Is there a way to harvest backgrounds from bad subs for later blending? - Forums

Topic starter

Mar 13, 2021

I have a noob question about background processing. First the setup.

I take a lot of subs, and usually end up throwing out about 1/2. I've been wondering whether I'm being stupid about throwing away those particular frames; not for the signal, but for the backgrounds. Is there no way to process those frames such that they are aligned, stacked, and then intensity subtracted (vs the "good" stack) such that they can be blended later to make for a cleaner background? Denoise is good, but if all those frames could be put to some beneficial use, I wouldn't mind a bit more processing to get value out of them. Maybe this wouldn't work on nebula, but what about galaxy fields?

Ideas?

Well Written Engaging

Rob Calfee

Mar 13, 2021 · Edited: Mar 13, 2021

I'm wondering if this would help: https://www.youtube.com/watch?v=HfeFsyG2NoU

Being able to take out the stars and mess with the background with clonestamp, maybe? By the way, I used this to fix an issue where the camera wasn't tight enough and went off-center in comparison to the flats.

Kevin Morefield

IOTD Staff

Mar 13, 2021

Doug,

I do this frequently when I have nights with different seeing conditions. Since the bright parts of the image don’t need as much integration time, I can do one integration of only sharp subs and use that for the highlights and do a second integration with all subs for the shadows. Once I have the “sharp” and “all subs” masters I combine them using a mask. The mask is a luminance mask made by copying the all subs master to be the mask. That all subs data will have slightly larger stars so I reverse it to mask out the bright areas in the all subs version.

The mask usually benefits from increasing the contrast to allow more of the all subs data to come through and more of the sharp data to hit the highlights. I prefer to do this in Photoshop with non-linear data so I can see the result in real time while I adjust the mask. Color data can come 100% from all subs usually. The key is that the two masters are normalized so their histograms match well.

Kevin

Helpful Respectful

Doug Summers

Topic starter

Mar 13, 2021

Thanks Kevin and Rob. I'll look into these approaches. Cheers, Doug

Well Written Respectful

Björn Arnold

Mar 14, 2021

Hi @Doug Summers ,

I'm curious, why exactly are you throwing half of your subs away?

CS!
Björn

Doug Summers

Topic starter

Mar 14, 2021

Either FWHM or eccentricity typically will not meet what I want for the stack. So, I discard the sub. I realize I could use weights, but I'm more comfortable just doing a pre-sort. Anyway, I was thinking that if I could save the backgrounds for these subs, it might result in a better end result (after I get the amount of signal I want of the quality needed).

Björn Arnold

Mar 14, 2021

I checked your profile quickly. You're shooting mostly with a RASA, or did I accidentally just pick the RASA images?

Are you guiding? If eccentricity is an issue then probably you should invest in guiding or improve it, if you're already into it. It would allow you to keep all images, you wouldn't have to sort them out manually as with sigma-clipping you'd get rid of singular events like satellite trails. I admit that I usually go through my subs but rather quickly just to see if there is any that is extremely bad. The rest should be taken care of through the stacking software.

Helpful Insightful Respectful Engaging

Doug Summers

Topic starter

Mar 14, 2021

yes, I'm shooting with a RASA-11, and guiding too. The mount I have has a 75lb payload capacity, and my RASA-11 is 53lbs more or less. So, I'm a bit over what I would like, but I don't want to spend 10K or more for a good mount. So, it is what it is. The CGX-L mount is ok, but it does have definite limits….

Björn Arnold

Mar 14, 2021

I don't know if I did check the right RASA but from what I can see on the Celestron web site, the RASA 1100 should be at 31lbs. With the CGX-L you should be at approx. 50% of max. rated load (added some guiding equipment). Personally, I'd say the mount should handle that with ease and there shouldn't be any need for a 10K mount. Also the focal length of 620mm isn't too aggressive, IMHO.

That makes me wondering about the guiding performance. What's your guiding setup and what software are you using?

I don't want to convince you of something, so please let me know if you're interested to elaborate a bit more on that or not, but I'd say that loosing half of your exposure time is very significant and if you could counter it with some more or less easy measure, it would be a good idea.

Helpful Respectful Engaging

Björn Hoffmann

Mar 14, 2021 · Edited: Mar 14, 2021

Quality weighting and advanced outlier rejection methods are your friend! I never throw away bad subs, even if a plane goes through. Todays stacking algorithms get rid of all things that are not visible in other frames, which includes bloated stars (at least when there is only a small fraction of those bad frames). It works on pixel level, so all other pixels without disturbance are still integrated. Quality weighting like APP does gives you even more control. You can weigh the frames for general quality or star size / shape. In that way all the background data still flows into your stack. By fine tuning the rejection parameters you can decide which types of outliers are still valid and what not.

Helpful Insightful

Doug Summers

Topic starter

Mar 14, 2021

Hello Bjorn,

The RASA11 is 43 lbs. Adding a 60mm finder, Image cam, guide cam, EAF, USB hub, Pi4, , dovetail mounts and associated wiring adds the additional 10 lbs.   About the guide performance, the CGX-L is ok for it's price range, and it's got a decent payload, but it has it's limits with that payload.   The mount has a 21 second sharp period spike associated with a gear/belt tooth exchange that can't be PEC'd out (only 1/2 second).   Worse, the mount has favored directions for performing with the load, and south isn't the strongest direction (but it's the favored direction for targets).   So, while it's not death by a thousand cuts, each little error budget term takes a toll.

About 620mm, it's true the RASA is designed for wide field, but I use it as a narrow field (1 degree) via use of a 2.4um ASI183 OSC.    That gives a platescale of 0.799 arcsecs/sec which opens up the deep sky nicely for galaxy work which I like.    Trading some field size for image quality (the spot diagram is nicely binned by a UV-IR filter and by off-axis chip size) was intentional.    The guide SW is PHD2.

So, maybe I was a bit harsh in stating that I lose 1/2 all the time….in reality I lose up to 1/2 in the worst conditions, and I like to only keep FWHM < 3 (ish).    On good nights I can keep pretty much all subs.     In any case, we've digressed a bit from the question (but I'm glad to answer).    I just was wondering if I could try glean backgrounds from unused subs.   I like the ideas others suggested and will play with them a bit to see how it bears out.    Cheers,    Doug

Helpful Insightful Respectful Engaging

Doug Summers

Topic starter

Mar 14, 2021

I use Pixinsight, but I confess I'm not using weighted processing. That's another area to be explored more fully. I know many people have nice Subframe Selector weighing formulas that I should probably be using, and that might allow the backgrounds to be used more fully even for the bad subs. I still have a lot to learn…..

Well Written

Related discussions

SPCC Thoughts and Questions

After reading a related post on some problems using SPCC, it gave rise to some questions I have sort of assembled upon my recent adoption of this feature in PI. Please feel free to dump on me over these questions and challenges below, but I am not an...

Mar 10, 2023

Both posts involve experienced astrophotographers seeking advice on advanced image processing techniques to improve their results.