Dithering technique

Dithering is a technique that simulates colours or grey levels that are not available in a limited palette by arranging pixels of the available colours in a pattern. The eye blends neighbouring pixels, so the image looks as if it has more colours or shades than the display really has.

Why it is needed

A display or file format with few colours (for example 1-bit black and white, or an 8-bit 256-colour palette) cannot show smooth gradients of a photograph. Direct reduction gives visible bands (false contours). Dithering trades spatial resolution for apparent intensity resolution.

Principle (black and white example)

A small block of pixels is used for one original pixel. A 2 x 2 block can show 5 grey levels depending on how many pixels are black:

 level 0   level 1   level 2   level 3   level 4
  . .       # .       # .       # .       # #
  . .       . .       . #       # #       # #
 (white)                                  (black)

Methods

1. Ordered (pattern) dithering. Each pixel intensity is compared with a threshold from a repeating dither matrix, e.g. for a 2 x 2 matrix [[0,2],[3,1]] scaled to 0-255. If the pixel value is above the matrix threshold the output is white, else black. Fast but gives a regular pattern.
2. Random dithering. A random noise value is added before thresholding. No pattern, but grainy.
3. Error diffusion (Floyd-Steinberg). The pixel is rounded to the nearest available colour, and the error is spread to the neighbours not yet processed: 7/16 to the right, 3/16 below-left, 5/16 below, 1/16 below-right. Gives the best quality.

Uses and effect

• Showing photographs on 256-colour or monochrome displays, in GIF images, newspaper (halftone) printing and fax.
• Advantage: a better look with a small palette. Disadvantage: image looks grainy and dithered images compress poorly.