FWIW I also use a system similar to what is presented in that Battlefield paper. On my Xeon W3550, my SSE optimized code clocks in at 16k sphere-frustum checks in at about 0.12ms (1 thread), which is plenty fast enough for me. It didn't take too much effort to rework the data structures so the spheres were accessed linearly in memory and really the code ended up considerably simpler (IMO).
Posted by synulation
on 21 January 2011 - 03:09 PM
To add to this there is a GLSL version available that I believe is similar (if not the same, not sure)if people are itching to play around with some code. The comments are in French though. http://igm.univ-mlv.fr/~biri/mlaa-gpu/
Did you try this one? I want to convert this from GLSL to HLSL... Is this is an easy job?? Any comments will be appreciated
I based my initial hlsl implementation off of that and found it relatively easy to work through. You can take a look at my experiments (including the hlsl code) with mlaa here: http://divergentcode...g/antialiasing/ I'm more than happy to answer any questions about it as well.