Re: [Algorithms] Polygon reduction and QEM

[Tom F. <tom...@ee...>]

> The problem we have though is that we have a lot of vertex on edge
> conditions in our geometry ( I know its bad )

Bad? "It's worse than that, he's dead, Jim." T-junctions are utterly evil -
none of these algorithms deal with T-junctions. They don't render correctly
even before you start collapsing. You need to fix those in some way - either
split the long edge at the T, or insert a degenerate triangle. *Then* you
can run the edge-collapse algorithm.


TomF.


> -----Original Message-----
> From: gda...@li...
> [mailto:gda...@li...] On Behalf Of
> Technomancer
> Sent: Friday, May 02, 2008 10:30 PM
> To: Game Development Algorithms
> Subject: [Algorithms] Polygon reduction and QEM
> 
> To whomever maybe able to help,
> 
> I am currently trying to take our artist generated models which are
> then
> placed in a 125m x 125m x 400m static scene with only welded position
> data and run a polygon reduction algorithm over it to generate a decent
> low lod version for use in game. I have looked at both the algorithms
> by
> Stan Melax and the Quadric Error Metric algorithm by Michael Garland
> and
> Paul Heckbert, and the QEM method seems to give me quite decent
> results.
> 
> The problem we have though is that we have a lot of vertex on edge
> conditions in our geometry ( I know its bad ) that the QEM algorithm
> i'm
> using doesn't like too much and we get holes appearing where there
> shouldn't be holes :( I have looked at a package made by Okino called
> Polytrans that uses the same QEM method of simplification and there
> implementation doesn't have this problem.
> 
> I was wondering what i'm doing wrong ? or is there something that they
> are doing extra, the white paper they have on their website
> http://www.okino.com/conv/polygon_reduction/geoman2/polygon_reduction8.
> htm
> has detailed information about their technique but for the life of me i
> can't figure out what extra steps they are doing to stop these holes
> appearing. They also seem to have the ability to stop collapsing edges
> when the model start to degrade again i don't know how they are doing
> this either.
> 
> if anyone has any insights it would be very much appreciated.
> 
> Thanks in advance
> 
> Garry Wallis
> 
> -----------------------------------------------------------------------
> --
> This SF.net email is sponsored by the 2008 JavaOne(SM) Conference
> Don't miss this year's exciting event. There's still time to save $100.
> Use priority code J8TL2D2.
> http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/
> javaone
> _______________________________________________
> GDAlgorithms-list mailing list
> GDA...@li...
> https://lists.sourceforge.net/lists/listinfo/gdalgorithms-list
> Archives:
> http://sourceforge.net/mailarchive/forum.php?forum_name=gdalgorithms-
> list