We forgot to update this code as part of D3108. I'd like to include this in 2.90,
it's entirely broken now so can't really get any worse.
Differential Revision: https://developer.blender.org/D8738
The previous fix loaded the pixels so existing tiles were not overwritten.
However the Cycles render buffer is expected to be scaled by the number of
sample, which was not taken into account.
This is not ideal in that previews could have a mismatched number of samples
between multiple objects, though the result will be correct. The better solution
would be to bake all objects together per tile, rather than one after the other.
But that is a bigger change than we can do in 2.90.
Differential Revision: https://developer.blender.org/D8704
The return value of scene_intersect must be checked, the isect struct members
can't be assumed to be initialized if that returns false.
Differential Revision: https://developer.blender.org/D8692
For 2.90 release this should not be exposed in the RNA API.
In master this needs to be ON by default, that's all.
Differential Revision: https://developer.blender.org/D8589
Mistake in cb578ca104. Before that, the extension vector was static,
to make sure the extension name strings wouldn't get destructed when
leaving the function. I didn't think that was an issue and couldn't
recreate one, because until the previous commit we wouldn't actually
add any extensions to the vector on Windows (the system I tested
with).
Use C++17's `std::string_view` now, which avoids the string copies
`std::string` creates for itself and thus its destruction when leaving
the local scope.
The OpenXR debug extension was disabled on Windows as a workaround. This
was an old leftover from when there was only the Windows Mixed Reality
runtime on Windows. The debug extension didn't work for it and we didn't
have a way to disable it just for Windows Mixed Reality.
Now it seems to work though, so we remove the workaround. If specific
runtimes still have trouble with the extension, we can disable it
specifically for these runtimes now.
Windows only workaround. I'll have to investigate Linux separately.
Steam's OpenGL compatibility is still new and doesn't work for us yet
(neither does it for standard OpenXR examples from what I've heard and
seen myself). We can work around that by falling back to our DirectX
compatibility layer.
Note that this DirectX compatibility still doesn't work for some
systems, see T76082.
Implementation note: Since the graphics binding extensions have to be
enabled before we can find out which runtime is in use (e.g. SteamVR vs.
Oculus, etc), we can now enable multiple graphics binding extensions but
settle for a single one to use later.
Once the SteamVR OpenGL backend works, we can remove this workaround
again.
Fixes T78267.
The previous algorithm was not using all of the requested grids to build a mesh
around the volume due to limitations regarding the use of a dense buffer to
gather information about the volume's topology. This resulted in artefacts during
rendering.
The mesh generation is now done by merging all of the input grids and using the
resulting grid's topology to create the mesh. The generation of the mesh
is still done in index space as before, and the vertices are converted to object
space by using the merged topology grid indexToWorld transform.
To be able to merge the grids together we have to make sure that their transformation
matrices and their index spaces match, thus, if they do not match we simply resample
the grids. This behaviour should tackle one other limitation of the current algorithm,
which is that only one transformation matrix was used to generate the final mesh.
If we do not have an OpenVDB grid for the requested volume data, we generate
a temporary OpenVDB grid for it.
Differential Revision: https://developer.blender.org/D8401
We can't exactly follow what we do for macOS here. On Windows special
characters can be inserted with Ctrl+Alt. So make sure we expect UTF-8
characters when Alt is held.
Mistake in 87062d4d67.
In an upcoming bugfix we'll use OpenVDB data structures directly to build mesh
for sparse OpenVDB volumes, loading them OpenVDB grids earlier and removing any
references to Blender data structures makes that easier.
This also makes changes to Blender volumes to support this, so Cycles can take
ownership of a grid without Blender having to keep its own reference to it.
This should also be useful in a future Python API.
Ref D8401
On windows, spacebar would be passed as UTF-8 text input, despite the
control key being pressed. On macOS, there already was an explicit
exception for this (command key in this case), on Linux XInput already
handled this case for us.
Note that Alt should still allow text input, for special character
sequences.
Issue also happened in the Text Editor if a text data-block was set.
Optimization was disabled in this function to work around a bug in MSVC, use
a different solution that does not come with such a big performance regression.
The decomposed transform would have consists of nan values if the input
transform had zero scale.
Now the decomposition will check for zero scale, and if it is detected
then the result will be ensured to be finite. Additionally, rotation
value will be copied from previous/next time step to help avoiding
obscure interpolation.
The latter step can become more comprehensive than the current simple
implementation.
Differential Revision: https://developer.blender.org/D8450
Caused by c7aa0f9d74.
Since above commit, BKE_image_user_frame_calc requires an image (not
just the iuser) to get the framenumber.
Cycles used to call this with NULL image (in `image_user_file_path` and
`image_user_frame_number`), now pass the image as well.
Maniphest Tasks: T79111
Differential Revision: https://developer.blender.org/D8439
Fix several issues in CLOG code:
* In `clg_str_reserve`, allocated memory may be bigger than requested
one, do not assign the latter back to `cstr->len_alloc`.
* `clg_str_vappendf` was mis-interpreting returned value from
`vsnprintf`, and completely mixing total allocated memory and extra
needed amount of memory to allocate...
Simplified code of `clg_str_vappendf` to only have allocating code
handled in one place, makes things easier to follow too.
Think this should also be beckported to 2.83.
The fisheye camera setup causes the edges of the image to not shoot primary rays. This was not
respected by OptiX because of an optimization that tried to reduce conditionals around trace calls.
Removing that does not seem to have an impact on performance anymore however and it fixes
the issue.
New option that lets users the define the maximum number of fluid particles that will be allowed in the simulation. This can come in handy, for example, to ensure that the particle count will not exceed the hardware capabilities, or to avoid excessive amounts of particles in a scene.
The OptiX kernels are compiled for target "compute_sm_52", which is only available on second
generation Maxwell GPUs, so disable support for older ones.
When definining static variables that own memory, you should
use the "construct on first use" idiom. Otherwise, you'll get
a warning when Blender exits.
More details are provided in D8354.
Differential Revision: https://developer.blender.org/D8354
The sky will appear brighter than before by default. To compensate for this,
lower exposure in the Film panel. The default altitude was also changed from
90 to 15 degrees.
Patch contributed by Marco with the help of Ryan Jones.
Differential Revision: https://developer.blender.org/D8285
This patch changes the discovery of pre-compiled kernels, to look for any PTX, even if
it does not match the current architecture version exactly. It works because the driver can
JIT-compile PTX generated for architectures less than or equal to the current one.
This e.g. makes it possible to render on a new GPU architecture even if no pre-compiled
binary kernel was distributed for it as part of the Blender installation.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D8332
This refactor is in response to reports in which the adaptive domain with noise caused a crash (e.g. T79009). It should also fix issues where the smoke appeared to be cut off when using the adaptive domain together with noise. It is also possible that some of these changes improve the lines issue from T74559.
This commit introduces a new way to build unit tests. It is now possible
for each module to generate its own test library. The tests in these
libraries are then bundled into a single executable.
The test executable can be run with `ctest`. Even though the tests
reside in a single executable, they are still exposed as individual
tests to `ctest`, and thus can be selected via its `-R` argument.
Not yet ported tests still build & run as before.
The following rules apply:
- Test code should reside in the same directory as the code under test.
- Tests that target functionality in `somefile.{c,cc}` should reside in
`somefile_test.cc`.
- The namespace for tests is the `tests` sub-namespace of the code under
test. For example, tests for `blender::bke` should be in
`blender::bke:tests`.
- The test files should be listed in the module's `CMakeLists.txt` in a
`blender_add_test_lib()` call. See the `blenkernel` module for an
example.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D7649
These flags need to be set correctly in order to distinguish between data that comes from cache files and raw data that comes directly from pointers to the data in Mantaflow.
