Hello all, Yes I've posted tangential posts in the past but I'm returning to this work after a hiatus and would like to make sure I'm using the correct library/technologies as I dive back in.
I'm looking to use an osg-compatible post-processing library that is capable of the following: input: 32-bit float framebufferobject processing step 1: get global min, max, mean, histogram required for step 2 processing step 2: histogram equalization converts to 8-bit unsigned int processing step 3: add motion blur processing step 4: add fixed and/or random noise output: screen-sized quad of the results I had a basic setup working last year using osgCuda, but I would like to try to move away from cuda/opencl and towards a shader approach for increased portability. As I see it this means using either osgPPU or osgFX::EffectCompositor, which is part of Wang Rui's osgRecipes. I know that these two libraries do similar things, but since neither is part of the officially maintained 1st-party OSG distribution I'm trying to figure out which is likely to see the most support going forward, and which might be best suited for the processing steps I've listed. In particular I worry about the min, max, mean, and histogram calculations and whether these can be done more easily using computer shaders than in the past when more complicated parallel reduction algorithms were required. Any comments on this? Thanks, -Ethan ------------------ Read this topic online here: http://forum.openscenegraph.org/viewtopic.php?p=56737#56737 _______________________________________________ osg-users mailing list [email protected] http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
