Path: utzoo!attcan!uunet!oddjob!ncar!boulder!sunybcs!dmark
From: dmark@cs.Buffalo.EDU (David Mark)
Newsgroups: comp.graphics
Subject: Re: Request for info on octrees...
Keywords: v is voxels....
Message-ID: <550@cs.Buffalo.EDU>
Date: 7 Aug 88 20:36:07 GMT
References: <1070@ucsd.EDU> <1074@ucsd.EDU> <157@edai.ed.ac.uk> <1081@ucsd.EDU>
Reply-To: dmark@sunybcs.UUCP (David Mark)
Distribution: comp.graphics
Organization: SUNY/Buffalo Geography
Lines: 19

In article <1081@ucsd.EDU> hutch@net1.UUCP (Jim Hutchison) writes:
>
>  [some material removed]      Unfortunately Voxels are not always as
>similiar in size as pixels.  Pixels are well behaved little polygons
>all the same size (squares, rectangles, hexagons, etc.).  Voxels are
>optimally the size of the largest volume of similiar material/color
>they can encompass.  Voxels are more a data structure than a sampling
>region.  Or do I misunderstand pixels?

No, you misunderstand *voxels*.  In the octtree literature I am familiar
with, all voxels are the same size.  If any sub-cube above some minimum
threshold is partly inside and partly outside the solid, it is subdivided
into 8 sub-cubes of the next lower level.  Sub-cubes which are uniformly
inside or uniformly outside the solid are "leaf-nodes" in the octtree.
THE voxel is the smallest allowable sub-cube.  This is, I believe, the way
voxels are defined in octtree papers by Gargantini, Samet, and others.

David M. Mark
dmark@joey.cs.buffalo.edu