Suggestions for rejecting frames with SubframeSelector

[DaveB]

I'm curious, for people that use the SubframeSelector script to reject frames, what kind of expression(s) do you use? And, if I missed a tutorial that describes it, let me know. I have seen suggestions/tutorials for what to use for Weighting, but I didn't see anything for Approval/rejection. I just use Blink, and perhaps visual inspection is the preferred method, but I figured that I'd check.

Thanks

[mschuster]

I like to look at the plots and reject outliers manually. IMO even better is improve your setup so that fewer outliers are generated next time. By improving focusing, alignment, and tracking over a couple year period, I was able to basically eliminate outliers due to those issues. Actually, observing from good sites, my most recents sets reject nothing.

For weighting I've moved to expressions that are frame set independent. Like the simple SNRWEIGHT or the more compex SNRWeight / FWHM^2.

[DaveB]

Thanks. I guess part of my question (which I didn't explicitly mention) is this: If I have frames that were taken in hazy or moonlit conditions, they have a much brighter background when compared to images on moonless, clear nights. Yet, they often have very good SNR and FWHM compared to other subs. Am I incorrectly assuming that these images aren't good images to use? I would guess that PI thinks that the background glow is signal rather than noise (which it is), but I've been thinking that the lack of contrast would make these candidates for rejection.

Also, with your weighting expression of SNRWeight / FWHM^2, that is self contained to the single image, correct? I take it that is what you mean by "frame set independent", i.e. no need to calculate a min and max of FWHM, SNRWeight, or Eccentricity of all frames like some of the tutorials suggest.

Thanks again.

[mschuster]

I'm usually targeting nebula so haze and moonlight make rejection candidates. On the other hand, if I wanted to resolve the center of a globular cluster, FWHM is more important and the rest would not be an issue. You might want to try several integrations to see the impact of differing rejections.

Yes, expressions that depend only on the frame itself.