Color adjustment of (3 channel) NB data

[dhalliday]

I continue to "aspire" to  relying on NB data from my crummy,urban skies...
I a have to admit that it is slow going...on a number of levels.
Two main question (working with stacks of Ha,OIII and SII...)
1) do I combine them (unproccessed) in "color combination" as RGB...?
Or do I do something more clever...use PM,process them first...?

2)In "Curves"...what channels do I play with...??
I am confused by all the choices...ie RGB and then "a"  and (?) "h", etc...
What am I doing here..?
Again SII is very faint,but I am using it...

Anyone who is a NB guru...if you could give some tips...I would be grateful..
Here are some of the (bad) results I am getting so far...the colors look odd,in my mind.
http://www.flickr.com/photos/daveh56/4628151524/

and
http://www.flickr.com/photos/daveh56/4636708014/

In these cases SII was a synthetic "mash up" of the Ha and OIII which Sivercup was kind enough to explain to me,in PM...

Dave

[Nocturnal]

Hi Dave,

I don't do NB so take this with a grain of salt. Your images aren't bad at all. I think they've improved actually. Still way too much flexure but we talked about that before so we'll leave it at that

I recommend not messing with synthetic anything right now. In fact I don't see the point of it at all. Mixing two NB channels doesn't add any data at all. Just settle on which NB channel you map to which RGB channel and use what you have. The picture is not natural and so you should not aim to make it look like RGB, IMO.

Curves allows you to play with hue, saturation, luminance and other image parameters. Understand curves in general and understand the parameter or channel you're processing and you understand what it's doing. I know that sounds cryptic but you need to apply divide and conquer. Use curves on L, R G and B and perhaps saturation, then move on from there. There's no magic wand waiting there that'll improve your images several times over

I only use saturation curves to 'amplify color'. This has nothing to do with NB vs. RGB so I don't think you should try to tackle both at the same time, certainly not in the same forum topic.

[dhalliday]

Sander

Thanks
Well..the "synthetic third channel"  (which many use..) is because often the SII lacks much/any structure detail...
You NEED three to have color...no ?

So I am trying with a Ha/OIII blend...and also with SII, as shot.I may well stumble on a solution myself here...
BUT...

ONE of my issues/questions is....well...the "signal" is only 5-7 nm wide in each channel...
So spectrally my "color image" is abnormal from the get go.
The RGB filters on my camera ( in "normal" color imaging)...provide a broad swath of spectral signal...
I just wondered if there is a BETTER way to adjust the NB colors than just "bodging around" in curves...
ie if there was a scientific rationale to doing this NB work with colors...(in Curves...)
I see R,G,B,RGB/K,A,L,a,b,c etc etc as options in Curves...
In Pixel Math I see R/K,G,B,A....etc etc.
In "Channel combination" I see HSV,HSI,...and "CIE"...(THREE options...)
Confused..??
Just a bit...

As for my flexure....its still there...
Looking at the last 9 frames I shot last nite (M57 in OIII,) the flexure is (at 1800mm fl) the flexure is 1.3 pixels in Dx and .6 in Dy...this is per frame,720 second exposures,7.4 micron pixels.

But again,its the color adjustment and processing of NB data that I am all mixed up about.
Its not a case of "NB vs RGB"....
Its about the way to work with NB data...to get a "color image" that has good (adjustable) final color.
Volunteers...??

Dave

[Nocturnal]

Dave, any image with 3 channels is technically a color capable image. You can store monochrome images in them if you like. If you keep one channel unused (0) you still have a color image. Give it a try and see how it looks.

Don't think too hard.

[dhalliday]

http://www.flickr.com/photos/daveh56/4664898045/

But again...I am not happy with the color..
Detail is "OK"....

Dave

[Nocturnal]

I think you need to reset your NB expectations. Not sure what you expected would happen but this looks like a decent image to me. Try experimenting with different color mappings to see if you like some colors better. Stars will always look a bit 'odd' as they won't be white. I'd say that's cool because the nebulae look nice.

[Andres.Pozo]

Well..the "synthetic third channel"  (which many use..) is because often the SII lacks much/any structure detail...
You NEED three to have color...no ?

You can create color images without needing three channels. A couple of weeks ago I did an experiment for finding a way of taking images of galaxies from a place with a very bad sky. Usually using RGB filters I can't get enough signal/noise for a decent image. I tried taking only L and H-alpha frames and then combining them.
The result was much better than I hoped, and I got a color image using only one filter!!

Image Data:

• Telescope: LX200R 8" with astrophysics CCD.67 reductor
• Camera: QSI540wsg
• Frames: (L)12x300" bin1x1, (Ha)12x300" bin2x2

[Nocturnal]

To be fair, when combining L and a single color channel you should end up with a monochrome image. Of course there's nothing stopping you from assigning L and Ha to different color channels but it makes little sense.

[Andres.Pozo]

To be fair, when combining L and a single color channel you should end up with a monochrome image.

I think (and my image proves it) that is not true. It is not a full three color channel image, but clearly it is not monochrome. It is not also a toned image (i.e. sepia ).

Of course there's nothing stopping you from assigning L and Ha to different color channels but it makes little sense.

Why it makes little sense? In the example that I have shown, the H-alpha areas are enhanced and IMO that it is interesting by itself.

[dhalliday]

Well...this (two channel creation of a "false" third signal) is the heart of (part of) my question.
I assume you used Pix Math ( ) to "create" a third channel out of the other two...
So this then allows you to have a RGB to use...
But of course the synthetic channel is actually DUPLICATING image structure seen in the other two...
So you MUST be losing structure detail...

The SECOND part of the question was about what way to use curves to adjust the color.
Again...the "color" has very little spectral range...(by nature of the filter...)
As far as (final) star color this can be fixed with a star mask...
Anyone..??

[Nocturnal]

But of course the synthetic channel is actually DUPLICATING image structure seen in the other two...

Exactly!

Again...the "color" has very little spectral range...(by nature of the filter...)

I don't know what that really means but you're thinking too much again After putting Ha in the R channel the R channel represents the intensity of light that passed through your filter. Ha plus whatever else made it through the bandpass. So dark areas mean little Ha, bright areas mean a lot of Ha. Red represents Ha, it does not represent Red. You can stick Ha in any channel you like. The decision on where to put what is referred to as the palette.