Can we apply any insights from image and video compression and partitioning to this kind of dimensional masking ?