4:2:2 is a chroma subsampling mode where colour resolution is halved horizontally but kept full vertically. Compared to the universal streaming default of 4:2:0 — which halves colour in both directions — 4:2:2 keeps twice as much colour data. Compared to 4:4:4 (no subsampling at all), 4:2:2 is half the colour data. It sits in the middle of the trade-off curve, and that's where it earns its keep.
The home of 4:2:2 is production. Professional cameras (Sony, ARRI, Canon), broadcast workflows, intermediate editing codecs like ProRes 422 and DNxHR HQ, and live broadcast equipment all use 4:2:2 because it's the lowest level of colour subsampling that survives multiple rounds of editing, colour grading, keying and compositing without producing visible colour artefacts. Greenscreen keying on 4:2:0 material is particularly painful; on 4:2:2 it works cleanly.
For a streaming product, you'll see 4:2:2 mostly on the input side of your pipeline — what comes in from cameras and editors — and then convert to 4:2:0 for delivery. The reason: 4:2:2 typically costs 30 % more storage and bandwidth than 4:2:0 for almost no perceived quality benefit on natural finished content. The exceptions are content where chroma detail genuinely matters (animation, screen recordings with coloured text, graphic-heavy news broadcasts) — for those, sticking with 4:2:2 through delivery can be worth the cost. HEVC Main 4:2:2 and AV1 4:2:2 profiles exist precisely for this niche.
