When encountering an error during context creation, the "OptiXDevice" constructor aborts early.
This means the "cuda_stream" vector is never resized and the destructor iterated over non-existent data.
The acceleration structure built by OptiX may be different between GPUs, so cannot assume the memory size is the same for all.
This fixes that by moving the memory management for all OptiX acceleration structures into the responsibility of each device (was already the case for BLAS previously, now for TLAS too).
The Principled BSDF uses Microfacet closures that include a fresnel term,
which are a special case since their weight tends to be near white even
if their average contribution is fairly low.
The sample weight is scaled by the average fresnel weight to account for
this, but the denoising albedo still used the unscaled weight.
This was fine for the original denoiser, but apparently OIDN can't handle
the resulting albedo pass well. Therefore, this commit adds the described
scaling to the albedo pass contribution as well.
This problem was described in T69770.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D6289
Extend options are currently stored per curve. This was not clearly
communicated to the user and they expected this to be a setting per
CurveMapping.
This change will move the option from `Curve` to `CurveMapping`. In
order to support this the API had to be changed.
BPY: CurveMap.evaluate is also moved to CurveMapping.evaluate what
breaks Python API. Cycles has been updated but other add-ons have
not. After release of 2.81 we can merge this to master and adapt
the add-ons.
Reviewed By: campbellbarton
Differential Revision: https://developer.blender.org/D6169
The Random Per Island attribute is a random float associated with each
connected component (island) of the mesh. It is particularly useful
when artists want to add variations to meshes composed of separate
units. Like tree leaves created using particle systems, wood planks
created using array modifiers, or abstract splines created using AN.
Reviewed By: Sergey Sharybin, Jacques Lucke
Differential Revision: https://developer.blender.org/D6154
Calling "OptiXDevice::load_kernels" multiple times would call "optixPipelineDestroy" on a pipeline
pointer that may have already been deleted previously (since the PIP_SHADER_EVAL pipeline is only
created conditionally).
This change also avoids a CUDA kernel reload every time this is called. The CUDA kernels are
precompiled and don't change, so there is no need to reload them every time.
The OptiX intersection program for curves uses "optixGetObjectRayDirection"
to get the ray direction in object space (which was inverse transformed
with the current transformation matrix). OptiX does no additional operations
on it, so if there is a scaling transform, the direction is not normalized.
But the curve intersection routine expects that. In addition, the distances
used in "optixGetRayTmax()" and "optixReportIntersection()" are in world
space, so need to adjust them accordingly.
When using OpenCL with Cycles the rendering time increased substantial.
After doing some tests the bottleneck was found in 4d voronoi and 2d and
3d smooth voronoi.
This change will hide these behind a specific compile directive so the
speed will improve.
AMD RX480 + BMW scene
2.80 (3:10)
2.81 (5:48)
2.81 excluding 4d voronoi+2d/3d smooth (3:50)
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D6231
The multi device code did not correctly handle cases where some GPUs store a
resource in device memory and others store it in host mapped memory.
Differential Revision: https://developer.blender.org/D6126
Cycles did not update the "is_enabled" flag on lights when they were synchronized again, which caused all lights disabled by "LightManager::disable_ineffective_light" to be disabled indefinitely. As a result the OptiX kernels were not reloaded with correct features when a change to a light was made. This fixes that by updating the "is_enabled" flag during synchronization.
Differential Revision: https://developer.blender.org/D6141
The cryptomatte sockets were incorrectly numbered using a step size of two. While the increment by two is necessary to get the correct number of render passes, they should be numbered consecutively matching the socket names of the cryptomatte node.
Reviewed By: lukasstockner97
Differential Revision: https://developer.blender.org/D6185
Was caused by D6068, which did not handle "MEM_PIXELS" memory
when not in background mode. Before that it always fell back to using
generic device memory, so restoring that behavior. In future this
should be changes to use OpenGL interop for optimal performance.
The OptiX implementation wasn't trying to allocate memory on the host if device allocation failed, while the CUDA implementation did. This copies the implementation over to OptiX to remedy that.
Differential Revision: https://developer.blender.org/D6068
This is intended for developers on Windows primarily:
Now, CUDA architectures of type compute_xx are supported. This allows for quicker builds,
at the expense of the CUDA driver running ptxas the first time a kernel is loaded.
Differential Revision: https://developer.blender.org/D5953
Now local collections are fully working with cycles preview, while the
collection visibility bug is fixed.
Local collections were not working with cycles viewport even before the recent
commit to allow users to show collections that are hidden in the view layer.
It just got worse with said commit (0812949bbc3d).
Differential Revision: https://developer.blender.org/D6034
disabled and no world assigned to scene
BlenderSync::sync_world still relied on a blender world (mixes the world
viewport color with the studio light), for now just take black if no
world is present.
Maybe we should we use the theme color in the future here (seems eevee
does this in that case) -- we'd have to pass down `b_userpref` from
`BlenderSession::render` down to `sync_data > sync_shaders > sync_world`
then afaics.
Reviewed By: jbakker, brecht
Maniphest Tasks: T70573
Differential Revision: https://developer.blender.org/D6005
The clamp option in the Map Range node doesn't work correctly when the
inputs are linked. The code didn't put that into considration.
Reviewers: brecht
Differential Revision: https://developer.blender.org/D5987
Rendering would produce invalid results or crash if the Vector pass was active but motion blur was inactive. This caused the OptiX BVH to be built with motion (because objects reported motion available), but the pipeline to be built without motion support (since with disabled motion blur this is not in the list of requested features). The two are not compatible and therefore caused issues. This patch fixes that by not building the BVH with motion if motion blur is not active (which makes sense).
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D5968
Curves with motion blur produced wrong results with OptiX (T69801). This is because the AABBs for the motion steps were calculated from incorrect attribute data because the offset into the attribute data array was incorrect.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D5961
The OptiX device only loads the denoising kernels when the "use_denoising" feature is active. This was not set by the calling code however and therefore they were never loaded and attempting to launch them failed (see T69801).
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D5946
The "optix_devices" array was not freed on exit, which caused a memory leak (see T69801).
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D5944