kannonier8/blender
1
0
Fork 0
forked from bartvdbraak/blender
Code Pull Requests Activity
51e8c167f4
blender/intern/cycles/kernel/kernel_path.h

959 lines
29 KiB
C
Raw Normal View History

Cycles render engine, initial commit. This is the engine itself, blender modifications and build instructions will follow later. Cycles uses code from some great open source projects, many thanks them: * BVH building and traversal code from NVidia's "Understanding the Efficiency of Ray Traversal on GPUs": http://code.google.com/p/understanding-the-efficiency-of-ray-traversal-on-gpus/ * Open Shading Language for a large part of the shading system: http://code.google.com/p/openshadinglanguage/ * Blender for procedural textures and a few other nodes. * Approximate Catmull Clark subdivision from NVidia Mesh tools: http://code.google.com/p/nvidia-mesh-tools/ * Sobol direction vectors from: http://web.maths.unsw.edu.au/~fkuo/sobol/ * Film response functions from: http://www.cs.columbia.edu/CAVE/software/softlib/dorf.php
2011-04-27 11:58:34 +00:00
/*
Cycles: relicense GNU GPL source code to Apache version 2.0. More information in this post: http://code.blender.org/ Thanks to all contributes for giving their permission!
2013-08-18 14:16:15 +00:00
* Copyright 2011-2013 Blender Foundation
Cycles render engine, initial commit. This is the engine itself, blender modifications and build instructions will follow later. Cycles uses code from some great open source projects, many thanks them: * BVH building and traversal code from NVidia's "Understanding the Efficiency of Ray Traversal on GPUs": http://code.google.com/p/understanding-the-efficiency-of-ray-traversal-on-gpus/ * Open Shading Language for a large part of the shading system: http://code.google.com/p/openshadinglanguage/ * Blender for procedural textures and a few other nodes. * Approximate Catmull Clark subdivision from NVidia Mesh tools: http://code.google.com/p/nvidia-mesh-tools/ * Sobol direction vectors from: http://web.maths.unsw.edu.au/~fkuo/sobol/ * Film response functions from: http://www.cs.columbia.edu/CAVE/software/softlib/dorf.php
2011-04-27 11:58:34 +00:00
*
Cycles: relicense GNU GPL source code to Apache version 2.0. More information in this post: http://code.blender.org/ Thanks to all contributes for giving their permission!
2013-08-18 14:16:15 +00:00
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
Cycles render engine, initial commit. This is the engine itself, blender modifications and build instructions will follow later. Cycles uses code from some great open source projects, many thanks them: * BVH building and traversal code from NVidia's "Understanding the Efficiency of Ray Traversal on GPUs": http://code.google.com/p/understanding-the-efficiency-of-ray-traversal-on-gpus/ * Open Shading Language for a large part of the shading system: http://code.google.com/p/openshadinglanguage/ * Blender for procedural textures and a few other nodes. * Approximate Catmull Clark subdivision from NVidia Mesh tools: http://code.google.com/p/nvidia-mesh-tools/ * Sobol direction vectors from: http://web.maths.unsw.edu.au/~fkuo/sobol/ * Film response functions from: http://www.cs.columbia.edu/CAVE/software/softlib/dorf.php
2011-04-27 11:58:34 +00:00
*
Cycles: relicense GNU GPL source code to Apache version 2.0. More information in this post: http://code.blender.org/ Thanks to all contributes for giving their permission!
2013-08-18 14:16:15 +00:00
* http://www.apache.org/licenses/LICENSE-2.0
Cycles render engine, initial commit. This is the engine itself, blender modifications and build instructions will follow later. Cycles uses code from some great open source projects, many thanks them: * BVH building and traversal code from NVidia's "Understanding the Efficiency of Ray Traversal on GPUs": http://code.google.com/p/understanding-the-efficiency-of-ray-traversal-on-gpus/ * Open Shading Language for a large part of the shading system: http://code.google.com/p/openshadinglanguage/ * Blender for procedural textures and a few other nodes. * Approximate Catmull Clark subdivision from NVidia Mesh tools: http://code.google.com/p/nvidia-mesh-tools/ * Sobol direction vectors from: http://web.maths.unsw.edu.au/~fkuo/sobol/ * Film response functions from: http://www.cs.columbia.edu/CAVE/software/softlib/dorf.php
2011-04-27 11:58:34 +00:00
*
Cycles: relicense GNU GPL source code to Apache version 2.0. More information in this post: http://code.blender.org/ Thanks to all contributes for giving their permission!
2013-08-18 14:16:15 +00:00
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
Cleanup: Fix Cycles Apache header. This was already mixed a bit, but the dot belongs there.
2014-12-25 01:50:24 +00:00
* limitations under the License.
Cycles render engine, initial commit. This is the engine itself, blender modifications and build instructions will follow later. Cycles uses code from some great open source projects, many thanks them: * BVH building and traversal code from NVidia's "Understanding the Efficiency of Ray Traversal on GPUs": http://code.google.com/p/understanding-the-efficiency-of-ray-traversal-on-gpus/ * Open Shading Language for a large part of the shading system: http://code.google.com/p/openshadinglanguage/ * Blender for procedural textures and a few other nodes. * Approximate Catmull Clark subdivision from NVidia Mesh tools: http://code.google.com/p/nvidia-mesh-tools/ * Sobol direction vectors from: http://web.maths.unsw.edu.au/~fkuo/sobol/ * Film response functions from: http://www.cs.columbia.edu/CAVE/software/softlib/dorf.php
2011-04-27 11:58:34 +00:00
*/
Fix #33375: OSL geom:trianglevertices gave wrong coordinates for static BVH. Also some simple OSL optimization, passing thread data pointer directly instead of via thread local storage, and creating ustrings for attribute lookup.
2012-12-01 19:15:05 +00:00
#ifdef __OSL__
Cycles: Cleanup, indent nested preprocessor directives Quite straightforward, main trick is happening in path_source_replace_includes(). Reviewers: brecht, dingto, lukasstockner97, juicyfruit Differential Revision: https://developer.blender.org/D1794
2016-02-12 17:33:43 +00:00
# include "osl_shader.h"
Fix #33375: OSL geom:trianglevertices gave wrong coordinates for static BVH. Also some simple OSL optimization, passing thread data pointer directly instead of via thread local storage, and creating ustrings for attribute lookup.
2012-12-01 19:15:05 +00:00
#endif
Cycles code refactor: move geometry related kernel files into own directory.
2014-03-29 12:03:45 +00:00
#include "kernel_random.h"
Cycles code refactor: move more code to geom folder, add some comments.
2014-03-29 12:03:48 +00:00
#include "kernel_projection.h"
#include "kernel_montecarlo.h"
#include "kernel_differential.h"
#include "kernel_camera.h"
Cycles code refactor: move geometry related kernel files into own directory.
2014-03-29 12:03:45 +00:00
Cycles code refactor: move more code to geom folder, add some comments.
2014-03-29 12:03:48 +00:00
#include "geom/geom.h"
Cycles: Move BVK kernel files to own directory BVH traversal is not really that much a geometry and we've got quite some traversals now. Makes sense to keep them separate in the name of source structure clarity.
2016-07-11 10:28:45 +00:00
#include "bvh/bvh.h"
Cycles code refactor: move geometry related kernel files into own directory.
2014-03-29 12:03:45 +00:00
Cycles: Render Passes Currently supported passes: * Combined, Z, Normal, Object Index, Material Index, Emission, Environment, Diffuse/Glossy/Transmission x Direct/Indirect/Color Not supported yet: * UV, Vector, Mist Only enabled for CPU devices at the moment, will do GPU tweaks tommorrow, also for environment importance sampling. Documentation: http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Passes
2012-01-25 17:23:52 +00:00
#include "kernel_accumulate.h"
Cycles render engine, initial commit. This is the engine itself, blender modifications and build instructions will follow later. Cycles uses code from some great open source projects, many thanks them: * BVH building and traversal code from NVidia's "Understanding the Efficiency of Ray Traversal on GPUs": http://code.google.com/p/understanding-the-efficiency-of-ray-traversal-on-gpus/ * Open Shading Language for a large part of the shading system: http://code.google.com/p/openshadinglanguage/ * Blender for procedural textures and a few other nodes. * Approximate Catmull Clark subdivision from NVidia Mesh tools: http://code.google.com/p/nvidia-mesh-tools/ * Sobol direction vectors from: http://web.maths.unsw.edu.au/~fkuo/sobol/ * Film response functions from: http://www.cs.columbia.edu/CAVE/software/softlib/dorf.php
2011-04-27 11:58:34 +00:00
#include "kernel_shader.h"
#include "kernel_light.h"
Cycles: Render Passes Currently supported passes: * Combined, Z, Normal, Object Index, Material Index, Emission, Environment, Diffuse/Glossy/Transmission x Direct/Indirect/Color Not supported yet: * UV, Vector, Mist Only enabled for CPU devices at the moment, will do GPU tweaks tommorrow, also for environment importance sampling. Documentation: http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Passes
2012-01-25 17:23:52 +00:00
#include "kernel_passes.h"
Cycles render engine, initial commit. This is the engine itself, blender modifications and build instructions will follow later. Cycles uses code from some great open source projects, many thanks them: * BVH building and traversal code from NVidia's "Understanding the Efficiency of Ray Traversal on GPUs": http://code.google.com/p/understanding-the-efficiency-of-ray-traversal-on-gpus/ * Open Shading Language for a large part of the shading system: http://code.google.com/p/openshadinglanguage/ * Blender for procedural textures and a few other nodes. * Approximate Catmull Clark subdivision from NVidia Mesh tools: http://code.google.com/p/nvidia-mesh-tools/ * Sobol direction vectors from: http://web.maths.unsw.edu.au/~fkuo/sobol/ * Film response functions from: http://www.cs.columbia.edu/CAVE/software/softlib/dorf.php
2011-04-27 11:58:34 +00:00
Cycles: initial subsurface multiple scattering support. It's not working as well as I would like, but it works, just add a subsurface scattering node and you can use it like any other BSDF. It is using fully raytraced sampling compatible with progressive rendering and other more advanced rendering algorithms we might used in the future, and it uses no extra memory so it's suitable for complex scenes. Disadvantage is that it can be quite noisy and slow. Two limitations that will be solved are that it does not work with bump mapping yet, and that the falloff function used is a simple cubic function, it's not using the real BSSRDF falloff function yet. The node has a color input, along with a scattering radius for each RGB color channel along with an overall scale factor for the radii. There is also no GPU support yet, will test if I can get that working later. Node Documentation: http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Nodes/Shaders#BSSRDF Implementation notes: http://wiki.blender.org/index.php/Dev:2.6/Source/Render/Cycles/Subsurface_Scattering
2013-04-01 20:26:52 +00:00
#ifdef __SUBSURFACE__
Cycles: Cleanup, indent nested preprocessor directives Quite straightforward, main trick is happening in path_source_replace_includes(). Reviewers: brecht, dingto, lukasstockner97, juicyfruit Differential Revision: https://developer.blender.org/D1794
2016-02-12 17:33:43 +00:00
# include "kernel_subsurface.h"
Cycles: initial subsurface multiple scattering support. It's not working as well as I would like, but it works, just add a subsurface scattering node and you can use it like any other BSDF. It is using fully raytraced sampling compatible with progressive rendering and other more advanced rendering algorithms we might used in the future, and it uses no extra memory so it's suitable for complex scenes. Disadvantage is that it can be quite noisy and slow. Two limitations that will be solved are that it does not work with bump mapping yet, and that the falloff function used is a simple cubic function, it's not using the real BSSRDF falloff function yet. The node has a color input, along with a scattering radius for each RGB color channel along with an overall scale factor for the radii. There is also no GPU support yet, will test if I can get that working later. Node Documentation: http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Nodes/Shaders#BSSRDF Implementation notes: http://wiki.blender.org/index.php/Dev:2.6/Source/Render/Cycles/Subsurface_Scattering
2013-04-01 20:26:52 +00:00
#endif
Cycles Volume Render: support for rendering of homogeneous volume with absorption. This is the simplest possible volume rendering case, constant density inside the volume and no scattering or emission. My plan is to tweak, verify and commit more volume rendering effects one by one, doing it all at once makes it difficult to verify correctness and track down bugs. Documentation is here: http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Materials/Volume Currently this hooks into path tracing in 3 ways, which should get us pretty far until we add more advanced light sampling. These 3 hooks are repeated in the path tracing, branched path tracing and transparent shadow code: * Determine active volume shader at start of the path * Change active volume shader on transmission through a surface * Light attenuation over line segments between camera, surfaces and background This is work by "storm", Stuart Broadfoot, Thomas Dinges and myself.
2013-12-28 15:56:19 +00:00
#ifdef __VOLUME__
Cycles: Cleanup, indent nested preprocessor directives Quite straightforward, main trick is happening in path_source_replace_includes(). Reviewers: brecht, dingto, lukasstockner97, juicyfruit Differential Revision: https://developer.blender.org/D1794
2016-02-12 17:33:43 +00:00
# include "kernel_volume.h"
Cycles Volume Render: support for rendering of homogeneous volume with absorption. This is the simplest possible volume rendering case, constant density inside the volume and no scattering or emission. My plan is to tweak, verify and commit more volume rendering effects one by one, doing it all at once makes it difficult to verify correctness and track down bugs. Documentation is here: http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Materials/Volume Currently this hooks into path tracing in 3 ways, which should get us pretty far until we add more advanced light sampling. These 3 hooks are repeated in the path tracing, branched path tracing and transparent shadow code: * Determine active volume shader at start of the path * Change active volume shader on transmission through a surface * Light attenuation over line segments between camera, surfaces and background This is work by "storm", Stuart Broadfoot, Thomas Dinges and myself.
2013-12-28 15:56:19 +00:00
#endif
Cycles Volume Render: add support for overlapping volume objects. This works pretty much as you would expect, overlapping volume objects gives a more dense volume. What did change is that world volume shaders are now active everywhere, they are no longer excluded inside objects. This may not be desirable and we need to think of better control over this. In some cases you clearly want it to happen, for example if you are rendering a fire in a foggy environment. In other cases like the inside of a house you may not want any fog, but it doesn't seem possible in general for the renderer to automatically determine what is inside or outside of the house. This is implemented using a simple fixed size array of shader/object ID pairs, limited to max 15 overlapping objects. The closures from all shaders are put into a single closure array, exactly the same as if an add shader was used to combine them.
2013-12-28 19:02:40 +00:00
#include "kernel_path_state.h"
Code cleanup: move shadow_blocked function into separate file.
2013-12-28 01:56:53 +00:00
#include "kernel_shadow.h"
Cycles: volume light sampling * Volume multiple importace sampling support to combine equiangular and distance sampling, for both homogeneous and heterogeneous volumes. * Branched path "Sample All Direct Lights" and "Sample All Indirect Lights" now apply to volumes as well as surfaces. Implementation note: For simplicity this is all done with decoupled ray marching, the only case we do not use decoupled is for distance only sampling with one light sample. The homogeneous case should still compile on the GPU because it only requires fixed size storage, but the heterogeneous case will be trickier to get working.
2014-04-04 14:45:49 +00:00
#include "kernel_emission.h"
Cycles: OpenCL kernel split This commit contains all the work related on the AMD megakernel split work which was mainly done by Varun Sundar, George Kyriazis and Lenny Wang, plus some help from Sergey Sharybin, Martijn Berger, Thomas Dinges and likely someone else which we're forgetting to mention. Currently only AMD cards are enabled for the new split kernel, but it is possible to force split opencl kernel to be used by setting the following environment variable: CYCLES_OPENCL_SPLIT_KERNEL_TEST=1. Not all the features are supported yet, and that being said no motion blur, camera blur, SSS and volumetrics for now. Also transparent shadows are disabled on AMD device because of some compiler bug. This kernel is also only implements regular path tracing and supporting branched one will take a bit. Branched path tracing is exposed to the interface still, which is a bit misleading and will be hidden there soon. More feature will be enabled once they're ported to the split kernel and tested. Neither regular CPU nor CUDA has any difference, they're generating the same exact code, which means no regressions/improvements there. Based on the research paper: https://research.nvidia.com/sites/default/files/publications/laine2013hpg_paper.pdf Here's the documentation: https://docs.google.com/document/d/1LuXW-CV-sVJkQaEGZlMJ86jZ8FmoPfecaMdR-oiWbUY/edit Design discussion of the patch: https://developer.blender.org/T44197 Differential Revision: https://developer.blender.org/D1200
2015-05-09 14:34:30 +00:00
#include "kernel_path_common.h"
Cycles code refactor: move some surface and volume path code to separate files.
2014-04-04 12:35:30 +00:00
#include "kernel_path_surface.h"
#include "kernel_path_volume.h"
Cycles: * Fix missing update when editing objects with emission materials. * Fix preview pass rendering set to 1 not showing full resolution. * Fix CUDA runtime compiling failing due to missing cache directory. * Use settings from first render layer for visibility and material override. And a bunch of incomplete and still disabled code mostly related to closure sampling.
2011-09-12 13:13:56 +00:00
Cycles: Add support for debug passes Currently only summed number of traversal steps and intersections used by the camera ray intersection pass is implemented, but in the future we will support more debug passes which would help checking what things makes the scene slow. Example of such extra passes could be number of bounces, time spent on the shader tree evaluation and so. Implementation from the Cycles side is pretty much straightforward, could only mention here that it's a build-time option disabled by default. From the blender side it's implemented as a PASS_DEBUG with several subtypes possible. This way we don't need to create an extra DNA pass type for each of the debug passes, saving us a bits. Reviewers: campbellbarton Reviewed By: campbellbarton Differential Revision: https://developer.blender.org/D813
2014-10-04 13:00:26 +00:00
#ifdef __KERNEL_DEBUG__
Cycles: Cleanup, indent nested preprocessor directives Quite straightforward, main trick is happening in path_source_replace_includes(). Reviewers: brecht, dingto, lukasstockner97, juicyfruit Differential Revision: https://developer.blender.org/D1794
2016-02-12 17:33:43 +00:00
# include "kernel_debug.h"
Cycles: Add support for debug passes Currently only summed number of traversal steps and intersections used by the camera ray intersection pass is implemented, but in the future we will support more debug passes which would help checking what things makes the scene slow. Example of such extra passes could be number of bounces, time spent on the shader tree evaluation and so. Implementation from the Cycles side is pretty much straightforward, could only mention here that it's a build-time option disabled by default. From the blender side it's implemented as a PASS_DEBUG with several subtypes possible. This way we don't need to create an extra DNA pass type for each of the debug passes, saving us a bits. Reviewers: campbellbarton Reviewed By: campbellbarton Differential Revision: https://developer.blender.org/D813
2014-10-04 13:00:26 +00:00
#endif
Code cleanup: move shadow_blocked function into separate file.
2013-12-28 01:56:53 +00:00
CCL_NAMESPACE_BEGIN
Cycles: Render Passes Currently supported passes: * Combined, Z, Normal, Object Index, Material Index, Emission, Environment, Diffuse/Glossy/Transmission x Direct/Indirect/Color Not supported yet: * UV, Vector, Mist Only enabled for CPU devices at the moment, will do GPU tweaks tommorrow, also for environment importance sampling. Documentation: http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Passes
2012-01-25 17:23:52 +00:00
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).
2015-11-22 10:48:33 +00:00
ccl_device void kernel_path_indirect(KernelGlobals *kg,
Cycles CUDA: reduce branched path stack memory by sharing indirect ShaderData. Saves about 15% for the branched path kernel.
2016-05-24 20:28:03 +00:00
ShaderData *sd,
Cycles CUDA: reduce stack memory by reusing ShaderData. 57% less for path and 48% less for branched path.
2016-05-22 20:35:47 +00:00
ShaderData *emission_sd,
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).
2015-11-22 10:48:33 +00:00
RNG *rng,
Ray *ray,
float3 throughput,
int num_samples,
PathState *state,
PathRadiance *L)
Cycles: reduce noise using regular path tracing + subsurface scattering with new cubic and gaussian falloff. Like the branched path tracer, this will now shade all intersection points instead of using one at random.
2013-08-24 15:02:08 +00:00
{
Cycles render engine, initial commit. This is the engine itself, blender modifications and build instructions will follow later. Cycles uses code from some great open source projects, many thanks them: * BVH building and traversal code from NVidia's "Understanding the Efficiency of Ray Traversal on GPUs": http://code.google.com/p/understanding-the-efficiency-of-ray-traversal-on-gpus/ * Open Shading Language for a large part of the shading system: http://code.google.com/p/openshadinglanguage/ * Blender for procedural textures and a few other nodes. * Approximate Catmull Clark subdivision from NVidia Mesh tools: http://code.google.com/p/nvidia-mesh-tools/ * Sobol direction vectors from: http://web.maths.unsw.edu.au/~fkuo/sobol/ * Film response functions from: http://www.cs.columbia.edu/CAVE/software/softlib/dorf.php
2011-04-27 11:58:34 +00:00
/* path iteration */
Code refactor: move random number and MIS variables into PathState. This makes it easier to pass this state around, and wraps some common RNG dimension computations in utility functions.
2014-01-03 01:48:48 +00:00
for(;;) {
Cycles render engine, initial commit. This is the engine itself, blender modifications and build instructions will follow later. Cycles uses code from some great open source projects, many thanks them: * BVH building and traversal code from NVidia's "Understanding the Efficiency of Ray Traversal on GPUs": http://code.google.com/p/understanding-the-efficiency-of-ray-traversal-on-gpus/ * Open Shading Language for a large part of the shading system: http://code.google.com/p/openshadinglanguage/ * Blender for procedural textures and a few other nodes. * Approximate Catmull Clark subdivision from NVidia Mesh tools: http://code.google.com/p/nvidia-mesh-tools/ * Sobol direction vectors from: http://web.maths.unsw.edu.au/~fkuo/sobol/ * Film response functions from: http://www.cs.columbia.edu/CAVE/software/softlib/dorf.php
2011-04-27 11:58:34 +00:00
/* intersect scene */
Intersection isect;
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).
2015-11-22 10:48:33 +00:00
uint visibility = path_state_ray_visibility(kg, state);
bool hit = scene_intersect(kg,
ray,
visibility,
&isect,
NULL,
0.0f, 0.0f);
Cycles render engine, initial commit. This is the engine itself, blender modifications and build instructions will follow later. Cycles uses code from some great open source projects, many thanks them: * BVH building and traversal code from NVidia's "Understanding the Efficiency of Ray Traversal on GPUs": http://code.google.com/p/understanding-the-efficiency-of-ray-traversal-on-gpus/ * Open Shading Language for a large part of the shading system: http://code.google.com/p/openshadinglanguage/ * Blender for procedural textures and a few other nodes. * Approximate Catmull Clark subdivision from NVidia Mesh tools: http://code.google.com/p/nvidia-mesh-tools/ * Sobol direction vectors from: http://web.maths.unsw.edu.au/~fkuo/sobol/ * Film response functions from: http://www.cs.columbia.edu/CAVE/software/softlib/dorf.php
2011-04-27 11:58:34 +00:00
Cycles: multiple importance sampling for lamps, which helps reduce noise for big lamps and sharp glossy reflections. This was already supported for mesh lights and the background, so lamps should do it too. This is not for free and it's a bit slower than I hoped even though there is no extra BVH ray intersection. I'll try to optimize it more later. * Area lights look a bit different now, they had the wrong shape before. * Also fixes a sampling issue in the non-progressive integrator. * Only enabled for the CPU, will test on the GPU later. * An option to disable this will be added for situations where it does not help. Same time comparison before/after: http://www.pasteall.org/pic/show.php?id=43313 http://www.pasteall.org/pic/show.php?id=43314
2013-01-09 21:09:20 +00:00
#ifdef __LAMP_MIS__
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).
2015-11-22 10:48:33 +00:00
if(kernel_data.integrator.use_lamp_mis && !(state->flag & PATH_RAY_CAMERA)) {
Cycles: multiple importance sampling for lamps, which helps reduce noise for big lamps and sharp glossy reflections. This was already supported for mesh lights and the background, so lamps should do it too. This is not for free and it's a bit slower than I hoped even though there is no extra BVH ray intersection. I'll try to optimize it more later. * Area lights look a bit different now, they had the wrong shape before. * Also fixes a sampling issue in the non-progressive integrator. * Only enabled for the CPU, will test on the GPU later. * An option to disable this will be added for situations where it does not help. Same time comparison before/after: http://www.pasteall.org/pic/show.php?id=43313 http://www.pasteall.org/pic/show.php?id=43314
2013-01-09 21:09:20 +00:00
/* ray starting from previous non-transparent bounce */
Ray light_ray;
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).
2015-11-22 10:48:33 +00:00
light_ray.P = ray->P - state->ray_t*ray->D;
state->ray_t += isect.t;
light_ray.D = ray->D;
light_ray.t = state->ray_t;
light_ray.time = ray->time;
light_ray.dD = ray->dD;
light_ray.dP = ray->dP;
Cycles: multiple importance sampling for lamps, which helps reduce noise for big lamps and sharp glossy reflections. This was already supported for mesh lights and the background, so lamps should do it too. This is not for free and it's a bit slower than I hoped even though there is no extra BVH ray intersection. I'll try to optimize it more later. * Area lights look a bit different now, they had the wrong shape before. * Also fixes a sampling issue in the non-progressive integrator. * Only enabled for the CPU, will test on the GPU later. * An option to disable this will be added for situations where it does not help. Same time comparison before/after: http://www.pasteall.org/pic/show.php?id=43313 http://www.pasteall.org/pic/show.php?id=43314
2013-01-09 21:09:20 +00:00
/* intersect with lamp */
float3 emission;
Cycles CUDA: reduce stack memory by reusing ShaderData. 57% less for path and 48% less for branched path.
2016-05-22 20:35:47 +00:00
if(indirect_lamp_emission(kg, emission_sd, state, &light_ray, &emission)) {
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).
2015-11-22 10:48:33 +00:00
path_radiance_accum_emission(L,
throughput,
emission,
state->bounce);
}
Cycles: multiple importance sampling for lamps, which helps reduce noise for big lamps and sharp glossy reflections. This was already supported for mesh lights and the background, so lamps should do it too. This is not for free and it's a bit slower than I hoped even though there is no extra BVH ray intersection. I'll try to optimize it more later. * Area lights look a bit different now, they had the wrong shape before. * Also fixes a sampling issue in the non-progressive integrator. * Only enabled for the CPU, will test on the GPU later. * An option to disable this will be added for situations where it does not help. Same time comparison before/after: http://www.pasteall.org/pic/show.php?id=43313 http://www.pasteall.org/pic/show.php?id=43314
2013-01-09 21:09:20 +00:00
}
#endif
Cycles Volume Render: support for rendering of homogeneous volume with absorption. This is the simplest possible volume rendering case, constant density inside the volume and no scattering or emission. My plan is to tweak, verify and commit more volume rendering effects one by one, doing it all at once makes it difficult to verify correctness and track down bugs. Documentation is here: http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Materials/Volume Currently this hooks into path tracing in 3 ways, which should get us pretty far until we add more advanced light sampling. These 3 hooks are repeated in the path tracing, branched path tracing and transparent shadow code: * Determine active volume shader at start of the path * Change active volume shader on transmission through a surface * Light attenuation over line segments between camera, surfaces and background This is work by "storm", Stuart Broadfoot, Thomas Dinges and myself.
2013-12-28 15:56:19 +00:00
#ifdef __VOLUME__
Code refactor: better distinguish scatter and absorption for volume integration.
2013-12-29 14:49:16 +00:00
/* volume attenuation, emission, scatter */
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).
2015-11-22 10:48:33 +00:00
if(state->volume_stack[0].shader != SHADER_NONE) {
Ray volume_ray = *ray;
Cycles Volume Render: heterogeneous (textured) volumes support. Volumes can now have textured colors and density. There is a Volume Sampling panel in the Render properties with these settings: * Step size: distance between volume shader samples when rendering the volume. Lower values give more accurate and detailed results but also increased render time. * Max steps: maximum number of steps through the volume before giving up, to protect from extremely long render times with big objects or small step sizes. This is much more compute intensive than homogeneous volume, so when you are not using a texture you should enable the Homogeneous Volume option in the material or world for faster rendering. One important missing feature is that Generated texture coordinates are not yet working in volumes, and they are the default coordinates for nearly all texture nodes. So until that works you need to plug in object texture coordinates or a world space position. This is work by "storm", Stuart Broadfoot, Thomas Dinges and myself.
2013-12-29 21:19:38 +00:00
volume_ray.t = (hit)? isect.t: FLT_MAX;
Cycles Volume Render: scattering support. This is done by adding a Volume Scatter node. In many cases you will want to add together a Volume Absorption and Volume Scatter node with the same color and density to get the expected results. This should work with branched path tracing, mixing closures, overlapping volumes, etc. However there's still various optimizations needed for sampling. The main missing thing from the volume branch is the equiangular sampling for homogeneous volumes. The heterogeneous scattering code was arranged such that we can use a single stratified random number for distance sampling, which gives less noise than pseudo random numbers for each step. For volumes where the color is textured there still seems to be something off, needs to be investigated.
2013-12-29 14:40:43 +00:00
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).
2015-11-22 10:48:33 +00:00
bool heterogeneous =
volume_stack_is_heterogeneous(kg,
state->volume_stack);
Code refactor: Split __VOLUME__ defines in Cycles. * __VOLUME__ is basic volume support with Emission and Absorption. * __VOLUME_SCATTER__ enables volume Scattering support. * __VOLUME_DECOUPLED__ enables Decoupled Ray Marching.
2014-08-20 21:15:30 +00:00
Cycles: Cleanup, indent nested preprocessor directives Quite straightforward, main trick is happening in path_source_replace_includes(). Reviewers: brecht, dingto, lukasstockner97, juicyfruit Differential Revision: https://developer.blender.org/D1794
2016-02-12 17:33:43 +00:00
# ifdef __VOLUME_DECOUPLED__
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).
2015-11-22 10:48:33 +00:00
int sampling_method =
volume_stack_sampling_method(kg,
state->volume_stack);
Cycles: volume sampling method can now be set per material/world. This gives you "Multiple Importance", "Distance" and "Equiangular" choices. What multiple importance sampling does is make things more robust to certain types of noise at the cost of a bit more noise in cases where the individual strategies are always better. So if you've got a pretty dense volume that's lit from far away then distance sampling is usually more efficient. If you've got a light inside or near the volume then equiangular sampling is better. If you have a combination of both, then the multiple importance sampling will be better.
2014-06-07 16:47:14 +00:00
bool decoupled = kernel_volume_use_decoupled(kg, heterogeneous, false, sampling_method);
Cycles Volume Render: scattering support. This is done by adding a Volume Scatter node. In many cases you will want to add together a Volume Absorption and Volume Scatter node with the same color and density to get the expected results. This should work with branched path tracing, mixing closures, overlapping volumes, etc. However there's still various optimizations needed for sampling. The main missing thing from the volume branch is the equiangular sampling for homogeneous volumes. The heterogeneous scattering code was arranged such that we can use a single stratified random number for distance sampling, which gives less noise than pseudo random numbers for each step. For volumes where the color is textured there still seems to be something off, needs to be investigated.
2013-12-29 14:40:43 +00:00
Cycles: volume light sampling * Volume multiple importace sampling support to combine equiangular and distance sampling, for both homogeneous and heterogeneous volumes. * Branched path "Sample All Direct Lights" and "Sample All Indirect Lights" now apply to volumes as well as surfaces. Implementation note: For simplicity this is all done with decoupled ray marching, the only case we do not use decoupled is for distance only sampling with one light sample. The homogeneous case should still compile on the GPU because it only requires fixed size storage, but the heterogeneous case will be trickier to get working.
2014-04-04 14:45:49 +00:00
if(decoupled) {
/* cache steps along volume for repeated sampling */
VolumeSegment volume_segment;
Cycles code refactor: deduplicate and symmetrize some path tracing code.
2014-04-04 12:21:35 +00:00
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).
2015-11-22 10:48:33 +00:00
shader_setup_from_volume(kg,
Cycles CUDA: reduce branched path stack memory by sharing indirect ShaderData. Saves about 15% for the branched path kernel.
2016-05-24 20:28:03 +00:00
sd,
Cycles: Refactor how we pass bounce info to light path node. This commit changes the way how we pass bounce information to the Light Path node. Instead of manualy copying the bounces into ShaderData, we now directly pass PathState. This reduces the arguments that we need to pass around and also makes it easier to extend the feature. This commit also exposes the Transmission Bounce Depth to the Light Path node. It works similar to the Transparent Depth Output: Replace a Transmission lightpath after X bounces with another shader, e.g a Diffuse one. This can be used to avoid black surfaces, due to low amount of max bounces. Reviewed by Sergey and Brecht, thanks for some hlp with this. I tested compilation and usage on CPU (SVM and OSL), CUDA, OpenCL Split and Mega kernel. Hopefully this covers all devices. :)
2016-01-06 22:38:13 +00:00
&volume_ray);
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).
2015-11-22 10:48:33 +00:00
kernel_volume_decoupled_record(kg,
state,
&volume_ray,
Cycles CUDA: reduce branched path stack memory by sharing indirect ShaderData. Saves about 15% for the branched path kernel.
2016-05-24 20:28:03 +00:00
sd,
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).
2015-11-22 10:48:33 +00:00
&volume_segment,
heterogeneous);
Cycles: volume sampling method can now be set per material/world. This gives you "Multiple Importance", "Distance" and "Equiangular" choices. What multiple importance sampling does is make things more robust to certain types of noise at the cost of a bit more noise in cases where the individual strategies are always better. So if you've got a pretty dense volume that's lit from far away then distance sampling is usually more efficient. If you've got a light inside or near the volume then equiangular sampling is better. If you have a combination of both, then the multiple importance sampling will be better.
2014-06-07 16:47:14 +00:00
volume_segment.sampling_method = sampling_method;
Cycles: volume light sampling * Volume multiple importace sampling support to combine equiangular and distance sampling, for both homogeneous and heterogeneous volumes. * Branched path "Sample All Direct Lights" and "Sample All Indirect Lights" now apply to volumes as well as surfaces. Implementation note: For simplicity this is all done with decoupled ray marching, the only case we do not use decoupled is for distance only sampling with one light sample. The homogeneous case should still compile on the GPU because it only requires fixed size storage, but the heterogeneous case will be trickier to get working.
2014-04-04 14:45:49 +00:00
/* emission */
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).
2015-11-22 10:48:33 +00:00
if(volume_segment.closure_flag & SD_EMISSION) {
path_radiance_accum_emission(L,
throughput,
volume_segment.accum_emission,
state->bounce);
}
Cycles: volume light sampling * Volume multiple importace sampling support to combine equiangular and distance sampling, for both homogeneous and heterogeneous volumes. * Branched path "Sample All Direct Lights" and "Sample All Indirect Lights" now apply to volumes as well as surfaces. Implementation note: For simplicity this is all done with decoupled ray marching, the only case we do not use decoupled is for distance only sampling with one light sample. The homogeneous case should still compile on the GPU because it only requires fixed size storage, but the heterogeneous case will be trickier to get working.
2014-04-04 14:45:49 +00:00
/* scattering */
VolumeIntegrateResult result = VOLUME_PATH_ATTENUATED;
if(volume_segment.closure_flag & SD_SCATTER) {
Workaround for T47213: branched path sampling issues with CUDA 7.5.
2016-02-18 23:39:00 +00:00
int all = kernel_data.integrator.sample_all_lights_indirect;
Cycles: volume light sampling * Volume multiple importace sampling support to combine equiangular and distance sampling, for both homogeneous and heterogeneous volumes. * Branched path "Sample All Direct Lights" and "Sample All Indirect Lights" now apply to volumes as well as surfaces. Implementation note: For simplicity this is all done with decoupled ray marching, the only case we do not use decoupled is for distance only sampling with one light sample. The homogeneous case should still compile on the GPU because it only requires fixed size storage, but the heterogeneous case will be trickier to get working.
2014-04-04 14:45:49 +00:00
/* direct light sampling */
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).
2015-11-22 10:48:33 +00:00
kernel_branched_path_volume_connect_light(kg,
rng,
Cycles CUDA: reduce branched path stack memory by sharing indirect ShaderData. Saves about 15% for the branched path kernel.
2016-05-24 20:28:03 +00:00
sd,
Cycles CUDA: reduce stack memory by reusing ShaderData. 57% less for path and 48% less for branched path.
2016-05-22 20:35:47 +00:00
emission_sd,
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).
2015-11-22 10:48:33 +00:00
throughput,
state,
L,
all,
&volume_ray,
&volume_segment);
Cycles: volume light sampling * Volume multiple importace sampling support to combine equiangular and distance sampling, for both homogeneous and heterogeneous volumes. * Branched path "Sample All Direct Lights" and "Sample All Indirect Lights" now apply to volumes as well as surfaces. Implementation note: For simplicity this is all done with decoupled ray marching, the only case we do not use decoupled is for distance only sampling with one light sample. The homogeneous case should still compile on the GPU because it only requires fixed size storage, but the heterogeneous case will be trickier to get working.
2014-04-04 14:45:49 +00:00
/* indirect 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 */
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).
2015-11-22 10:48:33 +00:00
float rphase = path_state_rng_1D_for_decision(kg, rng, state, PRNG_PHASE);
float rscatter = path_state_rng_1D_for_decision(kg, rng, state, PRNG_SCATTER_DISTANCE);
Cycles: volume light sampling * Volume multiple importace sampling support to combine equiangular and distance sampling, for both homogeneous and heterogeneous volumes. * Branched path "Sample All Direct Lights" and "Sample All Indirect Lights" now apply to volumes as well as surfaces. Implementation note: For simplicity this is all done with decoupled ray marching, the only case we do not use decoupled is for distance only sampling with one light sample. The homogeneous case should still compile on the GPU because it only requires fixed size storage, but the heterogeneous case will be trickier to get working.
2014-04-04 14:45:49 +00:00
result = kernel_volume_decoupled_scatter(kg,
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).
2015-11-22 10:48:33 +00:00
state,
&volume_ray,
Cycles CUDA: reduce branched path stack memory by sharing indirect ShaderData. Saves about 15% for the branched path kernel.
2016-05-24 20:28:03 +00:00
sd,
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).
2015-11-22 10:48:33 +00:00
&throughput,
rphase,
rscatter,
&volume_segment,
NULL,
true);
Cycles: volume light sampling * Volume multiple importace sampling support to combine equiangular and distance sampling, for both homogeneous and heterogeneous volumes. * Branched path "Sample All Direct Lights" and "Sample All Indirect Lights" now apply to volumes as well as surfaces. Implementation note: For simplicity this is all done with decoupled ray marching, the only case we do not use decoupled is for distance only sampling with one light sample. The homogeneous case should still compile on the GPU because it only requires fixed size storage, but the heterogeneous case will be trickier to get working.
2014-04-04 14:45:49 +00:00
}
/* free cached steps */
kernel_volume_decoupled_free(kg, &volume_segment);
if(result == VOLUME_PATH_SCATTERED) {
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).
2015-11-22 10:48:33 +00:00
if(kernel_path_volume_bounce(kg,
rng,
Cycles CUDA: reduce branched path stack memory by sharing indirect ShaderData. Saves about 15% for the branched path kernel.
2016-05-24 20:28:03 +00:00
sd,
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).
2015-11-22 10:48:33 +00:00
&throughput,
state,
L,
ray))
{
Cycles: volume light sampling * Volume multiple importace sampling support to combine equiangular and distance sampling, for both homogeneous and heterogeneous volumes. * Branched path "Sample All Direct Lights" and "Sample All Indirect Lights" now apply to volumes as well as surfaces. Implementation note: For simplicity this is all done with decoupled ray marching, the only case we do not use decoupled is for distance only sampling with one light sample. The homogeneous case should still compile on the GPU because it only requires fixed size storage, but the heterogeneous case will be trickier to get working.
2014-04-04 14:45:49 +00:00
continue;
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).
2015-11-22 10:48:33 +00:00
}
else {
Cycles: volume light sampling * Volume multiple importace sampling support to combine equiangular and distance sampling, for both homogeneous and heterogeneous volumes. * Branched path "Sample All Direct Lights" and "Sample All Indirect Lights" now apply to volumes as well as surfaces. Implementation note: For simplicity this is all done with decoupled ray marching, the only case we do not use decoupled is for distance only sampling with one light sample. The homogeneous case should still compile on the GPU because it only requires fixed size storage, but the heterogeneous case will be trickier to get working.
2014-04-04 14:45:49 +00:00
break;
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).
2015-11-22 10:48:33 +00:00
}
Cycles: volume light sampling * Volume multiple importace sampling support to combine equiangular and distance sampling, for both homogeneous and heterogeneous volumes. * Branched path "Sample All Direct Lights" and "Sample All Indirect Lights" now apply to volumes as well as surfaces. Implementation note: For simplicity this is all done with decoupled ray marching, the only case we do not use decoupled is for distance only sampling with one light sample. The homogeneous case should still compile on the GPU because it only requires fixed size storage, but the heterogeneous case will be trickier to get working.
2014-04-04 14:45:49 +00:00
}
Cycles: Small improvement for volume render (decoupled) Simplify branching here a bit, helps ~3% in volume_light_sampling.blend (Branched MIS scene).
2015-02-14 19:44:30 +00:00
else {
throughput *= volume_segment.accum_transmittance;
}
Cycles: volume light sampling * Volume multiple importace sampling support to combine equiangular and distance sampling, for both homogeneous and heterogeneous volumes. * Branched path "Sample All Direct Lights" and "Sample All Indirect Lights" now apply to volumes as well as surfaces. Implementation note: For simplicity this is all done with decoupled ray marching, the only case we do not use decoupled is for distance only sampling with one light sample. The homogeneous case should still compile on the GPU because it only requires fixed size storage, but the heterogeneous case will be trickier to get working.
2014-04-04 14:45:49 +00:00
}
Code refactor: Split __VOLUME__ defines in Cycles. * __VOLUME__ is basic volume support with Emission and Absorption. * __VOLUME_SCATTER__ enables volume Scattering support. * __VOLUME_DECOUPLED__ enables Decoupled Ray Marching.
2014-08-20 21:15:30 +00:00
else
Cycles: Cleanup, indent nested preprocessor directives Quite straightforward, main trick is happening in path_source_replace_includes(). Reviewers: brecht, dingto, lukasstockner97, juicyfruit Differential Revision: https://developer.blender.org/D1794
2016-02-12 17:33:43 +00:00
# endif
Code refactor: Split __VOLUME__ defines in Cycles. * __VOLUME__ is basic volume support with Emission and Absorption. * __VOLUME_SCATTER__ enables volume Scattering support. * __VOLUME_DECOUPLED__ enables Decoupled Ray Marching.
2014-08-20 21:15:30 +00:00
{
Cycles: volume light sampling * Volume multiple importace sampling support to combine equiangular and distance sampling, for both homogeneous and heterogeneous volumes. * Branched path "Sample All Direct Lights" and "Sample All Indirect Lights" now apply to volumes as well as surfaces. Implementation note: For simplicity this is all done with decoupled ray marching, the only case we do not use decoupled is for distance only sampling with one light sample. The homogeneous case should still compile on the GPU because it only requires fixed size storage, but the heterogeneous case will be trickier to get working.
2014-04-04 14:45:49 +00:00
/* integrate along volume segment with distance sampling */
VolumeIntegrateResult result = kernel_volume_integrate(
Cycles CUDA: reduce branched path stack memory by sharing indirect ShaderData. Saves about 15% for the branched path kernel.
2016-05-24 20:28:03 +00:00
kg, state, sd, &volume_ray, L, &throughput, rng, heterogeneous);
Cycles: volume light sampling * Volume multiple importace sampling support to combine equiangular and distance sampling, for both homogeneous and heterogeneous volumes. * Branched path "Sample All Direct Lights" and "Sample All Indirect Lights" now apply to volumes as well as surfaces. Implementation note: For simplicity this is all done with decoupled ray marching, the only case we do not use decoupled is for distance only sampling with one light sample. The homogeneous case should still compile on the GPU because it only requires fixed size storage, but the heterogeneous case will be trickier to get working.
2014-04-04 14:45:49 +00:00
Cycles: Cleanup, indent nested preprocessor directives Quite straightforward, main trick is happening in path_source_replace_includes(). Reviewers: brecht, dingto, lukasstockner97, juicyfruit Differential Revision: https://developer.blender.org/D1794
2016-02-12 17:33:43 +00:00
# ifdef __VOLUME_SCATTER__
Cycles: volume light sampling * Volume multiple importace sampling support to combine equiangular and distance sampling, for both homogeneous and heterogeneous volumes. * Branched path "Sample All Direct Lights" and "Sample All Indirect Lights" now apply to volumes as well as surfaces. Implementation note: For simplicity this is all done with decoupled ray marching, the only case we do not use decoupled is for distance only sampling with one light sample. The homogeneous case should still compile on the GPU because it only requires fixed size storage, but the heterogeneous case will be trickier to get working.
2014-04-04 14:45:49 +00:00
if(result == VOLUME_PATH_SCATTERED) {
/* direct lighting */
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).
2015-11-22 10:48:33 +00:00
kernel_path_volume_connect_light(kg,
rng,
Cycles CUDA: reduce branched path stack memory by sharing indirect ShaderData. Saves about 15% for the branched path kernel.
2016-05-24 20:28:03 +00:00
sd,
Cycles CUDA: reduce stack memory by reusing ShaderData. 57% less for path and 48% less for branched path.
2016-05-22 20:35:47 +00:00
emission_sd,
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).
2015-11-22 10:48:33 +00:00
throughput,
state,
L);
Cycles: volume light sampling * Volume multiple importace sampling support to combine equiangular and distance sampling, for both homogeneous and heterogeneous volumes. * Branched path "Sample All Direct Lights" and "Sample All Indirect Lights" now apply to volumes as well as surfaces. Implementation note: For simplicity this is all done with decoupled ray marching, the only case we do not use decoupled is for distance only sampling with one light sample. The homogeneous case should still compile on the GPU because it only requires fixed size storage, but the heterogeneous case will be trickier to get working.
2014-04-04 14:45:49 +00:00
/* indirect light bounce */
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).
2015-11-22 10:48:33 +00:00
if(kernel_path_volume_bounce(kg,
rng,
Cycles CUDA: reduce branched path stack memory by sharing indirect ShaderData. Saves about 15% for the branched path kernel.
2016-05-24 20:28:03 +00:00
sd,
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).
2015-11-22 10:48:33 +00:00
&throughput,
state,
L,
ray))
{
Cycles: volume light sampling * Volume multiple importace sampling support to combine equiangular and distance sampling, for both homogeneous and heterogeneous volumes. * Branched path "Sample All Direct Lights" and "Sample All Indirect Lights" now apply to volumes as well as surfaces. Implementation note: For simplicity this is all done with decoupled ray marching, the only case we do not use decoupled is for distance only sampling with one light sample. The homogeneous case should still compile on the GPU because it only requires fixed size storage, but the heterogeneous case will be trickier to get working.
2014-04-04 14:45:49 +00:00
continue;
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).
2015-11-22 10:48:33 +00:00
}
else {
Cycles: volume light sampling * Volume multiple importace sampling support to combine equiangular and distance sampling, for both homogeneous and heterogeneous volumes. * Branched path "Sample All Direct Lights" and "Sample All Indirect Lights" now apply to volumes as well as surfaces. Implementation note: For simplicity this is all done with decoupled ray marching, the only case we do not use decoupled is for distance only sampling with one light sample. The homogeneous case should still compile on the GPU because it only requires fixed size storage, but the heterogeneous case will be trickier to get working.
2014-04-04 14:45:49 +00:00
break;
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).
2015-11-22 10:48:33 +00:00
}
Cycles: volume light sampling * Volume multiple importace sampling support to combine equiangular and distance sampling, for both homogeneous and heterogeneous volumes. * Branched path "Sample All Direct Lights" and "Sample All Indirect Lights" now apply to volumes as well as surfaces. Implementation note: For simplicity this is all done with decoupled ray marching, the only case we do not use decoupled is for distance only sampling with one light sample. The homogeneous case should still compile on the GPU because it only requires fixed size storage, but the heterogeneous case will be trickier to get working.
2014-04-04 14:45:49 +00:00
}
Cycles: Cleanup, indent nested preprocessor directives Quite straightforward, main trick is happening in path_source_replace_includes(). Reviewers: brecht, dingto, lukasstockner97, juicyfruit Differential Revision: https://developer.blender.org/D1794
2016-02-12 17:33:43 +00:00
# endif
Cycles Volume Render: scattering support. This is done by adding a Volume Scatter node. In many cases you will want to add together a Volume Absorption and Volume Scatter node with the same color and density to get the expected results. This should work with branched path tracing, mixing closures, overlapping volumes, etc. However there's still various optimizations needed for sampling. The main missing thing from the volume branch is the equiangular sampling for homogeneous volumes. The heterogeneous scattering code was arranged such that we can use a single stratified random number for distance sampling, which gives less noise than pseudo random numbers for each step. For volumes where the color is textured there still seems to be something off, needs to be investigated.
2013-12-29 14:40:43 +00:00
}
Cycles Volume Render: support for rendering of homogeneous volume with absorption. This is the simplest possible volume rendering case, constant density inside the volume and no scattering or emission. My plan is to tweak, verify and commit more volume rendering effects one by one, doing it all at once makes it difficult to verify correctness and track down bugs. Documentation is here: http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Materials/Volume Currently this hooks into path tracing in 3 ways, which should get us pretty far until we add more advanced light sampling. These 3 hooks are repeated in the path tracing, branched path tracing and transparent shadow code: * Determine active volume shader at start of the path * Change active volume shader on transmission through a surface * Light attenuation over line segments between camera, surfaces and background This is work by "storm", Stuart Broadfoot, Thomas Dinges and myself.
2013-12-28 15:56:19 +00:00
}
#endif
Cycles: multiple importance sampling for lamps, which helps reduce noise for big lamps and sharp glossy reflections. This was already supported for mesh lights and the background, so lamps should do it too. This is not for free and it's a bit slower than I hoped even though there is no extra BVH ray intersection. I'll try to optimize it more later. * Area lights look a bit different now, they had the wrong shape before. * Also fixes a sampling issue in the non-progressive integrator. * Only enabled for the CPU, will test on the GPU later. * An option to disable this will be added for situations where it does not help. Same time comparison before/after: http://www.pasteall.org/pic/show.php?id=43313 http://www.pasteall.org/pic/show.php?id=43314
2013-01-09 21:09:20 +00:00
if(!hit) {
Cycles: fixes for OpenCL build after pass changes, patch by Daniel Genrich.
2012-01-26 15:37:33 +00:00
#ifdef __BACKGROUND__
Cycles: merging features from tomato branch. === BVH build time optimizations === * BVH building was multithreaded. Not all building is multithreaded, packing and the initial bounding/splitting is still single threaded, but recursive splitting is, which was the main bottleneck. * Object splitting now uses binning rather than sorting of all elements, using code from the Embree raytracer from Intel. http://software.intel.com/en-us/articles/embree-photo-realistic-ray-tracing-kernels/ * Other small changes to avoid allocations, pack memory more tightly, avoid some unnecessary operations, ... These optimizations do not work yet when Spatial Splits are enabled, for that more work is needed. There's also other optimizations still needed, in particular for the case of many low poly objects, the packing step and node memory allocation. BVH raytracing time should remain about the same, but BVH build time should be significantly reduced, test here show speedup of about 5x to 10x on a dual core and 5x to 25x on an 8-core machine, depending on the scene. === Threads === Centralized task scheduler for multithreading, which is basically the CPU device threading code wrapped into something reusable. Basic idea is that there is a single TaskScheduler that keeps a pool of threads, one for each core. Other places in the code can then create a TaskPool that they can drop Tasks in to be executed by the scheduler, and wait for them to complete or cancel them early. === Normal ==== Added a Normal output to the texture coordinate node. This currently gives the object space normal, which is the same under object animation. In the future this might become a "generated" normal so it's also stable for deforming objects, but for now it's already useful for non-deforming objects. === Render Layers === Per render layer Samples control, leaving it to 0 will use the common scene setting. Environment pass will now render environment even if film is set to transparent. Exclude Layers" added. Scene layers (all object that influence the render, directly or indirectly) are shared between all render layers. However sometimes it's useful to leave out some object influence for a particular render layer. That's what this option allows you to do. === Filter Glossy === When using a value higher than 0.0, this will blur glossy reflections after blurry bounces, to reduce noise at the cost of accuracy. 1.0 is a good starting value to tweak. Some light paths have a low probability of being found while contributing much light to the pixel. As a result these light paths will be found in some pixels and not in others, causing fireflies. An example of such a difficult path might be a small light that is causing a small specular highlight on a sharp glossy material, which we are seeing through a rough glossy material. With path tracing it is difficult to find the specular highlight, but if we increase the roughness on the material the highlight gets bigger and softer, and so easier to find. Often this blurring will be hardly noticeable, because we are seeing it through a blurry material anyway, but there are also cases where this will lead to a loss of detail in lighting.
2012-04-28 08:53:59 +00:00
/* sample background shader */
Cycles CUDA: reduce stack memory by reusing ShaderData. 57% less for path and 48% less for branched path.
2016-05-22 20:35:47 +00:00
float3 L_background = indirect_background(kg, emission_sd, state, ray);
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).
2015-11-22 10:48:33 +00:00
path_radiance_accum_background(L,
throughput,
L_background,
state->bounce);
Cycles: fixes for OpenCL build after pass changes, patch by Daniel Genrich.
2012-01-26 15:37:33 +00:00
#endif
Cycles: * Add alpha pass output, to use set Transparent option in Film panel. * Add Holdout closure (OSL terminology), this is like the Sky option in the internal renderer, objects with this closure show the background / zero alpha. * Add option to use Gaussian instead of Box pixel filter in the UI. * Remove camera response curves for now, they don't really belong here in the pipeline, should be moved to compositor. * Output full float values for rendering now, previously was only byte precision. * Add a patch from Thomas to get a preview passes option, but still disabled because it isn't quite working right yet. * CUDA: don't compile shader graph evaluation inline. * Convert tabs to spaces in python files.
2011-08-28 13:55:59 +00:00
Cycles render engine, initial commit. This is the engine itself, blender modifications and build instructions will follow later. Cycles uses code from some great open source projects, many thanks them: * BVH building and traversal code from NVidia's "Understanding the Efficiency of Ray Traversal on GPUs": http://code.google.com/p/understanding-the-efficiency-of-ray-traversal-on-gpus/ * Open Shading Language for a large part of the shading system: http://code.google.com/p/openshadinglanguage/ * Blender for procedural textures and a few other nodes. * Approximate Catmull Clark subdivision from NVidia Mesh tools: http://code.google.com/p/nvidia-mesh-tools/ * Sobol direction vectors from: http://web.maths.unsw.edu.au/~fkuo/sobol/ * Film response functions from: http://www.cs.columbia.edu/CAVE/software/softlib/dorf.php
2011-04-27 11:58:34 +00:00
break;
}
/* setup shading */
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).
2015-11-22 10:48:33 +00:00
shader_setup_from_ray(kg,
Cycles CUDA: reduce branched path stack memory by sharing indirect ShaderData. Saves about 15% for the branched path kernel.
2016-05-24 20:28:03 +00:00
sd,
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).
2015-11-22 10:48:33 +00:00
&isect,
Cycles: Refactor how we pass bounce info to light path node. This commit changes the way how we pass bounce information to the Light Path node. Instead of manualy copying the bounces into ShaderData, we now directly pass PathState. This reduces the arguments that we need to pass around and also makes it easier to extend the feature. This commit also exposes the Transmission Bounce Depth to the Light Path node. It works similar to the Transparent Depth Output: Replace a Transmission lightpath after X bounces with another shader, e.g a Diffuse one. This can be used to avoid black surfaces, due to low amount of max bounces. Reviewed by Sergey and Brecht, thanks for some hlp with this. I tested compilation and usage on CPU (SVM and OSL), CUDA, OpenCL Split and Mega kernel. Hopefully this covers all devices. :)
2016-01-06 22:38:13 +00:00
ray);
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).
2015-11-22 10:48:33 +00:00
float rbsdf = path_state_rng_1D_for_decision(kg, rng, state, PRNG_BSDF);
Cycles: Add multi-scattering, energy-conserving GGX as an option to the Glossy, Anisotropic and Glass BSDFs This commit adds a new distribution to the Glossy, Anisotropic and Glass BSDFs that implements the multiple-scattering microfacet model described in the paper "Multiple-Scattering Microfacet BSDFs with the Smith Model". Essentially, the improvement is that unlike classical GGX, which only models single scattering and assumes the contribution of multiple bounces to be zero, this new model performs a random walk on the microsurface until the ray leaves it again, which ensures perfect energy conservation. In practise, this means that the "darkening problem" - GGX materials becoming darker with increasing roughness - is solved in a physically correct and efficient way. The downside of this model is that it has no (known) analytic expression for evalation. However, it can be evaluated stochastically, and although the correct PDF isn't known either, the properties of MIS and the balance heuristic guarantee an unbiased result at the cost of slightly higher noise. Reviewers: dingto, #cycles, brecht Reviewed By: dingto, #cycles, brecht Subscribers: bliblubli, ace_dragon, gregzaal, brecht, harvester, dingto, marcog, swerner, jtheninja, Blendify, nutel Differential Revision: https://developer.blender.org/D2002
2016-06-23 20:56:43 +00:00
shader_eval_surface(kg, sd, rng, state, rbsdf, state->flag, SHADER_CONTEXT_INDIRECT);
Cycles: * Fix some compile errors, when building without Branched Path.
2013-09-08 17:20:47 +00:00
#ifdef __BRANCHED_PATH__
Cycles CUDA: reduce branched path stack memory by sharing indirect ShaderData. Saves about 15% for the branched path kernel.
2016-05-24 20:28:03 +00:00
shader_merge_closures(sd);
Cycles: * Fix some compile errors, when building without Branched Path.
2013-09-08 17:20:47 +00:00
#endif
Fix #35614: cycles objects on mask layer in render layers were writing passes such as motion/uv/normal but they should have been excluded from them.
2013-06-04 15:41:45 +00:00
/* blurring of bsdf after bounces, for rays that have a small likelihood
* of following this particular path (diffuse, rough glossy) */
if(kernel_data.integrator.filter_glossy != FLT_MAX) {
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).
2015-11-22 10:48:33 +00:00
float blur_pdf = kernel_data.integrator.filter_glossy*state->min_ray_pdf;
Fix #35614: cycles objects on mask layer in render layers were writing passes such as motion/uv/normal but they should have been excluded from them.
2013-06-04 15:41:45 +00:00
if(blur_pdf < 1.0f) {
float blur_roughness = sqrtf(1.0f - blur_pdf)*0.5f;
Cycles CUDA: reduce branched path stack memory by sharing indirect ShaderData. Saves about 15% for the branched path kernel.
2016-05-24 20:28:03 +00:00
shader_bsdf_blur(kg, sd, blur_roughness);
Fix #35614: cycles objects on mask layer in render layers were writing passes such as motion/uv/normal but they should have been excluded from them.
2013-06-04 15:41:45 +00:00
}
}
Cycles render engine, initial commit. This is the engine itself, blender modifications and build instructions will follow later. Cycles uses code from some great open source projects, many thanks them: * BVH building and traversal code from NVidia's "Understanding the Efficiency of Ray Traversal on GPUs": http://code.google.com/p/understanding-the-efficiency-of-ray-traversal-on-gpus/ * Open Shading Language for a large part of the shading system: http://code.google.com/p/openshadinglanguage/ * Blender for procedural textures and a few other nodes. * Approximate Catmull Clark subdivision from NVidia Mesh tools: http://code.google.com/p/nvidia-mesh-tools/ * Sobol direction vectors from: http://web.maths.unsw.edu.au/~fkuo/sobol/ * Film response functions from: http://www.cs.columbia.edu/CAVE/software/softlib/dorf.php
2011-04-27 11:58:34 +00:00
#ifdef __EMISSION__
/* emission */
Cycles CUDA: reduce branched path stack memory by sharing indirect ShaderData. Saves about 15% for the branched path kernel.
2016-05-24 20:28:03 +00:00
if(sd->flag & SD_EMISSION) {
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).
2015-11-22 10:48:33 +00:00
float3 emission = indirect_primitive_emission(kg,
Cycles CUDA: reduce branched path stack memory by sharing indirect ShaderData. Saves about 15% for the branched path kernel.
2016-05-24 20:28:03 +00:00
sd,
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).
2015-11-22 10:48:33 +00:00
isect.t,
state->flag,
state->ray_pdf);
path_radiance_accum_emission(L, throughput, emission, state->bounce);
Cycles: Render Passes Currently supported passes: * Combined, Z, Normal, Object Index, Material Index, Emission, Environment, Diffuse/Glossy/Transmission x Direct/Indirect/Color Not supported yet: * UV, Vector, Mist Only enabled for CPU devices at the moment, will do GPU tweaks tommorrow, also for environment importance sampling. Documentation: http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Passes
2012-01-25 17:23:52 +00:00
}
Cycles: * Add max diffuse/glossy/transmission bounces * Add separate min/max for transparent depth * Updated/added some presets that use these options * Add ray visibility options for objects, to hide them from camera/diffuse/glossy/transmission/shadow rays * Is singular ray output for light path node Details here: http://wiki.blender.org/index.php/Dev:2.5/Source/Render/Cycles/LightPaths
2011-09-01 15:53:36 +00:00
#endif
/* path termination. this is a strange place to put the termination, it's
style cleanup: block comments
2012-06-09 17:22:52 +00:00
* 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 */
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).
2015-11-22 10:48:33 +00:00
float probability =
path_state_terminate_probability(kg,
state,
throughput*num_samples);
Cycles render engine, initial commit. This is the engine itself, blender modifications and build instructions will follow later. Cycles uses code from some great open source projects, many thanks them: * BVH building and traversal code from NVidia's "Understanding the Efficiency of Ray Traversal on GPUs": http://code.google.com/p/understanding-the-efficiency-of-ray-traversal-on-gpus/ * Open Shading Language for a large part of the shading system: http://code.google.com/p/openshadinglanguage/ * Blender for procedural textures and a few other nodes. * Approximate Catmull Clark subdivision from NVidia Mesh tools: http://code.google.com/p/nvidia-mesh-tools/ * Sobol direction vectors from: http://web.maths.unsw.edu.au/~fkuo/sobol/ * Film response functions from: http://www.cs.columbia.edu/CAVE/software/softlib/dorf.php
2011-04-27 11:58:34 +00:00
Cycles: experimental correlated multi-jittered sampling pattern that can be used instead of sobol. So far one doesn't seem to be consistently better or worse than the other for the same number of samples but more testing is needed. The random number generator itself is slower than sobol for most number of samples, except 16, 64, 256, .. because they can be computed faster. This can probably be optimized, but we can do that when/if this actually turns out to be useful. Paper this implementation is based on: http://graphics.pixar.com/library/MultiJitteredSampling/ Also includes some refactoring of RNG code, fixing a Sobol correlation issue with the first BSDF and < 16 samples, skipping some unneeded RNG calls and using a simpler unit square to unit disk function.
2013-06-07 16:06:22 +00:00
if(probability == 0.0f) {
Cycles: * Add max diffuse/glossy/transmission bounces * Add separate min/max for transparent depth * Updated/added some presets that use these options * Add ray visibility options for objects, to hide them from camera/diffuse/glossy/transmission/shadow rays * Is singular ray output for light path node Details here: http://wiki.blender.org/index.php/Dev:2.5/Source/Render/Cycles/LightPaths
2011-09-01 15:53:36 +00:00
break;
Cycles: experimental correlated multi-jittered sampling pattern that can be used instead of sobol. So far one doesn't seem to be consistently better or worse than the other for the same number of samples but more testing is needed. The random number generator itself is slower than sobol for most number of samples, except 16, 64, 256, .. because they can be computed faster. This can probably be optimized, but we can do that when/if this actually turns out to be useful. Paper this implementation is based on: http://graphics.pixar.com/library/MultiJitteredSampling/ Also includes some refactoring of RNG code, fixing a Sobol correlation issue with the first BSDF and < 16 samples, skipping some unneeded RNG calls and using a simpler unit square to unit disk function.
2013-06-07 16:06:22 +00:00
}
else if(probability != 1.0f) {
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).
2015-11-22 10:48:33 +00:00
float terminate = path_state_rng_1D_for_decision(kg, rng, state, PRNG_TERMINATE);
Cycles: experimental correlated multi-jittered sampling pattern that can be used instead of sobol. So far one doesn't seem to be consistently better or worse than the other for the same number of samples but more testing is needed. The random number generator itself is slower than sobol for most number of samples, except 16, 64, 256, .. because they can be computed faster. This can probably be optimized, but we can do that when/if this actually turns out to be useful. Paper this implementation is based on: http://graphics.pixar.com/library/MultiJitteredSampling/ Also includes some refactoring of RNG code, fixing a Sobol correlation issue with the first BSDF and < 16 samples, skipping some unneeded RNG calls and using a simpler unit square to unit disk function.
2013-06-07 16:06:22 +00:00
if(terminate >= probability)
break;
Cycles: * Add max diffuse/glossy/transmission bounces * Add separate min/max for transparent depth * Updated/added some presets that use these options * Add ray visibility options for objects, to hide them from camera/diffuse/glossy/transmission/shadow rays * Is singular ray output for light path node Details here: http://wiki.blender.org/index.php/Dev:2.5/Source/Render/Cycles/LightPaths
2011-09-01 15:53:36 +00:00
Cycles: experimental correlated multi-jittered sampling pattern that can be used instead of sobol. So far one doesn't seem to be consistently better or worse than the other for the same number of samples but more testing is needed. The random number generator itself is slower than sobol for most number of samples, except 16, 64, 256, .. because they can be computed faster. This can probably be optimized, but we can do that when/if this actually turns out to be useful. Paper this implementation is based on: http://graphics.pixar.com/library/MultiJitteredSampling/ Also includes some refactoring of RNG code, fixing a Sobol correlation issue with the first BSDF and < 16 samples, skipping some unneeded RNG calls and using a simpler unit square to unit disk function.
2013-06-07 16:06:22 +00:00
throughput /= probability;
}
Cycles: * Add max diffuse/glossy/transmission bounces * Add separate min/max for transparent depth * Updated/added some presets that use these options * Add ray visibility options for objects, to hide them from camera/diffuse/glossy/transmission/shadow rays * Is singular ray output for light path node Details here: http://wiki.blender.org/index.php/Dev:2.5/Source/Render/Cycles/LightPaths
2011-09-01 15:53:36 +00:00
Cycles: ambient occlusion support, with AO factor and distance, and a render pass. http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/World#Ambient_Occlusion http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Passes#Lighting_Passes
2012-02-28 16:45:08 +00:00
#ifdef __AO__
/* ambient occlusion */
Cycles CUDA: reduce branched path stack memory by sharing indirect ShaderData. Saves about 15% for the branched path kernel.
2016-05-24 20:28:03 +00:00
if(kernel_data.integrator.use_ambient_occlusion || (sd->flag & SD_AO)) {
Fix #35665: more CUDA issues with recent kernel changes, tested on sm_20, sm_21 and sm_30 cards, so hopefully it should all work now. Also includes some warnings fixes related to nvcc compiler arguments, should make no difference otherwise.
2013-06-11 21:58:48 +00:00
float bsdf_u, bsdf_v;
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).
2015-11-22 10:48:33 +00:00
path_state_rng_2D(kg, rng, state, PRNG_BSDF_U, &bsdf_u, &bsdf_v);
Cycles: ambient occlusion support, with AO factor and distance, and a render pass. http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/World#Ambient_Occlusion http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Passes#Lighting_Passes
2012-02-28 16:45:08 +00:00
Cycles: ambient occlusion now takes per-BSDF normals into account.
2012-11-15 15:37:58 +00:00
float ao_factor = kernel_data.background.ao_factor;
float3 ao_N;
Cycles CUDA: reduce branched path stack memory by sharing indirect ShaderData. Saves about 15% for the branched path kernel.
2016-05-24 20:28:03 +00:00
float3 ao_bsdf = shader_bsdf_ao(kg, sd, ao_factor, &ao_N);
Cycles: ambient occlusion support, with AO factor and distance, and a render pass. http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/World#Ambient_Occlusion http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Passes#Lighting_Passes
2012-02-28 16:45:08 +00:00
float3 ao_D;
float ao_pdf;
Fix #36741: cycles AO pass giving values > 1.0 with transparency.
2013-09-17 13:22:42 +00:00
float3 ao_alpha = make_float3(0.0f, 0.0f, 0.0f);
Cycles: ambient occlusion support, with AO factor and distance, and a render pass. http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/World#Ambient_Occlusion http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Passes#Lighting_Passes
2012-02-28 16:45:08 +00:00
Cycles: ambient occlusion now takes per-BSDF normals into account.
2012-11-15 15:37:58 +00:00
sample_cos_hemisphere(ao_N, bsdf_u, bsdf_v, &ao_D, &ao_pdf);
Cycles: ambient occlusion support, with AO factor and distance, and a render pass. http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/World#Ambient_Occlusion http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Passes#Lighting_Passes
2012-02-28 16:45:08 +00:00
Cycles CUDA: reduce branched path stack memory by sharing indirect ShaderData. Saves about 15% for the branched path kernel.
2016-05-24 20:28:03 +00:00
if(dot(sd->Ng, ao_D) > 0.0f && ao_pdf != 0.0f) {
Cycles: ambient occlusion support, with AO factor and distance, and a render pass. http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/World#Ambient_Occlusion http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Passes#Lighting_Passes
2012-02-28 16:45:08 +00:00
Ray light_ray;
float3 ao_shadow;
Cycles CUDA: reduce branched path stack memory by sharing indirect ShaderData. Saves about 15% for the branched path kernel.
2016-05-24 20:28:03 +00:00
light_ray.P = ray_offset(sd->P, sd->Ng);
Cycles: ambient occlusion support, with AO factor and distance, and a render pass. http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/World#Ambient_Occlusion http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Passes#Lighting_Passes
2012-02-28 16:45:08 +00:00
light_ray.D = ao_D;
light_ray.t = kernel_data.background.ao_distance;
Cycles: Cleanup, indent nested preprocessor directives Quite straightforward, main trick is happening in path_source_replace_includes(). Reviewers: brecht, dingto, lukasstockner97, juicyfruit Differential Revision: https://developer.blender.org/D1794
2016-02-12 17:33:43 +00:00
# ifdef __OBJECT_MOTION__
Cycles CUDA: reduce branched path stack memory by sharing indirect ShaderData. Saves about 15% for the branched path kernel.
2016-05-24 20:28:03 +00:00
light_ray.time = sd->time;
Cycles: Cleanup, indent nested preprocessor directives Quite straightforward, main trick is happening in path_source_replace_includes(). Reviewers: brecht, dingto, lukasstockner97, juicyfruit Differential Revision: https://developer.blender.org/D1794
2016-02-12 17:33:43 +00:00
# endif
Cycles CUDA: reduce branched path stack memory by sharing indirect ShaderData. Saves about 15% for the branched path kernel.
2016-05-24 20:28:03 +00:00
light_ray.dP = sd->dP;
Fix #35207: addition to previous fix to avoid OSL getting uninitialized ray differentials for lighting, which could cause bad texture filtering artifacts or performance.
2013-05-03 21:34:51 +00:00
light_ray.dD = differential3_zero();
Cycles: ambient occlusion support, with AO factor and distance, and a render pass. http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/World#Ambient_Occlusion http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Passes#Lighting_Passes
2012-02-28 16:45:08 +00:00
Cycles CUDA: reduce stack memory by reusing ShaderData. 57% less for path and 48% less for branched path.
2016-05-22 20:35:47 +00:00
if(!shadow_blocked(kg, emission_sd, state, &light_ray, &ao_shadow)) {
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).
2015-11-22 10:48:33 +00:00
path_radiance_accum_ao(L,
throughput,
ao_alpha,
ao_bsdf,
ao_shadow,
state->bounce);
}
Cycles: reduce noise using regular path tracing + subsurface scattering with new cubic and gaussian falloff. Like the branched path tracer, this will now shade all intersection points instead of using one at random.
2013-08-24 15:02:08 +00:00
}
}
#endif
#ifdef __SUBSURFACE__
/* bssrdf scatter to a different location on the same object, replacing
* the closures with a diffuse BSDF */
Cycles CUDA: reduce branched path stack memory by sharing indirect ShaderData. Saves about 15% for the branched path kernel.
2016-05-24 20:28:03 +00:00
if(sd->flag & SD_BSSRDF) {
Cycles: reduce noise using regular path tracing + subsurface scattering with new cubic and gaussian falloff. Like the branched path tracer, this will now shade all intersection points instead of using one at random.
2013-08-24 15:02:08 +00:00
float bssrdf_probability;
Cycles CUDA: reduce branched path stack memory by sharing indirect ShaderData. Saves about 15% for the branched path kernel.
2016-05-24 20:28:03 +00:00
ShaderClosure *sc = subsurface_scatter_pick_closure(kg, sd, &bssrdf_probability);
Cycles: reduce noise using regular path tracing + subsurface scattering with new cubic and gaussian falloff. Like the branched path tracer, this will now shade all intersection points instead of using one at random.
2013-08-24 15:02:08 +00:00
/* modify throughput for picking bssrdf or bsdf */
throughput *= bssrdf_probability;
/* do bssrdf scatter step if we picked a bssrdf closure */
if(sc) {
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).
2015-11-22 10:48:33 +00:00
uint lcg_state = lcg_state_init(rng, state, 0x68bc21eb);
Cycles: reduce noise using regular path tracing + subsurface scattering with new cubic and gaussian falloff. Like the branched path tracer, this will now shade all intersection points instead of using one at random.
2013-08-24 15:02:08 +00:00
Cycles / SSS: * Remove the compatible falloff SSS implementation. We shouldn't support two implementations in the long term, and 2.7x is a good release number do break some compatibility as well. * Version patch added, so Files with Compatible falloff will automatically use Cubic now. It was already mentioned in the manual, that Compatible is deprecated. http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Nodes/Shaders#BSSRDF
2013-10-08 17:07:18 +00:00
float bssrdf_u, bssrdf_v;
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).
2015-11-22 10:48:33 +00:00
path_state_rng_2D(kg,
rng,
state,
PRNG_BSDF_U,
&bssrdf_u, &bssrdf_v);
subsurface_scatter_step(kg,
Cycles CUDA: reduce branched path stack memory by sharing indirect ShaderData. Saves about 15% for the branched path kernel.
2016-05-24 20:28:03 +00:00
sd,
Cycles: Refactor how we pass bounce info to light path node. This commit changes the way how we pass bounce information to the Light Path node. Instead of manualy copying the bounces into ShaderData, we now directly pass PathState. This reduces the arguments that we need to pass around and also makes it easier to extend the feature. This commit also exposes the Transmission Bounce Depth to the Light Path node. It works similar to the Transparent Depth Output: Replace a Transmission lightpath after X bounces with another shader, e.g a Diffuse one. This can be used to avoid black surfaces, due to low amount of max bounces. Reviewed by Sergey and Brecht, thanks for some hlp with this. I tested compilation and usage on CPU (SVM and OSL), CUDA, OpenCL Split and Mega kernel. Hopefully this covers all devices. :)
2016-01-06 22:38:13 +00:00
state,
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).
2015-11-22 10:48:33 +00:00
state->flag,
sc,
&lcg_state,
bssrdf_u, bssrdf_v,
false);
Cycles: ambient occlusion support, with AO factor and distance, and a render pass. http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/World#Ambient_Occlusion http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Passes#Lighting_Passes
2012-02-28 16:45:08 +00:00
}
}
#endif
Cycles Bake Expand Cycles to use the new baking API in Blender. It works on the selected object, and the panel can be accessed in the Render panel (similar to where it is for the Blender Internal). It bakes for the active texture of each material of the object. The active texture is currently defined as the active Image Texture node present in the material nodetree. If you don't want the baking to override an existent material, make sure the active Image Texture node is not connected to the nodetree. The active texture is also the texture shown in the viewport in the rendered mode. Remember to save your images after the baking is complete. Note: Bake currently only works in the CPU Note: This is not supported by Cycles standalone because a lot of the work is done in Blender as part of the operator only, not the engine (Cycles). Documentation: http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Bake Supported Passes: ----------------- Data Passes * Normal * UV * Diffuse/Glossy/Transmission/Subsurface/Emit Color Light Passes * AO * Combined * Shadow * Diffuse/Glossy/Transmission/Subsurface/Emit Direct/Indirect * Environment Review: D421 Reviewed by: Campbell Barton, Brecht van Lommel, Sergey Sharybin, Thomas Dinge Original design by Brecht van Lommel. The entire commit history can be found on the branch: bake-cycles
2014-01-02 21:05:07 +00:00
#if defined(__EMISSION__) && defined(__BRANCHED_PATH__)
Cycles: internal changes that should have no effect on user level yet, added shader flags for various purposes, and some code for light types other than points.
2011-09-27 20:37:24 +00:00
if(kernel_data.integrator.use_direct_light) {
Workaround for T47213: branched path sampling issues with CUDA 7.5.
2016-02-18 23:39:00 +00:00
int all = kernel_data.integrator.sample_all_lights_indirect;
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).
2015-11-22 10:48:33 +00:00
kernel_branched_path_surface_connect_light(kg,
rng,
Cycles CUDA: reduce branched path stack memory by sharing indirect ShaderData. Saves about 15% for the branched path kernel.
2016-05-24 20:28:03 +00:00
sd,
Cycles CUDA: reduce stack memory by reusing ShaderData. 57% less for path and 48% less for branched path.
2016-05-22 20:35:47 +00:00
emission_sd,
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).
2015-11-22 10:48:33 +00:00
state,
throughput,
1.0f,
L,
all);
Cycles: first step for implementation of non-progressive sampler that handles direct and indirect lighting differently. Rather than picking one light for each point on the path, it now loops over all lights for direct lighting. For indirect lighting it still picks a random light each time. It gives control over the number of AA samples, and the number of Diffuse, Glossy, Transmission, AO, Mesh Light, Background and Lamp samples for each AA sample. This helps tuning render performance/noise and tends to give less noise for renders dominated by direct lighting. This sampling mode only works on the CPU, and still needs proper tile rendering to show progress (will follow tommorrow or so), because each AA sample can be quite slow now and so the delay between each update wil be too long.
2012-06-13 11:44:48 +00:00
}
#endif
Cycles Volume Render: support for rendering of homogeneous volume with absorption. This is the simplest possible volume rendering case, constant density inside the volume and no scattering or emission. My plan is to tweak, verify and commit more volume rendering effects one by one, doing it all at once makes it difficult to verify correctness and track down bugs. Documentation is here: http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Materials/Volume Currently this hooks into path tracing in 3 ways, which should get us pretty far until we add more advanced light sampling. These 3 hooks are repeated in the path tracing, branched path tracing and transparent shadow code: * Determine active volume shader at start of the path * Change active volume shader on transmission through a surface * Light attenuation over line segments between camera, surfaces and background This is work by "storm", Stuart Broadfoot, Thomas Dinges and myself.
2013-12-28 15:56:19 +00:00
Cycles CUDA: reduce branched path stack memory by sharing indirect ShaderData. Saves about 15% for the branched path kernel.
2016-05-24 20:28:03 +00:00
if(!kernel_path_surface_bounce(kg, rng, sd, &throughput, state, L, ray))
Cycles Volume Render: support for rendering of homogeneous volume with absorption. This is the simplest possible volume rendering case, constant density inside the volume and no scattering or emission. My plan is to tweak, verify and commit more volume rendering effects one by one, doing it all at once makes it difficult to verify correctness and track down bugs. Documentation is here: http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Materials/Volume Currently this hooks into path tracing in 3 ways, which should get us pretty far until we add more advanced light sampling. These 3 hooks are repeated in the path tracing, branched path tracing and transparent shadow code: * Determine active volume shader at start of the path * Change active volume shader on transmission through a surface * Light attenuation over line segments between camera, surfaces and background This is work by "storm", Stuart Broadfoot, Thomas Dinges and myself.
2013-12-28 15:56:19 +00:00
break;
}
Cycles: first step for implementation of non-progressive sampler that handles direct and indirect lighting differently. Rather than picking one light for each point on the path, it now loops over all lights for direct lighting. For indirect lighting it still picks a random light each time. It gives control over the number of AA samples, and the number of Diffuse, Glossy, Transmission, AO, Mesh Light, Background and Lamp samples for each AA sample. This helps tuning render performance/noise and tends to give less noise for renders dominated by direct lighting. This sampling mode only works on the CPU, and still needs proper tile rendering to show progress (will follow tommorrow or so), because each AA sample can be quite slow now and so the delay between each update wil be too long.
2012-06-13 11:44:48 +00:00
}
Cycles: Tweak inline policy for some functions The goal is to make Experimental kernel closer in performance to the official kernel, avoiding spills and such. There should not be big impact on official kernel, own tests showed few percent performance drop on laptop's GPU. CPU was always the same speed on AVX, AVX2 and SSE4.1 CPUs i've been testing here. This seems to be the last essential step before we can get rid of Experimental kernel and enable SSS officially on GPU without causing some major performance issues. Surely some more tweaks are possibly required, but that we can do for until cows go home anyway.
2016-01-14 09:53:05 +00:00
ccl_device_noinline void kernel_path_ao(KernelGlobals *kg,
ShaderData *sd,
Cycles CUDA: reduce stack memory by reusing ShaderData. 57% less for path and 48% less for branched path.
2016-05-22 20:35:47 +00:00
ShaderData *emission_sd,
Cycles: Tweak inline policy for some functions The goal is to make Experimental kernel closer in performance to the official kernel, avoiding spills and such. There should not be big impact on official kernel, own tests showed few percent performance drop on laptop's GPU. CPU was always the same speed on AVX, AVX2 and SSE4.1 CPUs i've been testing here. This seems to be the last essential step before we can get rid of Experimental kernel and enable SSS officially on GPU without causing some major performance issues. Surely some more tweaks are possibly required, but that we can do for until cows go home anyway.
2016-01-14 09:53:05 +00:00
PathRadiance *L,
PathState *state,
RNG *rng,
float3 throughput)
Cycles Bake Expand Cycles to use the new baking API in Blender. It works on the selected object, and the panel can be accessed in the Render panel (similar to where it is for the Blender Internal). It bakes for the active texture of each material of the object. The active texture is currently defined as the active Image Texture node present in the material nodetree. If you don't want the baking to override an existent material, make sure the active Image Texture node is not connected to the nodetree. The active texture is also the texture shown in the viewport in the rendered mode. Remember to save your images after the baking is complete. Note: Bake currently only works in the CPU Note: This is not supported by Cycles standalone because a lot of the work is done in Blender as part of the operator only, not the engine (Cycles). Documentation: http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Bake Supported Passes: ----------------- Data Passes * Normal * UV * Diffuse/Glossy/Transmission/Subsurface/Emit Color Light Passes * AO * Combined * Shadow * Diffuse/Glossy/Transmission/Subsurface/Emit Direct/Indirect * Environment Review: D421 Reviewed by: Campbell Barton, Brecht van Lommel, Sergey Sharybin, Thomas Dinge Original design by Brecht van Lommel. The entire commit history can be found on the branch: bake-cycles
2014-01-02 21:05:07 +00:00
{
/* todo: solve correlation */
float bsdf_u, bsdf_v;
path_state_rng_2D(kg, rng, state, PRNG_BSDF_U, &bsdf_u, &bsdf_v);
float ao_factor = kernel_data.background.ao_factor;
float3 ao_N;
float3 ao_bsdf = shader_bsdf_ao(kg, sd, ao_factor, &ao_N);
float3 ao_D;
float ao_pdf;
float3 ao_alpha = shader_bsdf_alpha(kg, sd);
sample_cos_hemisphere(ao_N, bsdf_u, bsdf_v, &ao_D, &ao_pdf);
Cycles: OpenCL kernel split This commit contains all the work related on the AMD megakernel split work which was mainly done by Varun Sundar, George Kyriazis and Lenny Wang, plus some help from Sergey Sharybin, Martijn Berger, Thomas Dinges and likely someone else which we're forgetting to mention. Currently only AMD cards are enabled for the new split kernel, but it is possible to force split opencl kernel to be used by setting the following environment variable: CYCLES_OPENCL_SPLIT_KERNEL_TEST=1. Not all the features are supported yet, and that being said no motion blur, camera blur, SSS and volumetrics for now. Also transparent shadows are disabled on AMD device because of some compiler bug. This kernel is also only implements regular path tracing and supporting branched one will take a bit. Branched path tracing is exposed to the interface still, which is a bit misleading and will be hidden there soon. More feature will be enabled once they're ported to the split kernel and tested. Neither regular CPU nor CUDA has any difference, they're generating the same exact code, which means no regressions/improvements there. Based on the research paper: https://research.nvidia.com/sites/default/files/publications/laine2013hpg_paper.pdf Here's the documentation: https://docs.google.com/document/d/1LuXW-CV-sVJkQaEGZlMJ86jZ8FmoPfecaMdR-oiWbUY/edit Design discussion of the patch: https://developer.blender.org/T44197 Differential Revision: https://developer.blender.org/D1200
2015-05-09 14:34:30 +00:00
if(dot(ccl_fetch(sd, Ng), ao_D) > 0.0f && ao_pdf != 0.0f) {
Cycles Bake Expand Cycles to use the new baking API in Blender. It works on the selected object, and the panel can be accessed in the Render panel (similar to where it is for the Blender Internal). It bakes for the active texture of each material of the object. The active texture is currently defined as the active Image Texture node present in the material nodetree. If you don't want the baking to override an existent material, make sure the active Image Texture node is not connected to the nodetree. The active texture is also the texture shown in the viewport in the rendered mode. Remember to save your images after the baking is complete. Note: Bake currently only works in the CPU Note: This is not supported by Cycles standalone because a lot of the work is done in Blender as part of the operator only, not the engine (Cycles). Documentation: http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Bake Supported Passes: ----------------- Data Passes * Normal * UV * Diffuse/Glossy/Transmission/Subsurface/Emit Color Light Passes * AO * Combined * Shadow * Diffuse/Glossy/Transmission/Subsurface/Emit Direct/Indirect * Environment Review: D421 Reviewed by: Campbell Barton, Brecht van Lommel, Sergey Sharybin, Thomas Dinge Original design by Brecht van Lommel. The entire commit history can be found on the branch: bake-cycles
2014-01-02 21:05:07 +00:00
Ray light_ray;
float3 ao_shadow;
Cycles: OpenCL kernel split This commit contains all the work related on the AMD megakernel split work which was mainly done by Varun Sundar, George Kyriazis and Lenny Wang, plus some help from Sergey Sharybin, Martijn Berger, Thomas Dinges and likely someone else which we're forgetting to mention. Currently only AMD cards are enabled for the new split kernel, but it is possible to force split opencl kernel to be used by setting the following environment variable: CYCLES_OPENCL_SPLIT_KERNEL_TEST=1. Not all the features are supported yet, and that being said no motion blur, camera blur, SSS and volumetrics for now. Also transparent shadows are disabled on AMD device because of some compiler bug. This kernel is also only implements regular path tracing and supporting branched one will take a bit. Branched path tracing is exposed to the interface still, which is a bit misleading and will be hidden there soon. More feature will be enabled once they're ported to the split kernel and tested. Neither regular CPU nor CUDA has any difference, they're generating the same exact code, which means no regressions/improvements there. Based on the research paper: https://research.nvidia.com/sites/default/files/publications/laine2013hpg_paper.pdf Here's the documentation: https://docs.google.com/document/d/1LuXW-CV-sVJkQaEGZlMJ86jZ8FmoPfecaMdR-oiWbUY/edit Design discussion of the patch: https://developer.blender.org/T44197 Differential Revision: https://developer.blender.org/D1200
2015-05-09 14:34:30 +00:00
light_ray.P = ray_offset(ccl_fetch(sd, P), ccl_fetch(sd, Ng));
Cycles Bake Expand Cycles to use the new baking API in Blender. It works on the selected object, and the panel can be accessed in the Render panel (similar to where it is for the Blender Internal). It bakes for the active texture of each material of the object. The active texture is currently defined as the active Image Texture node present in the material nodetree. If you don't want the baking to override an existent material, make sure the active Image Texture node is not connected to the nodetree. The active texture is also the texture shown in the viewport in the rendered mode. Remember to save your images after the baking is complete. Note: Bake currently only works in the CPU Note: This is not supported by Cycles standalone because a lot of the work is done in Blender as part of the operator only, not the engine (Cycles). Documentation: http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Bake Supported Passes: ----------------- Data Passes * Normal * UV * Diffuse/Glossy/Transmission/Subsurface/Emit Color Light Passes * AO * Combined * Shadow * Diffuse/Glossy/Transmission/Subsurface/Emit Direct/Indirect * Environment Review: D421 Reviewed by: Campbell Barton, Brecht van Lommel, Sergey Sharybin, Thomas Dinge Original design by Brecht van Lommel. The entire commit history can be found on the branch: bake-cycles
2014-01-02 21:05:07 +00:00
light_ray.D = ao_D;
light_ray.t = kernel_data.background.ao_distance;
#ifdef __OBJECT_MOTION__
Cycles: OpenCL kernel split This commit contains all the work related on the AMD megakernel split work which was mainly done by Varun Sundar, George Kyriazis and Lenny Wang, plus some help from Sergey Sharybin, Martijn Berger, Thomas Dinges and likely someone else which we're forgetting to mention. Currently only AMD cards are enabled for the new split kernel, but it is possible to force split opencl kernel to be used by setting the following environment variable: CYCLES_OPENCL_SPLIT_KERNEL_TEST=1. Not all the features are supported yet, and that being said no motion blur, camera blur, SSS and volumetrics for now. Also transparent shadows are disabled on AMD device because of some compiler bug. This kernel is also only implements regular path tracing and supporting branched one will take a bit. Branched path tracing is exposed to the interface still, which is a bit misleading and will be hidden there soon. More feature will be enabled once they're ported to the split kernel and tested. Neither regular CPU nor CUDA has any difference, they're generating the same exact code, which means no regressions/improvements there. Based on the research paper: https://research.nvidia.com/sites/default/files/publications/laine2013hpg_paper.pdf Here's the documentation: https://docs.google.com/document/d/1LuXW-CV-sVJkQaEGZlMJ86jZ8FmoPfecaMdR-oiWbUY/edit Design discussion of the patch: https://developer.blender.org/T44197 Differential Revision: https://developer.blender.org/D1200
2015-05-09 14:34:30 +00:00
light_ray.time = ccl_fetch(sd, time);
Cycles: reduce noise using regular path tracing + subsurface scattering with new cubic and gaussian falloff. Like the branched path tracer, this will now shade all intersection points instead of using one at random.
2013-08-24 15:02:08 +00:00
#endif
Cycles: OpenCL kernel split This commit contains all the work related on the AMD megakernel split work which was mainly done by Varun Sundar, George Kyriazis and Lenny Wang, plus some help from Sergey Sharybin, Martijn Berger, Thomas Dinges and likely someone else which we're forgetting to mention. Currently only AMD cards are enabled for the new split kernel, but it is possible to force split opencl kernel to be used by setting the following environment variable: CYCLES_OPENCL_SPLIT_KERNEL_TEST=1. Not all the features are supported yet, and that being said no motion blur, camera blur, SSS and volumetrics for now. Also transparent shadows are disabled on AMD device because of some compiler bug. This kernel is also only implements regular path tracing and supporting branched one will take a bit. Branched path tracing is exposed to the interface still, which is a bit misleading and will be hidden there soon. More feature will be enabled once they're ported to the split kernel and tested. Neither regular CPU nor CUDA has any difference, they're generating the same exact code, which means no regressions/improvements there. Based on the research paper: https://research.nvidia.com/sites/default/files/publications/laine2013hpg_paper.pdf Here's the documentation: https://docs.google.com/document/d/1LuXW-CV-sVJkQaEGZlMJ86jZ8FmoPfecaMdR-oiWbUY/edit Design discussion of the patch: https://developer.blender.org/T44197 Differential Revision: https://developer.blender.org/D1200
2015-05-09 14:34:30 +00:00
light_ray.dP = ccl_fetch(sd, dP);
Cycles Bake Expand Cycles to use the new baking API in Blender. It works on the selected object, and the panel can be accessed in the Render panel (similar to where it is for the Blender Internal). It bakes for the active texture of each material of the object. The active texture is currently defined as the active Image Texture node present in the material nodetree. If you don't want the baking to override an existent material, make sure the active Image Texture node is not connected to the nodetree. The active texture is also the texture shown in the viewport in the rendered mode. Remember to save your images after the baking is complete. Note: Bake currently only works in the CPU Note: This is not supported by Cycles standalone because a lot of the work is done in Blender as part of the operator only, not the engine (Cycles). Documentation: http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Bake Supported Passes: ----------------- Data Passes * Normal * UV * Diffuse/Glossy/Transmission/Subsurface/Emit Color Light Passes * AO * Combined * Shadow * Diffuse/Glossy/Transmission/Subsurface/Emit Direct/Indirect * Environment Review: D421 Reviewed by: Campbell Barton, Brecht van Lommel, Sergey Sharybin, Thomas Dinge Original design by Brecht van Lommel. The entire commit history can be found on the branch: bake-cycles
2014-01-02 21:05:07 +00:00
light_ray.dD = differential3_zero();
Cycles CUDA: reduce stack memory by reusing ShaderData. 57% less for path and 48% less for branched path.
2016-05-22 20:35:47 +00:00
if(!shadow_blocked(kg, emission_sd, state, &light_ray, &ao_shadow))
Cycles Bake Expand Cycles to use the new baking API in Blender. It works on the selected object, and the panel can be accessed in the Render panel (similar to where it is for the Blender Internal). It bakes for the active texture of each material of the object. The active texture is currently defined as the active Image Texture node present in the material nodetree. If you don't want the baking to override an existent material, make sure the active Image Texture node is not connected to the nodetree. The active texture is also the texture shown in the viewport in the rendered mode. Remember to save your images after the baking is complete. Note: Bake currently only works in the CPU Note: This is not supported by Cycles standalone because a lot of the work is done in Blender as part of the operator only, not the engine (Cycles). Documentation: http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Bake Supported Passes: ----------------- Data Passes * Normal * UV * Diffuse/Glossy/Transmission/Subsurface/Emit Color Light Passes * AO * Combined * Shadow * Diffuse/Glossy/Transmission/Subsurface/Emit Direct/Indirect * Environment Review: D421 Reviewed by: Campbell Barton, Brecht van Lommel, Sergey Sharybin, Thomas Dinge Original design by Brecht van Lommel. The entire commit history can be found on the branch: bake-cycles
2014-01-02 21:05:07 +00:00
path_radiance_accum_ao(L, throughput, ao_alpha, ao_bsdf, ao_shadow, state->bounce);
}
}
Cycles: first step for implementation of non-progressive sampler that handles direct and indirect lighting differently. Rather than picking one light for each point on the path, it now loops over all lights for direct lighting. For indirect lighting it still picks a random light each time. It gives control over the number of AA samples, and the number of Diffuse, Glossy, Transmission, AO, Mesh Light, Background and Lamp samples for each AA sample. This helps tuning render performance/noise and tends to give less noise for renders dominated by direct lighting. This sampling mode only works on the CPU, and still needs proper tile rendering to show progress (will follow tommorrow or so), because each AA sample can be quite slow now and so the delay between each update wil be too long.
2012-06-13 11:44:48 +00:00
Cycles-Bake: Subsurface Scattering support (fix T40060) This fixes the SSS Direct/Indirect passes as well as the Combined pass. Patch reviewed and with fixes and contributions from Brecht van Lommel. Note: displacement/bump map (related to the report) will be handled separately Reviewers: brecht Differential Revision: https://developer.blender.org/D503
2014-05-07 14:55:09 +00:00
#ifdef __SUBSURFACE__
Cycles: Recent SSS inline changes broke CPU tests Very weird, but let's just fall back a bit for now.
2016-08-03 13:27:48 +00:00
# ifndef __KERNEL_CUDA__
ccl_device
# else
ccl_device_inline
# endif
bool kernel_path_subsurface_scatter(
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).
2015-11-22 10:48:33 +00:00
KernelGlobals *kg,
ShaderData *sd,
Cycles CUDA: reduce stack memory by reusing ShaderData. 57% less for path and 48% less for branched path.
2016-05-22 20:35:47 +00:00
ShaderData *emission_sd,
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).
2015-11-22 10:48:33 +00:00
PathRadiance *L,
PathState *state,
RNG *rng,
Ray *ray,
float3 *throughput,
SubsurfaceIndirectRays *ss_indirect)
Cycles-Bake: Subsurface Scattering support (fix T40060) This fixes the SSS Direct/Indirect passes as well as the Combined pass. Patch reviewed and with fixes and contributions from Brecht van Lommel. Note: displacement/bump map (related to the report) will be handled separately Reviewers: brecht Differential Revision: https://developer.blender.org/D503
2014-05-07 14:55:09 +00:00
{
float bssrdf_probability;
ShaderClosure *sc = subsurface_scatter_pick_closure(kg, sd, &bssrdf_probability);
/* modify throughput for picking bssrdf or bsdf */
*throughput *= bssrdf_probability;
/* do bssrdf scatter step if we picked a bssrdf closure */
if(sc) {
Cycles: Set of fixes for delayed SSS ray tracing There were multiple issues which are solved now: - It was possible that ray wouldn't be bounced off the BSSRDF, for example when PDF or shader eval is zero. In this case PathState might have been left in pre-bounced state which would have been gave incorrect shading results. This is solved by having separate PathState for each of the hits. - Path radiance summing wasn't happening correct as well, indirect rays were using wrong path radiance in the case when there were more than one hit recorded. This is now using a bit trickier state machine which calculates path radiance for just SSS (both direct and indirect) and then sums it back to the final radiance. - Previous commit wasn't totally correct either and was an induced bug due to wrong path state left from the "un-happened" ray bounce. There should be no special case happening here, BSSRDFs will be replaced with diffuse ones due to PATH_RAY_DIFFUSE_ANCESTOR flag. - Merged back codebases for "delayed" and "immediate" indirect SSS ray tracing, hopefully making it easier to maintain the codebase. Sure this changes brings memory usage back by about 4-5%, but overall it's still about 2x memory reduction for the experimental kernel here. Thanks Brecht for the review!
2015-11-28 14:30:35 +00:00
/* We should never have two consecutive BSSRDF bounces,
* the second one should be converted to a diffuse BSDF to
* avoid this.
*/
kernel_assert(!ss_indirect->tracing);
Cycles-Bake: Subsurface Scattering support (fix T40060) This fixes the SSS Direct/Indirect passes as well as the Combined pass. Patch reviewed and with fixes and contributions from Brecht van Lommel. Note: displacement/bump map (related to the report) will be handled separately Reviewers: brecht Differential Revision: https://developer.blender.org/D503
2014-05-07 14:55:09 +00:00
Cycles: Set of fixes for delayed SSS ray tracing There were multiple issues which are solved now: - It was possible that ray wouldn't be bounced off the BSSRDF, for example when PDF or shader eval is zero. In this case PathState might have been left in pre-bounced state which would have been gave incorrect shading results. This is solved by having separate PathState for each of the hits. - Path radiance summing wasn't happening correct as well, indirect rays were using wrong path radiance in the case when there were more than one hit recorded. This is now using a bit trickier state machine which calculates path radiance for just SSS (both direct and indirect) and then sums it back to the final radiance. - Previous commit wasn't totally correct either and was an induced bug due to wrong path state left from the "un-happened" ray bounce. There should be no special case happening here, BSSRDFs will be replaced with diffuse ones due to PATH_RAY_DIFFUSE_ANCESTOR flag. - Merged back codebases for "delayed" and "immediate" indirect SSS ray tracing, hopefully making it easier to maintain the codebase. Sure this changes brings memory usage back by about 4-5%, but overall it's still about 2x memory reduction for the experimental kernel here. Thanks Brecht for the review!
2015-11-28 14:30:35 +00:00
uint lcg_state = lcg_state_init(rng, state, 0x68bc21eb);
Cycles: Fallback to diffuse BSDF for the indirect SSS rays when BSSRDF is hit This is actually how it was intended to work, just didn't notice it wasn't really happening in the main ray loop. Solves some memory issues reported in T46880.
2015-11-28 10:38:12 +00:00
Cysles: Avoid having ShaderData on the stack This commit introduces a SSS-oriented intersection structure which is replacing old logic of having separate arrays for just intersections and shader data and encapsulates all the data needed for SSS evaluation. This giver a huge stack memory saving on GPU. In own experiments it gave 25% memory usage reduction on GTX560Ti (722MB vs. 946MB). Unfortunately, this gave some performance loss of 20% which only happens on GPU. This is perhaps due to different memory access pattern. Will be solved in the future, hopefully. Famous saying: won in memory - lost in time (which is also valid in other way around).
2015-11-22 10:00:29 +00:00
SubsurfaceIntersection ss_isect;
Cycles-Bake: Subsurface Scattering support (fix T40060) This fixes the SSS Direct/Indirect passes as well as the Combined pass. Patch reviewed and with fixes and contributions from Brecht van Lommel. Note: displacement/bump map (related to the report) will be handled separately Reviewers: brecht Differential Revision: https://developer.blender.org/D503
2014-05-07 14:55:09 +00:00
float bssrdf_u, bssrdf_v;
path_state_rng_2D(kg, rng, state, PRNG_BSDF_U, &bssrdf_u, &bssrdf_v);
Cysles: Avoid having ShaderData on the stack This commit introduces a SSS-oriented intersection structure which is replacing old logic of having separate arrays for just intersections and shader data and encapsulates all the data needed for SSS evaluation. This giver a huge stack memory saving on GPU. In own experiments it gave 25% memory usage reduction on GTX560Ti (722MB vs. 946MB). Unfortunately, this gave some performance loss of 20% which only happens on GPU. This is perhaps due to different memory access pattern. Will be solved in the future, hopefully. Famous saying: won in memory - lost in time (which is also valid in other way around).
2015-11-22 10:00:29 +00:00
int num_hits = subsurface_scatter_multi_intersect(kg,
&ss_isect,
sd,
sc,
&lcg_state,
bssrdf_u, bssrdf_v,
false);
Cycles: Cleanup, indent nested preprocessor directives Quite straightforward, main trick is happening in path_source_replace_includes(). Reviewers: brecht, dingto, lukasstockner97, juicyfruit Differential Revision: https://developer.blender.org/D1794
2016-02-12 17:33:43 +00:00
# ifdef __VOLUME__
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).
2015-11-22 10:48:33 +00:00
ss_indirect->need_update_volume_stack =
kernel_data.integrator.use_volumes &&
ccl_fetch(sd, flag) & SD_OBJECT_INTERSECTS_VOLUME;
Cycles: Cleanup, indent nested preprocessor directives Quite straightforward, main trick is happening in path_source_replace_includes(). Reviewers: brecht, dingto, lukasstockner97, juicyfruit Differential Revision: https://developer.blender.org/D1794
2016-02-12 17:33:43 +00:00
# endif
Cycles-Bake: Subsurface Scattering support (fix T40060) This fixes the SSS Direct/Indirect passes as well as the Combined pass. Patch reviewed and with fixes and contributions from Brecht van Lommel. Note: displacement/bump map (related to the report) will be handled separately Reviewers: brecht Differential Revision: https://developer.blender.org/D503
2014-05-07 14:55:09 +00:00
/* compute lighting with the BSDF closure */
for(int hit = 0; hit < num_hits; hit++) {
Cysles: Avoid having ShaderData on the stack This commit introduces a SSS-oriented intersection structure which is replacing old logic of having separate arrays for just intersections and shader data and encapsulates all the data needed for SSS evaluation. This giver a huge stack memory saving on GPU. In own experiments it gave 25% memory usage reduction on GTX560Ti (722MB vs. 946MB). Unfortunately, this gave some performance loss of 20% which only happens on GPU. This is perhaps due to different memory access pattern. Will be solved in the future, hopefully. Famous saying: won in memory - lost in time (which is also valid in other way around).
2015-11-22 10:00:29 +00:00
/* NOTE: We reuse the existing ShaderData, we assume the path
* integration loop stops when this function returns true.
*/
subsurface_scatter_multi_setup(kg,
&ss_isect,
hit,
sd,
Cycles: Refactor how we pass bounce info to light path node. This commit changes the way how we pass bounce information to the Light Path node. Instead of manualy copying the bounces into ShaderData, we now directly pass PathState. This reduces the arguments that we need to pass around and also makes it easier to extend the feature. This commit also exposes the Transmission Bounce Depth to the Light Path node. It works similar to the Transparent Depth Output: Replace a Transmission lightpath after X bounces with another shader, e.g a Diffuse one. This can be used to avoid black surfaces, due to low amount of max bounces. Reviewed by Sergey and Brecht, thanks for some hlp with this. I tested compilation and usage on CPU (SVM and OSL), CUDA, OpenCL Split and Mega kernel. Hopefully this covers all devices. :)
2016-01-06 22:38:13 +00:00
state,
Cysles: Avoid having ShaderData on the stack This commit introduces a SSS-oriented intersection structure which is replacing old logic of having separate arrays for just intersections and shader data and encapsulates all the data needed for SSS evaluation. This giver a huge stack memory saving on GPU. In own experiments it gave 25% memory usage reduction on GTX560Ti (722MB vs. 946MB). Unfortunately, this gave some performance loss of 20% which only happens on GPU. This is perhaps due to different memory access pattern. Will be solved in the future, hopefully. Famous saying: won in memory - lost in time (which is also valid in other way around).
2015-11-22 10:00:29 +00:00
state->flag,
sc,
false);
Cycles: Set of fixes for delayed SSS ray tracing There were multiple issues which are solved now: - It was possible that ray wouldn't be bounced off the BSSRDF, for example when PDF or shader eval is zero. In this case PathState might have been left in pre-bounced state which would have been gave incorrect shading results. This is solved by having separate PathState for each of the hits. - Path radiance summing wasn't happening correct as well, indirect rays were using wrong path radiance in the case when there were more than one hit recorded. This is now using a bit trickier state machine which calculates path radiance for just SSS (both direct and indirect) and then sums it back to the final radiance. - Previous commit wasn't totally correct either and was an induced bug due to wrong path state left from the "un-happened" ray bounce. There should be no special case happening here, BSSRDFs will be replaced with diffuse ones due to PATH_RAY_DIFFUSE_ANCESTOR flag. - Merged back codebases for "delayed" and "immediate" indirect SSS ray tracing, hopefully making it easier to maintain the codebase. Sure this changes brings memory usage back by about 4-5%, but overall it's still about 2x memory reduction for the experimental kernel here. Thanks Brecht for the review!
2015-11-28 14:30:35 +00:00
PathState *hit_state = &ss_indirect->state[ss_indirect->num_rays];
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).
2015-11-22 10:48:33 +00:00
Ray *hit_ray = &ss_indirect->rays[ss_indirect->num_rays];
float3 *hit_tp = &ss_indirect->throughputs[ss_indirect->num_rays];
Cycles: Set of fixes for delayed SSS ray tracing There were multiple issues which are solved now: - It was possible that ray wouldn't be bounced off the BSSRDF, for example when PDF or shader eval is zero. In this case PathState might have been left in pre-bounced state which would have been gave incorrect shading results. This is solved by having separate PathState for each of the hits. - Path radiance summing wasn't happening correct as well, indirect rays were using wrong path radiance in the case when there were more than one hit recorded. This is now using a bit trickier state machine which calculates path radiance for just SSS (both direct and indirect) and then sums it back to the final radiance. - Previous commit wasn't totally correct either and was an induced bug due to wrong path state left from the "un-happened" ray bounce. There should be no special case happening here, BSSRDFs will be replaced with diffuse ones due to PATH_RAY_DIFFUSE_ANCESTOR flag. - Merged back codebases for "delayed" and "immediate" indirect SSS ray tracing, hopefully making it easier to maintain the codebase. Sure this changes brings memory usage back by about 4-5%, but overall it's still about 2x memory reduction for the experimental kernel here. Thanks Brecht for the review!
2015-11-28 14:30:35 +00:00
PathRadiance *hit_L = &ss_indirect->L[ss_indirect->num_rays];
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).
2015-11-22 10:48:33 +00:00
*hit_state = *state;
*hit_ray = *ray;
*hit_tp = *throughput;
Cycles-Bake: Subsurface Scattering support (fix T40060) This fixes the SSS Direct/Indirect passes as well as the Combined pass. Patch reviewed and with fixes and contributions from Brecht van Lommel. Note: displacement/bump map (related to the report) will be handled separately Reviewers: brecht Differential Revision: https://developer.blender.org/D503
2014-05-07 14:55:09 +00:00
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).
2015-11-22 10:48:33 +00:00
hit_state->rng_offset += PRNG_BOUNCE_NUM;
Cycles-Bake: Subsurface Scattering support (fix T40060) This fixes the SSS Direct/Indirect passes as well as the Combined pass. Patch reviewed and with fixes and contributions from Brecht van Lommel. Note: displacement/bump map (related to the report) will be handled separately Reviewers: brecht Differential Revision: https://developer.blender.org/D503
2014-05-07 14:55:09 +00:00
Cycles: Set of fixes for delayed SSS ray tracing There were multiple issues which are solved now: - It was possible that ray wouldn't be bounced off the BSSRDF, for example when PDF or shader eval is zero. In this case PathState might have been left in pre-bounced state which would have been gave incorrect shading results. This is solved by having separate PathState for each of the hits. - Path radiance summing wasn't happening correct as well, indirect rays were using wrong path radiance in the case when there were more than one hit recorded. This is now using a bit trickier state machine which calculates path radiance for just SSS (both direct and indirect) and then sums it back to the final radiance. - Previous commit wasn't totally correct either and was an induced bug due to wrong path state left from the "un-happened" ray bounce. There should be no special case happening here, BSSRDFs will be replaced with diffuse ones due to PATH_RAY_DIFFUSE_ANCESTOR flag. - Merged back codebases for "delayed" and "immediate" indirect SSS ray tracing, hopefully making it easier to maintain the codebase. Sure this changes brings memory usage back by about 4-5%, but overall it's still about 2x memory reduction for the experimental kernel here. Thanks Brecht for the review!
2015-11-28 14:30:35 +00:00
path_radiance_init(hit_L, kernel_data.film.use_light_pass);
Cycles: Fix SSS object not properly reflected in glossy object with indirect clamping This fixes remained issues reported in T46908.
2015-12-02 10:59:12 +00:00
hit_L->direct_throughput = L->direct_throughput;
path_radiance_copy_indirect(hit_L, L);
Cycles CUDA: reduce stack memory by reusing ShaderData. 57% less for path and 48% less for branched path.
2016-05-22 20:35:47 +00:00
kernel_path_surface_connect_light(kg, rng, sd, emission_sd, *hit_tp, state, hit_L);
Cysles: Avoid having ShaderData on the stack This commit introduces a SSS-oriented intersection structure which is replacing old logic of having separate arrays for just intersections and shader data and encapsulates all the data needed for SSS evaluation. This giver a huge stack memory saving on GPU. In own experiments it gave 25% memory usage reduction on GTX560Ti (722MB vs. 946MB). Unfortunately, this gave some performance loss of 20% which only happens on GPU. This is perhaps due to different memory access pattern. Will be solved in the future, hopefully. Famous saying: won in memory - lost in time (which is also valid in other way around).
2015-11-22 10:00:29 +00:00
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).
2015-11-22 10:48:33 +00:00
if(kernel_path_surface_bounce(kg,
rng,
sd,
hit_tp,
hit_state,
Cycles: Set of fixes for delayed SSS ray tracing There were multiple issues which are solved now: - It was possible that ray wouldn't be bounced off the BSSRDF, for example when PDF or shader eval is zero. In this case PathState might have been left in pre-bounced state which would have been gave incorrect shading results. This is solved by having separate PathState for each of the hits. - Path radiance summing wasn't happening correct as well, indirect rays were using wrong path radiance in the case when there were more than one hit recorded. This is now using a bit trickier state machine which calculates path radiance for just SSS (both direct and indirect) and then sums it back to the final radiance. - Previous commit wasn't totally correct either and was an induced bug due to wrong path state left from the "un-happened" ray bounce. There should be no special case happening here, BSSRDFs will be replaced with diffuse ones due to PATH_RAY_DIFFUSE_ANCESTOR flag. - Merged back codebases for "delayed" and "immediate" indirect SSS ray tracing, hopefully making it easier to maintain the codebase. Sure this changes brings memory usage back by about 4-5%, but overall it's still about 2x memory reduction for the experimental kernel here. Thanks Brecht for the review!
2015-11-28 14:30:35 +00:00
hit_L,
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).
2015-11-22 10:48:33 +00:00
hit_ray))
{
Cycles: Cleanup, indent nested preprocessor directives Quite straightforward, main trick is happening in path_source_replace_includes(). Reviewers: brecht, dingto, lukasstockner97, juicyfruit Differential Revision: https://developer.blender.org/D1794
2016-02-12 17:33:43 +00:00
# ifdef __LAMP_MIS__
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).
2015-11-22 10:48:33 +00:00
hit_state->ray_t = 0.0f;
Cycles: Cleanup, indent nested preprocessor directives Quite straightforward, main trick is happening in path_source_replace_includes(). Reviewers: brecht, dingto, lukasstockner97, juicyfruit Differential Revision: https://developer.blender.org/D1794
2016-02-12 17:33:43 +00:00
# endif
Cycles: Fully roll-back to non-delayed SSS indirect rays for CPU There are some issues to be solved with the recent optimization we did for the indirect rays for the SSS. Those issues will take a bit of a time to be fully solved still and we need to unlock Caminandes team now, so let's revert some changes back. CUDA will still use delayed indirect rays since it's an experimental feature. For the details about what's to be done still please refer to T46880.
2015-11-27 12:12:44 +00:00
Cycles: Cleanup, indent nested preprocessor directives Quite straightforward, main trick is happening in path_source_replace_includes(). Reviewers: brecht, dingto, lukasstockner97, juicyfruit Differential Revision: https://developer.blender.org/D1794
2016-02-12 17:33:43 +00:00
# ifdef __VOLUME__
Cycles: Fix wrong volume stack after SSS bounce Was introduced by a recent fixes, now it should be all correct and additionally it solves the TODO mentioned in the code.
2015-11-28 15:06:44 +00:00
if(ss_indirect->need_update_volume_stack) {
Ray volume_ray = *ray;
/* Setup ray from previous surface point to the new one. */
volume_ray.D = normalize_len(hit_ray->P - volume_ray.P,
&volume_ray.t);
kernel_volume_stack_update_for_subsurface(
kg,
Cycles CUDA: reduce stack memory by reusing ShaderData. 57% less for path and 48% less for branched path.
2016-05-22 20:35:47 +00:00
emission_sd,
Cycles: Fix wrong volume stack after SSS bounce Was introduced by a recent fixes, now it should be all correct and additionally it solves the TODO mentioned in the code.
2015-11-28 15:06:44 +00:00
&volume_ray,
hit_state->volume_stack);
}
Cycles: Cleanup, indent nested preprocessor directives Quite straightforward, main trick is happening in path_source_replace_includes(). Reviewers: brecht, dingto, lukasstockner97, juicyfruit Differential Revision: https://developer.blender.org/D1794
2016-02-12 17:33:43 +00:00
# endif
Cycles: Fix SSS object not properly reflected in glossy object with indirect clamping This fixes remained issues reported in T46908.
2015-12-02 10:59:12 +00:00
path_radiance_reset_indirect(L);
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).
2015-11-22 10:48:33 +00:00
ss_indirect->num_rays++;
Cycles: Set of fixes for delayed SSS ray tracing There were multiple issues which are solved now: - It was possible that ray wouldn't be bounced off the BSSRDF, for example when PDF or shader eval is zero. In this case PathState might have been left in pre-bounced state which would have been gave incorrect shading results. This is solved by having separate PathState for each of the hits. - Path radiance summing wasn't happening correct as well, indirect rays were using wrong path radiance in the case when there were more than one hit recorded. This is now using a bit trickier state machine which calculates path radiance for just SSS (both direct and indirect) and then sums it back to the final radiance. - Previous commit wasn't totally correct either and was an induced bug due to wrong path state left from the "un-happened" ray bounce. There should be no special case happening here, BSSRDFs will be replaced with diffuse ones due to PATH_RAY_DIFFUSE_ANCESTOR flag. - Merged back codebases for "delayed" and "immediate" indirect SSS ray tracing, hopefully making it easier to maintain the codebase. Sure this changes brings memory usage back by about 4-5%, but overall it's still about 2x memory reduction for the experimental kernel here. Thanks Brecht for the review!
2015-11-28 14:30:35 +00:00
}
else {
path_radiance_accum_sample(L, hit_L, 1);
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).
2015-11-22 10:48:33 +00:00
}
}
return true;
}
return false;
}
Cycles: Fix wrong original ray used for SSS baking Also de-duplicated some code by moving to an utility function.
2015-11-28 14:43:54 +00:00
ccl_device_inline void kernel_path_subsurface_init_indirect(
SubsurfaceIndirectRays *ss_indirect)
{
ss_indirect->tracing = false;
ss_indirect->num_rays = 0;
}
Cycles: Set of fixes for delayed SSS ray tracing There were multiple issues which are solved now: - It was possible that ray wouldn't be bounced off the BSSRDF, for example when PDF or shader eval is zero. In this case PathState might have been left in pre-bounced state which would have been gave incorrect shading results. This is solved by having separate PathState for each of the hits. - Path radiance summing wasn't happening correct as well, indirect rays were using wrong path radiance in the case when there were more than one hit recorded. This is now using a bit trickier state machine which calculates path radiance for just SSS (both direct and indirect) and then sums it back to the final radiance. - Previous commit wasn't totally correct either and was an induced bug due to wrong path state left from the "un-happened" ray bounce. There should be no special case happening here, BSSRDFs will be replaced with diffuse ones due to PATH_RAY_DIFFUSE_ANCESTOR flag. - Merged back codebases for "delayed" and "immediate" indirect SSS ray tracing, hopefully making it easier to maintain the codebase. Sure this changes brings memory usage back by about 4-5%, but overall it's still about 2x memory reduction for the experimental kernel here. Thanks Brecht for the review!
2015-11-28 14:30:35 +00:00
ccl_device void kernel_path_subsurface_accum_indirect(
SubsurfaceIndirectRays *ss_indirect,
PathRadiance *L)
{
if(ss_indirect->tracing) {
path_radiance_sum_indirect(L);
path_radiance_accum_sample(&ss_indirect->direct_L, L, 1);
if(ss_indirect->num_rays == 0) {
*L = ss_indirect->direct_L;
}
}
}
Cycles: Trace indirect subsurface rays by restarting the integrator loop This gives much lower stack usage on GPU and reduces kernel memory size to around 448MB on GTX560Ti (comparing to 652MB with previous commit and 946MB with official release). There's also a barely measurable speedup of around 5%, but this is to be confirmed still. At this stage we're using only ~3% for the experimental kernel and SSS rendering seems to be faster by 40% and after some further testing we might consider making SSS and CMJ official features and remove experimental precompiled kernels.
2015-11-22 11:08:03 +00:00
ccl_device void kernel_path_subsurface_setup_indirect(
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).
2015-11-22 10:48:33 +00:00
KernelGlobals *kg,
Cycles: Trace indirect subsurface rays by restarting the integrator loop This gives much lower stack usage on GPU and reduces kernel memory size to around 448MB on GTX560Ti (comparing to 652MB with previous commit and 946MB with official release). There's also a barely measurable speedup of around 5%, but this is to be confirmed still. At this stage we're using only ~3% for the experimental kernel and SSS rendering seems to be faster by 40% and after some further testing we might consider making SSS and CMJ official features and remove experimental precompiled kernels.
2015-11-22 11:08:03 +00:00
SubsurfaceIndirectRays *ss_indirect,
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).
2015-11-22 10:48:33 +00:00
PathState *state,
Ray *ray,
Cycles: Set of fixes for delayed SSS ray tracing There were multiple issues which are solved now: - It was possible that ray wouldn't be bounced off the BSSRDF, for example when PDF or shader eval is zero. In this case PathState might have been left in pre-bounced state which would have been gave incorrect shading results. This is solved by having separate PathState for each of the hits. - Path radiance summing wasn't happening correct as well, indirect rays were using wrong path radiance in the case when there were more than one hit recorded. This is now using a bit trickier state machine which calculates path radiance for just SSS (both direct and indirect) and then sums it back to the final radiance. - Previous commit wasn't totally correct either and was an induced bug due to wrong path state left from the "un-happened" ray bounce. There should be no special case happening here, BSSRDFs will be replaced with diffuse ones due to PATH_RAY_DIFFUSE_ANCESTOR flag. - Merged back codebases for "delayed" and "immediate" indirect SSS ray tracing, hopefully making it easier to maintain the codebase. Sure this changes brings memory usage back by about 4-5%, but overall it's still about 2x memory reduction for the experimental kernel here. Thanks Brecht for the review!
2015-11-28 14:30:35 +00:00
PathRadiance *L,
Cycles: Trace indirect subsurface rays by restarting the integrator loop This gives much lower stack usage on GPU and reduces kernel memory size to around 448MB on GTX560Ti (comparing to 652MB with previous commit and 946MB with official release). There's also a barely measurable speedup of around 5%, but this is to be confirmed still. At this stage we're using only ~3% for the experimental kernel and SSS rendering seems to be faster by 40% and after some further testing we might consider making SSS and CMJ official features and remove experimental precompiled kernels.
2015-11-22 11:08:03 +00:00
float3 *throughput)
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).
2015-11-22 10:48:33 +00:00
{
Cycles: Set of fixes for delayed SSS ray tracing There were multiple issues which are solved now: - It was possible that ray wouldn't be bounced off the BSSRDF, for example when PDF or shader eval is zero. In this case PathState might have been left in pre-bounced state which would have been gave incorrect shading results. This is solved by having separate PathState for each of the hits. - Path radiance summing wasn't happening correct as well, indirect rays were using wrong path radiance in the case when there were more than one hit recorded. This is now using a bit trickier state machine which calculates path radiance for just SSS (both direct and indirect) and then sums it back to the final radiance. - Previous commit wasn't totally correct either and was an induced bug due to wrong path state left from the "un-happened" ray bounce. There should be no special case happening here, BSSRDFs will be replaced with diffuse ones due to PATH_RAY_DIFFUSE_ANCESTOR flag. - Merged back codebases for "delayed" and "immediate" indirect SSS ray tracing, hopefully making it easier to maintain the codebase. Sure this changes brings memory usage back by about 4-5%, but overall it's still about 2x memory reduction for the experimental kernel here. Thanks Brecht for the review!
2015-11-28 14:30:35 +00:00
if(!ss_indirect->tracing) {
ss_indirect->direct_L = *L;
}
ss_indirect->tracing = true;
Cycles: Trace indirect subsurface rays by restarting the integrator loop This gives much lower stack usage on GPU and reduces kernel memory size to around 448MB on GTX560Ti (comparing to 652MB with previous commit and 946MB with official release). There's also a barely measurable speedup of around 5%, but this is to be confirmed still. At this stage we're using only ~3% for the experimental kernel and SSS rendering seems to be faster by 40% and after some further testing we might consider making SSS and CMJ official features and remove experimental precompiled kernels.
2015-11-22 11:08:03 +00:00
/* Setup state, ray and throughput for indirect SSS rays. */
ss_indirect->num_rays--;
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).
2015-11-22 10:48:33 +00:00
Cycles: Trace indirect subsurface rays by restarting the integrator loop This gives much lower stack usage on GPU and reduces kernel memory size to around 448MB on GTX560Ti (comparing to 652MB with previous commit and 946MB with official release). There's also a barely measurable speedup of around 5%, but this is to be confirmed still. At this stage we're using only ~3% for the experimental kernel and SSS rendering seems to be faster by 40% and after some further testing we might consider making SSS and CMJ official features and remove experimental precompiled kernels.
2015-11-22 11:08:03 +00:00
Ray *indirect_ray = &ss_indirect->rays[ss_indirect->num_rays];
Cycles: Set of fixes for delayed SSS ray tracing There were multiple issues which are solved now: - It was possible that ray wouldn't be bounced off the BSSRDF, for example when PDF or shader eval is zero. In this case PathState might have been left in pre-bounced state which would have been gave incorrect shading results. This is solved by having separate PathState for each of the hits. - Path radiance summing wasn't happening correct as well, indirect rays were using wrong path radiance in the case when there were more than one hit recorded. This is now using a bit trickier state machine which calculates path radiance for just SSS (both direct and indirect) and then sums it back to the final radiance. - Previous commit wasn't totally correct either and was an induced bug due to wrong path state left from the "un-happened" ray bounce. There should be no special case happening here, BSSRDFs will be replaced with diffuse ones due to PATH_RAY_DIFFUSE_ANCESTOR flag. - Merged back codebases for "delayed" and "immediate" indirect SSS ray tracing, hopefully making it easier to maintain the codebase. Sure this changes brings memory usage back by about 4-5%, but overall it's still about 2x memory reduction for the experimental kernel here. Thanks Brecht for the review!
2015-11-28 14:30:35 +00:00
PathRadiance *indirect_L = &ss_indirect->L[ss_indirect->num_rays];
Fix T39823: SSS scatter doesn't update volume stack, causing shading artifacts Basically the title says it all, we need to update volume stack when doing ray scatter for SSS. This leads to speed regressions in cases scene does have both volume and SSS (performance in case there's no SSS or no volume should be the same). We might try optimizing kernel_path_subsurface_update_volume_stack() a bit by either recording all intersections or using some more appropriate visibility flags. Reviewers: brecht, juicyfruit, dingto Differential Revision: https://developer.blender.org/D795
2014-09-18 11:06:05 +00:00
Cycles: Set of fixes for delayed SSS ray tracing There were multiple issues which are solved now: - It was possible that ray wouldn't be bounced off the BSSRDF, for example when PDF or shader eval is zero. In this case PathState might have been left in pre-bounced state which would have been gave incorrect shading results. This is solved by having separate PathState for each of the hits. - Path radiance summing wasn't happening correct as well, indirect rays were using wrong path radiance in the case when there were more than one hit recorded. This is now using a bit trickier state machine which calculates path radiance for just SSS (both direct and indirect) and then sums it back to the final radiance. - Previous commit wasn't totally correct either and was an induced bug due to wrong path state left from the "un-happened" ray bounce. There should be no special case happening here, BSSRDFs will be replaced with diffuse ones due to PATH_RAY_DIFFUSE_ANCESTOR flag. - Merged back codebases for "delayed" and "immediate" indirect SSS ray tracing, hopefully making it easier to maintain the codebase. Sure this changes brings memory usage back by about 4-5%, but overall it's still about 2x memory reduction for the experimental kernel here. Thanks Brecht for the review!
2015-11-28 14:30:35 +00:00
*state = ss_indirect->state[ss_indirect->num_rays];
Cycles: Trace indirect subsurface rays by restarting the integrator loop This gives much lower stack usage on GPU and reduces kernel memory size to around 448MB on GTX560Ti (comparing to 652MB with previous commit and 946MB with official release). There's also a barely measurable speedup of around 5%, but this is to be confirmed still. At this stage we're using only ~3% for the experimental kernel and SSS rendering seems to be faster by 40% and after some further testing we might consider making SSS and CMJ official features and remove experimental precompiled kernels.
2015-11-22 11:08:03 +00:00
*ray = *indirect_ray;
Cycles: Set of fixes for delayed SSS ray tracing There were multiple issues which are solved now: - It was possible that ray wouldn't be bounced off the BSSRDF, for example when PDF or shader eval is zero. In this case PathState might have been left in pre-bounced state which would have been gave incorrect shading results. This is solved by having separate PathState for each of the hits. - Path radiance summing wasn't happening correct as well, indirect rays were using wrong path radiance in the case when there were more than one hit recorded. This is now using a bit trickier state machine which calculates path radiance for just SSS (both direct and indirect) and then sums it back to the final radiance. - Previous commit wasn't totally correct either and was an induced bug due to wrong path state left from the "un-happened" ray bounce. There should be no special case happening here, BSSRDFs will be replaced with diffuse ones due to PATH_RAY_DIFFUSE_ANCESTOR flag. - Merged back codebases for "delayed" and "immediate" indirect SSS ray tracing, hopefully making it easier to maintain the codebase. Sure this changes brings memory usage back by about 4-5%, but overall it's still about 2x memory reduction for the experimental kernel here. Thanks Brecht for the review!
2015-11-28 14:30:35 +00:00
*L = *indirect_L;
*throughput = ss_indirect->throughputs[ss_indirect->num_rays];
Revert "Cycles: Fix wrong SSS with regular path tracing and clamping enabled" This wasn't really a complete fix and only worked if there was a single scatter event recorded only. Proper fix requires some more thoughts to make it correct without memory use increase. This reverts commit bf9e88bfbebaf5c6228363560970fa526e779c8b.
2015-11-27 11:41:05 +00:00
Cycles: Set of fixes for delayed SSS ray tracing There were multiple issues which are solved now: - It was possible that ray wouldn't be bounced off the BSSRDF, for example when PDF or shader eval is zero. In this case PathState might have been left in pre-bounced state which would have been gave incorrect shading results. This is solved by having separate PathState for each of the hits. - Path radiance summing wasn't happening correct as well, indirect rays were using wrong path radiance in the case when there were more than one hit recorded. This is now using a bit trickier state machine which calculates path radiance for just SSS (both direct and indirect) and then sums it back to the final radiance. - Previous commit wasn't totally correct either and was an induced bug due to wrong path state left from the "un-happened" ray bounce. There should be no special case happening here, BSSRDFs will be replaced with diffuse ones due to PATH_RAY_DIFFUSE_ANCESTOR flag. - Merged back codebases for "delayed" and "immediate" indirect SSS ray tracing, hopefully making it easier to maintain the codebase. Sure this changes brings memory usage back by about 4-5%, but overall it's still about 2x memory reduction for the experimental kernel here. Thanks Brecht for the review!
2015-11-28 14:30:35 +00:00
state->rng_offset += ss_indirect->num_rays * PRNG_BOUNCE_NUM;
Cycles-Bake: Subsurface Scattering support (fix T40060) This fixes the SSS Direct/Indirect passes as well as the Combined pass. Patch reviewed and with fixes and contributions from Brecht van Lommel. Note: displacement/bump map (related to the report) will be handled separately Reviewers: brecht Differential Revision: https://developer.blender.org/D503
2014-05-07 14:55:09 +00:00
}
Cycles: Fully roll-back to non-delayed SSS indirect rays for CPU There are some issues to be solved with the recent optimization we did for the indirect rays for the SSS. Those issues will take a bit of a time to be fully solved still and we need to unlock Caminandes team now, so let's revert some changes back. CUDA will still use delayed indirect rays since it's an experimental feature. For the details about what's to be done still please refer to T46880.
2015-11-27 12:12:44 +00:00
Cycles: Set of fixes for delayed SSS ray tracing There were multiple issues which are solved now: - It was possible that ray wouldn't be bounced off the BSSRDF, for example when PDF or shader eval is zero. In this case PathState might have been left in pre-bounced state which would have been gave incorrect shading results. This is solved by having separate PathState for each of the hits. - Path radiance summing wasn't happening correct as well, indirect rays were using wrong path radiance in the case when there were more than one hit recorded. This is now using a bit trickier state machine which calculates path radiance for just SSS (both direct and indirect) and then sums it back to the final radiance. - Previous commit wasn't totally correct either and was an induced bug due to wrong path state left from the "un-happened" ray bounce. There should be no special case happening here, BSSRDFs will be replaced with diffuse ones due to PATH_RAY_DIFFUSE_ANCESTOR flag. - Merged back codebases for "delayed" and "immediate" indirect SSS ray tracing, hopefully making it easier to maintain the codebase. Sure this changes brings memory usage back by about 4-5%, but overall it's still about 2x memory reduction for the experimental kernel here. Thanks Brecht for the review!
2015-11-28 14:30:35 +00:00
#endif /* __SUBSURFACE__ */
Cycles-Bake: Subsurface Scattering support (fix T40060) This fixes the SSS Direct/Indirect passes as well as the Combined pass. Patch reviewed and with fixes and contributions from Brecht van Lommel. Note: displacement/bump map (related to the report) will be handled separately Reviewers: brecht Differential Revision: https://developer.blender.org/D503
2014-05-07 14:55:09 +00:00
Cycles: Tweak inline policy for some functions The goal is to make Experimental kernel closer in performance to the official kernel, avoiding spills and such. There should not be big impact on official kernel, own tests showed few percent performance drop on laptop's GPU. CPU was always the same speed on AVX, AVX2 and SSE4.1 CPUs i've been testing here. This seems to be the last essential step before we can get rid of Experimental kernel and enable SSS officially on GPU without causing some major performance issues. Surely some more tweaks are possibly required, but that we can do for until cows go home anyway.
2016-01-14 09:53:05 +00:00
ccl_device_inline float4 kernel_path_integrate(KernelGlobals *kg,
RNG *rng,
int sample,
Ray ray,
ccl_global float *buffer)
Cycles: first step for implementation of non-progressive sampler that handles direct and indirect lighting differently. Rather than picking one light for each point on the path, it now loops over all lights for direct lighting. For indirect lighting it still picks a random light each time. It gives control over the number of AA samples, and the number of Diffuse, Glossy, Transmission, AO, Mesh Light, Background and Lamp samples for each AA sample. This helps tuning render performance/noise and tends to give less noise for renders dominated by direct lighting. This sampling mode only works on the CPU, and still needs proper tile rendering to show progress (will follow tommorrow or so), because each AA sample can be quite slow now and so the delay between each update wil be too long.
2012-06-13 11:44:48 +00:00
{
Cycles: reduce noise using regular path tracing + subsurface scattering with new cubic and gaussian falloff. Like the branched path tracer, this will now shade all intersection points instead of using one at random.
2013-08-24 15:02:08 +00:00
/* 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);
Cycles CUDA: reduce stack memory by reusing ShaderData. 57% less for path and 48% less for branched path.
2016-05-22 20:35:47 +00:00
/* shader data memory used for both volumes and surfaces, saves stack space */
ShaderData sd;
/* shader data used by emission, shadows, volume stacks */
ShaderData emission_sd;
Cycles: reduce noise using regular path tracing + subsurface scattering with new cubic and gaussian falloff. Like the branched path tracer, this will now shade all intersection points instead of using one at random.
2013-08-24 15:02:08 +00:00
PathState state;
Cycles CUDA: reduce stack memory by reusing ShaderData. 57% less for path and 48% less for branched path.
2016-05-22 20:35:47 +00:00
path_state_init(kg, &emission_sd, &state, rng, sample, &ray);
Cycles: reduce noise using regular path tracing + subsurface scattering with new cubic and gaussian falloff. Like the branched path tracer, this will now shade all intersection points instead of using one at random.
2013-08-24 15:02:08 +00:00
Cycles: Add support for debug passes Currently only summed number of traversal steps and intersections used by the camera ray intersection pass is implemented, but in the future we will support more debug passes which would help checking what things makes the scene slow. Example of such extra passes could be number of bounces, time spent on the shader tree evaluation and so. Implementation from the Cycles side is pretty much straightforward, could only mention here that it's a build-time option disabled by default. From the blender side it's implemented as a PASS_DEBUG with several subtypes possible. This way we don't need to create an extra DNA pass type for each of the debug passes, saving us a bits. Reviewers: campbellbarton Reviewed By: campbellbarton Differential Revision: https://developer.blender.org/D813
2014-10-04 13:00:26 +00:00
#ifdef __KERNEL_DEBUG__
DebugData debug_data;
debug_data_init(&debug_data);
#endif
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).
2015-11-22 10:48:33 +00:00
#ifdef __SUBSURFACE__
SubsurfaceIndirectRays ss_indirect;
Cycles: Fix wrong original ray used for SSS baking Also de-duplicated some code by moving to an utility function.
2015-11-28 14:43:54 +00:00
kernel_path_subsurface_init_indirect(&ss_indirect);
Cycles: Trace indirect subsurface rays by restarting the integrator loop This gives much lower stack usage on GPU and reduces kernel memory size to around 448MB on GTX560Ti (comparing to 652MB with previous commit and 946MB with official release). There's also a barely measurable speedup of around 5%, but this is to be confirmed still. At this stage we're using only ~3% for the experimental kernel and SSS rendering seems to be faster by 40% and after some further testing we might consider making SSS and CMJ official features and remove experimental precompiled kernels.
2015-11-22 11:08:03 +00:00
Revert "Cycles: Fix wrong SSS with regular path tracing and clamping enabled" This wasn't really a complete fix and only worked if there was a single scatter event recorded only. Proper fix requires some more thoughts to make it correct without memory use increase. This reverts commit bf9e88bfbebaf5c6228363560970fa526e779c8b.
2015-11-27 11:41:05 +00:00
for(;;) {
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).
2015-11-22 10:48:33 +00:00
#endif
Cycles: first step for implementation of non-progressive sampler that handles direct and indirect lighting differently. Rather than picking one light for each point on the path, it now loops over all lights for direct lighting. For indirect lighting it still picks a random light each time. It gives control over the number of AA samples, and the number of Diffuse, Glossy, Transmission, AO, Mesh Light, Background and Lamp samples for each AA sample. This helps tuning render performance/noise and tends to give less noise for renders dominated by direct lighting. This sampling mode only works on the CPU, and still needs proper tile rendering to show progress (will follow tommorrow or so), because each AA sample can be quite slow now and so the delay between each update wil be too long.
2012-06-13 11:44:48 +00:00
/* path iteration */
Code refactor: move random number and MIS variables into PathState. This makes it easier to pass this state around, and wraps some common RNG dimension computations in utility functions.
2014-01-03 01:48:48 +00:00
for(;;) {
Cycles: first step for implementation of non-progressive sampler that handles direct and indirect lighting differently. Rather than picking one light for each point on the path, it now loops over all lights for direct lighting. For indirect lighting it still picks a random light each time. It gives control over the number of AA samples, and the number of Diffuse, Glossy, Transmission, AO, Mesh Light, Background and Lamp samples for each AA sample. This helps tuning render performance/noise and tends to give less noise for renders dominated by direct lighting. This sampling mode only works on the CPU, and still needs proper tile rendering to show progress (will follow tommorrow or so), because each AA sample can be quite slow now and so the delay between each update wil be too long.
2012-06-13 11:44:48 +00:00
/* intersect scene */
Intersection isect;
uint visibility = path_state_ray_visibility(kg, &state);
Cycles: reduce noise using regular path tracing + subsurface scattering with new cubic and gaussian falloff. Like the branched path tracer, this will now shade all intersection points instead of using one at random.
2013-08-24 15:02:08 +00:00
Cycles OpenCL: fix other build issues when enabling more features.
2013-05-09 15:28:38 +00:00
#ifdef __HAIR__
Cycles: reduce noise using regular path tracing + subsurface scattering with new cubic and gaussian falloff. Like the branched path tracer, this will now shade all intersection points instead of using one at random.
2013-08-24 15:02:08 +00:00
float difl = 0.0f, extmax = 0.0f;
uint lcg_state = 0;
if(kernel_data.bvh.have_curves) {
if((kernel_data.cam.resolution == 1) && (state.flag & PATH_RAY_CAMERA)) {
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;
Code refactor: move random number and MIS variables into PathState. This makes it easier to pass this state around, and wraps some common RNG dimension computations in utility functions.
2014-01-03 01:48:48 +00:00
lcg_state = lcg_state_init(rng, &state, 0x51633e2d);
Cycles: reduce noise using regular path tracing + subsurface scattering with new cubic and gaussian falloff. Like the branched path tracer, this will now shade all intersection points instead of using one at random.
2013-08-24 15:02:08 +00:00
}
bool hit = scene_intersect(kg, &ray, visibility, &isect, &lcg_state, difl, extmax);
Cycles OpenCL: fix other build issues when enabling more features.
2013-05-09 15:28:38 +00:00
#else
Cleanup: Avoid some defines for scene_intersect(), related to Min Width.
2014-09-24 09:32:29 +00:00
bool hit = scene_intersect(kg, &ray, visibility, &isect, NULL, 0.0f, 0.0f);
Cycles OpenCL: fix other build issues when enabling more features.
2013-05-09 15:28:38 +00:00
#endif
Cycles: first step for implementation of non-progressive sampler that handles direct and indirect lighting differently. Rather than picking one light for each point on the path, it now loops over all lights for direct lighting. For indirect lighting it still picks a random light each time. It gives control over the number of AA samples, and the number of Diffuse, Glossy, Transmission, AO, Mesh Light, Background and Lamp samples for each AA sample. This helps tuning render performance/noise and tends to give less noise for renders dominated by direct lighting. This sampling mode only works on the CPU, and still needs proper tile rendering to show progress (will follow tommorrow or so), because each AA sample can be quite slow now and so the delay between each update wil be too long.
2012-06-13 11:44:48 +00:00
Cycles: Add support for debug passes Currently only summed number of traversal steps and intersections used by the camera ray intersection pass is implemented, but in the future we will support more debug passes which would help checking what things makes the scene slow. Example of such extra passes could be number of bounces, time spent on the shader tree evaluation and so. Implementation from the Cycles side is pretty much straightforward, could only mention here that it's a build-time option disabled by default. From the blender side it's implemented as a PASS_DEBUG with several subtypes possible. This way we don't need to create an extra DNA pass type for each of the debug passes, saving us a bits. Reviewers: campbellbarton Reviewed By: campbellbarton Differential Revision: https://developer.blender.org/D813
2014-10-04 13:00:26 +00:00
#ifdef __KERNEL_DEBUG__
if(state.flag & PATH_RAY_CAMERA) {
debug_data.num_bvh_traversal_steps += isect.num_traversal_steps;
Cycles: Add debug pass which shows number of instance pushes during camera ray intersection TODO: We might want to refactor debug passes into PASS_DEBUG and some debug_type (similar to Blender's side passes) to avoid issue of running out of bits.
2015-06-11 22:12:03 +00:00
debug_data.num_bvh_traversed_instances += isect.num_traversed_instances;
Cycles: Add support for debug passes Currently only summed number of traversal steps and intersections used by the camera ray intersection pass is implemented, but in the future we will support more debug passes which would help checking what things makes the scene slow. Example of such extra passes could be number of bounces, time spent on the shader tree evaluation and so. Implementation from the Cycles side is pretty much straightforward, could only mention here that it's a build-time option disabled by default. From the blender side it's implemented as a PASS_DEBUG with several subtypes possible. This way we don't need to create an extra DNA pass type for each of the debug passes, saving us a bits. Reviewers: campbellbarton Reviewed By: campbellbarton Differential Revision: https://developer.blender.org/D813
2014-10-04 13:00:26 +00:00
}
Cycles: Add debug pass showing average number of ray bounces per pixel Quite straightforward implementation, but still needs some work for the split kernel. Includes both regular and split kernel implementation for that. The pass is not exposed to the interface yet because it's currently not really easy to have same pass listed in the menu multiple times.
2015-06-11 08:42:38 +00:00
debug_data.num_ray_bounces++;
Cycles: Add support for debug passes Currently only summed number of traversal steps and intersections used by the camera ray intersection pass is implemented, but in the future we will support more debug passes which would help checking what things makes the scene slow. Example of such extra passes could be number of bounces, time spent on the shader tree evaluation and so. Implementation from the Cycles side is pretty much straightforward, could only mention here that it's a build-time option disabled by default. From the blender side it's implemented as a PASS_DEBUG with several subtypes possible. This way we don't need to create an extra DNA pass type for each of the debug passes, saving us a bits. Reviewers: campbellbarton Reviewed By: campbellbarton Differential Revision: https://developer.blender.org/D813
2014-10-04 13:00:26 +00:00
#endif
Cycles: multiple importance sampling for lamps, which helps reduce noise for big lamps and sharp glossy reflections. This was already supported for mesh lights and the background, so lamps should do it too. This is not for free and it's a bit slower than I hoped even though there is no extra BVH ray intersection. I'll try to optimize it more later. * Area lights look a bit different now, they had the wrong shape before. * Also fixes a sampling issue in the non-progressive integrator. * Only enabled for the CPU, will test on the GPU later. * An option to disable this will be added for situations where it does not help. Same time comparison before/after: http://www.pasteall.org/pic/show.php?id=43313 http://www.pasteall.org/pic/show.php?id=43314
2013-01-09 21:09:20 +00:00
#ifdef __LAMP_MIS__
Cycles: make multiple importance sampling for lamps an option per lamp now, disabled by default for backwards compatibility. http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Integrator
2013-01-30 15:57:15 +00:00
if(kernel_data.integrator.use_lamp_mis && !(state.flag & PATH_RAY_CAMERA)) {
Cycles: multiple importance sampling for lamps, which helps reduce noise for big lamps and sharp glossy reflections. This was already supported for mesh lights and the background, so lamps should do it too. This is not for free and it's a bit slower than I hoped even though there is no extra BVH ray intersection. I'll try to optimize it more later. * Area lights look a bit different now, they had the wrong shape before. * Also fixes a sampling issue in the non-progressive integrator. * Only enabled for the CPU, will test on the GPU later. * An option to disable this will be added for situations where it does not help. Same time comparison before/after: http://www.pasteall.org/pic/show.php?id=43313 http://www.pasteall.org/pic/show.php?id=43314
2013-01-09 21:09:20 +00:00
/* ray starting from previous non-transparent bounce */
Ray light_ray;
Code refactor: move random number and MIS variables into PathState. This makes it easier to pass this state around, and wraps some common RNG dimension computations in utility functions.
2014-01-03 01:48:48 +00:00
light_ray.P = ray.P - state.ray_t*ray.D;
state.ray_t += isect.t;
Cycles: multiple importance sampling for lamps, which helps reduce noise for big lamps and sharp glossy reflections. This was already supported for mesh lights and the background, so lamps should do it too. This is not for free and it's a bit slower than I hoped even though there is no extra BVH ray intersection. I'll try to optimize it more later. * Area lights look a bit different now, they had the wrong shape before. * Also fixes a sampling issue in the non-progressive integrator. * Only enabled for the CPU, will test on the GPU later. * An option to disable this will be added for situations where it does not help. Same time comparison before/after: http://www.pasteall.org/pic/show.php?id=43313 http://www.pasteall.org/pic/show.php?id=43314
2013-01-09 21:09:20 +00:00
light_ray.D = ray.D;
Code refactor: move random number and MIS variables into PathState. This makes it easier to pass this state around, and wraps some common RNG dimension computations in utility functions.
2014-01-03 01:48:48 +00:00
light_ray.t = state.ray_t;
Cycles: multiple importance sampling for lamps, which helps reduce noise for big lamps and sharp glossy reflections. This was already supported for mesh lights and the background, so lamps should do it too. This is not for free and it's a bit slower than I hoped even though there is no extra BVH ray intersection. I'll try to optimize it more later. * Area lights look a bit different now, they had the wrong shape before. * Also fixes a sampling issue in the non-progressive integrator. * Only enabled for the CPU, will test on the GPU later. * An option to disable this will be added for situations where it does not help. Same time comparison before/after: http://www.pasteall.org/pic/show.php?id=43313 http://www.pasteall.org/pic/show.php?id=43314
2013-01-09 21:09:20 +00:00
light_ray.time = ray.time;
Possible fix for #35198: uninitialized memory access with background multiple importance sampling + OSL.
2013-05-03 05:24:05 +00:00
light_ray.dD = ray.dD;
light_ray.dP = ray.dP;
Cycles: multiple importance sampling for lamps, which helps reduce noise for big lamps and sharp glossy reflections. This was already supported for mesh lights and the background, so lamps should do it too. This is not for free and it's a bit slower than I hoped even though there is no extra BVH ray intersection. I'll try to optimize it more later. * Area lights look a bit different now, they had the wrong shape before. * Also fixes a sampling issue in the non-progressive integrator. * Only enabled for the CPU, will test on the GPU later. * An option to disable this will be added for situations where it does not help. Same time comparison before/after: http://www.pasteall.org/pic/show.php?id=43313 http://www.pasteall.org/pic/show.php?id=43314
2013-01-09 21:09:20 +00:00
/* intersect with lamp */
float3 emission;
Cycles CUDA: reduce stack memory by reusing ShaderData. 57% less for path and 48% less for branched path.
2016-05-22 20:35:47 +00:00
if(indirect_lamp_emission(kg, &emission_sd, &state, &light_ray, &emission))
Cycles: reduce noise using regular path tracing + subsurface scattering with new cubic and gaussian falloff. Like the branched path tracer, this will now shade all intersection points instead of using one at random.
2013-08-24 15:02:08 +00:00
path_radiance_accum_emission(&L, throughput, emission, state.bounce);
Cycles: multiple importance sampling for lamps, which helps reduce noise for big lamps and sharp glossy reflections. This was already supported for mesh lights and the background, so lamps should do it too. This is not for free and it's a bit slower than I hoped even though there is no extra BVH ray intersection. I'll try to optimize it more later. * Area lights look a bit different now, they had the wrong shape before. * Also fixes a sampling issue in the non-progressive integrator. * Only enabled for the CPU, will test on the GPU later. * An option to disable this will be added for situations where it does not help. Same time comparison before/after: http://www.pasteall.org/pic/show.php?id=43313 http://www.pasteall.org/pic/show.php?id=43314
2013-01-09 21:09:20 +00:00
}
#endif
Cycles Volume Render: support for rendering of homogeneous volume with absorption. This is the simplest possible volume rendering case, constant density inside the volume and no scattering or emission. My plan is to tweak, verify and commit more volume rendering effects one by one, doing it all at once makes it difficult to verify correctness and track down bugs. Documentation is here: http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Materials/Volume Currently this hooks into path tracing in 3 ways, which should get us pretty far until we add more advanced light sampling. These 3 hooks are repeated in the path tracing, branched path tracing and transparent shadow code: * Determine active volume shader at start of the path * Change active volume shader on transmission through a surface * Light attenuation over line segments between camera, surfaces and background This is work by "storm", Stuart Broadfoot, Thomas Dinges and myself.
2013-12-28 15:56:19 +00:00
#ifdef __VOLUME__
Code refactor: better distinguish scatter and absorption for volume integration.
2013-12-29 14:49:16 +00:00
/* volume attenuation, emission, scatter */
Cycles code refactor: replace magic ~0 values in the code with defines.
2014-03-29 12:03:47 +00:00
if(state.volume_stack[0].shader != SHADER_NONE) {
Cycles Volume Render: heterogeneous (textured) volumes support. Volumes can now have textured colors and density. There is a Volume Sampling panel in the Render properties with these settings: * Step size: distance between volume shader samples when rendering the volume. Lower values give more accurate and detailed results but also increased render time. * Max steps: maximum number of steps through the volume before giving up, to protect from extremely long render times with big objects or small step sizes. This is much more compute intensive than homogeneous volume, so when you are not using a texture you should enable the Homogeneous Volume option in the material or world for faster rendering. One important missing feature is that Generated texture coordinates are not yet working in volumes, and they are the default coordinates for nearly all texture nodes. So until that works you need to plug in object texture coordinates or a world space position. This is work by "storm", Stuart Broadfoot, Thomas Dinges and myself.
2013-12-29 21:19:38 +00:00
Ray volume_ray = ray;
volume_ray.t = (hit)? isect.t: FLT_MAX;
Cycles Volume Render: scattering support. This is done by adding a Volume Scatter node. In many cases you will want to add together a Volume Absorption and Volume Scatter node with the same color and density to get the expected results. This should work with branched path tracing, mixing closures, overlapping volumes, etc. However there's still various optimizations needed for sampling. The main missing thing from the volume branch is the equiangular sampling for homogeneous volumes. The heterogeneous scattering code was arranged such that we can use a single stratified random number for distance sampling, which gives less noise than pseudo random numbers for each step. For volumes where the color is textured there still seems to be something off, needs to be investigated.
2013-12-29 14:40:43 +00:00
Cycles: volume light sampling * Volume multiple importace sampling support to combine equiangular and distance sampling, for both homogeneous and heterogeneous volumes. * Branched path "Sample All Direct Lights" and "Sample All Indirect Lights" now apply to volumes as well as surfaces. Implementation note: For simplicity this is all done with decoupled ray marching, the only case we do not use decoupled is for distance only sampling with one light sample. The homogeneous case should still compile on the GPU because it only requires fixed size storage, but the heterogeneous case will be trickier to get working.
2014-04-04 14:45:49 +00:00
bool heterogeneous = volume_stack_is_heterogeneous(kg, state.volume_stack);
Code refactor: Split __VOLUME__ defines in Cycles. * __VOLUME__ is basic volume support with Emission and Absorption. * __VOLUME_SCATTER__ enables volume Scattering support. * __VOLUME_DECOUPLED__ enables Decoupled Ray Marching.
2014-08-20 21:15:30 +00:00
Cycles: Cleanup, indent nested preprocessor directives Quite straightforward, main trick is happening in path_source_replace_includes(). Reviewers: brecht, dingto, lukasstockner97, juicyfruit Differential Revision: https://developer.blender.org/D1794
2016-02-12 17:33:43 +00:00
# ifdef __VOLUME_DECOUPLED__
Cycles: Avoid call to volume_stack_sampling_method() on GPU, Decoupled is required for Equi-Angular/MIS.
2014-08-24 13:58:41 +00:00
int sampling_method = volume_stack_sampling_method(kg, state.volume_stack);
Cycles: volume sampling method can now be set per material/world. This gives you "Multiple Importance", "Distance" and "Equiangular" choices. What multiple importance sampling does is make things more robust to certain types of noise at the cost of a bit more noise in cases where the individual strategies are always better. So if you've got a pretty dense volume that's lit from far away then distance sampling is usually more efficient. If you've got a light inside or near the volume then equiangular sampling is better. If you have a combination of both, then the multiple importance sampling will be better.
2014-06-07 16:47:14 +00:00
bool decoupled = kernel_volume_use_decoupled(kg, heterogeneous, true, sampling_method);
Cycles Volume Render: scattering support. This is done by adding a Volume Scatter node. In many cases you will want to add together a Volume Absorption and Volume Scatter node with the same color and density to get the expected results. This should work with branched path tracing, mixing closures, overlapping volumes, etc. However there's still various optimizations needed for sampling. The main missing thing from the volume branch is the equiangular sampling for homogeneous volumes. The heterogeneous scattering code was arranged such that we can use a single stratified random number for distance sampling, which gives less noise than pseudo random numbers for each step. For volumes where the color is textured there still seems to be something off, needs to be investigated.
2013-12-29 14:40:43 +00:00
Cycles: volume light sampling * Volume multiple importace sampling support to combine equiangular and distance sampling, for both homogeneous and heterogeneous volumes. * Branched path "Sample All Direct Lights" and "Sample All Indirect Lights" now apply to volumes as well as surfaces. Implementation note: For simplicity this is all done with decoupled ray marching, the only case we do not use decoupled is for distance only sampling with one light sample. The homogeneous case should still compile on the GPU because it only requires fixed size storage, but the heterogeneous case will be trickier to get working.
2014-04-04 14:45:49 +00:00
if(decoupled) {
/* cache steps along volume for repeated sampling */
VolumeSegment volume_segment;
Cycles code refactor: deduplicate and symmetrize some path tracing code.
2014-04-04 12:21:35 +00:00
Cycles CUDA: reduce stack memory by reusing ShaderData. 57% less for path and 48% less for branched path.
2016-05-22 20:35:47 +00:00
shader_setup_from_volume(kg, &sd, &volume_ray);
Cycles: volume light sampling * Volume multiple importace sampling support to combine equiangular and distance sampling, for both homogeneous and heterogeneous volumes. * Branched path "Sample All Direct Lights" and "Sample All Indirect Lights" now apply to volumes as well as surfaces. Implementation note: For simplicity this is all done with decoupled ray marching, the only case we do not use decoupled is for distance only sampling with one light sample. The homogeneous case should still compile on the GPU because it only requires fixed size storage, but the heterogeneous case will be trickier to get working.
2014-04-04 14:45:49 +00:00
kernel_volume_decoupled_record(kg, &state,
Cycles CUDA: reduce stack memory by reusing ShaderData. 57% less for path and 48% less for branched path.
2016-05-22 20:35:47 +00:00
&volume_ray, &sd, &volume_segment, heterogeneous);
Cycles: volume light sampling * Volume multiple importace sampling support to combine equiangular and distance sampling, for both homogeneous and heterogeneous volumes. * Branched path "Sample All Direct Lights" and "Sample All Indirect Lights" now apply to volumes as well as surfaces. Implementation note: For simplicity this is all done with decoupled ray marching, the only case we do not use decoupled is for distance only sampling with one light sample. The homogeneous case should still compile on the GPU because it only requires fixed size storage, but the heterogeneous case will be trickier to get working.
2014-04-04 14:45:49 +00:00
Cycles: volume sampling method can now be set per material/world. This gives you "Multiple Importance", "Distance" and "Equiangular" choices. What multiple importance sampling does is make things more robust to certain types of noise at the cost of a bit more noise in cases where the individual strategies are always better. So if you've got a pretty dense volume that's lit from far away then distance sampling is usually more efficient. If you've got a light inside or near the volume then equiangular sampling is better. If you have a combination of both, then the multiple importance sampling will be better.
2014-06-07 16:47:14 +00:00
volume_segment.sampling_method = sampling_method;
Cycles: volume light sampling * Volume multiple importace sampling support to combine equiangular and distance sampling, for both homogeneous and heterogeneous volumes. * Branched path "Sample All Direct Lights" and "Sample All Indirect Lights" now apply to volumes as well as surfaces. Implementation note: For simplicity this is all done with decoupled ray marching, the only case we do not use decoupled is for distance only sampling with one light sample. The homogeneous case should still compile on the GPU because it only requires fixed size storage, but the heterogeneous case will be trickier to get working.
2014-04-04 14:45:49 +00:00
/* emission */
if(volume_segment.closure_flag & SD_EMISSION)
path_radiance_accum_emission(&L, throughput, volume_segment.accum_emission, state.bounce);
/* scattering */
VolumeIntegrateResult result = VOLUME_PATH_ATTENUATED;
if(volume_segment.closure_flag & SD_SCATTER) {
Workaround for T47213: branched path sampling issues with CUDA 7.5.
2016-02-18 23:39:00 +00:00
int all = false;
Cycles: volume light sampling * Volume multiple importace sampling support to combine equiangular and distance sampling, for both homogeneous and heterogeneous volumes. * Branched path "Sample All Direct Lights" and "Sample All Indirect Lights" now apply to volumes as well as surfaces. Implementation note: For simplicity this is all done with decoupled ray marching, the only case we do not use decoupled is for distance only sampling with one light sample. The homogeneous case should still compile on the GPU because it only requires fixed size storage, but the heterogeneous case will be trickier to get working.
2014-04-04 14:45:49 +00:00
/* direct light sampling */
Cycles CUDA: reduce stack memory by reusing ShaderData. 57% less for path and 48% less for branched path.
2016-05-22 20:35:47 +00:00
kernel_branched_path_volume_connect_light(kg, rng, &sd,
&emission_sd, throughput, &state, &L, all,
&volume_ray, &volume_segment);
Cycles: volume light sampling * Volume multiple importace sampling support to combine equiangular and distance sampling, for both homogeneous and heterogeneous volumes. * Branched path "Sample All Direct Lights" and "Sample All Indirect Lights" now apply to volumes as well as surfaces. Implementation note: For simplicity this is all done with decoupled ray marching, the only case we do not use decoupled is for distance only sampling with one light sample. The homogeneous case should still compile on the GPU because it only requires fixed size storage, but the heterogeneous case will be trickier to get working.
2014-04-04 14:45:49 +00:00
/* indirect 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, rng, &state, PRNG_PHASE);
float rscatter = path_state_rng_1D_for_decision(kg, rng, &state, PRNG_SCATTER_DISTANCE);
result = kernel_volume_decoupled_scatter(kg,
Cycles CUDA: reduce stack memory by reusing ShaderData. 57% less for path and 48% less for branched path.
2016-05-22 20:35:47 +00:00
&state, &volume_ray, &sd, &throughput,
Cycles: volume light sampling * Volume multiple importace sampling support to combine equiangular and distance sampling, for both homogeneous and heterogeneous volumes. * Branched path "Sample All Direct Lights" and "Sample All Indirect Lights" now apply to volumes as well as surfaces. Implementation note: For simplicity this is all done with decoupled ray marching, the only case we do not use decoupled is for distance only sampling with one light sample. The homogeneous case should still compile on the GPU because it only requires fixed size storage, but the heterogeneous case will be trickier to get working.
2014-04-04 14:45:49 +00:00
rphase, rscatter, &volume_segment, NULL, true);
}
/* free cached steps */
kernel_volume_decoupled_free(kg, &volume_segment);
if(result == VOLUME_PATH_SCATTERED) {
Cycles CUDA: reduce stack memory by reusing ShaderData. 57% less for path and 48% less for branched path.
2016-05-22 20:35:47 +00:00
if(kernel_path_volume_bounce(kg, rng, &sd, &throughput, &state, &L, &ray))
Cycles: volume light sampling * Volume multiple importace sampling support to combine equiangular and distance sampling, for both homogeneous and heterogeneous volumes. * Branched path "Sample All Direct Lights" and "Sample All Indirect Lights" now apply to volumes as well as surfaces. Implementation note: For simplicity this is all done with decoupled ray marching, the only case we do not use decoupled is for distance only sampling with one light sample. The homogeneous case should still compile on the GPU because it only requires fixed size storage, but the heterogeneous case will be trickier to get working.
2014-04-04 14:45:49 +00:00
continue;
else
break;
}
Cycles: Small improvement for volume render (decoupled) Simplify branching here a bit, helps ~3% in volume_light_sampling.blend (Branched MIS scene).
2015-02-14 19:44:30 +00:00
else {
throughput *= volume_segment.accum_transmittance;
}
Cycles: volume light sampling * Volume multiple importace sampling support to combine equiangular and distance sampling, for both homogeneous and heterogeneous volumes. * Branched path "Sample All Direct Lights" and "Sample All Indirect Lights" now apply to volumes as well as surfaces. Implementation note: For simplicity this is all done with decoupled ray marching, the only case we do not use decoupled is for distance only sampling with one light sample. The homogeneous case should still compile on the GPU because it only requires fixed size storage, but the heterogeneous case will be trickier to get working.
2014-04-04 14:45:49 +00:00
}
Cycles: Cleanup, indent nested preprocessor directives Quite straightforward, main trick is happening in path_source_replace_includes(). Reviewers: brecht, dingto, lukasstockner97, juicyfruit Differential Revision: https://developer.blender.org/D1794
2016-02-12 17:33:43 +00:00
else
# endif
Code refactor: Split __VOLUME__ defines in Cycles. * __VOLUME__ is basic volume support with Emission and Absorption. * __VOLUME_SCATTER__ enables volume Scattering support. * __VOLUME_DECOUPLED__ enables Decoupled Ray Marching.
2014-08-20 21:15:30 +00:00
{
Cycles: volume light sampling * Volume multiple importace sampling support to combine equiangular and distance sampling, for both homogeneous and heterogeneous volumes. * Branched path "Sample All Direct Lights" and "Sample All Indirect Lights" now apply to volumes as well as surfaces. Implementation note: For simplicity this is all done with decoupled ray marching, the only case we do not use decoupled is for distance only sampling with one light sample. The homogeneous case should still compile on the GPU because it only requires fixed size storage, but the heterogeneous case will be trickier to get working.
2014-04-04 14:45:49 +00:00
/* integrate along volume segment with distance sampling */
VolumeIntegrateResult result = kernel_volume_integrate(
Cycles CUDA: reduce stack memory by reusing ShaderData. 57% less for path and 48% less for branched path.
2016-05-22 20:35:47 +00:00
kg, &state, &sd, &volume_ray, &L, &throughput, rng, heterogeneous);
Cycles: volume light sampling * Volume multiple importace sampling support to combine equiangular and distance sampling, for both homogeneous and heterogeneous volumes. * Branched path "Sample All Direct Lights" and "Sample All Indirect Lights" now apply to volumes as well as surfaces. Implementation note: For simplicity this is all done with decoupled ray marching, the only case we do not use decoupled is for distance only sampling with one light sample. The homogeneous case should still compile on the GPU because it only requires fixed size storage, but the heterogeneous case will be trickier to get working.
2014-04-04 14:45:49 +00:00
Cycles: Cleanup, indent nested preprocessor directives Quite straightforward, main trick is happening in path_source_replace_includes(). Reviewers: brecht, dingto, lukasstockner97, juicyfruit Differential Revision: https://developer.blender.org/D1794
2016-02-12 17:33:43 +00:00
# ifdef __VOLUME_SCATTER__
Cycles: volume light sampling * Volume multiple importace sampling support to combine equiangular and distance sampling, for both homogeneous and heterogeneous volumes. * Branched path "Sample All Direct Lights" and "Sample All Indirect Lights" now apply to volumes as well as surfaces. Implementation note: For simplicity this is all done with decoupled ray marching, the only case we do not use decoupled is for distance only sampling with one light sample. The homogeneous case should still compile on the GPU because it only requires fixed size storage, but the heterogeneous case will be trickier to get working.
2014-04-04 14:45:49 +00:00
if(result == VOLUME_PATH_SCATTERED) {
/* direct lighting */
Cycles CUDA: reduce stack memory by reusing ShaderData. 57% less for path and 48% less for branched path.
2016-05-22 20:35:47 +00:00
kernel_path_volume_connect_light(kg, rng, &sd, &emission_sd, throughput, &state, &L);
Cycles: volume light sampling * Volume multiple importace sampling support to combine equiangular and distance sampling, for both homogeneous and heterogeneous volumes. * Branched path "Sample All Direct Lights" and "Sample All Indirect Lights" now apply to volumes as well as surfaces. Implementation note: For simplicity this is all done with decoupled ray marching, the only case we do not use decoupled is for distance only sampling with one light sample. The homogeneous case should still compile on the GPU because it only requires fixed size storage, but the heterogeneous case will be trickier to get working.
2014-04-04 14:45:49 +00:00
/* indirect light bounce */
Cycles CUDA: reduce stack memory by reusing ShaderData. 57% less for path and 48% less for branched path.
2016-05-22 20:35:47 +00:00
if(kernel_path_volume_bounce(kg, rng, &sd, &throughput, &state, &L, &ray))
Cycles: volume light sampling * Volume multiple importace sampling support to combine equiangular and distance sampling, for both homogeneous and heterogeneous volumes. * Branched path "Sample All Direct Lights" and "Sample All Indirect Lights" now apply to volumes as well as surfaces. Implementation note: For simplicity this is all done with decoupled ray marching, the only case we do not use decoupled is for distance only sampling with one light sample. The homogeneous case should still compile on the GPU because it only requires fixed size storage, but the heterogeneous case will be trickier to get working.
2014-04-04 14:45:49 +00:00
continue;
else
break;
}
Cycles: Cleanup, indent nested preprocessor directives Quite straightforward, main trick is happening in path_source_replace_includes(). Reviewers: brecht, dingto, lukasstockner97, juicyfruit Differential Revision: https://developer.blender.org/D1794
2016-02-12 17:33:43 +00:00
# endif
Cycles Volume Render: scattering support. This is done by adding a Volume Scatter node. In many cases you will want to add together a Volume Absorption and Volume Scatter node with the same color and density to get the expected results. This should work with branched path tracing, mixing closures, overlapping volumes, etc. However there's still various optimizations needed for sampling. The main missing thing from the volume branch is the equiangular sampling for homogeneous volumes. The heterogeneous scattering code was arranged such that we can use a single stratified random number for distance sampling, which gives less noise than pseudo random numbers for each step. For volumes where the color is textured there still seems to be something off, needs to be investigated.
2013-12-29 14:40:43 +00:00
}
Cycles Volume Render: support for rendering of homogeneous volume with absorption. This is the simplest possible volume rendering case, constant density inside the volume and no scattering or emission. My plan is to tweak, verify and commit more volume rendering effects one by one, doing it all at once makes it difficult to verify correctness and track down bugs. Documentation is here: http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Materials/Volume Currently this hooks into path tracing in 3 ways, which should get us pretty far until we add more advanced light sampling. These 3 hooks are repeated in the path tracing, branched path tracing and transparent shadow code: * Determine active volume shader at start of the path * Change active volume shader on transmission through a surface * Light attenuation over line segments between camera, surfaces and background This is work by "storm", Stuart Broadfoot, Thomas Dinges and myself.
2013-12-28 15:56:19 +00:00
}
#endif
Cycles: multiple importance sampling for lamps, which helps reduce noise for big lamps and sharp glossy reflections. This was already supported for mesh lights and the background, so lamps should do it too. This is not for free and it's a bit slower than I hoped even though there is no extra BVH ray intersection. I'll try to optimize it more later. * Area lights look a bit different now, they had the wrong shape before. * Also fixes a sampling issue in the non-progressive integrator. * Only enabled for the CPU, will test on the GPU later. * An option to disable this will be added for situations where it does not help. Same time comparison before/after: http://www.pasteall.org/pic/show.php?id=43313 http://www.pasteall.org/pic/show.php?id=43314
2013-01-09 21:09:20 +00:00
if(!hit) {
Cycles: reduce noise using regular path tracing + subsurface scattering with new cubic and gaussian falloff. Like the branched path tracer, this will now shade all intersection points instead of using one at random.
2013-08-24 15:02:08 +00:00
/* eval background shader if nothing hit */
if(kernel_data.background.transparent && (state.flag & PATH_RAY_CAMERA)) {
L_transparent += average(throughput);
#ifdef __PASSES__
if(!(kernel_data.film.pass_flag & PASS_BACKGROUND))
#endif
break;
}
Cycles: first step for implementation of non-progressive sampler that handles direct and indirect lighting differently. Rather than picking one light for each point on the path, it now loops over all lights for direct lighting. For indirect lighting it still picks a random light each time. It gives control over the number of AA samples, and the number of Diffuse, Glossy, Transmission, AO, Mesh Light, Background and Lamp samples for each AA sample. This helps tuning render performance/noise and tends to give less noise for renders dominated by direct lighting. This sampling mode only works on the CPU, and still needs proper tile rendering to show progress (will follow tommorrow or so), because each AA sample can be quite slow now and so the delay between each update wil be too long.
2012-06-13 11:44:48 +00:00
#ifdef __BACKGROUND__
/* sample background shader */
Cycles CUDA: reduce stack memory by reusing ShaderData. 57% less for path and 48% less for branched path.
2016-05-22 20:35:47 +00:00
float3 L_background = indirect_background(kg, &emission_sd, &state, &ray);
Cycles: reduce noise using regular path tracing + subsurface scattering with new cubic and gaussian falloff. Like the branched path tracer, this will now shade all intersection points instead of using one at random.
2013-08-24 15:02:08 +00:00
path_radiance_accum_background(&L, throughput, L_background, state.bounce);
Cycles: first step for implementation of non-progressive sampler that handles direct and indirect lighting differently. Rather than picking one light for each point on the path, it now loops over all lights for direct lighting. For indirect lighting it still picks a random light each time. It gives control over the number of AA samples, and the number of Diffuse, Glossy, Transmission, AO, Mesh Light, Background and Lamp samples for each AA sample. This helps tuning render performance/noise and tends to give less noise for renders dominated by direct lighting. This sampling mode only works on the CPU, and still needs proper tile rendering to show progress (will follow tommorrow or so), because each AA sample can be quite slow now and so the delay between each update wil be too long.
2012-06-13 11:44:48 +00:00
#endif
break;
}
/* setup shading */
Cycles: Refactor how we pass bounce info to light path node. This commit changes the way how we pass bounce information to the Light Path node. Instead of manualy copying the bounces into ShaderData, we now directly pass PathState. This reduces the arguments that we need to pass around and also makes it easier to extend the feature. This commit also exposes the Transmission Bounce Depth to the Light Path node. It works similar to the Transparent Depth Output: Replace a Transmission lightpath after X bounces with another shader, e.g a Diffuse one. This can be used to avoid black surfaces, due to low amount of max bounces. Reviewed by Sergey and Brecht, thanks for some hlp with this. I tested compilation and usage on CPU (SVM and OSL), CUDA, OpenCL Split and Mega kernel. Hopefully this covers all devices. :)
2016-01-06 22:38:13 +00:00
shader_setup_from_ray(kg, &sd, &isect, &ray);
Fix T39804: cycles smoke domain visible in rendering. Transparent objects could become subtly visible by the different sampling patterns for pixels covered and not covered by the object. It still converged to the right solution but that can take a while. Now we try to use the same sampling pattern here.
2014-05-29 12:12:12 +00:00
float rbsdf = path_state_rng_1D_for_decision(kg, rng, &state, PRNG_BSDF);
Cycles: Add multi-scattering, energy-conserving GGX as an option to the Glossy, Anisotropic and Glass BSDFs This commit adds a new distribution to the Glossy, Anisotropic and Glass BSDFs that implements the multiple-scattering microfacet model described in the paper "Multiple-Scattering Microfacet BSDFs with the Smith Model". Essentially, the improvement is that unlike classical GGX, which only models single scattering and assumes the contribution of multiple bounces to be zero, this new model performs a random walk on the microsurface until the ray leaves it again, which ensures perfect energy conservation. In practise, this means that the "darkening problem" - GGX materials becoming darker with increasing roughness - is solved in a physically correct and efficient way. The downside of this model is that it has no (known) analytic expression for evalation. However, it can be evaluated stochastically, and although the correct PDF isn't known either, the properties of MIS and the balance heuristic guarantee an unbiased result at the cost of slightly higher noise. Reviewers: dingto, #cycles, brecht Reviewed By: dingto, #cycles, brecht Subscribers: bliblubli, ace_dragon, gregzaal, brecht, harvester, dingto, marcog, swerner, jtheninja, Blendify, nutel Differential Revision: https://developer.blender.org/D2002
2016-06-23 20:56:43 +00:00
shader_eval_surface(kg, &sd, rng, &state, rbsdf, state.flag, SHADER_CONTEXT_MAIN);
Cycles: reduce noise using regular path tracing + subsurface scattering with new cubic and gaussian falloff. Like the branched path tracer, this will now shade all intersection points instead of using one at random.
2013-08-24 15:02:08 +00:00
/* holdout */
#ifdef __HOLDOUT__
if((sd.flag & (SD_HOLDOUT|SD_HOLDOUT_MASK)) && (state.flag & PATH_RAY_CAMERA)) {
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;
}
#endif
/* holdout mask objects do not write data passes */
Cycles: add pass alpha threshold value to render layers. Z, Index, normal, UV and vector passes are only affected by surfaces with alpha transparency equal to or higher than this threshold. With value 0.0 the first surface hit will always write to these passes, regardless of transparency. With higher values surfaces that are mostly transparent can be skipped until an opaque surface is encountered.
2014-02-06 14:18:34 +00:00
kernel_write_data_passes(kg, buffer, &L, &sd, sample, &state, throughput);
Cycles: first step for implementation of non-progressive sampler that handles direct and indirect lighting differently. Rather than picking one light for each point on the path, it now loops over all lights for direct lighting. For indirect lighting it still picks a random light each time. It gives control over the number of AA samples, and the number of Diffuse, Glossy, Transmission, AO, Mesh Light, Background and Lamp samples for each AA sample. This helps tuning render performance/noise and tends to give less noise for renders dominated by direct lighting. This sampling mode only works on the CPU, and still needs proper tile rendering to show progress (will follow tommorrow or so), because each AA sample can be quite slow now and so the delay between each update wil be too long.
2012-06-13 11:44:48 +00:00
/* blurring of bsdf after bounces, for rays that have a small likelihood
* of following this particular path (diffuse, rough glossy) */
if(kernel_data.integrator.filter_glossy != FLT_MAX) {
Code refactor: move random number and MIS variables into PathState. This makes it easier to pass this state around, and wraps some common RNG dimension computations in utility functions.
2014-01-03 01:48:48 +00:00
float blur_pdf = kernel_data.integrator.filter_glossy*state.min_ray_pdf;
Cycles: first step for implementation of non-progressive sampler that handles direct and indirect lighting differently. Rather than picking one light for each point on the path, it now loops over all lights for direct lighting. For indirect lighting it still picks a random light each time. It gives control over the number of AA samples, and the number of Diffuse, Glossy, Transmission, AO, Mesh Light, Background and Lamp samples for each AA sample. This helps tuning render performance/noise and tends to give less noise for renders dominated by direct lighting. This sampling mode only works on the CPU, and still needs proper tile rendering to show progress (will follow tommorrow or so), because each AA sample can be quite slow now and so the delay between each update wil be too long.
2012-06-13 11:44:48 +00:00
if(blur_pdf < 1.0f) {
float blur_roughness = sqrtf(1.0f - blur_pdf)*0.5f;
shader_bsdf_blur(kg, &sd, blur_roughness);
}
}
#ifdef __EMISSION__
/* emission */
if(sd.flag & SD_EMISSION) {
Cycles: reduce noise using regular path tracing + subsurface scattering with new cubic and gaussian falloff. Like the branched path tracer, this will now shade all intersection points instead of using one at random.
2013-08-24 15:02:08 +00:00
/* todo: is isect.t wrong here for transparent surfaces? */
Code refactor: move random number and MIS variables into PathState. This makes it easier to pass this state around, and wraps some common RNG dimension computations in utility functions.
2014-01-03 01:48:48 +00:00
float3 emission = indirect_primitive_emission(kg, &sd, isect.t, state.flag, state.ray_pdf);
Cycles: reduce noise using regular path tracing + subsurface scattering with new cubic and gaussian falloff. Like the branched path tracer, this will now shade all intersection points instead of using one at random.
2013-08-24 15:02:08 +00:00
path_radiance_accum_emission(&L, throughput, emission, state.bounce);
Cycles: first step for implementation of non-progressive sampler that handles direct and indirect lighting differently. Rather than picking one light for each point on the path, it now loops over all lights for direct lighting. For indirect lighting it still picks a random light each time. It gives control over the number of AA samples, and the number of Diffuse, Glossy, Transmission, AO, Mesh Light, Background and Lamp samples for each AA sample. This helps tuning render performance/noise and tends to give less noise for renders dominated by direct lighting. This sampling mode only works on the CPU, and still needs proper tile rendering to show progress (will follow tommorrow or so), because each AA sample can be quite slow now and so the delay between each update wil be too long.
2012-06-13 11:44:48 +00:00
}
#endif
/* 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 */
Cycles: reduce noise using regular path tracing + subsurface scattering with new cubic and gaussian falloff. Like the branched path tracer, this will now shade all intersection points instead of using one at random.
2013-08-24 15:02:08 +00:00
float probability = path_state_terminate_probability(kg, &state, throughput);
Cycles: first step for implementation of non-progressive sampler that handles direct and indirect lighting differently. Rather than picking one light for each point on the path, it now loops over all lights for direct lighting. For indirect lighting it still picks a random light each time. It gives control over the number of AA samples, and the number of Diffuse, Glossy, Transmission, AO, Mesh Light, Background and Lamp samples for each AA sample. This helps tuning render performance/noise and tends to give less noise for renders dominated by direct lighting. This sampling mode only works on the CPU, and still needs proper tile rendering to show progress (will follow tommorrow or so), because each AA sample can be quite slow now and so the delay between each update wil be too long.
2012-06-13 11:44:48 +00:00
Cycles: experimental correlated multi-jittered sampling pattern that can be used instead of sobol. So far one doesn't seem to be consistently better or worse than the other for the same number of samples but more testing is needed. The random number generator itself is slower than sobol for most number of samples, except 16, 64, 256, .. because they can be computed faster. This can probably be optimized, but we can do that when/if this actually turns out to be useful. Paper this implementation is based on: http://graphics.pixar.com/library/MultiJitteredSampling/ Also includes some refactoring of RNG code, fixing a Sobol correlation issue with the first BSDF and < 16 samples, skipping some unneeded RNG calls and using a simpler unit square to unit disk function.
2013-06-07 16:06:22 +00:00
if(probability == 0.0f) {
Cycles: first step for implementation of non-progressive sampler that handles direct and indirect lighting differently. Rather than picking one light for each point on the path, it now loops over all lights for direct lighting. For indirect lighting it still picks a random light each time. It gives control over the number of AA samples, and the number of Diffuse, Glossy, Transmission, AO, Mesh Light, Background and Lamp samples for each AA sample. This helps tuning render performance/noise and tends to give less noise for renders dominated by direct lighting. This sampling mode only works on the CPU, and still needs proper tile rendering to show progress (will follow tommorrow or so), because each AA sample can be quite slow now and so the delay between each update wil be too long.
2012-06-13 11:44:48 +00:00
break;
Cycles: experimental correlated multi-jittered sampling pattern that can be used instead of sobol. So far one doesn't seem to be consistently better or worse than the other for the same number of samples but more testing is needed. The random number generator itself is slower than sobol for most number of samples, except 16, 64, 256, .. because they can be computed faster. This can probably be optimized, but we can do that when/if this actually turns out to be useful. Paper this implementation is based on: http://graphics.pixar.com/library/MultiJitteredSampling/ Also includes some refactoring of RNG code, fixing a Sobol correlation issue with the first BSDF and < 16 samples, skipping some unneeded RNG calls and using a simpler unit square to unit disk function.
2013-06-07 16:06:22 +00:00
}
else if(probability != 1.0f) {
Fix T39804: cycles smoke domain visible in rendering. Transparent objects could become subtly visible by the different sampling patterns for pixels covered and not covered by the object. It still converged to the right solution but that can take a while. Now we try to use the same sampling pattern here.
2014-05-29 12:12:12 +00:00
float terminate = path_state_rng_1D_for_decision(kg, rng, &state, PRNG_TERMINATE);
Cycles: experimental correlated multi-jittered sampling pattern that can be used instead of sobol. So far one doesn't seem to be consistently better or worse than the other for the same number of samples but more testing is needed. The random number generator itself is slower than sobol for most number of samples, except 16, 64, 256, .. because they can be computed faster. This can probably be optimized, but we can do that when/if this actually turns out to be useful. Paper this implementation is based on: http://graphics.pixar.com/library/MultiJitteredSampling/ Also includes some refactoring of RNG code, fixing a Sobol correlation issue with the first BSDF and < 16 samples, skipping some unneeded RNG calls and using a simpler unit square to unit disk function.
2013-06-07 16:06:22 +00:00
if(terminate >= probability)
break;
Cycles: first step for implementation of non-progressive sampler that handles direct and indirect lighting differently. Rather than picking one light for each point on the path, it now loops over all lights for direct lighting. For indirect lighting it still picks a random light each time. It gives control over the number of AA samples, and the number of Diffuse, Glossy, Transmission, AO, Mesh Light, Background and Lamp samples for each AA sample. This helps tuning render performance/noise and tends to give less noise for renders dominated by direct lighting. This sampling mode only works on the CPU, and still needs proper tile rendering to show progress (will follow tommorrow or so), because each AA sample can be quite slow now and so the delay between each update wil be too long.
2012-06-13 11:44:48 +00:00
Cycles: experimental correlated multi-jittered sampling pattern that can be used instead of sobol. So far one doesn't seem to be consistently better or worse than the other for the same number of samples but more testing is needed. The random number generator itself is slower than sobol for most number of samples, except 16, 64, 256, .. because they can be computed faster. This can probably be optimized, but we can do that when/if this actually turns out to be useful. Paper this implementation is based on: http://graphics.pixar.com/library/MultiJitteredSampling/ Also includes some refactoring of RNG code, fixing a Sobol correlation issue with the first BSDF and < 16 samples, skipping some unneeded RNG calls and using a simpler unit square to unit disk function.
2013-06-07 16:06:22 +00:00
throughput /= probability;
}
Cycles: first step for implementation of non-progressive sampler that handles direct and indirect lighting differently. Rather than picking one light for each point on the path, it now loops over all lights for direct lighting. For indirect lighting it still picks a random light each time. It gives control over the number of AA samples, and the number of Diffuse, Glossy, Transmission, AO, Mesh Light, Background and Lamp samples for each AA sample. This helps tuning render performance/noise and tends to give less noise for renders dominated by direct lighting. This sampling mode only works on the CPU, and still needs proper tile rendering to show progress (will follow tommorrow or so), because each AA sample can be quite slow now and so the delay between each update wil be too long.
2012-06-13 11:44:48 +00:00
#ifdef __AO__
/* ambient occlusion */
Cycles: 4 new nodes. * Tangent: generate a tangent direction for anisotropic shading. Can be either radial around X/Y/Z axis, or from a UV map. The default tangent for the anisotropic BSDF and geometry node is now always radial Z, for UV tangent use this node now. http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Nodes/More#Tangent * Normal Map: generate a perturbed normal from an RGB normal map image. This is usually chained with an Image Texture node in the color input, to specify the normal map image. For tangent space normal maps, the UV coordinates for the image must match, and the image texture should be set to Non-Color mode to give correct results. http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Nodes/More#Normal_Map * Refraction BSDF: for best results this node should be considered as a building block and not be used on its own, but rather mixed with a glossy node using a fresnel type factor. Otherwise it will give quite dark results at the edges for glossy refraction. http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Nodes/Shaders#Refraction * Ambient Occlusion: controls the amount of AO a surface receives, rather than having just a global factor in the world. Note that this outputs a shader and not a color, that's for another time. http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Nodes/Shaders#Ambient_Occlusion
2012-11-06 19:59:02 +00:00
if(kernel_data.integrator.use_ambient_occlusion || (sd.flag & SD_AO)) {
Cycles CUDA: reduce stack memory by reusing ShaderData. 57% less for path and 48% less for branched path.
2016-05-22 20:35:47 +00:00
kernel_path_ao(kg, &sd, &emission_sd, &L, &state, rng, throughput);
Cycles: first step for implementation of non-progressive sampler that handles direct and indirect lighting differently. Rather than picking one light for each point on the path, it now loops over all lights for direct lighting. For indirect lighting it still picks a random light each time. It gives control over the number of AA samples, and the number of Diffuse, Glossy, Transmission, AO, Mesh Light, Background and Lamp samples for each AA sample. This helps tuning render performance/noise and tends to give less noise for renders dominated by direct lighting. This sampling mode only works on the CPU, and still needs proper tile rendering to show progress (will follow tommorrow or so), because each AA sample can be quite slow now and so the delay between each update wil be too long.
2012-06-13 11:44:48 +00:00
}
Cycles: * Fix missing update when editing objects with emission materials. * Fix preview pass rendering set to 1 not showing full resolution. * Fix CUDA runtime compiling failing due to missing cache directory. * Use settings from first render layer for visibility and material override. And a bunch of incomplete and still disabled code mostly related to closure sampling.
2011-09-12 13:13:56 +00:00
#endif
Cycles: first step for implementation of non-progressive sampler that handles direct and indirect lighting differently. Rather than picking one light for each point on the path, it now loops over all lights for direct lighting. For indirect lighting it still picks a random light each time. It gives control over the number of AA samples, and the number of Diffuse, Glossy, Transmission, AO, Mesh Light, Background and Lamp samples for each AA sample. This helps tuning render performance/noise and tends to give less noise for renders dominated by direct lighting. This sampling mode only works on the CPU, and still needs proper tile rendering to show progress (will follow tommorrow or so), because each AA sample can be quite slow now and so the delay between each update wil be too long.
2012-06-13 11:44:48 +00:00
Cycles: reduce noise using regular path tracing + subsurface scattering with new cubic and gaussian falloff. Like the branched path tracer, this will now shade all intersection points instead of using one at random.
2013-08-24 15:02:08 +00:00
#ifdef __SUBSURFACE__
/* bssrdf scatter to a different location on the same object, replacing
* the closures with a diffuse BSDF */
if(sd.flag & SD_BSSRDF) {
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).
2015-11-22 10:48:33 +00:00
if(kernel_path_subsurface_scatter(kg,
&sd,
Cycles CUDA: reduce stack memory by reusing ShaderData. 57% less for path and 48% less for branched path.
2016-05-22 20:35:47 +00:00
&emission_sd,
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).
2015-11-22 10:48:33 +00:00
&L,
&state,
rng,
&ray,
&throughput,
&ss_indirect))
{
Cycles / SSS: * Remove the compatible falloff SSS implementation. We shouldn't support two implementations in the long term, and 2.7x is a good release number do break some compatibility as well. * Version patch added, so Files with Compatible falloff will automatically use Cubic now. It was already mentioned in the manual, that Compatible is deprecated. http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Nodes/Shaders#BSSRDF
2013-10-08 17:07:18 +00:00
break;
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).
2015-11-22 10:48:33 +00:00
}
Cycles: reduce noise using regular path tracing + subsurface scattering with new cubic and gaussian falloff. Like the branched path tracer, this will now shade all intersection points instead of using one at random.
2013-08-24 15:02:08 +00:00
}
Cycles: Fully roll-back to non-delayed SSS indirect rays for CPU There are some issues to be solved with the recent optimization we did for the indirect rays for the SSS. Those issues will take a bit of a time to be fully solved still and we need to unlock Caminandes team now, so let's revert some changes back. CUDA will still use delayed indirect rays since it's an experimental feature. For the details about what's to be done still please refer to T46880.
2015-11-27 12:12:44 +00:00
#endif /* __SUBSURFACE__ */
Cycles Volume Render: support for rendering of homogeneous volume with absorption. This is the simplest possible volume rendering case, constant density inside the volume and no scattering or emission. My plan is to tweak, verify and commit more volume rendering effects one by one, doing it all at once makes it difficult to verify correctness and track down bugs. Documentation is here: http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Materials/Volume Currently this hooks into path tracing in 3 ways, which should get us pretty far until we add more advanced light sampling. These 3 hooks are repeated in the path tracing, branched path tracing and transparent shadow code: * Determine active volume shader at start of the path * Change active volume shader on transmission through a surface * Light attenuation over line segments between camera, surfaces and background This is work by "storm", Stuart Broadfoot, Thomas Dinges and myself.
2013-12-28 15:56:19 +00:00
Cycles code refactor: deduplicate and symmetrize some path tracing code.
2014-04-04 12:21:35 +00:00
/* direct lighting */
Cycles CUDA: reduce stack memory by reusing ShaderData. 57% less for path and 48% less for branched path.
2016-05-22 20:35:47 +00:00
kernel_path_surface_connect_light(kg, rng, &sd, &emission_sd, throughput, &state, &L);
Cycles Volume Render: support for rendering of homogeneous volume with absorption. This is the simplest possible volume rendering case, constant density inside the volume and no scattering or emission. My plan is to tweak, verify and commit more volume rendering effects one by one, doing it all at once makes it difficult to verify correctness and track down bugs. Documentation is here: http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Materials/Volume Currently this hooks into path tracing in 3 ways, which should get us pretty far until we add more advanced light sampling. These 3 hooks are repeated in the path tracing, branched path tracing and transparent shadow code: * Determine active volume shader at start of the path * Change active volume shader on transmission through a surface * Light attenuation over line segments between camera, surfaces and background This is work by "storm", Stuart Broadfoot, Thomas Dinges and myself.
2013-12-28 15:56:19 +00:00
Cycles code refactor: deduplicate and symmetrize some path tracing code.
2014-04-04 12:21:35 +00:00
/* compute direct lighting and next bounce */
if(!kernel_path_surface_bounce(kg, rng, &sd, &throughput, &state, &L, &ray))
Cycles Volume Render: support for rendering of homogeneous volume with absorption. This is the simplest possible volume rendering case, constant density inside the volume and no scattering or emission. My plan is to tweak, verify and commit more volume rendering effects one by one, doing it all at once makes it difficult to verify correctness and track down bugs. Documentation is here: http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Materials/Volume Currently this hooks into path tracing in 3 ways, which should get us pretty far until we add more advanced light sampling. These 3 hooks are repeated in the path tracing, branched path tracing and transparent shadow code: * Determine active volume shader at start of the path * Change active volume shader on transmission through a surface * Light attenuation over line segments between camera, surfaces and background This is work by "storm", Stuart Broadfoot, Thomas Dinges and myself.
2013-12-28 15:56:19 +00:00
break;
Cycles render engine, initial commit. This is the engine itself, blender modifications and build instructions will follow later. Cycles uses code from some great open source projects, many thanks them: * BVH building and traversal code from NVidia's "Understanding the Efficiency of Ray Traversal on GPUs": http://code.google.com/p/understanding-the-efficiency-of-ray-traversal-on-gpus/ * Open Shading Language for a large part of the shading system: http://code.google.com/p/openshadinglanguage/ * Blender for procedural textures and a few other nodes. * Approximate Catmull Clark subdivision from NVidia Mesh tools: http://code.google.com/p/nvidia-mesh-tools/ * Sobol direction vectors from: http://web.maths.unsw.edu.au/~fkuo/sobol/ * Film response functions from: http://www.cs.columbia.edu/CAVE/software/softlib/dorf.php
2011-04-27 11:58:34 +00:00
}
Cycles: reduce noise using regular path tracing + subsurface scattering with new cubic and gaussian falloff. Like the branched path tracer, this will now shade all intersection points instead of using one at random.
2013-08-24 15:02:08 +00:00
Cycles: Set of fixes for delayed SSS ray tracing There were multiple issues which are solved now: - It was possible that ray wouldn't be bounced off the BSSRDF, for example when PDF or shader eval is zero. In this case PathState might have been left in pre-bounced state which would have been gave incorrect shading results. This is solved by having separate PathState for each of the hits. - Path radiance summing wasn't happening correct as well, indirect rays were using wrong path radiance in the case when there were more than one hit recorded. This is now using a bit trickier state machine which calculates path radiance for just SSS (both direct and indirect) and then sums it back to the final radiance. - Previous commit wasn't totally correct either and was an induced bug due to wrong path state left from the "un-happened" ray bounce. There should be no special case happening here, BSSRDFs will be replaced with diffuse ones due to PATH_RAY_DIFFUSE_ANCESTOR flag. - Merged back codebases for "delayed" and "immediate" indirect SSS ray tracing, hopefully making it easier to maintain the codebase. Sure this changes brings memory usage back by about 4-5%, but overall it's still about 2x memory reduction for the experimental kernel here. Thanks Brecht for the review!
2015-11-28 14:30:35 +00:00
#ifdef __SUBSURFACE__
kernel_path_subsurface_accum_indirect(&ss_indirect, &L);
Cycles: Trace indirect subsurface rays by restarting the integrator loop This gives much lower stack usage on GPU and reduces kernel memory size to around 448MB on GTX560Ti (comparing to 652MB with previous commit and 946MB with official release). There's also a barely measurable speedup of around 5%, but this is to be confirmed still. At this stage we're using only ~3% for the experimental kernel and SSS rendering seems to be faster by 40% and after some further testing we might consider making SSS and CMJ official features and remove experimental precompiled kernels.
2015-11-22 11:08:03 +00:00
/* Trace indirect subsurface rays by restarting the loop. this uses less
* stack memory than invoking kernel_path_indirect.
*/
if(ss_indirect.num_rays) {
kernel_path_subsurface_setup_indirect(kg,
&ss_indirect,
Cycles: Set of fixes for delayed SSS ray tracing There were multiple issues which are solved now: - It was possible that ray wouldn't be bounced off the BSSRDF, for example when PDF or shader eval is zero. In this case PathState might have been left in pre-bounced state which would have been gave incorrect shading results. This is solved by having separate PathState for each of the hits. - Path radiance summing wasn't happening correct as well, indirect rays were using wrong path radiance in the case when there were more than one hit recorded. This is now using a bit trickier state machine which calculates path radiance for just SSS (both direct and indirect) and then sums it back to the final radiance. - Previous commit wasn't totally correct either and was an induced bug due to wrong path state left from the "un-happened" ray bounce. There should be no special case happening here, BSSRDFs will be replaced with diffuse ones due to PATH_RAY_DIFFUSE_ANCESTOR flag. - Merged back codebases for "delayed" and "immediate" indirect SSS ray tracing, hopefully making it easier to maintain the codebase. Sure this changes brings memory usage back by about 4-5%, but overall it's still about 2x memory reduction for the experimental kernel here. Thanks Brecht for the review!
2015-11-28 14:30:35 +00:00
&state,
Cycles: Trace indirect subsurface rays by restarting the integrator loop This gives much lower stack usage on GPU and reduces kernel memory size to around 448MB on GTX560Ti (comparing to 652MB with previous commit and 946MB with official release). There's also a barely measurable speedup of around 5%, but this is to be confirmed still. At this stage we're using only ~3% for the experimental kernel and SSS rendering seems to be faster by 40% and after some further testing we might consider making SSS and CMJ official features and remove experimental precompiled kernels.
2015-11-22 11:08:03 +00:00
&ray,
Cycles: Set of fixes for delayed SSS ray tracing There were multiple issues which are solved now: - It was possible that ray wouldn't be bounced off the BSSRDF, for example when PDF or shader eval is zero. In this case PathState might have been left in pre-bounced state which would have been gave incorrect shading results. This is solved by having separate PathState for each of the hits. - Path radiance summing wasn't happening correct as well, indirect rays were using wrong path radiance in the case when there were more than one hit recorded. This is now using a bit trickier state machine which calculates path radiance for just SSS (both direct and indirect) and then sums it back to the final radiance. - Previous commit wasn't totally correct either and was an induced bug due to wrong path state left from the "un-happened" ray bounce. There should be no special case happening here, BSSRDFs will be replaced with diffuse ones due to PATH_RAY_DIFFUSE_ANCESTOR flag. - Merged back codebases for "delayed" and "immediate" indirect SSS ray tracing, hopefully making it easier to maintain the codebase. Sure this changes brings memory usage back by about 4-5%, but overall it's still about 2x memory reduction for the experimental kernel here. Thanks Brecht for the review!
2015-11-28 14:30:35 +00:00
&L,
Cycles: Trace indirect subsurface rays by restarting the integrator loop This gives much lower stack usage on GPU and reduces kernel memory size to around 448MB on GTX560Ti (comparing to 652MB with previous commit and 946MB with official release). There's also a barely measurable speedup of around 5%, but this is to be confirmed still. At this stage we're using only ~3% for the experimental kernel and SSS rendering seems to be faster by 40% and after some further testing we might consider making SSS and CMJ official features and remove experimental precompiled kernels.
2015-11-22 11:08:03 +00:00
&throughput);
}
else {
break;
}
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).
2015-11-22 10:48:33 +00:00
}
Cycles: Set of fixes for delayed SSS ray tracing There were multiple issues which are solved now: - It was possible that ray wouldn't be bounced off the BSSRDF, for example when PDF or shader eval is zero. In this case PathState might have been left in pre-bounced state which would have been gave incorrect shading results. This is solved by having separate PathState for each of the hits. - Path radiance summing wasn't happening correct as well, indirect rays were using wrong path radiance in the case when there were more than one hit recorded. This is now using a bit trickier state machine which calculates path radiance for just SSS (both direct and indirect) and then sums it back to the final radiance. - Previous commit wasn't totally correct either and was an induced bug due to wrong path state left from the "un-happened" ray bounce. There should be no special case happening here, BSSRDFs will be replaced with diffuse ones due to PATH_RAY_DIFFUSE_ANCESTOR flag. - Merged back codebases for "delayed" and "immediate" indirect SSS ray tracing, hopefully making it easier to maintain the codebase. Sure this changes brings memory usage back by about 4-5%, but overall it's still about 2x memory reduction for the experimental kernel here. Thanks Brecht for the review!
2015-11-28 14:30:35 +00:00
#endif /* __SUBSURFACE__ */
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).
2015-11-22 10:48:33 +00:00
Cycles: Separation of Indirect and Direct clamping. Indirect and Direct samples can now be clamped individually. This way we can clamp the indirect samples (fireflies), while keeping the direct highlights. Example render: http://www.pasteall.org/pic/show.php?id=66586 WARNING: This breaks backwards compatibility. If you had Clamping enabled in an old file, you must re-enable either Direct/Indirect clamping or both again. Reviewed by: brecht Differential Revision: https://developer.blender.org/D303
2014-02-10 20:44:49 +00:00
float3 L_sum = path_radiance_clamp_and_sum(kg, &L);
Cycles: reduce noise using regular path tracing + subsurface scattering with new cubic and gaussian falloff. Like the branched path tracer, this will now shade all intersection points instead of using one at random.
2013-08-24 15:02:08 +00:00
kernel_write_light_passes(kg, buffer, &L, sample);
Cycles: Add support for debug passes Currently only summed number of traversal steps and intersections used by the camera ray intersection pass is implemented, but in the future we will support more debug passes which would help checking what things makes the scene slow. Example of such extra passes could be number of bounces, time spent on the shader tree evaluation and so. Implementation from the Cycles side is pretty much straightforward, could only mention here that it's a build-time option disabled by default. From the blender side it's implemented as a PASS_DEBUG with several subtypes possible. This way we don't need to create an extra DNA pass type for each of the debug passes, saving us a bits. Reviewers: campbellbarton Reviewed By: campbellbarton Differential Revision: https://developer.blender.org/D813
2014-10-04 13:00:26 +00:00
#ifdef __KERNEL_DEBUG__
kernel_write_debug_passes(kg, buffer, &state, &debug_data, sample);
#endif
Cycles: reduce noise using regular path tracing + subsurface scattering with new cubic and gaussian falloff. Like the branched path tracer, this will now shade all intersection points instead of using one at random.
2013-08-24 15:02:08 +00:00
return make_float4(L_sum.x, L_sum.y, L_sum.z, 1.0f - L_transparent);
Cycles: first step for implementation of non-progressive sampler that handles direct and indirect lighting differently. Rather than picking one light for each point on the path, it now loops over all lights for direct lighting. For indirect lighting it still picks a random light each time. It gives control over the number of AA samples, and the number of Diffuse, Glossy, Transmission, AO, Mesh Light, Background and Lamp samples for each AA sample. This helps tuning render performance/noise and tends to give less noise for renders dominated by direct lighting. This sampling mode only works on the CPU, and still needs proper tile rendering to show progress (will follow tommorrow or so), because each AA sample can be quite slow now and so the delay between each update wil be too long.
2012-06-13 11:44:48 +00:00
}
Cycles: change __device and similar qualifiers to ccl_device in kernel code. This to avoids build conflicts with libc++ on FreeBSD, these __ prefixed values are reserved for compilers. I apologize to anyone who has patches or branches and has to go through the pain of merging this change, it may be easiest to do these same replacements in your code and then apply/merge the patch. Ref T37477.
2013-11-15 23:17:10 +00:00
ccl_device void kernel_path_trace(KernelGlobals *kg,
ccl_global float *buffer, ccl_global uint *rng_state,
Cycles / Non-Progressive integrator: * Non-Progressive integrator is now available on the GPU (CUDA, sm_20 and above). Implementation details: * kernel_path_trace() has been split up into two functions: kernel_path_trace_non_progressive() and kernel_path_trace_progressive(). * We compile two CUDA kernel entry functions (in kernel.cu) for the two integrators, they are still inside one .cubin file but due to the kernel separation there should be no performance problem. I tested with the BMW file on my Geforce 540M and the render times were the same for 100 samples (1.57 min in my case). This is part of my GSoC project, SVN merge of r59032 + manual merge of UI changes for this from my branch.
2013-08-09 18:47:25 +00:00
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;
Ray ray;
kernel_path_trace_setup(kg, rng_state, sample, x, y, &rng, &ray);
Cycles render engine, initial commit. This is the engine itself, blender modifications and build instructions will follow later. Cycles uses code from some great open source projects, many thanks them: * BVH building and traversal code from NVidia's "Understanding the Efficiency of Ray Traversal on GPUs": http://code.google.com/p/understanding-the-efficiency-of-ray-traversal-on-gpus/ * Open Shading Language for a large part of the shading system: http://code.google.com/p/openshadinglanguage/ * Blender for procedural textures and a few other nodes. * Approximate Catmull Clark subdivision from NVidia Mesh tools: http://code.google.com/p/nvidia-mesh-tools/ * Sobol direction vectors from: http://web.maths.unsw.edu.au/~fkuo/sobol/ * Film response functions from: http://www.cs.columbia.edu/CAVE/software/softlib/dorf.php
2011-04-27 11:58:34 +00:00
/* integrate */
Fisheye Camera for Cycles For sample images see: http://www.dalaifelinto.com/?p=399 (equisolid) http://www.dalaifelinto.com/?p=389 (equidistant) The 'use_panorama' option is now part of a new Camera type: 'Panorama'. Created two other panorama cameras: - Equisolid: most of lens in the market simulate this lens - e.g. Nikon, Canon, ...) this works as a real lens up to an extent. The final result takes the sensor dimensions into account also. .:. to simulate a Nikon DX2S with a 10.5mm lens do: sensor: 23.7 x 15.7 fisheye lens: 10.5 fisheye fov: 180 render dimensions: 4288 x 2848 - Equidistant: this is not a real lens model. Although the old equidistant lens simulate this lens. The result is always as a circular fisheye that takes the whole sensor (in other words, it doesn't take the sensor into consideration). This is perfect for fulldomes ;) For the UI we have 10 to 360 as soft values and 10 to 3600 as hard values (because we can). Reference material: http://www.hdrlabs.com/tutorials/downloads_files/HDRI%20for%20CGI.pdf http://www.bobatkins.com/photography/technical/field_of_view.html Note, this is not a real simulation of the light path through the lens. The ideal solution would be this: https://graphics.stanford.edu/wikis/cs348b-11/Assignment3 http://www.graphics.stanford.edu/papers/camera/ Thanks Brecht for the fix, suggestions and code review. Kudos for the dome community for keeping me stimulated on the topic since 2009 ;) Patch partly implemented during lab time at VisGraf, IMPA - Rio de Janeiro.
2012-05-04 16:20:51 +00:00
float4 L;
Cycles Volume Render: support for rendering of homogeneous volume with absorption. This is the simplest possible volume rendering case, constant density inside the volume and no scattering or emission. My plan is to tweak, verify and commit more volume rendering effects one by one, doing it all at once makes it difficult to verify correctness and track down bugs. Documentation is here: http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Materials/Volume Currently this hooks into path tracing in 3 ways, which should get us pretty far until we add more advanced light sampling. These 3 hooks are repeated in the path tracing, branched path tracing and transparent shadow code: * Determine active volume shader at start of the path * Change active volume shader on transmission through a surface * Light attenuation over line segments between camera, surfaces and background This is work by "storm", Stuart Broadfoot, Thomas Dinges and myself.
2013-12-28 15:56:19 +00:00
if(ray.t != 0.0f)
Fix #36545: crash with branched path tracing, correlated multi-jittered sampling and subsurface scattering.
2013-08-23 23:04:50 +00:00
L = kernel_path_integrate(kg, &rng, sample, ray, buffer);
Cycles / Non-Progressive integrator: * Non-Progressive integrator is now available on the GPU (CUDA, sm_20 and above). Implementation details: * kernel_path_trace() has been split up into two functions: kernel_path_trace_non_progressive() and kernel_path_trace_progressive(). * We compile two CUDA kernel entry functions (in kernel.cu) for the two integrators, they are still inside one .cubin file but due to the kernel separation there should be no performance problem. I tested with the BMW file on my Geforce 540M and the render times were the same for 100 samples (1.57 min in my case). This is part of my GSoC project, SVN merge of r59032 + manual merge of UI changes for this from my branch.
2013-08-09 18:47:25 +00:00
else
L = make_float4(0.0f, 0.0f, 0.0f, 0.0f);
/* accumulate result in output buffer */
kernel_write_pass_float4(buffer, sample, L);
path_rng_end(kg, rng_state, rng);
}
Cycles render engine, initial commit. This is the engine itself, blender modifications and build instructions will follow later. Cycles uses code from some great open source projects, many thanks them: * BVH building and traversal code from NVidia's "Understanding the Efficiency of Ray Traversal on GPUs": http://code.google.com/p/understanding-the-efficiency-of-ray-traversal-on-gpus/ * Open Shading Language for a large part of the shading system: http://code.google.com/p/openshadinglanguage/ * Blender for procedural textures and a few other nodes. * Approximate Catmull Clark subdivision from NVidia Mesh tools: http://code.google.com/p/nvidia-mesh-tools/ * Sobol direction vectors from: http://web.maths.unsw.edu.au/~fkuo/sobol/ * Film response functions from: http://www.cs.columbia.edu/CAVE/software/softlib/dorf.php
2011-04-27 11:58:34 +00:00
CCL_NAMESPACE_END
Copy Permalink