Linked Armature with local proxy, using feature "Custom shape at other bone" stopped working
on undo/redo.
It was actually a bug in the original commit (r26600, april 2010), storing a pointer from
the library bone into the local proxy bone.
That's strictly forbidden in Blender, but it never showed up because on every undo-redo a
complete proxy-sync was called again. To allow undo/redo I had to disable this syncing,
except for file load. Hence the feature got lost :)
The fix is simple; just store the pointer to its own local bone instead.
Issue was caused by couple of circumstances:
- Normal Map node requires tesselated faces to compute tangent space
- All temporary meshes needed for Cycles export were adding to G.main
- Undo pushes would temporary set meshes tessfaces to NULL
- Moving node will cause undo push and tree re-evaluate fr preview
All this leads to threading conflict between preview render and undo
system.
Solved it in way that all temporary meshes are adding to that exact
Main which was passed to Cycles via BlendData. This required couple
of mechanic changes like adding extra parameter to *_add() functions
and adding some *_ex() functions to make it possible RNA adds objects
to Main passed to new() RNA function.
This was tricky to pass Main to RNA function and IMO that's not so
nice to pass main to function, so ended up with such decision:
- Object.to_mesh() will add temp mesh to G.main
- Added Main.meshes.new_from_object() which does the same as to_mesh,
but adds temporary mesh to specified Main.
So now all temporary meshes needed for preview render would be added
to preview_main which does not conflict with undo pushes.
Viewport render shall not be an issue because object sync happens from
main thread in this case.
It could be some issues with final render, but that's not so much
likely to happen, so shall be fine.
Thanks to Brecht for review!
Change Scene.frame_set so that it ensures subframe in range [0,1[ as Blender
expects, otherwise some things like physics point cache lookups don't get
evaluated properly.
now blenlib/BLI doesn't depend on any blenkern/BKE functions,
there are still some bad level includes but these are only to access G.background and the blender version define.
Path rebuild may fail after certain editing operations. Instead of blindly
assuming that path rebuilds will always succeed, we now give up trying to
evaluate in case of failure.
Now only one operator. Same options for vertices, edges and faces (so adds edges sorting, and some options to vertices sorting).
Face sorting should behave as previously. However, XSortVerts won’t pack anymore selected vertices at the begining of the vert array (as it used to), if you want such behavior you’ll have to first run SortElements with Selected action.
Also added bug ref I forgot in r46354 (armature.c).
- Make sure functions are named in way BKE_<object>_<action> (same way as RNA callbacks)
- Make functions which are used by mball.c only static and remove their prototypes
from public header file.
Further cleanup is coming.
In fact, problem was in vec_roll_to_mat3(), which has to detect when the bone is aligned with its Y axis, using a threshold. This one have been raised to quite a high value due to bug [#23954], then lowered a bit due to [#27675] (which is imho in fine the same problem as 30438).
Reset it to its org value (very low 1e-13), as testing file given with firt bug did not show any problem anymore... So now, instead of 1/1000 of bone length from Y axis, we have about 3.25*10-7... Only (hardly) noticeable at max zoom level in 3D view.
This commit:
* Removes the Wire and Color options from the UI for all object types but meshes, curves/surfaces/texts, and metas.
* Adds a basic bounding box drawing (and computing) for armatures.
Now constraints' space conversion code also uses generic armature_mat_(pose_to_bone/bone_to_pose). Previous own function (constraint_pchan_diff_mat) was somewhat inconsistent too with Hinge/NoScale/LocalLocation options...
As with previous similar changes, this might break some old rigs, in very specific cases (when constraint-evaluating an hinge/noscale/local_location bone in local space).
In the same part of code, removed unnecessary matrices copying, mult_m4_m4m4 can take the same matrix as input and output, nowadays...
Also found a bug-generator weakness in those armature_mat_ functions (if both input and output mat where the same, result was wrong, now systematically copying input mat, as done in LIB's matrix funcs).
Finally, factorized offset bone matrix generation into its own small function too, as it is used in two different places in armature.c (pchan_to_pose_mat itself, and restpose's where_is_armature_bone).
Note: I think all parts of blender's code related to that topic have now been tackled, but yet have to check BGE, it’s probably using that stuff too, one way or the other...
