Path: utzoo!mnetor!uunet!mcvax!enea!luth!cad!sow
From: sow@cad.luth.se (Sven-Ove Westberg)
Newsgroups: comp.graphics
Subject: Re: Help Wanted: Filling Voxels
Message-ID: <996@luth.luth.se>
Date: 6 Mar 88 13:51:04 GMT
References: <198@dutrun.UUCP> <23153@ucbvax.BERKELEY.EDU>
Sender: news@sm.luth.se
Reply-To: sow@cad.luth.se (Sven-Ove Westberg)
Organization: University of Lulea, Sweden
Lines: 26

In article <23153@ucbvax.BERKELEY.EDU> ph@degas.Berkeley.EDU.UUCP (Paul Heckbert) writes:
|[ In article <198@dutrun.UUCP> winffhp@dutrun.UUCP (ruud waij) asks
|  how to find the set of voxels intersecting various modeling primitives,
|  such as boxes, spheres, cones, and cylinders.  ]
|  
|Yes, interesting problem!  Fitting a bounding box around the object and listing
|that object in all voxels intersected by the bounding box will be inefficient as
|it can list the object in many voxels not intersected by the object itself.
|Imagine a long, thin cylinder at an angle to the voxel grid.
|

Yes this is a real interesting problem. But it is similar to the problem 
of converting a boundary representation to a block model. So it is
possible to find good algorithms in the cad literature.

One reference that converts a polyhedon to a block model is:
Tamminen M, Samet H, Efficient Octree Conversion by Connectivity
Labeling, Computer Graphics 18 (3), pp 43-51, July, 1984.

Perhaps Markko Tamminen or Martti Mantyla, at Helsinki University
of Technology, Finland could give us some good comments on this problem.

Sven-Ove Westberg, CAD, University of Lulea, S-951 87 Lulea, Sweden.
Tel:     +46-920-91677  (work)                 +46-920-48390  (home)
UUCP:    {uunet,mcvax}!enea!cad.luth.se!sow
Internet: sow@cad.luth.se