I thought the compression ratio of JPEG was a global setting and not something that could be varied for different parts of the image. Does this approach actually produce smaller files than other video compression techniques? Or is it done just to produce that effect?