Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!wuarchive!zaphod.mps.ohio-state.edu!ub!uhura.cc.rochester.edu!rochester!cornell!deb
From: deb@elli.cs.cornell.edu (David Baraff)
Newsgroups: comp.graphics
Subject: Re: Cumulative error in rotation matrices
Keywords: matrices, errors, aipdc@castle started this thread
Message-ID: <49139@cornell.UUCP>
Date: 30 Nov 90 19:49:49 GMT
References: <7386@castle.ed.ac.uk>
Sender: nobody@cornell.UUCP
Reply-To: deb@cs.cornell.edu (David Baraff)
Distribution: comp
Organization: Cornell Univ. CS Dept, Ithaca NY
Lines: 12

Try using unit quaternions to represent your rotations.
After each multiplication, the quaternion is renormalized,
preventing error from building up.

I got this idea from the SIGGRAPH '88 paper "A modeling system
based on dynamic constraints", by Ronen Barzel and Al Barr.
Computer Graphics (Proc. SIGGRAPH), vol 22, pp. 179 - 188.

I use this for my dynamic simulator and it works fine.

	David Baraff
	deb@graphics.cornell.edu