When images result from multiple images (e.g. averaging), we should consider writing out a .json sidecar stating which files the images came from, for transparency and bug hunting.
This should be pretty easy to do, but would be a nice touch to make sure our images are what we think they are