Currently, for any change to a frame sink, the capturer will execute a CopyOutputRequest on the whole frame. Instead, it should only do so for the damaged region; and then apply the change on top of a prior frame. This may or may not result in a significant performance improvement.