Color addition involving wavelengths

[Vacrin]

recently i have been thinking about how colors add when you have 2 different light sources of color and put the streams together, such as taking red and green beams and crossing them to make yellow. i was looking at wavelengths to determine if there was a mathematical way to say what color would be produced by the streams of light. however i ran into a problem when i looked at the red+blue = magenta (purple ish). most colors i can see average instead of add together.

red and green added together produce yellow. which if you take the wavelengths of red and green, add them, and divide by 2, produce yellows wavelength.

blue and green added together produce cyan, and again adding the wavelengths and dividing produce cyan's wavelength.

however adding blue and red wavelengths let's say (440nm and 640nm respectively) = 1080/2 = 540 which is green, and that is not possible. the wavelength for magenta is about 390-405nm (roughly) can anyone help with this problem?

 

[sophiecentaur]

Colour and wavelength are two very different things. Wavelength describes one single spectral line of light. Colour is the eye's response to a combination of one or more wavelengths. You cannot 'mix' wavelengths and produce another, different wavelength. You can give the eye a mixture of two Colours and produce a Third, percieved, Colour. It is very rare to see pure, 'spectral' lines in real life. Even what comes out of a prism is usually polluted by ambient light, despite it's very spectacular apearance.
My Mantra of Colourimetry is "The Eye is not a Spectrometer".

If you are referring to the colour mixing you get in TV, the phosphors (RG and B) are far from monochromatic - there is no need for them to be and you can often get far more light power out of a non-monochromatic source.

 

[Vacrin]

okay, so the color that i see from the two different sources is still red and blue, or still red and green, but the way i perceive the mixed wavelengths as being magenta or yellow respectively. so there is no mixing of wavelengths between the source lights and the back ground it is hitting?

 

[sophiecentaur]

okay, so the color that i see from the two different sources is still red and blue, or still red and green, but the way i perceive the mixed wavelengths as being magenta or yellow respectively. so there is no mixing of wavelengths between the source lights and the back ground it is hitting?

When you have a linear medium (that's pretty well every common substance you will come across) the light of any particular wavelength that reflects back from a surface will only have its amplitude (brightness) changed. Other light energy will have absolutely no effect on it. Two light beams will interact with the surface in a way that is totally independent of each other. Any perception of a 'resultant' colour is all inside your head, as your brain makes the best it can from the three sets of signals it gets from the three wavelength selective sensors.
Our ability to distinguish wavelengths of light is extremely 'crude' compared with our perception of sound, which involves a cochlea, consisting of hundreds of sensors which are each sensitive to a very narrow band of frequencies. Our Ears really are like audio spectrometers!

If you had a radio receiver as unselective as your eye's colour sensors, you would be picking up all the VHF FM stations at once and you could not make any sense of what came out of the loudspeaker. You could possibly be able to tell whether there were more transmissions in the higher, lower or middle parts of the band - that's all. Out colour vision has evolved to be 'just good enough' to give us some advantage to survive, just like all our other body functions.

One point worth making:
If there happens to be a high enough density of light energy arriving so that the surface heats up then, of course, you can get thermal radiation off the surface (e.g. it can get red hot) but that is another issue.

 

[Vacrin]

hmmm, makes me wonder if we would see the same kind of colors if our eyes adapted to different wavelengths, perhaps in the infrared or xray, I am sure there would be different colors and different things would be seen, but what of the colors... anyways thank you guys for the information, i wish it worked differently, but i guess not :/

hearing is a very interesting thing, we like perfect even number notes and frequencies, if it is slightly off it sounds completely wrong to us. maybe that is way music affects us so much more than paintings. well atleast for me it does. that's another matter though

 

[haruspex]

okay, so the color that i see from the two different sources is still red and blue, or still red and green, but the way i perceive the mixed wavelengths as being magenta or yellow respectively. so there is no mixing of wavelengths between the source lights and the back ground it is hitting?

Quite so.
The best way to understand it is by knowing the physiology of the eye. You have rod cells, detecting light level, and three types of cone cells for detecting colours. Each cone type has its own stimulus/response curve across the visible spectrum. Crudely, they're known as red, green and blue cones, because that's where the three curves peak, but there's considerable overlap.
Given a single incoming frequency, the four types of detector give their four different responses. After normalising for light level, that's three degrees of freedom for colour. The brain interprets a given combination of these three numbers as a particular hue.
Now suppose we have a mix of incoming frequencies from the same spot in the visual field. Each results in a set of three response numbers, and these sets just get averaged out to produce one set. So pure yellow light results in perception of a yellow colour, but a suitable mix of other frequencies can result in exactly the same set of numbers, so you perceive the identical colour.
It will not be as simple as averaging the wavelengths of the individual frequencies because the cones' responses are nonlinear.
Note that most response triads cannot be produced by a single incoming frequency, so there are many colours not in the spectrum. Also, if you could directly stimulate individual cones you could produce the perception of colours that "don't exist", in the sense that no combination of incoming light could produce them.

 

[Vacrin]

Quite so.
Note that most response triads cannot be produced by a single incoming frequency, so there are many colours not in the spectrum. Also, if you could directly stimulate individual cones you could produce the perception of colours that "don't exist", in the sense that no combination of incoming light could produce them.

Now that is a very interesting idea, to produce a color in the brain that cannot be produced by any natural mean. i wish that i could see this color, it would be quite spectacular, or maybe quite the opposite.
it might have to be fine tuned to each person, given that we all see a bit differently, although we don't really know that because of how we were raised. perhaps though a wavelength of light could be fined tuned to someones given rods so that it only stimulates one of the rods

 

[sophiecentaur]

Now that is a very interesting idea, to produce a color in the brain that cannot be produced by any natural mean. i wish that i could see this color, it would be quite spectacular, or maybe quite the opposite.
it might have to be fine tuned to each person, given that we all see a bit differently, although we don't really know that because of how we were raised. perhaps though a wavelength of light could be fined tuned to someones given rods so that it only stimulates one of the rods

I don't quite know what you mean by that but some colours - such as the colours of oil films on water and the metallic appearance of insect bodies and bird feathers are very striking because they are 'unfamiliar'. They are not produced by pigments but work on interference. Also, the 'dayglo' colours are particularly striking because the dyes convert UV light and converts it to visible light and the colours produced are not 'natural' combinations of wavelengths that you could normally get from a surface that was just reflecting straight sunlight. You get 'whiter than white', for instance.
Imo, there is not much point in discussing what a different set of sensitivity curves would produce because it would really be 'anybody's guess' how we would perceive things. All animals have different sensitivity to colours, of course, because they will have evolved with different requirements.

 

[haruspex]

I don't quite know what you mean by that but some colours - such as the colours of oil films on water and the metallic appearance of insect bodies and bird feathers are very striking because they are 'unfamiliar'. They are not produced by pigments but work on interference. Also, the 'dayglo' colours are particularly striking because the dyes convert UV light and converts it to visible light and the colours produced are not 'natural' combinations of wavelengths that you could normally get from a surface that was just reflecting straight sunlight. You get 'whiter than white', for instance.

Those still consist of a mix of photons of whatever frequencies. They appear striking mostly, I suspect, for the reason you give: they are brighter than they 'should' be for the general level of illumination.
Stimulation of M ('green') cones only cannot be achieved that way since the L cones respond to all the same frequencies (but differently). See http://en.wikipedia.org/wiki/File:Cone-fundamentals-with-srgb-spectrum.svg.
I find yellow intriguing; we all agree it is qualitatively different from red and green, but in sliding from green to blue everything looks sort of green and/or sort of blue.

 

[sophiecentaur]

Those still consist of a mix of photons of whatever frequencies. They appear striking mostly, I suspect, for the reason you give: they are brighter than they 'should' be for the general level of illumination.
Stimulation of M ('green') cones only cannot be achieved that way since the L cones respond to all the same frequencies (but differently). See http://en.wikipedia.org/wiki/File:Cone-fundamentals-with-srgb-spectrum.svg.
I find yellow intriguing; we all agree it is qualitatively different from red and green, but in sliding from green to blue everything looks sort of green and/or sort of blue.

I would rather put it that the metameric match with spectral yellow by using red and green is
a qualitative judgement, where there is a difference in the quantitative situation (i.e the spectra of light involved)

Also, the detailed response curves for M and L cones are (necessarily) different, although there is a overlap. What is your point about "Stimulation of M ('green'). . . . . ."? The perceived 'colour' is still the result of the relative output levels from the three L,M and S sensors. The 'auto colour balance', in our brains, decides that a certain illuminant is there and assumes that reflectivity cannot be greater than unity at any particular frequency. When the evidence it gets, with the reflection from dayglo surface, conflicts with this, no wonder the colour stands out! It just shouldn't be there (not from the past millions of years of evolutionary experience). No wonder we are confused.

 

[haruspex]

I would rather put it that the metameric match with spectral yellow by using red and green is
a qualitative judgement, where there is a difference in the quantitative situation (i.e the spectra of light involved)

I agree with that, but don't understand what it has to do with anything I wrote. If it was apropos of my remark about yellow, I meant that we find it counter-intuitive that the colour yellow can be built from red and green light, but perfectly reasonable that cyan can be built from blue and green. Why the perceptual difference?

Also, the detailed response curves for M and L cones are (necessarily) different, although there is a overlap. What is your point about "Stimulation of M ('green'). . . . . ."? The perceived 'colour' is still the result of the relative output levels from the three L,M and S sensors.

Sure, but some combinations of output ("M only" is one example) may be impossible to achieve through photonic^* input. Other examples would be cone output with no rod output. This would be a kind of super-daylight sensation, in the sense that the converse arises naturally under low light conditions.

The 'auto colour balance', in our brains, decides that a certain illuminant is there and assumes that reflectivity cannot be greater than unity at any particular frequency. When the evidence it gets, with the reflection from dayglo surface, conflicts with this, no wonder the colour stands out! It just shouldn't be there (not from the past millions of years of evolutionary experience). No wonder we are confused.

Again, no disagreement there - except that as you previously noted there are natural examples, so it would not be counter to evolutionary experience.

*EDIT: That is, photonic input to a large enough region of the retina to encompass all receptor cell types.

 

[Vacrin]

only thing that i didnt understand is dayglo surface. what exactly is that? and how does it manage to convert a wavelength that we can't see to one that we can? wouldn't that have a energy dispersion and turn it to the next step down so it would appear blueish/violet?

 

[haruspex]

only thing that i didnt understand is dayglo surface. what exactly is that? and how does it manage to convert a wavelength that we can't see to one that we can? wouldn't that have a energy dispersion and turn it to the next step down so it would appear blueish/violet?

AFAIK, it doesn't necessarily mean just a slightly lower energy. It could absorb in UV and emit in green or red.

 

[Vacrin]

Well, that is true, your talking about electron absorption and emission right?

 

[sophiecentaur]

I agree with that, but don't understand what it has to do with anything I wrote. If it was apropos of my remark about yellow, I meant that we find it counter-intuitive that the colour yellow can be built from red and green light, but perfectly reasonable that cyan can be built from blue and green. Why the perceptual difference?

I think I misunderstood what you meant ("qualitative" for me was referring to the perception of colour - which is not directly quantitative; any actual electronic measurements of U and V are related to qualitative / physcometric results) But there is a great difference in principle with producing a match for a spectral colour and just producing a non spectral colour. Half the problems that people have with the 'Colour' thing is when they confuse colours with wavelengths and Yellow is an added complication. The CIE diagram should be required viewing for anyone who wants to make any comment at all on Colourimetry.

Sure, but some combinations of output ("M only" is one example) may be impossible to achieve through photonic^* input. Other examples would be cone output with no rod output. This would be a kind of super-daylight sensation, in the sense that the converse arises naturally under low light conditions.

I see no point in discussing low light conditions with respect to Colourimetry. By the time you get to low levels, your eye's colour system is broken and all bets are off.The three Cone response curves are only relevant when there is enough energy to stimulate them. It's only Luminance that counts in the 'dark'.

Again, no disagreement there - except that as you previously noted there are natural examples, so it would not be counter to evolutionary experience.

*EDIT: That is, photonic input to a large enough region of the retina to encompass all receptor cell types.

Are there any naturally occurring dayglo colours? Even if there were, could one say there was any evolutionary advantage in the brain being able to distinguish / deal with them? Bearing in mind the wide variation of colour perception ability from animal to animal, I reckon that our particular system must relate to our particular requirements - e.g. social interaction (facial expressions / blushing etc.) and spotting blood etc. whilst hunting - and those only require us to deal with pigments rather than other sources of colour. The fact that we are gobsmacked by a rainbow, which is a really washed out and feeble set of very non-spectral colours, supports this view, I think.

 

[haruspex]

I think I misunderstood what you meant ("qualitative" for me was referring to the perception of colour

Yes, that's how I was using it, but I was not referring to the fact that the sensation of yellow can be produced either by yellow light, as a single frequency, or by a mix of red and green lights. That's easily explained by the fact that the outputs from the M and L cones are the same in the two cases. Rather, I'm intrigued by the qualia - the higher level perception. When we look at cyan, whether by mix of blue and green light or by a single wavelength on the blue/green boundary, we perceive it as blueish/greenish. Yet we do not perceive yellow as greenish/reddish. Well, I don't.

The CIE diagram

Thanks for the link. 45 years ago I read Prof Richard Gregory's popular work Eye and Brain. It included response charts for the three cone types. Based on that I drew more-or-less that diagram, though mine was a bit more ear-shaped. Nice to know it has a name.
One thing could have been made a little clearer in the text, though. The three points labelled R, B and G are somewhat arbitrary. They could in principle have been anywhere in the area.

I see no point in discussing low light conditions with respect to Colourimetry.

Again, you misread me. I mentioned that as a kind of opposite to the theoretical possibility of signal from cones but not from rods.

Are there any naturally occurring dayglo colours?

Dayglo is not really a colour in the sense of the eye's response, is it? I thought it was just that the intensity was greater than the brain expects given the ambient light level. Scorpions glow green in UV. (I thought some flowers appeared brighter to our eyes in the presence of UV, but I can't find a reference, so maybe it's just that insects and birds can see them in UV.)

Even if there were, could one say there was any evolutionary advantage in the brain being able to distinguish / deal with them?

Only to the extent of not being spooked by them.

The fact that we are gobsmacked by a rainbow, which is a really washed out and feeble set of very non-spectral colours, supports this view, I think.

Rainbow non-spectral? How so?
Actually, there is a slight dayglo effect with rainbows, the sky above the main bow being darker than the bow itself.

 

[LURCH]

A small question on the side: do interferons ever produce perceived colors in a wavelength not physically present?

 

[sophiecentaur]

A small question on the side: do interferons ever produce perceived colors in a wavelength not physically present?

Proteins? Or is there another meaning of 'interferon'?

 

[sophiecentaur]

@haruspex
I think the difference for 'yellow' is that yellow does occur as a spectral colour and is identified by the brain as 'part of the rainbow' etc. etc.. The brain always seems to do its best to categorise things as simply as possible (this can be a real disadvantage in life) and it does that here. There is, of course, another example of this with the colour we call 'Violet'. It corresponds to a spectral lines but can be matched quite closely by adding R&B (with a bit of green). See where Violet turns up on the CIE diagram in the Wiki Article. It's the violet we see most, I think as the spectral violet is a very low luminance colour and will usually be drowned out by ambient light.
You query my statement about the rainbow not being a pure spectrum. Even when it show against a dark cloud there is a lot of additional 'whiteish' light, mixed in with it, so it will always be desaturated and sometimes very much so.
How would you achieve a situation where the the cones were selectively illuminated without the rods - are you suggesting crocodile clips on selected nerves?
The three points, R,G &B refer to the phosphors, used in a three colour display. They are, indeed, a bit arbitrary. Their choice is based on the availability of suitable, efficient phosphors or LEDs. There is a compromise between getting the primaries as widly spaced as possible (near the spectral curve) - to achieve a wide gamut of displayable colours - and to get them as bright as possible. As colour TV has developed, there have been a number of different phosphor sets to work to. Translating the signals from a three band sensor to any given set of RGB phosphors requires matrixing the chrominance vectors - and then there's the white point to consider too. Quite a nightmare and I was thrown head first into it all in late 1967, with only an Elliott 803 computer to help me. That accounts for my evangelic attitude to colour and the way many people mis-use it. (but not you, I think).
You may well be right about some dayglo effect from some flowers.
Must return to my painting (cupboards and not pictures).

 

[haruspex]

@haruspex
I think the difference for 'yellow' is that yellow does occur as a spectral colour and is identified by the brain as 'part of the rainbow' etc. etc..

That doesn't account for it. You can also find a spectral point between blue and green (which I'm calling cyan, but the name doesn't matter), yet we perceive it as a greenish blue or a blueish green. Likewise, although we give orange a unique name (in many languages), it's no surprise that it can be simulated by mixing red and yellow light; it looks sort of red and sort of yellow.

another example of this with the colour we call 'Violet'. It corresponds to a spectral lines but can be matched quite closely by adding R&B (with a bit of green).

That you consider that another example suggests to me you're still missing my point. It is a surprise if you're comparing a single spectral input with the perceived colour, but my puzzlement about yellow is in comparing the cell output with perception. As you say, the cell output for violet is simulable by a mix of red and blue inputs, and we perceive violet as a reddish blue, so there's no paradox in that case.

You query my statement about the rainbow not being a pure spectrum. Even when it show against a dark cloud there is a lot of additional 'whiteish' light, mixed in with it, so it will always be desaturated and sometimes very much so.

Yes, I realized later that's probably what you meant. But the eye and brain are so good at adjusting for white level (don't forget we have the rods to take into account) that the higher level perception is quite close to a genuine spectral experience.

How would you achieve a situation where the the cones were selectively illuminated without the rods

I was not suggesting any actual experiment.

I was thrown head first into it all in late 1967, with only an Elliott 803 computer to help me.

Well, well. I was programming an 803 at Elliott Bros' Rochester Airport factory in '68!

 

[sophiecentaur]

The reason that Yellow is so 'identifiable' may be because of the way that the primaries sit on the CIE colour space(see that Wiki article again. The line between B and G primaries is miles away from the spectral curve (compared with the line between R and G). Our colour discrimination is not uniform all over the chart. See the MacAdam Ellipse and it shows that, around the 'cyan' region, we are a lot more approximate than around the Yellow region. We are more sensitive to skin tones and really don't care too much about greeny blues and bluey greens as they don't affect our lives as much as the health, attractiveness or emotional state of our fellow humans - which are all signalled by skin tones. I remember seeing and using an alternative colour space diagram which is a distorted version of the CIE chart to give equal steps in perception for equal distances. I cannot find it or any reference to it in a brief search. Somewhere, I still have the famous BBC Engineering Monograph on Colourimetry . . . . . .
What happens outside the gamut of colours which a normal display is not usually a problem in our appreciation of the quality of colour reproduction but the limits are sometimes made very obvious when looking at the clothing of the crowd at a sporting event. There are a lot of brightly coloured jackets that seem to be the same selection of a few colours in strong sunlight. This is because highly saturated colours all end up bunched along the RG and GB lines (the limits of the triangle).
Just found that diagram. Uniform Chromaticity Chart.

803: 4k words of 39 bits each and a clock speed of a few kHz. EEh, they don't know they're born, these days.

 

[haruspex]

The reason that Yellow is so 'identifiable' may be because of the way that the primaries sit on the CIE colour space(see that Wiki article again. The line between B and G primaries is miles away from the spectral curve (compared with the line between R and G). Our colour discrimination is not uniform all over the chart. See the MacAdam Ellipse and it shows that, around the 'cyan' region, we are a lot more approximate than around the Yellow region. We are more sensitive to skin tones and really don't care too much about greeny blues and bluey greens as they don't affect our lives as much as the health, attractiveness or emotional state of our fellow humans - which are all signalled by skin tones.

Some interesting suggestions there - thanks.

 

[LURCH]

Proteins? Or is there another meaning of 'interferon'?

I'm talking about the way waves from two different sources can interfere with each other to produce a wave pattern not present in either of the original two sources (as in interferometry).

 

[sophiecentaur]

I'm talking about the way waves from two different sources can interfere with each other to produce a wave pattern not present in either of the original two sources (as in interferometry).

Two coherent sources of multi-coloured light? I'd need to have a bit more explanation before having an opinion about this. It doesn't seem to fit with the normal ideas of wave interference.

 

[Khashishi]

A very thorough web book on color: http://www.handprint.com/HP/WCL/wcolor.html

 

[sophiecentaur]

A very thorough web book on color: http://www.handprint.com/HP/WCL/wcolor.html

Colour yes. Colourimetry, no (from what I read of it). It seems to miss out the quantitative level which is essential if you want to predict and get things right.