it seems to be a property of optical systems with a secondary mirror that's mounted on spider vanes. i get bloated stars from newts and R-Cs on a regular basis.
the masked stretch script works pretty well to keep the stars under control, but the problem is you can't do the entire stretch with that script (i've modified the script to allow me to reach larger medians and the results are not pretty - the background of the image becomes almost completely grey and there's no contrast left). when you do a final HT stretch, the bloat is back.
just today i was working on combatting this and tried to use Vicent's tutorial
http://pixinsight.com/tutorials/NGC7023-HDR/en.html to copy the stars from a version of the image stretched with masked stretch to the same median as the HT stretched version. the results were encouraging but not perfect. the haloes were still there but at least at the center of the halo there was a star, albeit sort of flattened out. i tried to restore the gaussian profile with Atrous but that's not really right either...
