Hi All, I now have enough of the new ViewDependentShadowMap technique implemented for members of the community to start testing it out on their own data. From the testing I've done so far it looks to be signifcantly faster and produces more consistently high quality results than any of the other techniques we have had so far. The algorithm isn't based on any one paper, but one could say it in similar to perspective/light space perspective shadow maps but I abdoned the maths behind these approaches as I just couldn't get them to work reliably enough over a wide range of data.
The new technique isn't finished yet, it's certainly far enough along for others to start testing and to report back results. The new technique isn't compatible with depth partitioning yet, and while it does handle spot lights it do so avoiding tackling the perspective transform in light space. My hope with the shdow implementation is that it'll be robust and flexible enough to replace all the old shadow techniques. The new techhnique also lays down the framework for future support of mltiple shadow maps such as with cascaded or parallel split shadow maps, as well handling multiple lights+shadows. If we can get it working robust enough I'll removal all the old techniques and enable us to have a lighter weight osgShadow library. All the changes are now checked into svn/trunk. To test use the --vdsm option in osgshadow, or use the new osgShadow::ViewDepedentShadowMap technique class in place of your existing shadow technique used in your application. Please let me know how you get on. Robert. _______________________________________________ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org