Astro Image Stacking (optimize DSS image stacking and post processing)

[Drakkith]

What do you mean stacking invalidated the flats?

 

[davenn]

Update:...
- stacking the 16-bit images invalidated the flat frames- but I'm hoping to show some dramatic improvement within a few days.. ...Progress?

What do you mean stacking invalidated the flats?

yup, didn't make sense to me either
The only thing I could think of that would do that is if your flats were taken differently than your lights
... different exposure time, focal length, ISO setting etc

Flats all need to be done in the same way as your lights, otherwise DSS will reject them

 

[Drakkith]

Hmmm. That's odd if it rejects them if they are different exposure times. Flats are almost never the same exposure time as your lights.

 

[Andy Resnick]

Now I have some data supporting my hypothesis that DSS wasn't interpreting the RAW data files correctly (thanks to Tom.G!). Here are two linescans comparing the 'old' interpolation method ('Bilinear Interpolation') and the new, *correct* interpolation method ('Adaptive Homogeneity-Directed (AHD) Interpolation'). No flats were used during stacking- these images were acquired at f/4.

The stairstep effect is completely gone now (the background is a lot flatter, as well). I used the original RAW files, not 16-bit TIF images. The DSS 32-bit histogram looks identical- this problem was really subtle...

Ok- regarding my comment about 'stacking invalidated the flats'. Last year (or so), when I got more serious about flat field correction, DSS generated a whole range of Master Flat 16-bit TIF files and then I deleted all the original RAW files- I figured, why keep 100GB of files when I don't need them anymore? What I didn't know is that the Bilinear TIF interpolation from DSS and TIF interpolation algorithm from Nikon's Capture NX program created differently-sized pixel arrays. Consequently, DSS simply ignored the flat file when stacking 16-bit TIFs.

Worth noting, once you have 16-bit Master Flats, you can manipulate them to improve performance- that's how I generated multiple master flats to eek out that final 0.01% difference: for example, I generated a series of mixed Master f/2.8 and a Master f/3.5 in different proportions (75/25, 50/50, 25/75). The RAW flats do not have to have the same image parameters as the bright images- however, all the RAW flats have to have the *same* image parameters (I can't mix and match different ISOs, for example).

So, I'm happy to report that the major problem has been solved (thanks, everyone!). Since I'm teaching QM this semester, I'm calling this 'first quantization correction'. Last night, I took some new RAW flat images and today will generate new Master Flats with the AHD interpolation method ('second quantization') and see how that impacts my low f-number images- presumably I won't have to contort the flats so much anymore. I'll post results in the usual thread rather than this one.

 

[Tom.G]

I'll post results in the usual thread rather than this one.

Usual? What usual? Enlighten us. I sure don't want to miss those new results!

Cheers,
Tom

 

[Andy Resnick]

Usual? What usual? Enlighten us. I sure don't want to miss those new results!

Cheers,
Tom

I meant the 'amazing universe' thread that everyone posts their pics onto. :)

 

[Drakkith]

What does that have to do with flats?

Not really true. PixInsight's "automaticbackgroundextraction" tool does wonders for light pollution.

 

[Andy Resnick]

bias frames:

https://dslr-astrophotography.com/shoot-bias-frames-dslr/

These will help reduce your camera's operational noise

dark frames:

http://www.rawastrodata.com/pages/typesofimages.html

Both are important to creating a better image.

Those are claims. What is your evidence that my stacked images would be significantly improved by either of those reference images?

 

[Drakkith]

Go grab a trial copy of PixInsight and use the tool in question. i think that'll best answer your question.

I'd prefer if you explained why you mentioned a background correction tool in an image processing program, but you quoted the part of my post about flat field images. The two don't appear to be related at all other than being a part of image processing. I'm not going to download the program and potentially spend a few hours guessing about what you meant when you could simply tell me.

Those are claims. What is your evidence that my stacked images would be significantly improved by either of those reference images?

I don't know about significantly improving your images, but I always shoot dark and bias frames when imaging. They absolutely help my images.

 

[davenn]

A very noisy image, looks like it's been stretched too far imho.

hardly touched, and definitely no stretching
you are obviously referring to the wrong image

Even worse than the above image imho.

yeah, that was my whole point ... that you seemed to miss
His 100+ frames doesn't even compare to an image of much lesser frames

show us your work using the same setup and settings !
put your money where your mouth is

 

[Andy Resnick]

I don't know about significantly improving your images, but I always shoot dark and bias frames when imaging. They absolutely help my images.

I tried incorporating dark and bias frames once, a while ago, and felt that the effort-to-result ratio was too high.

 

[Drakkith]

I tried incorporating dark and bias frames once, a while ago, and felt that the effort-to-result ratio was too high.

I could see there being little noticeable effect if you're shooting short exposures in heavily light polluted skies. The effect from the skyglow is probably orders of magnitude more than dark current and bias. But they are something to be aware of if you start moving to longer exposures, darker skies, or if you want to do 'sciency' stuff that requires precision measurements.

 

[jim mcnamara]

This is a great thread. To make an analogy it is like how to develop a recipe for lasagna. It has examples and experiences and lots of informative content. This means opinions, supported with examples, are part of the process. So if there are problems with posts, please report them.

 

[DennisN]

Thanks for this thread, it is interesting, and I will read it through when I have more time.