This bumps OSL to 1.11.10.0. OSL Has a new build time
dependency: Clang, and more importantly it expects
clang and llvm to share a library folder, which it
previously for us did not.
This patch changes:
-OSL Update to 1.11.10.0
-refactor the llvm/clang/clang-tools-extra builds into the llvm
build using the llvm-project tarball for building that has all
of the subprojects in it.
-update ispc/openmp builds since clang no longer its own dependency
and they have to depend on the llvm build now.
-Update the windows builder to use the 64 bit host tools since it
ran out of ram linking clang
-Since OSL now needs clang to link successfully a findclang.cmake
has been provided for linux/OSX
Differential Revision: https://developer.blender.org/D10212
Reviewed By: brecht, sebbas, sybren
Specular color is set to black instead of white inside the Principled BSDF
when the base color is set to fully black. This is contradictory to the sample
code of the Disney BRDF in BRDF Explorer. This patch aligns both
implementations.
Differential Revision: https://developer.blender.org/D10448
With very large distances there were precision / overflow errors, normalize
the average albedo to avoid that. This was causing test failures on macOS
Arm, but also other architectures had slightly wrong results.
Ref T78710
Cycles has supported path-traced subsurface scattering for a while, but while it's
more accurate than other approaches, the increase in noise makes it an expensive option.
To improve this, this patch implements Dwivedi guiding, a technique that is based on
zero-variance random walk theory from particle physics and helps to produce shorter
random walks with more consistent throughput.
The idea behind this is that in non-white materials, each scattering event inside the
medium reduces the path throughput. Therefore, the darker the material is, the lower the
contribution of paths that travel far from the origin is.
In order to reduce variance, Dwivedi guiding uses modified direction and distance sampling
functions that favor paths which go back towards the medium interface.
By carefully selecting these sampling distributions, variance can be greatly reduced, and
as a neat side effect shorter paths are produced, which speeds up the process.
One limitation of just blindly applying this is that the guiding is derived from the
assumption of a medium that covers an infinite half-space. Therefore, at corners or thin
geometry where this does not hold, the algorithm might lead to fireflies.
To avoid this, the implementation here uses MIS to combine the classic and guided sampling.
Since each of those works on one of the three color channels, the final estimator combines
six sampling techniques. This results in some unintuitive math, but I tried to structure
it in a way that makes some sense.
Another improvement is that in areas where the other side of the mesh is close (e.g. ears),
the algorithm has a chance to switch to guiding towards the other side. This chance is based
on how deep the random walk is inside the object, and once again MIS is applied to the
decision, giving a total of nine techniques.
Combining all this, the noise of path-traced subsurface scattering is reduced significantly.
In my testing with the Rain character model and a simple lighting setup, the path-traced
SSS is now actually less noisy than the Christensen-Burley approximation at same render time
while of course still being significantly more realistic.
Differential Revision: https://developer.blender.org/D9932
Baking vertex colors per-corner leads to unwanted discontinuities when there is
sampling noise, for example in ambient occlusion or with a bevel shader node for
normals. For this reason the code used to always average results per-vertex.
However when using split normals, multiple materials or UV islands, we do want to
preserve discontinuities. So now bake per corner, but make sure the sampling seed
is shared for vertices.
Fix T85550: vertex color baking crash with split normals, Ref D10399
Fix T84663: vertex color baking blending at UV seams
Double floating point precision is an extension of OpenCL, which might
not be implemented by certain drivers, such as Intel Xe graphics.
Cycles does not use double floating point precision, and there is no
need on keeping doubles unless there is an explicit decision to use
them.
This is a simple fix from Cycles side to replace double floating point
type with a type of same size and alignment rules. Inspired by Brecht
and Patrick.
Tested on NVidia Titan V, Radeon RX Vega M, and TGL laptop.
Differential Revision: https://developer.blender.org/D10143
In my testing this works, but it requires me to remove the min(start_sample...) part in the
adaptive sampling kernel, and I assume there's a reason why it was there?
Reviewed By: brecht
Maniphest Tasks: T82351
Differential Revision: https://developer.blender.org/D9445
Commit d259e7dcfb increased the instance limit, but only provided
a fall back for the host code for older OptiX SDKs, not for kernel code. This caused a mismatch when
an old SDK was used (as is currently the case on buildbot) and subsequent rendering artifacts. This
fixes that by moving the bit that is checked to a common location that works with both old an new
SDK versions.
For a while now OptiX had support for 28-bits of instance IDs, instead of the initial 24-bits (see also
value reported by OPTIX_DEVICE_PROPERTY_LIMIT_MAX_INSTANCE_ID). This change makes use of
that and also adds an error reported when the number of instances an OptiX acceleration structure is
created with goes beyond the limit, to make this clear instead of just rendering an image with artifacts.
Manifest Tasks: T81431
This is relatively expensive and as per the OSL spec, this value is not
expected to be meaningful for non-light shaders. This makes viewport updates
a little faster.
As a side effect also fixes T82723, viewport refresh issue with volume density.
Adds support for building multiple BVH types in order to support using both CPU and OptiX
devices for rendering simultaneously. Primitive packing for Embree and OptiX is now
standalone, so it only needs to be run once and can be shared between the two. Additionally,
BVH building was made a device call, so that each device backend can decide how to
perform the building. The multi-device for instance creates a special multi-BVH that holds
references to several sub-BVHs, one for each sub-device.
Reviewed By: brecht, kevindietrich
Differential Revision: https://developer.blender.org/D9718
The SVM AO node calls "scene_intersect_local" with a NULL pointer for the intersection
information, which caused a crash with OptiX since it was not checking for this case and
always dereferencing this pointer. This fixes that by checking whether any hit information
was requested first (like is done in the BVH2 intersection routines).
Support for the AO and bevel shader nodes requires calling "optixTrace" from within the shading
VM, which is only allowed from inlined functions to the raygen program or callables. This patch
therefore converts the shading VM to use direct callables to make it work. To prevent performance
regressions a separate kernel module is compiled and used for this purpose.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D9733
NanoVDB includes "assert.h" and makes use of "assert" in several places and since the compile
pipeline for CUDA/OptiX kernels does not define "NDEBUG" for release builds, those debug
checks were always added. This is not intended, so this patch disables "assert" for CUDA/OptiX
by defining "NDEBUG" before including NanoVDB headers.
This also fixes a warning about unknown pragmas in NanoVDB thrown by the CUDA compiler.
This separates out PugiXML that was previously
bundled by OIIO.
As this linux/mac libs are not available
this commit only contains the builder and windows
changes, and the option to enable pugixml is
guarded by a platform if, this can be removed
once all platforms have committed the svn libs.
For details see D8628
Recent changes introduced `acc` parameter into the texture read
functions. When nanovdb isn't enabled this leads to compilation errors
as the `acc` variable wasn't defined. OpenCL only compiles needed
features what made it more prominent.
Reviewed By: Patrick Mours
Differential Revision: https://developer.blender.org/D9629
ROCm 3.9 already defined `NULL`. This patch will first check if it was
already defined to remove compilation warnings.
NOTE: This doesn't add official support for ROCm as it still fails to
render correctly (crashes with default cube).
Reviewed By: Brecht van Lommel
Differential Revision: https://developer.blender.org/D9610
The Normal Map node was falling back to (0, 0, 0) when it was missing
the required attributes to calculate a new normal.
(0, 0, 0) is not a valid normal and can lead to NaNs when it is
normalized later in the shader. Instead, we now return sd->N,
the unperturbed surface normal.
The names of the parameters are based on those of those of the sockets, so they also need to be updated. This was forgotten about in the previous commit (rBa284e559b90e).
Ref T82561.
There were some changes to the NanoVDB API that broke the way Cycles was previously using it.
With these changes it compiles successfully again and also still compiles with the NanoVDB revision
that is currently part of the Blender dependencies. Ref T81454.
forceinline attribute is only applicable for function which are
marked inline. Interestingly, it can be used for class methods
without explicit inline statement. But for functions it is another
story.
Volumes using tricubic sampling were producing different results with NanoVDB compared
to dense textures. This fixes that by using the same tricubic sampling algorithm in both
cases. It also fixes some remaining offset issues and some minor things that broke OpenCL
kernel compilation on NVIDIA.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D9491
The NanoVDB sampling implementation behaves different from dense texture sampling, so this
adds a small offset to the voxel indices to correct for that.
Also removes the need to modify the sampling coordinates by moving all the necessary
transformations into the image transform. See also T81454.
Cycles defines some basic integer types since it cannot use the standard headers when
compiling with NVRTC. NanoVDB however only does this when the "__CUDACC_RTC__" define
is set and otherwise includes the standard "stdint.h" header which clashes with those typedefs.
So for compatibility do the same thing in the Cycles kernel headers. See also T81454.
The existing code for this was incomplete. Each instance can now have a set
of attributes stored separately from geometry attributes. Geometry attributes
take precedence over instance attributes.
Ref D2057
This avoids OpenCL inlining heavy volume interpolation code once for every
data type, which could cause a performance regression when we add a float4
data type in the next commit.
Ref D2057
Corrects incorrect usage of contraction for 'it is', when possessive 'its' was required.
Differential Revision: https://developer.blender.org/D9250
Reviewed by Campbell Barton
