How should I combine Ha with LRGB?

Hello Andy,
Another alternative I find works well for galaxies is to extract channels after a channelCombination, and use the Blend script in Pixinsight to overlay 20% (tweak the opacity percent until I am happy) Ha using Overlay Mode = Screen (negativ multiply) and disable Linear STF. Upper image is Ha and lower image is Red and adjust the opacity around 30%. Then use the result to recombine the Ha+R with GB to make a new RGB.

I used the channel extraction + pixelmath method for a while but then I read Mastering PI and I found very effective the way RBA explains in his book. He uses a Max expression that will combine your Ha pixels with the broadband image IF they have a higher value. From here you can play with the stretch levels and expression symbols to obtain the "blending percentage" you want. Really effective.

I've seen a lot of articles quoting the following PixelMath formula for combining Ha and Red into the red channel:

    ((HA * R_bandwidth) - (R * HA_bandwidth)) / (R_bandwidth - HA_bandwidth)

I never had much joy with it and now I have just realised that it does not take into account the different exposure times that you may use between RGB and narrowband.  I think the formula should be:   

    ((Ha*Red_Exposure*Red_bandwidth)-(Red*Ha_Exposure*Ha_Bandwidth))/((Red_Exposure*Red_bandwidth)-(Ha_Exposure*Ha_Bandwidth))+Red

I've tried my new version of the formula and it actually seems to work.

If you leave out the final "+Red" in the formula you can actually see what the Ha is adding on top of the Red.

Thoughts?

OK, I used my pixelmath formula:

    ((Ha*Red_Exposure*Red_bandwidth)-(Red*Ha_Exposure*Ha_Bandwidth))/((Red_Exposure*Red_bandwidth)-(Ha_Exposure*Ha_Bandwidth))+Red

and a similar one to combine Ha and Lum to create the following in 15 mins from the LRGBHa masters.




M81 and M82 - Early posting ... need much more acquisition time

It's not perfect, but I was impressed by how much detail I could get in M82 whilst almost bringing out M81 in such a short time.  I will process it properly in the future.  

FWIW:  process was:

* Dynamic crop all images
* DBE all images based on points chosen on red frame
* Use above formula to create a pseudo-red channel
* Channel combination of that pseudo red channel with green and blue
* SCNR to remove excess green on this new RGB image
* Used the above pixelmath formula, but with Lum instead of red to create a pseudo luminance channel
* Used LRGBCombination to apply this new pseudo-luminance to the RGB image
* EZDenoise
* HDRMultiScaleTransform


I will redo this using Starnet++ to help me retain the star colour and stop the stars from bloating, but I am impressed that pixelmath can get me 90% of the way there in 10% of the time

John Hayes:
Most of the methods here simply try to adjust the contrast of the Ha signal relative to the bandwidth difference between the red and Ha signals.  I think that it's better to isolate the actual Ha signal and to use PixelMath to add it in the correct ratio to the red and blue channels of the RGB image.  Vincent Perez and Martin Pugh describe the method in this article.  I've done this a number of times and found that it maintains the correct color balance of the image while adding just the Ha regions on top of the broadband colors.

http://www.arciereceleste.it/tutorial-pixinsight/cat-tutorial-eng/85-enhance-galaxy-ha-eng

The one additional thing that I do is to adjust the Lum channel by adding the Ha signal in PixelMath.  You have to be very careful to control noise so I use a mask for this operation.

This image shows the result of adding pure Ha:  https://www.astrobin.com/uijojf/E/

John

Thank you John! I was feeling a bit uneasy with Andy's formula because in my humble understanding it could lead to negative pixel values in theory - maybe not in practice? The formula you present looks safe in this respect while taking the different exposure times into consideration.

Clear skies
Wolfgang