PSF wings

[gbernstein]

Erin's photometry tests show that 17x17 PIFF PSFs are not large enough to capture behavior of aperture vs PSF mags properly. It might be very useful to add to the pixelized PSF model the capability to fit a lower-resolution extended wing (maybe a power law or other assumption of circular symmetry) that would make it possible to build bigger PSFs without introducing infeasible numbers of DOF.

[rmjarvis]

I suspect this will be best done via the composite model that we are considering for the optical + atm PSF model. There the full model will be the convolution of several components. I think PixelGrid x Airy would probably do a decent job at this with only a couple extra degrees of freedom over the full PixelGrid in the center.
I don't know that convolution is ideal. Possibly addition is better. But once the Convolution composite is working, I think it will be trivial to also do a Sum one. (You could even have an arbitrary hierarchy of nested composites of the different types if you were so inclined.)

[gbernstein]

I was thinking of something simple (and additive), like a power circular power law that covers pixels beyond the PixelGrid and might be feasibly added as a single extra DOF for this model to keep it linear. [I think the large-R behavior of a typical ground-based PSF is dominated by scattering and doesn't look like the Airy tails.]