Hi Cyril, Rather than spend lots of time investigating this issue with the present osgShadow implementation I would recommend waiting till I have the new codes checked in. I haven't competed the work yet but have early results that show that the light space/shadow volume computation is far more robust than the old view dependent shadow techniques - I don't see clipping of the shadow or wild jumps in shadow resolution. I'm testing the new code on models that pick out the problem areas for the old techniques, and would suspect that as my new codes address the problems in these problem areas that there is good chance it'll address the problems you are seeing as well.
So far I've written a new base class, osgShadow::ViewDependentShadowMap, that provides a framework for setting up and rendering view dependent shadow techinques. This base class currently implements a minimal bounds shadow map, this base class will be usable out of the box for doing minimal bounds shadow mapping, and provide the ability to provide custom light space/shadow map projections such as perspective/LispSM shadow maps just by subclassing or callbacks. Today I'll be writing some provisional shaders for the minimal bounds shadow map that should give a basic drop in replacement for the current osgShadow::MinimalBoundsShadowMap implementation. Through next week I'll be making the shader codes more powerful and flexible, and looking at add multiple shadow map techniques and alternative projections for the light space/shadow maps. If all goes well in a couple of weeks I'll have replacements for most of the exist shadow map implementations. Robert. _______________________________________________ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
