Weights Optimizer script

Phil Misiowiec

Hey Adam!

Would love to learn more about Weights Optimizer script. It was mentioned by robyx in a PI forum post, specifically:

Don't consider the minimum weight value a reliable way to "discard bad frames" in the sense you're suggesting. It is not equivalent to a rejecting formula that you may use proficiently in Subframe Selector (which can also work on several parameters, not just the overall weight), it is a threshold that saves computational time by rejecting frames that contribute insignificantly to the result if included.

If you want to focus on specific details you still have the Weighting Optimizer script, which is the best option to maximize or minimize an integration measurement, but at a higher cost in terms of computation time."

It looks like the computed values end up in the FITS header, but WBPP does not have a way to reference them (I do see the "FITS keyword" option in the stand-alone ImageIntegration tool though).

This came up in the context of playing around with different minimum weights for image integration in WBPP. I was doing a qualitative analysis of different rejected images vs. various weights to try to get a sense of what a rejected image looked like relative to the "good" ones. Overall very happy with the output of WBPP and its consolidated functionality but wondering if Weights Optimizer is worth exploring further.

Thank you,

Phil

BlockHead

I agree... it is another area to explore.

BUT... I would highly recommend exploring the difference between PSF Signal Weight and PSF Scale SNR.

I have gone into this extensively... but no one believes me. lol

My argument boils down to signal is more important (usually much more) than resolution or background brightness. If you are concerned about optimizing results *between* weighting method... this is a good place to look. I believe this has a LARGER effect than optimizing within a given weighting scheme.

-the Blockhead

Phil Misiowiec

Thanks, Adam. I know the exact videos you are referring to and will re-watch.