Colour vision and colour science for photographers

Colour management exists because colour is not a property of objects alone. It is produced by light, modified by surfaces and inks, captured by sensors, reproduced by displays and printers, and finally interpreted by a human visual system that adapts constantly to its surroundings. A photographer who wants a print to match the screen, a client's product to look right in a catalogue, or a series of prints to look consistent on a gallery wall needs to understand each of these steps. This course takes you from the science of colour through the ICC colour management system, monitor calibration, viewing conditions, profiling, soft proofing, printers, inks and papers, to making and preserving fine prints. This first lesson lays the foundation: how light and surfaces create colour, how human vision works, how colour is measured, the CIE colour spaces that make colour management possible, colour difference and Delta E, and the perceptual effects that make colour matching harder than it looks.

Light, surfaces and colour

Visible light is electromagnetic radiation with wavelengths from roughly 380 to 730 nanometres. A light source has a spectral power distribution (SPD): the amount of energy it emits at each wavelength. Daylight has a smooth, broad SPD; incandescent bulbs emit more energy toward the red end; fluorescent and LED sources often have peaks and gaps. When light falls on a surface, the surface reflects a fraction of the light at each wavelength, described by its spectral reflectance curve. A red apple reflects much of the long-wavelength light and absorbs most of the short wavelengths. The light that reaches the eye, or a camera, is the product of the source SPD and the surface reflectance at every wavelength. This is the first reason colour management is complex: the same print looks different under different light, because the light it reflects changes.

Here \(E(\lambda)\) is the illuminant's spectral power, \(R(\lambda)\) the surface reflectance and \(S(\lambda)\) the reflected light reaching the viewer at wavelength \(\lambda\).

How human vision sees colour

The retina contains rods, which handle low-light vision and do not distinguish colour, and three types of cone cell, sensitive to long (L), medium (M) and short (S) wavelengths, sometimes loosely called red, green and blue cones. Every colour sensation is reduced to three numbers: the responses of the three cone types. This trichromacy has a remarkable consequence: two lights or surfaces with completely different spectra can produce identical cone responses and therefore look identical. Such pairs are called metamers, and metamerism is what makes colour reproduction possible at all, since a display with three primaries or a printer with a handful of inks can never reproduce the spectrum of a real flower, only a metameric match to it.

Metamerism also creates problems. A match that works under one light may fail under another, because the two spectra respond differently when the illuminant changes. This is called illuminant metamerism, or metameric failure, and it explains why two prints that match in the studio may look different in a gallery lit by tungsten spotlights, or why a dyed fabric may not match its photograph under shop lighting. Observer metamerism is related: people's cone sensitivities differ slightly, so a match that looks perfect to one person may not to another, a problem that has grown with wide-gamut displays with narrow-band primaries.

About 8 % of men and 0.5 % of women of Northern European descent have some form of colour vision deficiency, most commonly a reduced sensitivity in the L or M cones. Photographers with colour vision deficiency can still manage colour reliably by relying on measurement, numbers and calibrated tools rather than visual judgement alone.

Measuring colour: from spectra to numbers

Colour science needed a way to describe colours numerically, independent of any particular device. In 1931, the Commission Internationale de l'Éclairage (CIE) defined the standard observer: a set of three colour-matching functions, derived from experiments in which observers matched spectral colours with mixtures of three primaries. Integrating a light's spectrum against these functions gives three numbers, the tristimulus values X, Y and Z. Y corresponds to luminance, perceived brightness. Any two stimuli with the same XYZ values match for the standard observer. A 1964 supplementary observer, using a 10-degree field of view instead of 2 degrees, is used in some industries.

Instruments measure colour in two main ways. Colorimeters use three or more filtered sensors that imitate the colour-matching functions; they are fast and inexpensive and are used to calibrate monitors. Spectrophotometers measure the full spectrum in narrow bands, typically every 10 nanometres, and calculate XYZ for any chosen illuminant and observer; they are used to profile printers and measure prints, and can also measure displays.

Chromaticity and the CIE diagram

Dividing X and Y by the sum X + Y + Z gives the chromaticity coordinates x and y, which describe a colour's hue and saturation without its brightness. Plotting them produces the familiar horseshoe-shaped CIE 1931 chromaticity diagram, with pure spectral colours around the curved edge and the line of purples joining the ends. White points such as D50 and D65 sit near the middle. Device gamuts, the ranges of colours a display or printer can reproduce, appear as triangles or irregular shapes inside. The diagram is excellent for showing gamuts and white points, but it is not perceptually uniform: equal distances in different regions do not correspond to equal perceived differences, with the green region greatly exaggerated.

CIELAB: a perceptual colour space

To describe colour differences meaningfully, the CIE introduced CIELAB (L*a*b*) in 1976. L* represents lightness from 0 (black) to 100 (diffuse white); a* runs from green (negative) to red (positive); b* runs from blue (negative) to yellow (positive). CIELAB is calculated from XYZ relative to a reference white, usually D50 in graphic arts and photography, using cube-root functions that roughly match the eye's response to lightness. It is approximately perceptually uniform, so similar distances represent similar perceived differences across most of the space. CIELAB is the language of colour management: printer profiles, measurement tools and quality standards express colours as Lab values. The related LCh form describes the same colours as lightness, chroma (colourfulness) and hue angle, which is often more intuitive: a skin tone might be described as L* 65, C* 25, h 55°.

Here \(X_n, Y_n, Z_n\) are the tristimulus values of the reference white and \(f(t) = t^{1/3}\) for most values, with a linear segment near black.

Colour difference and Delta E

Delta E (\(\Delta E\)) measures how different two colours are. The original 1976 formula is simply the straight-line distance between two points in CIELAB:

Because CIELAB is not perfectly uniform, especially for saturated colours and blues, later formulas weight the differences in lightness, chroma and hue. CIEDE2000, usually written \(\Delta E_{00}\), is the current standard in graphic arts and display calibration; it is more complex but agrees better with what people see. As a rough guide, a \(\Delta E_{00}\) below 1 is generally imperceptible in side-by-side comparison, 1 to 2 is perceptible only to trained eyes, 2 to 3.5 is noticeable on close inspection, and above 5 colours look clearly different. Tolerances depend on context: a product colour on a website, a reproduction of a painting and a neutral grey in a black-and-white print each need different precision.

Chromatic adaptation and appearance

Human vision adapts to the colour of the light around us. A white sheet of paper looks white under tungsten light and under daylight, even though the light it reflects differs greatly, because the visual system rebalances its sensitivity, much like automatic white balance. Colour management uses mathematical chromatic adaptation transforms, such as the Bradford transform, to convert colours between different white points, for example from a display's D65 white to the D50 white used in print workflows, so that a colour appears the same to an adapted observer under each condition.

Many other appearance effects complicate matching. Simultaneous contrast makes a grey patch look darker on a white background and lighter on a black one, and makes colours appear shifted away from their surroundings. Brightness adaptation means a print that looks right in a dim room looks dark in bright light. The Hunt effect makes colours appear more colourful at higher luminance, and the Stevens effect makes contrast appear higher in brighter light, which is why prints viewed under dim light look flat and muddy. These effects explain why viewing conditions (Lesson 5) are as important as accurate devices, and why the same image may need different adjustments for a backlit screen, a print in a gallery and a print in a living room.

Why photographers need colour management

Every device in a photographic workflow speaks its own colour language. A camera records colour according to its sensor's spectral sensitivities; a monitor displays colour according to its primaries, white point and tone response; a printer reproduces colour according to its inks, paper and driver settings. The same RGB numbers produce different colours on different devices. Colour management translates between them through a device-independent reference, CIE XYZ or CIELAB, using profiles that describe each device. The result is not perfect matching, which is impossible when devices have different gamuts and media look different from screens, but predictable, consistent and controllable colour. The next lesson explains how the ICC system makes this translation work.

Colour and photography: a brief history

Photographers have wrestled with colour reproduction since the first colour processes. James Clerk Maxwell demonstrated additive colour photography in 1861 by projecting three black-and-white images taken through red, green and blue filters. The autochrome plates of the Lumière brothers, introduced in 1907, used dyed potato starch grains as tiny colour filters. Kodachrome (1935) and later colour negative films brought colour photography to millions, with each film stock rendering colour in its own characteristic way. Darkroom colour printing required careful filtration and test prints to match negatives. Digital imaging brought new possibilities and new problems, which colour management was created to solve.