HELP PLEASE ! Trying to make a colour image from R, G B and Luminance frames

[stevewanstall]

I have just started using a ZWO 1600mm pro with a filter wheel and ZWO LRGB filters. I have taken frames of a target using the different filters and am now trying to combine them into a   colour image.  I have tried following various tutorials but I end up with a monochrome image!

I attach my files in case anyone wants to have a go, in case the files are at fault.

What I have done is :

1. Using DSS, I have used the best frame as a reference image then registered and stacked the R, G B and L frames (I haven't bothered with bias, flats and dark for the moment). The alignment is fine, so  Iam at least doing that bit right.

2. I saved each of the files as a .tif without doing anything in DSS.

3. In Photoshop CS5 I  created a RGB file from the three separate files using the Channels ... Merge Channel. I have made sure the red goes in the red, green in green etc.

4. I then opened the luminance file,  did Select All, Copy and pasted that onto the RGB.

5. I  set the layers blend to Luminosity.

6. I then used Levels and Curves to stretch the image. 

This is what I get:

Untitled-3.tif

 

What am I doing wrong? I am sure it is something blindingly obvious to someone.

blue test.TIF gren test.TIF NGC4565 luminance.TIF red test.TIF

[Adreneline]

Well I am afraid I didn't use PS - I used PI to register and combine the RGB and stretch the image and then PS to lift/balance the colour and perform a little noise reduction.

There is some amp glow visible on the right but the darks should sort that; there is also some dust bunnies which the flats will sort.

I'm sure there is more colour there but I didn't push it too far. I didn't use the Lum data.

I'm sure a PS guru will pop up and do the whole thing in PS.

HTH

Adrian

Edited March 27, 2020 by Adreneline
Typo

[stevewanstall]

Thanks Adrian, I guess it shows there is some colour here somewhere !

[Bignerdguy]

As Adreneline mentioned i did this in Photoshop using a action i got for an older version of PS.  Its Called ColorComposite.atn.  It is designed to add and convert the color channels into a single color image while giving you several layer controls to adjust each channel separately (Histogram, Level and Curve).  Once you have it the way you like, you simply flatten the image to combine it.  In this one i did the action on the RGB channels and then added the luminance layer with only a levels layer for it.  I have added the action here and we will see if this attaches for you. Not sure if it will work in your version of PS as i am using CS3 still.  Anyway this was the best i could do as there was a lot of noise in the image (which more shots plus dark frames would cure) and i had to reduce and cut the luminance frame down as not only was it a larger frame but the bottom right corner of the file was cut off.  So i also cropped the photo on you.

NGC4565.TIF ColourComposite.atn

[stevewanstall]

Thank you Bignerdguy,  I will try the action. In the meanwhile, I have managed to make some progress using PS. I found it is better to process the RGB as far as I can in PS, then do the same to Luminance. Once they are 'complete', then to merge them either as  RGB or as Lab.  This is my LRGB version:

 

[StargeezerTim]

Here is my try using PS... Thanks for posting the stacks, it's good to practice. I only use DSLRs so rarely process like this.

In brief this is what I did:-

1. Combined the rgb into one one image and stretched and boosted colours then smoothed the image.

2. Resized the luminence, then stretched using curves and sharpened the bright areas. 

3. Added the luminence as a layer, changed blend mode to luminosity. Increase curves to upper layer. Reduced curves on rgb layer.

4. Flattened the image, final colour boost and adjust background. Get rid of gradients and hot pixels.

Cheers...

Edited April 2, 2020 by StargeezerTim

[wornish]

The Luminance file you shared is a totally different size to the R,G and B files .  

Can you share one the same size?

[stevewanstall]

Stargeezer Tim, very nice. What do you use to remove gradients in PS?

[stevewanstall]

4 hours ago, wornish said:

The Luminance file you shared is a totally different size to the R,G and B files .  

Can you share one the same size?

The luminance was binned 1x1 and the RGBs were binned 2x2. What I have done is to process the RGB file then increase the image size to match the luminance (4656 pixels on the long edge).

[StargeezerTim]

8 minutes ago, stevewanstall said:

Stargeezer Tim, very nice. What do you use to remove gradients in PS?

I have a very old copy of Pixinsight LE that has a DBE function. 

What I am doing more and more though is duplicating the image. Then applying enough gaussian blur on the duplicate to get rid of the stars and galaxies. Then go back to the original, go 'apply image', select the copy and 'subtract' the copy from the original. It usually does a good job' especially with galaxies and clusters.

[stevewanstall]

1 minute ago, StargeezerTim said:

I have a very old copy of Pixinsight LE that has a DBE function. 

What I am doing more and more though is duplicating the image. Then applying enough gaussian blur on the duplicate to get rid of the stars and galaxies. Then go back to the original, go 'apply image', select the copy and 'subtract' the copy from the original. It usually does a good job' especially with galaxies and clusters.

I shall try that! Gradients are a bit of a bugbear on most of the images I take, even after taking flats, etc.

[wornish]

Resized R G and B to same as Lum as that carries the detail.

Integrated in APP and then cropped and tweaked in PS.

Colours are a real challenge have to say.

Thanks for sharing your data.

[wimvb]

in PixInsight:

I think that the main problem is lack of data. How much was the total integration time?

[stevewanstall]

Just over 2 hours luminance and about 90 mins for RGB I think,

 

[wimvb]

I combined the unstretched r, g, and b images to create a colour image. I then used colour calibration (PI photometric color calibration) and arcsinh stretch (is also available for PS). I desaturated the background protecting the galaxy and stars.

I processed the luminance separately:

gradient removal (DBE in PixInsight)

deconvolution (sharpening)

stretch using histogram transform (levels in PS)

HDR compression to get more contrast in the dust lanes of the galaxy

Then I combined the L with the RGB

Topped off with star reduction.

[stevewanstall]

That's basically what I have determined I need to do. I don't have PI, but use DSS, PS and Lightroom. I process the luminance and get the best image I can, then combine with an RGB that has been similarly processed.