Free the BVH tree immediately along with the mesh, otherwise we might access
invalid mesh data.
Differential Revision: https://developer.blender.org/D4201
This is actually a workaround for the crash in OpenSubdiv.
Topology refiner will have a crash when special conditions
are met:
- Refiner is configured to use infinitely sharp patches.
- Refinement happens for the level 1 (which we call Quality 1 on
Blender side).
- Mesh has non-quad faces.
The workaround is to force refinement to happen to level 2 (or
quality 2 on Blender side) when those conditions are met.
Later on with the next OpenSubdiv update we can remove this
workaround, since there was work done on OpenSubdiv side to
deal better with such configurations.
The modifier will now be somewhat slower, but this will be
compensated with upcoming topology cache enabled by default.
The workaround is done when initializing settings, so the
comparison of topology refiner settings is happening without
any extra workarounds there.
Will document the new options in release notes, then in manual.
Still a bit of work to do on the bulging shape that appears
on cube corners if using arc inner miters, but will do that later.
Also need to do something smarter in clamp overlap.
There was no documentation at all, some very bad practices (like using
G.debug_value > 0 as some sort of global debug print switch), and even
an overlapping use of '1' value...
Also, python setter did not check for valid range (since this is a
short, not an int).
This commit makes it so OpenSubdiv's topology refiner is kept
in memory and reused for until topology changes. There are the
following modifications which causes topology refiner to become
invalid:
- Change in a mesh topology (for example, vertices, edges, and
faces connectivity).
- Change in UV islands (adding new islands, merging them and
so on),
- Change in UV smoothing options.
- Change in creases.
- Change in Catmull-Clark / Simple subdivisions.
The following limitations are known:
- CPU evaluator is not yet cached.
- UV islands topology is not checked.
The UV limitation is currently a stopper for making this cache
enabled by default.
That was indeed not working properly, not at all. Except for
the basic case, but as soon as you used another object to define the
mirror plane, it would be utterly broken, in several different ways!
The units scaling was inappropriate when the bevel value was
to be interpreted as a percent, so added a separate rna property
for "Width Percent" and made UI show the width appropriate for
current offset_type.
Harden normals causes normal splitting, which will not give the
appearance expected due to autosmooth unless some edges are sharpened,
so this change fixes that. Also bevel tool will turn on autosmooth
if not already on if hardening normals.
Move the bevel hardening code all into bmesh_bevel.c.
Based on user feedback, rewrote the bevel hardening algorithm
to be more what users want.
Based on user feedback, changed the UI, removing some
not-useful options. Now hardening normals while beveling
is enabled by a simple checkbox.
Now setting face strength gives options for which faces
get their face strength set.
This is a second attempt to get the crash fixed. The original fix
worked, but it was reverted by d3e0d7f082.
Now the logic goes as:
- All pointers which we can not have shared (the ones which are
owned by the runtime) are cleared.
- The rest of runtime stays untouched.
This seems to be enough to keep particles happy.
Fix T58237: Exporters: Curve Modifier not applied when "apply modifiers" are selected.
Fix T58856: Python: "to_mesh" broken in 2.8.
...And many other cases... ;)
Thing is, we need target IDs to always be evaluated ones (at least I
cannot see any case where having orig ones is desired effect here).
Depsgraph/Cow system ensures us that when modifiers are evaluated by it,
but they can also be called outside of this context, e.g. when doing
binding, or object conversion...
So we need to ensure in modifiers code that we actually are always
working with eval data for those targets.
Note that I did not touch to physics modifiers, those are a bit touchy
and rather not 'fix' something there until proven broken!
better have this vertex color layer cover the whole 0-1 range
thx @sergey for checking
Maniphest Tasks: T57994
Differential Revision: https://developer.blender.org/D3976
There were at least three copies of those:
- OB_RECALC* family of flags, which are rudiment of an old
dependency graph system.
- PSYS_RECALC* which were used by old dependency graph system
as a separate set since the graph itself did not handle
particle systems.
- DEG_TAG_* which was used to tag IDs.
Now there is a single set, which defines what can be tagged
and queried for an update. It also has some aggregate flags
to make queries simpler.
Lets once and for all solve the madness of those flags, stick
to a single set, which will not overlap with anything or require
any extra conversion.
Technically, shouldn't be measurable user difference, but some
of the agregate flags for few dependency graph components did
change.
Fixes T58632: Particle don't update rotation settings
We had two different ways of doing it, SurfaceDeform and LaplacianDeform
would do it through a special modifier stack evaluation triggered from
binding operator, while MeshDeform would do it through a regular
depsgraph update/eval (also triggered from its binding op).
This enforces the later to search back for orig modifier data inside
modifier code (to apply binding on that one, and not on useless CoW
one).
Besides the question of safety about modifying orig data from threaded
despgraph (that was *probably* OK, but think it's bad idea in general),
it's much better to have a common way of doing that kind of things.
For now it remains rather dodgy, but at least it's reasonably consistent
and safe now.
This commit also fixes a potential memleak from binding process of
MeshDeform, and does some general cleanup a bit.
This fixes/clarifies Surface Deform evaluation code that does the
binding, since that part should only be called outside of depsgraph
evaluation, with orig data-blocks and not CoW ones.
Now we have a decent amount of asserts and checks to ensure eveything
works as expected.
Also had to add a special case to get target's mesh in binding case,
since often target's evaluated mesh is not available, in that case (and
in that case only), we can actually compute that mesh (because we are
out of depsgraph evaluation).
