Dithering

Trade spatial noise for perceived smoothness when the output format cannot represent all the shades you want directly.

graphics

color

quantization

output

Interactive playground

Tweak the operator or scene live so the article connects to an immediate visual result.

Dithering

Use spatial structure or error diffusion to fake smooth gradients when the output only supports a small number of intensity levels.

Continuous gradient

Reference view

Quantized output

Interactive compare

Ordered dithering uses a threshold pattern, while error diffusion pushes quantization mistakes into nearby pixels.

Quantization levels 4

Dither method

Family

Color & tone

Color spaces, gamma, tone mapping, compositing, and distribution-aware image remapping.

Builds on

2 topics

Read these first if you want the surrounding pipeline context.

Unlocks

0 next topics

Use these follow-ups when you want to keep turning the image-processing pipeline forward.

Learning paths

This topic appears in curated graphics progressions so the next step is obvious.

Problem

A gradient can look perfectly smooth in a high-precision buffer but break into obvious bands when stored with fewer levels or mapped to a tiny palette.

Intuition

Dithering deliberately scatters error or patterning across neighboring pixels so the eye perceives intermediate shades that the hardware or format cannot represent directly.

Core idea

Quantize each pixel to the available levels or palette.
Either compare against an ordered threshold pattern or spread the quantization error into neighbors.
Let the viewer integrate the resulting spatial noise into a smoother perceived tone.

Worked example

A black-to-white ramp rendered with only a handful of gray levels bands badly without dithering. With a Bayer pattern or error diffusion, the same ramp looks much smoother from normal viewing distance.

Complexity

Ordered dithering is constant work per pixel. Error-diffusion methods are still linear in the number of pixels but require carrying local quantization error forward.

When to choose it

Choose it when output precision or palette size is limited.
Tone mapping should usually happen first if the scene is still HDR-like.
Histogram equalization changes contrast; dithering changes how limited output levels are spatially arranged.

Key takeaways

Dithering fights banding by adding structured or distributed noise.
It is most relevant near the end of the pipeline when output precision becomes limited.
Ordered and error-diffusion variants make different quality and cost trade-offs.
Good dithering often looks noisy up close and smoother from a normal viewing distance.

Practice ideas

Render the same gradient with no dithering, Bayer dithering, and error diffusion.
Quantize a tone-mapped image to very low bit depth and compare the result with and without dithering.
Experiment with dithering before and after gamma-aware output conversion.

Relation to other topics

Tone mapping and gamma correction usually happen before final dithering decisions.
Histogram equalization may increase contrast, which can make banding more visible and therefore increase the need for dithering.
Alpha compositing can produce subtle gradients that later need protection from banding.