How to best combine StarNet star removal and deconvolution?

I am learning how to use deconvolution tool. If I use PSF from stars and modify step size and iterations I can control deconvolution and generate nice results from a fits file with nebulae and stars together. I am using SIRIL from git master.

PSF from stars shall be generated from linear data since documentation states “…the use of a PSF from star PSF can only be done on linear images”.

What is the best process if I also want to use StarNet star removal and deconvolution and then recompose for final stretching? Is it to:

  1. Open linear result.fits, select stars and generate PSF for deconvolution and save PSF.

  2. Run StarNet star removal.

  3. Run deconvolution using saved PSF on starmask (stars only). Save new deconvolved starmask.

  4. Run Star recomposition and use deconvolved starmask + starless. Stretch for final image.

or, is there a better way?

EDIT: I mean, is it better to deconvolve the result fits file first and then save a linear deconvolved file, which is then separated into starless and starmask using StarNet star removal?

I am replying to myself, but if 1-2-3-4 above includes a ‘pre stretch’ as part of StarNet star removal I break “…the use of a PSF from star PSF can only be done on linear images” => better to deconvolve first?