It would be sufficient to do sample based post processing and do the pixel construction after that... I think I already mentioned this on the forum. It would give nice speedups for GI (and maybe line/edge rendering?) and it would also make the post processing more flexible (as in no need to double the resolution for post DOF). It might slow down the post processing quite a bit though...
On 08/06/06, David Coombes <[EMAIL PROTECTED]> wrote:
This video (50 MB, 50 minutes long) is very worth watching for how GI calcs can be sped up 10 fold including AA as standard. Definitely something the Meskannens should be looking into! http://www.acm.uiuc.edu/conference/archive/2005/video/bala.wmv There's about 3-4 mins of space at the front as the lecturer hall tries to sort itself out. David Coombes ...
