There is no need to have both `BKE_pbvh_api.hh` and `BKE_pbvh.hh`.
This commit removes the latter, moving the few remaining items to
the former, which will be renamed later.
There are two issues here. First, the instanced panel freeing uses
the panel type after it's been freed a few lines up in this function.
Second, while panel types are registered for a single space type,
there are cases where they're used in multiple space/region types
anyway. Because of that, removing the dangling pointer of the freed
type from just the registered region & space type combination isn't
enough, we need to process all regions.
Pull Request: https://projects.blender.org/blender/blender/pulls/129676
Running Xcode memory graphs and the Instruments tools revealed
memory leaks caused, in the main, by over-retained objects.
This removes the unnecessary 'retains' and adds some asserts
to guard against over-retaining in the future.
There are a few memory leaks remaining involving PyUnicode_DecodeUTF8
but I am unable to identify the cause of these at this time.
Authored by Apple: James McCarthy
Pull Request: https://projects.blender.org/blender/blender/pulls/129117
Similar to 5e46e3d28a.
This commit replaces the C-API version of `OpenSubdiv_Evaluator`
with direct calls to `EvalOutputAPI`. This removes a level of indirection,
theoretically reducing function call overhead, but also making the whole
system easier to understand and easier to modify. The downside is
further spread of `WITH_OPENSUBDIV` into the code, but I think that
can be improved in the future relatively easily once more of this sort
of change is finished.
Pull Request: https://projects.blender.org/blender/blender/pulls/128278
There were lot of places, where timeline frame was incorrectly
mapped to frame index.
For deleting transitions, internal function `seq_retiming_add_key`
was added. It can work in frame index domain directly. This is to avoid
adding more complexity, since key positions are stored and re-created
later. Going through index to frame and back to index would introduce
float precision errors.
Pull Request: https://projects.blender.org/blender/blender/pulls/129305
This commit applies the following changes:
* Use c++ math types instead of float / int arrays where possible
* Use c++ math library where c++ math types are used
* Use const where possible
* Reduces scope of variables where possible
* Clarifies some variable naming (e.g. d prefix to _delta suffix)
The main areas left untouched are anything that is part of a public
function signature to avoid making changes in many different files at
once.
Pull Request: https://projects.blender.org/blender/blender/pulls/129576
The recent refactor of sculpt mode drawing (d601bf7e) removed usage of
the update frustum planes. This makes storage of this data inside the
PBVH also unnecessary. If this feature needs to be reimplemented in the
future, the storage of the planes should be outside of the PBVH `struct`
Pull Request: https://projects.blender.org/blender/blender/pulls/129649
`SculptSession` is a runtime struct that should really be stored in
`ObjectRuntime` instead of `Object`. Writing pointers to it in files
is misleading because the struct itself isn't serialized.
Pull Request: https://projects.blender.org/blender/blender/pulls/129233
The issue was that when inserting a key, other keys weren't deselected
even though that feature was implemented with 6ef77a0d22
That only happened if no FCurve channel was selected in the channel list.
It turns out, that the intent of using the animfilter flags from
the editor was wrong. The animfilter code already checks the editor
flags based on the `bAnimContext` that is passed in.
In fact the drawing code itself has hardcoded flags.
So the fix is to use hardcoded flags as well.
Pull Request: https://projects.blender.org/blender/blender/pulls/129636
Changing 3D curve properties while viewport rendering was active
resulted in an error, because Cycles would attempt to update the
acceleration structure containing the curves, but that acceleration
structure was built without the
`OPTIX_BUILD_FLAG_ALLOW_UPDATE` flag allowing updates. This
fixes that by adding the flag to all curve build inputs.
Ideally could just use the same flags as for other build inputs and
differentiate between viewport and final rendering (based on
`bvh_type`), but that's not currently an option since the same flags
have to be specified to query the curve intersection module in
`load_kernels()`, where that differentiation is not known. See also
commit 5c6053ccb1.
Pull Request: https://projects.blender.org/blender/blender/pulls/129634
The primitive tool didn't write to the `fill_opacity` attribute meaning that primitives wouldn't
show up correctly when using fill materials with a strength < 1.
The fix does multiple things:
* Make sure to create and write to the `fill_opacity` attribute if necessary.
* Remove the `skipped_attribute_ids` function and build the set of attributes to skip
where the attributes are written to. This is more flexible.
Pull Request: https://projects.blender.org/blender/blender/pulls/129644
Snapping the 3D cursor to a grease pencil selection would include
locked layers in the computation of the centroid. This is inconsistent
with 4.2.
The fix makes sure that we skip over layers that are locked.
Pull Request: https://projects.blender.org/blender/blender/pulls/129639
This commit removes support for Vega GPUs from the AMD HIP backend of
Cycles. This is being done as:
- AMD no longer provides official support for Vega GPUs in their
ROCm software.
- Vega GPUs have rendering artifacts on all supported platforms,
and as a result of the reduction of support from AMD, are unlikely
to be fixed. Rendering artifacts include.
- The incorrect shading of volumes (Windows and Linux)
- Missing intersections on many meshes with HIPRT
- Crashing rendering subsurface scattering materials (Linux)
- And more.
Pull Request: https://projects.blender.org/blender/blender/pulls/129523
Previously, we initialized the layers and groups to hide masks.
For layers this makes sense, because they don't have any masks
by default, but for groups it makes more sense to show masks
by default for all the layers inside.
Pull Request: https://projects.blender.org/blender/blender/pulls/129528
The default value for `FLOAT_COLOR` attributes is white. We can't
change this default easily.
This fix will initialize the attributes accessed through the high-level
python API with their expected default value. In the case of
vertex colors, this is fully transparent black.
Pull Request: https://projects.blender.org/blender/blender/pulls/129638
In lieu of fixing the root cause, this commit requires the object to be
in object mode to duplicate particle systems, which works around the
issue because the problem is with particle edit data. This is a pragmatic
choice to focus development efforts on replacing the particle system.
Example:
Step is 3
Action starts at 1, so the action will try to export frames 1 / 4 / 7 / 10, etc...
But full object backing is enable, and scene starts at 0.
So cached data are on frame 0 / 3 / 6 / 9 / 12, etc...
When we are trying to export the action on armature after the full bake on empty, we are trying to retrieve frame 1, but cache was on 0 / 3 / 6 / 9
error log will go to stderr, and others to stdout
A new parameter in operator to set the loglevel from command line bpy.ops.export_scene.gltf(loglevel=logging.INFO)
If not set from the ops, value is -1, and is set, as before, from the bpy.app.debug_value
