This is from a recent thread
on the same subject;
Note that this is discussing OSC/DSLR data (e.g. with different channels in one file). If you already have separate stacks for each band, you can skip the first step below;
There are a couple of ways of doing that.
The most flexible, fool-proof way is to create a new red channel (for the purpose of coloring only, but not luminance!). You can use the Compose module to extract the red channel from your full-color datasets. Just load your color stack into the red channel only and it will extract just the red channel and create a "black and white" dataset out of just that channel (if Channel Interpolation is On). You can then save that dataset and extract the next one.
If you already have separate stacks for each band, you can skip the above. Following on;
Once you got these two mono datasets (one with red channel, one with Ha), you can use the Layer module to create a blend, or you can use the Compose module (set Luminance, Color to L + Synthetic L, Mono) to create a blend (also in mono). Then save this blend. This will be your new red color channel.
It should be stated that adding Ha to an RGB image doees not have an "exact" answer or "correct" method. You are combining narrow band with wide band data with very different signals and response curves, so any notion of exposure times or weightings goes out of the window. As a matter of fact, you can choose to use a very different approach to making you new red channel. The Layer module is perfect for this sort of thing due to its different filter types and compositing methods ("Layer Modes"). You could create a HDR composite with the "Min Distance to 1/2 Unity" filter, or use the "Lighten", "Add" or "Screen" Layer Modes. You may wish to chain different operations.
It goes without saying that when you have your new red channel, you don't want to use that as a basis for a synthetic luminance dataset. If you want to use a synthetic luminance frame,
Now use the Compose module again, set Luminance, Color to L, RGB, and load your full-color dataset as Luminance, Green and Blue, while loading your newly created red channel as Red.
You should now be ready to process the luminance of your full color stack as normal, while the Color module will bring in coloring that consists of R+Ha, G and B.
The other way is to process you color stack as normal and save it away.
Then to process your Ha stack as normal and overlay the two processed stacks in the Layer module using, for example, a Lighten blend mode.
The latter completely side-steps the method of the creation of a new red channel. E.g. you just process your RGB or LRGB as normal. Then process your Ha-only, save it, re-import it as red channel in the Compose module and turn off channel interpolation and set "Luminance, Color" to "RGB, RGB". This should yield a completely pure red (processed) image. Now open the Layer module and open the color image in the background. Set the Layer Mode to "Lighten", "Add", "Screen" or even "Color of Foreground". Try different Brightness Mask modes and Brightness Mask powers, try different filters and kernel radii to make the Ha "glow". Over-drive the effect with Blend Amount.
This is one of those answers that are a lot less straightforward, as there are so many ways to achieve different goals. That said, I've been thinking of adding dedicated functionality to make adding visual Ha easier...