Inochi2D
MeshGroups.
Me handing off existing work after the final merge of ECS refactor. Here provides math for implementing meshgroups.
Three parts:
- Testing whether a point is in a triangle.
- Efficient (but not guaranteed correct) testing of a point is in which triangle among those in a given mesh, with a bitmask constructed for the mesh.
- Given parent mesh and parent deformation, deform child mesh. Essentially letting one parent triangle drag a child vertex that is contained in it, and this drag can be chacterized by an inverse matrix, which can be reused for efficiency.
Intensive test for bitmask included (with a .png depicting the base test case), testing that bitmask yields correct test results for points with the whole test space transforming.
However, as mentioned, this bitmask method is not guaranteed to be correct, thus it is hard to define a "good enough" test case. It is easy to construct counterexamples.
An experimental attempt on dynamic bitmask step size is commented out cause test suites do not like it. A step size of 1 should be usable in actual rendering anyways.
Out first step should be bothering Luna and try to get the 0.9 spec on meshgroups (now called "Mesh Deformer", with the legacy one named "Deformer") out.