Do you have a tutorial on splitting the individual channels and restacking? I'm using PixInsight, I've gotten somewhat comfortable with the calibration, registration, and integration process (including debayering and drizzle), but I've not read about that kind of process.
The CA removal is ok for most of the image, but it definitely creates weird artifacts in the Horsehead region (colorless rings around stars) mostly due to the fact that the color in that region is the same as the CA being removed. What I ended up doing was removing the CA in one version of the image, then layering another version of the image without any CA removal and masking in the Horsehead region. This is with the 135mm lens, btw, the 85mm lens appears to have a bit less CA and so removal didn't really create any artifacts or problems. I also had the 85mm lens stopped down quite a bit to f4 to reduce coma and taking 5 minute exposures whereas with the 135mm I was shooting f2.8 to capture more light with 2 minute exposures.
I don't have a specific "tutorial" but in PixInsight the essential steps are to:
1) Split the single RGB color image into it's constituent color channels (use "Split RGB Channels" which has an icon along the top or if you prefer the menu navigation it's under "Image" -> "Extract" -> "Split RGB Channels". This will create three views which will all have the same name as your original view but they'll append a suffix "..._R", "..._G", and "..._B" after the name.
2) You can now treat those images as if they were all taken independently. This means you can use the same tools you would normally use to align the stars between images if you were taking lots of "lights" and wanted to stack them. True chromatic aberration caused by a lens (as opposed to atmospheric dispersion caused by the sky) causes the "blue" light to bend inward more and focuses at a closer distance than the blue, which focuses at a closer distance than the red. That means that your "blue" channel image stars are actually just fractionally closer together than your "red" stars. PixInsight has options when doing star alignment which allows it to resize the image when registering the image.
One of the cool things about PI is that images do not need to be in the same scale. If you were doing a mosaic of a section of sky, you can use manual star alignment tools where YOU pick the same star in two different images (so PI knows that this star in image #1 is the same star in image #2 even though the images are just overlapping areas of sky and not the same piece of sky). You normally do this with at least 2 or three stars to get a good fit.
But another cool thing is that multiple astro-images can collaborate on imaging... e.g. one astrophotographer captures some data using their scope & camera and another gets data using a different scope & camera... but they merge their data. The obvious problem is since the scopes may be completely different models, the area of sky can be completely different, the orientation of the cameras may have been completely different, and the image resolution and size of the cameras may have been completely different. But PI can rescale the images to make them fit. Automatic star alignment may fail (and probably will) because the scale is off... but if you use the same technique of identifying the same 2 or 3 stars in different images, it will understand how to do any manipulation to make them fit.
In essence this is what you're doing with the CA correction... except in your case the stars will nearly fit because they were all from the same image, but the CA means there's just a tiny amount of scale difference between the images.
Dynamic Alignment can do it, and I think there are options (modes) of Star Alignment that can do it. If what your really have is atmospheric dispersion and not CA then the atmosphere is separating the light and usually that causes the R, G, & B channels to all be slightly shifted (offset from each other) but not necessarily re-scaled.
You do have to pick a reference channel. I pick green because that's in the middle of the color spectrum but I'm not sure it really matters... and then tell PixInsight to register the red & blue channels to my green channel.
3) Now that you have "registered" images, you can re-combine them using a tool such as "Channel Combination".
You should now be back to a single RGB image, but the color channels for your stars should now line up nicely.
4) Since you fractionally adjusted your images to make the color channels align nicely, your borders no longer align so nicely. You'll want to do a very slight crop to get rid of any ragged edges caused by the star alignment process.
