Processing issues with blotchy areas in background

[tcln1456]

I've probably processed this picture at least 10 times, and each time I get these splotches of brown/blue/red around the stars (you can zoom in on the background near the bottom and top right)

http://i.imgur.com/LB2euq9.jpg

I can't seem to figure out where it's coming from. For example, when I do noise reduction with ACDNR (first round at 2std in lum, 3std in chrom and second at 5/6 in lum/chrom, basically whatever Harry does), they become more visible. I don't think it's from the ACDNR though because when I undo both and redo, it only seems to be removing noise, nothing else.

As an alternative to noise reduction, I've tried ATWT with 4 layers when the image was linear (along with an inverted luminance mask to protect the Milky Way), and I'm still getting some RGB splotches http://i.imgur.com/CEb3jtg.png.

I tried a third method where I did ATWT in the linear stage (S(3,1,2) S(2,1,2) S(1,1,2) S(.5,1,1) for 5 layers, 5th untouched, dyadic), stretched the image, HDR Wavelets applied, and did a second ATWT (S(3.1,1,1), S(.5,1,1) for 4 layers, 3rd and 4th untouched) and got the same splotches http://i.imgur.com/pgjFRwF.png

Any help as to what may be the problem?

If you'd like to work with my master light, here's the file: https://www.dropbox.com/s/3syl37p9r3492pp/light-BINNING_1.xisf?dl=0

Edit: I forgot to add the image details

12x13s lights @ ISO800
I didn't have a chance to take darks or bias (I was on a beach during vacation and didn't have a mount either and I was in a rush) so I used an older master dark and bias for integration. The darks were 10s exposures (~50 subs), bias were 1/8000s (~50 subs), and both were ISO3200

These were all loaded into batch preprocessing and run at default settings.

[chrisvdberge]

Please let us know what other processes you have used.

In any case I personally never would use noise reduction if this is the resulting background. Better to have some noise that is 'evenly spread' than these blobs/patches

[Carlos Milovic]

This is just medium scale noise. It is just natural random photon noise, that was amplified after your noise reduction (now has greater relative contrast). The solution really is to have more frames and longer exposure times, to increase SNR. Also, you should decrease contrast and brightness of the background sky.

[jkmorse]

Check out this post: http://pixinsight.com/forum/index.php?topic=8618.msg55913#msg55913

In it Mike Smythers has some great scripts and one includes a background smoothing routine that is fantastic.  It uses the Morphological Transformation tool to great effect.  I like it so much I have it saved in my Process Container and use it regularly.  It helped me solve problems on a number of images where I was having the same blotchy background problems. 

Best,

Jim

[tcln1456]

Please let us know what other processes you have used.

I started out with a dynamic crop to cut out misaligned frames and did a DBE http://i.imgur.com/p9NIGan.png. I made a preview near the bottom right of the image for background neutralization and then created an preview aggregate of the stars in the image, excluding the Milky Way for color calibration. I checked the post-background neutralization background values and set the lower limit for the white reference to be slightly above those values. upper limit for background was 0.01.

I stretched the image with the STF process icon applied to histogram transformation and did HDRWavlets with 4 layers, 1 iteration, and deringing set to 0.03 small and 0.00 large scale.

Then I did ACDNR according to Harry's video (details in my parent post)

I stretched it a little again and did color saturation with the curves transformation tool using a luminance mask, and then used SCNR to remove the greens.

[jkmorse]

Also, in the linear phase, to the extent you are going to try noise reduction, you really should try Morphological Linear Transform.  It gives superior results in the linear stage to the others.  You can also try running a convolution kernel to smooth things out as well.

[tcln1456]

This is just medium scale noise. It is just natural random photon noise, that was amplified after your noise reduction (now has greater relative contrast). The solution really is to have more frames and longer exposure times, to increase SNR. Also, you should decrease contrast and brightness of the background sky.

I was afraid this was the issue. Granted, I only had time to take 12 exposures before my dad made my get off the beach because it was already dark and empty. Maybe next time

[tcln1456]

Also, in the linear phase, to the extent you are going to try noise reduction, you really should try Morphological Linear Transform.  It gives superior results in the linear stage to the others.

I will definitely give it a try. Thank you!

[tcln1456]

Check out this post: http://pixinsight.com/forum/index.php?topic=8618.msg55913#msg55913

In it Mike Smythers has some great scripts and one includes a background smoothing routine that is fantastic.  It uses the Morphological Transformation tool to great effect.  I like it so much I have it saved in my Process Container and use it regularly.  It helped me solve problems on a number of images where I was having the same blotchy background problems. 

Best,

Jim

I will check this out, thank you!

[msmythers]

Hi

Let me add my 2 cents on the subject. I shoot a lot of short exposure images because of poor local environment and poor equipment. What Carlos said is the exact thing to do. More total exposure time and longer subs. If longer subs are not possible then try to collect as many subs as possible.

That being said you can help yourself  by not shooting so low to the horizon. You are looking through thick, turbulent air. Especially next to a body of water. If you are going to take more of these short exposures from a tripod try shooting areas that are more over head. That will help a bit, thinner atmosphere.

You can only process to what the data in your image supports. If you find that the noise is to much or doesn't look good it's ok to have less contrast or less brightness. Short exposure shots will have a lot of apparent chromatic noise so again it's ok to not have as much color saturation. It's even ok to only process for luminance only without color. One other thing is you can reduce the apparent resolution. Not all images need to be seen as a high resolution image. Poor, noisy images will always look better at lower resolutions.

As you learn to process and get better you can always reprocess an image. I do all the time. Shots that I took when I first started I am reprocessing now because of what I have learned. It's the great thing about digital processing.

Here is a shot I took in January. It's what I call my Ugly shot. It was on a tripod using 10 second subs. I took 101 subs for a total exposure time of 16 minutes. I was able to capture what I wanted to capture. It's not a good image but by doing some of the thing mentioned I have a shot that is 'ok'. I think it is processed to what the data supports.
http://www.astrobin.com/149220/


Mike