The functionality of this function: https://docs.astropy.org/en/stable/api/astropy.wcs.utils.fit_wcs_from_points.html is really begging for a tutorial. It could conceivable be added to the photometry guide, but sometimes you just want to do this for astrometry too, so I think it could be a good stand-alone tutorial. I would imagine the following (I've split this into 3 parts which could be a 3-part tutorial or just three sections of one tutorial):

Part 1:

• Start from an image that is slightly distorted with no wcs but where you can say "we know about where this is pointed" and can identify by eye a few bright stars.
• Use those with fit_wcs_from_points to get a "rough" WCS solution (i.e. just basic TAN projection)
  Part 2:
• Use the rough solution to cross-match to a catalog with known-good astronometry (e.g. Gaia, SDSS, or POSS)
• Re-fit the wcs with fit_wcs_from_points but using a fancier projection that includes some sort of distortion correction.
  Part 3:
• Find a second image that's similar to the first but offset just a bit (e.g., a dither, or from a different night for time-series photometry)
• Repeat parts 1-2 (or maybe just 2 if the "rough" solution is good enough) to get a WCS including distortion corrections
• Use reproject (https://reproject.readthedocs.io/en/stable/) to project the second image to the wcs of the first image
• Either coadd the two images (yielding a better image) or subtract them (to see variable stars pop out)