Q: PixInsight - SubframeSelector options

[lazjen]

I've been heavily leaning on the SubframeSelector process in PI to help sort out what calibrated frames to keep.

The output of this process generates files with a weighting keyword added to use during the integration process. I've been following this tutorial: https://www.lightvortexastronomy.com/tutorial-pre-processing-calibrating-and-stacking-images-in-pixinsight.html and using this formula for the weighting:

(15*(1-(FWHM-FWHMMin)/(FWHMMax-FWHMMin)) + 15*(1-(Eccentricity-EccentricityMin)/(EccentricityMax-EccentricityMin)) + 20*(SNRWeight-SNRWeightMin)/(SNRWeightMax-SNRWeightMin))+50

Now comes the choices and questions.

I can also add approval formula as well, e.g. FWHM < X && Eccentricity < Y to exclude gross outliers.

However, given the weighting formula, is it worth just letting the weight value being lower as good enough for the integration, or is it better to aggressively weed out earlier and add the extra approval formula in to do exclusions? And if I do the exclusions approval formula, how aggressive should I go?

[ChrisV]

I use blink to first get rid of the obvious duds due to movement (my deck etc) or clouds etc. I run through the subs twice. Once showing the whole world to get rid of planes etc. Again zoomed in to look at star shakes

Subframe selector. I have some sort of algorithm like you have shown. I also use a rejection (I prefer to call it inclusion)
Fwhm < a && eccentricity < b && snrweight < c

Choose the values to prune off what look like outliers on the graphs for each - like the hedge in my backyard. I usually cut about 10% of subs when doing this. Maybe I should be more aggressive ...

[lazjen]

That's similar to what I do now and depending on the conditions of the various night(s) of data I might lose up to a similar amount. It's that nagging question though is whether to be more aggressive in the rejections or let the weight figure handle it.