Now add asserts to make sure this layer is only added once the modifier stack has been calculated.
this saves normal layer being calculated whenver vertex normals need updating.
many modifiers were calculating normals, when those normals were ignored by the next modifier.
now flag normals as dirty and recalculate for modifiers that set use `dependsOnNormals()` callback.
Quick test on mesh with 12 modifiers (mostly build type), calculated normals 6 times, now it only runs once - so this will give some speedup too.
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!
reported as [#29376] BMESH_TODO: remove tessface CD_ORIGINDEX layer
for a single mesh there could be 3 origindex mappings stored, one on the polygons and 2 on the tessfaces.
(CD_POLYINDEX and CD_ORIGINDEX).
as Andrew suggests, now tessfaces (which are really a cache of polygons), using origindex to point to polygons on
the same derived mesh, and polygons only store the original index values.
Added four new functions as shortcuts to creating GHashes that use the
standard ptr/str/int/pair hash and compare functions.
GHash *BLI_ghash_ptr_new(const char *info);
GHash *BLI_ghash_str_new(const char *info);
GHash *BLI_ghash_int_new(const char *info);
GHash *BLI_ghash_pair_new(const char *info);
Replaced almost all occurrences of BLI_ghash_new() with one of the
above functions.
CDDM_tessfaces_to_faces wasbt updating the polyindex (missed incrementing the polyindex pointer).
also added an assert so non release builds will complain when CDDM_calc_edges_tessface() needs to run before CDDM_tessfaces_to_faces().
Seems that after converting tessfaces into faces normals should be re-calculated.
At least that's what happening in some other modifiers and what makes boolean
work fine with smooth normals.
Issue was caused by resetting face's mat_nr to zero if there's no material
map sent to ConvertCSGDescriptorsToDerivedMesh. In case of boolean modifier
we can't use such map because we can't affect on materials present in object.
So the only way which can give reasonable result is:
- Dot change mat_nr for faces from left operand (they should be fine, because
materials aren't deleting by modifier)
- For faces from right operand check if needed material exists in left operand
and if so, use it's index as new mat_nr.
- If there are materials in right operand which doesn't exist in left operand,
they'll be changed to first material from left operand.
Original indices from right operand were used in boolean result derived mesh
which lead to crash if right operand has got more entities (faces/edges/vertices)
than left operand.
Issue was caused by left boolean operand consist of several intersecting manifolds
which make Carve triangulator confused and which can't be resolved in general case.
Added mesh pre-processing before actual applying boolean operator on it. This
preprocessing applies union operation on intersecting manifolds of the same object
so intersection edge loop with second object wouldn't confuse tesselator and correct
result would be returned.
Detecting of intersecting manifolds is based on AABB intersection check which leads
to some extra union operation called, but it's possible to speed things up from
Carve side so union operation of two not intersecting meshes would work faster.
Additional condition for running union for manifold is this manifolds intersects
AABB of second operand, so mesh topology wouldn't be changed at all in areas
where there's definitely no intersection between operands. It might be improved
so only manifolds which actually intersects second operand would be joined
together, but it'll slow things down a bit and prefer to do it only if it'll
be really a problem.
Additional change is fixed memory leak when boolean operation fails to run -
it was missed "delete" call if exception happens in Carve library.
From side effects of this change might be named boolean operation between
suzanne and another object: suzanne is consist of three intersecting open
manifolds, so this new meshes preprocessing leads to missed eyes in result
because of failure of merging two open manifolds. Don't think making suzanne
work for all setups should really be a goal, it's a bit crappy mesh for CSG
algorithms.
inconsistent with similar functions & math notation:
mul_m4_m4m4(R, B, A) => mult_m4_m4m4(R, A, B)
mul_m3_m3m4(R, B, A) => mult_m3_m3m4(R, A, B)
For branch maintainers, it should be relatively simple to fix things manually,
it's also possible run this script after merging to do automatic replacement:
http://www.pasteall.org/27459/python
most local modifier,GPU,ImBuf and Interface functions are now static.
also fixed an error were the fluid modifier definition and the header didnt have the same number of args.
only tags the ID and does the actual flush/update delayed, before the next
redraw. For objects the update was already delayed, just flushing wasn't
yet.
This should help performance in python and animation editors, by making
calls to RNA property update quicker. Still need to add calls in a few
places where this was previously avoided due to bad performance.
