diff options
author | Sergey Sharybin <sergey.vfx@gmail.com> | 2015-11-22 13:48:33 +0300 |
---|---|---|
committer | Sergey Sharybin <sergey.vfx@gmail.com> | 2015-11-25 11:01:22 +0300 |
commit | 2a5c1fc9ccbabfaef4eeaf90093dfb2ac0acfc90 (patch) | |
tree | 98ee26786419e78d9fa0e325c527bf2e1fa3d905 /intern/cycles/kernel/kernel_path_branched.h | |
parent | 8bca34fe326d10cc2f20df7fa541179e9ba835d2 (diff) |
Cycles: Delay shooting SSS indirect rays
The idea is to delay shooting indirect rays for the SSS sampling and
trace them after the main integration loop was finished.
This reduces GPU stack usage even further and brings it down to around
652MB (comparing to 722MB before the change and 946MB with previous
stable release).
This also solves the speed regression happened in the previous commit
and now simple SSS scene (SSS suzanne on the floor) renders in 0:50
(comparing to 1:16 with previous commit and 1:03 with official release).
Diffstat (limited to 'intern/cycles/kernel/kernel_path_branched.h')
-rw-r--r-- | intern/cycles/kernel/kernel_path_branched.h | 59 |
1 files changed, 51 insertions, 8 deletions
diff --git a/intern/cycles/kernel/kernel_path_branched.h b/intern/cycles/kernel/kernel_path_branched.h index b6f95d6b0d2..ee507364b30 100644 --- a/intern/cycles/kernel/kernel_path_branched.h +++ b/intern/cycles/kernel/kernel_path_branched.h @@ -91,10 +91,27 @@ ccl_device_noinline void kernel_branched_path_surface_indirect_light(KernelGloba float3 tp = throughput; Ray bsdf_ray; - if(!kernel_branched_path_surface_bounce(kg, &bsdf_rng, sd, sc, j, num_samples, &tp, &ps, L, &bsdf_ray)) + if(!kernel_branched_path_surface_bounce(kg, + &bsdf_rng, + sd, + sc, + j, + num_samples, + &tp, + &ps, + L, + &bsdf_ray)) + { continue; + } - kernel_path_indirect(kg, rng, bsdf_ray, tp*num_samples_inv, num_samples, ps, L); + kernel_path_indirect(kg, + rng, + &bsdf_ray, + tp*num_samples_inv, + num_samples, + &ps, + L); /* for render passes, sum and reset indirect light pass variables * for the next samples */ @@ -312,12 +329,25 @@ ccl_device float4 kernel_branched_path_integrate(KernelGlobals *kg, RNG *rng, in 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); + 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 */ @@ -353,15 +383,28 @@ ccl_device float4 kernel_branched_path_integrate(KernelGlobals *kg, RNG *rng, in 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); + 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 */ |