UHD, wide colour and a large container
A complete BT.2020 guide: why UHD needed a new gamut, exact spectral-like primaries and D65, 10/12-bit coding, the OETF, Y′ coefficients, constant/non-constant-luminance YCbCr, how it differs from HDR/BT.2100, P3 mastering inside a Rec.2020 container and the role of gamut mapping.
Rec.2020 is a wide-gamut RGB/UHD colorimetry standard, not a synonym for HDR. It defines very distant primaries, D65, signal/coding parameters and YCbCr relationships. HDR often uses the same BT.2020/BT.2100 colour primaries, but PQ or HLG is a separate transfer layer.
Quick reference
≈ 630 nm monochromatic equivalent in the BT.2100 table.
≈ 532 nm monochromatic equivalent.
≈ 467 nm monochromatic equivalent.
Bit depth does not change gamut geometry.
Different from Rec.709 matrix coefficients.
PQ/HLG belong to the BT.2100 transfer layer.
The triangles use the standard xy coordinates. A large encoding gamut does not imply that every real display — or every master — reaches its outer edges.
A primaries label cannot tell you the HDR transfer function. Always inspect colour primaries and transfer characteristics separately.
Bit depth controls quantisation precision. It does not move the RGB primaries and therefore does not make the gamut larger.
A wrong matrix can create colour errors even when the primaries metadata itself is correct.
The move to UHD was not treated merely as scaling 1920×1080 to 3840×2160 or 7680×4320. ITU wanted a system with higher spatial resolution, deeper quantisation and a substantially wider colour gamut. BT.2020 was published as a new UHD baseline for production and international programme exchange. Rec.2020 is therefore a system recommendation, not merely a triangle on a CIE diagram.
CIE 1931 xy: Red 0.708/0.292, Green 0.170/0.797, Blue 0.131/0.046, with D65 white at 0.3127/0.3290. BT.2100 gives informative spectral equivalents of roughly 630 nm, 532 nm and 467 nm monochromatic light for these primaries. BT.2020 therefore places the vertices very close to the spectral locus. The gamut is far wider than P3 or Rec.709 and includes chromaticities that many real displays still cannot physically reproduce.
Despite the new gamut, BT.2020 retained D65. This improves interoperability with Rec.709 and later BT.2100: the neutral axis and reference-white chromaticity stay compatible while the primaries expand. A Rec.709↔Rec.2020 conversion at D65 therefore does not require white-point chromatic adaptation; it requires the appropriate primary/gamut matrix plus transfer conversion.
BT.2020-2 describes UHD system parameters and a relative OETF whose shape is close to BT.709. HDR is specified separately in Recommendation BT.2100, which retains the same wide primaries/D65 but introduces PQ and HLG transfer systems, an HDR reference environment and other parameters. A file tagged with BT.2020 colour primaries does not by itself tell you whether the transfer is PQ, HLG or an SDR-like one. Primaries and transfer are separate parts of the system.
For relative scene signal E, BT.2020 uses E′=4.5E in the linear toe and E′=αE^0.45−(α−1) above the breakpoint. The exact smooth-join solutions are α≈1.0992968268 and β≈0.0180539685. For practical coding BT.2020 allows α=1.099, β=0.018 in 10-bit systems and α=1.0993, β=0.0181 in 12-bit. This is another example of how a 'Gamma' dropdown can hide a specific standardised piecewise transfer function.
The wider the gamut, the larger the perceptual volume that must be quantised. BT.2020 specifies 10- or 12-bit component coding rather than relying on legacy 8-bit Rec.709-era workflows. More bit depth does not make the gamut wider — the primaries already define its geometry — but it reduces quantisation steps and makes large gradients and colour transforms more practical, especially when the material later passes through HDR transfer functions and aggressive grading.
For R′G′B′/Y′ nominal black and peak white, 10-bit coding uses 64 and 940, while 12-bit uses 256 and 3760. Chroma neutral sits at 512/2048 respectively. This extends the familiar 16–235 Rec.709 logic and is not a property of the gamut. Full/data range versus video/narrow range remains a separate metadata/processing decision that must be interpreted correctly.
For non-linear R′G′B′, BT.2020 uses Y′=0.2627R′+0.6780G′+0.0593B′. Those are not the Rec.709 coefficients 0.2126/0.7152/0.0722 because the luma matrix is tied to the colour primaries. If a decoder uses the wrong matrix, hue and saturation errors can look like a 'wrong gamut' problem even when the chromaticity tags themselves are correct.
BT.2020 defines both conventional non-constant-luminance Y′C′BC′R and a constant-luminance variant Y′C C′BC C′RC. The latter is intended to retain luminance information more accurately and can potentially improve coding efficiency for very wide-gamut signals. Ecosystem support and historical production practice, however, have long leaned more heavily on conventional non-constant-luminance behaviour. The existence of two variants shows why the label 'BT.2020 YCbCr' still does not describe the complete matrix path.
BT.2020 allows several chroma-sampling structures. 4:2:0 reduces spatial chroma resolution, but the primaries, D65 and allowed chromaticities remain the same. `10-bit 4:2:0 Rec.2020` therefore combines three independent properties: quantisation precision, chroma sampling and colour space. They cannot be collapsed into a single 'higher/lower quality' ranking.
HDR delivery often signals BT.2020 primaries because the standard provides a common wide-gamut signalling space. The mastering display may physically cover only P3-D65 or a little more, while the actual images may use an even smaller subset. That is perfectly valid: the RGB code space defines a container, not a requirement that every pixel approach its boundaries. Problems arise when a downstream display cannot reproduce a chromaticity and the output pipeline fails to perform controlled gamut mapping.
Input compatibility means the device recognises the signalling and can convert it into its native panel gamut. Coverage is a separate physical property of the display primaries/technology. A television can correctly accept BT.2020/PQ and then tone/gamut-map the content onto a panel closer to P3. Validating an HDR master on a consumer TV therefore requires awareness of the real display colour volume, not just the input-signal label.
Do not choose Rec.2020 as an 'HDR preset'. First identify four layers: are the Input/Output Colour Space primaries BT.2020? Is the transfer BT.2020 SDR-like, PQ or HLG? Is the range video or full? Is the matrix RGB or a BT.2020 YCbCr path? In managed Resolve workflows Colour Space and Gamma are separated precisely because Rec.2020 primaries can be paired with several transfer systems. Output gamut mapping then determines what happens to colours the target display cannot physically reproduce.
In Resolve, `Rec.2020` primarily answers the primaries/colorimetry question. A complete state still needs Gamma/EOTF, range, matrix and target display. Never use the single term Rec.2020 as shorthand for an entire HDR pipeline.
Technical definitions are anchored to standards bodies and primary specifications wherever that distinction matters.
The in-force Recommendation covering primaries, D65, OETF, Y′/chroma equations, sampling and 10/12-bit coding.
The full primary BT.2020-2 specification.
The separate HDR standard: the same wide primaries/D65 paired with PQ/HLG and an HDR system context.
A useful modern comparison reference for sRGB, Display P3 and Rec.2020 gamut geometry.
