Formally, both the ray origin and the direction derived from the
inside_vector for the inside mesh test would see an inverse transform
into the mesh's coordinate space. This would sometimes result in a
change in the inside_vector/ray direction with respect to the mesh.
Such direction changes are problematic given test ray directions roughly
perpendicular to a triangle's normal are noisy with respect to
ray-triangle intersections.
Running the attached test scene we correctly get the attached image where a planes texture is controlled by an object pattern based upon the mesh. The star mesh has a flat back very set just inside the plane surface. If the scene is run against the current master, the inside of the star mesh is missed.
Formally, both the ray origin and the direction derived from the inside_vector for the inside mesh test would see an inverse transform into the mesh's coordinate space. This would sometimes result in a change in the inside_vector/ray direction with respect to the mesh. Such direction changes are problematic given test ray directions roughly perpendicular to a triangle's normal are noisy with respect to ray-triangle intersections.
Running the attached test scene we correctly get the attached image where a planes texture is controlled by an object pattern based upon the mesh. The star mesh has a flat back very set just inside the plane surface. If the scene is run against the current master, the inside of the star mesh is missed.
MeshInisideVect.pov.txt