For those who don't know this, a brief explanation as to why colours are
treated differently as ink on paper as compared to pixels on screens.

On the screen, each pixel is a light source (or in the case of LCD, a light
filter). It has an emission (or transmission) spectrum which looks like
this (where the X axis is decreasing wavelength and the Y axis is amplitude,
typically from zero to 255 for each channel):


            /\
           /  \
          /    \
----------      --------------------
           Red  Green  Blue

                  /\
                 /  \
                /    \
----------------      --------------
           Red  Green  Blue

                        /\
                       /  \
                      /    \
----------------------      --------
           Red  Green  Blue

However, we note that paper is a reflective, not emissive, medium. When you
print out black text, although the colour value is triple-zero on a 24-bit
display, it is C=0,M=0,Y=0,K=1 on the page. Similarly, when you print out
red pixels, you do so with ink that absorbs everything other than red light
- which gets reflected.

----------      --------------------
          \    /
           \  /
            \/

           Red  Green  Blue

----------------      --------------
                \    /
                 \  /
                  \/

           Red  Green  Blue

----------------------      --------
                      \    /
                       \  /
                        \/

           Red  Green  Blue

So, if we just had R, G and B inks, we would need to use equal parts of all
three to get black, and none of each to get white.

However, it gets more complicated than this. Firstly, no-one uses R, G and B
inks, rather colours are generated using cyan, magenta and yellow ink.
Secondly, colour ink is expensive, so better printers tend to use a separate
black cartridge. This is invoked when all colours are present in equal
quantities (i.e. the image is grayscale). Since most pictures are stored in
RGB format it is necessary to turn one into the other. Fortunately this is
as easy since CMY is the complement of RGB:

[C]   [1 - R]
[M] = [1 - G]
[Y]   [1 - B]

Also, if R == G == B, then we don't need to perform this operation at all -
simply use the K (black) ink at the given intensity.

This is what percentcolour uses to calculate the percentage of a page which
is C, Y, M and K.

Some printers also use additional colours - newer Epson printers use 5
colour and one black ink (I think there are two `light' versions of the
primary complimentary colours, but I can't remember which). As to how to
bill that - I have a very large Bucket of Vyno.
