Wolski, Allebach & Bouman (1994)

Here a system is proposed which combines some of the techniques suggested by previous authors - i.e. soft clipping (Stone and Wallace, 1991) and separate mapping algorithms for different parts of colour space (CARISMA, 1992; Haneishi et al., 1993). The algorithm, which is implemented in the CIELUV colour space and maps between the original image and reproduction medium gamuts, consists of the following steps (Figure 1):

1. map source colours into smallest rectangle containing the target gamut in a given plane of constant hue by soft clipping saturation (chroma), shifting and then soft clipping lightness (i.e. use 75 per cent between L*=50 and maximum L* as cut off point in soft compression and clip colours with too low L*, after the whole range was shifted in an unspecified manner by between -5 and 2 L* units depending on the target gamut)
2. divide input gamut into three regions: cylinder around neutral axis (radius between 0.3 and 1 saturation units), upper chromatic region (L*>50) and lower chromatic region (L*<50)
3. compress saturation and maintain lightness for the neutral cylinder (how it is done is not specifiedÖ)
4. maintain saturation and compress lightness for upper chromatic region
5. map lightness and saturation simultaneously in a ratio of two lightness units to one saturation unit (however, this ratio can be different for some images)

The method proposed in this paper recognises the need for different kinds of mapping for different regions. However, the techniques suggested for the actual mapping are in some cases not particularly well-defined.