif  /* __AO__ */
 
 #ifdef __SUBSURFACE__
                /* bssrdf scatter to a different location on the same object */
                if(sd.flag & SD_BSSRDF) {
-                       kernel_branched_path_subsurface_scatter(kg, &sd, &L, &state,
-                                                               rng, &ray, throughput);
+                       kernel_branched_path_subsurface_scatter(kg, &sd, &indirect_sd, emission_sd,
+                                                               L, &state, &ray, throughput);
                }
-#endif
+#endif  /* __SUBSURFACE__ */
 
-               if(!(sd.flag & SD_HAS_ONLY_VOLUME)) {
-                       PathState hit_state = state;
+               PathState hit_state = state;
 
 #ifdef __EMISSION__
-                       /* direct light */
-                       if(kernel_data.integrator.use_direct_light) {
-                               bool all = kernel_data.integrator.sample_all_lights_direct;
-                               kernel_branched_path_surface_connect_light(kg, rng,
-                                       &sd, &hit_state, throughput, 1.0f, &L, all);
-                       }
-#endif
+               /* direct light */
+               if(kernel_data.integrator.use_direct_light) {
+                       int all = (kernel_data.integrator.sample_all_lights_direct) ||
+                                         (state.flag & PATH_RAY_SHADOW_CATCHER);
+                       kernel_branched_path_surface_connect_light(kg,
+                               &sd, emission_sd, &hit_state, throughput, 1.0f, L, all);
+               }
+#endif  /* __EMISSION__ */
 
-                       /* indirect light */
-                       kernel_branched_path_surface_indirect_light(kg, rng,
-                               &sd, throughput, 1.0f, &hit_state, &L);
+               /* indirect light */
+               kernel_branched_path_surface_indirect_light(kg,
+                       &sd, &indirect_sd, emission_sd, throughput, 1.0f, &hit_state, L);
 
-                       /* continue in case of transparency */
-                       throughput *= shader_bsdf_transparency(kg, &sd);
+               /* continue in case of transparency */
+               throughput *= shader_bsdf_transparency(kg, &sd);
 
-                       if(is_zero(throughput))
-                               break;
-               }
+               if(is_zero(throughput))
+                       break;
 
                /* Update Path State */
-               state.flag |= PATH_RAY_TRANSPARENT;
-               state.transparent_bounce++;
+               path_state_next(kg, &state, LABEL_TRANSPARENT);
+
+#ifdef __VOLUME__
+               }
+               else {
+                       if(!path_state_volume_next(kg, &state)) {
+                               break;
+                       }
+               }
+#endif
 
                ray.P = ray_offset(sd.P, -sd.Ng);
                ray.t -= sd.ray_length; /* clipping works through transparent */
 
-
 #ifdef __RAY_DIFFERENTIALS__
                ray.dP = sd.dP;
                ray.dD.dx = -sd.dI.dx;
                ray.dD.dy = -sd.dI.dy;
-#endif
+#endif  /* __RAY_DIFFERENTIALS__ */
 
 #ifdef __VOLUME__
                /* enter/exit volume */
                kernel_volume_stack_enter_exit(kg, &sd, state.volume_stack);
-#endif
+#endif  /* __VOLUME__ */
        }
-
-       float3 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
-
-       return make_float4(L_sum.x, L_sum.y, L_sum.z, 1.0f - L_transparent);
 }
 
 ccl_device void kernel_branched_path_trace(KernelGlobals *kg,
-       ccl_global float *buffer, ccl_global uint *rng_state,
+       ccl_global float *buffer,
        int sample, int x, int y, int offset, int stride)
 {
        /* buffer offset */
        int index = offset + x + y*stride;
        int pass_stride = kernel_data.film.pass_stride;
 
-       rng_state += index;
        buffer += index*pass_stride;
 
        /* initialize random numbers and ray */
-       RNG rng;
+       uint rng_hash;
        Ray ray;
 
-       kernel_path_trace_setup(kg, rng_state, sample, x, y, &rng, &ray);
+       kernel_path_trace_setup(kg, sample, x, y, &rng_hash, &ray);
 
        /* integrate */
-       float4 L;
-
-       if(ray.t != 0.0f)
-               L = kernel_branched_path_integrate(kg, &rng, sample, ray, buffer);
-       else
-               L = make_float4(0.0f, 0.0f, 0.0f, 0.0f);
-
-       /* accumulate result in output buffer */
-       kernel_write_pass_float4(buffer, sample, L);
+       PathRadiance L;
 
-       path_rng_end(kg, rng_state, rng);
+       if(ray.t != 0.0f) {
+               kernel_branched_path_integrate(kg, rng_hash, sample, ray, buffer, &L);
+               kernel_write_result(kg, buffer, sample, &L);
+       }
 }
 
+#endif  /* __SPLIT_KERNEL__ */
+
 #endif  /* __BRANCHED_PATH__ */
 
 CCL_NAMESPACE_END
-