cs781 - Colour for Computer Graphics - Winter 2012

Course Notes

Lecture 8 - Characterizing an Additive Device

And Now from our Sponsor

A Generic Additive Device

These assumptions must be checked and the size of deviation quantified.

We are, of course, going to calibrate using physical measurements., In terms of typical physical measurements the colour matches might be equivalent to

Possible Defects of Generic Devices

Tristimulus Values

What do We Measure

  1. Chromaticities of the phosphors
    1. Maybe you get them from the manufacturer
    2. More likely, and better, you measure them at several values, turning on only one primary.
      • If they are not constant, you probably have a background to subtract
  2. The value of gamma
    1. Turn on each primary in turn to many levels
      • If you notice clipping at high and/or low values, which is likely,
        • censor the data,
        • note the true range of useful values of the primary.
    2. Measure Y
    3. Plot log(Y) against log(R[GB])
    4. It should be a straight line, the slope of which is gamma
  3. The normalization constants
    1. Turn on the three primaries together to get white
    2. Measure the chromaticity of the display, which is its colour temperature
    3. Measure the luminance and calculate X, Y, Z
    4. Calculate the inverse of the matrix of chromaticies
    5. Multiply the measured tristimulus values by the chromaticity matrix to give
      e_R (R/R0)^g           X
      e_G (G/G0)^g = [M^-1]  Y
      e_B (B/B0)^g           Z

      Of course, the multiplicatinve factors, (R/R0)^g, (G/G0)^g, and B/B0)^g should be equal

The constants required to calculate the tristimulus values without further measurement are now known.

Getting RGB from the Tristimulus Values

More often we know what tristimulus values should appear on the screen, and want to know what to write into the frame buffer.

  1. Use the inverse equation above.
    A_R           X
    A_G = [M^-1]  Y
    A_B           Z
  2. Divide each element of the resulting vector by the corresponding normalization factor
    (R/R0)^g = A_R / e_R

    and so on.

  3. Take the logarithm
    log(R/R0) = (1/g) * log(A_R/e_R)

    and so on

  4. Exponentiate
    R = R0 * (A_R/e_R)^(1/g)

It is common to find that R/R0 > 1. Then the desired colour cannot be produced by the display. What to do in that case is addressed later in the course.

Other Additive Devices

Colour LCD

These are widely used because they occupy little volume, use little energy, and weigh little compared to CRTs. Their colour performance is comparable to CRTs, but the structure of the images they produce can be intrusive

Technological basis

What if we had four filters?

Colour OLED

Solid state photodetectors and LEDs are very closely related

Plasma Display Panel (PDP)

Fuorescent Lights

They consist of

  1. A gas tube coated on the inside with a phosphor.
  2. It is filled mostly with inert gases, to which a small amount of mercury vapour is added.
  3. Electrodes at each end of the tube

They make light by


A PDP is an array of electrodes and phosphors which, in the off state maintain a gas just below it breakdown level.

Fluorescent light is turned off and on by applying varying voltages to the electrodes.

Thus, a PDP is in effect an array of tiny fluorescent lamps..

Field Effect Display (FED)

Just as the PDP is an array of tiny fluorescent lamps, an FED is an array of tiny CRTs.

Return to: