Building EPSF model from PSF images

[anahita25kh]

I am trying to use photutils to perform PSF photometry on my simulated images. I already have the PSF images and I am trying to make the PSF models in the format appropriate for Photutils. I use photutils.psf.EPSFModel and my PSF image as an input to make the PSF model. When I check my final flux value with the true flux values (I know the brightness of my input sources), all sources are systematically brighter for about ~25%. I have checked all of the other possible causes of this issue including background, etc. Everything seems to be estimated fine. Another uncertainty comes from the PSF model. As I understand EPSFModel code is supposed to take the stars as input to build the model assuming an oversampling value from the user. However, I am using an oversampled PSF image as an input to build the model. The only reason for doing such is to have a psf in the proper format for photutils. So I am wondering if something is not right with my epsf model and so it is overestimating my fluxes.
I would appreciate your help.

[larrybradley]

@anahita25kh If you already have an oversampled PSF image, then I would try FittableImageModel (https://photutils.readthedocs.io/en/stable/api/photutils.psf.FittableImageModel.html#photutils.psf.FittableImageModel) instead of EPSFModel. By default the EPSFModel is normalized within a circular radius of 5.5 pixels (I think to match Jay Anderson's ePSF paper). The FittableImageModel may give you a normalization that you expect.