After thinking this through some more, it seems like if you are starting with existing geometry, then iterating over existing primitivesets and creating new primitivesets with one primitive per primitiveset (and then using BIND_PER_PRIMITIVE_SET) is the easiest solution with regards to code complexity.
Using duplicate vertices seems the better option if you are creating your geometry from scratch and have control over that geometry creation. For example, one of my requirements was to use the building extrusion geometry from osgEarth and to color those per-primitive. I was able to write a fairly simple nodeVisitor that accomplishes this by splitting the primitivesets up so that I got one primitive per primitive set (and removing the original primitivesets). As a second example, I also had some code that was converting hdf files to osg files and ended up using the second method of using duplicate vertices because I was generating everything from scratch including the vertex array. I realize I could do either of these examples either way, but if the efficiency is in the same ballpark I'll just use whichever is easier on a case by case basis. There isn't a silver bullet for this problem. Thanks to all for the feedback and patience, it took me a while to wrestle with this but it definitely deepened my understanding of osg::Geometry in general. ------------------ Read this topic online here: http://forum.openscenegraph.org/viewtopic.php?p=57898#57898 _______________________________________________ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
