home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.graphics
- Path: sparky!uunet!cs.utexas.edu!zaphod.mps.ohio-state.edu!hobbes.physics.uiowa.edu!news.iastate.edu!pi.eai.iastate.edu!boylan
- From: boylan@pi.eai.iastate.edu (Terran Boylan)
- Subject: POLYGON DECIMATION of big geometries
- Message-ID: <C187qp.EFA@news.iastate.edu>
- Sender: news@news.iastate.edu (USENET News System)
- Organization: Engineering Animation, Inc.
- Date: Thu, 21 Jan 1993 22:56:45 GMT
- Lines: 55
-
- We are currently working on writing routines to take big geometry
- files and make them smaller geometry files in an intelligent way.
- Sound fun? You bet!
-
- The most prevalent examples would come from marching cubes data, GIS
- data, or Cyberware data. This is not necessarily our case, but
- we _do_ generate a lot of big geometries (>50,000 polygons).
-
- I have the papers from SIGGRAPH '92, but they don't really provide
- the level of implementation details that I need. (Call me a SIGbaby,
- but it still seems like there are some SIGGRAPH papers that are
- written to help people solve problems and some that are written to
- publicize results! %^( )
-
- Is there any software (PD or otherwise) that is available to do this
- polygon decimation process? I talked to somebody from Cyberware a while
- back and they said that they had a program to do this but that it was only
- available with their systems. That may have changed by now, though. We're
- probably going to end up writing code to do the decimation process somehow,
- but if we write it it'll end up being specially tailored to our data.
-
- It seems to me like there are a couple of approaches to the problem:
-
- -- polygon removal
- -- shrink all pts on the poly to a single pt
- -- vertex removal
- -- resulting vertices are a strict subset of original vts
- -- resulting vertices are a combination of original vts plus
- new vertices based on the originals
-
- If anybody has done this before I'd be interested in what you ended up
- using for the "vertex/polygon" kill criteria. Obviously its made up of
- some of the following (plus some stuff I haven't thought of):
-
- -- whether a vt is an interior/exterior vt
- -- how many adjacent vts exist within a vt's "kill radius"
- -- difference between a poly's normal and adjacent poly normals
-
- I realize that the problem is somewhat different based on the following:
-
- -- whether original geometry topology is "nice" (like a uniform grid)
- -- whether polys with >3 sides are allowed
- -- whether concave polys are allowed
-
- The best possible case would be for somebody to mail me the C code for a
- user-friendly GUI program written in GL, but I'm not holding my breath. I
- promise that this is _not_ for a homework assignment and I am _not_ asking
- anybody to do my job for me. :^)
-
- Any information or ponters would be greatly appreciated.
-
- ---
- Terran J. Boylan, Sr. Artist/Programmer | "It's better to have loved
- Engineering Animation, Inc., Ames, IA | and lost than just to have
- (515) 296-9908 / (515) 296-7934 (> 5PM) | lost." -- Dorky Dog
-