Colors of light

Red, green and blue are not primary colors of light, but primary colors of our eyes. It's eye that has three types of color sensors with peak sensitivity at red, green and blue wavelength. Electromagnetic radiation by itself doesn't care about how humans detect it, but we do.

Electromagnetic radiation, commonly called light only in certain narrow range of frequencies, has continuous spectrum from zero Hz to infinity (a physicist will correct this statement if it's wrong).

There are two "systems" for color: additive and subtractive. Additive is adding light of different colors to make other colors; color TV is additive. The primaries for this are red, blue and green. Subtractive color is the color that is absorbed by a reflective surface; printed material is subtractive, and the primaries are magenta, yellow and cyan (NOT red yellow and blue!

"Exploring Colour Photography" by R. Hirsch; "Creative Colour" by F. Birren; available at Borders.

In colour photography, to use filters to their full potential, simply get rid of them. Why would you want to use filters to corrupt the natural colours of nature?

Peace.

You've already corrupted "The Colors Of Nature" by passing them through mulitcoated glass lenses and fixing them on a substrate using a photochemical process. Do you think you're a "Purist" by not using filters?

Alpha Omega probably thinks that photography is about replicating what his eyes see. Most good photographers think it's about interpreting what the eyes see to create a more pleasing image. The process of interpretation can be carried out without making the image seem unnatural. Alpha Omega is missing out on a lot.

Wayne, Dave, and Alpha Omega: Photography is a wonderful thing in that it can't be defined by one person or group of people. While it is fine to express your preferences or opinion, the "light modifiers" and "purists" should learn to express their ideas with belittling others or claiming superiority.

As in many things, there is more than one right answer.

light

The human eye sees electromagnetic radiation roughly in the range of wavelengths from 400 nanometers to 700nm.

400-500nm is blue light (below this is Ultraviolet UV)

500-600nm is green light

600-700nm is red light (beyond this is infrared IR)

When these three "primary colors" of light are in balance for the human eye, then we see white (B+G+R).

B+G (lacking red) appears cyan. G+R (lacking blue) appears yellow. B+R (lacking green) appears magenta. These are the "complementary colors."

In photographic materials which are designed to reproduce colors like the human eye sees, the layer sensitive to red light produces cyan dye, the layer sensitive to green light produces magenta dye, and the layer sensitive to blue light produces yellow dye. In negatives, this makes pretty good sense because the image turns out the complementary color to what the film saw. In slides (color reversal), the "negative" is developed first without color dyes, then the image is "reversed" and then the dyes are produced to make a positive image.

When I think of on-camera filtration, I like to consider what color the film won't be seeing due to filtration (that is, don't consider the color of the filter). To get an image that is lacking yellow, you need a blue filter. The filter appears blue because it passes all colors of light except yellow. So, if a scene is too yellow (tungsten lighting for example), you need a filter which appears blue to correct the tonality back to neutral (if so desired). Now, there are specific names for such filters, so don't get too caught up in the science.

<The filter appears blue because it passes all colors of light except yellow. >

Oops, my statement above is incorrect. The filter appears blue because it allows more blue light to pass through it than it does red or green light. My clue that I made an error was that there really isn't such thing as "yellow light."

This points out why I like to think in terms of primary colors of light. If you only have to think of 3 colors, the reasoning becomes a lot simpler.

Dan Sapper-- RGB monitors and film emulsion layers may not deal with yellow except as additive components of green and red, but to say that "there really isn't such thing as 'yellow light'" is not correct. A continuous spectrum of visible light gives us the Roy G. Biv sequence of how we interpret visible wavelengths, and the "y" for yellow is very real. In fact yellow is the predominant wavelength radiated by the surface of the sun or any "black body radiator" of that temperature. Granted that much of what we perceive as yellow both in natural and in synthetic materials does appear so in whole or in part as a combination of red and green wavelengths, we shouldn't say that pure yellow light does not exist.

No yellow light? Sure there is, at least as much as there is green, red or blue light. So called unique yellow is around 580 nm as a monochromatic light

Thanks for the great explanations on light. Now, what about light reflecting off paint/pigment on a surface? How does yellow become a primary color??? Subtractive?

Thanks Bruce. I just re-read your post and I see you already explained about subtractive colors. Should I blame this on my elementary school teachers that taught us about colors? Am I the only one who was told red, blue and yellow?

Should I blame this on my elementary school teachers that taught us about colors?

Yes.

Am I the only one who was told red, blue and yellow?

No.

Using the colors red, yellow, and blue as primaries was practiced years ago as a dumbed-down and incomplete explanation of the Munsell color system - literally "color theory for artists" (I mean the fine arts "oil on canvas" crowd). This was before the average 10 year old American kid had access to CMYK inkjet printers and 800MHz computers. Back then, it was too difficult to explain to the average layperson what cyan was and how to spell it so teachers bailed out and used simple terms for the expediency of teaching (and at the expense of accuracy).

If you're an adult, you've probably noticed that lots of things that teachers taught you in grade school were incomplete versions of The Truth^TM.

Amusingly, some old art books like Ralph Mayer's The Painter's Craft refer to cyan, magenta, and yellow as the primary components of additive color. Newer books often point out that the "red, yellow, blue" explanation is popularly and incorrectly practiced (e.g., Inglis and Luther's Video Engineering).

For more information on basic color theory in regard to photography, turn to chapter 15 of your copy of Photographic Materials and Processes. In my opinion, the book's 70+ pages on color are far better written than anything I've seen on the web about color theory. You do read books, don't you?

{{Can someone "shed some light" on this relationship of colors, and explain additive/subtractive priciples?}}

Jeff, let's say you had a white sheet of paper. If you poured cyan, magenta and yellow dye in the right quantities together on that white sheet of paper you'd get black reflected back. Those are the subtractive colors.

Let's say you had a black sheet of paper. If you poured red, green and blue dye together on that black sheet you'd get the color white reflected back. Those are the additive colors.

{{Why would you want to use filters to corrupt the natural colours of nature? }}

Well, when I took the dichroic filters out of my enlarer to be in peace with the universe, I got these nasty, blue and orange looking prints. Otherwise, I agree that photogs that shoot Velvia with a color enhancing filter need to switch to Sanka.

{{Back then, it was too difficult to explain to the average layperson what cyan was and how}}

Darn, and I thought the color wheel model was also proof of a closed universe and grand unified field theory. Thanks for busting my bubble :^)

<Forgive me for asking such a fundamental question>

I have just finished reading a 200 page book on Light and Vision (the old Time-Life series), and can assure you that while the question is fundamental, the answers took more than 300 years to solve (from Isaac Newton in the 17th century, till the 20th century molecular biology).

<but even though I have been told, and accepted the fact, that the primary colors of light are red, blue and green >

Light is part of the continuous spectrum of electromagnetic radiation. The fact that humans see 3 primary colors is a function of the 3 rhodopsin pigments in the eye, and is not inherent in the spectrum of light itself. A different multispectral system was used for Landsat images, and had to be converted to false colors for display on maps.

The brain fills in the rest of the colors to create the appearance of a continuous spectrum. The red and green wavelengths preceived by the eye are closer than Dan Sapper indicates. About 8% of males have difficulty distinguishing red from green (color-blind).

A rainbow or prism is the best example of a continuous spectrum. If the light source changes (e.g. gets redder at sunset), the appearance of the rainbow will change (the blue end vanishes).

Not all light sources have the same quality or color balance. Incandescent lights (tungsten) are yellowish, and needs a slight blue filter when used with ordinary film. Fluorescent lights and low pressure vapor lamps (mercury and sodium) emit a non-continuous spectrum which doesn't match either the eye well, and is hard to correct for film. Pure sky light tends to blue, and needs a slight red (warming) filter to look natural.

<the primary colors of paints are red, blue and yellow>

The following is from the Paintshop Pro help file: "The CMYK model, a subtractive color model, is based on light being absorbed and reflected by paint and ink. This model is often used when printing. The primary colors, cyan, magenta, and yellow, are mixed to produce the other colors. When all three are combined, they produce black. Because impurities in the ink make it difficult to produce a true black, a fourth color, black (the K), is added when printing."

<Is there a website that shows visual examples of combined colors of light including complementary colors?>

As Sean indicates you would be well off with a good book that shows spectral diagrams and color examples. Your image editor should allow you to play with a color chooser, and also convert images to negative which will gives you the complementary colors.

<To use filters to their full potential>

Filters absorb light implying subtractive terminology. They can be of any color, not limited to the primary subtractive colors above. The fraction of light absorbed by a filter can be either narrow (a small part of the spectrum) or wide (most of the spectrum). If you stack more than 2 different narrow-pass filters together you will see darkness, since each will absorb most of the light and let nothing through.

A red filter will make both sky and grass dark (on B&W film). An orange filter will make the sky dark, but not the grass. Color negative film has an orange base, so the result has a blue cast color when turned into a postive (converted to the complimentary colours). Filtering is necessary, and can be done digitally, which has an effect equivalent to an optical filter.

<and for photography in general, I need to "get the big picture" about colors.>

The eye tends to judge colors by comparing them to adjacent areas, while film does not. The eye is also biased by the light source and tends to compensate, while film does not. The eye is more sensitive to certain wavelengths (green and red), and only perceives color when the light source is bright enough (not in darkness at night), while film undegoes color shifts with long exposures (reciprocity failure).

The RGB scheme is used in computer monitors, TV tubes, transparencies, scanning, imaging and digital cameras. The CMY(K) scheme is used in paints, inks, pigments and negative film. Other 3 color schemes include HSL (hue, saturation, luminance) for image editing, YCC for Jpeg compression, and NTSC and PAL for TV transmision. All of these are derivatives of the RGB scheme.

Just a few notes/followups on Gordon's comments. The colorblindness gene is carried on the 23rd pair of chromosomes; it is why colorblindness is far more common in males than females. Also, the human eye is far more sensitive to changes in luminance (brightness) rather than chrominance (color).

Your eyes do in fact adjust to the prevalent light. One experiment: stare at a blank piece of paper or a white wall for a couple of minutes, then look at your computer screen. It should appear unnaturally blue.

An example of blatantly artificial representation of the electromagnetic spectrum can be seen with color infrared film. Colors in the visible spectrum are pretty much randomly assigned to varying frequencies in the film's recording spectrum. There are more subtle inaccuracies in the way color film records what we see. People who photograph flowers will notice that certain colors simply are recorded incorrectly. If you've watched some nature programs on TV, you might have run across an episode that shows how bees might see the world differently (due to pronounced sensitivity to the UV spectrum).

If you're confused by RGB, "red, yellow, blue", CMYK, you need a modern tutorial, especially if digital imaging is something you wish to learn. It's a tall task to explain digital imaging accurately without being able to look at a CIE chromacity chart.

To "get the big picture" about most things, a book is typically the best starting point. This includes organization of information, presentation thereof, overall writing quality, and efficiency in which it is delivered. Also nice are indexes, glossaries, and bibliographies - all things that the average Web article sadly lacks. Good luck.

>>…but even though I have been told, and accepted the fact, that the primary colors of light are red, blue and green( and the primary colors of paints are red, blue and yellow) I still feel like I'm "in the dark"…<<

Jeffrey, actually there is no single set of "THE" 3 primaries for human viewing. The only real rule for selecting a (minimum) set of primaries is that no one of the colors can be matched by mixing the other two. That is, each of the 3 colors is unique. Apparently, every physically possible set of primaries has problems in that there are some colors that cannot be properly duplicated. The red, green and blue primaries are apparently able to handle a greater range of colors than nearly any other set. As an example, the "Feynman Lectures on Physics" (1963, FWIW), 35-3, indicates that the red, blue and yellow set has difficulty in producing a good green. A brief quote from same: "In elementary books they (the primary colors) are said to be red, green and blue, but that is merely because with these a wider range of colors is available without minus signs for some of the combinations."

So your elementary school teacher was right about red, yellow and blue being primary colors. It's just that they are not the ONLY SET of primaries.

To learn more about filter response, etc, I concur on the recommendations for (Basic) Photographic Materials and Processes. Ultimately, I think filter effects can only be well understood on a spectral (not RGB, etc) basis. But one can also learn to use filters very effectively without having a deep understanding; manufacturer's (professional) film data sheets generally give a variety of filter recommendations under various conditions. Also, books like "Kodak Photographic Filter Handbook" (publication B-3) contains a lot of specific filter data.

Scott Eaton wrote

>>Jeff, let's say you had a white sheet of paper. If you poured cyan, magenta and yellow dye in the right quantities together on that white sheet of paper you'd get black reflected back. Those are the subtractive colors. Let's say you had a black sheet of paper. If you poured red, green and blue dye together on that black sheet you'd get the color white reflected back. Those are the additive colors.<<

Actually you'd get a dark muddy color - any time you use pigments/dye in that manner it is an example of subtractive colors. They would be additive if you head red green and blue colored lights.

Gordon Richardson wrote: >>...but even though I have been told, and accepted the fact, that the primary colors of light are red, blue and green

Actually there a large number of sets of 3 primaries, not just red, blue and green. In part, I believe the use of those primaries, and sometimes yellow, is that those colors are psychologically unique in the sense that you can perceive unique yellow, red, blue and green colors as opposed to for example purple which is both red and blue.

>>Light is part of the continuous spectrum of electromagnetic radiation. The fact that humans see 3 primary colors is a function of the 3 rhodopsin pigments in the eye

There is only 1 rhodopsin photopigment (in rods) and 3 cone photopigments.

>>The eye is more sensitive to certain wavelengths (green and red), and only perceives color when the light source is bright enough (not in darkness at night), while film undegoes color shifts with long exposures (reciprocity failure).

Interestingly the eye undergoes perceived "brightness" shifts at different amounts of illumination. E.g blue green objects in the dark become relatively darker compared to yellow/orange objects as teh illumination is increased. (Although the blue green will only appear grey!)

Sean Yamamoto writes:

>>Also, the human eye is far more sensitive to changes in luminance (brightness) rather than chrominance (color).

This seems to me to besort of like apples and oranges or like saying smell is more sensitive than hearing. Luminance changes would be measure in quanta (ergs, whatever...) while changes in color would be presumable measured in wavelength (wavenumber, Hz, whatever) and don't see

Jeffrey, Here is a great web site that explains the functions of color: http://home.att.net/~B-P.TRUSCIO/COLOR.htm Note that combining colors with light is far different than combining colors of paint! The illustrations in the web site will help much. As you will see, when you combine red light, green light, and blue light, you get white light! If you take a triple exposure, where you put on a different one of these three filters over the lens for each shot; things that were stationary in the photo will appear their natural color, because you have combined these primary colors which will equal white. Things that were moving in your photograph, such as clouds, water, leaves, etc., will be either red, green, blue, or a rainbow of colors where they overlapped as they moved through the three exposures.

It is a bit misleading to say there are many sets of primary colors.

First, one has to accept that primary colors are only defined because of the structure of the retina. There are three precise wavelengths to which cones are sensitive: "pure" red, "pure" blue and "pure" green. There are two ways thus of generating colors and this defines two sets of primary colors. No more.

The first way is to shine into the eye three lights (red, blue and green) of varying intensity. If they all have the same intensity, we see white, or gray... or black (all turned off). This is why it is called additive: add them all up and you get the brightest color of all. Look very closely at your screen and you'll have the most easily reacheable example of additive colors.

The second way is to shine a white light into the eye and remove the unwanted primary (red, blue or green). To remove a color, you need a filter of the opposite color. The opposite of a primary color is formed by adding the other two primary colors. For instance, if you want to filter out red, you create a filter that has the color of blue and green together (i.e., cyan). Do that for the three additive primaries and you have three filters: cyan, magenta (red and blue, filters green out) and yellow (red and green, filters blue out). So if you put together these three filters, nothing goes through (which is why it is called subtractive: once you've removed everything, there is nothing left). This is what happens when you use paint or ink.

I would be very interested to see what other set of colors allow the full spectrum of colors to be reproduced.

My primary school teacher taught me that ;-)

Regards,

Francis

There are three precise wavelengths to which cones are sensitive

No, each type of cones has a spectral sensitivity curve (you can look at typical curves in Color Vision article). In other words, they are sensitive to a range of wavelengths.

I would be very interested to see what other set of colors allow the full spectrum of colors to be reproduced

No set of primary colors can reproduce all colors. If I recall correctly, if you put your three primary colors on the CIE Color Chart, the range of colors reproducible with this set is constrained to the triangle with your primary colors at its apices. As you may see, the choice of primary colors close to red, green and blue helps to reproduce more colors, but still not all them. In fact, the range of colors reproducible by real systems (TVs, monitors, color printing, etc.) is not a triangle and covers less than half of the area on the chart, due to various real-world imperfections.

OK, this is from the Websters dictionary:

--primary colors 1 The principal colors at which blue light is separated by a prizm; the colors of the rainbow. 2 The colors red, yellow, green and blue, by mixing which any desired color or hue may be obtained; to these white and black may be added.

I wish I had my TV design handbook with me, but it's pretty much unreachable now. I might got something wrong without it.

Vadim is exactly right regarding the primaries (assuming three are used). Except to point out that I think use of three non-interacting primaries will produce a GENERALLY triangular shaped gamut outline on the CIE "shark fin" chart. Photo paper dyes should be substantially non-interacting.

But back to the filter-on-camera issue: the CIE stuff and red, green, blue, etc, is based on HUMAN PERCEPTION and is not necessarily how film sees it. To repeat from my first post, "Ultimately, I think filter effects can only be well understood on a spectral (not RGB, etc) basis.".

Vadim quoted a dictionary definition for primary colors. To go a step further, it might be educational for interested people to now try to find a legitimate, unambiguous definition in physically measurable terms of what produces the colors cited; ie, red, green , blue and yellow. As far as I know, such definitions do not exist. I mean in terms of such and such a percentage of power in the wavelength range such and such. Until one actually goes on a search for such a definition, I think they tend to presume that one exists.

Still following this thread? here is another tack on the subject of light, color and how it relates to digital and analog photography: http://www.plumeltd.com/lightsources.htm rel="nofollow"> A white paper on Continuous Light Technologies for Digital Still and Video Work compiled Fall 1995. An oldy but a goody and a still easily understood approach to this subject as it relates to imaging.

>>No set of primary colors can reproduce all colors.

This is not true. In fact the the CIE chromaticity diagram which was shown above was produced precisely by using 3 spectral primaries to match a 4th test light. The realization that 3 primaries (a bad choice of word since it, incorrectly I think, implies some sort of uniqueness) was critical in the development of the theory that 3 cone photopigments underly human color vision. I think part of the confusion about whether or not 3 primaries can be used to match a 4th arises from a misunderstanding of how matches are generally made when obtaining something like the CIE chromaticity diagram. Since this is getting sort of far afield I'll just reference a pretty good, dated, intro text to color vision: Human Color Vision by RM Boynton. This even has an appendix on the history of the CIE diagram for thos who must really know.

Just to make things even more confusing, the "primaries" in the CIE diagram are not even physically realizable and were obtained by extrapolation from the actual data obtained using, obviously, physical stimuli. In other words you can transform the CIE chart into a large number of other charts based on different sets of primaries.

>>Vadim quoted a dictionary definition for primary colors. To go a step further, it might be educational for interested people to now try to find a legitimate, unambiguous definition in physically measurable terms of what produces the colors cited; ie, red, green , blue and yellow.

First you'd have to define what red,blue,green and yellow mean. There are a wide range of stimuli which people would call "red" "green" etc. As such there will be an even larger set of physical stimuli which would produce a "red". Even if you specified, e.g., a "unique" yellow (that is it has no red, nor green, nor blue in it - only yellow) which for a monochromatic light is around 580 nm, there are many, many mixtures of light which can match it exactly. That is sort of what the CIE chromaticity diagram allows you to do although you need to work backwards to get back to the physical properties of the lights (i.e. the amount of energy).

produced precisely by using 3 spectral primaries to match a 4th test light... and I think they had to add one of the primary lights to the test light for some ranges of color, which equals to "negative" intensity of that primary. If you don't do that, match is impossible.

I must confess that I was unaware of the excellent series of articles on Spectral Selectivity by Ed Scott and Hollis Bewley. (The title is slightly obscure, and is hard to find in the Learn menu). This has many excellent diagrams, and will answer most of the original questions in a much clearer way than I was able to.

>>produced precisely by using 3 spectral primaries to match a 4th test light

>... and I think they had to add one of the primary lights to the test light for some ranges of color, which equals to "negative" intensity of that primary. If you don't do that, match is impossible.

That's correct. Color matches generally follow the rules of algebra - multiplicative, additive, associative adn distributive. E.g. if you have a match and you add a color to both matches they will still match.

I meant, when they already turned off one of the spectral primaries and still didn't get match, they had to move it to the side of the test light and add some of that primary light to the test light.

>>I meant, when they already turned off one of the spectral primaries and still didn't get match, they had to move it to the side of the test light and add some of that primary light to the test light.

I know what you meant.

I said that by using 3 spectral primaries you could match a 4th test light. In color matching studies in which you match lighted patches you can add a primary to either patch for a match to obtain. The law of trichromacy states, among other things, that given 4 lights T, A, B, C there exist scalars a, b, c such that T = aA + bB + cC where = means "matches" here. For a, b or c < 0 obvisouly, since you cannot have negative light, in the experiment you need to *add* the light to the test patch. This is consistent with trichromacy, the existene of 3 cone photopigments and the CIE chromaticity diagram and has been known for > 100 years.

A problem would arise in non experimental situations such as a tv screen where there is no physical test patch ( I suppose the test patch is in the mind of the viewer) and you have 3 colored phosphors. Because of that arrangement there will no easy way to add the negative light. Striclty speaking tho' you could add "negative" light to the 3 primary side by preadapting the viewers eye by preexposing it to the "negative" primary. Obviously not practical, nor easy even theoretically since the adaptation will be time varying, bbut this would allow a match without adding light to the test patch

>> In fact the the CIE chromaticity diagram which was shown above was produced precisely by using 3 spectral primaries to match a 4th test light. <<

No, no, no! The shark fin part of the diagram has nothing to do with three primaries. It is simply a plot with every "visible" wavelength around the curved part of the perimeter. It theoretically contains every color that can be seen by a human since every visible wavelength is shown. It is on an XY grid which is CIE's 1931 Yxy color space.

>> Just to make things even more confusing, the "primaries" in the CIE diagram are not even physically realizable. <<

The "primaries" only come into play when a color-matching function is defined. There can be any number of such primaries. Two common examples I have seen in books are: 1) the CIE 1931 standard colorimetric observer: this set of color matching functions have no negative lobes and are based on "imaginary" primaries (called XYZ) which are (as Edwin said) "not even physically realizable". 2) second "common" example uses REAL monochromatic primaries (denoted by script RGB?) at 435.8nm, 546.1nm and 700.0 nm. The color matching functions for this set of primaries DO have negative lobes, indicating some colors (roughly saturated greens and blues) cannot be matched. Unless, as Vadim pointed out, one "adds" the appropriate color to the reference sample.

Either set uses the same graph (supplied by Vadim) to designate its chromaticities.

>> That is sort of what the CIE chromaticity diagram allows you to do although you need to work backwards to get back to the physical properties of the lights (i.e. the amount of energy)<<

No, no, no. You CANNOT work backwards to get the physical properties of the light. All you can do is determine whether another set of stimuli matches the color for an (average) human observer. As Edwin indicated, there are typically many combinations of primaries that can match a given color specification on the CIE chart. Given that, one would NOT be able to tell specifically which (if any) of those combinations was the actual physical light. These are so called metameric matches and have a lot to do with the situation where a color photographs way different than how we saw it.

Sorry about the long off topic post, but I've got into a situation where I'm not willing to let some things go unanswered.

Finally, this colorimetric stuff relates to human vision, only. If a film system does not exactly match human visual response (and currently, they do NOT), the CIE colorimetric system does not apply. As I said in an earlier post, "Ultimately, I think filter effects can only be well understood on a spectral (not RGB, etc) basis".

>> In fact the the CIE chromaticity diagram which was shown above was produced precisely by using 3 spectral primaries to match a 4th test light. << >>>No, no, no! The shark fin part of the diagram has nothing to do with three primaries.

YEs, yes, yes it absolutely has to do with primaries. It is generated by usgin 3 primaries (e.g. 435.8, 546.1 and 700 nm) to produce colormatching functinos. These are tehn transformed into a chromaticity diagram, the precise shape of which (the shark-fin shape) is *exactly* determined by the colormatching functions. The CIE diagram is one of many which can be generated.

>>>It is simply a plot with every "visible" wavelength around the curved part of the perimeter.

No it is not simply a plot of wavelengths... Rather, the coordinate for each wavelength specifies the ratio of each of the primaries used in deriving the diagram such that the sum equals 1.0. For example lets say that for 500 nm x=0.05, y = 0.6 then z =0.35 (zis not shown in these diagrams since z can be derived by z= 1.0 -(x+y). So if this were not a transformed diagram such as is the CIE this would give you exactly the ratio of, say, the 435.8, 546.1 and 700 nm light needed to make a match.

>>The "primaries" only come into play when a color-matching function is defined.

The point was that simply 3 primaries were used to derive the entire CIE chormaticity diagram, which to beat a dead horse, was obtained from real colormatching functions using 3 real spectral stimuli. It was then, for several reasons, transformed into the CIE space.

>>>There can be any number of such primaries. Two common examples I have seen in books are: 1) the CIE 1931 standard colorimetric observer: this set of color matching functions have no negative lobes and are based on "imaginary" primaries (called XYZ) which are (as Edwin said) "not even physically realizable".

Obtained useing 3 physically realizable, spectral primaries.

>>>2) second "common" example uses REAL monochromatic primaries (denoted by script RGB?) at 435.8nm, 546.1nm and 700.0 nm. The color matching functions for this set of primaries DO have negative lobes, indicating some colors (roughly saturated greens and blues) cannot be matched.

As of course did the color matching functions which were the basis of the CIE chromaticity diagram.

>> That is sort of what the CIE chromaticity diagram allows you to do although you need to work backwards to get back to the physical properties of the lights (i.e. the amount of energy)<< >>>No, no, no. You CANNOT work backwards to get the physical properties of the light. All you can do is determine whether another set of stimuli matches the color for an (average) human observer.

This is dead wrong. (Note I'm not saying that a person with nothing more than the CIE daigram can do so, nor am I saying that is is straightforward to do so). The CIE diagram gives the the absolute ratios of the imaginary primaries. The imaginary primaies were generated by transforming coordinates from color matching functions using real stimuli.

>>>As Edwin indicated, there are typically many combinations of primaries that can match a given color specification on the CIE chart. Given that, one would NOT be able to tell specifically which (if any) of those combinations was the actual physical light.

I said nothing about trying to decide which if any was the actual physical light (I'm not even sure what that means). I said "Even if you specified [ a light] there are many, many mixtures of light which can match it exactly." Obviously it follows that thtere is not a unique 1-to-1 mapping rather it is many-to-1.

My point was simply that given the CIE coordinates this minimally gives you the values of the primaries **in that coordinate system**. If you are tied to thtat system you can derive values for the color matching functinos used to derive the CIE diagram using a series of published papers in the 1920's and 1930's which specif the stimuli used and the methhods of transforming coordinates. Will the results be unique? No, I've already said as much.

I won't be posting any more on this (the beaten dead horses are pounding on my door) since it's way off topic/relevance, bbut may respond to email.

>> I won't be posting any more on this (the beaten dead horses are pounding on my door) since it's way off topic/relevance, bbut may respond to email. <<

OK, I will do likewise after two last paragraphs. I still disagree with most points of contention; the last post has not changed any of my views except to wonder if semantics are playing a part here. I will say that there is a big difference between the GRAPH shown and color matching functions (I concur that the matching functions themselves originate from physically real primaries and that transforms can create CMFs for imaginary primaries). However all the CMFs will continue to work the same in the total absence of that graph.

Also, that this does not apply to conventional color photography; the most glamorous simple proof I can think of is to photograph something like "morning glory" or "bluebell" flowers on several films and see how they compare to your visual (with unchanged chromaticity coordinate) interpretation. Or for a more technical bent, try to work out the spectral makeup of a complicated filter from the chromaticity coordinates (both available in Kodak's filter handbook).. I don't believe it's possible.

Bill C & Edwin - If you're taking the remainder of your um. . .discussion. . .off the forum, would you mind posting a summary of your consensus (once all is said and done)? Now I'm curious to hear the final word.

Thanks for all the responses, I certainly got more than I expected! I can't say I understand all of this, but I am getting "the big picture" and most of us seem to agree on some basic facts: 1. Film is not the same as the human eye, and we are constantly reminded of this in photography. 2. Light (and "colors" of) is defined in terms of wave and quantum theories (including measuring frequencies, and how the eye reacts to stimuli at these frequencies), but we may never truly understand it's nature. 3. When we "create" a relationship between colors (frequencies) of light, it is based partly on the application (photography, painting, printing, TV screen, ect) and not just "pure physics". Thanks again, and please don't bang your heads too hard over this.

so may different responses for this subject. this article may be of assistance:

Basic Color Theory