Poweramp selects mp4 image based on image size. Unfortunately, tracks in the wild often have very low quality as first image if multiple images are embedded, this is why this approach as chosen. Also, there is no notion of front, back, or leaflet images for the format (there is no standard on "covr" atom at all) - just a list of images is there which we must interpret one way or another.
Also, this logic was slightly modified to choose first image if it's size is large enough (50kb).