07 novembre 2006

Théorie de la couleur - les bases



Excerpted from Philip Ball´s book Bright Earth. The information can be a valuable review for experienced colorists or an introduction to color physics and physiology for others. Particularly interesting is the section on color space as applied to the CIE chromaticity chart.
source: telecine internet group

The light-sensitive entities in the eye come in two classes, distinguishable under the microscope because of their different shapes. They sit in the retina at the ends of millions of filaments from the optic nerve, and they are either rod-shaped or cone-shaped. There are 120 million rods and 5 million cones in each human retina. Most of the cones are located in a depression of the retina called the fovea centralis, which lies at the focal point of the eye´s lens. This little pit is devoid of rods, which outnumber cones everywhere else on the retina.

Rods and cones stimulate nerve signals when they are struck by light. The rods absorb light over the entire visible spectrum, but do so most strongly (that is, the chance of the light being absorbed is greatest) for blue-green light. Absorption of light by a rod triggers an identical neural response regardless of the weavelength. So rods do not discriminate between colours, but only between light and dark. They are extremely sensitive, and are the main light receptors that we use in very dim illumination such as starlight. This is why it is hard to identify colours under such conditions. Because their response is greatest for blue-green light, objects that reflect these wavelengths (such as leaves) appear brighter than red objects at night.

In bright sunlight, the colour-sensitive cones supply the visual signal to the brain. Under these conditions, the rod cells are 'bleached' -- saturated with light and unable to absorb photons. Only when the bright light is shut off can the rods relax back to their initial state, ready to absorb photons and trigger nerve impulses. This relaxation takes many minutes, which is why we gain night vision only gradually after leaving a brightly lit building. If we are outside at dusk, night vision takes over quite smoothly as the sun´s last rays disappear. The differing colour sensitivity of rods and cones results in a change in the perceived intensity of blue/green objects relative to red as twilight deepens. This effect was first clearly identified in 1925 by the Bohemian pysiologist J.E. Purkinje, although artists had noticed it previously.

The hypothesis for colour vision was verified by experiments in the 1960s which measured the absorption properties of single cones and confirmed that they fall into three classes with different colour sensitivity. The blue-light cones are the least sensitive, which is why fully saturated blue looks relatively dark. Blue´s late historical arrival as a true colour, as opposed to a kind of black, is thus ultimately for biological reasons. [editor´s note: the history of color is fascinating and is well-researched by Mr. Ball in the rest of his book.]

The overall sensitivity of the eye to the colours of the spectrum is the sum of the responses for all three types of cone, and it rises steadily from red to yellow and then falls off steadily from yellow to violet. So yellow is percieved as the brightest colour. The yellow band in a rainbow stands out not because it is more intense (not, that is, because there are more yellow photons than others) but because the yellow photons generate the biggest optical response from the eye. Curiously, yellow is regarded in many cultures as the least attractive colour, and its metaphorical and symbolic associations are often denigrating. It is traditionally the shade of treachery and cowardice, and clothing designers admit that it is a terribly difficult colour to sell. Yellow is popular in China (it is the emperor´s colour, huang); but in the West you had better call it gold.

Each 'seen' colour is constructed in the visual system from the combined stimuli from the three types of cone cell. Red light excites mostly the 'red' cones. But a mixture of red and green rays can stimulate red and green cones in the same ratio as does pure yellow light -- and so the colour sensations are identical. If blue light is added, we see white. (Although the three types of cone are often linked to Maxwell´s primaries of red, green and blue, this is a crude shorthand. Their peak sensitivities are in fact in the yellow, green and violet respectively.)

The rod and cone cells are studded with many thousands of individual light receptors called photopigments. Each of these is a single protein molecule, embedded in the stacked folds of the cell membranes. All photopigments contain a light-absorbing molecular unit called retinal, which has a zigzagging, smeared cloud of electrons very similar to that in the carotenoid pigments of plants. Retinal acts as a kind of switch. There we stand, say, before Yves Klein´s blue sculptures, flooding us with reflected blue light. A blue-sensitive photopigment absorbs a photon of blue light, and in response its retinal unit changes shape from kinked to straight. This enables the photopigment to set in train a sequence of molecular events taht leads to a change in the electrical impulses in the nerve to which the cone cell is attached. Some region in the visual cortex of the brain stirs into life, and we register 'blue'. Where we go from there is our own business. [editor´s note: Yves Klein patented a color, International Klein Blue, for use in paints.]

Measuring colour

The colour wheel has come a long way since Newton. Its most popular modern incarnation is less pleasing to the eye, but a lot more informative: a colour diagram drawn up by the Commission Internationale de l´Eclairage (CIE), starchily called the CIE chromaticity curve. The 'pure' wavelengths of Newton´s spectrum lie on the tounge-shaped periphery, while the colours inside it are the result of various additive mixtures of these rays. Any colour that lies along a line connecting two points on rhe edge may be mixed from those spectral colours. If the line passes through the white region in the centre, the two peripheral colours may be mixed to white. Thus white light can be created from blue and yellow alone (as it is in monochrome television screens), but not from red and green.

The artificiality of the union of red and violet in the 'colour wheel' is emphasized by the flat base of the tongue -- along here the colours are, as Newton confessed, not found in even the finest unweaving of the rainbow´s strands. [editor´s note: the 'colour wheel' was an early attempt to categorize the color spectrum, and though it looked nice on paper, actually presented an artificial connection of colours at the two ends of the spectrum.)

Yet for all its glory, the CIE diagram doesn´t show us all colours. Where is brown? Where is pink? There is clearly a lot more colour space than the mandalas of colour wheels can accomodate.

The defining characterisitic of a coloured material is not whether its hue sits closer to the kingdom of red than of blue or whatever, but what its total spectral composition is: how it absorbs and reflects light across the continuum of the visible spectrum. A colour´s most discriminating signature is thus a wiggly line that traces the variation in intensity of the reflected light as the wavelength varies. The signature of 'pure' white (though not of sunlight) is a straight line: all wavelengths are reflected fully. Black makes the same mark, but at zero rather than full intensity: every wavelength is negated. What, then, is grey? Along with black and white, grey is sometimes classified as an oxymoronic 'achromatic colour' -- we might say that grey has no 'colour' as such, but is more of an intermediary between light and dark. Grey is what we perceive when all wavelengths are absorbed partially, yet more or less equally, from the light. It is, if you will, white light with the volume turned down.

Brown is another difficult one. It sits on the border between a real colour and an achromatic one -- a 'dirty' colour akin to grey. Brown is in fact a kind of grey biased towards yellow or orange. A brown surface absorbs all wavelengths to some extent, but orange/yellow somewhat less than others. Another way of saying this is that brown is a low-brightness yellow or orange, the sensation generated when low-intensity light of these wavelengths impinges on your eye. It is a physiological and linguistic curiosity that, whereas we might classify low-intensity blues, greens, and reds still as blues, greens, and reds, we feel the need for a new basic colour term for low-intensity yellow.

Brown and grey don´t feature on the CIE diagram because it doesn´t show the colours produced by brightness variations. To do that requires a whole stack of CIE diagrams, with the white centre getting progressively greyer. As it does so, the orange/yellow part of the diagram gets progressively browner.

This illustrates the fact that colour space -- the kind of thing you see in trade paint catalogues -- is in fact three-dimensional. The CIE diagram shows just two of the three parameters of colour -- two 'dimensions', portrayed on a flat plane.
One of these is hue, which is what we usually mean colloquially by 'colour'. Strictly speaking, the hue is the dominant wavelength in the colour, and it is what enables us to characterize a colour as basically red, green or whatever. In this sense, the hue of brown is yellow or orange, while grey has no hue -- no dominant wavelength -- and so can be regarded as achromatic. In the CIE diagram, the hue varies around the perimeter of the tongue. Purples lie along the sloping bottom side, between violet at the lower left-hand corner and red at the lower right. The diagram brings home rather forcefully the oddity that in English and most other European languages there is still no generally accepted colour term for the hue between yellow and green, or that between green and blue, even though these occupy appreciable parts of the perimeter. [editor´s note: what about 'tourquoise' 'aqua', 'cyan'? it would seem that there are some terms for the hues between green and blue.]

The second parameter of colour on the CIE diagram is saturation, sometimes caled the purity or (potentially misleadingly) the intensity. This refers to the extent to which white (or black or grey) is mixed in with a pure hue. Roughly speaking, the saturation of a colour varies along the line between the 'pure' hue on the periphery of the diagram and the pure white spot in the centre. Notice, incidentally, how large the white area is: there is a wide range of whites. True white is defined in the CIE scheme as 'equal energy' white, the white obtained from an equal mixture of the three primaries that lie at the extremities: red light of 770-nanometre wavelength at the lower right corner, violet light of 380 nanometres at the lower left, and green light of 520 nanometres at the topmost point of the upper curve. Sunlight lies slightly to the yellow side of true white.

Omitted from the CIE diagram is the third parameter of colour: brightness, which can be crudely considered as the shade of grey the colour generates in a black-and-white photograph. By the early nineteenth century, colour theorists were already beginning to appreciate that flat colour wheels gave only a partial picture of colour space -- a mere slice through the landscape. Some theorists expanded their wheels to include tertiary colours, which are made by mixing the three primaries in different ratios. The German Romantic painter and colour theorist Philipp Otto Runge went further, presenting a color sphere in his book Farben-Kugel (Colour Sphere) (1810) that, roughly speaking, made allowance for variations in brightness of Newton´s spectral colours. The fully saturated primary and secondary colours are situated around the equator of the globe-like sphere. Toward one pole, the colours get progressively lighter; towards the other, darker. So one pole is pure white and one fully black.

Yet even this will not suffice, for it does not properly accomodate independent variations in saturation and brightness: grey appears nowhere on the sphere. Its surface is still two-dimensional, whereas real colour space is three-dimensional. In the early 1900s the American artist and teacher Albert Munsell made one of the first attempts to codify all of this space. Munsell hoped that his scheme would allow him to classify colours perceived in nature so that he could reproduce them accurately on canvas in his studio. His first colour scale was published in 1905, and was later expanded in the Atlas of the Munsell Color System in 1915. The full Munsell scheme is somewhat like a 3D CIE chart, except that the profile is more like a polychromatic spider than a tongue. As in the CIE chart, hue changes around the perimeter while saturation varies along radial lines towards white at the centre. The brightness varies in the vertical direction, as in our hypothetical stacks of CIE charts, so that the central point runs from pure black through grey to pure white.

Munsell updated his colour notation scheme again in 1929, dividing the colour space into discrete blocks that were intended to progress, in any direction, through equal perceptual steps. Careful psychological tests were conducted by the Optical Society of America to try to ensure that Munsell´s colour space was as 'even' as possible.

The Munsell colour scale, in the form of coloured plastic counters or chips, has been used extensively by psychologists and anthropologists conducting research into colour perception. But its value in this arena remains limited by its attempt to impose scientific quantification on concepts of colour that inevitable carry a lot of cultural baggage. John Gage recounts with some glee how Danish antrhopologists arrived on a Polynesian island in 1971 ready to test their Munsell chips on the indigenous people, only to receive the deflating response, 'We don´t talk much about colour here.' The sociologist M. Sahlins expressed the point very neatly in 1976: 'a semiotic theory of color universals must take for "significance" exactly what colors do mean in human societies. They do not mean Munsell chips.'

By the same token, colour does not mean Newton´s rainbow, nor (as the Oxford English Dictionary suggests) a material´s propensity for light absorption, nor a sensation produced by stimulation of the optic nerve. It is all of these things, but to artists they are mere abstractions. Painters need colour to be embodied in stuff, they need to be able to purchase it and get it smeared across their overalls. That is the bottom line, and I would not like to see it obscured (as it sometimes has been) among multi-hued wheels and globes and charts. Painters need paint. Colour is their medium of expression and communication, but to make their dreams visible it needs substance.