Adding Stars Back Whilst Still Linear

M

midnightlightning

Well-known member
I would like to be able to remove stars to run DBE, one of those all nebula and no background scenarios, and then put them back in to be able to run SPCC.

I have regularly removed stars and put them back after stretching with ~((~starless)*(~Stars)) without issue but I cant find a way to put them back whilst everything is still linear.

Any suggestions?
 
If you are still linear then remove the stars by subtraction: stars = (starry - starless). Do what you need to do Then add back in with <new starry> = <modified starless> + stars
 
I would strongly recommend running SPCC first. DBE (and any other processing) will work better once you have the colour roughly right. It is a common mistake to think that "tidying up" the background will "help" SPCC. SPCC does not need this sort of help.
 
midnightlightning said:
I would like to be able to remove stars to run DBE, one of those all nebula and no background scenarios, and then put them back in to be able to run SPCC.

I have regularly removed stars and put them back after stretching with ~((~starless)*(~Stars)) without issue but I cant find a way to put them back whilst everything is still linear.

Any suggestions?
Click to expand...
I agree with Fred. Run SPCC before you do anything else.

What I sometimes do is then remove the stars and use DBE to create a nice background model. But I don't apply it to the starless image and add the stars back. I apply it to the original image that still has the stars. I find that works very well.
 
If (one of these days) PixInsight includes a native star removal tool (as it once did when it included StarNet - though I would want much better quality than StarNet or StarNet2) I may play with star removal. While this is still the domain of 3rd party paid apps and non-standard library configurations, I will (quite happily) live without it.
 
Great responses - thanks.

I thought it was important to do DBE prior to SPCC but doing the opposite certainly does what I need.

@cloudbait Don't you find that you end up with stars in some of your sample points doing it like that?

@fredvanner I understand re having to buy 3rd party apps but STX is a joy to use and processing nebula and stars separately makes a massive difference in my experience, especially now we have GHS.
 
midnightlightning said:
Great responses - thanks.

I thought it was important to do DBE prior to SPCC but doing the opposite certainly does what I need.

@cloudbait Don't you find that you end up with stars in some of your sample points doing it like that?

@fredvanner I understand re having to buy 3rd party apps but STX is a joy to use and processing nebula and stars separately makes a massive difference in my experience, especially now we have GHS.
Click to expand...
I don't run the DBE against the image with stars, I use PixelMath to subtract the clean background that DBE generated on the starless image.
 
fredvanner said:
If (one of these days) PixInsight includes a native star removal tool (as it once did when it included StarNet - though I would want much better quality than StarNet or StarNet2) I may play with star removal. While this is still the domain of 3rd party paid apps and non-standard library configurations, I will (quite happily) live without it.
Click to expand...
Yes, star extraction is not usually an essential operation. I do make use of it in a few places where alternatives are few. One is comet processing. Another is placing RGB stars in a narrowband image.
 
cloudbait said:
I don't run the DBE against the image with stars, I use PixelMath to subtract the clean background that DBE generated on the starless image.
Click to expand...

I didnt get this at first but just to play back to be sure :
  • Run DBE on the starless image - generate and keep the DBE_Background.
  • Pixelmath : Single expression "RGB (including stars) - DBE_Background"
What a great solution :)
 
only problem is by default DBE creates a background model that is 1/2 the size of the image, so you have to go into the DBE interface and change the downsample to 1 (in the Model Image section)
 
midnightlightning said:
I didnt get this at first but just to play back to be sure :
  • Run DBE on the starless image - generate and keep the DBE_Background.
  • Pixelmath : Single expression "RGB (including stars) - DBE_Background"
What a great solution :)
Click to expand...
Yeah, that's what I do. And as noted by pfile, it's best to let DBE generate a full size background image (although you could resize the scaled down one and it would probably work as well).

I don't use this approach all the time. I find it useful for images with a significant area of background and centered, isolated targets. Like a galaxy in the middle and a lot of space around it. Then I can let DBE autogenerate a dense grid and just manually delete the points on top of the target (to the extent that the tolerance engine allows them to be created there). It's an easy way to get a lot of sample points without much work.
 
cloudbait said:
Yeah, that's what I do. And as noted by pfile, it's best to let DBE generate a full size background image (although you could resize the scaled down one and it would probably work as well).

I don't use this approach all the time. I find it useful for images with a significant area of background and centered, isolated targets. Like a galaxy in the middle and a lot of space around it. Then I can let DBE autogenerate a dense grid and just manually delete the points on top of the target (to the extent that the tolerance engine allows them to be created there). It's an easy way to get a lot of sample points without much work.
Click to expand...
and @pfile

The Downsample is new to me. Just playing around with it and notice there is a setting of "Use Model Widths;Heights". Is this the same as "1" if you use default width/heights?

I played around with it and couldn't see any difference between "2" and "use model...".

This link is interesting, it suggests there shouldn't be much if any difference. https://pixinsight.com/forum/index.php?threads/automatic-resampling-in-dbe.651/
However, I don't like any compromise when processing hard won data so will definitely use full size going forwards - its a quick process to run anyway.

Thanks for the help, great to learn some new approaches.
 
You must log in or register to reply here.
Back
Top