Image Stacking Weights?

[dmcclain]

I have a bunch of frames, of several exposure durations, of the same nebular region. I arbitrarily chose a 2-min integration frame as my stacking reference frame. After alignment projection the stacking operation (ImageIntegration) assigns an arbitrary weight of 1.0 to each of R,G,B for the first image in the stack. Thereafter, each additional image is assigned a mixing weight according to some statistical valuation of the image.

I find it interesting that a 3-min exposure typically has a weights of 1.23-1.38 against my 2 minute reference frame with weight 1.0. The remaining 2-min exposures typically have weights around 0.78-1.00.

What statistical attributes of an image would cause a 50% increase in integration time to weigh in only 30% better? And why the same duration exposures weigh in only 80% as much as the first one?

- DM

[dmcclain]

Achh... I just realized that the SNR increase in the 3 min exposures would only be square-root of 1.5 better than the 2-min exposures, or about 1.22. Sorry for that. But I'm still curious about the 80% weights on remaining same duration exposures?

- DM
 

[jkmorse]

A great way to check is to load the images into the SubFrameSelector script.  It will show you very clearly where you might have issues, ranging from star count to focus to SNR as compared to your other subs.  I often find that in a series there will often be some that are just weaker than others and they stand out in the script.  Could have been haze or passing clouds, or again focus that causes a longer iteration to be weighted less.

I then use the best image as measured by the script as my reference image for star alignment and image integration.

Also, be sure to run the images through Blink.  That will always show visually defective subs.

Best,

Jim

[dmcclain]

Thanks for the suggestions. I always do the Blink check, but wasn't aware of SubFrameSelector. I'll give that a try.