From: twatson@hfnet.bt.co.uk (Toby Watson)
Subject: Re: SCI: Voxels vs. Polygons
Date: Fri, 21 Aug 1992 08:12:28 GMT
Organization: BT Laboratories



From article <1992Aug20.181228.9880@u.washington.edu>, by bobp@hal.com
(Bob Pendelton):
  
>> How do you define rotation with voxel-objects, it seems that any
>> inaccuracy (inevitable?) in final coordinates would randomize the
>> voxel-object after significant rotations?
> 
> I haven't done this, so this could be pure ****, but I suspect that
> the shearing based rotation techniques that work for pixmaps can be
> generalized from 2 to 3 dimensions. Cumulative rotations might have
> nasty end results. But, if you keep track of the total rotation and
> always generate the rotated object based on the base object and the
> total rotation, instead of doing incremental rotations, you should get
> good results. And, I doubt that it is cheaper to use incremental
> rotations.

  Voxel objects can be rotated, you use the same maths that you use
  for polygon vertice rotations. On the issue of cumulative errors,
  this is the same as 2D texture rotation, you have to keep track
  of calculation accuracy and ensure that the sample frequency is
  high enough to avoid spatial aliasing. Easy enough and voxels will
  be a part of VR someday....but not on a PC !

----------------
Toby Watson - I really must get a signature sorted out...
----------------