My investigation with the "sbad.osg" model reveals that the Simplifier is not removing the centermost edge because it thinks it is a boundary edge. It thinks this because the two vertices are repeated twice in the model, each time with a different normal. The Simplifier encounters the edge twice, each time with only one triangle. If I go in and hack the model to use the same normal everywhere (as in the "sgood.osg" model), then the Simplifier encounters the edge once, with two triangles (one on either side), and does not consider it a boundary edge, and proceeds to collapse it. I can see the value in not collapsing boundary edges for the purpose of generating terrain tiles. However, I wonder if there is some value in providing a mode that would allow boundary edges to be collapsed? I'll run some tests and see if I can obtain good results with some trivial changes. -Paul
_____ From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Paul Martz Sent: Thursday, October 16, 2008 11:40 AM To: 'OpenSceneGraph Users' Subject: Re: [osg-users] Simplifier algorithm Robert -- Thanks for the previous explanation. I believe that the EdgeCollapse is confused about its definition of an error. I've attached two nearly identical models. To simplify them: osgconv --simplify .3 sgood.osg sgood-out.osg osgconv --simplify .3 sbad.osg sbad-out.osg sgood.osg simplifies as I would expect -- the center edge collapses, and the model retessellates. The model sbad.osg does not simplify at all. Note, however, that if the analogous center edge were removed from sbad.osg, and the model retessellated (as was done for sgood.osg), the resulting model would be identical in appearance to the original. Thus, collapsing sbad.osg's center edge would not introduce any error to the model. That right angle along the x-axis in sbad.osg appears to make the EdgeCollapse class incorrectly conclude that removing the center edge would introduce an error, when, in fact, it would not. I understand the Simplifier is used by VPB for terrain. However, I'm wondering if it could be enhanced to handle the sbad.osg model. I can probably spend a few hours digging in. I'd appreciate any pointers (yes, I know it's been a while since you coded it :-). Paul Martz Skew Matrix Software LLC http://www.skew-matrix.com <http://www.skew-matrix.com/> +1 303 859 9466
_______________________________________________ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org