| Age | Commit message (Collapse) | Author |
|---|
|
and fix Rip Region double proportional checkbox
This fixes the double prop edit checkbox in the redo menu.
This also makes it so that proportional edit in connected mode now
matches how it behaves in mesh edit mode.
Without this change, ripping in UV edit mode with proportional edit on
would be useless as the UV verts you ripped will still be stuck together
even if they were not connected anymore.
Reviewed By: Campbell
Differential Revision: http://developer.blender.org/D8289
|
|
These flags need to be set correctly in order to distinguish between data that comes from cache files and raw data that comes directly from pointers to the data in Mantaflow.
|
|
Thanks to @JacquesLucke for the heads up.
|
|
For this mode is possible use strokes of 2 points.
Differential revision: https://developer.blender.org/D8138
|
|
|
|
The move constructor of `mpq_class` from GMP currently
allocates when it is moved. So, it cannot be noexcept.
Since we want to use this type, this static assert cannot
stay there.
|
|
Reviewed By: sebbas
Differential Revision: https://developer.blender.org/D8284
|
|
|
|
This was disabled during 2.8x for smooth porting of 2.7x scripts,
Now '@' is used for matrix multiplication,
support '*' to multiple vector elements.
See T56276.
|
|
parents.
This will override all linked data-blocks in the tree branch leading to
the selected one.
|
|
This is fairly generic code that can be re-used in other places.
|
|
As we can see in `initVertSlide_ex`, `sld` can be `NULL`.
`sld` is dereferenced, but can still be `NULL`.
|
|
Reviewers: sebbas
Differential Revision: https://developer.blender.org/D8281
|
|
|
|
Should not use copy_v3_v3 on a 2 element vectors.
|
|
|
|
|
|
|
|
|
|
Internally the "show_expanded" property stores the expansion for every
subpanel, but for RNA we should only check the first bit of the flag that
corresponds to the main panel.
|
|
|
|
`Kfv-> sco` was being treated as a local variable, and can be
confusing since this value is not updated when navigating the viewport.
|
|
span.size() returns an uint, causing a signed/unsigned comparison
using 3u sidesteps the issue
|
|
Properly normalize buffers now. Also expose option to not use albedo and normal
just like OptiX.
|
|
The smear brush was using the stroke direction to slide colors across
the mesh surface (this is called drag in other sculpt tools). Similarly,
other deformations can be included. The most common ones in image
editing are pinch and expand, which can be used to sharpen transitions
between colors.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D8270
|
|
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D8269
|
|
Enables the color palette subpanel for brushes that have color
capabilities (only the paint brush for now)
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D8268
|
|
Only update pointer of the template if we actually changed it...
|
|
|
|
Clang-tidy reported that those parameters could be const,
but that is not true on windows.
|
|
Code is mature enough now to not need this anymore, people who do not
want to use liboverrides can just not create them.
|
|
|
|
shobjidl_core.h only exists in the windows 10 SDK in the 8.1
SDK ShObjIdl.h will have the definitions we need, which still
exists in the 10 SDK and implicitly includes shobjidl_core.h.
so ShObjIdl.h will work on both SDK versions.
|
|
Careless use of acos() in spherical coordinates transformation was
deteriorating the precision near zenith (and nadir) and producing
glitchy pixels (best seen in longer focal lengths).
Reviewed By: fclem
Differential Revision: https://developer.blender.org/D8266
|
|
Simply remove that check ob userdef's themes, we are never read any
userdef from startup file anymore, so this check makes no more sense.
To be backported to 2.83.
|
|
Implemented G.moving suggestion from comments.
|
|
Previously it only picked the appropriate version with the
blender-vX.XX-release branches.
|
|
This commit adds a new read-only boolean property `Screen.is_scrubbing`.
The related property `Screen.is_animation_playing` is set to `True` in
two situations:
- Animation is actually playing (for example via the Play button in the
timeline)
- The user is scrubbing through time (in the timeline, dopesheet, graph
editor, etc.)
To distinguish between these two cases, the property
`Screen.is_scrubbing` has been added.
Concept approved by @brecht.
|
|
It was already possible to create Sound and Image strips that reference
non-existing files. Now it's also possible to create Movie strips
referencing missing files via the Python API call
`Sequences.new_movie()`. In this case, the duration of the strip will be
set to 1 frame.
Note that this commit does not change anything in the user interface.
The Python API of the `MovieStrip` class is extended with a function
`reload_if_needed()`. This function only performs disk I/O if the movie
strip cannot produce frames, that is either when its filepath points to
a non-existing file, or when the video sequence editor has not been
shown yet (for example because it is in an inactive workspace).
This allows for the following:
```
import bpy
scene = bpy.context.scene
vse = scene.sequence_editor_create()
filepath = bpy.path.abspath('//demo.mkv')
strip = vse.sequences.new_movie("movie", filepath,
channel=2,
frame_start=47,
file_must_exist=False)
strip.frame_final_end = 327
```
This will create a new movie strip, even when `demo.mkv` does not exist.
Once `demo.mkv` has appeared at the expected location, either
`strip.reload_if_needed()` or `strip.filepath = strip.filepath` will
load it.
Differential Revision: https://developer.blender.org/D8257
Reviewed By: Sergey, ISS
|
|
overrides.
This is long work, we are still likely missing a lot of cases...
|
|
|
|
I don't see the need for a BVH Tree to have root but not have leafs.
But apparently this case is possible.
|
|
Called when canceling a transform operation.
|
|
|
|
|
|
This function is more expansive than the simpler `rna_ensure_property()`
one, and should only be used when IDProperty data is actually needed.
If one only needs to ensure it has a valid PropertyRNA pointer,
`rna_ensure_property()` is much more efficient.
Also add compiler warnings when results of those functions are unused,
this should never be the case.
|
|
Only consider a full IDProperty as set if it actually exists in given
PointerRNA data.
|
|
|
|
The warnings were introduced in rB725973485a909c2b732c5.
|
|
Clang Tidy reported a couple of false positives. I disabled
those `NOLINTNEXTLINE`.
Differential Revision: https://developer.blender.org/D8199
|
