The Map UV and Displace nodes produce unexpected outputs. That's because
the derivatives computed for anisotropic filter were computed in the
sampler's space, while it should be in texel space, as expected bu the
textureGrad function.
In larger scenes Blender could crash when duplicating frames in GPv3.
This was caused by a dangling reference in `insert_duplicate_frame`.
The source frame could become invalid when the frames map was
reallocated due to `layer.add_frame(dst_frame_number, ...)` a few lines
later in the code.
The fix replaces the reference and assigns by value.
Pull Request: https://projects.blender.org/blender/blender/pulls/121414
This updates OIDN to 2.3.0-beta. The final version is planned to be
released in time for Blender 4.2 Beta. The most relevant changes:
1. Much higher quality when denoising with accurate prefiltering in *high*
quality mode, but at the cost of lower performance (use *balanced*
quality mode to revert to previous *high* quality mode)
2. Added new *fast* quality mode for 1.5-2x higher performance viewport
denoising
3. Lazy device module loading to avoid potential stability issues caused by
drivers of unused devices
4. Release CUDA primary context as soon as the OIDN device gets destroyed
to prevent potential memory leaks
To enable 2 and 3, code changes are needed in Blender, to be committed
separately.
Ref #118455
Pull Request: https://projects.blender.org/blender/blender/pulls/121359
Change the preview size comparison threshold, so that we only use linear
interpolation for the render preview when zoomed out, and use nearest
neighbor interpolation at 1:1 zoom.
Pull Request: https://projects.blender.org/blender/blender/pulls/120325
Grease Pencil overlay drawing code uses the existence of shading groups
to determine which parts of the overlay need to be drawn. If the tool
settings change the shading groups need to be set to null, otherwise
the drawing code will still draw points in curve selection mode.
Pull Request: https://projects.blender.org/blender/blender/pulls/121472
The Corner Pin node as well as the Plane Track Deform nodes always
return a single color that appears to be the average of the input.
That's because the derivatives were computed in the sampler's space,
while they should be in texel space. Large derivatives meant that the
textureGrad function would always sample the lowest MIP level, hence the
constant average color.
This might be an issue with other uses of textureGrad in the compositor,
so their use should be investigated.
The Blur node takes too long to execute even though it is in a simple
configuration. That's because the CPU compositor uses variable size
blurring even if the size is constant. So ensure the input size is
actually variable before using variable size blurring.
Without this canceling cppcheck part way through meant comparing old/new
log-files wasn't possible without manual renaming.
Now the new log-files are only written (and previous execution renamed
to ".old.log") once generating the log has completed.
Previously, menu sockets were sometimes shown as integers or strings
in socket tooltips. Now, they are always shown as "Menu" type. This also
changes how these values are logged. Previously, they were logged as
strings. Now, only the integer identifier is logged and the name is looked
up when drawing the tooltip.
Pull Request: https://projects.blender.org/blender/blender/pulls/121236
Material index of faces could be a invalid value after booleaning in
exact boolean in "Index Based" mode against objects with no material
slots, this is caused by the lack of initilization material remapping
array. Now all initilized to 0.
Assigning 0 explicitly in Fast mode as well for consistency.
Pull Request: https://projects.blender.org/blender/blender/pulls/121380
When moving importers from Python to C++ validation was removed
(from PLY) & disabled by default (STL & OBJ) which re-introduces
crashes reported by users such as #31835.
Enable validation by default because crashes caused by imported data
should be avoided, especially since the crashes may happen later when
users enter edit-mode or run certain editing operations.
This does slow down importing, from testing a 236MB .OBJ it takes around
twice as long to import (~1.5 to ~3 seconds). Although validation can
be optimized to reduce the overhead ad well as run in parallel for
importers that load multiple objects.
The defaults for USD and Alembic remain unchanged since this was never
enabled by default, although we could consider enabling these as well.
Rather than storing asset shelf types only for the space type defined
with `bl_space_type`, store them statically/globally so that asset shelf
popups can later be displayed in any space. In earlier designs the asset
shelf was more space dependent than it is now.
For example the brush selector asset shelf popup being implemented for
the brush assets project should be available in the tool settings UI,
which is also displayed in the Properties. If the asset shelf type is
space dependent, this popup won't work in the Properties.
To be clear, `bl_space_type` will still be what defines where the asset
shelf region will be available. But the popup version can be displayed
anywhere.
Pull Request: https://projects.blender.org/blender/blender/pulls/121411
Add a function for asset shelves, `get_active_asset` in RNA, that
can be used to sync the active item of an asset shelf with an asset
reference elsewhere in Blender. In the brush asset system this is
used to retrieve the asset reference from the active `Paint` struct.
Pull Request: https://projects.blender.org/blender/blender/pulls/121405
The operator stored when registering the asset shelf will be called
with properties that reference the asset (a weak reference, split into
three properties already used in other asset-related operators). The
operator is called whenever a grid view item is activated, and is
meant to do things like import and activate a brush asset.
Previously Reviewed: https://projects.blender.org/blender/blender/pulls/117861
Pull Request: https://projects.blender.org/blender/blender/pulls/121402
This adds the `Draw on back` option to GPv3.
Note:
In the legacy system the stroke being drawn would be displayed
in front and only moved to the back upon finishing the operator.
Whereas this will display the stroke on back throughout the
drawing operation.
Pull Request: https://projects.blender.org/blender/blender/pulls/120894
Adds a new asset shelf option (`STORE_ENABLED_CATALOGS_IN_PREFERENCES`
option in RNA) to use the Preferences for storing the enabled catalogs.
This way asset shelf types can decide if for their use-case, they want
to synchronize the enabled catalogs over Blender sessions and files, or
keep the stored locally in the file.
This is important because for example on one hand, it would be annoying
if for brush assets you'd have to enable the visible catalog tabs for
every 3D View and every file, while on the other hand you need that
level of control for the pose library where the catalogs the rigger/
animator cares about varies from project to project, character to
character and shot to shot.
Conceptually this also makes some sense: The new brush assets workflow
synchronizes brush assets and their catalogs across Blender sessions
and files, basically making them globally accessible independent of
the current file/project, so treating the enabled catalogs the same
is consistent.
Previously reviewed in #120264
Pull Request: https://projects.blender.org/blender/blender/pulls/121363
When Blender is built using `WITH_EXPERIMENTAL_FEATURES=OFF`, the GP
custom properties panel still tries to access `bpy.types.GreasePencilv3`,
which then doesn't exist.
This is a very simple, local fix that checks whether that type exists,
and only then defines `DATA_PT_grease_pencil_custom_props`.
Pull Request: https://projects.blender.org/blender/blender/pulls/121394
This assertion was added in assumption that sizes of input argument are always correct.
But this is not such. More correct is to depend on assertions of span access methods.
The size of the multi-function parameters is only guaranteed to be large enough so that
every index in the mask can be accessed. It may be larger.
Pull Request: https://projects.blender.org/blender/blender/pulls/121393
Remapping of internal ID pointers in 'undo swap' case (i.e. when new ID
data had to be read from undo buffer, but current ID address is kept)
needs to also ignore 'fake user' handling, otherwise it would reverse
the undoing effect.
Previously, the Realize Instances node would always realize everything in the geometry.
Now it's possible to more selectively realize parts of the geometry:
* One can choose which top level instances should be realized.
* The realization depth can be controlled per top-level instance. For example, if there are
5 top level instances, each of which contains 10 other instances, it's now possible to
realize only one level so that one ends up with 50 instances.
Pull Request: https://projects.blender.org/blender/blender/pulls/116582
All official Blender platforms use the SIMD code path, with pow() approximations
for 2.4 and 1/2.4 powers. The non-SIMD code path would only be used on other
platforms like PowerPC etc. Make that fallback scalar code path use the same
math approximation for consistency. This is part of #121312.
This also makes srgb_to_linearrgb_v3_v3 and linearrgb_to_srgb_v3_v3 functions
non-inlined. They are 50-100 CPU instructions, and thus hardly good candidates
for forced inlining into each and every call site.
Also _bli_math_blend_sse now uses actual SSE4 blend instruction instead of doing
it in a roundabout way.
Pull Request: https://projects.blender.org/blender/blender/pulls/121368
Issue was in the fact that log should be propagated on source node and on parent node.
Instead of `group_node_id`, loggers need to have id of actual parent (zone node or node group) node.
This also fix issue with timings propagation on frame nodes.
Pull Request: https://projects.blender.org/blender/blender/pulls/120842
