What colour is Ha?

[gregbradley]

I posted this on an Astromart forum and I thought it might be of interest to imagers here.

It is about what colour Ha is and thus what colour it should be assigned to it in Photoshop.

I also wonder if one reason some imagers complain about "Salmon" colours from some HaLRGB combination is because the wrong colour is being assigned to Ha. Photoshop assigns a basic red (around 700nm) when you use the red channel. This is not the correct colour.

Here is the article I wrote:



I researched this a bit more.

The following link shows a spectrum of colours and their associated wavelengths.

http://www.midnightkite.com/color.html

Ha light is 656.281nm. So per this spectrum it is an orangy red which
is often depicted in astroimages as a dark red, per this spectrum, incorrectly.

H beta wavelength is 486.13nm and per this spectrum is a light sky blue. But a Ha
filter would not be letting through any Ha beta as that is a long way from 656nm.
Assigning some blue to it would thus be twisting the data as it would really need to be
a H beta filtered image to get the correct H beta intensity. Also the blue assigned by
Photoshop would have to be that light sky blue shown in the link and not the heavy bright and darkish
blue Photoshop uses as default.

Also, I wonder what shade of "red" Photoshop uses when one assigns a channel to "red"?

"Red" of course is a scale from lightish tones to deep tones closer to infrared.The programmers of
Photoshop, perhaps arbitrarily, select a particular shade of red, which, in common usage considered "red".

Using the above spectrum then I wonder if it is possible to define which shade of red you want Photoshop
to use when it assigns red to Ha.

Photoshop has a palette of colours one can select for other actions. I wonder if
it lets you define a red from that palette for the purposes of assigning red to a channel.
I imagine it would if someone suffered through the Help file!

So in conclusion I take Ha to be a orangy red, not deep red, not orange, not pinkish but
a red leaning towards the orange side of the scale of red.

Using the above I have attached a Ha image I took of Eta Carina using a 12.5 inch RCOS and
adjusted the red channel assigned so that it matches quite closely the orangy red from the spectrum
in that link.

I think you'll agree this is not how it is normally represented. So I conclude that noone
is showing Ha in its more accurate colours and I will reprocess this Ha into an LRGB image to
see what it looks like and post it.
Greg.

[Bassnut (Fred)]

Attached ?

[gregbradley]

Here are 3 example images to show what I was talking about.

Firstly, a straight Ha image using Eta Carina using an RCOS 12.5 inch and STL camera with an Astrodon 6nm Ha filter. I have worked to match the colour assigned to the Ha to match the spectrum example in the link in my article. I think it is pretty close.

http://www.pbase.com/image/103334252

Secondly I got a straight LRGB colour image of the same object/scope/camera:

http://www.pbase.com/gregbradley/image/103334257

And thirdly I added the spectrum matched Ha image above to this
colour image and this is the resulting blend:

http://www.pbase.com/gregbradley/image/103334261

I personally like the resulting blend but this is only one example. I wonder how it will work out with different types of objects imaged.

Time will tell as I begin using this technique.

Greg.

[[1ponders] (Paul)]

Thanks for the write up Greg. It gives me hope that maybe my initial salmon pink efforts are not so far from reality after all.

[Satchmo]

I remember catching visual sensations of colour in outer parts of Eta Carinae with a 20"binocular under superb conditions and they were definately the kind of red you depict here

[Jeffkop (Jeff)]

An interesting and im sure, quite valid theory Greg. Explained well and the examples are excellent. As usual, Im in awe of your knowledge.

[Prickly]

Off topic on colours I know but this link shows in one image how the excited electrons dropping from higher into lower energy orbital states release radiation to make up the various parts of the spectrum for hydrogen. Mind the maths though - which Im no longer game to try and figure out!

Its quite interesting really to see how you get the various bands.

http://hyperphysics.phy-astr.gsu.edu/hbase/hyde.html

Cheers
David

[Ken]

Hi Greg, I think you are onto something with the Ha red leaning towards the orange. I also think the deep red Ha images can look wrong. I have also recently found blending deep Ha data with RGB a difficult task when the aim is to make a natural looking image. Your bi colour images look great, maybe this is the future. I guess you are creating an artificial green image to go with the Ha red and O111 blue. Some more information on this would be great. Looking forward to more of your images

Clear skies
Ken C

[avandonk]

Colour does not exist! It is only because of the ratios of different frequencies of light due to the receptors in our eyes that give us the sensation of colour. To try to 'assign' a colour to HA is meaningless as our visual system barely perceives it. It should be a very lipstick red if anything as it is far down the spectrum to what humans perceive as red.

Bert

[Peter Ward]

Originally Posted by avandonk

..... It should be a very lipstick red if anything as it is far down the spectrum to what humans perceive as red.

Bert

As Bert suggests, H-alpha is deep red. Just look at the sun with a H-alpha solar filter
...which passes *only* H-alpha light to sub-angstrom precision...

The perceived colour is unmistakable. H-alpha emissions from nebulae are *exactly* the same colour....just a lot less bright.

A nebula however may have more than one emission (eg H-Beta) which will may give it a more magenta hue.

But as for the colour of H-Alpha, I would not have thought there is any debate.

[Peter Ward]

...I should have also mentioned in my previous post.... it is the phosphors of a display that define what its version of pure "red" is (ie not software, eg photoshop).

All any computer monitor can do is light up all its red phosphors for *any* deep red...which is a poor approximation to a real spectrum....which varies continuously in colour.

As you move toward the blue, other phosphors light up and trick the eye (which doesn't resolve the individual RGB phosphors...unless you look really closely!) in to perceiving a gamut of colours until we run into the same problem at the Blue end (eg CaK line)

[Peter Ward]

Further still to my previous two posts..... (I have this thing about red it seems )

A little research on the web shows the red phosphors of CRT's and LED/LCD monitors to have a wavelength of about 620nm: a fair way short of the 652nm of H-Alpha.

Hence even if you make a H-alpha image as deep a red as the display allows, it isn't quite deep enough.

Making the image an orange, salmon, tuna or whatever fruit/fish colour you like simply takes you up the spectrum: in short, going the wrong way.

I suppose the best analogy I can give is: viewing CRT/LCD images is a bit like listening to a melody being played with the middle two octaves of a piano. The melody is generally recognizable, but we really need all 88 keys to fully appreciate it.

[gregbradley]

Hi Peter,

I don't think that is quite right.

Firstly LCD monitors do not use phosphors - that is used in CRT or Plasma screen per this article:

"So what about the LCD, today’s most obvious alternative to plasma? A liquid-crystal television is, in effect, a sandwich with many ingredients. Its layers include a bright white backlight, a layer of liquid-crystal molecules, a matrix of thin-film transistors, two pieces of polarized glass, and colored filters. The transistors control the voltage applied to the three groups of liquid-crystal molecules that make up each picture element.
When the voltage is on, it twists the molecules, allowing light through the layers of glass and color filters; the molecules untwist when the voltage is off, blocking light. Each picture element consists of liquid-crystal molecules above a red, a green, and a blue filter. Switching the appropriate molecules on and off gives myriad combinations of red, green, and blue light, and therefore the palette of human vision."

Secondly, there is very little data on the internet about this and this surprised me but the few spectrums I could find that showed wavelengths show "red" as a band from 620nm to 780nm. So Ha at 656.281 is not that far from light red. Some posts state red at 650nm others at 680 - so it seems theres not a lot of agreement about what "red" is.

My point was red is a scale, specifically it is 620nm to 780nm in wavelengths.

Photoshop arbitrarily assigns a non defined red to your Ha channel when you use that technique.

My point is to question the correctness of the red Photoshop is assigning to your Ha layer. It may not be right. It may be too red or indeed not dark enough red.

Per that link, given the frailities of what the computer screen displays etc, the 656nm area defintely showed a bit of orangy red which is the colour that images of the Sun are often displayed as.

If you disagree that is fine but it would boil down to disagreeing with the accuracy of the spectrum I linked to rather than theory about monitors and what they can or cannot display in the red band of the visible spectrum.

Deep red is for sure closer to 700nm per any spectrum I have been able to find on the net.

I am sure not all monitors have the same red response. Certainly checking out different laptop monitors showed a huge variance in ability to display different colours from images of my choosing rather than the sales ones.
So basing the red you use based on your monitors performance is a wild variable.

Greg.

[h0ughy (David)]

excellent topic - and thanks for the research and discussions so far.

[Garyh]

Very interesting! Must agree with Peter here but. I think these simulated spectrum's are not quiet right...As you cones and rods in your eyes can`t see out of the 380 to 680nm range, how can you put a color on the spectrum above or below these values? Would be like describing the color red to someone that`s blind...
652nm is getting very close to what we can see so has to be a very deep red and deeper than what consumer visual products can simulate.

[Peter Ward]

Originally Posted by gregbradley

Hi Peter,

I don't think that is quite right.


Greg.

Well, sorry I think it is correct.

The web and many manufacturers give Gamut data on their monitors...including LCD/TFT/CRT

eg http://www.atscope.com.au/BRO/images/data/samsung.jpg

As far as I am aware there is not a single monitor on the market that can display colours longer (ie redder) than a 620nm wavelength (this is the point at which red is displayed at in intensity of 255 with G, B at zero).

What a monitor tries to come up with and reality are invariably two very different colours. (unless the colour being displayed lies within the monitor's triangular gamut boundary)

But to keep it really simple.....I invite anyone to take look at the sun through a H-Alpha filter/scope. 652nm, viewed directly, appears as a *very* deep red to the human eye.

You can make H-Alpha look whatever colour you like on a monitor, but unless the monitor has a Red gamut boundary lower than 620nm (if you find one, let me know! ) the representation will be too orange.

[Peter Ward]

Of course, sometimes the bleedin' obvious passes me by.....what colour is
H-Alpha even if you don't have easy access to a solar scope?

Simple!

Just grab a the narrow band H-Alpha filter you are using to image with, and look at a bright white light through it (duh)

Pretty deep red eh?

P.S.
10nm pass filters don't look quite so deep as they obviously pass a little 645nm light...as opposed to 3nm pass versions....they are very crimson.

[winensky]

Thanks for this fascinating discussion. I have asked a few times in the past if any of our processing programes have rendering tools for the narrow band colours. This discussion has given me some new thoughts on how to approach my HaR blending in the future.

[multiweb (Marc)]

Here's an interesting link. No extreme spectrum though

http://www.xrite.com/custom_page.aspx?PageID=77

See how good you go. Got 28 in the 40-49 age range.

[gregbradley]

[quote=Peter Ward;365861]Of course, sometimes the bleedin' obvious passes me by.....what colour is
H-Alpha even if you don't have easy access to a solar scope?

Simple!

Just grab a the narrow band H-Alpha filter you are using to image with, and look at a bright white light through it (duh)

Pretty deep red eh?

P.S.
10nm pass filters don't look quite so deep as they obviously pass a little 645nm light...as opposed to 3nm pass versions....they are very crimson.



Hi Peter,

Yes I thought of the same thing. And yes they are quite a deep red.
So that spectrum linked is obviously quite inaccurate. That is a simple and workable test.

Even so the red Photoshop assigns then is not as deep as what you see
through the filter. Suggests that it might be good practice to select a dark crimson colour to Ha layers or channels.

Greg.