Explore need for delta-function galaxy model basis.

[wmwv]

Is assuming smooth functional forms sufficient?
Do we need more pixel-to-pixel variation to capture the variations of the underlying galaxy.

1. Nearby galaxies, where the variation in surface brightness is resolved.
2. Lensed galaxies, where the magnification creates resolved surface brightness.

Scene modeling for SNLS, SDSS, and DES use sub-pixel basis with some smoothing/regularization. Will that be necessary?

While AstroPhot could easily apply a sub-pixel model, the more challenging question is
(a) how stable it would be to fit for a sub-pixel model in this framework.
(b) if that fit should be simultaneous (just more parameters) or iterative (perhaps based on a coadd).