More than a tag. A coordinated SDR path.

BT.709 is an agreement on HDTV parameters for production and programme exchange. It links colorimetry to signal structure, picture dimensions and frame rates. It is not a creative look: scenes with very different lighting can both be represented correctly in Rec.709.

This chapter uses BT.709-6, the 2015 edition. In post-production, a “Rec.709 master” usually implies a coordinated chain: colour coordinates, output tone response, levels, metadata and viewing conditions. A filename alone cannot express that agreement.

Separate four layers: RGB primaries and D65 white; transfer encoding; RGB or Y′CbCr representation; digital code range. These properties are related but are not interchangeable.

BT.709 specifies the source OETF; BT.1886 defines reference HDTV display behaviour. BT.2035 addresses evaluation conditions separately. “Everything is set to 709” therefore does not establish what the application is doing or what the monitor is showing.

The CIE 1931 xy coordinates are R (0.640, 0.330), G (0.300, 0.600), B (0.150, 0.060), with D65 (0.3127, 0.3290). They determine a linear RGB ↔ XYZ matrix. Equal linear RGB components point towards D65.

An xy triangle represents chromaticities of primary mixtures, not luminance or the complete three-dimensional colour volume. It does not specify the light associated with code 0.5 or the brightest saturated colour a particular screen can display.

sRGB and Rec.709 share primaries and white. Their linear RGB coordinates require no change of basis when normalization matches. However, sRGB has its own piecewise transfer function, while video must be considered together with its display path.

Finished sRGB artwork already has a display-related meaning. It should not automatically be treated as camera data encoded by the BT.709 OETF. Establish the graphic’s interpretation first, then integrate it into the working pipeline and intended output.

For normalized linear input L, BT.709 specifies V = 4.5L below L = 0.018 and V = 1.099L^0.45 − 0.099 above the breakpoint. This maps light to a nonlinear signal; it is not display behaviour. The standard uses rounded coefficients, which the lab retains instead of smoothing the join.

The calculation teaches encoding. It does not imply that every finished Rec.709 master should automatically be decoded using the inverse OETF: the image may already have been creatively prepared for a reference display.

BT.1886 relates normalized signal V to screen luminance L. Its 2.4 exponent is combined with black Lb and white Lw. With zero black, L = Lw × V^2.4. Nonzero black changes the response, particularly in shadows.

The laboratory fixes white at 100 cd/m² and lets you change black. Its cd/m² readout is an equation result, not a reading of your screen. Checking a physical display requires an instrument and controlled conditions.

An OETF → EOTF chain need not reproduce the original relative light value. Television links a scene to a screen image with contrast, creative preparation and surround in mind. That differs from checking the reversibility of one encoding function.

The experiment displays input L, code V and model display luminance together. Different values of L and Lscreen/Lw do not automatically indicate an error. Applying the wrong function to a signal with an established meaning does.

Colour decisions need a reference view. BT.2035 addresses HDTV evaluation: display characteristics, background, ambient illumination and viewing geometry. Its references include white around 100 cd/m², background relative to display white, and controlled low illumination.

These are decision-making conditions, not a requirement that every audience device have identical brightness. Compensating for an overly bright grading monitor inside the grade transfers that compensation into the file. Check the viewing path before changing the master for each player.

Video may store nonlinear R′G′B′ or the component representation Y′CbCr. BT.709 luma is Y′ = 0.2126R′ + 0.7152G′ + 0.0722B′. Colour-difference components express blue and red differences relative to luma.

Y′ is not the linear CIE Y coordinate: the prime indicates nonlinear signal values. Y′CbCr alone does not specify gamut either. The decoder needs the correct matrix coefficients and primary interpretation.

Nominal 10-bit video RGB/Y′ black and white are 64 and 940; the 8-bit equivalents are 16 and 235. Colour-difference components use a different range. Full range maps normalized values to integers differently. This does not shrink gamut or change the transfer function.

Switch Video/Full and 8/10 bits in the experiment. With matching readback, the normalized signal is nearly preserved; small differences come from rounding. Reading the same integers under a different range without remapping changes levels.

Metadata can identify primaries, transfer characteristics and matrix coefficients separately. The decoder must also understand levels. A tag does not recalculate the signal, but it can change how an application interprets it.

Check both data and description. Files matching in one player may be interpreted differently elsewhere if tags differ or an application assumes defaults. Assigning a desired label cannot replace a missing conversion.

Resolve exposes Rec.709, Gamma 2.4 and specialized interpretation options. Do not treat them as interchangeable because they share a gamut label. Input settings describe the incoming signal; output settings prepare an image for a target.

Rec.709-A is an application option, not a separate ITU recommendation. Its behaviour and macOS viewing settings should be checked against your version’s documentation and a test export. This chapter does not prescribe it as a universal cure for Viewer/player differences.

Define the target and monitoring path first, then identify the clips’ source spaces. In a managed project, transforms can run outside the clip node graph. In a manual chain, a CST must receive the actual gamut and encoding present at its input.

A matching output-gamut name does not guarantee identical tone rendering between RCM and manual CST: tone mapping, gamut mapping and image state matter. After output, check levels, tags, re-import and a controlled external view.

Investigate in order: source → input interpretation → working space → output → levels → metadata → viewing application → screen. Start with a neutral ramp and several colour samples rather than a complex creative shot.

Keep a control export and change one layer at a time. Fixing black while damaging colours does not prove success: one error may have compensated for another. Check tonal steps and coloured regions together.