From patch D11780 from Erik Abrahamsson.
It parallelizes making the vertices, destruction of map entries,
finding if the result is PWN, finding triangle adjacencies,
and finding the ambient cell.
The latter needs a parallel_reduce from tbb, so added one into
BLI_task.hh so that if WITH_TBB is false, the code will still work.
On Erik's 6-core machine, the elapsed time went from 17.5s to 11.8s
(33% faster) on an intersection of two spheres with 3.1M faces.
On Howard's 24-core machine, the elapsed time went from 18.7s to 10.8s
for the same test.
The main goal here is to remove the need for a C API to the code in
`mesh_boolean_convert.cc`. This is achieved by moving `MOD_boolean.c`
to C++ and making the necessary changes for it to compile. On top of
that there are some other slight simplifications possible to the
direct mesh boolean code: it doesn't need to copy the material
remaps, and the modifier code can use some other C++ types directly.
This commit improves the performance of the node by up to 40% in some
cases when there are only two input meshes, mainly by skipping the
conversion to and from BMesh.
When there are more than two input meshes (note the distinction from
"Geometries", a geometry set can have many mesh instances), the
performance is actually worse, since boolean currently always does
self intersection in that case. Theoretically this could be improved
in the boolean code, or another option is automatically realizing
instances for each input geometry set.
Another improvement is using multi-input sockets for the inputs, which
removes the need to have a separate boolean node for every operation,
which can hopefully simplify some node trees.
The changes necessary for transforms in `mesh_boolean_convert.cc` are
somewhat subtle; they come from the fact that the collecting the
geometry set instances already gives transforms in the local space
of the modifier object. There is also a very small amount of cleanup
to those lines, using `float4x4::identity()`.
This commit also fixes T87078, where overlapping difference meshes
makes the operation not work, though I haven't investigated why.
Differential Revision: https://developer.blender.org/D10599
I had done some experiments to see what Fast boolean did for material
mapping and thought it just used the same slot in the target as the
slot in the source. The truth is more complicated: if the target material
exists in any slot of the destination, we need to remap to whatever
slot has the matching material. I fixed Exact Boolean to do this.
Since the materials may be in the object, this means that BKE_mesh_boolean
had to get another argument, the remapping arrays.
I will note that the current behavior of Fast, and now Exact, is not ideal.
Ideally, if the source material does not exist in the target, a new material
slot should be created in the target and the source material copied there
(and incrementing the material's reference count). Maybe a future project,
but for now, I want the behavior of Exact to match that of Fast.
The commit rB6f63417b500d that made exact boolean work on meshes
with holes (like Suzanne) unfortunately dramatically slowed things
down on other non-manifold meshes that don't have holes and didn't
need the per-triangle insideness test.
This adds a hole_tolerant parameter, false by default, that the user
can enable to get good results on non-manifold meshes with holes.
Using false for this parameter speeds up the time from 90 seconds
to 10 seconds on an example with 1.2M triangles.
This commit includes a few simple improvements to the direct mesh
boolean code added recently.
- Passing the transforms and meshes to `direct_mesh_boolean` as spans
makes the function easier to call from C++.
- The definition of `TransMat`, was unecessary when we have the
`float4x4` type already used elsewhere in C++ code.
Differential Revision: https://developer.blender.org/D10592
The Exact modifier code had been written to avoid using BMesh but
in the initial release the modifier still converted all Meshes to
BMeshes, and then after running the boolean code on the BMeshes,
converted the result back to a Mesh.
This change skips that. Most of the work here is in getting the
Custom Data layers right. The approach taken is to merge default
layers from all operand meshes into the final result, and then
use the original verts, edges, polys, and loops to copy or interpolate
the appropriate custom data layers from all operands into the result.
