CompuServe Thread

Hi David, I'm assuming that you're trying to determine whether a point's inside or outside an arbritrary mesh. In other words, you aren't creating the mesh, just checking your point(s) against an already present mesh. I'd approach this problem in two parts; Start with the mesh's bounding box. You can get this info from the object's matrix and it's accessible from within a script or an IPAS routine. Once you have the bound box, determine whether the point falls within its boundaries. If the point is outside the bounding box, you can be sure that the point won't fall inside your mesh. If it falls inside the bounding box, you then have to determine whether it actually falls within the mesh object itself. There are a couple of ways of determining that. I'd probably use a marching cubes routine that subdivides the object's bounding box into smaller cubes and then I'd test each cube for the presence of a face or vertex. If a face or vertex falls within or intersects one of the cubes, mark that cube. Once you've got all the cubes marked (or unmarked), you have a pretty good idea of the surface of the mesh. Discard any cubes that are outside the marked cubes, they represent a volume that's outside your mesh. What you're left with is a bunch of cubes that intersect or are inside the surface of the mesh. Now, test each of these remaining cubes for the presence of your point(s). That's it. If the point is inside one of these cubes, it's probably inside the mesh. I say probably because it all depends on the level of accuracy you're using. It's possible to get very accurate results by controlling the amount of subdivision that you apply to the original bounding box to create your cubes. The more cubes you create, the more accurate your representation of the mesh's surface will be and the more accurate your determination of whether a point is inside the mesh will be. Of course, this makes the scan slower and more memory intensive too. Using the initial bounding box test helps minimize the amount of work you have to do by throwing out points that are obviously outside the mesh. This helps speed thing up quite a bit. I've written something similar in the past and it worked out great. Hope this helps you out. Let me know if you need a hand or if you would like more detail in the explanation. Cya, Danny…

>> I'm curious: what's your application? He he. In my (very limited) spare time, I'm still experimenting with various TSE modules and the basic graphics principles. In this case, I'm trying to fill a volume with evenly spaced objects. In another experiment, I'm also tinkering with meshing an mathematically defined surface. (I'm feeling like Tim Brannan who in his message to you just said he felt like "a monkey trying to comprehend quantum mechanics.") Don't count on any results in the near future. I'm really going to require some time and effort on these. I keep taking five steps back for each one forward. (On the positive side, it's gotten me back into mathematics – something I've ignored since college.) – Dave

>> Should be rather simple <wg>. Yeah, and so is quantum physics! I like your suggestion though. As time permits, I'll tinker with it. I'll post news of my success or failure here. – Dave