There are also differences between "additive" and "subtractive" color models.

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

Good explanation

Dave T2

There are two different sets of primary colors, one is for transmitted light, like that from a television or computer monitor, and one set for reflected light, like from a sheet of white paper. One set is additive and one is subtractive.

If you start with black, like a TV screen and add color, then the primary colors are red, blue, green. Full-up R,G & B makes white. R,G & B at zero make black (since that's where you started.)

If you start with a sheet of white paper and full-spectrum white light and add color, you are subtracting from the white and the primary colors are cyan, red (actually magenta) and yellow. Full-up C,M & Y makes black because all colors from the white light get absorbed and nothing is reflected. Obviously, a white sheet of paper with no ink will be white because it reflects all the white light.

You'll notice that your color printer uses cyan, magenta and yellow ink. It also uses a black ink cartridge since adding C, M & Y doesn't actually produce the richest blacks and it uses a lot of ink. This is a practical matter. It is often referred to a the "four color process." Most color printing of any kind uses this. Some color inkjet printers will use two extra inks, light cyan and light magenta, for a six color process as using these two extra inks will allow it to print more subtle color variations which produces better prints of photographs.

John

I believe the extra inks are to extend the range (gamut) of colors that can be printed... not necessarily more subtle color variations.

I believe the extra inks are to extend the range (gamut) of colors that can be printed... not necessarily more subtle color variations.

From what I have seen in high end photographic printing, it's both.

And Canon's new ProGRAF printers have 12 ink cartridges!

To quote Ben Allen, "There are not seven colours in a rainbow and red, blue and yellow aren't the primary colours of anything".
Light is the same as sound, it's a continuous spectrum of wavelengths and there aren't three primary notes on a piano. It's only due to how our vision works that we can trick our eyes into seeing a gamut of colours by adding or subtracting using three colours.

Bob.

Yeah, that's very true. And it amazes me that despite all the anatomical & chemical differences that exist from one person to the next, that those same three primary colors work well enough for everyone. Why should those three colors be the same for me that they are for you?

Then again, maybe they're not the same. Maybe that's why some people see a television image looking very good while someone else looking at the same screen sees it looking bad.

Yeah, this is the sort of thing that used to keep me up all night wondering about it, until i finally got old enough to not really care anymore. ;)

Some great comments... thanks all... looks like the answer is "both" depending on subtractive vs additive... great wiki article, I showed it to my family, thanks for that...

Always gets me when something I've taken for granted for years, eg "primary colors are RGB" isn't getting taught to my kid, so I thought I'd ask you folks, the best-qualified in the world for video stuff, about all of it.

Thanks very much, now I can sleep at night, not worrying about whether I screwed up telling my kid that Red Green Blue are the primary colors... the correct answer is "both" depending on the situation...

As a visual person, I always encourage her to think in colors. What made the Biggest impact on me, was listening to Star Wars' producer George Lucas talk about color themes in his movies, which had completely escaped me, til I heard him talk about it, on the extended/bonus sections of the DVDs, eg:

1) Star Wars III: New Hope (original first one): mostly sand and browns, then black and whites...

2) Empire Strikes Back (IV, the best one imo): Lots of white... hoth.. mainly bright colors... whites, vivid..

3) Return of the Jedi (VI, boo): Mostly greens, forests, lots of natural green colors..

So those distinctions, the color themes of the movies, that discussion has had a huge impact on me, listening to George Lucas.. and so I try to help my daughter visualize in color themes, too.. quizzing her on "ok what's the main theme of this building? movie? etc, red/green/blue.. discussing primaries a lot..."

anyways, thanks very much - great insights from everyone!

may the RGB force be with you,

-k

> To quote Ben Allen, "There are not seven colours in a rainbow and
> red, blue and yellow aren't the primary colours of anything".

I disagree with what I think is the underlying point here. : ) It's a very poetic sentiment but it doesn't tell you much if you're trying to reproduce color.

There are three primary colors of light, red, green, and blue, simply because that's how the coders in our eyes work. The number of colors in the rainbow doesn't matter here.

Also: Red, green, and blue are properly considered primary colors (of additive color) because there is just one way to see them: by exposing your retina to light of the appropriate wavelength. Yellow and cyan are secondary colors (of additive color) because there are two ways to see them: e.g. yellow either by mixing red and green light, OR by using yellow, monochromatic light. SImilarly for cyan.

But you can't mix any two colors and come up with "red". You have to start with red. That's why red is primary.

Magenta is strange in that it is not a spectral color: There is no single wavelength that will cause your eyes to tell your brain "magenta". (To put it another way: It isn't in the rainbow.) The only way to see magenta is by mixing red and blue. (Or, in subtractive color, by subtracting green.)

Conversely, indigo and (true) violet are spectral colors that can't be simulated by adding other colors together. They don't look anything like magenta, and additive color systems like color TV and computer screens can't reproduce them. (And color TV cameras won't pick them up except as "dark blue", so there's no point in the TV screens trying!) They look sort of "bluer than blue," that is, a blue that is more deeply saturated than even 100% saturated blue.

What's the eye doing with these, when it has no coder for "violet"? I think what is going on here is that the spectral response of the green coders slops over into the blue a lot, so the green coders respond some to true blue (so true blue never looks as saturated as, say, green can), but not as much to indigo and even less to violet.

These are often (mis)represented in online color charts by using dark purple, but true indigo and violet don't look like that at all; there is no reddish component in them whatsoever.

There is not much indigo or violet in sunlight and very little in artificial light, so we don't get to see these very often. So we're not missing much by leaving them out of color TV.

The true primary colors of paints and pigments -- subtractive color -- are indeed magenta, yellow, and cyan, simply because these are complementary to the primary colors of light (additive color). That's how subtractive color works. These are taught as "red, yellow, and blue" in elementary schools, but that's an approximation. You could pick three others, but you wouldn't have as wide a gamut.

As a practical matter: The primary colors are the ones which, when mixed in various proportions, will give you the widest color gamut! A lot of practical research has gone into this. The choice for "IRE blue", for example, is not necessarily the peak of the eye's blue response curve. It was a compromise between usefulness (which wavelength gives us the widest gamut?), availability and brightness of phosphors, many other things.

Did you know that for a long time -- through, I think, the mid-60s -- color TV sets weren't displaying true IRE red? the only phosphors that could do it were very dim, and oculdn't keep up with the green and blue. Only professional monitors had true IRE red, and they had to be viewed in subdued lighting (not just for this reason, of course). Then rare earth phosphors were developed that were a lot closer to the IRE colors. "Rare earth phosphors" were a huge deal when they first came out in consumer sets.

A different way of thinking about it is:

The eye has S, M, and L cones that determine color perception (if there is enough light / for photopic vision).

The S cones are most sensitive to blue/violet light, M to green light, and L to red light.

Any particular wavelength of light will cause some response in all three types of cones. This overlap means that for 'perfect' color reproduction, you need a lot of colors (each distinct wavelength gets its own color). This will let you reproduce all possible colors / would reproduce the whole possible gamut of colors.
In practice that isn't really practical.

To reproduce most colors, some additive mixture of red green and blue will work.
*If it were possible to produce negative light, then 3 colors would be enough to make all colors. But monitors can't make negative light so that doesn't work.



AFAIK the IRE didn't define any standard primary chromaticities. It was the original NTSC committee that did that.

The original NTSC primaries actually called for a very large color gamut. However this approach was abandoned in an effort to make consumer TVs brighter. Later on Conrac monitors were very popular for monitoring... SMPTE adopted the primaries of those monitors and set the SMPTE C standard for colorimetry. The SMPTE C primaries/colors are a much smaller gamut compared to the original NTSC primaries.

Something else to consider:

Our visual system does not process color the same way a light meter would. I'd argue that our perception of "color" has some other dimensions to it... certain objects appear neon, metallic, etc. There are some perceived qualities/attributes to what we see... and that goes beyond the basic red / green / blue (or hue chroma brightness).

RYB was what they taught us in elementary school 40 years ago for use with crayons and tempera paints. Remember how yellow and blue crayons make green?

Rob Mack

Also some of the materials that create colours can fluoresce, we have a plant that flowers once a year and in the shade the flowers don't look that spectacular but when sunlight hits them they take on a whole new look, it's a red that defies description and all efforts to capture it on camera, either film or digital have failed me.

A small percentage of the population have more than 3 colour receptors and must therefore see more colours than the rest of us. It seems to me that vision is not our most refined sensory system, especially compared to our auditory system. Probably just as well too, if we could resolve the wavelenghts of light as well as we do sound building an image recording system would be centuries ahead of us.

I don't think what Ben Allen has to say is out of place, it's the subtext to his lecture on colour correction, maybe I should wait until I've heard the rest of what he has to say before commenting too much. I think where he's coming from is that we should understand what it is we're really working with before we dive into the deep end.

I guess I should also mention colours like IKB, there's one of Kleins works in our state art gallery and again it's a colour that I think defies all efforts to capture it on film or display it on any monitor, you just have to go see the real thing. As Glenn has said part of the magic would defy any instrument, by suspending pigments in resin you have a depth of colour within the object that cannot be captured in a flat medium.

Bob.

"certain objects appear neon, metallic"

. . and then, there are the words we use to convey just WHAT we are talking about.

Is yellow still yellow to someone who IS colour blind? It maybe pink? If there were enough "colour-blind" yellowists around would pink needed to be renamed as yellow? When does an accepted "norm" appear to be the truth?

Grazie

Most languages are set up to describe visual images, I can simply say "red" or "blue" and most of us have that common reference. No language of which I am aware has a lot of words to accurately describe sound. Read some of the high-end "audiophile" magazines and look at all the essentially meaningless words they use to describe what they think they're hearing.

Back when I was a recording engineer, a guitar player might say he wanted more "crunch" on his guitar, but his idea of what "crunch" sounded like and my idea were probably completely different. (Plus there was no knob labeled "crunch" on my mixer.) The best we could do was for him to play a CD which had the sound that he was after and once we had a common reference, I could attempt to replicate the sound. It sure wasn't as easy as saying "make it more blue."

John

> Our visual system does not process color the same way a light
> meter would. I'd argue that our perception of "color" has some
> other dimensions to it... certain objects appear neon, metallic, etc.
> There are some perceived qualities/attributes to what we see...
> and that goes beyond the basic red / green / blue (or hue chroma
> brightness).

I believe it was established (in the early 80s.. at least, that's when I remember seeing the article in Science News :) ) by CGI researchers that the difference between metallic, plastic, marble, etc., surfaces had to do with how different layers just *below* the surface color, diffuse, etc., the light. Before this discovery, everything was based on modeling just the top surface, and everything looked "plastic" or near to it. Sub-surface modeling was the big breakthrough in, well, "surface" modeling. :) Somehow the visual system sees the net result of all of this and figures out "that's metal" vs. "that's water" vs. "that's glass", etc.

All of this is true and I'd add water and human flesh. These seem to be the two holy grails of CGI and I've yet to see CGI water that's convincing. Since some of the recent work on adding fluid dynamics it flows better but still isn't convincing enough.

I think perhaps the real problem with specular light and sub surfaces is that we see the image as is a flat plane. Our eyes and heads are not static, locked in a vice, when we look at the real world. When we look at a reflective surface we expect the reflections to move as our eyes and heads move. This behaviour would be extremely difficult to reproduce in a monitor or on a cinema screen. Even3D imaging probably misses out on capturing this.

Bob.

I had a college classmate who had a t-shirt that was a fluorescent yellow color. Even though the yellow pigment must have absorbed some light, it still seemed like it was brighter than a white shirt would have been. It seemed to actually glow with it's own light source.

She would wear it to breakfast on exam days so that when she took off her jacket the blinding yellow would wake us all up! It sure worked.

Well, maybe the fact that it was her wearing it helped all us male engineering students wake up too.