Vertical banding issue

[b0balagurak]

Im a beginner and this is my first image of the north American nebula and have some issues with banding. I'm shooting with a Fuji xt1 and am getting some magenta vertical bands. Only visible after running starnett on a linear image and then stretching. I also have some weird pattern when zooming in. I'm using siril for processing and when I open my flats in histogram mode, there is a pattern when I zoom in but I did my flats correctly. This is about 8 hours of data from last night. Any tips and tricks would be appreciated. The bright one is doing starnett last while still linear and the one with less banding is running starnett first and small stretching.

[Elp]

Is it a square pattern, a bit difficult to see in the images. I do sometimes see this with my astro cams but once stacked it goes away. I suspect my issue might be when I'm imaging near LED or the moon light sources.

Have you tried Sirils banding reduction, you can apply vertically as well as horizontally.

 

[b0balagurak]

Yes it's like a square pattern but its only in the flats calibration is what it looks like. Very artificial. I do use a led panel for flats and make sure it's properly exposed. I should try morning sky with tshirt on lens to see if that helps. I tried the banding reduction but no dice, kind of makes it worse. Also images don't look particularly sharp. This lens doesn't focus past infinity so i can't go back and forth but stars are pinpoint.

[Elp]

I used to get a lot of issues with flats using my led tracer panel, I have since dimmed it to its lowest setting and dimmed the light further by using grey and black perspex (you can also use ND film), so my flats have to be at least a few seconds to get a right exposure (I'm using an Asiair so it does it automatically). You could try a number of layers of t shirt over the panel to achieve something similar.

Sometimes flats issues can also be due to the bias signal not being removed from the flat properly prior to calibration, I'm assuming you also use dark flats in your process?

Edited September 21, 2023 by Elp

[b0balagurak]

Insteresting, I'll give it a shot with more layers and I got a new light panel that has output control. I did use a a fairly fast shutter speed too, shooting at 1600iso. I don't use dark flats as I don't think siril can process them at the moment. I use the stock script to run everything. It does have master files but only for darks, flats, and biases. The flats are processed because the master file is renamed to preprocessed flats.

[b0balagurak]

Here's a zoomed in view of my flats file

[alacant]

9 hours ago, b0balagurak said:

Fuji xt1 and am getting some magenta vertical bands

Hi

It's the Fuji bayer. Siril has an option for fixing it:

[b0balagurak]

10 hours ago, alacant said:

Hi

It's the Fuji bayer. Siril has an option for fixing it:

Does the mean that I won't be able to use the scripts? If I change the setting, then would I be able to run with scripts with those settings saved and also point to master calibration files? No big deal if I have to do each step separately

[alacant]

You've just found one of the limitations of using scripts.

Whilst you could alter the script, we find it just as quick to use Siril manually anyway; you're in control, rather than the script.

Cheers 

Edited September 22, 2023 by alacant

[b0balagurak]

3 hours ago, alacant said:

You've just found one of the limitations of using scripts.

Whilst you could alter the script, we find it just as quick to use Siril manually anyway; you're in control, rather than the script.

Cheers 

Ah ok! Ill give it a shot manually! Thanks!

[b0balagurak]

So I tried it manually and my files look better, the debayer got rid of the weird line pattern but when stack is complete, my image is like black with a few stars and when I hit auto stretch, nothing. Histogram mode turns it all black. For calibration I do median stacking and for the lights I choose windsor sigma clipping and no weight setting for testing. No idea why it's happening. I think I had this issue a year back and why I chose scripts because they somehow worked.

[Elp]

The main window updates every time you do an operation, are you to say throughout the whole process the histogram stretch preview was showing as black? Go through it step by step.

[b0balagurak]

Yeah it was black after stacking. I think I solved it late last night. Followed a great tutorial and I didn't preprocess the flats and I think I had a wrong setting on the stacking. Stacking was successful manually! Just now gotta figure out the tiny line pattern which I think is from flats and the walking noise. Time for the guide scope and phd2!

[alacant]

1 hour ago, b0balagurak said:

Time for the guide scope and phd2

... and watch your imaging tahe a leap forward.

1 hour ago, b0balagurak said:

walking noise

Go for a big -at least 10 pixels- dither😉

1 hour ago, b0balagurak said:

think I solved it

Well done. If not, post the siril log if you like.

Cheers

Edited September 23, 2023 by alacant

[b0balagurak]

ok so looks like it still has issues. I stack the calibration files median staking with no normalization then with the flats, I preprocess the sequence with bias. Stack those processed flats with median but this time with multiplicative normalization. Lights are then calibrated with the bias, darks, and flats option ticked and the xtrans correction, equalize cfa and the optimization boxes ticked. Once done, registered with global star and then stacked with average with rejection and additive with scaling normalization. The pixel rejection is linear fit with no weight but image rejection is fwhm. Here are some screenshots of the issues. Also where would I get the log

[b0balagurak]

first one shows the bands that im having issues with, second is asinh with remove green noise and background extration after stack, last one is autostretch after stack

[b0balagurak]

looks like output normalization and rgb equalization did the trick. So I guess the only question is the horizontal lines in the first image when I zoom in. Also manual stacking messes with background extraction to where it wont detect nebulosity so I have to remove a bunch of squares.

[alacant]

12 hours ago, b0balagurak said:

Lights are then calibrated with the bias, darks, and flats option

mmm... dslr, so don't use dark frames of any type. Subtract the offset only, but not from an in-camera bias. No need for median stacking anywhere.

For questions about the Fuji problem, best to ask the developers:
https://discuss.pixls.us/c/software/siril/34

Location of log files. On Linux we simp'ly redirect the console output to a file. No idea how other OSs do it. Sorry.
https://discuss.pixls.us/t/saving-console-output-to-a-log/29711

10 hours ago, b0balagurak said:

manual stacking messes with background extraction to where it wont detect nebulosity so I have to remove a bunch of squares

Adjust the sensitivity using the smoothing, samples per line, grid tolerance and interpolation method. Or, you guessed it, add the points manually yourself; much quicker. You only need a handful over the background areas anyway.

Cheers and HTH
 

Edited September 24, 2023 by alacant

[b0balagurak]

All worked out, thanks!