sandey/blender
1
0
Fork 0
forked from bartvdbraak/blender
Code Pull Requests Activity
55640288d2
blender/intern/cycles/kernel/kernel_path.h

1205 lines
34 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
/*
* Copyright 2011, Blender Foundation.
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public License
* as published by the Free Software Foundation; either version 2
* of the License, or (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software Foundation,
* Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
*/
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__
#include "osl_shader.h"
#endif
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_differential.h"
#include "kernel_montecarlo.h"
Fix most of #31307: cycles panorama camera not working correct with speed vectors and window texture coordinates. Only for Fisheye Equisolid it's still not working correct yet. Patch from Dalai with modifications.
2012-05-07 10:53:09 +00:00
#include "kernel_projection.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_object.h"
Cycles: support for motion vector and UV passes. Most of the changes are related to adding support for motion data throughout the code. There's some code for actual camera/object motion blur raytracing but it's unfinished (it badly slows down the raytracing kernel even when the option is turned off), so that code it disabled still. Motion vector export from Blender tries to avoid computing derived meshes when the mesh does not have a deforming modifier, and it also won't store motion vectors for every vertex if only the object or camera is moving.
2012-04-30 12:49:26 +00:00
#include "kernel_triangle.h"
Cycles Hair: refactoring to support generic attributes for hair curves. There should be no functional changes yet. UV, tangent and intercept are now stored as attributes, with the intention to add more like multiple uv's, vertex colors, generated coordinates and motion vectors later. Things got a bit messy due to having both triangle and curve data in the same mesh data structure, which also gives us two sets of attributes. This will get cleaned up when we split the mesh class.
2013-01-03 12:08:54 +00:00
#include "kernel_curve.h"
#include "kernel_primitive.h"
#include "kernel_projection.h"
Cycles Hair: Strand Minimum Pixel Size Code is added to restrict the pixel size of strands in cycles. It works best with ribbon primitives and a preset for these is included. It uses distance dependent expansion of the strands and then stochastic strand removal to give a fading. To prevent a slowdown for triangle mesh objects in the BVH an extra visibility flag has been added. It is also only applied for camera rays. The strand width settings are also changed, so that the particle size is not included in the width calculation. Instead there is a separate particle system parameter for width scaling.
2013-04-15 21:38:31 +00:00
#include "kernel_random.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_bvh.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_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_camera.h"
#include "kernel_shader.h"
#include "kernel_light.h"
#include "kernel_emission.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__
#include "kernel_subsurface.h"
#endif
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_BEGIN
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
typedef struct PathState {
Code cleanup: avoid some warnings due to implicit uint/int/float/double conversion.
2013-06-07 16:06:17 +00:00
int flag;
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
int bounce;
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: * 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
int diffuse_bounce;
int glossy_bounce;
int transmission_bounce;
int transparent_bounce;
} PathState;
__device_inline void path_state_init(PathState *state)
{
Cycles: add transparent shadow support, i.e. shadows through Transparent BSDF shaders, enabled by default.
2011-10-16 17:54:43 +00:00
state->flag = PATH_RAY_CAMERA|PATH_RAY_SINGULAR|PATH_RAY_MIS_SKIP;
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
state->bounce = 0;
state->diffuse_bounce = 0;
state->glossy_bounce = 0;
state->transmission_bounce = 0;
state->transparent_bounce = 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: add transparent shadow support, i.e. shadows through Transparent BSDF shaders, enabled by default.
2011-10-16 17:54:43 +00:00
__device_inline void path_state_next(KernelGlobals *kg, PathState *state, int label)
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: * 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
/* ray through transparent keeps same flags from previous ray and is
style cleanup: block comments
2012-06-09 17:22:52 +00:00
* not counted as a regular bounce, transparent has separate max */
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
if(label & LABEL_TRANSPARENT) {
state->flag |= PATH_RAY_TRANSPARENT;
state->transparent_bounce++;
Cycles: add transparent shadow support, i.e. shadows through Transparent BSDF shaders, enabled by default.
2011-10-16 17:54:43 +00:00
if(!kernel_data.integrator.transparent_shadows)
state->flag |= PATH_RAY_MIS_SKIP;
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
return;
}
state->bounce++;
/* reflection/transmission */
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
if(label & LABEL_REFLECT) {
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
state->flag |= PATH_RAY_REFLECT;
state->flag &= ~(PATH_RAY_TRANSMIT|PATH_RAY_CAMERA|PATH_RAY_TRANSPARENT);
if(label & LABEL_DIFFUSE)
state->diffuse_bounce++;
else
state->glossy_bounce++;
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
}
else {
kernel_assert(label & LABEL_TRANSMIT);
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
state->flag |= PATH_RAY_TRANSMIT;
state->flag &= ~(PATH_RAY_REFLECT|PATH_RAY_CAMERA|PATH_RAY_TRANSPARENT);
state->transmission_bounce++;
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
}
/* diffuse/glossy/singular */
if(label & LABEL_DIFFUSE) {
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
state->flag |= PATH_RAY_DIFFUSE;
Cycles: add transparent shadow support, i.e. shadows through Transparent BSDF shaders, enabled by default.
2011-10-16 17:54:43 +00:00
state->flag &= ~(PATH_RAY_GLOSSY|PATH_RAY_SINGULAR|PATH_RAY_MIS_SKIP);
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
}
else if(label & LABEL_GLOSSY) {
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
state->flag |= PATH_RAY_GLOSSY;
Cycles: add transparent shadow support, i.e. shadows through Transparent BSDF shaders, enabled by default.
2011-10-16 17:54:43 +00:00
state->flag &= ~(PATH_RAY_DIFFUSE|PATH_RAY_SINGULAR|PATH_RAY_MIS_SKIP);
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
}
else {
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
kernel_assert(label & LABEL_SINGULAR);
Cycles: add transparent shadow support, i.e. shadows through Transparent BSDF shaders, enabled by default.
2011-10-16 17:54:43 +00:00
state->flag |= PATH_RAY_GLOSSY|PATH_RAY_SINGULAR|PATH_RAY_MIS_SKIP;
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
state->flag &= ~PATH_RAY_DIFFUSE;
}
}
Cycles: code refactoring, to do render layer visibility test a bit different, replacing the camera visibility flag with object layer flags.
2011-12-21 20:51:43 +00:00
__device_inline uint path_state_ray_visibility(KernelGlobals *kg, PathState *state)
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
{
uint flag = state->flag;
/* for visibility, diffuse/glossy are for reflection only */
if(flag & PATH_RAY_TRANSMIT)
flag &= ~(PATH_RAY_DIFFUSE|PATH_RAY_GLOSSY);
Cycles: code refactoring, to do render layer visibility test a bit different, replacing the camera visibility flag with object layer flags.
2011-12-21 20:51:43 +00:00
/* for camera visibility, use render layer flags */
if(flag & PATH_RAY_CAMERA)
flag |= kernel_data.integrator.layer_flag;
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: * 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
return flag;
}
__device_inline float path_state_terminate_probability(KernelGlobals *kg, PathState *state, const float3 throughput)
{
if(state->flag & PATH_RAY_TRANSPARENT) {
/* transparent rays treated separately */
if(state->transparent_bounce >= kernel_data.integrator.transparent_max_bounce)
return 0.0f;
else if(state->transparent_bounce <= kernel_data.integrator.transparent_min_bounce)
return 1.0f;
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: * 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
else {
/* other rays */
if((state->bounce >= kernel_data.integrator.max_bounce) ||
(state->diffuse_bounce >= kernel_data.integrator.max_diffuse_bounce) ||
(state->glossy_bounce >= kernel_data.integrator.max_glossy_bounce) ||
(state->transmission_bounce >= kernel_data.integrator.max_transmission_bounce))
style cleanup: assignment & indentation.
2012-06-09 18:56:12 +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
return 0.0f;
style cleanup: assignment & indentation.
2012-06-09 18:56:12 +00:00
}
else if(state->bounce <= kernel_data.integrator.min_bounce) {
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
return 1.0f;
style cleanup: assignment & indentation.
2012-06-09 18:56:12 +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
}
/* probalistic termination */
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
return average(throughput); /* todo: try using max here */
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: 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
__device_inline bool shadow_blocked(KernelGlobals *kg, PathState *state, Ray *ray, float3 *shadow)
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
{
Fix #30955: cycles render issue with lights that have shadow casting disabled.
2012-04-15 15:35:09 +00:00
*shadow = make_float3(1.0f, 1.0f, 1.0f);
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
if(ray->t == 0.0f)
return 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
Intersection isect;
Cycles OpenCL: fix other build issues when enabling more features.
2013-05-09 15:28:38 +00:00
#ifdef __HAIR__
bool result = scene_intersect(kg, ray, PATH_RAY_SHADOW_OPAQUE, &isect, NULL, 0.0f, 0.0f);
#else
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
bool result = scene_intersect(kg, ray, PATH_RAY_SHADOW_OPAQUE, &isect);
Cycles OpenCL: fix other build issues when enabling more features.
2013-05-09 15:28:38 +00:00
#endif
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: * 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
#ifdef __TRANSPARENT_SHADOWS__
if(result && kernel_data.integrator.transparent_shadows) {
/* transparent shadows work in such a way to try to minimize overhead
style cleanup: block comments
2012-06-09 17:22:52 +00:00
* in cases where we don't need them. after a regular shadow ray is
* cast we check if the hit primitive was potentially transparent, and
* only in that case start marching. this gives on extra ray cast for
* the cases were we do want transparency.
*
* also note that for this to work correct, multi close sampling must
* be used, since we don't pass a random number to shader_eval_surface */
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
if(shader_transparent_shadow(kg, &isect)) {
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
float3 throughput = make_float3(1.0f, 1.0f, 1.0f);
float3 Pend = ray->P + ray->D*ray->t;
int bounce = state->transparent_bounce;
for(;;) {
if(bounce >= kernel_data.integrator.transparent_max_bounce) {
return true;
}
else if(bounce >= kernel_data.integrator.transparent_min_bounce) {
/* todo: get random number somewhere for probabilistic terminate */
#if 0
float probability = average(throughput);
Cycles: add transparent shadow support, i.e. shadows through Transparent BSDF shaders, enabled by default.
2011-10-16 17:54:43 +00:00
float terminate = 0.0f;
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
if(terminate >= probability)
return true;
throughput /= probability;
#endif
}
Cycles OpenCL: fix other build issues when enabling more features.
2013-05-09 15:28:38 +00:00
#ifdef __HAIR__
if(!scene_intersect(kg, ray, PATH_RAY_SHADOW_TRANSPARENT, &isect, NULL, 0.0f, 0.0f)) {
#else
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
if(!scene_intersect(kg, ray, PATH_RAY_SHADOW_TRANSPARENT, &isect)) {
Cycles OpenCL: fix other build issues when enabling more features.
2013-05-09 15:28:38 +00:00
#endif
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
*shadow *= throughput;
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
return false;
}
Cycles: add transparent shadow support, i.e. shadows through Transparent BSDF shaders, enabled by default.
2011-10-16 17:54:43 +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
if(!shader_transparent_shadow(kg, &isect))
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
return true;
ShaderData sd;
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
shader_setup_from_ray(kg, &sd, &isect, ray);
Cycles OSL minor optimizations: recycle shading context, don't do memory allocations for trace data, avoid some virtual function calls. Only helps a few percentages.
2012-12-15 10:18:42 +00:00
shader_eval_surface(kg, &sd, 0.0f, PATH_RAY_SHADOW, SHADER_CONTEXT_SHADOW);
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
throughput *= shader_bsdf_transparency(kg, &sd);
ray->P = ray_offset(sd.P, -sd.Ng);
Cycles: add transparent shadow support, i.e. shadows through Transparent BSDF shaders, enabled by default.
2011-10-16 17:54:43 +00:00
if(ray->t != FLT_MAX)
Cycles: * Revert r57203 (len() renaming) There seems to be a problem with nVidia OpenCL after this and I haven't figured out the real cause yet. Better to selectively enable native length() later, after figuring out what's wrong. This fixes [#35612].
2013-06-04 17:20:00 +00:00
ray->D = normalize_len(Pend - ray->P, &ray->t);
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
bounce++;
}
}
}
#endif
return result;
}
Fix #35665: cycles CUDA crash after recent changes. This works around a compiler bug in CUDA 4.2 (solved in 5.5) with typedef'd function parameters.
2013-06-09 16:37:04 +00:00
__device float4 kernel_path_progressive(KernelGlobals *kg, RNG *rng, int sample, Ray ray, __global float *buffer)
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
{
/* initialize */
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
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 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: 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
float min_ray_pdf = FLT_MAX;
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
float ray_pdf = 0.0f;
Cycles: different fix for perlin noise generating nan values, now check for the result to be finite afterwards which is a bit faster and works for OSL too without needing to slow down OSL itself.
2013-01-09 22:06:03 +00:00
#ifdef __LAMP_MIS__
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
float ray_t = 0.0f;
Cycles: different fix for perlin noise generating nan values, now check for the result to be finite afterwards which is a bit faster and works for OSL too without needing to slow down OSL itself.
2013-01-09 22:06:03 +00:00
#endif
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
PathState state;
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
int rng_offset = PRNG_BASE_NUM;
Fix #35804: NVidia OpenCL render issue after sampler changes, workaround what looks like a compiler bug.
2013-06-21 12:34:34 +00:00
#ifdef __CMJ__
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
int num_samples = kernel_data.integrator.aa_samples;
Fix #35804: NVidia OpenCL render issue after sampler changes, workaround what looks like a compiler bug.
2013-06-21 12:34:34 +00:00
#else
int num_samples = 0;
#endif
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: * 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
path_state_init(&state);
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 */
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
for(;; rng_offset += PRNG_BOUNCE_NUM) {
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: code refactoring, to do render layer visibility test a bit different, replacing the camera visibility flag with object layer flags.
2011-12-21 20:51:43 +00:00
uint visibility = path_state_ray_visibility(kg, &state);
Cycles Hair: Strand Minimum Pixel Size Code is added to restrict the pixel size of strands in cycles. It works best with ribbon primitives and a preset for these is included. It uses distance dependent expansion of the strands and then stochastic strand removal to give a fading. To prevent a slowdown for triangle mesh objects in the BVH an extra visibility flag has been added. It is also only applied for camera rays. The strand width settings are also changed, so that the particle size is not included in the width calculation. Instead there is a separate particle system parameter for width scaling.
2013-04-15 21:38:31 +00:00
#ifdef __HAIR__
Cycles: code refactoring to deduplicate the various BVH traversal variations. Now there is a single BVH traversal code with #ifdefs for various features. At runtime it will then select the appropriate variation to use depending if instancing, hair or motion blur is in use. This makes scenes without hair render a bit faster, especially after the minimum width feature was added. It's not the most beautiful code, but we can't use c++ templates and there were already 4 copies, adding 4 more to handle the hair case separately would be too much.
2013-04-17 20:07:22 +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;*/
Cycles: * Revert r57203 (len() renaming) There seems to be a problem with nVidia OpenCL after this and I haven't figured out the real cause yet. Better to selectively enable native length() later, after figuring out what's wrong. This fixes [#35612].
2013-06-04 17:20:00 +00:00
difl = kernel_data.curve_kernel_data.minimum_width * len(pixdiff) * 0.5f;
Cycles: code refactoring to deduplicate the various BVH traversal variations. Now there is a single BVH traversal code with #ifdefs for various features. At runtime it will then select the appropriate variation to use depending if instancing, hair or motion blur is in use. This makes scenes without hair render a bit faster, especially after the minimum width feature was added. It's not the most beautiful code, but we can't use c++ templates and there were already 4 copies, adding 4 more to handle the hair case separately would be too much.
2013-04-17 20:07:22 +00:00
}
extmax = kernel_data.curve_kernel_data.maximum_width;
Fix #35665: cycles CUDA crash after recent changes. This works around a compiler bug in CUDA 4.2 (solved in 5.5) with typedef'd function parameters.
2013-06-09 16:37:04 +00:00
lcg_state = lcg_init(*rng + rng_offset + sample*0x51633e2d);
Cycles Hair: Strand Minimum Pixel Size Code is added to restrict the pixel size of strands in cycles. It works best with ribbon primitives and a preset for these is included. It uses distance dependent expansion of the strands and then stochastic strand removal to give a fading. To prevent a slowdown for triangle mesh objects in the BVH an extra visibility flag has been added. It is also only applied for camera rays. The strand width settings are also changed, so that the particle size is not included in the width calculation. Instead there is a separate particle system parameter for width scaling.
2013-04-15 21:38:31 +00:00
}
bool hit = scene_intersect(kg, &ray, visibility, &isect, &lcg_state, difl, extmax);
#else
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
bool hit = scene_intersect(kg, &ray, visibility, &isect);
Cycles Hair: Strand Minimum Pixel Size Code is added to restrict the pixel size of strands in cycles. It works best with ribbon primitives and a preset for these is included. It uses distance dependent expansion of the strands and then stochastic strand removal to give a fading. To prevent a slowdown for triangle mesh objects in the BVH an extra visibility flag has been added. It is also only applied for camera rays. The strand width settings are also changed, so that the particle size is not included in the width calculation. Instead there is a separate particle system parameter for width scaling.
2013-04-15 21:38:31 +00:00
#endif
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: 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;
light_ray.P = ray.P - ray_t*ray.D;
ray_t += isect.t;
light_ray.D = ray.D;
light_ray.t = ray_t;
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 */
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
float light_t = path_rng_1D(kg, rng, sample, num_samples, rng_offset + PRNG_LIGHT);
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
float3 emission;
if(indirect_lamp_emission(kg, &light_ray, state.flag, ray_pdf, light_t, &emission))
path_radiance_accum_emission(&L, throughput, emission, state.bounce);
}
#endif
if(!hit) {
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
/* eval background shader if nothing hit */
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
if(kernel_data.background.transparent && (state.flag & PATH_RAY_CAMERA)) {
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
L_transparent += average(throughput);
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
#ifdef __PASSES__
if(!(kernel_data.film.pass_flag & PASS_BACKGROUND))
#endif
break;
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: 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
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 */
float3 L_background = indirect_background(kg, &ray, state.flag, ray_pdf);
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 */
ShaderData sd;
shader_setup_from_ray(kg, &sd, &isect, &ray);
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
float rbsdf = path_rng_1D(kg, rng, sample, num_samples, rng_offset + PRNG_BSDF);
Cycles OSL minor optimizations: recycle shading context, don't do memory allocations for trace data, avoid some virtual function calls. Only helps a few percentages.
2012-12-15 10:18:42 +00:00
shader_eval_surface(kg, &sd, rbsdf, state.flag, SHADER_CONTEXT_MAIN);
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: 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
/* holdout */
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
#ifdef __HOLDOUT__
Fix #31168: cycles mask layer should only affect objects for camera rays. Fix: texture coordinate normal output was not correct, still changed under object transform.
2012-05-02 09:33:45 +00:00
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
Fix #32205: Holdout shader + transparent background stays black with Alpha = 1 It was read of initialized memory around holdout_weight in cases when holdout material is used. Seems that it should be assigned to result of shader_holdout_eval here. If Brecht could double check this it'll be great. This could potentially fix #32224: Holdout Error with CUDA Cycles Render
2012-08-01 12:59:47 +00:00
holdout_weight = shader_holdout_eval(kg, &sd);
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 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
/* any throughput is ok, should all be identical here */
L_transparent += average(holdout_weight*throughput);
Fix #31168: cycles mask layer should only affect objects for camera rays. Fix: texture coordinate normal output was not correct, still changed under object transform.
2012-05-02 09:33:45 +00:00
}
Cycles: small code cleanup + fix SSS closure mixed with other closures doing a bit too much work.
2013-04-02 16:37:28 +00:00
if(sd.flag & SD_HOLDOUT_MASK)
Fix #31168: cycles mask layer should only affect objects for camera rays. Fix: texture coordinate normal output was not correct, still changed under object transform.
2012-05-02 09:33:45 +00:00
break;
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
}
#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
/* holdout mask objects do not write data passes */
kernel_write_data_passes(kg, buffer, &L, &sd, sample, state.flag, throughput);
/* 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) {
float blur_pdf = kernel_data.integrator.filter_glossy*min_ray_pdf;
if(blur_pdf < 1.0f) {
float blur_roughness = sqrtf(1.0f - blur_pdf)*0.5f;
shader_bsdf_blur(kg, &sd, blur_roughness);
}
}
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: 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
if(sd.flag & SD_EMISSION) {
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
/* todo: is isect.t wrong here for transparent surfaces? */
float3 emission = indirect_primitive_emission(kg, &sd, isect.t, state.flag, ray_pdf);
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
path_radiance_accum_emission(&L, throughput, emission, state.bounce);
}
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: * 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
float probability = path_state_terminate_probability(kg, &state, throughput);
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) {
float terminate = path_rng_1D(kg, rng, sample, num_samples, rng_offset + PRNG_TERMINATE);
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: 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__
/* bssrdf scatter to a different location on the same object, replacing
* the closures with a diffuse BSDF */
if(sd.flag & SD_BSSRDF) {
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) {
Fix #35665: cycles CUDA crash after recent changes. This works around a compiler bug in CUDA 4.2 (solved in 5.5) with typedef'd function parameters.
2013-06-09 16:37:04 +00:00
uint lcg_state = lcg_init(*rng + rng_offset + sample*0x68bc21eb);
Fix #34852: multilayer SSS material rendering different in progressive and non-progressive integrator.
2013-04-03 16:12:13 +00:00
subsurface_scatter_step(kg, &sd, state.flag, sc, &lcg_state, false);
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: 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: 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: 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
/* todo: solve correlation */
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;
path_rng_2D(kg, rng, sample, num_samples, rng_offset + 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;
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;
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
Fix #33188: AO pass wrong after recent change.
2012-11-15 21:32:05 +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;
light_ray.P = ray_offset(sd.P, sd.Ng);
light_ray.D = ao_D;
light_ray.t = kernel_data.background.ao_distance;
Cycles: camera motion blur enabled. Still more work needed to get object motion blur ready.
2012-10-09 18:37:14 +00:00
#ifdef __OBJECT_MOTION__
Cycles: support for motion vector and UV passes. Most of the changes are related to adding support for motion data throughout the code. There's some code for actual camera/object motion blur raytracing but it's unfinished (it badly slows down the raytracing kernel even when the option is turned off), so that code it disabled still. Motion vector export from Blender tries to avoid computing derived meshes when the mesh does not have a deforming modifier, and it also won't store motion vectors for every vertex if only the object or camera is moving.
2012-04-30 12:49:26 +00:00
light_ray.time = sd.time;
#endif
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.dP = sd.dP;
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: ambient occlusion now takes per-BSDF normals into account.
2012-11-15 15:37:58 +00:00
if(!shadow_blocked(kg, &state, &light_ray, &ao_shadow))
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
path_radiance_accum_ao(&L, throughput, ao_bsdf, ao_shadow, state.bounce);
}
}
#endif
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
#ifdef __EMISSION__
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) {
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
/* sample illumination from lights to find path contribution */
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
if(sd.flag & SD_BSDF_HAS_EVAL) {
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
float light_t = path_rng_1D(kg, rng, sample, num_samples, rng_offset + PRNG_LIGHT);
#ifdef __MULTI_CLOSURE__
float light_o = 0.0f;
#else
float light_o = path_rng_1D(kg, rng, sample, num_samples, rng_offset + PRNG_LIGHT_F);
#endif
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 light_u, light_v;
path_rng_2D(kg, rng, sample, num_samples, rng_offset + PRNG_LIGHT_U, &light_u, &light_v);
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
Ray light_ray;
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
BsdfEval L_light;
Fix #33984: cycles shadow pass problem with CUDA.
2013-01-30 17:04:51 +00:00
bool is_lamp;
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: camera motion blur enabled. Still more work needed to get object motion blur ready.
2012-10-09 18:37:14 +00:00
#ifdef __OBJECT_MOTION__
Cycles: support for motion vector and UV passes. Most of the changes are related to adding support for motion data throughout the code. There's some code for actual camera/object motion blur raytracing but it's unfinished (it badly slows down the raytracing kernel even when the option is turned off), so that code it disabled still. Motion vector export from Blender tries to avoid computing derived meshes when the mesh does not have a deforming modifier, and it also won't store motion vectors for every vertex if only the object or camera is moving.
2012-04-30 12:49:26 +00:00
light_ray.time = sd.time;
#endif
Fix #33984: cycles shadow pass problem with CUDA.
2013-01-30 17:04:51 +00:00
if(direct_emission(kg, &sd, -1, light_t, light_o, light_u, light_v, &light_ray, &L_light, &is_lamp)) {
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
/* trace shadow ray */
float3 shadow;
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: 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(!shadow_blocked(kg, &state, &light_ray, &shadow)) {
/* accumulate */
Fix non-progressive lamps with multiple samples not giving correct intensity after recent fix.
2013-02-05 13:33:24 +00:00
path_radiance_accum_light(&L, throughput, &L_light, shadow, 1.0f, state.bounce, is_lamp);
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: 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: 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
/* no BSDF? we can stop here */
Cycles: small code cleanup + fix SSS closure mixed with other closures doing a bit too much work.
2013-04-02 16:37:28 +00:00
if(!(sd.flag & SD_BSDF))
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;
/* sample BSDF */
float bsdf_pdf;
BsdfEval bsdf_eval;
float3 bsdf_omega_in;
differential3 bsdf_domega_in;
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;
path_rng_2D(kg, rng, sample, num_samples, rng_offset + PRNG_BSDF_U, &bsdf_u, &bsdf_v);
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
int label;
label = shader_bsdf_sample(kg, &sd, bsdf_u, bsdf_v, &bsdf_eval,
&bsdf_omega_in, &bsdf_domega_in, &bsdf_pdf);
if(bsdf_pdf == 0.0f || bsdf_eval_is_zero(&bsdf_eval))
break;
/* modify throughput */
path_radiance_bsdf_bounce(&L, &throughput, &bsdf_eval, bsdf_pdf, state.bounce, label);
/* set labels */
if(!(label & LABEL_TRANSPARENT)) {
ray_pdf = bsdf_pdf;
Cycles: different fix for perlin noise generating nan values, now check for the result to be finite afterwards which is a bit faster and works for OSL too without needing to slow down OSL itself.
2013-01-09 22:06:03 +00:00
#ifdef __LAMP_MIS__
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_t = 0.0f;
Cycles: different fix for perlin noise generating nan values, now check for the result to be finite afterwards which is a bit faster and works for OSL too without needing to slow down OSL itself.
2013-01-09 22:06:03 +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
min_ray_pdf = fminf(bsdf_pdf, min_ray_pdf);
}
/* update path state */
path_state_next(kg, &state, label);
/* setup ray */
ray.P = ray_offset(sd.P, (label & LABEL_TRANSMIT)? -sd.Ng: sd.Ng);
ray.D = bsdf_omega_in;
Fix #33257: cycles camera clipping did not work through transparent objects.
2012-11-21 13:22:24 +00:00
if(state.bounce == 0)
ray.t -= sd.ray_length; /* clipping works through transparent */
else
ray.t = FLT_MAX;
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 __RAY_DIFFERENTIALS__
ray.dP = sd.dP;
ray.dD = bsdf_domega_in;
#endif
}
float3 L_sum = path_radiance_sum(kg, &L);
#ifdef __CLAMP_SAMPLE__
path_radiance_clamp(&L, &L_sum, kernel_data.integrator.sample_clamp);
#endif
kernel_write_light_passes(kg, buffer, &L, sample);
return make_float4(L_sum.x, L_sum.y, L_sum.z, 1.0f - L_transparent);
}
#ifdef __NON_PROGRESSIVE__
Fix #35665: cycles CUDA crash after recent changes. This works around a compiler bug in CUDA 4.2 (solved in 5.5) with typedef'd function parameters.
2013-06-09 16:37:04 +00:00
__device void kernel_path_indirect(KernelGlobals *kg, RNG *rng, int sample, Ray ray, __global float *buffer,
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
float3 throughput, int num_samples, int num_total_samples,
Fix #33824: cycles non-progressive render mode did not do correct path termination, leading to too much noise when the min bounce setting was lower than max bounce.
2013-01-10 19:28:16 +00:00
float min_ray_pdf, float ray_pdf, PathState state, int rng_offset, PathRadiance *L)
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: different fix for perlin noise generating nan values, now check for the result to be finite afterwards which is a bit faster and works for OSL too without needing to slow down OSL itself.
2013-01-09 22:06:03 +00:00
#ifdef __LAMP_MIS__
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
float ray_t = 0.0f;
Cycles: different fix for perlin noise generating nan values, now check for the result to be finite afterwards which is a bit faster and works for OSL too without needing to slow down OSL itself.
2013-01-09 22:06:03 +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
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 */
for(;; rng_offset += PRNG_BOUNCE_NUM) {
/* intersect scene */
Intersection isect;
uint visibility = path_state_ray_visibility(kg, &state);
Cycles OpenCL: fix other build issues when enabling more features.
2013-05-09 15:28:38 +00:00
#ifdef __HAIR__
bool hit = scene_intersect(kg, &ray, visibility, &isect, NULL, 0.0f, 0.0f);
#else
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
bool hit = scene_intersect(kg, &ray, visibility, &isect);
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: 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;
light_ray.P = ray.P - ray_t*ray.D;
ray_t += isect.t;
light_ray.D = ray.D;
light_ray.t = ray_t;
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 */
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
float light_t = path_rng_1D(kg, rng, sample, num_total_samples, rng_offset + PRNG_LIGHT);
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
float3 emission;
if(indirect_lamp_emission(kg, &light_ray, state.flag, ray_pdf, light_t, &emission))
path_radiance_accum_emission(L, throughput, emission, state.bounce);
}
#endif
if(!hit) {
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 */
float3 L_background = indirect_background(kg, &ray, state.flag, ray_pdf);
path_radiance_accum_background(L, throughput, L_background, state.bounce);
#endif
break;
}
/* setup shading */
ShaderData sd;
shader_setup_from_ray(kg, &sd, &isect, &ray);
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
float rbsdf = path_rng_1D(kg, rng, sample, num_total_samples, rng_offset + PRNG_BSDF);
Cycles OSL minor optimizations: recycle shading context, don't do memory allocations for trace data, avoid some virtual function calls. Only helps a few percentages.
2012-12-15 10:18:42 +00:00
shader_eval_surface(kg, &sd, rbsdf, state.flag, SHADER_CONTEXT_INDIRECT);
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
shader_merge_closures(kg, &sd);
/* 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) {
float blur_pdf = kernel_data.integrator.filter_glossy*min_ray_pdf;
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: 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
float3 emission = indirect_primitive_emission(kg, &sd, isect.t, state.flag, 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
path_radiance_accum_emission(L, throughput, emission, state.bounce);
}
#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: 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
float probability = path_state_terminate_probability(kg, &state, throughput*num_samples);
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) {
float terminate = path_rng_1D(kg, rng, sample, num_total_samples, rng_offset + PRNG_TERMINATE);
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
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__
/* bssrdf scatter to a different location on the same object, replacing
* the closures with a diffuse BSDF */
if(sd.flag & SD_BSSRDF) {
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) {
Fix #35665: cycles CUDA crash after recent changes. This works around a compiler bug in CUDA 4.2 (solved in 5.5) with typedef'd function parameters.
2013-06-09 16:37:04 +00:00
uint lcg_state = lcg_init(*rng + rng_offset + sample*0x68bc21eb);
Fix #34852: multilayer SSS material rendering different in progressive and non-progressive integrator.
2013-04-03 16:12:13 +00:00
subsurface_scatter_step(kg, &sd, state.flag, sc, &lcg_state, false);
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: 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)) {
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;
path_rng_2D(kg, rng, sample, num_total_samples, rng_offset + PRNG_BSDF_U, &bsdf_u, &bsdf_v);
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: 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;
float3 ao_bsdf = shader_bsdf_ao(kg, &sd, ao_factor, &ao_N);
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
float3 ao_D;
float ao_pdf;
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: 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
Fix #33188: AO pass wrong after recent change.
2012-11-15 21:32:05 +00:00
if(dot(sd.Ng, ao_D) > 0.0f && ao_pdf != 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
Ray light_ray;
float3 ao_shadow;
light_ray.P = ray_offset(sd.P, sd.Ng);
light_ray.D = ao_D;
light_ray.t = kernel_data.background.ao_distance;
Cycles: camera motion blur enabled. Still more work needed to get object motion blur ready.
2012-10-09 18:37:14 +00:00
#ifdef __OBJECT_MOTION__
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
light_ray.time = sd.time;
#endif
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.dP = sd.dP;
light_ray.dD = differential3_zero();
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: ambient occlusion now takes per-BSDF normals into account.
2012-11-15 15:37:58 +00:00
if(!shadow_blocked(kg, &state, &light_ray, &ao_shadow))
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_radiance_accum_ao(L, throughput, ao_bsdf, ao_shadow, state.bounce);
}
}
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
#ifdef __EMISSION__
if(kernel_data.integrator.use_direct_light) {
/* sample illumination from lights to find path contribution */
if(sd.flag & SD_BSDF_HAS_EVAL) {
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
float light_t = path_rng_1D(kg, rng, sample, num_total_samples, rng_offset + PRNG_LIGHT);
#ifdef __MULTI_CLOSURE__
float light_o = 0.0f;
#else
float light_o = path_rng_1D(kg, rng, sample, num_total_samples, rng_offset + PRNG_LIGHT_F);
#endif
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 light_u, light_v;
path_rng_2D(kg, rng, sample, num_total_samples, rng_offset + PRNG_LIGHT_U, &light_u, &light_v);
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
Ray light_ray;
BsdfEval L_light;
Fix #33984: cycles shadow pass problem with CUDA.
2013-01-30 17:04:51 +00:00
bool is_lamp;
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: camera motion blur enabled. Still more work needed to get object motion blur ready.
2012-10-09 18:37:14 +00:00
#ifdef __OBJECT_MOTION__
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
light_ray.time = sd.time;
#endif
/* sample random light */
Fix #33984: cycles shadow pass problem with CUDA.
2013-01-30 17:04:51 +00:00
if(direct_emission(kg, &sd, -1, light_t, light_o, light_u, light_v, &light_ray, &L_light, &is_lamp)) {
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
/* trace shadow ray */
float3 shadow;
if(!shadow_blocked(kg, &state, &light_ray, &shadow)) {
/* accumulate */
Fix non-progressive lamps with multiple samples not giving correct intensity after recent fix.
2013-02-05 13:33:24 +00:00
path_radiance_accum_light(L, throughput, &L_light, shadow, 1.0f, state.bounce, is_lamp);
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 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
}
}
#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
/* no BSDF? we can stop here */
Cycles: small code cleanup + fix SSS closure mixed with other closures doing a bit too much work.
2013-04-02 16:37:28 +00:00
if(!(sd.flag & SD_BSDF))
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
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
/* sample BSDF */
float bsdf_pdf;
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
BsdfEval bsdf_eval;
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
float3 bsdf_omega_in;
differential3 bsdf_domega_in;
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;
path_rng_2D(kg, rng, sample, num_total_samples, rng_offset + PRNG_BSDF_U, &bsdf_u, &bsdf_v);
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
int label;
label = shader_bsdf_sample(kg, &sd, bsdf_u, bsdf_v, &bsdf_eval,
&bsdf_omega_in, &bsdf_domega_in, &bsdf_pdf);
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
if(bsdf_pdf == 0.0f || bsdf_eval_is_zero(&bsdf_eval))
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;
/* modify 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
path_radiance_bsdf_bounce(L, &throughput, &bsdf_eval, bsdf_pdf, state.bounce, label);
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
/* set labels */
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
if(!(label & LABEL_TRANSPARENT)) {
Fix #30034: bug in multiple importance sampling + transparency, would give slightly wrong lighting behind transparent objects.
2012-01-31 14:57:46 +00:00
ray_pdf = bsdf_pdf;
Cycles: different fix for perlin noise generating nan values, now check for the result to be finite afterwards which is a bit faster and works for OSL too without needing to slow down OSL itself.
2013-01-09 22:06:03 +00:00
#ifdef __LAMP_MIS__
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_t = 0.0f;
Cycles: different fix for perlin noise generating nan values, now check for the result to be finite afterwards which is a bit faster and works for OSL too without needing to slow down OSL itself.
2013-01-09 22:06:03 +00:00
#endif
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
min_ray_pdf = fminf(bsdf_pdf, min_ray_pdf);
}
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: * 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
/* update path state */
Cycles: add transparent shadow support, i.e. shadows through Transparent BSDF shaders, enabled by default.
2011-10-16 17:54:43 +00:00
path_state_next(kg, &state, label);
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
/* setup ray */
ray.P = ray_offset(sd.P, (label & LABEL_TRANSMIT)? -sd.Ng: sd.Ng);
ray.D = bsdf_omega_in;
ray.t = FLT_MAX;
#ifdef __RAY_DIFFERENTIALS__
ray.dP = sd.dP;
ray.dD = bsdf_domega_in;
#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
}
Fix #35665: cycles CUDA crash after recent changes. This works around a compiler bug in CUDA 4.2 (solved in 5.5) with typedef'd function parameters.
2013-06-09 16:37:04 +00:00
__device_noinline void kernel_path_non_progressive_lighting(KernelGlobals *kg, RNG *rng, int sample,
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
ShaderData *sd, float3 throughput, float num_samples_adjust,
float min_ray_pdf, float ray_pdf, PathState state,
int rng_offset, PathRadiance *L, __global float *buffer)
{
Fix #35804: NVidia OpenCL render issue after sampler changes, workaround what looks like a compiler bug.
2013-06-21 12:34:34 +00:00
#ifdef __CMJ__
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
int aa_samples = kernel_data.integrator.aa_samples;
Fix #35804: NVidia OpenCL render issue after sampler changes, workaround what looks like a compiler bug.
2013-06-21 12:34:34 +00:00
#else
int aa_samples = 0;
#endif
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
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 __AO__
/* ambient occlusion */
if(kernel_data.integrator.use_ambient_occlusion || (sd->flag & SD_AO)) {
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
int num_samples = ceil_to_int(kernel_data.integrator.ao_samples*num_samples_adjust);
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
float num_samples_inv = num_samples_adjust/num_samples;
float ao_factor = kernel_data.background.ao_factor;
float3 ao_N;
float3 ao_bsdf = shader_bsdf_ao(kg, sd, ao_factor, &ao_N);
for(int j = 0; j < num_samples; j++) {
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;
path_rng_2D(kg, rng, sample*num_samples + j, aa_samples*num_samples, rng_offset + PRNG_BSDF_U, &bsdf_u, &bsdf_v);
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
float3 ao_D;
float ao_pdf;
sample_cos_hemisphere(ao_N, bsdf_u, bsdf_v, &ao_D, &ao_pdf);
if(dot(sd->Ng, ao_D) > 0.0f && ao_pdf != 0.0f) {
Ray light_ray;
float3 ao_shadow;
light_ray.P = ray_offset(sd->P, sd->Ng);
light_ray.D = ao_D;
light_ray.t = kernel_data.background.ao_distance;
#ifdef __OBJECT_MOTION__
light_ray.time = sd->time;
#endif
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.dP = sd->dP;
light_ray.dD = differential3_zero();
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
if(!shadow_blocked(kg, &state, &light_ray, &ao_shadow))
path_radiance_accum_ao(L, throughput*num_samples_inv, ao_bsdf, ao_shadow, state.bounce);
}
}
}
#endif
#ifdef __EMISSION__
/* sample illumination from lights to find path contribution */
if(sd->flag & SD_BSDF_HAS_EVAL) {
Ray light_ray;
BsdfEval L_light;
bool is_lamp;
#ifdef __OBJECT_MOTION__
light_ray.time = sd->time;
#endif
/* lamp sampling */
for(int i = 0; i < kernel_data.integrator.num_all_lights; i++) {
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
int num_samples = ceil_to_int(num_samples_adjust*light_select_num_samples(kg, i));
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
float num_samples_inv = num_samples_adjust/(num_samples*kernel_data.integrator.num_all_lights);
Fix #35665: cycles CUDA crash after recent changes. This works around a compiler bug in CUDA 4.2 (solved in 5.5) with typedef'd function parameters.
2013-06-09 16:37:04 +00:00
RNG lamp_rng = cmj_hash(*rng, i);
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
if(kernel_data.integrator.pdf_triangles != 0.0f)
num_samples_inv *= 0.5f;
for(int j = 0; j < num_samples; j++) {
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 light_u, light_v;
path_rng_2D(kg, &lamp_rng, sample*num_samples + j, aa_samples*num_samples, rng_offset + PRNG_LIGHT_U, &light_u, &light_v);
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
if(direct_emission(kg, sd, i, 0.0f, 0.0f, light_u, light_v, &light_ray, &L_light, &is_lamp)) {
/* trace shadow ray */
float3 shadow;
if(!shadow_blocked(kg, &state, &light_ray, &shadow)) {
/* accumulate */
path_radiance_accum_light(L, throughput*num_samples_inv, &L_light, shadow, num_samples_inv, state.bounce, is_lamp);
}
}
}
}
/* mesh light sampling */
if(kernel_data.integrator.pdf_triangles != 0.0f) {
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
int num_samples = ceil_to_int(num_samples_adjust*kernel_data.integrator.mesh_light_samples);
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
float num_samples_inv = num_samples_adjust/num_samples;
if(kernel_data.integrator.num_all_lights)
num_samples_inv *= 0.5f;
for(int j = 0; j < num_samples; j++) {
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
float light_t = path_rng_1D(kg, rng, sample*num_samples + j, aa_samples*num_samples, rng_offset + PRNG_LIGHT);
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 light_u, light_v;
path_rng_2D(kg, rng, sample*num_samples + j, aa_samples*num_samples, rng_offset + PRNG_LIGHT_U, &light_u, &light_v);
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
/* only sample triangle lights */
if(kernel_data.integrator.num_all_lights)
light_t = 0.5f*light_t;
if(direct_emission(kg, sd, -1, light_t, 0.0f, light_u, light_v, &light_ray, &L_light, &is_lamp)) {
/* trace shadow ray */
float3 shadow;
if(!shadow_blocked(kg, &state, &light_ray, &shadow)) {
/* accumulate */
path_radiance_accum_light(L, throughput*num_samples_inv, &L_light, shadow, num_samples_inv, state.bounce, is_lamp);
}
}
}
}
}
#endif
for(int i = 0; i< sd->num_closure; i++) {
const ShaderClosure *sc = &sd->closure[i];
if(!CLOSURE_IS_BSDF(sc->type))
continue;
/* transparency is not handled here, but in outer loop */
if(sc->type == CLOSURE_BSDF_TRANSPARENT_ID)
continue;
int num_samples;
if(CLOSURE_IS_BSDF_DIFFUSE(sc->type))
num_samples = kernel_data.integrator.diffuse_samples;
else if(CLOSURE_IS_BSDF_GLOSSY(sc->type))
num_samples = kernel_data.integrator.glossy_samples;
else
num_samples = kernel_data.integrator.transmission_samples;
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
num_samples = ceil_to_int(num_samples_adjust*num_samples);
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
float num_samples_inv = num_samples_adjust/num_samples;
Fix #35665: cycles CUDA crash after recent changes. This works around a compiler bug in CUDA 4.2 (solved in 5.5) with typedef'd function parameters.
2013-06-09 16:37:04 +00:00
RNG bsdf_rng = cmj_hash(*rng, i);
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
for(int j = 0; j < num_samples; j++) {
/* sample BSDF */
float bsdf_pdf;
BsdfEval bsdf_eval;
float3 bsdf_omega_in;
differential3 bsdf_domega_in;
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;
path_rng_2D(kg, &bsdf_rng, sample*num_samples + j, aa_samples*num_samples, rng_offset + PRNG_BSDF_U, &bsdf_u, &bsdf_v);
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
int label;
label = shader_bsdf_sample_closure(kg, sd, sc, bsdf_u, bsdf_v, &bsdf_eval,
&bsdf_omega_in, &bsdf_domega_in, &bsdf_pdf);
if(bsdf_pdf == 0.0f || bsdf_eval_is_zero(&bsdf_eval))
continue;
/* modify throughput */
float3 tp = throughput;
path_radiance_bsdf_bounce(L, &tp, &bsdf_eval, bsdf_pdf, state.bounce, label);
/* set labels */
float min_ray_pdf = FLT_MAX;
if(!(label & LABEL_TRANSPARENT))
min_ray_pdf = fminf(bsdf_pdf, min_ray_pdf);
/* modify path state */
PathState ps = state;
path_state_next(kg, &ps, label);
/* setup ray */
Ray bsdf_ray;
bsdf_ray.P = ray_offset(sd->P, (label & LABEL_TRANSMIT)? -sd->Ng: sd->Ng);
bsdf_ray.D = bsdf_omega_in;
bsdf_ray.t = FLT_MAX;
#ifdef __RAY_DIFFERENTIALS__
bsdf_ray.dP = sd->dP;
bsdf_ray.dD = bsdf_domega_in;
#endif
#ifdef __OBJECT_MOTION__
bsdf_ray.time = sd->time;
#endif
kernel_path_indirect(kg, rng, sample*num_samples + j, bsdf_ray, buffer,
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
tp*num_samples_inv, num_samples, aa_samples*num_samples,
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
min_ray_pdf, bsdf_pdf, ps, rng_offset+PRNG_BOUNCE_NUM, L);
/* for render passes, sum and reset indirect light pass variables
* for the next samples */
path_radiance_sum_indirect(L);
path_radiance_reset_indirect(L);
}
}
}
Fix #35665: cycles CUDA crash after recent changes. This works around a compiler bug in CUDA 4.2 (solved in 5.5) with typedef'd function parameters.
2013-06-09 16:37:04 +00:00
__device float4 kernel_path_non_progressive(KernelGlobals *kg, RNG *rng, int sample, Ray ray, __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
{
/* 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);
float ray_pdf = 0.0f;
PathState state;
int rng_offset = PRNG_BASE_NUM;
Fix #35804: NVidia OpenCL render issue after sampler changes, workaround what looks like a compiler bug.
2013-06-21 12:34:34 +00:00
#ifdef __CMJ__
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
int aa_samples = kernel_data.integrator.aa_samples;
Fix #35804: NVidia OpenCL render issue after sampler changes, workaround what looks like a compiler bug.
2013-06-21 12:34:34 +00:00
#else
int aa_samples = 0;
#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_state_init(&state);
for(;; rng_offset += PRNG_BOUNCE_NUM) {
/* intersect scene */
Intersection isect;
uint visibility = path_state_ray_visibility(kg, &state);
Cycles Hair: Strand Minimum Pixel Size Code is added to restrict the pixel size of strands in cycles. It works best with ribbon primitives and a preset for these is included. It uses distance dependent expansion of the strands and then stochastic strand removal to give a fading. To prevent a slowdown for triangle mesh objects in the BVH an extra visibility flag has been added. It is also only applied for camera rays. The strand width settings are also changed, so that the particle size is not included in the width calculation. Instead there is a separate particle system parameter for width scaling.
2013-04-15 21:38:31 +00:00
#ifdef __HAIR__
Cycles: code refactoring to deduplicate the various BVH traversal variations. Now there is a single BVH traversal code with #ifdefs for various features. At runtime it will then select the appropriate variation to use depending if instancing, hair or motion blur is in use. This makes scenes without hair render a bit faster, especially after the minimum width feature was added. It's not the most beautiful code, but we can't use c++ templates and there were already 4 copies, adding 4 more to handle the hair case separately would be too much.
2013-04-17 20:07:22 +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;*/
Cycles: * Revert r57203 (len() renaming) There seems to be a problem with nVidia OpenCL after this and I haven't figured out the real cause yet. Better to selectively enable native length() later, after figuring out what's wrong. This fixes [#35612].
2013-06-04 17:20:00 +00:00
difl = kernel_data.curve_kernel_data.minimum_width * len(pixdiff) * 0.5f;
Cycles: code refactoring to deduplicate the various BVH traversal variations. Now there is a single BVH traversal code with #ifdefs for various features. At runtime it will then select the appropriate variation to use depending if instancing, hair or motion blur is in use. This makes scenes without hair render a bit faster, especially after the minimum width feature was added. It's not the most beautiful code, but we can't use c++ templates and there were already 4 copies, adding 4 more to handle the hair case separately would be too much.
2013-04-17 20:07:22 +00:00
}
extmax = kernel_data.curve_kernel_data.maximum_width;
Fix #35665: cycles CUDA crash after recent changes. This works around a compiler bug in CUDA 4.2 (solved in 5.5) with typedef'd function parameters.
2013-06-09 16:37:04 +00:00
lcg_state = lcg_init(*rng + rng_offset + sample*0x51633e2d);
Cycles Hair: Strand Minimum Pixel Size Code is added to restrict the pixel size of strands in cycles. It works best with ribbon primitives and a preset for these is included. It uses distance dependent expansion of the strands and then stochastic strand removal to give a fading. To prevent a slowdown for triangle mesh objects in the BVH an extra visibility flag has been added. It is also only applied for camera rays. The strand width settings are also changed, so that the particle size is not included in the width calculation. Instead there is a separate particle system parameter for width scaling.
2013-04-15 21:38:31 +00:00
}
if(!scene_intersect(kg, &ray, visibility, &isect, &lcg_state, difl, extmax)) {
#else
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(!scene_intersect(kg, &ray, visibility, &isect)) {
Cycles Hair: Strand Minimum Pixel Size Code is added to restrict the pixel size of strands in cycles. It works best with ribbon primitives and a preset for these is included. It uses distance dependent expansion of the strands and then stochastic strand removal to give a fading. To prevent a slowdown for triangle mesh objects in the BVH an extra visibility flag has been added. It is also only applied for camera rays. The strand width settings are also changed, so that the particle size is not included in the width calculation. Instead there is a separate particle system parameter for width scaling.
2013-04-15 21:38:31 +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
/* eval background shader if nothing hit */
if(kernel_data.background.transparent) {
L_transparent += average(throughput);
#ifdef __PASSES__
if(!(kernel_data.film.pass_flag & PASS_BACKGROUND))
#endif
break;
}
#ifdef __BACKGROUND__
/* sample background shader */
float3 L_background = indirect_background(kg, &ray, state.flag, ray_pdf);
path_radiance_accum_background(&L, throughput, L_background, state.bounce);
#endif
break;
}
/* setup shading */
ShaderData sd;
shader_setup_from_ray(kg, &sd, &isect, &ray);
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
shader_eval_surface(kg, &sd, 0.0f, state.flag, SHADER_CONTEXT_MAIN);
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
shader_merge_closures(kg, &sd);
/* holdout */
#ifdef __HOLDOUT__
if((sd.flag & (SD_HOLDOUT|SD_HOLDOUT_MASK))) {
if(kernel_data.background.transparent) {
float3 holdout_weight;
if(sd.flag & SD_HOLDOUT_MASK)
holdout_weight = make_float3(1.0f, 1.0f, 1.0f);
else
Fix #32205: Holdout shader + transparent background stays black with Alpha = 1 It was read of initialized memory around holdout_weight in cases when holdout material is used. Seems that it should be assigned to result of shader_holdout_eval here. If Brecht could double check this it'll be great. This could potentially fix #32224: Holdout Error with CUDA Cycles Render
2012-08-01 12:59:47 +00:00
holdout_weight = shader_holdout_eval(kg, &sd);
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
/* any throughput is ok, should all be identical here */
L_transparent += average(holdout_weight*throughput);
}
Cycles: small code cleanup + fix SSS closure mixed with other closures doing a bit too much work.
2013-04-02 16:37:28 +00:00
if(sd.flag & SD_HOLDOUT_MASK)
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;
}
#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
/* holdout mask objects do not write data passes */
kernel_write_data_passes(kg, buffer, &L, &sd, sample, state.flag, 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
#ifdef __EMISSION__
/* emission */
if(sd.flag & SD_EMISSION) {
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
float3 emission = indirect_primitive_emission(kg, &sd, isect.t, state.flag, 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
path_radiance_accum_emission(&L, throughput, emission, state.bounce);
}
#endif
/* transparency termination */
if(state.flag & PATH_RAY_TRANSPARENT) {
/* path termination. this is a strange place to put the termination, it's
* mainly due to the mixed in MIS that we use. gives too many unneeded
* shader evaluations, only need emission if we are going to terminate */
float probability = path_state_terminate_probability(kg, &state, throughput);
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) {
float terminate = path_rng_1D(kg, rng, sample, aa_samples, rng_offset + PRNG_TERMINATE);
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(terminate >= probability)
break;
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
}
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__
/* bssrdf scatter to a different location on the same object */
if(sd.flag & SD_BSSRDF) {
for(int i = 0; i< sd.num_closure; i++) {
ShaderClosure *sc = &sd.closure[i];
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: 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
if(!CLOSURE_IS_BSSRDF(sc->type))
continue;
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: 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
/* set up random number generator */
Fix #35665: cycles CUDA crash after recent changes. This works around a compiler bug in CUDA 4.2 (solved in 5.5) with typedef'd function parameters.
2013-06-09 16:37:04 +00:00
uint lcg_state = lcg_init(*rng + rng_offset + sample*0x68bc21eb);
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
int num_samples = kernel_data.integrator.subsurface_samples;
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
float num_samples_inv = 1.0f/num_samples;
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
/* do subsurface scatter step with copy of shader data, this will
* replace the BSSRDF with a diffuse BSDF closure */
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
for(int j = 0; j < num_samples; j++) {
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
ShaderData bssrdf_sd = sd;
Fix #34852: multilayer SSS material rendering different in progressive and non-progressive integrator.
2013-04-03 16:12:13 +00:00
subsurface_scatter_step(kg, &bssrdf_sd, state.flag, sc, &lcg_state, true);
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
/* compute lighting with the BSDF closure */
kernel_path_non_progressive_lighting(kg, rng, sample*num_samples + j,
&bssrdf_sd, throughput, num_samples_inv,
ray_pdf, ray_pdf, state, rng_offset, &L, 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
}
}
}
#endif
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
/* lighting */
kernel_path_non_progressive_lighting(kg, rng, sample, &sd, throughput,
1.0f, ray_pdf, ray_pdf, state, rng_offset, &L, 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
/* continue in case of transparency */
throughput *= shader_bsdf_transparency(kg, &sd);
if(is_zero(throughput))
break;
path_state_next(kg, &state, LABEL_TRANSPARENT);
ray.P = ray_offset(sd.P, -sd.Ng);
Fix #33257: cycles camera clipping did not work through transparent objects.
2012-11-21 13:22:24 +00:00
ray.t -= sd.ray_length; /* clipping works through 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 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: 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
float3 L_sum = path_radiance_sum(kg, &L);
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 rejection of inf/nan samples, in principle these should not happen but this makes it more reliable for now. Also add an integrator "Clamp" option, to clamp very light samples to a maximum value. This will reduce accuracy but may help reducing noise and speed up convergence.
2012-04-05 15:17:45 +00:00
#ifdef __CLAMP_SAMPLE__
path_radiance_clamp(&L, &L_sum, kernel_data.integrator.sample_clamp);
#endif
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
kernel_write_light_passes(kg, buffer, &L, sample);
return make_float4(L_sum.x, L_sum.y, L_sum.z, 1.0f - L_transparent);
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: 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: 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
__device void kernel_path_trace(KernelGlobals *kg,
__global float *buffer, __global uint *rng_state,
int sample, int x, int y, int offset, int stride)
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: 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
/* buffer offset */
int index = offset + x + y*stride;
int pass_stride = kernel_data.film.pass_stride;
rng_state += index;
buffer += index*pass_stride;
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
/* initialize random numbers */
RNG rng;
float filter_u;
float filter_v;
Fix #35804: NVidia OpenCL render issue after sampler changes, workaround what looks like a compiler bug.
2013-06-21 12:34:34 +00:00
#ifdef __CMJ__
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
int num_samples = kernel_data.integrator.aa_samples;
Fix #35804: NVidia OpenCL render issue after sampler changes, workaround what looks like a compiler bug.
2013-06-21 12:34:34 +00:00
#else
int num_samples = 0;
#endif
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
path_rng_init(kg, rng_state, sample, num_samples, &rng, x, y, &filter_u, &filter_v);
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
/* sample camera ray */
Ray ray;
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
Cycles: * Move some hair width related code into a dedicated branch. * Don't calculate time/lens RNG when Motion Blur or Depth of Field are disabled
2013-05-30 11:05:02 +00:00
float lens_u = 0.0f, lens_v = 0.0f;
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
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
if(kernel_data.cam.aperturesize > 0.0f)
path_rng_2D(kg, &rng, sample, num_samples, PRNG_LENS_U, &lens_u, &lens_v);
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
float time = 0.0f;
Cycles: * Move some hair width related code into a dedicated branch. * Don't calculate time/lens RNG when Motion Blur or Depth of Field are disabled
2013-05-30 11:05:02 +00:00
#ifdef __CAMERA_MOTION__
if(kernel_data.cam.shuttertime != -1.0f)
Fix #35665: cycles CUDA crash after recent changes. This works around a compiler bug in CUDA 4.2 (solved in 5.5) with typedef'd function parameters.
2013-06-09 16:37:04 +00:00
time = path_rng_1D(kg, &rng, sample, num_samples, PRNG_TIME);
Cycles: support for motion vector and UV passes. Most of the changes are related to adding support for motion data throughout the code. There's some code for actual camera/object motion blur raytracing but it's unfinished (it badly slows down the raytracing kernel even when the option is turned off), so that code it disabled still. Motion vector export from Blender tries to avoid computing derived meshes when the mesh does not have a deforming modifier, and it also won't store motion vectors for every vertex if only the object or camera is moving.
2012-04-30 12:49:26 +00:00
#endif
camera_sample(kg, x, y, filter_u, filter_v, lens_u, lens_v, time, &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: 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 (ray.t != 0.0f) {
#ifdef __NON_PROGRESSIVE__
if(kernel_data.integrator.progressive)
#endif
Fix #35665: cycles CUDA crash after recent changes. This works around a compiler bug in CUDA 4.2 (solved in 5.5) with typedef'd function parameters.
2013-06-09 16:37:04 +00:00
L = kernel_path_progressive(kg, &rng, sample, ray, 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
#ifdef __NON_PROGRESSIVE__
else
Fix #35665: cycles CUDA crash after recent changes. This works around a compiler bug in CUDA 4.2 (solved in 5.5) with typedef'd function parameters.
2013-06-09 16:37:04 +00:00
L = kernel_path_non_progressive(kg, &rng, sample, ray, 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
#endif
}
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
else
Code cleanup / Cycles: * Simplify shaperadius() function a bit to avoid castings. * Style cleanup 1.f -> 1.0f, to follow rest of Cycles code.
2013-05-18 11:04:29 +00:00
L = make_float4(0.0f, 0.0f, 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
/* accumulate result in output buffer */
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
kernel_write_pass_float4(buffer, sample, L);
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: 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
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