Pixel Formats and Bit Depth

Pixel Formats

Common pixel formats determine how colour information is stored:

Format	Bits per Pixel	Number of Colours
Greyscale	8 bits (1 byte)	256
Highcolour	16 bits (2 bytes)	65,536
Truecolour	24 bits (3 bytes)	16.7 million
Deepcolour	≥32 bits (4+ bytes)	Billions+

Why Bit Depth Matters

Storage Requirements

Each colour channel typically uses 8 bits (values 0-255). A 5-megapixel uncompressed truecolour image requires: $5 \times 10^6 \times 3 = 15 \text{ MB}$

Why 8 Bits Suffice

Gamma encoding makes the 8-bit scale perceptually uniform. Without gamma:

• Perceptible banding in smooth gradients
• At least 12 bits per channel needed

Colour Banding

When there are insufficient bits to represent smooth colour gradients, banding artefacts appear.

Causes

• Limited bit depth
• Dark regions where gamma encoding compresses values
• Smooth gradients across large areas

Visual Illusions That Exacerbate Banding

Mach band illusion: The visual system exaggerates perceived intensity differences at edges, making band steps more visible.

Chevreul illusion: Similar edge enhancement effect.

Solutions

• Dithering: Add controlled noise to break up visible bands
• Higher bit depth: 10-bit or 12-bit encoding
• Avoid smooth gradients in design where possible

Deep Colour

For HDR content and professional workflows:

• 10-bit: 1024 levels per channel (used in HDR video)
• 12-bit: 4096 levels per channel
• 16-bit float: Per-channel floating-point precision

Summary

• Truecolour (24-bit) is standard for most applications
• 8 bits per channel suffices because gamma encoding makes it perceptually uniform
• Insufficient bit depth causes visible banding
• Dithering can reduce visible banding artefacts