Your comments

Good point! The implementation that made it into 1.2 supports either using the JPL backend *or* some polynomial approximations that aren't quite as precise.

This topic is pretty broad, but clearly has a lot of interest. In the aim of understanding a bit better what should be focused on, I've created three somewhat more specific topics:

But note that this suggestion is still relevant as a large-scale view, and a need for these to be well-integrated into a full workflow!

Note that there's a lot of activity in this direction going on as part of the JWST Data Analysis Tools sprints, along side the Google Summer of Code project. So folks should watch the photutils.psf space in the near future!

For context: the IRAF geomap/geotran tasks are used to generate distortion maps *and* do the "undistorting".

In the current astropy ecosystem the former task might be taken up by (non-existent) tools that generate astropy.wcs or gwcs representations. Then other tools like reproject could use that to do the image distortion itself. That makes it easier to mix-and-match different needs for different science cases.

@Fred Moolekamp: have you had a look at the changes to PSF machinery in photutils (the PR I linked above)? It might be worth your while to start thinking sooner rather than later about how to fit your machinery into that framework. If we do end up having a GSoC student working on this for the coming summer, that might be a good opportunity.

Note that a lot of progress in this direction has been happening in this pull request of photuils. That will probably be in very soon, and be in need of some user testing and feedback.