Hi Guys, I do wonder if we should avoid using polygon offset as its implementation varies across different graphics hardware. Using the projection matrix, the depth range, or shifting of the texgen matrix might produce a more portable result. Which to deploy depends upon the context. This is a general issue for all of osgShadow algorithms.
Might I suggest a general discussion on osg-users about this topic, perhaps when Wojtek returns from his vacation. Robert. On Wed, Mar 5, 2008 at 8:31 PM, <[EMAIL PROTECTED]> wrote: > Hi sorry i am on vacations writing from cell so i cannot verify if i did > modify polygon offset. But if i did it was by mistake and completely > unintended. In fact i doubt i did it because i tried to stay away from > delicate matters like depth bias or shaders used by technique. Version i > updated indded used self shadowing and i did not change that. I tried to > fix three issues that were genuine bugs imho. Improper LOD behaviour, > wrong lighting direction when database was not centered at 0 and non > functioning debug hud. I also added one precision tweak but this one was > carefully described and hopefully carefully examined by Robert. > Cheers,Wojtek > ,> Hi, > > > > > > I've merged my changes with the current SVN and experienced some strange > > things too. > > Wojciech I think has added negative polygonoffset when culling back > > faces. This leads to selfshadowing everywhere on my ATI card. > > When supplying positive values however it works on ATI and NVIDIA (when > > choosing a bigger value though) > > > > I've attached the merge as is, since my changes involve a function to > > set the polyoffset from outside (see attached description for further > > details). > > I also agree to Robert, that we need definitely more unit tests for > > this, but I have that strange feeling, that won't be enough, since I had > > very different experiences with different card/driver combinations. > > (e.g. ATI .. they have inverted the t-coordinate of the FBO-Texture in > > their newest drivers ...) > > > > I hope my changes are of any value. > > > > cheers > > psy > > > > _______________________________________________ > > osg-submissions mailing list > > [email protected] > > http://lists.openscenegraph.org/listinfo.cgi/osg-submission > > _______________________________________________ > 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
