Colour Vision

Retina receptors

representation of the visual spectrum (monitor etc will distort this a lot!)

Very rough idea of luminance response of the eye. Solid curve is high level (Photopic) and dotted curve is low level (Scotopic). The eye's sensitivity drops dramatically towards the edges of the spectrum.

Rough sensitivity of the cones against wavelength (hand drawn!). These three signals are mixed in a complex way in the brain/eye into what we perceive. For example it has been suggested that nerve signals include: rho-gamma, gamma-beta, beta-rho and differences of these intermediates! Basically there seems to be a redness-greeness signal and a blue-yellow signal as well as brightness.

Basic perceptual attributes

Here we see some main hues.
Note that opposites contrast strongly and adjacent colours would not.

The effect of surrounding colours

The brightness of a surrounding colour affects the preception of the central colour.

here the square on the left looks darker because it is surrounded by a relatively lighter area. So the contrast of an area depends on its surround. This means surround areas affect the overall brightness of a picture for example.

This is also true with brighter colours. The yellow area above right makes the central square seem darker, however it can also make the central area change hue.

in the above example the surrounds are roughly the same brightness (depending on your monitor!) and the central patch is also the same. Notice the disturbing appearance due to the lack of brightness edge around the borders! This can happen when text and background colours are chosen poorly. This emphasises the importance of brightness contrast over colour contrast for clarity.

In general avoid overlaying adjacent colours in the hue circle, to ensure higher contrast and avoid problems with colour-deficient sight.


Our eyes adapt to bright/dark scenes but also to the overall colour of a scene. Try this experiment with your eyes:

Experiment to show adaption and slow recovery

Colour Spaces

"Plain" RGB values do not correspond well to visual stimuli, so colour spaces have evolved to store more psychovisual values. These include XYZ, Luv, Yxy, Lab etc. The newer spaces are more uniform perceptually. The principle attributes of vision are
3d view of colour wheel

XYZ tristimulus values

X,Y,Z values are obtained from three integrals using the CIE 1931 standard observer colour matching functions illustrated roughly below:

Rough idea of the CIE colour-matching functions for the 1931 Standard Colorimetric Observer (hand-drawn)

The XYZ values are obtained from the integrals of these three xyz-bar curves with an illuminant curve and object reflectance spectrum.

This is a chromaticity diagram and shows two colour coordinates x,y. Y correlates approximately with brightness but X and Z are harder to visualise so x,y chromaticity values are based on ratios of the tristimulus values XYZ:

x = X/(X + Y + Z)

y = Y/(X + Y + Z)

z = Z/(X + Y + Z)

because they are ratios x + y + z = 1 so z is not needed if x and y are used. Usually Yxy is quoted for a colour.

Note that around the curved area of the above diagram lies the pure spectral colours. A CRT monitor will occupy a triangle inside its three phosphor colours and can not display all colours. Also note that the space is not uniform: moving a distance in the green area gives a smaller colour difference than the same distance in the blue area for example. So it is difficult to predict colour differences (and quantisation is not efficient).

Note how moving from the centre out increases saturation and rotating around the centre changes hue. These are more perceptually understandable parameters.


There are several more uniform colour spaces but CIELab is popular (and can be tried in Photoshop!). It is based on XYZ and the values for the reference white XnYnZn: Converting the a,b coordinates into polar gives CIE 1976 hue angle and chroma. Euclidean distances in Lab space are more uniform: a distance of 1 is just noticeably different. It is better at predicting colour contrasts than RGB. There are a couple of more complex models which can predict colour appearance taking into account surround colours and adaption factors but these are complex.

See slices through Lab space

Colour "weight" and "feeling"

This is very complex!
Look at a Mondrian painting to see that colours can be balanced by making smaller areas of the "strong" colours such as yellow:

Mondrian. composition with red, yellow and blue (1928)
Note the balancing of the colour areas, the yellow block is smallest for example. Images can look "top-heavy" etc by the placement of colours.

Similarly colours have certain associations:
Blue is "cool"
Redish colours are "warm"
Yellows are "lively" and bright
Greys can be "formal" and dull
Red can be used to attract attention etc.

However cultural differences occur! - a negative red used in a CANCEL makes sense in the west but in eastern cultures read can be positive/affirmative, so there would be a clash.

Look at the EPSON Colour Guide.
Look at further examples of colour design

Further reading

R. W. G. Hunt. Measuring Colour, Ellis Horwood 1987

W. N. Sproson, Colour Science in Television and Display Systems, Hilger, Bristol, 1983

W. S. Stiles and G. Wyszeski, Colour Science, 2nd Ed., Wiley, 1982.

Design for people with colour-deficient sight

Kirk Martinez 2000 (some content from Hunt and the Web!) EPSON Colour guide is copyright Seiko Epson.