Hi Jarmo,
We haven’t spoken in a while.
I hope you are well.
I came across a free utility that produces some very nice Narrowband results - Foraxx Palette Utility.
Where Pixelmath uses Static combinations e.g. 0.75Ha * 0.25SII, Foraxx creates Dynamic combinations by taking into account the pixel values from the image itself to produce a better and more accurate result.
If you wanted to implement this in AutoIntegrate then I imagine it would be an option on the “Extra processing” tab, in the section “Extra processing for narrowband”
The reason that I would put it in Extra processing is because it works on a stretched image.
If you were to include it in autoIntegrate, I think the process would would involve 5 steps that I am aware of:
Create Starless image using StarXterminator or Starnett
Run PixInsight Channel Extraction on Starless image AND run again on star image.
3.Run Foraxx Palette Script
For SHO (or HSO which also works out well), select “three channels option”
For HOO select “two channels” option
Assign Starless Channels - R channel to Sii for SHO or G for HSO and so on.
Assign the Star Channels the same way then Execute the Script.
This produces a Foraxx Palette image.
I run these steps manually and it’s not too much effort, but it might be a nice feature to add if you are looking for ideas.
Good to hear from you. It is summer here so not much astro but some other activities.
There is already Foraxx palette in AutoIntegrate but it is under the name Dynamic SHO. I have never been too happy when using that palette but from your comments I realized that maybe it should be run with non-linear images.
When you select ‘Dynamic SHO’ palette and click ‘Narrowband mapping using non-linear data’ you get image close to what you have posted.
It is not quite as good so I need to check if it could be improved. Or maybe it should be run later as you suggested.
Hi Jarmo,
I tried Narrowband mapping using non-linear data but the script errored because I had BXT enabled (error message explained that it is not compatible with BXT).
As you know, BXT needs to be run on linear data early in the process.
I think Foraxx would work best after all process is done (which is how I am applying it manually).
What are your thoughts?
It might be easier to implement as an extra option because of the need to create a starless image.
If the user selects multiple narrowband mappings - SHO, HSO, etc then the program logic would need to remember to do the extra processing right at the end on the SHO image (remove stars, extract starred and starless channels and run Foraxx). I’m not sure…you would know better than me.
Also, I’m not sure if it’s intentded to be used this way, but I have also run Foraax on HSO images to test the results of Dynamic vs Static pixel math and (if I remember correctly) I was pleased with the result.
Having Foraax as and Extra option would allow this flexability.
I looked into code and I agree that extra options looks easiest and best overall. I will keep old Dynamic palettes but will move Foraxx to the extra options. I have all the code to extract and combine channels so it should not be too hard.
In GitHub there is now a version where Foraxx palette is available as a narrowband extra option. There are SHO and HOO options.
I did not get exactly similar images as the ForaxxPelatteUtility script creates. But then I noticed that it uses some curves and saturation enhancements so I think that explains the different end result.
Please let me know how it works for you and if some more tuning/fixing/changes are needed.
I added an option to adjust reds after Foraxx palette. It also add saturation on orange and blue. Below is my before and after when using Foraxx mapping, Leave some green and Orange/blue colors.
Maybe still a bit too red? Feel free to suggest a hue curve from CurvesTransformation tool if you want a different look.
Hi Jarmo,
That’s great.
The images look fantastic.
I will try and compare the results of running foraxx manually vs AutoIntegrate.
When I run it manually, I remove the stars first then run Foraxx on starred and then starless, then recombine (which it what they recommend doing). I’m assuming I could just remove stars then just use the Foraxx checkbox on my starred and starless images separately, then recombine in AutoIntegrate as well?
