I’m finding that the Nvidia Background TOP either takes 2 or 3 frames to do its thing. Unfortunately, if you composite it with the original shot, and use a cache TOP to match the lag, if you’re off by even one frame, the original footage won’t align with the mask when there’s motion. Would it be possible to somehow have the Nvidia TOP pass along an integer of how many frames behind its input it is so that we can use that info for the cache top?