In Figs. 11.7 and 11.8, chromaticities are shown for preferred colour reproduction of blue sky, green grass, and Caucasian skin colours (Hunt, Pitt, and Winter, 1974). These results were obtained by making colour photographs of outdoor scenes containing well-defined areas of one of these test colours, and then varying the colours of those areas only. This was achieved by using pairs of opaque masks to obscure either the test part of the picture or the rest of it: by making two successive exposures in register in an enlarger, series of reflection prints for each scene were made in which the colour of the area of sky, grass, or skin was var- ied, but the colour of the rest of the picture was kept constant; and, by using the masks with pairs of slides projected on a screen, a similar result was obtained for transparencies. The colour of the sky, grass, or skin area was varied by covering that area with uniform pale colour-filters when enlarging or while projecting; in this way, these areas retained their inher- ent variety of tones and colours, and only the overall average colour was altered; neutral filters were used to control the luminances of the two parts of the picture.
Each reflection print and projected picture thus obtained was judged by a panel of observers for the quality of the colour reproduction of its blue sky, green grass, or Caucasian skin. The average chromaticity and relative luminance of each of these colours was then measured and correlated with the observers’ judgments to obtain the results shown in Figs. 11.7 and 11.8. For the reflection prints, the judgments were made in typical indoor daylight, and the
colorimetry was evaluated for Standard Illuminant C. For the projected transparencies, the colorimetry was evaluated for the actual projector illuminant. The results of a similar invest- igation on the preferred reproduction of Caucasian skin on a typical television display (Novick, 1972) are also shown in Fig. 11.7. The broken lines in the figures indicate areas of chromatic- ity giving acceptable colour reproduction. Chromaticities achieved by a negative-positive sys- tem of colour photography used for producing reflection prints are also shown in Fig. 11.7 (); and chromaticities achieved by two different reversal films used for producing transparencies are shown in Fig. 11.8 (and ).
It is clear from Fig. 11.7 that, for the reflection prints, the preferred skin colour (D) lies, as expected, on the yellowish (sun-tanned) side of typical average real skin (0) (Thomas, 1973), but the difference is small; and the preferred grass colour (D) lies on the yellowish side of typical average real grass (0) (Thomas, 1973), but again, the difference is small. The chro- maticities for real skin and grass lie within the area of acceptable colour reproduction, and this suggests that, for these colours, colorimetric and preferred colour reproduction are sim- ilar (although the relative luminances are rather different for grass). But, for the blue sky colour, although the dominant wavelength of the preferred (0) and real (0) (Hendley and Hecht, 1949) colours are closely similar, the preferred colour has an appreciably higher purity. The preferred skin colour on television (•) has a dominant wavelength similar to that of real (Illuminant C) skin (0), but is of considerably higher purity; this is partly because the associated subjective neutral point (•) was displaced towards yellow, but perhaps also partly because of a desaturating effect of the dim surround in which the television display was viewed.
The preferred colours shown in Fig. 11.8 for the projected pictures are displaced in an orange direction relative to those for the reflection prints; this is because of visual adaptation to the tungsten light of the projectors: the point marked ‘Grey’ was the chromaticity that appeared neutral to observers in the viewing conditions used for the projection, and it is shifted in the orange direction from points representing daylight (such as SC). The point
marked ‘I’ shows the chromaticity of the open-gate light from the projectors and its displace-
ment from the ‘Grey’ point indicates that the adaptation to the projector light was not quite complete. In Fig. 11.9 the main results of Figs. 11.7 and 11.8 are shown together, and the shift to more orange chromaticities in the case of the pictures projected with tungsten light is very apparent. It is interesting also to note that there is a small area of chromaticity that is accept- able as grass in reflection prints seen in daylight and as blue sky in tungsten-light projection. If the chromaticities of the real colours of Fig. 11.7 could be converted into the corresponding chromaticities for the viewing conditions used in obtaining the results of Fig. 11.8, it would be possible to say if the preferred colour reproduction in the projected pictures involved any sub- jective distortion of the colours. Compared to the chromaticity of a typical sample of real Caucasian skin colour illuminated by the light of the projector (0), the preferred colour has an appreciably higher purity, and this is perhaps to offset a desaturating effect of the dark sur- round to the projected picture. The results for the two films are within acceptable areas, except in one case.
The preferred relative luminances found were, no doubt, affected by the relative luminance corresponding to the apparent white level, and this level may be expected to vary from scene to scene and from area to area within a scene. This makes the specification of preferred relative luminances in pictures a rather complicated matter.
Comparison of the results shown in Figs. 11.7 and 11.8 with those obtained in earlier investigations (MacAdam, 1951; MacAdam, 1954; Bartleson, 1959; Bartleson and Bray, 1962) suggest that the latter may have been influenced to some extent by the generally lower levels of colour purity available in the systems then used. Results for preferred colour reproduction may, therefore, be influenced by the nature of the reproduction system used, and it also seems likely that they may be influenced by the particular form of the viewing conditions, and by cul- tural, ethnic, and psychological features of the observers: results such as shown in Figs. 11.7
Fig. 11.9. The main results of Fig. 11.7 (full lines) and of Fig. 11.8 (broken lines) shown together.
and 11.8 can therefore only be regarded as examples and not as definitive for particular applica- tions. The naturalness of an image has also been shown to be an important factor affecting the qualities of reproductions (Yendrikhovskij, Blommaert, and de Ridder, 1999).
Some commercial processes are run at higher contrasts than would be indicated by corres- ponding colour reproduction and possible reasons for this include the following (see Sections
6.4 and 6.5). When the original is a scene lit by natural daylight, the reproduction usually has a much lower luminance, and this lower luminance results in some reduction in perceived contrast. The lower luminance also results in lower brightnesses and colourfulnesses, and raising the contrast may result in images which appear more like the original scene. There is also evidence that memory colours tend to be more saturated than original colours, and rais- ing the contrast increases colour saturation. And, in pictures, the effects of atmospheric haze tend to be discounted less than in real scenes.
Preferred colour reproduction is an appropriate aim for portraiture and for reflection prints for the consumer market.
11.11 DEGREE OF METAMERISM
In the case of colorimetric or exact colour reproduction, the degree of metamerism can be assessed by a direct comparison of the spectral reflectances (or relative power distributions) of the original and reproduction. But in the cases of equivalent, corresponding, and preferred colour reproduction, the colours in the reproduction must, in general, be physically different from those in the original: hence there must always be some metamerism. However, some reproduction colorants will tend to produce corresponding colours with greater degrees of metamerism than others, and some means of assessing this would be desirable. For this pur- pose it would probably be good enough to assess the degree of metamerism for an appropriate
colorimetric colour reproduction situation, and to regard the results as indicative of the degree of metamerism for the other cases. Thus, for a picture of a sun-lit outdoor scene projected with a dark surround by tungsten light, the degree of metamerism could be assessed by comparing the spectral reflectance curves of various original colours with those of dye-concentration combinations in the film that are metameric matches to them for tungsten light.
11.12 CONCLUSIONS
Spectral colour reproduction (equality of spectral reflectances or of relative spectral power dis- tributions), is a desirable objective in proofing systems and in the duplication of images. Although not attainable in most other imaging situations, it provides a useful basis for deter- mining the degree of metamerism of reproduction systems; this is an important feature in the mail-order catalogue business.
Colorimetric colour reproduction (equality of chromaticities and relative luminances) is a useful criterion when the original and reproduction have the same viewing conditions and use illuminants of the same colour, and is applicable to colour photocopying.
Exact colour reproduction (equality of chromaticities, relative luminances, and absolute luminances) ensures equality of appearance for original and reproduction if the viewing condi- tions are the same for both, and is applicable to virtual reality systems.
Equivalent colour reproduction (chromaticities, relative luminances, and absolute lumin- ances such as to ensure equality of appearance) can allow for all effects of viewing conditions, and is applicable to marketing on the web.
Corre