@thomasfuchs Like I said, I understand that's the most efficient and obviously accurate method. I'm contemplating whether there might be a more sophisticated, computationally intensive way to get similar results from a single set of images.
As in, there's a lot of additional information (the star field as a reference, manually entered position of the comet in the image, the comet ephemeris data, etc) that maybe could be leveraged to tease out the comet image? Just idle speculation.