[RCC] M17 Swan Nebula in Narrowband - Forums

Hi all,

I'm working on learning narrowband processing.

My process:
- BXT/NXT on each of H, O, S
- NBColourMapper
- Colors mapped to 0, 120, 240 with saturation = 1 (this may have been going overboard!) and adjusting the histogram to align each channel
- StarXterminator to separate nebula and stars
- STF/HT on stars
- STF/HT, HDRMT, and Curves on nebula
- PixelMath to recombine

I've tried GHS on this one but just couldn't seem to get it to look as good as STF did in auto. This is my first time to play with HDRMT.

I'm happy for any advice for making this better.

Thanks,

Antha

Hi.

With SCNR you can play with Green, and can remove some amount Green excess.

Watch Sarah Mathews (SarahMaths) on youtube, SHO processing video and try her method for stretching the individual channels manually. https://www.youtube.com/watch?v=PfdOJZzyLEs&list=WL&index=2 Then switch over to PaulyMann Astro youtube video for Foraxx processing. https://www.youtube.com/watch?v=cl3_r3bL8Ys&t=6s
Just a suggestion. You can incorporate the parts you need into your workflow... or not.

You state on your image page the following: "This is a narrow band color mapped image where H-alpha is mapped to green, Oiii is mapped to blue, and Sii is mapped to red. I used STF/HT and HDRMT and curves to process the nebula in PixInsight."

If I'm not mistaken, Ha goes to red, Oiii goes to green and Sii goes to blue. If I'm incorrect, see disclaimer.

Also, if I'm wrong on that, someone straighten me out, please.

Disclaimer: This advice comes from a noob. I'm working on learning mono processing also.

DavesView:
If I'm not mistaken, Ha goes to red, Oiii goes to green and Sii goes to blue. If I'm incorrect, see disclaimer. Also, if I'm wrong on that, someone straighten me out, please.

For typical SHO combination, Sii is red, HA is green and Oiii is blue. The OP used the normal narrowband combination. The other typical use is HOO combination which is HA to red, Oiii to green and blue.

DavesView:
Watch Sarah Mathews (SarahMaths) on youtube, SHO processing video and try her method for stretching the individual channels manually. https://www.youtube.com/watch?v=PfdOJZzyLEs&list=WL&index=2 Then switch over to PaulyMann Astro youtube video for Foraxx processing. https://www.youtube.com/watch?v=cl3_r3bL8Ys&t=6s
Just a suggestion. You can incorporate the parts you need into your workflow... or not.

Exactly the kind of pointer I was looking for, thank you!

Iosif Bodnariu:
With SCNR you can play with Green, and can remove some amount Green excess.

Thanks for the hint! So a general question ... since the Hubble palette assigns Ha to green, and there's a LOT of Ha, why are so many of the Hubble palette images I see not predominantly green? Why is the green considered "excess"?

Antha Adkins:
Thanks for the hint! So a general question ... since the Hubble palette assigns Ha to green, and there's a LOT of Ha, why are so many of the Hubble palette images I see not predominantly green? Why is the green considered "excess"?

A good question and I don't know the answer except the the predominant of Ha and green is just visually not appealing.
I like to leave a slight amount but most people get rid of it. I think the edit by Iosef is more appealing than yours.
I also think you could experiment with tools like localhistagramequalization with masks and HDRmultiscale to bring our more detail in the nebula.

Dan Kearl:
DavesView:
If I'm not mistaken, Ha goes to red, Oiii goes to green and Sii goes to blue. If I'm incorrect, see disclaimer. Also, if I'm wrong on that, someone straighten me out, please.

For typical SHO combination, Sii is red, HA is green and Oiii is blue. The OP used the normal narrowband combination. The other typical use is HOO combination which is HA to red, Oiii to green and blue.

Thanks for clearing that up for me. Although, this chart muddies it up again. If 656 nm is Ha, how can Ha be green?

From Wiki: "Hydrogen-alpha, typically shortened to H-alpha or Hα, is a deep-red visible spectral line of the hydrogen atom with a wavelength of 656.28 nm in air and 656.46 nm in vacuum."

Dan Kearl:
Antha Adkins:
Thanks for the hint! So a general question ... since the Hubble palette assigns Ha to green, and there's a LOT of Ha, why are so many of the Hubble palette images I see not predominantly green? Why is the green considered "excess"?

A good question and I don't know the answer except the the predominant of Ha and green is just visually not appealing.
I like to leave a slight amount but most people get rid of it. I think the edit by Iosef is more appealing than yours.
I also think you could experiment with tools like localhistagramequalization with masks and HDRmultiscale to bring our more detail in the nebula.

Thanks for the tips! Exactly the feedback I was looking for. I have another version using the Canada France Hawaii Telescope palette, now posted, and I think I like that palette better. Thoughts?

Antha Adkins:
I have another version using the Canada France Hawaii Telescope palette, now posted, and I think I like that palette better. Thoughts?

I've only processed one image in SHO and I used the ForaxxPaletteUtility script, so Foraxx Palette for me for now. Also, if Ha is red, how can it be green? See above. When you look at the color chart above, there really is no blue in the filter set of SHO. Oiii is 500nm, which is green. Ha is 656nm, which is red and Sii is 672nm which is even more red. It's all very confusing to me. Maybe in the subs where the images overlap is where the variant comes from in the colors other than those indicated in the chart.

DavesView:
Antha Adkins:
I have another version using the Canada France Hawaii Telescope palette, now posted, and I think I like that palette better. Thoughts?

I've only processed one image in SHO and I used the ForaxxPaletteUtility script, so Foraxx Palette for me for now. Also, if Ha is red, how can it be green? See above.

I'm using the NBColourMapper script, which is really cool and allows you to map each narrowband channel to any color and allows you to see the result and play with it before saving. So I've posted two versions (I've actually made four) - one using the Hubble palette where Ha=green, and one using the Canada France Hawaii Telescope palette, where Ha=red.

DavesView:
Thanks for clearing that up for me. Although, this chart muddies it up again. If 656 nm is Ha, how can Ha be green?

That's a really neat chart! The answer is this is false color mapping with the purpose of having more contrast in the final image. I'm using the NBColourMapper script, which is really cool and allows you to map any filter to any color and allows you to see the result and tweak it before saving. So H-alpha is red in reality, but can be mapped to another color to generate contrast. If you look, H-alpha and Sii are very close on the spectrum, so there's not much distinction between them without false color mapping.

Antha Adkins:
I'm using the NBColourMapper script

I will have to play with that one, definitely.

Antha Adkins:
Iosif Bodnariu:
With SCNR you can play with Green, and can remove some amount Green excess.

Thanks for the hint! So a general question ... since the Hubble palette assigns Ha to green, and there's a LOT of Ha, why are so many of the Hubble palette images I see not predominantly green? Why is the green considered "excess"?

When mapping to the Hubble palette, it usually will be predominantly green. Depending on the object, that can vary from just mostly green, all the way to overwhelmingly and obnoxiously green. The reason you don't see SHO (Hubble Palette) images as predominantly green, is because of removing a bunch of the green (which is what Iosif was mentioning with removing excess green). Once you start to pull the green back, the Hubble Palette you're used to seeing starts to show up.

Few different ways you can do it depending on how careful you want to be. Very quick and easy, but less nuance, is SCNR as they mentioned. Just removes the green, which leaves you with the gold and blue you're used to seeing with the Hubble Palette, but it comes across a bit bicolor (this is just straight up removing a color after all). Taking a little more time, you can also do it with curves which is my preferred method. Doing it this way will give you some better color variety since you're not just removing a color. You'll get the blue and gold from before, but also some burnt oranges, some green will stick around, you'll see some different shades of blue, etc. That's where you start to get what you think of when you talk about Hubble palette images that you've seen. Is when you map Sii to R, Ha to G, Oii to B (SHO - Hubble palette), and then pull back the green to reveal the rest of the colors since by default, the green is so overbearing.

I suggest some experimentation with Power-of-Inverted-Pixels palettes in PixelMath, such as:SHO: // R

Oiii^~Oiii)*(N*Sii + (1-N)*Ha) + ~(Oiii^~Oiii)*Ha// G

(Oiii*Ha)^~(Oiii*Ha))*Ha + ~((Oiii*Ha)^~(Oiii*Ha))*Oiii// B

iii // SymbolsN=0.6 OR: where f = Oiii with contrast from curves added// R:f*(N*Sii + (1-N)*Ha) + ~f*Ha// G:f*(M*Ha + (1-M)*Sii) + ~f*Oiii// B

iii // Symbols:N=0.75,M=0.6

Take a look at https://thecoldestnights.com/2020/06/pixinsight-dynamic-narrowband-combinations-with-pixelmath/ and https://www.youtube.com/watch?v=ZmML4UekFMg

Todd Anderson:
Antha Adkins:
Iosif Bodnariu:
With SCNR you can play with Green, and can remove some amount Green excess.

Thanks for the hint! So a general question ... since the Hubble palette assigns Ha to green, and there's a LOT of Ha, why are so many of the Hubble palette images I see not predominantly green? Why is the green considered "excess"?

When mapping to the Hubble palette, it usually will be predominantly green. Depending on the object, that can vary from just mostly green, all the way to overwhelmingly and obnoxiously green. The reason you don't see SHO (Hubble Palette) images as predominantly green, is because of removing a bunch of the green (which is what Iosif was mentioning with removing excess green). Once you start to pull the green back, the Hubble Palette you're used to seeing starts to show up.

Few different ways you can do it depending on how careful you want to be. Very quick and easy, but less nuance, is SCNR as they mentioned. Just removes the green, which leaves you with the gold and blue you're used to seeing with the Hubble Palette, but it comes across a bit bicolor (this is just straight up removing a color after all). Taking a little more time, you can also do it with curves which is my preferred method. Doing it this way will give you some better color variety since you're not just removing a color. You'll get the blue and gold from before, but also some burnt oranges, some green will stick around, you'll see some different shades of blue, etc. That's where you start to get what you think of when you talk about Hubble palette images that you've seen. Is when you map Sii to R, Ha to G, Oii to B (SHO - Hubble palette), and then pull back the green to reveal the rest of the colors since by default, the green is so overbearing.

Very helpful, thank you!

Roger Nichol:
I suggest some experimentation with Power-of-Inverted-Pixels palettes in PixelMath, such as:SHO: // ROiii^~Oiii)*(N*Sii + (1-N)*Ha) + ~(Oiii^~Oiii)*Ha// G(Oiii*Ha)^~(Oiii*Ha))*Ha + ~((Oiii*Ha)^~(Oiii*Ha))*Oiii// Biii // SymbolsN=0.6 OR: where f = Oiii with contrast from curves added// R:f*(N*Sii + (1-N)*Ha) + ~f*Ha// G:f*(M*Ha + (1-M)*Sii) + ~f*Oiii// Biii // Symbols:N=0.75,M=0.6

Take a look at https://thecoldestnights.com/2020/06/pixinsight-dynamic-narrowband-combinations-with-pixelmath/ and https://www.youtube.com/watch?v=ZmML4UekFMg

Very helpful, thank you!

Iosif Bodnariu:
Hi.

With SCNR you can play with Green, and can remove some amount Green excess.

If you want more control than SCNR offers, use the Curves tools. You can nuke the green entirely, or not, or shift it and control it in ways you'll never be able to achieve with SCNR. You can transfer the signal power from green to bluer or yellower colors as you see fit. You can even use multiple curves to accomplish the task of managing the green and redistributing that signal to other channels. The Lab channels a and b are one of the best ways to shift the signal from green, but you also have the Hue curve, Chroma curve, and then of course the RGB curves themselves. You can work in all concurrently, or just tweak them one at a time until you like the colors you end up with.

Hello,

There are some great tips here and to expand on your green question here:

Dan Kearl:
Antha Adkins:
Thanks for the hint! So a general question ... since the Hubble palette assigns Ha to green, and there's a LOT of Ha, why are so many of the Hubble palette images I see not predominantly green? Why is the green considered "excess"?

A good question and I don't know the answer except the the predominant of Ha and green is just visually not appealing.
I like to leave a slight amount but most people get rid of it. I think the edit by Iosef is more appealing than yours.
I also think you could experiment with tools like localhistagramequalization with masks and HDRmultiscale to bring our more detail in the nebula.

I would not use SCNR to remove the Green from your image. SCNR is "Subtractive Chromatic Noise Reduction" and the green is here is not noise. SCNR is better suited for green pixel background noise which is very common in OSC RGGB since 50% of the pixels captured are green. Now yes it does work in accomplishing removal of the "green" here for a "better Hubble Pallet" looking image but I would balance/fit the images instead.

In Mono you are using filters to isolate the gases and the each mono image captured will have different intensities/brightness. Since the HA is so dominate, if you look at the histogram of each mono image you will see that the HA peak and data is much farther to the right/brighter than the other images. By balancing (fitting) the images before combination it wont allow the HA to be so dominate and will show the other gases you have captured creating some nice colors in your image.

You can accomplish this different ways. The two most common Pixinsight that I know are LinearFit and SPCC. You can Google both of these processes for more detail. I now prefer SPCC with the proper sensor selection, filter selection, and background selection, for neutralization, it does an amazing job of balancing the colors of your image.

If you look at my recent images I don't use the Hubble Pallet (SHO) anymore as I much prefer HSO Pallet using ImageMapper script. Best of luck with your journey into Mono.

Jon Rista:
If you want more control than SCNR offers, use the Curves tools. You can nuke the green entirely, or not, or shift it and control it in ways you'll never be able to achieve with SCNR. You can transfer the signal power from green to bluer or yellower colors as you see fit. You can even use multiple curves to accomplish the task of managing the green and redistributing that signal to other channels. The Lab channels a and b are one of the best ways to shift the signal from green, but you also have the Hue curve, Chroma curve, and then of course the RGB curves themselves. You can work in all concurrently, or just tweak them one at a time until you like the colors you end up with.

Very helpful, thank you!

James Hawks:
I would not use SCNR to remove the Green from your image. SCNR is "Subtractive Chromatic Noise Reduction" and the green is here is not noise. SCNR is better suited for green pixel background noise which is very common in OSC RGGB since 50% of the pixels captured are green. Now yes it does work in accomplishing removal of the "green" here for a "better Hubble Pallet" looking image but I would balance/fit the images instead.

In Mono you are using filters to isolate the gases and the each mono image captured will have different intensities/brightness. Since the HA is so dominate, if you look at the histogram of each mono image you will see that the HA peak and data is much farther to the right/brighter than the other images. By balancing (fitting) the images before combination it wont allow the HA to be so dominate and will show the other gases you have captured creating some nice colors in your image.

You can accomplish this different ways. The two most common Pixinsight that I know are LinearFit and SPCC. You can Google both of these processes for more detail. I now prefer SPCC with the proper sensor selection, filter selection, and background selection, for neutralization, it does an amazing job of balancing the colors of your image.

If you look at my recent images I don't use the Hubble Pallet (SHO) anymore as I much prefer HSO Pallet using ImageMapper script. Best of luck with your journey into Mono.

Thank you, this was very helpful.

Antha Adkins:
Iosif Bodnariu:
With SCNR you can play with Green, and can remove some amount Green excess.

Thanks for the hint! So a general question ... since the Hubble palette assigns Ha to green, and there's a LOT of Ha, why are so many of the Hubble palette images I see not predominantly green? Why is the green considered "excess"?

It's about how you choose to color balance, or color calibrate, your image. Take a look at the Pixinsight tutorials on narrowband processing:

https://www.youtube.com/watch?v=WYPVtmhXJYk
https://www.youtube.com/watch?v=c-JnE7LXtn8

I found them interesting and helpful. Even if you don't use their processing flow, I think it gives a good overview of a rationale for choosing a white color reference to color balance an image that ultimately reduces the level of green in your image.

Tom Boyd:
It's about how you choose to color balance, or color calibrate, your image. Take a look at the Pixinsight tutorials on narrowband processing:

https://www.youtube.com/watch?v=WYPVtmhXJYk
https://www.youtube.com/watch?v=c-JnE7LXtn8

I found them interesting and helpful. Even if you don't use their processing flow, I think it gives a good overview of a rationale for choosing a white color reference to color balance an image that ultimately reduces the level of green in your image.

Thank you, this was very helpful.

Thanks to all for your suggestions and links. I am learning a lot! Here is a new version, where I have done the following: - Used LinearFit to normalize the images to the same median value prior to NBColourMapper using Ha as the reference. This made a huge difference.- Ran NBColourMapper to make a SHO version and a HOS version, using background neutralization. I didn’t need to do anything to the input histograms since LinearFit had taken care of that.- Played with using BackgroundNeutralization and DBE after NBColourMapper. I liked what DBE did – made the background blacker, and running BackgroundNeutralization after DBE didn’t seem to have much effect. So this image used DBE.- Tried using SPCC in narrowband mode. This seemed to want to push the nebula back to the original colors – very green for SHO and red and gold for HOS (I really liked the latter). This image did not use SPCC.- Tried using ColorCalibration. This brought out the reds and golds in the SHO image, so this image includes that effect, though I liked it without ColorCalibration as well.- Ran StarXterminator. Stretched the stars and used ChannelExtraction to make a luminance channel and then used ChannelCombination to make a RGB image so the stars would be white. I thought that looked better than the magenta SHO stars.- Stretched the nebula. - Ran HDRMT on the nebula to pull out details in the core.- Ran Curves on the nebula for a final stretch. - Used PixelMath to add the stars back in. What do y’all think? Any more suggestions? Thanks so much!! Antha

Narrowband colour mapping is highly subjective.
The Hubble team assigns these colours to easily (visually) determine the composition of the gases in the nebula. Otherwise, it would appear teal, red, and ultra-red, and it would be much harder to discern the difference between the Ha and S2 components.

This is a typical result when three filters are assigned these colours. It varies with the Ha component of different nebulae.

Most people would agree though, that for a pleasing visual aesthetic, 'blue & green should never be seen without a colour in-between', so typically the green is replaced with warmer tones, and pink stars have no business appearing anywhere except Disneyland!

I did this one in photoshop using selective colour. Here is a tutorial that may help. http://bf-astro.com/hubbleP.htm

Finally, here is an example of M17 in Narrowband using a pleasing colour palatte. - https://www.astrobin.com/jgy3xo/
CS
Andy