Right now each Cutouts object is also a FileOnDiskMixin which is a waste because a lot of cutouts (at least 10 if not 100 or 1000) are going to share the same file. That makes the Cutouts table wider than it needs to be.
Instead, link all the cutouts from the same image to one CutoutsFile (or ListFile??) where we can keep track of things like filepath and MD5sum. This may also make it easier to load/save to file or copy to/from cache.
Right now each Cutouts object is also a FileOnDiskMixin which is a waste because a lot of cutouts (at least 10 if not 100 or 1000) are going to share the same file. That makes the Cutouts table wider than it needs to be.
Instead, link all the cutouts from the same image to one CutoutsFile (or ListFile??) where we can keep track of things like filepath and MD5sum. This may also make it easier to load/save to file or copy to/from cache.