- Add icons for Sculpt Cloth, Clay Thumb and Draw Face Sets, as well as GP Tint, Replace and Transform Fill tools
- Tweak icons for Sculpt Rotate, Pinch, Multiplane Scrape, Inflate, Blob, Draw Sharp, based on feedback on Devtalk
Fixed memory leak that showed up after the original issue (crash) had been fixed in 93ac4709eb. The fix ensures that light cache bakes free up GPU smoke textures and the smoke domain list correctly.
This commit also removes the workaround (f3a33a9298) that disabled light cache bakes for fluid objects.
The `check_is_animated()` function will be used by the upcoming Alembic
exporter as well. There is nothing USD-specific in the function.
No functional changes.
Owner of filelisting job was changed, without proper update of all
access/usages of that owner to reach the job, leading to failure of
timer removal from the WM, and attempt to double-free the job...
Caused by rB2c4dfbb00246ff.
The issue was that we were creating temporary mesh copies and storing
them in bmain and then later using BKE_mesh_nomain_to_mesh which would
add them to bmain once more (duplicates).
This would lead to crashes later as the custom data of the mesh could be
trashed quite easily.
Was happening when object does not have CD_MDISPS allocated yet.
Need to make sure totdisp and level is specified on CD_MDISPS data
prior to loading (as the load expects them to be properly set).
Part of the fix was to get gputexture to use an array to accomodate each
eye. This takes care of viewports showing individual Left or Right
views.
For the combined view the fix was in overlay_image.c:camera_background_images_stereo_setup.
Note 1: Referece images are still not supporting stereo.
Note 2: For painting, and getting image bindcode I'm hardcording a
single-view experience.
Note 3: Without D6922 stereo is too broken to even test this patch.
With D6922 + this patch the fullscreen modes work (anaglyph/interlace
not yet).
Differential Revision: D7143
This change fixes artifacts produced by these operations.
On a technical aspect this is done by porting all of the operations
to the new subdivision surface implementation which ensures that
tangent space used to evaluate modifier and those operations is
exactly the same (before modifier will use new code and the operations
will still use an old one).
The next step is to get sculpting on a non-top level to work, and
that actually requires fixes in the undo system.
Makes it work better "out of the box" for irregular topology like
Suzanne mesh.
There might be some performance impact on non-regular meshes, but
those are not very common usecase for multires and for those its
always possible to lower the quality if needed.
Loose vertices and vertices of loose edges callback was not working
correct if some of other callbacks were set to NULL.
Was caused by missing bitmask set in the callbacks which were set
to NULL.
This allows fast access to various arrays in the Python API.
Most notably, `image.pixels` can be accessed much more efficiently now.
**Benchmark**
Below are the results of a benchmark that compares different ways to
set/get all pixel values. I do the tests on 2048x2048 rgba images.
The benchmark tests the following dimensions:
- Byte vs. float per color channel
- Python list vs. numpy array containing floats
- `foreach_set` (new) vs. `image.pixels = ...` (old)
```
Pixel amount: 2048 * 2048 = 4.194.304
Byte buffer size: 16.8 mb
Float buffer size: 67.1 mb
Set pixel colors:
byte - new - list: 271 ms
byte - new - buffer: 29 ms
byte - old - list: 350 ms
byte - old - buffer: 2900 ms
float - new - list: 249 ms
float - new - buffer: 8 ms
float - old - list: 330 ms
float - old - buffer: 2880 ms
Get pixel colors:
byte - list: 128 ms
byte - buffer: 9 ms
float - list: 125 ms
float - buffer: 8 ms
```
**Observations**
The best set and get speed can be achieved with buffers and a float image,
at the cost of higher memory consumption. Furthermore, using buffers when
using `pixels = ...` is incredibly slow, because it is not optimized.
Optimizing this is possible, but might not be trivial (there were multiple
attempts afaik).
Float images are faster due to overhead introduced by the api for byte images.
If I profiled it correctly, a lot of time is spend in the `[0, 1] -> {0, ..., 255}`
conversion. The functions doing that conversion is `unit_float_to_uchar_clamp`.
While I have an idea on how it can be optimized, I do not know if it can be done
without changing its functionality slightly. Performance wise the best solution
would be to not do this conversion at all and accept byte input from the api
user directly, but that seems to be a more involved task as well.
Differential Revision: https://developer.blender.org/D7053
Reviewers: JacquesLucke, mont29
The code was copied from the Alembic exporter, and some of the options are
no longer used.
Not updating the Alembic exporter itself, as this will be done in a much
larger rewrite.
Both are doing almost the same and can be merged. This reduce complexity for user and less code to maintain.
Reviewed By: mendio, pepeland, fclem
Differential Revision: https://developer.blender.org/D7134
The Deform modifiers was reallocating buffers that only fit the vertices
of the inner loop. This patch first counts the maximum needed buffer and
allocates one.
When using the daily dweebs animation file the playback performance went
from 0.66 fps to 0.93 fps.
Reviewed By: sybren
Differential Revision: https://developer.blender.org/D7132
If one of the faces connected to a vertex is hidden in the face sets, we
can assume that the vertex is part of a boundary edge, so it should be
cosidered like that in all automasking and edge detection functions.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D7126
The idea of the visibility system is that tools should behave like
hidden vertices do not exist, so the flood fill operation should ignore
hidden vertices for all operators.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D7125
The face set ID is sequential, so implementing this was straightforward.
Suggested by Jeroen Bakker
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D7123
The face_set_set function which sets a face sets given a vertex index
can ignore all modifications to hidden face sets, so we can skip all
vertex visibility checks outside that function. This makes the code
faster, simpler and fixes multiple bugs.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D7122
The previous solution was also working fine as the white face set has no
meaning, but now it is a little bit more random. Also, bigger face sets
have more chance of getting the white color.
Reviewed By: jbakker
Maniphest Tasks: T74646
Differential Revision: https://developer.blender.org/D7111
The mesh provided in the report has 0 area faces and overlapping
vertices, causing the relax code to fail when calculating the plane to
constraint the vertex movement. Now it works fine both in the brush and
in the mesh filter.
Reviewed By: jbakker
Maniphest Tasks: T74648
Differential Revision: https://developer.blender.org/D7109
If the relax mesh filter was used on a non manifold mesh with open
boundaries, all the vertices were relaxed and the mesh was shrinking.
This was an unintended behavior that was making the filter unusable with
these meshes.
The mesh filter is now initializing an automasking buffer using the same
boundary automasking function from the brush code. Now edges are
preserved and the relax filter works as it should.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D7097
Embedded data should always be considered as outside of Main database
here.
Note that it's a bit of an edge case to decide whether those should
always have their `LIB_TAG_NOMAIN` set too, or not? For now, let's keep
things as they are here.
The behaviour of GP layers is the same as annotation layers: they show
in the dope sheet regardless of whether they have frames or not. This is
easily resolved by adding some extra filtering.
Part of the function was following an "if-ok: do-this" pattern, and then
mid-function switched to a "if-bad: skip" pattern. The function now just
uses the latter.
No functional changes.
