Why does the colour wheel exist?

[sophiecentaur]

The truth is that there is no 'hidden' information needed to resolve this particular question.

People keep going on about cross connections between receptors. There is no need for all that, if you want to explain the problem. The brain has only three signals to for its input and it can get identical combinations of signals for a whole range of received spectra. The 'non-spectral' colours, which cannot be produced by a single spectral line are not really different from the 'spectral' colours. No Artist ever painted a picture using spectral light sources to produce their fantastic coloured pictures of sunsets or flowers.
With additive colour synthesis, Red and green phosphors can produce a yellow colour which will match the Sodium D yellow - to an average Human Colour Vision and an appropriate combination of blue and red will give the same colour sensation as a spectral line that we would call violet. There is no contradiction or paradox in this. It's complex, in detail, because we are talking about human perception, but the basic principle is not hard and doesn't require anything 'fancy' from the hardware.
I have said this before but the human eye is not a spectrometer - it is (a fairly crude) colour matching engine and can be 'fooled' into thinking that the spectra from two sources are identical when, in fact they can be wildly different. The brain sees things in a way that can be conveniently and simplistically mapped onto the "colour wheel" which is mentioned in the OP. Better than the Colour Wheel is the CIE Chromaticity chart, which has a numerical scale associated with it and allows the design of good colour imaging systems.

 

[I like Serena]

Can you mix colors to a color that fools an observer into thinking it is spectral violet?
Or spectral red near the infrared?

 

[sophiecentaur]

What you need to do is to look at the CIA diagram. If you have phosphors way out at the violet corner of it, you could produce spectral violet. That means you would need a spectral violet phosphor. I don't think that is a practical proposition.
I have a feeling that you knew that already(?)
Haha

 

[I like Serena]

What you need to do is to look at the CIA diagram. If you have phosphors way out at the violet corner of it, you could produce spectral violet. That means you would need a spectral violet phosphor. I don't think that is a practical proposition.
I have a feeling that you knew that already(?)
Haha

Fair enough! :)
I guess I was only trying to make the point that violet can not be mixed from other colors than violet, or we would not perceive it as violet.

My problem remains that I'm still confused about the difference between blue and violet.

If I accept your explanation, then spectral blue would only be observed to be blue because it triggers the red/green cones.
And if the red/green cones were not triggered (by some means), we would be fooled into seeing violet.

Somehow, that seems to be off, since violet has a tinge of red in it.

 

[schip666!]

I decided against the "can I really see violet" experiment due to both practical and philosophical difficulties. But I did look at the macaque data in the previously ref'ed paper (Table 1. in http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1192171/pdf/jphysiol00526-0162.pdf) and believe that it indicates a small up-blip for red-cone sensitivity in the far blue meaning that there may be significant differences in the 3-space signals (combined with green response) to either side of "pure" blue. Still would be nice to see the human-Nature paper, but who am I to ask...

Some of this discussion may be off-point due to conflation of magenta and violet -- and whatever happened to "indigo" which was previously touted to be out there beyond violet? Magenta is a _non-spectral_ color and is produced by exciting blue and red cones near both their peaks. Violet is spectral and does -- in theory -- have a specific wavelength somewhat below 400nm. One of my philosophical difficulties with experimenting was that I have no idea how to point to the actual BLUE stripe from a prism in order to say that there is something I see on the other side. I suppose that could be solved with a spectrometer...but I can't get that at the library either.

I also noticed -- finally -- that the two response graphs on the wiki pages (color vision, color space) have different vertical scales. The vision one mumbles something about being linear and seems to be normalized, whereas the space one (which shows the red-blip) is clearly log. So maybe we've been talking about the same thing all along.

As to seeing color change into the IR, the tabular data and wiki graphs both indicate that green and blue sensors level out as the red tails off. So I presume all one sees is a gradual dimming of red rather than a color shift.

 

[sophiecentaur]

I decided against the "can I really see violet" experiment due to both practical and philosophical difficulties. But I did look at the macaque data in the previously ref'ed paper (Table 1. in http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1192171/pdf/jphysiol00526-0162.pdf) and believe that it indicates a small up-blip for red-cone sensitivity in the far blue meaning that there may be significant differences in the 3-space signals (combined with green response) to either side of "pure" blue. Still would be nice to see the human-Nature paper, but who am I to ask...

Some of this discussion may be off-point due to conflation of magenta and violet -- and whatever happened to "indigo" which was previously touted to be out there beyond violet? Magenta is a _non-spectral_ color and is produced by exciting blue and red cones near both their peaks. Violet is spectral and does -- in theory -- have a specific wavelength somewhat below 400nm. One of my philosophical difficulties with experimenting was that I have no idea how to point to the actual BLUE stripe from a prism in order to say that there is something I see on the other side. I suppose that could be solved with a spectrometer...but I can't get that at the library either.

I also noticed -- finally -- that the two response graphs on the wiki pages (color vision, color space) have different vertical scales. The vision one mumbles something about being linear and seems to be normalized, whereas the space one (which shows the red-blip) is clearly log. So maybe we've been talking about the same thing all along.

As to seeing color change into the IR, the tabular data and wiki graphs both indicate that green and blue sensors level out as the red tails off. So I presume all one sees is a gradual dimming of red rather than a color shift.

Do you not see the difference between a curve that gives the sensitivity of each receptor to a set of wavelengths and a curve that shows the way that human colour vision MATCHES a colour to one which which has been synthesised with three standard phosphors? Try to think out of the box!

@I like serena
Violet is an exceptional 'colour', in that it lies outside the gamut of most phosphor triads. Consequently, it cannot actually be synthesised on a TV screen or by any other normal colour syntheses. It is a bad example because we actually 'see' violet very seldom. The violet we see is mostly a very desaturated magenta and not spectral violet at all. If you try to look at the (spectral) violet in a rainbow, it it very low luminance and is lost in the white / blue background of the clouds /sky. To see a synthesised violet you would need a phosphor that is already violet - you couldn't do it with a normal RGB set of phosphors and 'Ultra Violet' is, by definition, invisible and wouldn't register on any of out sensors.
I assume you accept my argument about the synthesis of Yellow, as an example - so you can see how violet isn't included in any practical system.
Colours that lie outside the gamut of colour TV phosphors are all reproduced as colours lying on the perimeter of the triangle. If you look at crowd scenes, you will notice an awful lot of brightly coloured clothes that appear to be of the same colour (rain jackets at a tennis match, for instance). They aren't all really the same colour in the original scene but the TV system gives them all the same CIE coordinates on the TV screen, which is the best it can do. Violet is never shown, any more than a really saturated orange or yellow.

Also, we really do see the violet when we look at spectral violet. It's not "fooling" our sensors - what we get from the three of them is a combination of three signals that is unique to spectral violet. Violet (and saturated spectral red) are the only two colours that will only match to their own monochromatic wavelength. We can "fool" our eyes pretty well about other spectral colours.

"Fooling" is not the word to apply to the analysis of our eyes. All we do is put lots of spectral patterns into the same 'memory compartment" and group them as the same colour. Like I said already - our eye is not a spectrometer - it just gives us a very crude analysis of the spectra we see, using just three parameters. Nature never knowingly does more than it needs to. What would we do with a spectral plot of everything we saw? Would it be of any more use to us than 'YUV' values?

 

[schip666!]

Do you not see the difference between a curve that gives the sensitivity of each receptor to a set of wavelengths and a curve that shows the way that human colour vision MATCHES a colour to one which which has been synthesised with three standard phosphors? Try to think out of the box!

yah yah, I get the idea that the CIE graph is generated by color-matching experiments and the other one is -- maybe -- the measured sensitivities of each cone type. It did take a bit for me to grok that, but I get it...Chalk it up to language difficulties... But I believe one needs a blue bump in the red receptors response in order to get a real violet sensation and the macaque data indicates that there is such a bump. This is not shown on the putative cone sensitivity graph. My hypothesis is that the graph's Y axis is linear and what we are looking for is down in the noise so would only show up on a log plot.

Further, to my simple mind the easiest way to get to the CIE matching results is to have that blue response bump. Otherwise we'd be having to mix some green into the magentas, no?

Plus I would still like to see some definitive measurements of human cone sensitivity.

I think we are talking about the same thing, just using different terms...

 

[I like Serena]

we actually 'see' violet very seldom. The violet we see is mostly a very desaturated magenta and not spectral violet at all.

That's a good point!

I guess the only place we ever really see violet in practice, is indeed in a rainbow.

So I guess I'm wrong thinking it has a tinge of red in it.

 

[sophiecentaur]

I am so pleased that people are taking this on board at last! I was beginning to feel like Cassandra.

I think the reason for that 'bump' on the red primary matching curve relates to the fact that both red and yellow cone responses extend right over there and it is necessary to exaggerate the amount of red primary presented to the eye for it to notice any difference between the M and L receptor responses - else it would be difficult to differentiate between colours that lay in the blue region of the chart.

Furthermore, there are all those colours that lie on the 'straight bit' of the CIE chart and which are below the central 'white' spot. The only way to match those is to present the eye with high levels of R and B phosphors and little, or no Y phosphor. To represent a good, saturated bluish, magenta you would also need a large contribution of red primary, although the colour you are after is way over on the blue side of the triangle. But you can't draw a simple graph of that which would be along the lines of the one 'with a red bump'.

Now we all have access to computers and colour sliders we can play at producing all sorts of colour matches and get a better clue about just how the CIE diagram and the triangle of primaries work. It can be very 'illuminating' to play with RGB values and see what comes out on your colour control panel.

http://en.wikipedia.org/wiki/CIE_1931_color_space" link gives a good description of things. They make the point that there is a practical limit to the wavelengths usable for the R and B primaries which can be used because our sensitivity is very low at either end of what we actually call 'visible'. We would need kW of output for such phosphors and to no point.

 

[sophiecentaur]

Further, to my simple mind the easiest way to get to the CIE matching results is to have that blue response bump. Otherwise we'd be having to mix some green into the magentas, no?

Not sure I agree with your there, actually. If you added Green primary, you would just get a desaturated Magenta. Although I do believe that, as I have said before, the 'violet' we see on TV is probably just a desaturated magenta. We just forgot what real violet actually looks like 'cos we always see it on TV. (Just like so many people have forgotten what real cooking tastes like!)

 

[sophiecentaur]

Gor blimey strike a light guv'nor. I just looked at the lower, expanded, CIE chart on that link and just look at the actual names of the colours. Take a look at where the name 'violet' appears. (Press the 'expand' button and it's all very readable). Well I never.

 

[I like Serena]

Gor blimey strike a light guv'nor. I just looked at the lower, expanded, CIE chart on that link and just look at the actual names of the colours. Take a look at where the name 'violet' appears. (Press the 'expand' button and it's all very readable). Well I never.

 

[DaveC426913]

At one time, all such information would cost you money. You had to buy books for it or go to a library. There is no justification for complaining that it's not all freely available.

(Aside:
What kind of logic is that?? At one point, a fire department and health care would have cost you money too. This is the 21st century. Standard of lving means there are things we can expect, such as a rich electronic and mostly free cyberspace.)

 

[sophiecentaur]

That's a fair point but the fact that a lot of information is 'free' has nothing to do with morals and rights. It has to do with fact that it is not actually free. The bits of cyberspace that you think you are getting for nothing are funded by the fact that someone pays for the privilege of being able to target you with advertisements. You will notice that the adverts on this very forum reflect the content of the posts.
The difference between 'free' internetting and 'free' fire and health services is that we pay up-front with taxes for one and the other cost is a stealth cost. We have no rights associated with the latter because he who pays the piper calls the tune.
I pay for an email service and do not have to put up with all the garbage on hotmail etc.. If my service is interrupted, then have the right to complain and I do. If hotmail stops for a few hours - only the advertisers can complain.
A subscription service for information can be expected to be better and more useful and exclusive. No one can complain about that.

 

[schip666!]

Not sure I agree with your there, actually. If you added Green primary, you would just get a desaturated Magenta. Although I do believe that, as I have said before, the 'violet' we see on TV is probably just a desaturated magenta. We just forgot what real violet actually looks like 'cos we always see it on TV. (Just like so many people have forgotten what real cooking tastes like!)

That's what I meant. Given our "current understanding" of color -- or colour for thouse of you reading along in europe -- adding green to magenta just desaturates it. If our violet sense is from slightly different responses in red and green cones _all in the same decreasing direction_ it implies (again modulo my simplistic understanding) that one needs green to see violet. However, if there is a red-bump-out-there-in-the-blue we get violet in a way that is congruent with our CIE color matching model.

That fact that they call it "violet" on the chart is part of the language problem. I still want to have a good look at a real color spectrum from a prism or suchlike to remind myself what violet really looks like after all these years of RGB screens. Will probably have to wait for a visit to a science museum or something -- and they may have gotten it wrong anyway.

<RANT>
To continue beating the information availability horse... It's worse then previously described. Most journals actually CHARGE the authors for publication as well as extracting free labor in the form of peer-review, both in exchange for the prestige of appearing in their august pub. Authors retain some copyright so they can put the paper on their website but usually can't print the thing elsewhere. Using the pay-per-view model journals end up enforcing the stratification of disciplines. As in my IEEE society example, if one only does computer-vision one can probably just subscribe to a couple journals and get all one needs. But if you cross a disciplinary boundary you end up hunting through multiple pubs, each for a price, and much of the material you pay for is not relevant...Just one of my a little high-horses.
</RANT>

 

[I like Serena]

That fact that they call it "violet" on the chart is part of the language problem.

Let's not call it "violet", since that is a very ambiguous color (as you can find http://en.wikipedia.org/wiki/List_of_colors#V").
Usually it has red in it, but it can have many different colors.

Let's call it "spectrum violet", so we know what we're talking about.

 

[schip666!]

Let's call it "spectrum violet", so we know what we're talking about.

OK by me. The next time someone actually SEES spectrum violet, report your findings back to us.

 

[sophiecentaur]

There's an evolutionary aspect to all this. There would be little point in an early homo saps having a separate brain pigeon hole for a spectral colour that it doesn't meet very often. The only spectral violet likely to be seen would be in very desaturated form in a rainbow (sitting 'inboard' of blue, not beyond it).Would this deserve a special 'colour' sensation? I don't think so. It would make good design sense to lump it in with some other bluish, desaturated colours. Otoh, we are particularly discriminated about the so-called flesh tones because the colours of our faces are a good way to assess our mood.

 

[I like Serena]

Well, I have seen spectrum red near infrared, which is easy with all the infrared LEDs lying around.
However, IMO this red has a definite reddish look and feel to it.
Where's the difference?

 

[DaveC426913]

Well, I have seen spectrum red near infrared, which is easy with all the infrared LEDs lying around.

Don't forget that bell curve thing works both ways. Some erstwhile "infra-red" LEDs may emit into the far visible red.

 

[DaveC426913]

There's an evolutionary aspect to all this. There would be little point in an early homo saps having a separate brain pigeon hole for a spectral colour that it doesn't meet very often. The only spectral violet likely to be seen would be in very desaturated form in a rainbow (sitting 'inboard' of blue, not beyond it).Would this deserve a special 'colour' sensation? I don't think so. It would make good design sense to lump it in with some other bluish, desaturated colours. Otoh, we are particularly discriminated about the so-called flesh tones because the colours of our faces are a good way to assess our mood.

Why is everyone assuming there is little of this spectral violet around? What makes you think all of nature does not freely reflect in the violet range? It happily reflects in the UV (as witnessed in flowers by birds and bees), so...

 

[schip666!]

There's an evolutionary aspect to all this...Otoh, we are particularly discriminated about the so-called flesh tones because the colours of our faces are a good way to assess our mood.

I can't think of a good evolutionary reason for my hypothesized red-bump-in-the-blue which supposedly supplies color discrimination into the ultra-violet. If it really exists it might be a chemical accident in the red cone. But a nice accidental color none-the-less...

I believe we are most sensitive to shades of green, supposedly to distinguish food sources. Any red gradation sensitivity for facial recognition would have to be a much later development since a) most faces were not "pink-white" to start with; and, b) the social necessity of mood recognition is also a recent thing. Chimps for instance are all about body-language, not subtle facial expression.

But this is all off topic, no? Can we hijack the thread?

 

[I like Serena]

Why is everyone assuming there is little of this spectral violet around? What makes you think all of nature does not freely reflect in the violet range? It happily reflects in the UV (as witnessed in flowers by birds and bees), so...

That's because we don't *know* this is spectrum violet.
We see it, but we can't discern it. Especially since it will be mixed with other colors.

Don't forget that bell curve thing works both ways. Some erstwhile "infra-red" LEDs may emit into the far visible red.

I worked on a machine that relied heavily on infrared leds.
Some people could see whether they were on or off, but I couldn't.
I blamed my eyes for not going into the infrared far enough.
That is, until I took my glasses off!
Then I could *just* see it.

 

[sophiecentaur]

Why is everyone assuming there is little of this spectral violet around? What makes you think all of nature does not freely reflect in the violet range? It happily reflects in the UV (as witnessed in flowers by birds and bees), so...

There IS very little 'spectral' violet around. The violet we see is very de-saturated and nowhere near the envelope of that CIE chart. A rainbow would be the only natural occurrence with loads of scattered white light mixed in. There is a significant difference between the shape of the curved sections of the chart around blue and around red. I think this is probably because there are actually more natural saturated red sources in life (red hot objects at night, for instance).

 

[DaveC426913]

There IS very little 'spectral' violet around.

I don't understand by you say this. Every white object is reflecting loads of violet. Just because we don't see it or can't discern it doesn't mean it's not there.

 

[sophiecentaur]

A component of a perceived colour is not that colour. Of course ther will be some very short wavelengths in all the colours we see when outside in the day. But our eye is not a spectrometer. All we can be aware of is the combination of the outputs of our three sensors at once and we cannot discriminate closer than that. It wasn't until Newton started using prisms that anyone ever experienced near-pure spectral violet entering their eyes. Evolution had already provided us with a mechanism fir interpreting this which is not 'adequate' to analyse the new phenomenon accurately.

 

[Drakkith]

I think sophie is saying that while violet light IS in white light, you won't see the violet because of the mix of other colors causing you to perceive white. In other cases even if you have violet you would probably also have red or green or blue, causing you to not perceive the spectral violet by itself, but as another color.

 

[sophiecentaur]

@Dave
Pinky or black skins actually have very similar chrominance values. The difference is largely related to luminance.

 

[DaveC426913]

A component of a perceived colour is not that colour. Of course ther will be some very short wavelengths in all the colours we see when outside in the day. But our eye is not a spectrometer. All we can be aware of is the combination of the outputs of our three sensors at once and we cannot discriminate closer than that. It wasn't until Newton started using prisms that anyone ever experienced near-pure spectral violet entering their eyes. Evolution had already provided us with a mechanism fir interpreting this which is not 'adequate' to analyse the new phenomenon accurately.

Ok, so you're saying our colour perception didn't evolve to be sensitive to violet, since pure violet (alone) is an uncommon occurrence in nature - unlike pure green and pure red, which are common occurences in their pure form in nature.

 

[sophiecentaur]

I think you are getting near. There is no essential difference between the way we respond to spectral violet (400nm) and spectral green (550nm or so) etc., just the three parameters we get. We very seldom, if ever, see any monochromatic light in nature. When watching colours on a TV system there are never any monochromatic 'colours' produced. A monochromatic phosphor isn't very bright so afaik they aren't used.