Struggling with background extraction

[Ivor]

I am working on an image of the cocoon and I'm strugggling with separating the background from the black nebula around the cocoon nebula. It is a rich starfield so the placement of the DBE sample points is a challenge, the image below shows the placement I went with. I avoided the black nebula and focused on the areas I want to remove

I've used division and substraction on the subsequent outputs and this is the best I've managed to acheive (starnet2 used highlight the challenge).

 

 

I've attached the integrated image for other to play with and I'd be grateful for suggestions on how to solve the problem.

 

C19_V2_INT.xisf

[ONIKKINEN]

You have what looks like flat issues in the corners, best shown here in this Siril false colour rendering mode:

Could also be some other issue, like light leaks but this is what shifted flats would look like.

If you crop a little bit the gradient is quite easy to work with. I placed some samplers on the dusty parts, but im not sure you have to. I think its a decent sample for background though.

The resulting image is not too hard to work with to get the dust out:

[Fegato]

Just a tip on using DBE in PI...    I use StarXTerminator, and if you remove the stars prior to DBE (with unscreened option not ticked) it becomes so much easier. You can clearly see what's gradient and what's dust with no stars, particularly in a rich star field like this. Perform DBE with the normalize option ticked, and then add the stars back (simple $T+stars). I now do this on every image, and it's improved my results no end. I've realised some previous images had swathes of dust and nebulosity destroyed by poor sample positions in DBE.

[geeklee]

9 minutes ago, Fegato said:

Just a tip

Thanks for posting this Robin, I'll definitely try this approach next time.

[Ivor]

Thanks to everyone for the tips I'll work through this and see what I can pull out.

 

[ollypenrice]

I would advise against large numbers of sample points. What you are trying to remove is a broad gradient and the more points you put in, the more you are likely to pick up local brightness variations - or create them. I rarely use more than 8 and, sometimes, half that.

Olly

[tooth_dr]

Is DBE in PI the same principal as Light Pollution Removal tool in Astro Pixel Processor?  I’ve not moved that step over to PI yet but with a sense star field it’s hit and miss. 

[david_taurus83]

11 hours ago, ollypenrice said:

I would advise against large numbers of sample points. What you are trying to remove is a broad gradient and the more points you put in, the more you are likely to pick up local brightness variations - or create them. I rarely use more than 8 and, sometimes, half that.

Olly

I'm the same. I find less points works better most times unless I have a difficult gradient or residual flats issue where I need to place more specific points. If struggling with DBE you could also try the free software GraXpert. It has gotten me out of a jam a few times when DBE proves difficult.

[david_taurus83]

9 hours ago, tooth_dr said:

Is DBE in PI the same principal as Light Pollution Removal tool in Astro Pixel Processor?  I’ve not moved that step over to PI yet but with a sense star field it’s hit and miss. 

Yes it's usually pretty good (once you get the hang of PI) For narrowband I usually place 4 squares in the corners and that does the job by enhancing contrast slightly.

[ollypenrice]

10 hours ago, david_taurus83 said:

I'm the same. I find less points works better most times unless I have a difficult gradient or residual flats issue where I need to place more specific points. If struggling with DBE you could also try the free software GraXpert. It has gotten me out of a jam a few times when DBE proves difficult.

Residual flats issues would, in theory, require you to apply the DBE gradient map via division, whereas gradients are removed by subtraction. I've only ever used subtraction since I take flats and find they work with current setups.

Olly