Color depth & chroma subsampling
Bit depth, RGB vs. YCbCr, and the 4:4:4 / 4:2:2 / 4:2:0 shorthand on every spec sheet.
Resolution tells you how many pixels there are. Colour encoding tells you how much information each pixel carries — and it's the reason spec sheets are littered with figures like 10-bit and 4:2:0.
Bit depth: how many shades
Screens build every colour from three primaries: red, green and blue (RGB). Each channel is stored as a number, and the bit depth sets how many steps it has — exactly as in the sampling lab from Chapter 1.
| Bit depth (per channel) | Steps per channel | Total colours | Typical use |
|---|---|---|---|
| 8-bit | 256 | ≈ 16.7 million | Standard (SDR) video and computer desktops |
| 10-bit | 1,024 | ≈ 1.07 billion | HDR video, professional work |
| 12-bit | 4,096 | ≈ 68.7 billion | High-end cinema and mastering |
- 8-bit
- Steps per channel256
- Total colours≈ 16.7 million
- Typical useStandard (SDR) video and computer desktops
- 10-bit
- Steps per channel1,024
- Total colours≈ 1.07 billion
- Typical useHDR video, professional work
- 12-bit
- Steps per channel4,096
- Total colours≈ 68.7 billion
- Typical useHigh-end cinema and mastering
With too few steps, smooth gradients — a sky, a dim wall — show visible stripes called banding. HDR (high dynamic range) content stretches the range from dark to bright, which spreads those steps further apart, so HDR needs at least 10-bit to look smooth.
RGB vs. YCbCr
Video is often carried not as RGB but as YCbCr — the same picture, rearranged:
- Y (luma) — brightness alone. A black-and-white version of the image.
- Cb and Cr (chroma) — how far each pixel leans towards blue or red. The colour, with the brightness removed.
Why bother? Because human vision is far sharper for brightness than for colour. Separate the two and you can keep every bit of brightness detail while storing colour at lower resolution — and most people will never notice.
Chroma subsampling
The three-number shorthand describes how much colour detail is kept, using a reference block four pixels wide and two rows tall:
- 4:4:4 — full colour for every pixel. No subsampling.
- 4:2:2 — colour shared between each horizontal pair. Half the horizontal colour detail.
- 4:2:0 — colour shared across each 2 × 2 block. A quarter of the colour detail.
100%
Every pixel keeps its own colour.
Where it matters
For movies and camera video, 4:2:0 is the norm: streaming services, broadcast and Blu-ray all use it, and natural images hide the loss well.
For computer content the story changes. Spreadsheets, code, thin coloured lines and small coloured text on a coloured background are exactly the sharp colour edges subsampling destroys. That's why 4:4:4 matters in presentation and control-room systems, and why it appears so often on the spec sheets of switchers, extenders and AV-over-IP encoders.
Key takeaways
- 01Bit depth sets the number of shades per colour channel; HDR needs at least 10-bit.
- 02YCbCr separates brightness (luma) from colour (chroma).
- 03Human vision is sharper for brightness than colour, so chroma can be reduced with little visible loss.
- 044:4:4 is full colour; 4:2:2 carries about ⅔ of the data and 4:2:0 about ½.
- 05Computer content with fine coloured detail needs 4:4:4; video usually looks fine at 4:2:0.
Check your understanding
Q1Which component of YCbCr carries the brightness detail?
Q2In 4:2:0 subsampling, how many pixels share a single colour sample?
Q3A control room shows spreadsheets with small red text on grey. Which format best preserves legibility?