Am 17.08.2017 um 21:18 schrieb Bald Eagle:

> (b) if you only "scan" along that single vector - let's say, starting from any
> point along the other object's bounding box - then how do you handle the case(s)
> where the triangle being tested is perfectly parallel to the ray?
> How many intersection would it be calculated as?

In practice, this should only be a problem when all of the following
conditions come together:

1) The mesh must be in a non-union CSG.
2) The `inside_vector` must be parallel to one of the mesh triangles.
3) That triangle must lie in the same plane as one of the sibling object
surfaces.

If conditions 1) and 2) aren't met, this whole thing is a non-issue for
presumably obvious reasons. If condition 3) isn't met, the set of points
for which the test would become problematic is collapsed into
infinitesimally thin lines where the problematic triangles are projected
onto the sibling object surfaces, and the risk of any ray ever hitting
such a region is rather low.

That said, the described scenario may be one of the motivations (if not
the primary one) for making the `inside_vector` a parameter in the first
place, rather than just using a constant direction and toggling the
mechanism on and off with a boolean parameter.

Post a reply to this message