Using Red and Blue to create a synthetic Green?

[iksose7]

Hi guys,

After a 6 month hiatus from imaging i managed to capture some data. Unfortunately its only L, R and B. Put simply, how to i combine my grayscale R and B images to make my G. Btw i know creating synthetic green is frowned upon by some but i wont get another chance to image for a few weeks at least now so thought this can be a learning experience. I have looked at some tutorials but they are all different and deal only in narrowband, i just can seem to make anyhthing work.

Any tips would be welcomed!

Thanks and Merry Christmas

[msmythers]

I would say technically you can't get green from red and blue but you have L also. I don't do filtered images but I would try something like this. Since L is made up of the full spectrum, including green, I would use pixelmath  and subtract r and b from L. Without knowing the precise filters transmission characteristics the percentages to make a correct green will be off and that is where you would need to do some trial and error testing. Anyway that is how I would approach this. I could be way off since this is not what I work with.



Mike

[msmythers]

I just gave what I was suggesting a try with a M11 image from Jim Misty of Misty Mountain("Image Acquisition by Jim Misti".) I basically subtract the R from the L, then the B from the L. Added those together for a new almost green. Then I added the L back into the almost G for the new Green. I combined the R,B and new Green. I used autohistogram for a quick color balance, SCNR at 70 percent and some color saturation.

The original I combined the R,G and B. Autohistogram, SCRN@70% and color saturation.

Here is the result.


Mike

[msmythers]

Thought I'd show a better view of the 2 images. While not perfect I think it's not bad for a synthetic green. More playing with the color balance would help. And just to be clear these images are still linear in nature using STF linked to stretch.

Mike

[iksose7]

Hi Mike,

You may be right about not really being able to create the green, this is new to me but i thought i had heard about it in the past. Other than the gradient your SG has introduced, it seems to have worked well for the star clusters color. I cant seem to get your process to work for me. Did you work with linear images? Is the Pixel math a simple L-R?

Thanks
Callum

[msmythers]

Callum

First yes all of this is done linear, even the AutoHistogram operation. I just used the median value of the new G image as the target. I did not stretch the image like you would normally do. Color saturation and SCNR was done to the linear image. The pictures are nothing more then screen captures of the images with linked AutoSTF applied.

The pixel math is that simple. L-R is one. L-B is another. New L-R + New L-B which is New Almost Green. And them L+ New Almost Green for the New G.

The gradient is actually the background not being color balanced and the saturation of the background being too high. I would normally never just color saturate the stars without a mask. The Autohistogram only works the median of the image not the minimums like BackgroundNeutralization. This was just a quick process test. Bottom line once you get the synthetic green just combine with the R and B and threat like any RGB image and do proper color balancing. Then if you needed one can separate the balanced channels and do whatever is needed for the finished image.



Mike.