Core concept
A complete explanation of colour spaces: primaries, gamut, white point, transfer functions, encoding and transforms so Rec.709, sRGB, P3 and Rec.2020 stop looking like arbitrary names.
A colour space is an agreement about the colour meaning of numbers. An RGB system needs at least defined primaries and a white point; a real video or image signal also normally requires a known transfer function and encoding context.
Quick reference
They define the geometry of an RGB gamut.
For example D65 or D60.
Gamma, log, PQ and HLG are different classes of functions.
Gamut is part of a colour space, not a complete synonym.
In practical post-production terminology, a colour space links digital coordinates to measurable colours. For an RGB system, the key geometric elements are the primary chromaticities and the white point. But interpreting a stored video signal requires more: the transfer function, nominal range, matrix/YCbCr context and sometimes ICC metadata. One space name therefore does not always fully describe a file.
Red, green and blue primaries have specific CIE xy coordinates. Connect them on a chromaticity diagram and you obtain the triangle of possible positive RGB mixtures. Rec.709/sRGB, Display P3 and Rec.2020 differ primarily in the positions of these points, so the same RGB triplet can represent different chromaticities.
A gamut is the set of colours a system can represent under defined constraints. For ideal RGB systems we often discuss the triangular gamut on CIE xy, but a real device gamut also depends on luminance and physical display behaviour. A wider gamut does not automatically mean a better image; it offers more potential range while increasing the demands on transforms and monitoring.
RGB = 1,1,1 is intended to correspond to a defined reference chromaticity. Rec.709, sRGB, Display P3 and Rec.2020 normally use D65; ACES uses a reference white near D60; cinema P3 uses a different white convention. Converting between systems with different white points may require chromatic adaptation.
Primaries and white point describe colour geometry but do not tell you what a numerical value of 0.5 means in terms of light. That is the role of the transfer function. Different pipeline stages use OETF, EOTF and OOTF concepts; camera log, the sRGB curve, gamma-style EOTFs, PQ and HLG solve different problems. This is why CST exposes Colour Space and Gamma as separate fields.
Rec.709 can be encoded as YCbCr 4:2:0 10-bit video range, RGB full range or a floating-point buffer. These can describe the same colour system while storing numbers very differently. Codec, bit depth, chroma subsampling and range belong to encoding and transport; they do not replace a colour-space description.
A colour-space name defines how coordinates should be interpreted. If a Rec.709 signal is merely labelled Rec.2020 without conversion, the system does not create new real colours; it reads the same numbers in a different coordinate system and the image becomes wrong. A proper conversion requires a transform, not just metadata relabelling.
Camera space conveniently describes a specific sensor and encoding. A working space should provide headroom for grading and predictable math. An output/display space constrains the result for a particular display or delivery standard. A good colour-managed pipeline does not force one space to perform all three roles equally.
A correct transform accounts for source primaries, source transfer, target primaries, target transfer and, where needed, white-point adaptation, tone mapping and gamut mapping. CST, ACES Transform, RCM and ICC pipelines automate different parts of this job. A LUT can be part of the solution but does not automatically make a pipeline correct.
RGB is a colour model. A gamut is the available range of colours. A colour space defines a particular coordinate system and reference conditions. An ICC profile describes a device or standard space together with transforms used for colour management. Casual speech often mixes these terms, but pipeline diagnosis requires keeping them separate.
Input Colour Space answers 'which RGB geometry are these numbers in?', Input Gamma asks 'how are they encoded relative to light?', Output Colour Space asks 'which RGB geometry should they become?', and Output Gamma asks 'how should the result be encoded?'. Once those four answers are known, most of CST stops being mysterious.
Before a CST, LUT or manual transform, write down four states: source primaries/space, source transfer/gamma, target space and target transfer. Only then choose the Resolve menus.
Technical definitions are anchored to standards bodies and primary specifications wherever that distinction matters.
