I would love to see a DAOPHOT style of star subtraction. I know we can approximate it crudely with selective wavelet filtering and morphological operations, but that still falls short of directly subtracting a star profile.
DAOPHOT defaults, I believe, to a Gaussian profile over a constant background in the vicinity of a star. Higher level operations augment that profile with a matrix of excesses from the actual star profile, or an average of an aggregate of stars.
