home *** CD-ROM | disk | FTP | other *** search
- In article <D95284.Lo5@avalon.chinalake.navy.mil>, Des McPherson <des_mcpherson@cl_63smtp_gw.chinalake.navy.mil> says:
- >
- >collision is the term that they use and yes it does mean more than 2
- >polys sharing an edge. penetrating is ok, collinears are ok also, watch
- >out for merging points though. They have a page of examples that a
- >model must pass before approval.
-
- Yes, but Viewpoint also confuses the Wavefront notions of group
- and material, too. (That is, they use materials for sub-objects,
- and 'g' groups to describe materials.) It's backwards.
-
- >>I'm also not sure why any of these things would be a real problem in
- >>practice. What are the consequences if an object doesn't pass the
- >>polygon "collision" test?
- >
- >it wouldn't be a problem in LW, but when they translate the models into
- >some 50 other different formats, it sounds like other 3D programs go
- >crazy when the model is translated. A good example is boolean functions
- >in LW. Say you punch a hole through a polygon which then makes it
- >"concave" and then try to convert it into a format that only takes 3
- >point polys (Imagine for instance). What would happen is that the hole
- >would get covered up upon conversion. "No Concave Polygons" is also
- >"bad".
-
- You mention two problems here: one, if any program that purports
- to convert N-sided convex polygons to triangles is "covering up"
- areas, then that program is severely broken and you should stop
- using it.
-
- Two, there's the issue of how different 3D programs determine
- the outward normal, and what's allowable as polygons.
-
- When my company's program, InterChange, translates from LightWave
- to Wavefront, it properly re-arranges the faces so that Wavefront's
- notion of convexity and outward normal are preserved.
-
-
-
-