Path: utzoo!utgpu!news-server.csri.toronto.edu!dgp.toronto.edu!mccool
Newsgroups: comp.graphics
From: mccool@dgp.toronto.edu (Michael McCool)
Subject: Re: 3d from 2D
Message-ID: <1991Apr12.114158.9410@jarvis.csri.toronto.edu>
References: <5d6f84k@rpi.edu> <1991Apr9.193032.28077@ohrd.uucp>
Date: 12 Apr 91 15:41:58 GMT
Lines: 16

>In article <5d6f84k@rpi.edu> onders@raphael.ipl.rpi.edu (Timothy E. Onders) writes:
>>A couple of years ago, there was an article on Nova(on PBS), about research
>>somewhere on converting a series of two dimensional images to a three
>>dimensional representation.  The example given was panning a camera past
>>a model of a town and generating the 3D image from that.

I think what you are looking for is optical flow, which can convert
a sequence of images into a depth map.  Jepson here at the 
University of Toronto has done some work in this area.  Note that
this does not give you a 3D representation, but only a depth map
(this might be alright for displacement maps, etc. though).  Deriving
a full 3D representation is basically the full vision problem,
which is NP-hairy.

Michael McCool@dgp.utoronto.ca, 
Dynamic Graphics Project, University of Toronto.