Random strip of color from polynomial background extraction on sequence

Hi, when extracting a background gradient using a polynomial of degree 1 on a sequence, some of my output images (<5%) have random strips of color on them.

I calculated the difference between the original and output images, and the gradient subtracted also contains the artifact. This only happens when processing in a sequence – polynomial background extraction on the same images individually works fine.

Am I using the background extraction tool wrong, or is there any way to fix this? I’ve attached some screenshots below (all auto-stretched) for reference.

Before background extraction:


After background extraction:

Difference:

Hello and welcome! That looks like a bug, thanks for reporting it.
Please tell us what version of siril you are using and on what operating system.

I am on macOS 14.2.1 using Siril 1.2.5

Could you provide the original image?

I got your images, the calibrated images already show these bands, the blue channel was improperly generated apparently. Look for example at images 17, 37 or 46. The problem is not with background extraction but with either calibration or demosaicing.

Could you share the raw image associated with light_00037 and maybe the master dark and flat too?
Thanks

Maybe you could delete all the processed images (pp_light_*) including the .seq files and restart to see if it happens again on the same images first?

Sorry for the late response - was away for the New Year.

@lock042 Here is the calibrated image and the image with the extracted background – the artifacts only appear when processing the image in a long sequence (it happened with 600 images, but not with 200 images, and the affected images are random).

https://drive.google.com/drive/folders/1X1iPZi98GzEuhAgzeG0gS1U5uJNdhd6s?usp=sharing

@vinvin I don’t believe that is the case, as the images seem to be affected at random. The calibrated and processed file might be useful for you too.

Could you share the raw image associated with light_00037 and maybe the master dark and flat too?

I think you may have been looking at a different set of images before? The images above aren’t mine, and for me image 37 doesn’t have any issues.

yes apparently someone else has the same problem, sorry

I have tried with 400 images (400 times the image you shared) and could not see a problem. This is possibly only happening with mac…
Strangely for the other person seeing this problem, it was visible before background removal, during calibration and demosaicing. We will continue investigating.