I'm working an HaLRGB data set and I'm following Philippe Bernhard's "Processing of LHa-RHaGB images with RGB stars" workflow.
See here: LINK
I'm stalled at the point where I combine my stretched starless L and my stretched starless Ha with a process identified on the workflow diagram as "MixSHO AIP".
A little google-fu shows that the script has been deprecated.
What should I be using in it's place??
TIA, CS
Gordon
MixSHO AIP: what to use?
Well Written Respectful Engaging
Blend Ha with L at your leisure. But I can't find "MixSHO AIP" in the linked diagram.
I’d say this method is pointlessly overcomplicated.
You should combine LRGB first, either still in the linear phase or the standard LRGBCombination after stretching. Narrowband addition should come as one of the final steps to an image, I often do it just before adding the stars back in. Star extraction can sometimes eat some signal in the narrowband galaxy images, so be careful when using it. Noise reduction can sometimes work fine before continuum subtraction but it can lead to artifacts, so you’ll need to see if it works on your data. Same with blurx.
So a simple Ha workflow would look like this:
Correct gradients > (optional) blurx, NR > continuum subtraction (optionally with starless images) > stretch > colorize > screen back onto the final LRGB image
This way you avoid needless steps like matching the Ha addition strength in L and RGB and further boosting the Ha later on and the result is basically the same.
Helpful
andrea tasselli · Sep 9, 2025, 02:06 PM
Blend Ha with L at your leisure. But I can't find "MixSHO AIP" in the linked diagram.
Sorry, should have included a link to the diagram icons: LINK
I’ve copied the diagram here, with the step circled in yellow:
📷 Diagram.png
And as to whether or not the flow is “pointlessly overcomplicated”, with all due respect, I’m certainly not able to argue that as I’m too inexperienced but I AM at a point in my workflow on this project where I’d just like to know what to use to replace the deprecated script, not backup and/or start over.
OK, so that is "MixLHa" then. Use PixelMath to blend (which at its simplest is to average the L with the Ha, assuming LinearFit has been used beforehand) the two channels together, L and Ha. But that is just one of the possible combinations you can make.
andrea tasselli · Sep 9, 2025, 03:09 PM
OK, so that is "MixLHa" then. Use PixelMath to blend (which at its simplest is to average the L with the Ha, assuming LinearFit has been used beforehand) the two channels together, L and Ha. But that is just one of the possible combinations you can make.
Thank you sir.
Not having done anything yet with pixelmath, do you have an expression that would get me started?
Well Written Respectful
If you want to average both images and assuming your L would be called “image1”and your Ha “image2”, you could write in Pixelmath:
mean(image1,image2)
Concise
ArchStarGazer · Sep 9, 2025, 05:30 PM
If you want to average both images and assuming your L would be called “image1”and your Ha “image2”, you could write in Pixelmath:
mean(image1,image2)
Boy, that’s an easy one… My ignorance speaking here, but i thought it would be more complex than that!
Thanks!
Gordon Pegue · Sep 9, 2025, 03:03 PM
And as to whether or not the flow is “pointlessly overcomplicated”, with all due respect, I’m certainly not able to argue that as I’m too inexperienced but I AM at a point in my workflow on this project where I’d just like to know what to use to replace the deprecated script, not backup and/or start over.
According to the chart, you would only need to backup one step on your RGB image, basically undo Ha addition to the RGB image. The rest is about moving the Ha addition to the final steps in the processing. Other suggestions I made were supposed to be notes/tips for you to consider for future edits. You don’t need to start anything over.
Helpful Concise Supportive
As a side note I aways add Ha to R and to L in a LHaRGB image, as shown in the linked diagram above.
Mikołaj Wadowski · Sep 9, 2025, 06:45 PM
Gordon Pegue · Sep 9, 2025, 03:03 PM
And as to whether or not the flow is “pointlessly overcomplicated”, with all due respect, I’m certainly not able to argue that as I’m too inexperienced but I AM at a point in my workflow on this project where I’d just like to know what to use to replace the deprecated script, not backup and/or start over.
According to the chart, you would only need to backup one step on your RGB image, basically undo Ha addition to the RGB image. The rest is about moving the Ha addition to the final steps in the processing. Other suggestions I made were supposed to be notes/tips for you to consider for future edits. You don’t need to start anything over.
Ah OK.
FWIW here’s where I’m at in my flow (using the boxes on the diagram as a reference); I have these intermediate images:
“Stars”, non-linerar, stretched with StarStretch script;
“Starless RGB”, linear, not yet stretched with GHS;
“Layer Ha”, linear, not yet stretched with GHS;
“Layer L”, linear, not yet stretched with GHS.
I actually have not yet done any stretching or combining. My original question was posed because I was anticipating getting to that step where I “Mix LHa” and couldn’t determine what to use to do that.
For combining “layer Ha” and “Starless RGB”, @rqfugate advised me to use the Cosmic Photons ImageBlend script for whichever “Method” I choose (the boxes have the big orangish “S” against the sky-blue diamond). My project is LBN 437 so I’ll probably try “Method 1: Linear” for that combination since my data is still linear. I believe that simplifies the process a bit…
If you think I am indeed taking the “harder road”, I’m all ears! What changes to the flow would you then suggest (based on what I have now in my 4 intermediate images?
Helpful Respectful Engaging
Gordon Pegue · Sep 9, 2025, 07:50 PM
If you think I am indeed taking the “harder road”, I’m all ears! What changes to the flow would you then suggest (based on what I have now in my 4 intermediate images?
So I would go about combining all of them as follows: L → RGB, then Ha → LRGB, and finally Stars → HaLRGB.
You can take two routes for combining LRGB. Either stretching both L and RGB to about the same level and using the LRGBCombination process or doing it while they’re both still linear. Personally, I prefer doing it the former way as there’s less room for error and it’s just much simpler. I was shown this method by Charles Hagen.
Extract the average of the linear, color calibrated RGB image with
avg($T[0], $T[1], $T[2])in pixelmath.Linear fit the linear L image to the image you just created.
Use
$T L / avg($T[0], $T[1], $T[2])on the RGB image.
If you did it right, you should end up with a perfect LRGB combination, with minimal changes to the color intensity. I always do it pre-starx, but doing it on starless should also be fine.
Next, proceed to stretch the LRGB image and do any adjustments you’d like, be that saturation, contrast, HDR etc. The point is, your LRGB should be as close to the final version as possible before combining the Ha.
After that, stretch the continuum subtracted Ha and edit it to your liking. After you’re done, colorize it and use screen blending to combine it with the LRGB image.
Then do the same for the stars (except colorizing of course ;))
The method described in that flowchart will work too, it’s just that it’s more work and there is more room for error by adding Ha before combining LRGB. I’d say traditional LRGB combination, that is using LRGBCombination after stretching, is one of if not the hardest step in processing, so complicating it further should be avoided.
Helpful
Mikołaj Wadowski · Sep 9, 2025, 08:50 PM
Gordon Pegue · Sep 9, 2025, 07:50 PM
If you think I am indeed taking the “harder road”, I’m all ears! What changes to the flow would you then suggest (based on what I have now in my 4 intermediate images?
So I would go about combining all of them as follows: L → RGB, then Ha → LRGB, and finally Stars → HaLRGB.
You can take two routes for combining LRGB. Either stretching both L and RGB to about the same level and using the LRGBCombination process or doing it while they’re both still linear. Personally, I prefer doing it the former way as there’s less room for error and it’s just much simpler. I was shown this method by Charles Hagen.
Extract the average of the linear, color calibrated RGB image with
avg($T[0], $T[1], $T[2])in pixelmath.Linear fit the linear L image to the image you just created.
Use
$T L / avg($T[0], $T[1], $T[2])on the RGB image.If you did it right, you should end up with a perfect LRGB combination, with minimal changes to the color intensity. I always do it pre-starx, but doing it on starless should also be fine.
Next, proceed to stretch the LRGB image and do any adjustments you’d like, be that saturation, contrast, HDR etc. The point is, your LRGB should be as close to the final version as possible before combining the Ha.
After that, stretch the continuum subtracted Ha and edit it to your liking. After you’re done, colorize it and use screen blending to combine it with the LRGB image.
Then do the same for the stars (except colorizing of course ;))
The method described in that flowchart will work too, it’s just that it’s more work and there is more room for error by adding Ha before combining LRGB. I’d say traditional LRGB combination, that is using LRGBCombination after stretching, is one of if not the hardest step in processing, so complicating it further should be avoided.
Thank you sir! When I get back on this project, I’ll give your methodology a try.
Well Written Respectful