Lazy triangulation after boolean operations about manifold

[elalish]

If they don't intersect, then already the triangulation is being skipped. The increase in time is simply due to having to test more faces for intersection, and it should be fairly linear. Compose is for exactly situations like this, and it allows you to do it all in a single operation too. If you're worried, it might be enough to test a couple of neighbors for intersection (do an intersect opt and check the result is empty). If that works, then Compose should be safe.

[pca006132]

But their bounding box do intersect and it seems that this will still perform triangulation after union? If you set kVerbose to true in the boolean_result.cu and do grep Triangulation, this is the result:

----------- 1 ms for Triangulation
----------- 0 ms for Triangulation
----------- 0 ms for Triangulation
----------- 1 ms for Triangulation
----------- 1 ms for Triangulation
----------- 1 ms for Triangulation
----------- 2 ms for Triangulation
----------- 2 ms for Triangulation
----------- 2 ms for Triangulation
----------- 2 ms for Triangulation
----------- 3 ms for Triangulation
----------- 3 ms for Triangulation
----------- 3 ms for Triangulation
----------- 4 ms for Triangulation
----------- 4 ms for Triangulation
----------- 4 ms for Triangulation
----------- 4 ms for Triangulation
----------- 5 ms for Triangulation
----------- 5 ms for Triangulation
----------- 5 ms for Triangulation
----------- 6 ms for Triangulation
----------- 6 ms for Triangulation
----------- 6 ms for Triangulation
----------- 7 ms for Triangulation
----------- 7 ms for Triangulation
----------- 7 ms for Triangulation
----------- 7 ms for Triangulation
----------- 8 ms for Triangulation
----------- 8 ms for Triangulation
----------- 8 ms for Triangulation
----------- 8 ms for Triangulation
----------- 9 ms for Triangulation
----------- 9 ms for Triangulation
----------- 9 ms for Triangulation
----------- 9 ms for Triangulation
----------- 10 ms for Triangulation
----------- 10 ms for Triangulation
----------- 10 ms for Triangulation
----------- 11 ms for Triangulation
----------- 11 ms for Triangulation
----------- 12 ms for Triangulation
----------- 12 ms for Triangulation
----------- 12 ms for Triangulation
----------- 12 ms for Triangulation
----------- 12 ms for Triangulation
----------- 15 ms for Triangulation
----------- 13 ms for Triangulation
----------- 13 ms for Triangulation
----------- 13 ms for Triangulation
----------- 14 ms for Triangulation
----------- 14 ms for Triangulation
----------- 14 ms for Triangulation
----------- 15 ms for Triangulation
----------- 15 ms for Triangulation
----------- 15 ms for Triangulation
----------- 15 ms for Triangulation
----------- 15 ms for Triangulation
----------- 16 ms for Triangulation
----------- 16 ms for Triangulation
----------- 16 ms for Triangulation
----------- 16 ms for Triangulation
----------- 16 ms for Triangulation
----------- 17 ms for Triangulation
----------- 17 ms for Triangulation
----------- 18 ms for Triangulation
----------- 18 ms for Triangulation
----------- 18 ms for Triangulation
----------- 18 ms for Triangulation
----------- 18 ms for Triangulation
----------- 19 ms for Triangulation
----------- 19 ms for Triangulation
----------- 19 ms for Triangulation
----------- 19 ms for Triangulation
----------- 20 ms for Triangulation
----------- 20 ms for Triangulation
----------- 20 ms for Triangulation
----------- 20 ms for Triangulation
----------- 21 ms for Triangulation
----------- 21 ms for Triangulation
----------- 21 ms for Triangulation
----------- 22 ms for Triangulation
----------- 22 ms for Triangulation
----------- 22 ms for Triangulation
----------- 22 ms for Triangulation
----------- 23 ms for Triangulation
----------- 23 ms for Triangulation
----------- 23 ms for Triangulation
----------- 23 ms for Triangulation
----------- 23 ms for Triangulation
----------- 24 ms for Triangulation
----------- 24 ms for Triangulation
----------- 24 ms for Triangulation
----------- 24 ms for Triangulation
----------- 24 ms for Triangulation
----------- 25 ms for Triangulation
----------- 25 ms for Triangulation
----------- 25 ms for Triangulation
----------- 25 ms for Triangulation
----------- 26 ms for Triangulation
----------- 26 ms for Triangulation

I am using OpenMP instead of CUDA, not sure if this is related.

[elalish]

You need to take those labels with a grain of salt; in this case there won't be any polygons to triangulate, so the triangulation step is only copying the old triangle lists to the new one. This should be a pretty small fraction of the time taken for the operation, am I right?

[pca006132]

Indeed, it seems that it is mostly copying old triangle lists.