PixInsight help needed: adding stars back in changes colour

[Lee_P]

I'm hoping someone can help with this issue. I'm probably doing something really dumb, but I can't quite work out what.

So, I've got this starless image:

When I use PixelMath to add the stars back in, the resulting image has different colours:

I've used this technique a lot, and it's never done this. Any idea what I'm doing wrong?

 

[scotty38]

What maths are you using and which images?

[Lee_P]

13 minutes ago, scotty38 said:

What maths are you using and which images?

PixelMath: "v1_NeedsStars+star_mask1"

 

Edited July 17, 2022 by Lee_P

[Lee_P]

The two images are attached, in case anyone has time to diagnose where I'm going wrong 😁

v1_NeedsStars.xisf

star_mask1.xisf

[Lee_P]

Never mind, I've fixed this now. I basically started again! I think the problem was with editing the starless version in Photoshop, and then I must have saved it with the incorrect settings or somesuch.

[scotty38]

ah ok, was just downloading to have a play 🙂

[wimvb]

I don’t use starmasks generated by starnet. Instead I first copy the image w stars (copy1), make the original starless and copy that again (copy2). I process the original, and add the stars back in with

$T + copy1 - copy2

This tends to give me slightly better stars in the recombined image than if I use the starmask from starnet.

[Lee_P]

15 minutes ago, wimvb said:

I don’t use starmasks generated by starnet. Instead I first copy the image w stars (copy1), make the original starless and copy that again (copy2). I process the original, and add the stars back in with

$T + copy1 - copy2

This tends to give me slightly better stars in the recombined image than if I use the starmask from starnet.

You've blown my mind, I just tried this and it works. I've finished my Crescent image already, but may use this technique on future images. Two quick questions, if I may:

* What does $T mean? It's selecting the starless image you've been editing?
* How do you go about editing just your starfield? For example, removing halos; adjusting saturation of stars; reducing the number of very faint stars... all the things I currently do to the starmask before putting the stars back in.

 

[wimvb]

$T is the image that you apply PixelMath to. It will be replaced. In this case, it is the starless enhanced image.

Copy1 and copy2 don’t need to be stretched as much as the starless, enhanced image. Copy1 will therefore have smaller stars without halos.

You can also create copy1 when the image still is linear and use a colour preserving stretch on it, such as arcsinh stretch. Then create copy2 from copy1 and use starnet on that. It’s all a matter of experimentation.

[Lee_P]

50 minutes ago, wimvb said:

$T is the image that you apply PixelMath to. It will be replaced. In this case, it is the starless enhanced image.

Copy1 and copy2 don’t need to be stretched as much as the starless, enhanced image. Copy1 will therefore have smaller stars without halos.

You can also create copy1 when the image still is linear and use a colour preserving stretch on it, such as arcsinh stretch. Then create copy2 from copy1 and use starnet on that. It’s all a matter of experimentation.

Thanks, this is really helpful.