Video Signal Formats and Chroma Subsampling: A Reference

The short answer

A digital video signal is defined by four independent variables, and their product is the uncompressed bandwidth the cable, extender, and display must carry: resolution (active pixels), refresh rate (frames per second), bit depth (bits per color channel), and chroma subsampling (how much color detail is discarded). Multiply them out. 1080p is 1920x1080 (about 2.07 million pixels), 4K UHD is 3840x2160 (about 8.29 million, four times 1080p), and 8K UHD is 7680x4320 (about 33.18 million, four times 4K). At 4:4:4 8-bit each pixel is 24 bits; 4K at 60 Hz therefore needs roughly 3840 x 2160 x 60 x 24 = 11.94 Gbps of active payload before blanking and link-encoding overhead push it toward the ~18 Gbps of HDMI 2.0.

Chroma subsampling reduces color resolution because human vision is far more sensitive to brightness than to color. 4:4:4 keeps full color (24 bits/pixel at 8-bit) and is correct for text and computer graphics. 4:2:2 halves horizontal color (16 bits/pixel). 4:2:0 keeps a quarter of the color detail (12 bits/pixel) and is the standard for Blu-ray, broadcast, and streaming. Bit depth sets tonal gradation: 8-bit is 256 levels per channel (16.7 million colors), 10-bit is 1024 levels (about 1.07 billion), and 12-bit is 4096 levels. HDR needs 10-bit or higher to avoid visible banding across its wider brightness range. HDR10 uses static metadata at 10-bit; Dolby Vision uses dynamic per-scene metadata up to 12-bit; HLG (ITU-R BT.2100) is metadata-free and backward compatible with SDR displays.

The Four Variables That Define Any Video Signal

Every uncompressed digital video signal is described by four independent settings, and the entire distribution chain succeeds or fails on their product. Read a spec in this order: resolution (how many active pixels), refresh rate (frames per second, in Hz), bit depth (bits per color channel), and chroma subsampling (how much color information is kept). Change any one and the required bandwidth changes proportionally. A common field mistake is treating 4K as a single requirement; 4K/60 4:4:4 10-bit demands roughly two and a half times the data of 4K/60 4:2:0 8-bit, and only one of those fits an older extender.

The bandwidth formula is direct: active pixels x refresh x bits-per-pixel. Bits-per-pixel depends on both bit depth and subsampling. At 8-bit, 4:4:4 is 24 bits/pixel, 4:2:2 is 16, and 4:2:0 is 12; at 10-bit those become 30, 20, and 15. The result is the active-video payload. Real cables must also carry blanking intervals (roughly 5 to 25 percent extra depending on timing) and pay a link-encoding tax, so the physical link rate always exceeds the raw payload figure.

Use this decision order before selecting hardware: confirm the source's native format, confirm the display's accepted format, then size the transport (cable, extender, matrix) to the higher of the two. When the payload exceeds a transport tier, your options are to lower one variable (drop 4:4:4 to 4:2:0, or 10-bit to 8-bit), move to a faster link (HDMI 2.1, HDBaseT 3.0, or fiber), or apply visually lossless compression (VESA DSC).

• Bandwidth = active pixels x refresh rate x bits-per-pixel; every variable scales the result linearly
• Bits-per-pixel at 8-bit: 4:4:4 = 24, 4:2:2 = 16, 4:2:0 = 12; at 10-bit: 30, 20, 15; at 12-bit 4:4:4 = 36
• 1080p = 1920x1080 (2.07 MP); 4K UHD = 3840x2160 (8.29 MP, 4x 1080p); 8K UHD = 7680x4320 (33.18 MP, 4x 4K)
• Physical link rate exceeds the active payload: add blanking overhead plus 8b/10b or 16b/18b encoding tax
• When payload exceeds the link: lower a variable, move to a faster transport, or apply VESA DSC compression
• Size transport to the higher of the source and display formats, not to the label '4K' alone

Resolutions and Refresh Rates

Consumer UHD resolutions quadruple pixel count at each step. 1080p (Full HD) is 1920x1080, or 2,073,600 active pixels. 4K UHD is 3840x2160, or 8,294,400 pixels, exactly four 1080p frames. 8K UHD is 7680x4320, or 33,177,600 pixels, four 4K frames and sixteen 1080p frames. Note that UHD (3840x2160) is distinct from DCI 4K (4096x2160), the wider cinema format used in digital projection; consumer displays and the specifications on this page use the UHD dimensions.

Refresh rate multiplies the pixel load by frames per second. Film-sourced content is mastered at 24 Hz; broadcast and most video runs at 30 Hz (technically 29.97 in NTSC regions) or 60 Hz; gaming and high-motion displays target 120 Hz. Because bandwidth scales linearly with refresh, 4K/120 requires exactly twice the data of 4K/60 at identical bit depth and subsampling. This is why 120 Hz at 4K only became practical with HDMI 2.1-class links.

Resolution and refresh together set the raw pixel throughput. 1080p/60 moves about 124 million pixels per second; 4K/60 moves about 498 million; 4K/120 about 995 million; 8K/60 about 1.99 billion. Every later calculation multiplies these pixel rates by the bits carried per pixel, so choosing a resolution and refresh rate effectively fixes the floor of your bandwidth requirement before color settings are even considered.

• 1080p = 1920x1080 = 2,073,600 pixels; 4K UHD = 3840x2160 = 8,294,400; 8K UHD = 7680x4320 = 33,177,600
• UHD 4K (3840x2160) differs from DCI/cinema 4K (4096x2160); consumer gear uses the UHD dimensions
• Common refresh rates: 24 Hz (film), 30 Hz (broadcast; 29.97 NTSC), 60 Hz (video), 120 Hz (gaming/high motion)
• Bandwidth scales linearly with refresh: 4K/120 needs exactly 2x the data of 4K/60 at the same color settings
• Active pixel rates: 1080p/60 = 124 MP/s; 4K/60 = 498 MP/s; 4K/120 = 995 MP/s; 8K/60 = 1.99 GP/s
• Each UHD step quadruples pixel count: 8K is 4x 4K and 16x 1080p

Bit Depth and HDR Formats

Bit depth is the number of bits used per color channel (red, green, blue, or luma/chroma), and it sets how finely brightness and color are quantized. 8-bit gives 256 levels per channel, which combine across three channels to 16.7 million colors. 10-bit gives 1024 levels per channel, about 1.07 billion colors. 12-bit gives 4096 levels per channel. More levels mean smoother gradients and fewer visible steps in areas like skies, shadows, and lighting falloff.

High dynamic range (HDR) stretches luminance across a far wider range than standard dynamic range using the SMPTE ST 2084 Perceptual Quantizer (PQ) transfer function, whose signaling range reaches up to 10,000 nits. Spreading only 256 8-bit steps across that range produces visible banding, so HDR requires 10-bit or greater. HDR content also typically uses the wide ITU-R BT.2020 color gamut, versus BT.709 for HD SDR, which further rewards the extra bit depth.

The major HDR formats differ in metadata and licensing. HDR10 is an open standard: 10-bit, PQ, with static metadata (SMPTE ST 2086 mastering-display data plus MaxCLL and MaxFALL) describing the whole title. HDR10+ and Dolby Vision add dynamic metadata that adapts tone mapping per scene or frame; Dolby Vision is licensed and supports up to 12-bit. HLG (Hybrid Log-Gamma), defined in ITU-R BT.2100 alongside PQ, carries no metadata and remains viewable on SDR displays, which suits live broadcast.

• 8-bit = 256 levels/channel (16.7M colors); 10-bit = 1024 (about 1.07B); 12-bit = 4096 levels/channel
• HDR needs 10-bit minimum to avoid banding; SDR 1080p/4K is fine at 8-bit
• HDR uses the SMPTE ST 2084 PQ curve (up to 10,000 nits) and typically the wide ITU-R BT.2020 gamut
• HDR10: open, 10-bit, static metadata (SMPTE ST 2086 mastering data plus MaxCLL/MaxFALL)
• Dolby Vision: licensed, dynamic per-scene/frame metadata, up to 12-bit; HDR10+ adds dynamic metadata at 10-bit
• HLG (ITU-R BT.2100): 10-bit, metadata-free, backward compatible with SDR displays, suited to broadcast

Chroma Subsampling: 4:4:4, 4:2:2, and 4:2:0

Video is usually encoded as luma (Y, brightness) plus two chroma channels (Cb, Cr) rather than raw RGB, because the eye resolves brightness detail more sharply than color detail. Chroma subsampling exploits this by sending full-resolution luma but fewer color samples. The notation is written as three numbers over a conceptual 4-pixel-wide, 2-pixel-tall reference block: the first number (4) is the luma width, the second is the number of chroma samples in the top row, and the third is the number of chroma samples that change in the bottom row.

4:4:4 keeps every chroma sample: full color, no subsampling, 24 bits/pixel at 8-bit. It is the correct choice for computer text, UI, CAD, and fine graphics, where colored edges must stay crisp. 4:2:2 halves horizontal chroma (top row samples 2 of 4), giving 16 bits/pixel at 8-bit; it is common in professional capture and broadcast production. 4:2:0 samples chroma at half resolution both horizontally and vertically (2 in the top row, 0 changing in the bottom), yielding 12 bits/pixel at 8-bit, half the data of 4:4:4.

4:2:0 is the delivery standard for Ultra HD Blu-ray, ATSC/DVB broadcast, and streaming services because it roughly halves the color payload with almost no visible loss on natural photographic and video content. Its weakness shows only on sharp synthetic color edges, most notably small colored text, where it produces fringing. The practical rule: deliver movies and broadcast at 4:2:0, but drive computer and signage sources that render text at 4:4:4 whenever the transport has the headroom.

• Notation J:a:b over a 4x2 reference block: J = luma width (4), a = top-row chroma samples, b = bottom-row chroma changes
• 4:4:4 = full color, no subsampling, 24 bits/pixel at 8-bit; use for text, UI, CAD, and graphics
• 4:2:2 = half horizontal chroma, 16 bits/pixel at 8-bit; common in broadcast and pro capture
• 4:2:0 = quarter chroma (half horizontal and half vertical), 12 bits/pixel at 8-bit; half the data of 4:4:4
• 4:2:0 is standard for Ultra HD Blu-ray, broadcast, and streaming; near-invisible loss on natural video
• 4:2:0 causes color fringing on fine text/graphics; prefer 4:4:4 for computer and signage sources

Bandwidth Math and HDMI/HDBaseT Limits

Compute the active payload as pixels x refresh x bits-per-pixel. Worked examples at 8-bit unless noted: 1080p/60 4:4:4 = about 2.99 Gbps; 4K/60 4:2:0 = about 5.97 Gbps; 4K/60 4:4:4 = about 11.94 Gbps; 4K/60 4:4:4 10-bit = about 14.93 Gbps; 4K/120 4:4:4 10-bit = about 29.86 Gbps; 8K/60 4:2:0 10-bit = about 29.86 Gbps; 8K/60 4:4:4 10-bit = about 59.72 Gbps. These are active-video figures; the physical link must add blanking and encoding overhead on top.

HDMI links have hard ceilings. HDMI 1.4 carries 10.2 Gbps of raw signaling (about 8.16 Gbps effective after 8b/10b TMDS encoding), enough for 4K/30 4:4:4 or 4K/60 4:2:0. HDMI 2.0 carries 18 Gbps raw (about 14.4 Gbps effective), enough for 4K/60 4:4:4 8-bit. HDMI 2.1 uses Fixed Rate Link at 48 Gbps raw (about 42.6 Gbps effective after 16b/18b encoding), enabling 4K/120 and, with VESA Display Stream Compression (visually lossless, roughly 3:1), 8K/60, since uncompressed 8K/60 4:4:4 10-bit at about 59.7 Gbps exceeds even 48 Gbps.

For distance, HDBaseT sends video, audio, control, Ethernet, and power over one Cat cable to 100 meters (328 feet). HDBaseT 1.0/2.0 links carry up to 10.2 Gbps, matching HDMI 1.4-class formats (4K/30 4:4:4 or 4K/60 4:2:0). HDBaseT Spec 3.0 raises the native link to 16 Gbps and carries uncompressed 4K/60 4:4:4 8-bit to 100 meters over Cat6A. Beyond those tiers, installers reach for HDMI 2.1 active optical cable (fiber), compressed transport, or AV-over-IP. Always match the extender class to the specific format the payload table demands, not to the word '4K'.

• Active payload examples: 1080p/60 4:4:4 8-bit = 2.99 Gbps; 4K/60 4:2:0 8-bit = 5.97 Gbps; 4K/60 4:4:4 8-bit = 11.94 Gbps
• Higher tiers: 4K/60 4:4:4 10-bit = 14.93 Gbps; 4K/120 4:4:4 10-bit = 29.86 Gbps; 8K/60 4:4:4 10-bit = 59.72 Gbps
• HDMI 1.4 = 10.2 Gbps raw / ~8.16 effective (4K/30 4:4:4 or 4K/60 4:2:0); HDMI 2.0 = 18 Gbps raw / ~14.4 (4K/60 4:4:4 8-bit)
• HDMI 2.1 = 48 Gbps raw / ~42.6 effective via FRL; enables 4K/120 and, with VESA DSC, 8K/60
• HDBaseT reaches 100 m (328 ft) over Cat: 1.0/2.0 up to 10.2 Gbps; Spec 3.0 = 16 Gbps native, uncompressed 4K/60 4:4:4 to 100 m on Cat6A
• 8K/60 4:4:4 10-bit (~59.7 Gbps) exceeds 48 Gbps, so it requires DSC or another compression scheme

Putting It Together and When a Pro Install Helps

To specify a system, start from the most demanding source and display in the room and work backward. A 4K/60 HDR movie source produces a 10-bit 4:2:0 or 4:2:2 signal that fits HDMI 2.0 and HDBaseT 3.0 comfortably. A 4K/60 desktop or signage source that must render crisp text wants 4:4:4 8-bit at about 11.94 Gbps, which needs an HDMI 2.0-class or better path end to end, including every adapter, plate, and matrix port. A gaming or 4K/120 requirement, or any 8K plan, moves the whole chain to HDMI 2.1 or DSC.

The pitfalls that cause field failures are rarely the display; they are the transport details. A single HDMI 2.0-rated cable in an otherwise HDMI 2.1 path caps the link. An HDBaseT 1.0 extender silently forces 4:2:0 on a source that wanted 4:4:4. EDID handshakes can renegotiate a display down to a lower format when a receiver in the chain cannot pass the full bandwidth. Category cable quality, length, and termination all bear on whether a 16 Gbps HDBaseT link holds at 100 meters.

A DIYer or in-house team can absolutely handle a short direct-connect 4K setup by matching cable and device tiers to the payload table above. Longer runs, multi-display distribution, HDR that must survive the whole chain, and mixed 4:4:4/4:2:0 sources are where planning gets unforgiving, because one under-rated component quietly degrades the signal. That is where working with an installer to model the bandwidth, select matched HDMI 2.1 or HDBaseT 3.0 hardware, and verify EDID and format at each hop pays off. EVOTECH IT LLC designs and installs low-voltage video distribution for homes and businesses in the Houston area.

• Spec from the most demanding source/display in the room, then size every hop (cable, plate, matrix) to that payload
• 4K/60 HDR video (10-bit 4:2:0/4:2:2) fits HDMI 2.0 and HDBaseT 3.0; 4K/60 4:4:4 text needs an HDMI 2.0-class or better full path
• 4K/120, gaming, and any 8K plan require HDMI 2.1 (FRL) or VESA DSC end to end
• One under-rated cable, extender, or port caps the whole link and can force 4:2:0 or a lower resolution
• EDID renegotiation can silently drop a display to a lower format when a mid-chain device lacks the bandwidth
• Short direct 4K runs suit DIY; long runs, multi-display distribution, and chain-wide HDR benefit from professional design and verification

Frequently asked questions