Yet more dragons

[kosborn]

Hi all,

This is my take on NGC 6188. It is 24x 5min Ha, 22x 5min OIII an 18x 5min SII for a total 0f 5.3 hours.

A lot of SHO images posted on astrobin seem to be presented as blue/orange with all the green removed. As the hydrogen is represented in the green channel, and because hydrogen is the main element in emission nebulae, I think I prefer the blue/green/gold combination. I would be interested to know what others think. Pixel peeping version at astrobin (https://www.astrobin.com/full/357208/0/?real=&mod=).

I have some nice RGB data and would like to add RGB stars. RickS, if you happen to wander past this post, I found the following you posted on a different forum a couple of years ago:

"I usually add RGB stars to a NB image by extracting the L, a* and b* channels from each then using PixelMath and a star mask to blend the a* and b* from the RGB with the a* and b* from the NB image. Recombine the L, a* and b* for the NB image and voila! Sounds easy but there's always a bit of messing around and tuning for any specific image. The advantage of this approach is that the original NB star profiles are preserved and only the colour is changed."

Can you expand on that for a PI wannabe?

Kevin

[lazjen]

Nice.

I believe when it comes to Narrowband images, you can pretty much display as you want.

Those images that appear to remove the green are really "shifting" it via hue changes. The data is still there, not removed.

Up to you, but you might want to look into getting rid of the magenta tinge on the stars. I realise you want to add RGB stars, so that may not matter when it's done.

[multiweb]

Very nice. Good star colours too. I'd leave as is. That's what good NB stars look like. :thumbsup:

[Placidus]

No need to change a thing! Very good as is.

Our philosophy on colour using three narrowband filters:

(1) Our primary goal is to convey meaningful astrophysical information. If Mike wants to be primarily artistic (https://mikeberthonjones.smugmug.com/Sculpture/i-Hf8xsCS/O), he gets out the clay or the oils.

(2) We make the picture as beautiful as we can consistent with the primary goal.


(3) The kinds of astrophysical information we want to show:

- Hydrogen is very easy to ionize, so H-alpha tends to show where the bulk stuff is. It also shows shock fronts nicely.

- Oxygen requires high energy to ionize. OIII can be seen near the extremely hot white dwarf in a planetary nebula. Otherwise, it tends to indicate extreme violence. It is mostly seen near hot young OB stars that will soon go supernova, but can be excited by extremely violent shock fronts, e.g. in an SNR.

- Sulphur is found only where there is dredged-up material from the guts of a star where nucleosynthesis has progressed a long way. Sulphur is rare rather than ubiquitous. It also requires high energy to ionize, so SII emission is rare. The [SII] to H-alpha ratio is unusually high in supernova remnants. You can also get [SII] in the atmosphere of a PN or WR nebula.

(4) Now for the nitty gritty:

- Because OIII is less common than H-alpha, it requires more exposure. Since SII is dead rare, it requires long exposure.

- We don't try to show the relative abundances of H-alpha, OIII, and SII. That would be like a map of the mineral wealth of Australia trying to show the relative abundances of sand, coal, and gold. All you'd see was sand.

- Instead, we're trying to show action and structure: to show the locations of what little OIII and SII there might be, as these are special places.

- Since our primary goal is to convey true and genuine astrophysical information, we use the Hubble palette, because everyone immediately understands what they are looking at. If you translate a beautiful poem into let's say ancient Hittite or Martian, only very old Hittites and Uncle Martin will understand. Hence we map SII, H-alpha, and OIII to red, green, and blue respectively.

- We remove gradients from moonlight or light pollution, and set the black point very carefully in narrowband images. Otherwise one can falsely fill the background with OIII and SII, when all you've really got is moonlight and air glow and the floodlights at the Council car park. Makes a beautiful picture, but 'tis a lie.

- As regards the stars, they are incidental to the story, and their colour is of little or no interest. Consequently, we divide the image into two components: the stars and the nebulosity, and we process them separately.

- To avoid having the image flooded by H-alpha (with OIII getting lost and SII undetectable) we colour balance the nebulosity so that the image as a whole is colour neutral. That is to say the mean values for red, green, and blue for the whole image should be the same.

- At this point, it is easy to use reasonably strong wavelet sharpening on the nebulosity-only image. (It is very difficult if the stars are still in).

- To avoid distracting magenta haloes around the stars, and to avoid implying that the colours of the stars have significance, we suppress the OIII and SII channels in the star image, keep only H-alpha, and map the H-alpha to white.

- We then recombine the stars with the nebulosity to produce the final image.


Having said all that once again, your image, exactly as it is, is very pleasing. The goal of the diatribe is to save you from the Dark Side. We've tried to write the above description in a non-commercial way, concentrating on the rationale and goals, rather than saying "We use Bob's WobbleBobble action from his GobbleBobble package", when we have no real idea of what those actions do.

Very best,
Mike (in charge of maths) and Trish (colour and aesthetics)

[Benjamin]

Lovely image Kevin and a super informative post Mike and Trish.

[kosborn]

Thanks Chris. I've seen a northern hemisphere youtuber use SCNR in PixInsight as the first step in SHO processing to remove all the green, ending up with bright blue and bright orange. I didn't think that was a good approach!





Thanks Marc.





Thanks Benjamin.





Thanks Mike and Trish. Aesthetics is important but not at the expense of the science. Far from being a diatribe, your post is very informative and much appreciated!


Cheers, Kevin

[Andy01]

Nice image, well composed and good details :thumbsup:

As to colour - depends on your intent. :question:

You can do a M&T and be as scientifically informative as possible, whilst still producing a pleasing image.

... or you can go for aesthetics, using complimentary colours while still producing a scientifically valid image, which becomes an work of art.
Examples of these can be found in the works of Rick Stevenson, Suavi, Strongman Mike and well, yours truly.

Each to their own, there are no rules in NB imaging, only trends :)

[kosborn]

Thanks Andy,

I've only been doing astrophotography since February (and narrow band for considerably less time) so I'm still finding my way. I was lurking quietly on this forum for a while before posting and have been using the images posted by yourself and all the others as inspiration for targets and processing techniques. So much to learn!

By the way, how much are you enjoying your RC8? When I am comfortable with my Esprit and looking for more focal length an RC8 would be something to consider. I've heard horror stories though about collimation difficulties...

Kevin

[Andy01]

Hi Kevin, there's always lot's to learn and you're in the right forum for generous advice and assistance. Keep going, you'll be fine :thumbsup:

Re: The RC8, it has been a source of both frustration and Joy.:shrug::)

I love the longer FL - and it's about as much FL & weight as my EQ6 can carry.

I've had teething issues though, I had to ditch the stock focuser & immediately replace it with a Moonlite. Drill a hole in the tube to fit a heater to the secondary, add a counterweight to the front, and finally and most importantly engage a colleague's services to collimate the thing properly using a custom adapted Tak microscope! (It proved to be beyond my skills to do it properly).

So nine long months later, it's all sorted nicely and I'm gathering some quality data for my first NB image that is (hopefully) going to be up to par with the results I'm used to from my refractors! :D

[muletopia]

Hello Kevin,
As an aspiring astro photographer I found the range of colours in your image most interesting. The I then read the crontribution of Placidus. What a well reasoned and educational exposition. I shall certainly refer to it when I reach this level in practice,oh so far to go.
So many thanks for this thread and its contributers.
Chris

[Benjamin]

Just peeping at your pixels (lovely flat background by the way) I wonder if you have exactly the right distance between the camera and the Field Flattener? The pic is gorgeous as it is and I don’t really care about the corners myself but a few millimeters extra either way might make all the stars lovely and round?

[kosborn]

As I suspected. As much as I want the extra FL I think I'll stick with the Esprit for a bit longer!

Kevin

[kosborn]

Hi Chris,

Bite the bullet and go for it! In my limited experience, astrophotography is a lot of fun.

Kevin

[kosborn]

Hi Benjamin,

Absolutely correct. My sensor is exactly 3mm too far back from the field corrector. I thought I was close enough but your eyes are obviously better than mine. I will correct it one day soon.

Kevin

[Benjamin]

No rush to correct it I think as there are so many lovely elements in the picture that make that element almost irrelevant. :thumbsup:

[multiweb]

:bowdown:

[Andy01]

Lol ;) :rofl::rofl::rofl:

[RickS]

A nice image, Kevin. I'm not a fan of nuking all the green either, though I do tend to shift it somewhat towards the blues.

Some insightful posts in this thread as well :thumbsup:



Extracting the CIELAB from NB and RGB images is easy: ChannelExtraction. Then I normally do some PixelMath like this: "iif(star_mask>0.1,mean(NB_a,RGB_a),$T)" Apply to the NB a* data and this blends the data in the areas covered by the star mask (the magic 0.1 can be adjusted to tweak the star coverage - a smaller number exposes a larger area around the stars.) A similar PixelMath expression is used for b*. Recombine the L* and modified a* and b* with ChannelCombination and see how it looks. You'll probably need to redo the PixelMath a couple more times to get a stronger effect.

Cheers,
Rick.