How does colour addition / subtraction work?

[dotancohen]

I am familiar with the ideas of colour subtraction and addition, such as adding blue and yellow to make green. However, how does this work from an electromagnetic standpoint? Is there basis for the addition or subtraction, or averages, of the wavelengths or frequencies? Is the colour change a phenomenon of the physiology of the human eye (implying that different animals will experience colour addition differently). All resources that I find on Google refer to the mixing of dyes, none explain the phenomenon on the basis of each colour's wavelength.

Thanks!

 

[Bloodthunder]

Multiple waves of different wavelength, all of which are picked up by our eyes.

 

[dotancohen]

‎Thanks, Bloodthunder. So if a blue dye reflects only light at 455nm and a yellow dye only reflects light at 575nm, then the resulting green light is not 532nm (which is conveniently close to the average) but rather _two_ wavelengths of 575nm and 455nm that our eye perceives exactly as it would a single 532nm wavelength?

 

[Andy Resnick]

Color vision is not completely understood. The anatomy is well-known: cells in the retina have rods or cones (cones are generally confined to the fovea, while rods are in the periphery), and there are three forms of opsin which have broad overlaps in spectral sensitivity, and thus allow for precise color discrimination:

http://en.wikipedia.org/wiki/Opsin

However, the origin of some colors- purple, brown, pastel hues, for example- is not known (AFAIK).

Shining two colors of light on a photodetector can, under some conditions, result in color mixing as well- this is 'additive' mixing. Color mixing from dyes and paints are different- that is known as 'subtractive' mixing.

http://en.wikipedia.org/wiki/Subtractive_color
http://en.wikipedia.org/wiki/Additive_color

 

[sophiecentaur]

"So if a blue dye reflects only light at 455nm and a yellow dye only reflects light at 575nm, then the resulting green light is not 532nm (which is conveniently close to the average) but rather _two_ wavelengths of 575nm and 455nm that our eye perceives exactly as it would a single 532nm wavelength?"

No no no. If one dye absorbs all wavelengths but one and another absorbs all wavelengths but another then Nothing will be reflected --- Black will be the result. Colour dyes used in printing do not absorb just one narrow band of wavelengths. They are, essentially, very broad band absorbers. You just see the result of what they have all let through.

What you call colour addition and subtraction is really a totally subjective thing. A colour is something 'in your head'.
You can't change the spectrum of the light that comes from a coloured object (the wavelengths and intensities of the light from it). What you can do, though, is to use the way that our eyes and brains appreciate what we call Colour and fool them into 'seeing' a particular colour by providing them with a so-called metameric match, using a suitable mix of other colours.
You can do the mixing by adding light from three different sources (the Red Green Blue 'primaries' of colour TV) or by using layers of dyes or pigments to absorb light (the Yellow Cyan Magenta primaries in colour printing).
Additive mixing is much easier to understand and easier to 'get right' in practice, if you only have three primaries available. Get that sorted out in your head first -I recommend.
This thread - https://www.physicsforums.com/showthread.php?t=499677" is a recent one and may be worth while going through.

 

[sophiecentaur]

However, the origin of some colors- purple, brown, pastel hues, for example- is not known (AFAIK).

I'm not sure I'd agreed with that although you may want to enlarge on that bald statement, in which case I may agree. However, the tristimulus theory of colour vision is pretty well established, based on the three broadband analyses of an 'average' eye. The 'non-spectral' colours (and also, of course, the desaturated 'pastels' are the result of a more complex spectrum that cannot be matched against one particular spectral wavelength.

The perception of these browns, pastels and purples involves stimulation of all three sensors at the appropriate ratios and actual levels. For instance, you can produce a very convincing brown colour using a very low luminance (brightness) yellow. How low? It depends on the general ambient light level because the brain does the best it can by auto exposure and auto colour balancing systems that leave yer average camera standing. (We use a huge amount of knowledge and context which a poor old electronic system couldn't hope to do.) DYK that the ratio of RGB signals needed to produce a convincing brown / black skin tone is very near that required to produce a fair. European skin tone? Not surprising really, I suppose because all skin pigments are produced by the same chemical - just in different densities.

 

[obafgkmrns]

Color perception really is largely in your head. It depends not only on the wavelengths emanating from an object, but also on wavelengths emanating from adjacent objects. Here's a link to a great demonstration of that effect. Go to illusion #47 at the bottom of the page:

http://www.ukpuzzle.com/puzzles.htm

(#14 is a pretty interesting demonstration of color perception, as well.)

 

[sophiecentaur]

TOTALLY in your head, actually.

 

[Khashishi]

Read this website. http://www.handprint.com/HP/WCL/wcolor.html
In it's own words: "Here is the most comprehensive discussion for artists of color perception, color psychology, "color theory" and color mixing available online, and one of the most comprehensive available anywhere in any format."
It doesn't lie!

 

[GRB 080319B]

As sophiecentaur indicated, color perception is a subjective synthesis of different wavelengths of light absorbed by the retina. The cone cells responsible for human color vision are (peak wavelengths given): Short (420-440 nm), Medium (534-545 nm), and Long (564-580 nm). I believe the color perceived depends on how stimulated one type of cone cell is by light in the respective range relative to the others through the http://en.wikipedia.org/wiki/Opponent_process_of_color_vision" . I don't think the color perceived is a simple average of wavelengths absorbed. The color perceived can depend on the wavelengths as well as the intensities of the light, among other factors.

This page describes why colors like brown and pink are artificial/subjective (i.e., do not correspond to a single wavelength in the em spectrum): http://www.Newton.dep.anl.gov/askasci/phy99/phy99125.htm"

 

[sophiecentaur]

It (that last link) certainly is pretty comprehensive, from an artist's point of view- which is who it is aimed at.
It seems to be angled at the print (pigment based) media rather than the additive (phosphor based) media. Which is what it was produced for, of course. The ancient (1950s and 1960s) writing about colour TV development is still more relevant to understanding additive (light ) mixing, I think.
It makes good reading though and the question arises about how 'extra' colour channels for colour TV could help. Three colour printing is absolute rubbish, in comparison with RGB TV.

 

[sophiecentaur]

As sophiecentaur indicated, color perception is a subjective synthesis of different wavelengths of light absorbed by the retina. The cone cells responsible for human color vision are: Short (420-440 nm), Medium (534-545 nm), and Long (564-580 nm). I believe the color perceived depends on how stimulated one type of cone cell is by light in the respective range relative to the others through the http://en.wikipedia.org/wiki/Opponent_process_of_color_vision" . I don't think the color perceived is a simple average of wavelengths absorbed. The color perceived can depend on the wavelengths as well as the intensities of the light, among other factors.

This page describes why colors like brown and pink are artificial/subjective (i.e., do not correspond to a single wavelength in the em spectrum): http://www.Newton.dep.anl.gov/askasci/phy99/phy99125.htm"

The sensitivity range of the three sensors is way beyond those limits. The system could not work at all with such narrow band filtering.

 

[GRB 080319B]

The sensitivity range of the three sensors is way beyond those limits. The system could not work at all with such narrow band filtering.

Sorry, I forgot to label those as http://en.wikipedia.org/wiki/Cone_cell#Types".

 

[sophiecentaur]

Fair enough - but is central to the whole system.
People are so wedded to thinking simple RGB analysis and they don't realize that it just couldn't work like that.

 

[Andy Resnick]

I'm not sure I'd agreed with that although you may want to enlarge on that bald statement, in which case I may agree. However, the tristimulus theory of colour vision is pretty well established, based on the three broadband analyses of an 'average' eye. The 'non-spectral' colours (and also, of course, the desaturated 'pastels' are the result of a more complex spectrum that cannot be matched against one particular spectral wavelength.

The perception of these browns, pastels and purples involves stimulation of all three sensors at the appropriate ratios and actual levels. For instance, you can produce a very convincing brown colour using a very low luminance (brightness) yellow. How low? It depends on the general ambient light level because the brain does the best it can by auto exposure and auto colour balancing systems that leave yer average camera standing. (We use a huge amount of knowledge and context which a poor old electronic system couldn't hope to do.) DYK that the ratio of RGB signals needed to produce a convincing brown / black skin tone is very near that required to produce a fair. European skin tone? Not surprising really, I suppose because all skin pigments are produced by the same chemical - just in different densities.

As you pointed out, color vision is intimately tied into brain function- it should not be surprising then, that our brains can manufacture colors that do not physically exist:

http://en.wikipedia.org/wiki/Impossible_color
http://en.wikipedia.org/wiki/Imaginary_color

And from http://www.wdwradio.com/2007/09/epcot-trivia-general/ :

'After a great deal of research by Disney Imagineers and Kodak photography experts, the pavement in Epcot was painted a specific, custom pink color, in order to make the grass look greener and pictures look brighter.'

 

[sophiecentaur]

As you pointed out, color vision is intimately tied into brain function- it should not be surprising then, that our brains can manufacture colors that do not physically exist:

We are clearly at odds here. Only the brain 'manufactures' colours because colours are the interpretation of the spectrum of the light that comes into the eye. Colours do not 'exist' in any other way. Colour is our response to a set of stimuli - which are not colour themselves.
To say otherwise is like saying that the adjectives we use to describe something are actual, physical entities. A 'pretty' ornament is 'pretty' only because the observer thinks it so - the prettiness is not a physical quantity in the same way that its mass, temperature or hardness are.

Furthermore, 'Colour' is a subset of all the possible spectral shapes that can exist because there are many alternative spectral shapes between which we cannot distinguish. Our eyes are not spectrometers.

Sorry to bat on about this but it is an important distinction which many people just seem to ignore.

 

[Drakkith]

So your saying that the perception of color is ONLY in the brain because color does not exist elsewhere. A photon at X frequency isn't that color photon, it is simply a photon of X frequency that your brain may interpret as a color once received.

 

[sophiecentaur]

Of course.
As I said before, your eye / brain is not a spectrometer; it only does its best to catagorise the combination of three sensor outputs and decides that this combination matches some other input that its memory contains. "That's pale blueish green", you say to yourself.

 

[Andy Resnick]

Of course.
As I said before, your eye / brain is not a spectrometer; it only does its best to catagorise the combination of three sensor outputs and decides that this combination matches some other input that its memory contains. "That's pale blueish green", you say to yourself.

But that's being silly, or at least pedantic. HeNe laser light is the color red- the word 'red' is sufficiently ill-defined to encompass a whole range of hues and saturations, but nobody should claim that HeNe light has no color because color is a fabrication of our brain.

We agree color vision is not completely understood. Hopefully we also agree that HeNe light is red and that a clear daytime sky is blue.

 

[sophiecentaur]

It's a good thing that pedantry is noT dead.
;-)

The light from a HeN laser shares its redness with an infinite number of other combinations of more than one wavelengths. I would say that makes its colour a very imprecise description of it; particularly as no two people could rely on agreeing about its precise colour. As far as I can see, that makes its colour a very different attribute from the wavelength. The colour is essentially subjective. A race of dog Scientists could evolve a Science that mimics our own in most respects (they would have exactly the same E=hf as us, for instance) but their 'colours' would not map onto our own at all.
What 'colour' does a Bee see when it approaches a flower with a UV reflective 'landing strip' on it?

 

[Andy Resnick]

Sure- and when I speak with a metrologist or spectroscopist, I provide the detailed spectral distribution function.

I couldn't tell you what a bee, dog, or any other person sees.

 

[sophiecentaur]

Sure- and when I speak with a metrologist or spectroscopist, I provide the detailed spectral distribution function.

I couldn't tell you what a bee, dog, or any other person sees.

OK. Here's another angle. Does the electronics of an audio amplifier produce 'Music' or does it produce a varying electrical signal / sound waves? And would that 'Music' necessarily be universally acknowledged to be music or described in the same way by everyone who hears it? A loudspeaker produces Sound, which we interpret in different ways. Who, on these forums, would be discussing Music when they are talking the Science associated with Sound?

Does a camera lens designer concern himself with the aesthetics of the pictures or with the performance of the lens in terms of MDF, flare and aberrations? A camera is completely blind to the 'meaning' of any photograph it records.

Likewise, a source of light produces light of different wavelengths which we interpret, in different ways, and may talk of in terms of Colour for aesthetic purposes or for shorthand description

Why treat Colour differently from Music? They are both totally subjective sensations.

You can't say what 'colour' any other creature sees so why use the term in the context of Science and measurement - which, surely, is what these forums are about? It's far too vague a term and using the word Colour can only cloud the issue, scientifically.

 

[Drakkith]

I can see both points of view here. Calling something "red" or "blue" means that when we look at it our brains will see it that color. I don't think it is necessary to explain that the light is a certain wavelength that our brains will interpret as color. MUCH easier to just call something blue.

 

[sophiecentaur]

. . .
. I don't think it is necessary to explain that the light is a certain wavelength that our brains will interpret as color. MUCH easier to just call something blue.

But, don't you see? It's not just "one wavelength" that we see as a particular colour. All but the totally saturated spectral colours consist of light with a whole range of wavelengths. Furthermore, the same perceived colour can be made up of an infinite range of mixes of different wavelengths. This is my whole point and people seem to be missing the significance of the distinction. Does no one who reads this understand what I'g getting at? Sob sob.
I really am not a loony. Please read some colourimetry, folks.
Honest.

 

[Drakkith]

But, don't you see? It's not just "one wavelength" that we see as a particular colour. All but the totally saturated spectral colours consist of light with a whole range of wavelengths. Furthermore, the same perceived colour can be made up of an infinite range of mixes of different wavelengths. This is my whole point and people seem to be missing the significance of the distinction. Does no one who reads this understand what I'g getting at? Sob sob.
I really am not a loony. Please read some colourimetry, folks.
Honest.

I understand perfectly what you are saying. And you are correct as far as I know. I think the argument here is over semantics or whatever.

 

[sophiecentaur]

It's only the same 'semantics' that draw a distinction between Momentum and Mass or between Current and Power.
I am surprised that this particular distinction isn't taken more seriously.
But perhaps you need to have worked on it?

 

[Drakkith]

It's only the same 'semantics' that draw a distinction between Momentum and Mass or between Current and Power.
I am surprised that this particular distinction isn't taken more seriously.
But perhaps you need to have worked on it?

I think you are being a little ridiculous actually lol. Everyone here knows and agrees with the fact that our perception of color comes directly from the processing of input in the eye and brain. Someone that is red-green color blind WILL perceive color differently than most.

I'm not sure what you mean by me needing to work on whatever "it" is, but ok.

 

[sophiecentaur]

Suffice it to say that all wavelengths (visible) have a colour but most colours do not consist of only one wavelength. That's my problem.
I was referring to colourimetry as "it". :-)