Quote:
Originally Posted by The_bluester
Thought I would pipe in here (After checking with Andy first) The decision on RGB sub length more less was made by me by default while I was testing stuff for Andy.
The intent was to get the background comfortably off the noise floor but have few saturated pixels to preserve star colours, trying to fit the dynamic range of the camera best to the absolute range the sky gave. We applied the same logic to the NB data as well.
In the end at the time used, the minimum and median pixel values are each a couple of hundred ADC above a comparable dark frame and around 3000 pixels (Less than 0.2% of the frame) were saturated so basically only the very centers of the brightest stars would saturate and throw colours out of whack. We might have gotten away with slightly shorter subs too and get even those away from saturation but clear sky was too hard to come by to toss existing data so Andy pushed on at the same sub length. Some shorter subs threw calibration warnings in Astro Pixel Processor about excessive numbers of pixels clipping to zero, which I assumed to mean the subs were not long enough to get the background above the noise floor.
I have been applying the same logic to my ASI2600's of taking test subs to work out how long I can go before an arbitrary number of pixels saturate and going with that, to try to get the best star colours out of those cams too.
|
Very interesting, Paul. Thanks for the write-up!
I had thought that all you were doing (I say "all you were doing" like it's a small thing.. I know it's not!) was stripping out the stars via Starnet or StarXterminator and then layering the star image on the bi-colour.
If that's the case, why does the noise floor matter? All you'd want are the stars? They won't be anywhere near the noise floor.
Total noob here, so forgive my ignorance with these things.