Bright/Dark ring in background after DBE

[Matariki]

Hello together,

I just tried to process 2 datasets and they both suffer from the same problem, a huge bright ring in the background after I use DBE. The data has not been calibrated so I try to flatten the image with a DBE (division).

You can see the stack with its DBE samples here, I have checked every sample and they seem to be fine (not stars etc.)

https://i.imgur.com/QulBroB.png

and this is the resulting image after I execute the DBE, you can see this ring that spans the entire background of the image.

https://i.imgur.com/IhwajOj.png

I am not sure what I'm doing wrong as I already selected and checked as many samples as I can and played with some of the parameters. I guess its a artefact from overcorrecting the image with DBE but I am not sure at all. Has anyone seen this?

Here is a link to another album on some other data that has the same problems:

https://images.discordapp.net/.eJwFwVsOhCAMAMC7cADKY6XU2xAkaFRKoMaPzd59Z77qGZda1S7S5wqwHTPz2PQUHqkWXZnrVVI_ps58QxJJeb9Lkwnug95HS0jOBvKBEBx64ywaE9ziiaKJ8LSz8dt0b1X9_gloIuw.yyJCM9RxvaHUC310f4Z-uKZZ_EI?width=893&height=595

In this example you can spot the ring on the lower left.

Would appreciate any pointer or help!

[Carlos Milovic]

Since this image has a lot of background sky, I should try ABE instead, since it gives a more rigid background model.

[Matariki]

Since this image has a lot of background sky, I should try ABE instead, since it gives a more rigid background model.

Thanks for the response, but that doesn't help, see: https://i.imgur.com/XDeCkey.png

I have uploaded both stacks so if anyone could have a look at it I'd be very thankful. I already heavily crop the images to reduce the vignetting but it still doesn't work out.

Link to stacks: https://1drv.ms/f/s!AqyAoT9oZJvFgddGV59RnKJkUV7wYA

[Carlos Milovic]

Change the degree of the polynomials to better fit the background. Use even numbers.
If still doesn't work, decrease the number of samples in DBE and increase the softness parameter.

Another idea is to iterate a bit in DBE.

[aworonow]

What camera are you using? what is the per-pixel resolution (in arc seconds)? What is your drizzle factor? Maybe drizzle is just not appropriate for these images and there is some kind of ringing or beat frequency.

Alex

[Matariki]

Change the degree of the polynomials to better fit the background. Use even numbers.
If still doesn't work, decrease the number of samples in DBE and increase the softness parameter.

Another idea is to iterate a bit in DBE.

I will try changing the sample amount and parameters today. How would a workflow look to iterate through DBE, just do multiple runs?

What camera are you using? what is the per-pixel resolution (in arc seconds)? What is your drizzle factor? Maybe drizzle is just not appropriate for these images and there is some kind of ringing or beat frequency.

Alex

It is a Canon 6D with 1.88"/px scale (720mm FL). I will try to integrate the images without drizzle and see how this turns out.

[aworonow]

I wouldn't think pixel resolutions that small would be very amenable to dithering. Would best be up around 3x seeing, maybe more. Also, with a DSLR, there may (?) be other factors, since adjacent pixels do not see the same image (due to color), and dithering is a process of finding more resolution from what adjacent have in common and as differences. OK, it's a bit more complex than that, but I'm still leery of using an image captured through a Bayer pattern in dithering. There are references that tell various ways to do it, but I'm not sure what facilities, if any, PI implements. Maybe someone else could comment?

Finally, from the background color and galaxy intensities, I'd guess you've pushed the images you posted pretty hard. Both using a Bayered sensor and dithering require more images than a mono camera would for a single color filter. There just may not be enough data to get where you wanted to go.

Let us know what you get in the non-dithered experiment, please. Good Luck!

Alex