Dithering

Dithering is a technique used in digital imaging, audio, and other signal processing disciplines to minimize quantization artifacts when reducing the resolution of a signal. It achieves this by introducing or distributing noise or error so that the overall result appears more continuous than would be possible from a simple quantization step.

In image processing, dithering is used when reducing color depth or displaying images on devices with a limited color palette. After quantizing a pixel to the nearest color in the target palette, the quantization error is distributed to neighboring pixels according to a weighting pattern. This helps preserve perceived brightness and detail, reducing visible banding. Common methods include error-diffusion algorithms such as Floyd–Steinberg, and ordered dithering using a threshold matrix like the Bayer pattern. Random or stochastic dithering adds noise to the quantization process to avoid structured artifacts.

In audio processing, dithering adds a small amount of noise to a signal before reducing bit depth or sampling resolution. The aim is to decorrelate the quantization error and reduce audible distortion, especially in low-level passages. Types include triangular probability density function (TPDF) dithering and more complex shaped dither spectra.

History and usage: Dithering has roots in halftone printing and was adapted for digital displays in the 20th century, with error-diffusion algorithms becoming widely used in the 1970s and 1980s. Its effectiveness depends on the content and target medium; it can introduce a grainy texture or a perceptible noise pattern that is aesthetically desirable for photographs but may be undesirable for sharp text or graphics.

See also: color quantization, halftoning, error diffusion, palette reduction, quantization error, dithering in audio.