From: cmcl2!panix.com!entropy@uunet.UU.NET (Daniel Gross)
Subject: Re: TECH: Collision detection -- a possible new approach
Date: Tue, 10 Mar 1992 18:50:33 GMT
Organization: PANIX Public Access Unix, NYC



Perhaps useless, but.... PV-Ray (and just about every other CSG-based
ray-tracer I know) allows CSG primitives to be defined as the bounding
shapes of CSG complexes (unions, diff's, etc.). The ray-trace compares
against the simpler bounding shape (almost invariably sphere, cylinder,
or set of planes) and evaluates complex-CSG intersection iff the bounding-
shape test completes. The other advantage of this approach applied to
collision-detect would be that at a specified difference from POV, you
could consider the bounding-shape test to be conclusive.

Another trick for speed would be to specify CSG primitive bounding shapes
in fixed-point (i.e. int) math. Depending on your hardware, and certainly
in the case of the 486, this would give you some advantages from instruction
pipelining (in the case of multiple-object tests, you could actually begin
the integer bounding-shape test on another object before the floating-point
test on the previous complex-CSG object completes).

Is this do-able, or rather, how committed are you (in lines of code) to
the poly-line method at this point?

"La critique est facile, l'Art difficile..."

:-) Daniel.

-- 
  Daniel Gross              \  My opinions ALWAYS
  FLOW Research, Inc.       |  reflect those of my company.
  entropy@panix.com         |  If yours don't, consider quitting.