Path: utzoo!attcan!uunet!cs.utexas.edu!tut.cis.ohio-state.edu!n8emr!uncle!oink!jep
From: jep@oink.UUCP (James E. Prior)
Newsgroups: comp.graphics
Subject: Re: Ray Traced Bounding Spheres
Keywords: sphere, ray, trace
Message-ID: <25@oink.UUCP>
Date: 16 May 89 14:17:19 GMT
References: <17241@versatc.UUCP> <17249@versatc.UUCP> <8411@phoenix.Princeton.EDU> <SARREL.89May14223947@galley.cis.ohio-state.edu> <4057@rpi.edu> <11444@bloom-beacon.MIT.EDU>
Reply-To: jep@oink.UUCP (James E. Prior)
Organization: Random Prime Research Institute  Columbus, Ohio
Lines: 21

In article <11444@bloom-beacon.MIT.EDU> tada@athena.mit.edu (Michael Zehr) writes:
...
<The real problem is to speed up a ray-tracing program by providing some 
<sort of bounding sphere for an object.  This is really 2 problems: 
< 
<1. given a set of points, find the minimum-bounding sphere 
<2. given a set of points, quickly find a "good" bounding sphere 
...
<if the data isn't static (the points move with respect to each other 
<between each run of the program) then you want to find the bounding 
<sphere quickly.  and it's better to quickly find a sphere that's 
<slightly larger than what's needed if it takes a less time to compute 
<it. 

I would always want the closest fit, not merely a close fit, regardless of
how many frames I was calculating.  Ray traced images are so expensive
computationally, that spending the time at the beginning of each frame 
for the closest fit will speed it up.  If it's a good optimization for
a single frame, you should expect the improvement for doing many frames.
-- 
Jim Prior    jep@oink    osu-cis!n8emr!oink!jep    N8KSM