The robust nature of digital signals makes them a very desirable vehicle for communication; but the number of bits required for pictorial images can be very high unless systems are carefully designed to fit the perceptual properties of human vision. Features where important savings can be made in the bit requirements include: adjusting the light level in scanners; using a log or power function instead of a linear function for tonal quantization; allowing for the limited reproduction gamut; using luminance and chrominance signals to achieve bandwidth compression; and allowing for the modulation transfer function of the eye. By these means, large savings in bit requirements have been achieved in both desk-top publishing and broadcast television.