For some of the tests we want to run, we want to be able to compute some ellipticity parameters with a very simple estimator. We can start by a moments based approach, although in principle we can also back-propagate through a fitting algorithm, so we could do an ngmix-like model afterwards. And ultimately, we anyway probably want to use some neural estimator ;-)
So the goal of this issue is to track the development of a simple shape measurement method, a tf function, that would take a postage stamp as input, maybe a PSF (although we don't really care because we are doing metacal and we reconvolve by a circular PSF), and return e1, e2 (in batch ideally ;-) )
For some of the tests we want to run, we want to be able to compute some ellipticity parameters with a very simple estimator. We can start by a moments based approach, although in principle we can also back-propagate through a fitting algorithm, so we could do an ngmix-like model afterwards. And ultimately, we anyway probably want to use some neural estimator ;-)
So the goal of this issue is to track the development of a simple shape measurement method, a tf function, that would take a postage stamp as input, maybe a PSF (although we don't really care because we are doing metacal and we reconvolve by a circular PSF), and return e1, e2 (in batch ideally ;-) )