Weird columns/bands in stacked photo

[kentnek]

Hi everyone, I'm trying to use PixInsight to process my stack output from DeepSkyStacker. I'm getting this weird reddish column patterns after using AutomaticBackgroundExtractor (and an auto stretch), not to mention a dark circle in the middle:

This was shot with an unmodified Canon R6 with a RF 70-200f4 lens. I took:

• 44 light frames (30s each)
• 21 dark frames (30s)
• 30 flat frames
• 50 Bias

All of these are ISO400, f5.6.

The master calibration frames look normal to me:

If I open a single light frame and do an auto stretch, I think I can see the red bands faintly so it's likely due to the light frame. However, the sharp circle in the stacked photo does not appear here, and I believe any vignette should have been removed with the flats.

Does anyone know what the red bands and the dark circle are, and how to edit them out? Thank you!

 

[Ouroboros]

Have you tried removing them with the canon banding reduction script in PS?  Odd thing is whenever I’ve seen them they run horizontally. 

[kentnek]

7 hours ago, Ouroboros said:

Have you tried removing them with the canon banding reduction script in PS?  Odd thing is whenever I’ve seen them they run horizontally. 

Hi @Ouroboros, thanks for the tip! Running the script AFTER AutomaticBackgroundExtractor does help:

When reading about the script I saw others running it right after integration and before ABE/DBE, so I experimented doing so. While it was able to reduce the original columns, it introduced some new bands:

I guess if I perform a DynamicBackgroundExtractor I can reduce the dark circle to some extent. 

But I'm still curious about what causes these artifacts in the first place.

[Ouroboros]

After quite a few forays into this I eventually resorted to applying the canon banding utility to each sub-frame individually using the batch processing in PI.  I got this advice from this thread on SGL, and in the book Inside PixInsight.  This seemed to give the best, though often not perfect result. The banding can be significantly reduced towards the end of processing by using the Curves process to suppress the background values.

As for cause, banding is said to be introduced during the conversion from analogue to digital and has been variously put down to temperature effects and electronic interference from nearby USB cables.   

[kentnek]

Thanks for the thread! It looks quite informative, so I'll have a more careful read in the morning. It appears I should try stacking in PI itself, as mentioned in that thread and other posts I found.

Regarding the circle artifact, is it due to my optics (unlikely as I even see it after stacking pictures taken with my wide angle 15-35mm lens), or my flats? If I understand correctly, the flats should be able to help remove that circle right?

[ONIKKINEN]

Unrelated to the banding issue but your flats dont seem to be working all that well, given the weird circle effect.

The circle is present in your uncalibrated lights but not the flats, so it does not calibrate out. How did you take the flats?

[kentnek]

@ONIKKINEN I placed an iPad on my lens 🤣 All other settings are the same except for shutter speed, which was set for 50% peak in the histogram.

[ONIKKINEN]

31 minutes ago, kentnek said:

@ONIKKINEN I placed an iPad on my lens 🤣 All other settings are the same except for shutter speed, which was set for 50% peak in the histogram.

Looks very bright, although it is difficult to judge from a picture. Too bright flats can be bad and not capture properly because the exposure time is so short. I suppose there could also be internal reflections between the lens elements and the ipad surface that are different from your lights?

Try putting something in between and/or reducing brightness. A white t-shirt folded in layers and tightly rubberbanded on the lens would work. 

Also light leaks from the area you are shooting from to the lens surface could be an issue, but i doubt it would leave such a perfect circle. The top left bright spot looks more like that and is not that big of a deal.

[kentnek]

11 minutes ago, ONIKKINEN said:

Too bright flats can be bad and not capture properly because the exposure time is so short

Ohh this is interesting. I wasn't aware that too short exposure times would have an effect on the flats.

11 minutes ago, ONIKKINEN said:

I suppose there could also be internal reflections between the lens elements and the ipad surface that are different from your lights?

Hmm I didn't think of that. Perhaps next time I should put the iPad on my lens hood to allow a bit of distance between the iPad's screen and the lens elements, and use a white shirt or cloth to diffuse the light further. Thanks for the input!

Edited November 23, 2021 by kentnek

[ONIKKINEN]

5 minutes ago, kentnek said:

Hmm I didn't think of that. Perhaps next time I should put the iPad on my lens hood to allow a bit of distance between the iPad's screen and the lens elements, and use a white shirt or cloth to diffuse the light further. Thanks for the input!

If the lens hood was on for the lights, it should be for the flats aswell it could have an effect too.

You can actually just retake flats now/anytime later and stack again. Try to have the lens on good infinity focus and the aperture/iso as it was for the lights. Wont be perfect in terms of dust on the lens but otherwise should work. Many folks reuse flats taken months ago so it will probably be fine.

I used to take flats from a computer monitor on the day after shooting and as far as i know they were at least decent.

[newbie alert]

12 hours ago, kentnek said:

Thanks for the thread! It looks quite informative, so I'll have a more careful read in the morning. It appears I should try stacking in PI itself, as mentioned in that thread and other posts I found.

Regarding the circle artifact, is it due to my optics (unlikely as I even see it after stacking pictures taken with my wide angle 15-35mm lens), or my flats? If I understand correctly, the flats should be able to help remove that circle right?

Your flats aren't working as they're over correcting... if you look at the image and the flat, once they've been applied it makes the dark areas light and the light area dark.. so reduce your adu,  or aim for 30%... not really sure where this 50% comes from as they rarely work 

[Ouroboros]

Taking flats with the camera set to aperture priority worked OK for me. 

[kentnek]

After a few days busy with work, I finally got some time to continue editing my photos. I reshot the flats:

And the result is marginally worse than before 🤣

However looking at these more carefully, I notice that the top left quarter of the circle is missing, which is the direction of the moon (yes I shot this on a gibbous moon night). So I guess the moon is causing this shadowy circle thing.

DBE could remove the moon glare but not the circle:

While reading on this topic I came across "synthetic flats". Not sure if making those would solve this, so I'll try it later when I have time.

[sharkmelley]

It is quite common that when shooting in moonlight, stray light causes an obvious central circle which flats will not remove.  However, I've no idea about the cause of the vertical bands.

Mark

[kentnek]

8 hours ago, sharkmelley said:

stray light causes an obvious central circle which flats will not remove.

Yeah, thanks for confirming my suspicion! It's a good idea to avoid nights with the Moon up in the sky then. I was curious and tried to push my equipment (stock DSLR + 70-200 lens) to the extreme, under a light-polluted Bortle 8 sky + moonlight. And unsurprisingly my Pleiades were totally washed out haha

Edited November 29, 2021 by kentnek