When creating depsgraph relationships for drivers, avoid creating the
'unshare' depsgraph node for drivers on custom properties and on
shapekey `value` properties.
This should fix a significant part of the performance regression
mentioned in #140706.
Pull Request: https://projects.blender.org/blender/blender/pulls/140724
Resolve reference counting error in the RNA API which mixed up ownership
between ExtensionRNA::data & StructRNA::py_data.
In practice this generally worked as RNA_struct_free_extension behaved
as if ExtensionRNA owned the reference instead of StructRNA, so as long
as only one reference was removed, there wasn't any difference.
When re-registering Python classes with RNA, ownership wasn't handled
properly which could result in negative reference counts,
crashing in some cases.
After using the Purge Unused Data operator, hovering mouse over outliner
makes outliner tree dirty as the tree is not being rebuilt. The operator is
defined in python which calls `ANIM_armature_bonecoll_remove`. Add
notifier in that function code so that `outliner_main_region_listener`will tag
the outliner region for full redraw.
Pull Request: https://projects.blender.org/blender/blender/pulls/140256
Python 3.14 has moved some functionality into the public API,
use the updated names even with older Python versions.
Also resolve an error caused by variable reuse with delayed annotation
evaluation for TextureProperties_MixIn on startup.
Resolve#140695.
When dragging an area from one window to another it is possible for the
bScreen->active_area to be be invalid for a short period of time. This
is constantly updated so normally not noticeable but will cause ASAN
errors when trying to show keymaps in the status bar as this is use
after free. This PR just sets these to nullptr.
Pull Request: https://projects.blender.org/blender/blender/pulls/140683
`UI_icon_from_idcode` is used e.g. from the Action Editor to indicate an
Action's Slot Type (and without the icon it is kinda hard to tell what
the slot is suited for...)
Before
After
Discovered while investigating #140618
Pull Request: https://projects.blender.org/blender/blender/pulls/140658
After a recent commit, multiple drawings were always written using the
same address because we used the same stack memory in a loop. This
causes the blend file reading to not be able to distinguish between the
structs, meaning the file is corrupt. However, we already had the same
problem in some cases because of the inline buffer in `BlendWriteData`.
To resolve this, make all "temporary" data for writing live as long as
the writing is going on for the ID. This is somewhat inefficient since
it makes memory reuse impossible for this temporary data. In the future
we should use a technique like #127706 to address this.
For testing, I saved and loaded multiple production files with Grease
Pencil objects, with and without the "write with attribute storage"
option enabled. The fix still goes to 4.5 though, because the first
mentioned issue is present there, and the `CustomDataLayer` vector
address reuse is potentially a problem too.
Pull Request: https://projects.blender.org/blender/blender/pulls/140667
Fix#79163 bug related to the bevel operation producing disconnected UVs for
new bevel faces. This change replaces previous approach using scattered and
selective usage of functions: bev_merge_uvs, bev_merge_edge_uvs and
bev_merge_end_uvs with one coherent technique for all stages of the bevel operation.
It is utilizing a concept of loop (BMLoop) buckets to keep track of UV vertices
that should be merged at the end of bevel operation by a single call to
bevel_merge_uvs function. This approach doesn't touch initial UV position
calculation done by interpolation algorithm in bev_create_ngon function and
keeps the concept of representative faces (called frep, facerep or rep_face in
code) to help decide to which bucket specific loops should be assigned.
This is from PR https://projects.blender.org/blender/blender/pulls/139595,
which has more explanation and discussion.
GreasePencil ID type did not list `ID_OB` as its potential ID dependency
type, leading to lack of remapping of its layers' parents objects
pointers (e.g. on Object deletion), leading to crash from accessing
freed data.
We detected that the gpu/draw tests could lead to
crashes after the tests where finished. It was
identified that the crash happened as the submission
runner wasn't exited, uses a mutex, that is being
destroyed by the main thread.
Solution is to wait until the submission runner has
finished, before destroying resources.
Pull Request: https://projects.blender.org/blender/blender/pulls/140650
Avoid modifying the order of transform system data. Instead, create an
index map and use that to traverse the data arrays in sorted order.
The issue observed in #139042 stems from the assumption, in _some_ of
the code, that `tc->data[i]`, `tc->data_ext[i]`, and `tc->data_2d[i]`
all contain information about the same "transformable thing". Since
`tc->data` was sorted (by selection state and, optionally for
proportional editing, by distance) but the other arrays were not, this
caused issues.
The most obvious solution, sorting all arrays the same way, turned out
to be hard to do, as some elements in one array have pointers to
elements in another array. Reordering those arrays would therefore
also make it necessary to find and update those pointers.
Instead, I decided to implement a sorted index map. The arrays can
then be kept in their original order, and the index map can be used to
visit them in sorted order.
Pull Request: https://projects.blender.org/blender/blender/pulls/140132
The issue was that the wrong flag was passed to
`ANIM_nla_mapping_apply_if_needed_fcurve` resulting
in the handles not moved back in this method.
This issue was only visible when pasting on top of a single keyframe,
or when the key had free handles. That is because the handles would
be recalculated otherwise.
Pull Request: https://projects.blender.org/blender/blender/pulls/140542
MESH_GGT_spin_redo (for the operator) depended on MESH_GGT_spin
(for the tool) having already been initialized however the order
gizmos are initialized currently isn't guaranteed.
Workaround the problem by adding a null check.
See the report for details.
Co-authored-by: Philipp Oeser <philipp@blender.org>
The logic to access the edit-mesh for the selection engine didn't match
the logic used for drawing.
Resolve using the edit-mesh from: BKE_object_get_pre_modified_mesh
Ref !140525
Some USD files have the displacement input specified in the shader but
do not have it marked as an output in either the shader or the material.
Skip processing if we do not have a valid displacement source.
Pull Request: https://projects.blender.org/blender/blender/pulls/140619
`blend_write_prepare` expects to work on a shallow copy of the
`CurvesGeometry` struct. This is necessary because it changes
the struct to contain only the layers being written. This is
particularly important with writing with the new format enabled,
where attributes are written as part of `AttributeStorage`.
Writing with a shallow copy of the struct also gives us the
flexibility to zero the runtime pointer in the written file which
is generally a good practice.
Pull Request: https://projects.blender.org/blender/blender/pulls/140560
The distributed memory access toggle in Cycles preferences would show up
when a user has two GPUs that can access each other's memory, but only one
of them is supported by Cycles.
For example the AMD RX 5700XT and AMD Vega 64 can access each other's
memory, but only the 5700XT is supported by Cycles.
Pull Request: https://projects.blender.org/blender/blender/pulls/140521
For UILists that doesn't use `filter_items()`, search filter doesn't work
along with invert button. For filter_items, invert case is already
handled in filter_items_by_name. In absense of that function,
handle the invert case inside `UI_list_filter_and_sort_items`. Existing logic
is a bit wrong. Now flags are adjusted, i.e. when search string matches, check
for invert status to decide whether items should be in the list or not.
Pull Request: https://projects.blender.org/blender/blender/pulls/139523
This was caused by a hash colision caused by an incorrect
auto conversion from `Object*` to `ObjectRef`. The `ObjectRef`
would then not contain any instancing data and would create
the same hash for the same particle system on each instance
(since only `foreach_hair_particle_handle` was constructing
`ObjectKey` from an `Object*` it did not affect the instance
itself).
Marking the incriminating constructor `explicit` to avoid
more issues. Changing the `ObjectKey` creation for psys to use
`ObjectRef` fixes the issue.
Pull Request: https://projects.blender.org/blender/blender/pulls/140544
The performance of the sorted_paths_array kernel on B570 is problematic.
Relying on local sorting+partitioning instead gives a 25% overall rendering
speedup and no regression in shade_surface when rendering Agent 327 Barbershop scene.
On Arc A770, it still gives a 2% speedup when rendering Barbershop.
Pull Request: https://projects.blender.org/blender/blender/pulls/140308
We should never select something that is hidden in the viewport.
Generally, there can be multiple reasons why keyframe handles are not
shown in the Graph Editor:
- the `View` > `Show Handles` option is disabled
- the `Only Show Selected Keyframes Handles` is enabled
- a keyframe `Interpolation` method is anything but `Bezier`
This PR corrects the first.
Details:
- In `box_select_graphkeys()`, the `incl_handles` parameter is only
be passed to `initialize_box_select_key_editing_data()`
- add a separate `KEYFRAME_ITER_HANDLES_INVISIBLE` flag (representing
SIPO_NOHANDLES)
- "include_handles" purely represents operator setting / user intend now
- "final" decision about handles/their visibility is done in
`keyframe_ok_checks()` etc
For consistency, add "include_handles" option to Lasso/Circle as well
Part of #139314
Pull Request: https://projects.blender.org/blender/blender/pulls/139349
Add an optional init function which operators
An alternative to [0] which missed Python API support (causing #140451).
While that could be resolved, tracking which "slots" have been set
would have to be flagged on every map/hash insertion which seems
excessive and is prone to bmesh operators failing if the flag is ever
missed. Prefer a simpler init function so dissolve edges doesn't have
a zero threshold.
Also support multi-line comment blocks in the generated API docs.
[0]: bd3a66a416
The "delete this F-Curve" function assumes that the F-Curve is either a
driver or contained in `adt->action`. This does not have to be true, as
the Action editor can also show F-Curves from other uses (for example an
NLA strip). The channel delete operator (`ANIM_OT_channels_delete`) now
takes this into account.
Pull Request: https://projects.blender.org/blender/blender/pulls/140291
LoongArch is a new RISC ISA developed by Loongson, this adds the missing
entry to BLI_build_config.h. It has been verified to work on Arch Linux's
loong64 port.
Pull Request: https://projects.blender.org/blender/blender/pulls/130916
Also true (and reported) for the general File Browser.
In the implementation from 5a67407d5a it was noted that including the
active file was desirable(next to `FileSelection` `first` and `last` of
course).
`FileSelection` indices get correctly updated after search is altered/
cleared (by `file_current_selection_range_get`), the `FileSelectParams`
`active_file` however isnt.
This makes the framing go wrong when search is altered or cleared (it
uses an index from the previous/outdated filter result).
I case of the Asset Browser (in main) this was also obvious when you
looked at an active asset in the sidebar -> clearing a search (after
having selected something while searching) for example would make a
"completely different" asset active.
Since keeping track of a "previous" `active_file` (and trying to update
to an updated/correct index) turned out to be quite hairy, this PR goes
the route of actually clearing the `active_file` on search updates. Like
mentioned, this is still not ideal (at least for the Asset Browser, it
looses the "active" of something selected), but at least now you can
easily frame it again -- and is is better than in main (where "active"
would point to something completely unrelated and you could not even
frame it to click-activate it again).
Pull Request: https://projects.blender.org/blender/blender/pulls/140365
Create a good default name for saving individual frames of a movie file loaded
as an image datablock, instead of the movie file name.
Changes ImBuf to store the frame separate from the filepath, to implement this.
Seems more clear for ImBuf.filepath to be an actual filepath anyway.
Thanks to Jesse and Aras for investigating this bug.
Pull Request: https://projects.blender.org/blender/blender/pulls/140471
Currently we have two entries in the Select menu:
- Box Select
- Box Select (Include Handles)
But since b037ba2665, Include Handles is the default, so both entries
behave exactly the same.
To resolve, make it:
- Box Select (Include Handles)
- Box Select
with the second one setting the "include_handles" option to False.
NOTE: behavior of "include_handles" option False might be a bit flaky in
certain situations as well (can be checked later), better to actually
have two entries that represent what the code does.
Noticed while checking on #139314
Pull Request: https://projects.blender.org/blender/blender/pulls/139347
