As indicated in an earlier section, computer systems store images in their memory as a matrix of pixels. If the image changes with time, then a succession of these matrices is stored; one for each time frame. The data recorded for each pixel must contain its position on the screen (indicated by its location in the matrix), its luminance and in the case of colour images its colour. In the simplest case, such as a laser printer, each pixel can only be black or white, being a presence or absence of toner, so only one bit of information need be stored for each point. In a typical personal computer graphics system each pixel normally has 8bits, or one byte, allocated to it. This number can be thought of as a code that is used to select the final displayed colour and luminance from an array of pre-selected ones. The analogy commonly made is with an artist's palette. Although he is theoretically able to mix combinations of paints and thinners to make millions of different colours, he is only able to have a small selection of them on his palette at one time. Note that numerically consecutive pixel values need not translate to adjacent luminances or colours and that the palette may contain colours and luminances selected from the entire gamut of the display device. In this respect, the pixel value acts only as a place-holder. It is worth pointing out that nothing can be inferred about the resolution of luminances and colours with which the palette can be filled by knowing that each pixel is only represented by 8bits. What can be known though is that it is only possible to have 256 different luminances (and colours) displayed simultaneously on the same frame. As computer memory becomes cheaper we are seeing the rise in popularity of so-called true-colour graphics systems. These typically use 24 or 32bits in which to store each pixel, some of the bits (often 14) being used to store the luminance information and the remainder used to store the colour information the idea being that all the data is contained within the pixel so that the palette is superfluous. True-colour systems are superficially attractive but they have the drawback that each image occupies 3 or four times more memory than in the 8bit systems which, while not of itself a problem as memory is now fairly cheap, means that it takes 3 or 4 times longer to load or construct thus limiting the dynamic capabilities of the system.

^ Back to top