What is colour? From light to perception.

Physically, a source emits energy across wavelengths. Human vision responds only to a small region we call visible light. It is more useful to think in terms of spectral power distribution, or SPD, than a single number describing the 'colour of a lamp'. Two sources with similar correlated colour temperature can have very different SPDs and can render materials and cameras differently.

Monochromatic light can be associated with one wavelength, but most real colours are produced by mixtures of many wavelengths. Purple colours do not correspond to a single spectral wavelength at all; they arise from combinations of long- and short-wavelength energy. That is why the simple formula 'colour equals wavelength' is too crude for colour science.

A source is a physical lamp or emitter. An illuminant describes a lighting spectrum, including a standardised one. A white point reduces that description to a reference chromaticity; it no longer contains the whole spectrum. D65 can denote a standard daylight illuminant, while in an RGB specification it usually identifies the corresponding white point. A lamp labelled “6500 K” need not reproduce the D65 spectrum.

For a simple opaque, non-fluorescent surface, the reflected spectrum is proportional to illumination E(λ) multiplied by reflectance ρ(λ), with geometry accounted for. This is a useful starting model, not a complete law for all materials: transmission, specular reflection, fluorescence and emission require additional descriptions. Saying that a material “only filters light” without qualifications is therefore misleading.

After light interacts with a material, a spectrum reaches the eye and the visual system converts it into a small set of receptor responses. The brain also uses surround, adaptation and relative brightness and colour relationships. In a strict sense, colour is therefore not just a physical spectrum but a perception produced for an observer under defined conditions.

If two different spectral distributions produce sufficiently similar tristimulus responses, an observer may perceive them as the same colour. These matches are called metamers. Colour reproduction fundamentally relies on this: a display does not reproduce the spectrum of a real object, but its RGB mixture can trigger a similar visual response.

A sensor measures light through its spectral sensitivities and filters. RAW stores sensor data together with supporting metadata; it is not the same as a finished RGB image encoded with a log curve. Decoding and processing establish the chosen colour representation. Two cameras may render a fabric or LED source differently even with matching white-balance settings.

A display mixes its own red, green and blue primaries. Their spectra are normally very different from those of the original object, but a correct colour pipeline makes them produce the required CIE tristimulus values and a convincing visual match. This is where RGB primaries, white point, gamut, transfer functions and colour spaces enter the story.

For post-production, keep one chain in mind: illuminant → spectral scene → camera response → encoded signal → working colour space → creative grade → output transform → display emission → observer. An error at any stage may look like 'wrong colour' even though the underlying cause is completely different.