Fig. 11.2. Relative spectral power distributions of television phosphors: full lines, blue, green, and red sulphide phosphors; broken line, europium yttrium vanadate red phosphor. The use of the lat- ter phosphor in place of the sulphide red phosphor caused colours reproduced by television to become more metameric.
useful, in that it defines the requirement for independence of illuminant colour and of observers’ colour vision; and the extent to which any colour reproduction system is sensitive to these factors can be assessed by considering the effects of specified changes in illuminant or observer.
11.5 COLORIMETRIC COLOUR REPRODUCTION
Observer metamerism cannot be eliminated in practical situations, but it has been found that, if computations are made with the CIE Standard (2°) Observer data, the results usually accord well with assessments made by (non-colour-defective) real observers. It then becomes possible to define colorimetrically the particular metamer in the reproduction that would match any colour in the original. Such metameric matches are characterized by the original and the reproduction colours having the same CIE chromaticities and relative luminances. This is called colorimetric colour reproduction (Clapper and DeMarsh, 1969), which may therefore be defined as reproduction in which the colours have chromaticities and relative luminances equal to those of the original. In the case of reflection prints this would normally imply that the original and reproduction illuminants had the same chromaticities (but their spectral power distributions could be different). The colorimetry is usually carried out relative to a well-lit reference white in the original, and relative to its reproduction in the picture. This procedure makes the relative luminances independent of changes in the intensity of either the original or the reproduction illuminant (or, in television, the luminance of the screen). This is a simplifica- tion that has some limitations, which we shall discuss later, but it enables the usual type of colour-difference formula to be used. Thus, for daylight viewing of reflection reproduc- tions of scenes lit by daylight, the colorimetric evaluations could be made using a daylight type Standard Illuminant, and departures from colorimetric colour reproduction (Pitt, 1967) could be calculated using the colour-difference formula currently recommended by the CIE.
TABLE 11.1
Typical levels of illumination met with in practice
TYPICAL DAYLIGHT ILLUMINATION LEVELS
Bright Sun 50 000 to 100 000 lux
Hazy Sun 25 000 to 50 000 “
Cloudy Bright 10 000 to 25 000 “
Cloudy Dull 2000 to 10 000 “
Very Dull 100 to 2000 “
Sunset 1 to 100 “
Full Moon 0.01 to 0.1 “
Star Light 0.0001 to 0.001 “
TYPICAL ARTIFICIAL LIGHT ILLUMINATION LEVELS
Operating Theatre 5000 to 10 000 lux
Shop Windows 1000 to 5000 “
Drawing Offices 300 to 500 “
Offices 200 to 300 “
Living Rooms 50 to 200 “
Corridors 50 to 100 “
Good Street Lighting 20 “
Poor Street Lighting 0.1 “
However, it must be remembered that, in pictures, the colours of some objects (such as skin, blue sky, grass, foliage, and greys) are more important than others, and errors in some direc- tions (such as hue) are more serious than in others: a distinction therefore has to be made between the perceptibility and the acceptability of colour differences.
Colorimetric colour reproduction is an appropriate aim for colour photocopying
If the appearance of colours were independent of illuminant intensity, then the concept of colorimetric colour reproduction might be applicable to all cases where the original and repro- duction illuminants had the same colour (chromaticity co-ordinates); but the appearance of colours certainly is affected, sometimes quite markedly, by the illuminant intensity, which, as can be seen from Table 11.1, can vary very widely, and hence the achievement of colorim- etric colour reproduction does not necessarily imply equality of appearance of colours in the original and in the picture; moreover, other factors affecting the appearance of colours are also important.
11.6 EXACT COLOUR REPRODUCTION
If, in addition to the chromaticities and relative luminances being equal, the absolute lumin- ances of the colours in the original and in the picture are also equal, we have a situation in which differences in illuminant intensity (or screen luminance in the case of television) have been eliminated (see Section 7.4): this is called exact colour reproduction. Hence the reproduc- tion of a colour in a picture is exact if its chromaticity, its relative luminance, and its absolute luminance are the same as those in the original scene. This would result in equality of appear- ance of the reproduced and original colours providing that the state of adaptation of the eye was the same when viewing the picture as when viewing the original scene; factors that can have an important effect on the adaptation of the eye include the luminance and colour of the surround, the angular subtense, and glare, and only if all these viewing conditions are similar will the adaptation be the same.
Thus, if the reproduction of a certain colour is exact, the observer will only see the same colour as when looking at the original scene, if a number of important conditions are simultan- eously met. In general, there would be a difference in colour appearance: if the viewing condi- tions were not the same for the original object and for the reproduction; or if the observer differed appreciably from the CIE 2° Standard Observer; and, in practice, it is frequently the case that the spectral power distributions of the illuminants are not quite identical to those assumed for calculating the chromaticities and relative luminances (so that colorimetric errors may be present).
Exact colour reproduction is an appropriate aim for virtual reality systems (see Section 5.9).
11.7 EQUIVALENT COLOUR REPRODUCTION
There are many situations where colorimetric and exact reproduction are known to be erro- neous objectives. For instance, if a scene lit by tungsten light is reproduced in a viewing situ- ation in which the ambient lighting is daylight, then colorimetric and exact colour reproduction would both produce results that are too yellow. This situation commonly occurs in colour tele- vision: a studio scene lit by tungsten light, if reproduced on a colour receiver with colorimetric or exact colour reproduction, would look too yellow when viewed in ambient daylighting; this is because the eye would be adapted mainly to the daylight, as a result of its larger area, whereas, in the case of the original, the eye would have been adapted to tungsten light and hence would have had its blue sensitivity increased, and its red sensitivity decreased, relative to its green sensitivity. (The optimum colour balance to choose for a colour television display when viewing it in a variety of ambient illuminant colours is discussed in Section 21.13.)
Because of the effects of the viewing conditions, such as those just described, it is necessary to define a fourth type of objective, equivalent colour reproduction; this is defined as reproduc- tion in which the chromaticities, relative luminances, and absolute luminances of the colours are such that, when seen in the picture-viewing conditions, they have the same appearance as the colours in the original scene.
There are at least three types of effect that are of practical importance in this connec- tion: the effects of differences in colour between the original illuminant and the reproduction illuminant; the effects of differences in intensity between the two illuminants; and the effects of differences in the surround of the original and of the reproduction. The following examples of results obtained by haploscopic matching (see Section 8.10) illustrate these effects.
In Fig. 11.3 the chromaticities of pairs of equivalent colours (Hunt, 1957) are shown for tungsten light (dots) and daylight (arrow-heads); it is seen that, as expected, stimuli have to be bluer in daylight adaptation to elicit the same sensations as in tungsten-light adaptation. Equations relating equivalent colours have been proposed (Burnham, Evans, and Newhall, 1957; Nayatani, Takahama, and Sobagaki, 1981; Hunt, 1998; see Chapter 34).
In Fig. 8.20 the chromaticities of series of equivalent colours are shown for a series of changes in illuminant intensity (Hunt, 1952 and 1953). It is seen that as the illuminant intens- ity is decreased there is a gradual decrease in colourfulness. Fig. 13.7 shows the luminances of a series of equivalent colours for a series of greys from white to black viewed under a range of illuminant intensities (Hunt, 1965a). It is seen that, as the level of illumination drops, the brightnesses decrease, and there is also a slight reduction in apparent contrast (in this figure this could be an artefact of the scale used as ordinate, but other investigations also support this finding (Bartleson and Breneman, 1967a)).
In Fig. 11.4 an example of the effect of the surround is given; equivalent colours were meas- ured for a grey scale seen first with a grey surround and then with a dark surround; it can be seen that (as discussed in Chapter 6) the dark surround has the effect of decreasing the apparent gamma (Hunt, 1965b). A dark surround may also decrease the apparent colourful- ness of colours (Hunt, 1950; Rowe, 1972; Hunt, 1973; Pitt and Winter, 1974; Breneman,
Fig. 11.3. The chromaticities of colours that appear the same in 8.1 cd/m2 of Standard Illuminants A (dots) and C (arrow-heads).