Quality Score

I'm curious, is there an official formula published for how the Quality Score is calculated, similar to the "acceptance criteria" often used in PI?  I assume it includes FHWM, Star Count, Eccentricity, PSF in the Star Analysis phase, but is there any weighting for each of those?  Are there any others, like perhaps contrast detection to identify subs that might have been taken with dew on the mirror/objective?  My apologies if I've missed this documented somewhere.

Also, I know that the calculated quality score changes with the Registration and Normalization phases.  The general way that I use this score is to sort my Star Analysis results, exclude the poor-quality outliers, then Register, sort on Quality again and repeat, then Normalize, sort on Quality again and repeat (though I often don't have any to throw away at this stage).

I tend to vacillate on what my acceptance threshold is (for only the Star Analysis phase).  I would normally shoot for no less than 1 standard deviation lower than the mean... but APP doesn't calculate S.D.  If I'm quick and lazy, I'll de-select the bottom third of the quality scores for each filter.  If I'm really being picky but still lazy I'll kill the bottom half, but that's throwing away a lot of data and my pickiness usually doesn't produce a noticeably better image.  Much of the time I will just drop the obvious outliers.

It would be awesome if the Star Analysis tab allowed me to define my acceptance criteria, then auto-select only those subs that meet it after Star Analysis completes.  Choose .5, 1 or 2 Standard Deviations from a dropdown menu and Star Analysis would select only the subs that meet that threshold.  (I'm not sure I'd do the same for Registration or Normalization, as it may result in a more data loss than needed, but it'd be interesting to try it).