Cr8-xyz/blender
1
0
Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity

Cycles: Delay shooting SSS indirect rays

Browse Source 
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).
This commit is contained in:
Sergey Sharybin
2015-11-22 15:48:33 +05:00
parent 8bca34fe32
commit 2a5c1fc9cc
4 changed files with 353 additions and 112 deletions
Download Patch File Download Diff File Expand all files Collapse all files

59
intern/cycles/kernel/kernel_path_branched.h
View File

@@ -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 */