The issue was caused by render engine duplicating the curve object because
it then converts the object to mesh. The mesh deform duplication code didn't
duplicate binded data, so after duplication modifier is no longer applyable.
So now copyData of mesh deform would copy data needed for binding. This solves
reported render bug and also solves possible frustration when duplicating an
object with mesh deform in the viewport with Shift-D.
Checked other modifiers and laplacian deform already was copying binded data.
Didn't see other modifiers which might also need to copy extra data.
It now supports different scheduling schemas: dynamic and static.
Static one is the default and it splits work into equal number of
range iterations.
Dynamic one allocates chunks of 32 iterations which then being
dynamically send to a thread which is currently idling.
This gives slightly better performance. Still some tricks are
possible to have. For example we can use some smarter static scheduling
when one thread might steal tasks from another threads when it runs
out of work to be done.
Also removed unneeded spin lock in the mesh deform evaluation,
on the first glance it seemed to be a reduction involved here but
int fact threads are just adding value to the original vertex
coordinates. No write access to the same element of vertexCos
happens from separate threads.
Quite straightforward change, don't think some extra explanation is needed.
This gives about 15% speedup of the modifier evaluation on my laptop.
Reviewers: campbellbarton
Differential Revision: https://developer.blender.org/D836
This commit switches meshdeform modifier to use threads to evaluate
the vertices positions using the central task scheduler.
SO now we've got an utility function to help splitting the for loop
into tasks using BLI_task module which is pretty straightforward to
use: it gets range (which is an integer lower and higher bounds) and
the function and userdata to be invoked for each of the iterations.
The only weak point for now is the passing the data to the callback,
this isn't so trivial to improve in pure C.
Reviewers: campbellbarton
Differential Revision: https://developer.blender.org/D838
This is not needed when we tag normals as dirty, but in simple cases, if offset has some
rotation (or non-uniform scaling) component, vertex normals has to be corrected.
Now the bevel tool, modifier, and internal operator have a material
slot # parameter that the user can set. If left at default of -1,
behavior is as current -- bevel face material is taken from the
closest original face (this may be ambiguous). If material slot
is >= 0, it gives the material slot index number for the material
to use.
It was a regression since 5d49eff. Not really sure about proper solution
here, so used a bit workaround-ish way for now.
Hopefully new cloth will be landed after this GSoC anyway.
Own regression since e08db08. CustomData_copy_data() would fail in cases
when mvert/medge/mloop/mpoly arrays were lazy allocated since that change.
Now made it so this layers are copying from own copy of the arrays.
Not sure if this still misses some CD to be copied, from quick glance
seems no, but some further testing wouldn't hurt at all.
The issue was caused by the temporary CD layers being allocated for subsurf
meshes, same as we've got back in 881fb43.
In the long run this temporary storage is to be re-considered, but it'll also
imply re-considering of the Derivedmesh interaction as well. For now let's
use a simpler solution which is forbidding modifiers to call getArray for other
objects' derivedMeshes but use an API calls which would allocate local copy of
the data preventing race condition of shared data in DM.
