From ucsd!ames!nrl-cmf!mailrus!tut.cis.ohio-state.edu!bloom-beacon!mit-eddie!uw-beaver!tektronix!tekcrl!tekgvs!mikec Mon Mar 28 11:39:27 PST 1988


George Drapeau recently asked:

 "Why is it that for video, the three primary colors are (usually) red, green,
and blue, whereas for painting, the primaries are red, *yellow*, and blue? [1]
If you look at a color wheel, red and green are at opposite ends, as opposed
to forming two parts of the primary triangle." [2]


To which Grady Ward responded:

 "The question is not so naive.  While frequency of light is a continuum, the
eye believe [sic] there are primary colors because of its own physiology, not
because colors exist external to it.  Any set of three colors can be "primary"
for non-color blind humans, [3] beacuse [sic], like a TV set, we have three
sensing rod varieties. [4] [5]

  The colors red, blue, green are only one example of primaries (ones which
require the least addition of white to make all the rest of the colors). [6]
Paint pigments combine in a subtractive way, though, and red blue yellow [1]
are usually chosen by people using pigments (paints) rather than light becasue
[sic] they, too, require the least addition of white to make all the rest of
the non-primary colors." [6]


To which Matthew Turk subsequently responds:

"> ...  Any set of three colors can be "primary" for non-
 > color blind humans, because, like a TV set, we have three sensing
 > rod varieties.

 Well, almost.  The three colors must be "perceptually independent",
 i.e. no mixture of two of them can make the third.  Also, in additive
 color the choice of the three primaries limits the gamut of colors
 that can be represented (to the inclusive triangle in the CIE 
 chromaticity diagram)." [3]


To which I feel hopelessly compelled (ACKK!!!) to correct:

[1].  Artists and color scientists do not use the same color naming convention.
    The most common color primaries for additive color systems (e.g. color CRT)
    are Red, Green, and Blue.  The most common color primaries for subtractive
    color systems (e.g. paint pigments) are Cyan, Magenta, and Yellow.  Artists
    are usually taught to call Cyan "blue", Magenta "red", and Yellow "yellow".
    (While one out of three may seem bad -- it could have been worse.)  Most of
    us (including me) were (mis-)taught the subtractive system in grade school.
    Thus, paint pigment primaries are really Cyan, Magenta, and Yellow.  Note
    that in color hardcopy (ink printing) black is also used as a "primary".

[2].  Pathetic teletype graphics of a standard color wheel:

                                       \  Red  /         Additive  Subtractive
                                        \     /
    Do not take the chirality   Magenta  \   /  Yellow     C=G+B      R=M+Y
     or the rotational phase    __________\ /__________    M=R+B      G=C+Y
    of this diagram literally.            / \              Y=R+G      B=C+M
                                  Blue   /   \  Green
                                        /     \
                                       / Cyan  \

     Note that Cyan (artist's Blue) is the complement of Red, Magenta (artist's
     Red) is the complement of Green, and Yellow is the complement of Blue.  It
     would seem that George Drapeau was referring to artist's Red (Magenta) and
     Green as being complementary (opposite on the color wheel) colors.  A less
     likely explanation is the NCS (Natural Color System) proposed in Sweden
     several years ago (where Red and Green are complementary, Yellow and Blue
     are complementary, and White and Black are complementary) but this is not
     in common use (generally only found in color perception research).  Since
     Yellow and Blue are complementary (diagonally opposite) colors no additive
     or subtractive combination can be used to generate a good Green or Cyan.
     Since Red and Yellow are additive and subtractive primaries, respectively,
     they make a poor choice, as a pair, for primaries in any color system.  A
     (bizarre) full color system using Red, Yellow, and Blue as primaries would
     only be viable if both additive and subtractive color processes were used.

[3].  Grady Ward's suggestion that `Any set of three colors can be "primary"'
    is inaccurate.  Matthew Turk's explanation clarifies this point quit well.

[4].  Color television sets, generally, convert non-visible energy to visible
    energy.  They have no color "sensing" capability.  There are exceptions to
    this in some color laboratories.  Grady Ward's analogy is poorly chosen,
    since a color CRT generally has 3 color emitters, while a human generally
    has 3 color sensors.  Not every human is a trichromat (three primary color
    sensors).  In addition to the obvious color blindness (monochromats and
    dichromats) a few people are tetrachromats (four primary color sensors).
    This is a good example of evolution (i.e. change) in action.

[5].  Retinal rods are part of the Scotopic visual system and do not take part
    in the discrimination of color.  Retinal cones, being part of the Photopic
    visual system, do discriminate colors.  Grady Ward's claim is "confused".

[6].  In any additive color system, adding white decreases the saturation of a
    color (i.e. decreases the radial distance from the equal energy point on a
    CIE diagram).  Grady Ward's definition of what constitutes a primary color
    is incorrect while Matthew Turk's definition is correct.  Subtractive color
    systems generally "subtract" light from the equal energy point by absorbing
    the complementary wavelengths (i.e. Cyan pigment absorbs Red wavelengths).

                                     Micheal Lucifer Cranford