Chroma Subsampling: Why Video Uses 4:2:0

The notation describes how much color information survives: 4:4:4 keeps independent color data for every pixel; 4:2:2 halves it horizontally; 4:2:0 halves it both ways, so each 2×2 block of pixels shares one color sample. That is three quarters of the color data thrown away.

The reason it works: color-sensitive cone cells are far sparser in the human retina than brightness-sensitive rods. Eyes are extremely sharp on luminance detail and surprisingly blunt on color detail. Luma (Y) is kept at full resolution; the two chroma channels (Cb, Cr) get halved — and at normal viewing distance you simply cannot tell.

Blu-ray discs, mainstream streaming and phone/camera footage are all 4:2:0. 4:2:2 lives in professional cameras and HDMI capture pipelines (paired with intermediate codecs like ProRes to leave headroom for keying and grading). 4:4:4 exists only in screen recording, graphics and professional post — scenes full of pure-color text and fine graphics, where halved chroma would give itself away.

For everyday transcoding, source and output are both 4:2:0, so subsampling is not a variable you need to manage. The one real mistake is going the other way: re-encoding a 4:2:0 source to 4:4:4 output adds no information — you just store interpolated data at four times the size.

The visible cost of 4:2:0 is chroma bleed: saturated block edges, red-on-white text and fine colored lettering go soft because one color sample is averaged across four pixels. Subtitles, logos and graphic overlays suffer most; natural footage edges are practically immune.

The discipline: keep 4:2:0 untouched — bitrate and resolution decide picture quality, not chroma. If screen-recording text looks muddy, raise bitrate and resolution first. Only text-critical screen content justifies 4:2:2 output, at a clear size premium.