These updates are used to recalculate normals and average values between
grid boundaries during multires sculpting. In main the affected faces
are passed as an array of pointers. Using an `IndexMask` instead reduces
memory usage by 4x per affected face (8 byte pointer to 2 byte integer),
simplifies iteration and threading, and can also improve performance.
Finding which faces are affected is now multithreaded, with its runtime
changing from 0.63 ms to 0.12 ms in a simple test sculpting on a portion
of a 1 million face grid.
Also switch to VectorSet instead of GHash for finding affected adjacent
elements. That's a friendlier data structure that probably has better
performance as well.
`ED_view3d_win_to_ray_clipped_ex` requires `ray_end` when
`do_clip_planes` is true.
Since the snap code uses `ray_depth` instead of `ray_end`, the solution
is to always calculate the `ray_end` and convert to `ray_depth` when
necessary.
The new grid socket (#115270) will make these nodes obsolete and
provide more elegant ways of implementing the features. Removing
these nodes now to clean up and make future changes simpler.
Pull Request: https://projects.blender.org/blender/blender/pulls/115567
This avoids some duplication between the modifier and operator evaluation
contexts and also makes it easier to make independent from a specific
evaluation context (so e.g. the simulation nodes code shouldn't care whether
it's used from a modifier or operator.
Pull Request: https://projects.blender.org/blender/blender/pulls/115512
This makes so that locked layers don't show the points and the lines aren't visibly selected.
This separate the batch ensure function into two function, one for handling the visible grease pencil batch and one for handling the edit mode overlay batch.
This also makes it so that only selected curves show the points that make up it, (this matches legacy grease pencil).
Pull Request: https://projects.blender.org/blender/blender/pulls/114751
Because reading memfile undo steps doesn't send the mesh through the
versioning code that converts legacy layers back to the new generic
format, we can't write the old format in undo steps. This same check
existed during the mesh struct of array refactor.
Use blender::Set which is similar but offsers better type safety
and likely better performance as well. The only remaining user
was the mesh edit mode knife tool, and replacing that usage
with `Set` and `Map` was straightforward.
Inlining the functions is simpler nowadays, since there are utility
functions to copy spans and tag the mesh caches dirty. Also use an
array instead of a raw pointer for multires.
Resolves#103789
The issue was that bone colors are only specified with 3 channels (RGB),
but they're stored as 4 channels for some reason. So the fourth bogus
channel gets initialized to zero by default, which was then being
interpreted as zero alpha in some drawing modes.
This fixes the issue by ensuring that the fourth unused channel always
gets initialized to 255, and therefore can be validly interpreted as an
alpha channel.
Pull Request: https://projects.blender.org/blender/blender/pulls/115477
This utility was already duplicated in two places and planned to be used
more. While we should usually avoid creating arrays the size of the
indexed array (rather than the size of the mask), sometimes it does seem
to be the best option, and we're helped by the fact that most memory
stays unintialized for a small mask (allocating but not writing to memory
pages at all generally isn't too expensive).
Pull Request: https://projects.blender.org/blender/blender/pulls/115491
Description added in 9c2330d821.
It is unclear what "external" in this case means, why these asset
libraries are considered "normal" (are the others not "normal"?) and the
term "asset repository" is wrong (should be "asset library").
I think this isn't actually shown in the UI currently in fact.
Problem is the `UI_context_active_but_get()` lookup doesn't respect menu
regions, unlike the earier call to `UI_context_active_but_prop_get()`
(which is also used to validate the existence of an active button).
Using `UI_context_active_but_get_respect_menu()` instead would work, but
in fact the earlier lookup returns the active button if any, we just
didn't use that.
`wmDragActiveDropState` contains a `bContextStore` unique pointer, so
when constructing/destructing an object of this type `BKE_context.hh`
had to be included. It wouldn't be obvious why this include is necessary
in some cases. And the pointer is otherwise managed internally in the
source file, so the need for this include shouldn't bleed into other
files.
Make sure the unique pointer construction/destruction happens in the
translation unit of the source file that otherwise also manages it.
Avoids include in #107230.
New FCurves for bones were not automatically set to `FCURVE_COLOR_AUTO_RGB`
because the RNA path couldn't be resolved.
That was because the `PointerRNA` passed into the function points to the bone,
but the rna path is relative to the ID.
Fixed by resolving the path from the bones owner ID
Pull Request: https://projects.blender.org/blender/blender/pulls/115517
- Remove unnecessary exception if Py_CompileStringObject fails.
- Remove unnecessary PyErr_Occurred check if Py_CompileStringObject
succeeds (based on incorrect assumption from [0]).
- Pass closeit=1 to PyRun_FileExFlags so it's responsible for closing
the file before executing code.
- Remove unnecessary temporary ReportList.
[0]: b9f6d66328
Some Editors like the Filebrowser or Spreadsheet set up custom masks for
scrollbars (they dont cover the whole region width or height, some rows
or colums are excluded here -- e.g. the "index" column in the
Spreadsheet), this was working fine for drawing, but mouse interaction
was ignoring these custom masks.
Symptoms were that clicking on the "focus bubble" of the scrollbar and
dragging would not result in smooth scrolling [because internally code
would assume the mouse was clicking outside], instead scroll by 1 'page'
was happening [was perceived as a massive lag]. Especially noticable
with many rows/columns (where the "focus bubble" gets small).
So for drawing (`UI_view2d_scrollers_draw`) this wasnt a problem, masks
were passed in for `view2d_scrollers_calc`.
But for interaction (`scroller_activate_init`) `view2d_scrollers_calc`
is called again and a NULL mask was passed.
The already set up rects need to be considered though, otherwise coords
for `mouse_in_scroller_handle` later are not compatible.
To fix this, use existing `View2D` rects fors scrollbars, and pass a
union of those as `mask_custom` to `view2d_scrollers_calc`.
NOTE: this can temporarily extend the `View2D` scrollbar rect by a
pixel, the next draw call will correct this again though.
Pull Request: https://projects.blender.org/blender/blender/pulls/114631
Related commit: 82aa300e1c / 5596f7982182aa300e1c made it so that if only linked images are used, painting
would be prohibited. But since painting in Material mode actualy paints
across all materials it could happen that one particular image is linked
whereas another image isnt. In this case, painting on the linked image
would not be prohibited.
Two issues:
- [1] logic in above commit(s) seems slightly wrong, we could still pass
the check for linked images (also no image here should also not result
in a positive check I think).
- [2] code in `project_paint_prepare_all_faces` would still work with
linked images in slots, so
-- `ProjPaintState` -> `image_tot` was incremented
-- `ProjPaintImage` was created
-- this then ends up in the undo push - and crashes on undo
So now [1] is rectified and for [2] we skip faces with associated linked
images.
In addition, slots with linked images are now also greyed out in the UI
(`VIEW3D_PT_slots_projectpaint`) to give some sort of hint. For the
future it would probably also be good to dissallow selection of such
paint slots, but that is for another commit.
Pull Request: https://projects.blender.org/blender/blender/pulls/115263
Workaround potential C-API `FILE` incompatibility by reading the
file data into memory, compiling & running it - matching existing logic
for text buffers text buffers. This replaces the in-lined stub-script
that re-opened the file from Python.
While the down-side of the stub-script was minor, it required some
non-obvious logic and had the disadvantage of requiring 2x scripts to
execute whenever a file was executed on WIN32.
Expose BLI_file_read_data_as_mem_from_handle as a public function
since it's useful to be able to read an existing FILE into memory.
Passing a directory to --python would run it as a script without errors
since the directory would read as an empty file.
Raise an error instead because this is what Python it's self does.
This would already happen on WIN32 because the stub script was
executing the file from Python.
This effectively reverts [0] which required the installed Blender
along with the `bpy` module to load for the install target to generate
the man-page. It turns out running a full Blender at install time can be
quite involved.
Now the man-page generator runs `blender --help` & `blender --version`,
converting the output into a man-page from a Python script.
Some minor changes have also been made:
- Use PYTHON_EXECUTABLE so the systems Python
(which may not be compatible) isn't used.
- Remove leading blender version from the description
which was unintentionally being included.
[0]: 61d99d450e
