Chroma Subsampling Chroma Subsampling (4:4:4 / 4:2:2 / 4:2:0)

What 4:4:4, 4:2:2, and 4:2:0 mean

The J:a:b notation describes a conceptual sampling region J pixels wide and 2 pixels tall. J is the horizontal sampling reference, almost always 4 luma samples. a is the number of chrominance samples (Cr, Cb) taken in the first row of those J pixels. b is the number of changes of chrominance samples between the first and second rows. Luma is always sampled at every pixel; only chroma is reduced.

4:4:4 has no subsampling at all — every pixel carries its own luma and chroma sample. 4:2:2 halves chroma horizontally, with two chroma samples shared across four luma pixels in each row. 4:2:0 halves chroma both horizontally and vertically: two chroma samples in row 1 and zero new chroma samples in row 2 (row 2 reuses row 1's chroma), giving roughly one quarter the chroma sample density of 4:4:4.

Why it works and what it saves

Chroma subsampling exploits an asymmetry in the human visual system: the eye is much more sensitive to variations in brightness than to variations in color. Spending bits on luminance — what the eye actually resolves at high acuity — and economizing on chrominance produces large bandwidth savings with surprisingly little perceptible quality loss.

Relative to a 4:4:4 baseline, the 4:2:2 Y'CbCr scheme requires roughly two-thirds the bandwidth of non-subsampled 4:4:4. 4:2:0 halves chroma in both axes, cutting overall stream bandwidth roughly in half versus 4:4:4. Those savings are why 4:2:0 is the universal choice for compressed delivery formats.

HDMI bandwidth at the playback end is the practical constraint that forces these choices into the open. HDMI 2.0's 18 Gbps ceiling assumes 4K/60 with HDR, 4:4:4 chroma, and 10-bit color would saturate the entire link, so 4K/60 over HDMI 2.0 in practice runs at 4:2:0 (or 4:2:2 with reduced color depth). Carrying 4K/60 at full 4:4:4 with 10-bit HDR exceeds 18 Gbps and requires HDMI 2.1 FRL (Fixed Rate Link, up to 48 Gbps).

Where each scheme actually lives

UHD Blu-ray is mastered and stored at YUV 4:2:0 10-bit, encoded with the HEVC (H.265) Main 10 profile at up to 100 Mbps. 4:2:0 is not an option in the spec — it is a requirement of the 4K UHD Blu-ray standard. There is no consumer disc format that ships 4:4:4 video.

Streaming and broadcast are the same story. All consumer video delivery — Blu-ray, UHD Blu-ray, broadcast, cable, and streaming services such as Netflix, Disney+, and Apple TV+ — ships at 4:2:0. The H.264/AVC and H.265/HEVC codecs that power consumer delivery use 4:2:0 in their main profiles by default.

4:2:2 lives in professional intermediate and acquisition codecs used in production and post: Apple ProRes 422 and its variants, Avid DNxHD/DNxHR, Sony XDCAM HD422, DVCPRO HD, and Digital Betacam. All ProRes 422 variants use 4:2:2 chroma subsampling at 10-bit color depth. 4:2:2 also appears in pro-grade HDMI capture cards and broadcast SDI workflows. It is essentially absent from consumer delivery.

4:4:4 matters in three places: PC desktop output where text and fine UI graphics are rendered pixel-by-pixel (subsampled red text on a black background famously gets fringed and unreadable), PC gaming where rendered geometry has hard high-contrast edges, and professional video editing and color grading. For movie watching from a Blu-ray or streaming source, 4:4:4 capability on the display is irrelevant because the source itself is 4:2:0.

One field-relevant wrinkle on the AVR side: some older HDMI 2.0-era AVRs strip incoming 4:4:4 to 4:2:2 (or even 4:2:0) when bandwidth headroom is tight — for example, when the receiver's on-screen display overlay is enabled, which adds bandwidth overhead. On many such AVRs, disabling the OSD restores full 4:4:4 passthrough. This is purely a bandwidth-management compromise inside the AVR's HDMI subsystem, not an audio decision.

Common confusions

4:4:4 is not the same as 10-bit color depth. Chroma subsampling and bit depth are independent axes. Subsampling describes the spatial resolution of the color planes — how many chroma samples per region of pixels. Bit depth describes the precision of each sample (8, 10, or 12 bits per channel). A signal can be 4:4:4 at 8-bit, 4:2:0 at 10-bit (UHD Blu-ray's actual format), or any other combination, and the two consume HDMI bandwidth separately.

Buying a 4:4:4-capable TV does not improve movie image quality. Because every consumer movie source is encoded at 4:2:0, the display can never receive more chroma resolution than the source provides. Upsampling 4:2:0 to 4:4:4 inside the display does not recover information that was discarded at encode — there is no intrinsic benefit to the source upsampling to 4:4:4 or converting to RGB. 4:4:4 capability matters for PC and console desktop or rendered output, not for movie playback.

4K HDR is not delivered at 4:4:4. All current consumer 4K HDR delivery is 4:2:0. UHD Blu-ray, Dolby Vision streaming, and HDR10 streaming are all encoded at 4:2:0 with 10-bit color depth — even those discs and streams with wide color gamuts and HDR still use 4:2:0 chroma. The 4:2:0 is a bandwidth-driven delivery limitation across all consumer HDR pipelines, not a quality cap on what HDR (high dynamic range, wide color gamut, PQ EOTF) can carry. HDR's quality wins come from bit depth, peak luminance, and gamut — not chroma resolution.

RGB and YCbCr 4:4:4 from a graphics card are functionally similar but not identical. Both transmit full per-pixel color with no chroma subsampling and consume the same HDMI bandwidth. RGB sends red/green/blue values directly; YCbCr 4:4:4 sends luma plus two full-resolution chroma planes that the display converts back to RGB. NVIDIA and AMD control panels expose both options. For PC desktop work either preserves text quality, while YCbCr 4:2:2 or 4:2:0 visibly degrades small text and UI elements — subsampling is less noticeable in movies and games but is not recommended for computer use because all desktop and text interfaces will suffer noticeably.