| Age | Commit message (Collapse) | Author |
|---|
|
|
|
|
|
|
|
|
|
|
|
Since 2d80f814cc249d26b01f, curves always have evaluated points,
but single point curves do no have any evaluated segments, and the
leading zero length isn't stored in the curves length cache.
|
|
|
|
Previously the function had a fair amount of ugly boilerplate to avoid
allocating the temporary layers array, and then free it if necessary.
`blender::Vector` solves that problem more elegantly. Passing a span,
using references in a few cases, and using a switch statement also make
the functions simpler.
This refactoring is in preparation for D14583 and D14685.
Differential Revision: https://developer.blender.org/D15011
|
|
Also early exit in some functions.
|
|
|
|
The logic of skipping selected handles was inverted
and confusing.
|
|
When the 'compare' is not used for the resulting selection, just hide
it. This is the case for 'Vertex Groups' atm (where only membership is
taken into account).
Similar to rB9dc9692b0979.
Differential Revision: https://developer.blender.org/D14979
|
|
This could happen when switching between custom action sets that both
had controller pose actions. Problem was that controller data is
cleared when changing action sets, and this clearing was done when
handling WM events, which always occurs after XR controller data is
updated from GHOST.
Now, instead of activating the action set immediately, delay activation
until just before the next XR actions sync.
|
|
The nil terminator character shouldn't be included, causing an extra
character to be added in some cases.
|
|
Since [0], using the view navigation gizmo crashed with Wayland.
This only worked previously because GHOST_kGrabWrap was ignored.
Now the previous grab state is disabled before switching to a new
grab state.
[0]: da9e14b0b91c81d29c4e44f40ac299ae847367de
|
|
|
|
The original fix for T97366 was too restrictive and breaks real-world
cases of single-file UDIM textures. See D13297 for an example.
This patch effectively reverts the original fix and instead fixes the
downstream code to accept single-file ranges as necessary.
Note: This means it is very important for users to make use of the
"UDIM detection" option during `image.open` or drag n' drop scenarios in
order to declare their intent when loading their files.
Differential Revision: https://developer.blender.org/D14853
|
|
An ASAN build highlighted a longstanding bug during ctrl+c operations
inside various text widgets. The existing code had mismatched memory
lock/unlock calls and was using the wrong allocator.
Fix the code surrounding `SetClipboardData` to be correct per MSDN:
https://docs.microsoft.com/en-us/windows/win32/api/winuser/nf-winuser-setclipboarddata
Differential Revision: https://developer.blender.org/D15039
|
|
|
|
After recent changes to Nishita sky to clamp negative colors, the pixels ended
up a bit brighter which lead to them exceeding the half float max value. The
CUDA float to half function seems to need clamping.
|
|
Uses the macro introduced in b45f410b315 where it makes sense.
|
|
|
|
|
|
Differential Revision: https://developer.blender.org/D14878
|
|
IPO data-block types are deprecated since 2.5. Don't show them in the
Outliner at all.
Differential Revision: https://developer.blender.org/D15049
|
|
Such as AMD Radeon RX 6700S.
Differential Revision: https://developer.blender.org/D13495
|
|
Code was not yet updated to use generic attributes for vertex colors. This also
makes generic attributes work for adaptive subdivision.
|
|
|
|
This bug was introduced in commit https://developer.blender.org/rB5188c14718c56e4d088d3c5bb3ce3ed9ed8b7bdc because the object transform was not applied when baking the object.
Thanks to @frogstomp for the head up.
|
|
Fix "parameter unused" warning that shows up when building without NanoVDB.
No functional changes.
|
|
|
|
After removing the default 'Home' shortcut for "Frame All", a NDOF (3D
Mouse) default shortcut was still available for the operator. The event
filtering introduced in 4357fb63db77 was missing the NDOF filtering
logic. So while the context menu correctly found the NDOF keymap item,
its actual shortcut change/removal code incorrectly filtered out the
NDOF keymap items and thus failed to find the item.
|
|
|
|
This is a regression caused by a230445caec6.
The internal cause of the issue was that the synchronization component
was no longer tagged as visible (and hence not evaluated) as it not
connected to any directly visible IDs.
Changed the logic in a way that if any component of an ID is evaluated
the synchronization component will be evaluated as well.
The naming of the flag in the component node is a bit confusing, but
for the simplicity of the change for the upcoming release left it
unchanged.
|
|
Fix T97500 by removing the logic that for some unknown reason draws the
entire string if the min/max were swapped.
This function is only called in two places, once here in the NLA, and
the VSE. The bug only materializes in the NLA though.
Reviewed By: sybren
Differential Revision: https://developer.blender.org/D14742
|
|
Fix T97974 by having the marker rendering code explicitly set the
required line width.
Reviewed By: sybren
Differential Revision: https://developer.blender.org/D14890
|
|
|
|
full scaled image isn't used anymore. It was added to use a different scale when
displaying an image in the image editor. This was replaced by the image engine
redesign.
This change will reduce complexity of {T98375}.
|
|
`BKE_id_delete` should only check for consistency of user count with
regards to the tags and flags of the ID, not 'protect' nor even warn in
case a 'fake user' ID is deleted (such higher-level checks are to be
handled by higher-level code).
Also replace the assert + debug print by a CLOG error, this avoids
'assert crash' while still failing tests, and always producing a useful
message.
Fixes T98374 and T98260.
|
|
Dragging number buttons wasn't grabbing the cursor and would stop
when the pointer reached the screen edge & wasn't setting the cursor
visible on completion.
|
|
|
|
|
|
Currently all cursors are square, so this didn't show up as a bug.
|
|
Error in [0] caused the `set` callback be checked as the search callback.
[0]: 3f3d82cfe9cefe4bfd9da3d283dec4a1923ec22d
|
|
The day is currently specified as 90000 seconds which is 25 hours.
Obviously this is wrong, and this commit changes it to 86400s/24 hours.
Differential Revision: https://developer.blender.org/D15034
|
|
Mostly using built-in `__atomic` functions, with a special code path
using `MemoryBarrier()`on windows.
Authored By: Sergey Sharybin (sergey)
Reviewed By: Sergey Sharybin (sergey), Ray molenkamp (LazyDodo)
Ref D15020
|
|
Contrary to coarse extraction, GPU extraction uses the same buffer for
the coarse data, only the final GPU buffer needs to be offset.
|
|
The face dots normals may not be always requested, thus leading to a
crash by null pointer dereference.
|
|
|
|
|
