I think an extension to the OSG:::Geomentery makes sense, but to make it extensible to end user, I think that its important to let the end user design what part of the transformed vertices get included in the vertex feedback stream (the vbo).
The application I was looking to use TF was to couple it with the gpu tesselation for mesh refinement... ( Start with a mesh, split it with the tesselator, write it to a vbo).. Will that being said, I think the mods I made to the core would still be necessary, no matter where you make the TF calls .. Let me know what you think, Aaron Hagan On Wed, Oct 10, 2012 at 12:18 AM, Roman Grigoriev <[email protected]>wrote: > Hi,Robert > > > > When you say objects, what specifically do you mean. Subgraphs? Leaf > nodes? > > I mean geometries because for now main purpose as I see is precomputing > something. > For example for culling geometries as I saw in this > http://rastergrid.com/blog/2010/02/instance-culling-using-geometry-shaders/ > Aaron make it perfect in his example. > In first stage we use TF to cull invisible nodes and after we render only > visible nodes so we use first geometry to cull and second to render. > Could you please provide me example that uses TF in different way? > Using drawcallbacks is Ok because you need to cull every frame. > BTW in OGL 4.3 we can use compute shaders to make this work. > > Thank you! > > Cheers, > Roman > > ------------------ > Read this topic online here: > http://forum.openscenegraph.org/viewtopic.php?p=50535#50535 > > > > > > _______________________________________________ > osg-submissions mailing list > [email protected] > > http://lists.openscenegraph.org/listinfo.cgi/osg-submissions-openscenegraph.org > >
_______________________________________________ osg-submissions mailing list [email protected] http://lists.openscenegraph.org/listinfo.cgi/osg-submissions-openscenegraph.org
