Hi Robert,
2010/11/9 Robert Osfield <[email protected]> > Hi David, > > I'm just testing your changes and can't comment too much on the > specifics of the algorithm/maths as I'm not the author of this shadow > technique, so I mainly have to go on what the end results look like, > and whether the code looks sensible. > > In testing I used your "osgshadow --noUpdate --pssm --maxFarDist 500 > --minNearSplit 500 --mapcount 6 --debug-color model_test.3ds" > command line, and also tested it without the --maxFarDist etc. options > so used: > > osgshadow --noUpdate --pssm --debug-color model_test.3ds > > With this command line, ( without --minNearSplit 500), shadow camera is too near of the split when shadowMap is created, and some part of the model which cast shadow is clipped. As you can see in your screenshoot, error in red split is bigger than one in green split, because red split volume is smallest of the green one, and the wall which cast shadow is more clipped by red split that by green split when shadowMap is created. if you move the camera near the wall on right, error become smallest, if you move the camera far the wall on right, error become biggest, You can see minNearSplit option as : What is the minimal distance between split and an element in the scene for this element can cast a shadow in split ? default is 5.0. User must adapte this value to the scene. When I did this and then zoomed in close to the object I got an error > in the shadow region, with part of the shadow being clipped out. I've > attached the a pick of the problem. > > Do you see a similar problem? Any ideas? > > Robert. > > HTH David > > _______________________________________________ > 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
