| Age | Commit message (Collapse) | Author |
|---|
|
of Normal2DF0D with smooth FEdges. The fix is to avoid generating new FEdgeSmooth
instances whose length is shorter than epsilon.
|
|
Conflicts resolved:
source/blender/blenloader/intern/readfile.c
|
|
VERSION
|
|
|
|
members multiple times
|
|
* GaussianYBlurOperation accessed sizex instead of sizey.
|
|
|
|
solver operates in world reference, therefore armature scale is used to build the ik scene. But the scaling was not taken out when applying the pose at the end of the simulation.
|
|
|
|
|
|
|
|
finally works
This commit finally hooks up the Mask Modifier's "Armature" option with the
relevant depsgraph updates on bone selection. Hence, this feature finally works
as it was originally intended - that is, bone selections can be used to control
which parts of the mesh that the mask modifier is applied to are displayed,
giving riggers more freedom to experiment with rigs that don't necessarily
feature overbearing/cluttering widgets.
Regarding the implementation ("has_viz_deps" flag):
This feature is just the "tip of the iceberg" of a number of related set of
rigging/visual animation tools I've had in mind for a while now (dating back to
the introduction of this modifier). Key considerations
- Not all rigs will use this, so we don't want an extra (depsgraph-flush +
search) recalc cost for those that don't use this.
- There are some planned features which will also use this
|
|
'Armature' mode for the Mask Modifier was not working at all anymore even when
the selection <-> depsgraph recalc issue was patched to work (this latter fix is
coming in another commit). It appears that this probably happened during one or
more of the refactors which may have taken place around here over the years
since I first introduced it.
This commit does two things:
* Removed the unused/redundant "vgroupHash"
* Fixed the incorrect assumption used for determining if the vertex actually
belonged to a vgroup corresponding to a selected bone.
|
|
|
|
|
|
Note that I had to script-tag all sokets' names, as they are currently completely unknown from bpy.types (and hence unreachable for our po generating scripts).
|
|
not agreed on so probably this will be disabled for release.
the data is stored here so more dev can commit:
./release/scripts/addons/modules/rna_wiki_reference.py
|
|
|
|
|
|
having either a "texture" Tex, an "area" SpaceImage, or both "edit_image" Image and "edit_image_user" ImageUser).
Thanks to Campbell who pointed me to uiLayoutSetContextPointer() func!
|
|
high repulsions.
|
|
Spoted by MiikaH.
|
|
wrong file version for do_versions.
|
|
|
|
In response to [#31670], I've reviewed the way that the Paste Keyframes tool for
the DopeSheet and Graph Editors works. Previously, it required you to always
select the F-Curves to paste the keyframes into before allowing you to paste
keyframes. This was because it is quite difficult to infer which ID-block's set
of curves is intended if more than one ID-block has similar curves (e.g. a scene
with two materials, and both have their diffuse color animated). The underlying
assumption and intention of the feature here was that the copy+paste were only
being used by animators to copy animation between similar curves, to transfer
and offset animation across block boundaries.
However, it turns out that many people were by far more familiar with the
simpler copy/paste paradigm from everywhere else (i.e. instead of trying to use
duplicate to copy keyframes around within their respective F-Curves).
Furthermore, in most cases there is only going to be a single character being
animated at a time (vs multiple), which means that most of the time the matching
problem is much simpler.
Hence, the Paste now works as follows:
- If there are selected F-Curves, we limit the paste-matching to only consider
those in the selected F-Curves. This makes it possible to still explicitly
specify where to paste.
- In the more general case (no prior selections), pasting will try to match
anything relevant it finds.
TODO:
- Check on whether the strictest matching level needs adjustments to limit the
number of false positives
- Testing and feedback of the new behaviour needed <--- ANIMATORS! PLEASE TEST
|
|
|
|
used when drawing frames with a custom color (which is just RGB).
|
|
- Added mask sculpt brush
- Draw brush was removed from sculpt mode -- there's SculptDraw brush
- Made default strength for draw brush 1.0 which makes much more sense
especially for weight painting
|
|
|
|
Colors used by Bone Groups are now copied/assigned to Action Groups too when
they're created now. This completes the work started in r.46960 to restore this
functionality from 2.48.
Currently, there is no control over when/whether these colors are copied over
(although it is possible to disable the display of these colors for relevant
animation editors if desired). Originally I was going to make this a more
generic Keying Sets feature, though that turned out to be a bit too complex to
manage.
Other notes:
* Split out the code for copying colors to a common library function
|
|
|
|
By holding shift and "cutting" a node link a new reroute helper node can be inserted. This consists of a single socket that can be used to insert additional connection points into a link. This can be used to keep a connection point in the tree when deleting a node, or to control the path of long connections for layout cleanup.
|
|
|
|
|
|
|
|
Previous patch allocated temporarily buffers.
|
|
- should increase speed with large node setups
- enables caching of buffers in the node editor (in the future)
- OpenCL part still needs some work
|
|
|
|
|
|
input of math node
|
|
Solution based on patch #31635 by Philipp Oeser.
|
|
|
|
cause flickering as you paint.
It also changes multi-paint's "zero weight color" to black again--it was unintentionally changed to red some time late last year when an aspect of the colors were changed.
|
|
scale that sets the default size.
|
|
|
|
|
|
|
|
|
|
[#31562] New compositor crashes due to incorrect opencl initialization
|
|
dependency sorting is correct (avoid red links).
|
