The human colour perception system uses three varieties of cones in the retina, each responding to a different wavelength spectrum. This is known as tristimulus colour, and is the reason colour specifications require three attributes to accurately describe an arbitrary colour.
There are two main types of colour systems employed in video and graphic systems, component colour systems and composite colour systems.
Component systems transmit independant colour signals, and such systems include the additive RGB system and the subtractive CMY(K) system. Component colour systems are most often found in graphics systems.
Composite systems use frequency interleaving to combine three attributes into a single signal for transmission. The three attributes are a luminance and two colour differences. The luminance is a measure of the brightness of the signal (and so alone is suitable for black and white displays), and the colour differences are the difference between two primary colour values, such as the red minus the yellow components. The two most common examples are NTSC's YIQ (Luminance (Y), Interphase (I) and Quadrature (Q)) and PAL's YUV. Composite systems are most often found in video systems as they use less bandwidth (see Section 5.5.2) than component systems with no loss of information. In addition, the colour difference signals can be modulated onto the primary luminance signal, allowing backward-compatible introduction of colour television sets.
The colour gamuts of the composite video systems are also such that high saturation13 colours are to be avoided. NTSC has problems with colours of high saturation and low luminance, and complementary saturated colours may colour-bleed on screen and videotape. Also, the colour gamut of PAL is smaller than that of NTSC, so some colours visible in NTSC cannot accurately be reproduced in PAL.