CIE Division 8 - TC8-03: Survey of Gamut Mapping Papers

Viggiano & Wang (1992)

The aim set out in this paper was to compare various lightness mapping techniques both in XYZ and in CIELAB, which was done by looking at the following methods:
 

(a) Compress Y and keep X and Z constant

This was rejected as it resulted in a shift towards yellowish green.
 

(b) Compress Y and keep x and y constant

Even though hue and grey balance were maintained, chroma was increased and this was found to be particularly unacceptable for dark colours - therefore this method was also rejected.
 

(c) Compress L* and keep C* and hab constant

The compression suggested for L* is a linear compression in Bartleson and Brenemanís Darkness (V) (Bartleson and Breneman, 1967). The compression is carried out using the following formulæ:
 
Vr = TCR * Vo

V = 1.16 - 0.175*[100*(Y/Yn) + 0.6]0.41 (dim surround)


Vr and Vo are the darkness values for the reproduction and original respectively and TCR is the tone compression ratio (which should be the ratio of the darkness range of the reproduction and the original). Formulæ for calculating darkness values for other backgrounds are also given in the paper.

Even though this approach results in a reduction of saturation, it yields good results, as the constancy of chroma has already been identified by some earlier work (Pobboravsky et al., 1971) as an important condition for good colour reproduction.
 

(d) Compress L* and C* equally and keep hab unaltered

Here the same compression ratio is applied to chroma and darkness, i.e.
 
Cr = CCR *Co

CCR = TCR


Here CCR is the chroma compression ratio, Cr and Co are the C* coordinates of the reproduction and original respectively. Note, that this technique is based on earlier work (Gordon et al., 1987).
 

(e) Compress L* and C* independently and keep hab unaltered

What CCR and TCR values give best results was experimentally tested and the best results were achieved by the following relationship between the two ratios (i.e. a chroma compression half way between the tone compression ratio and unity):
 
CCR = 0.5*(1 + TCR)


The experimental evaluation of the above methods found that the independent compression of lightness and chroma with an unchanged hue angle (method (e)) gave the best results. However, the experimental results need to be treated with caution, as the effects of the different compression methods were simulated on the monitor, rather than comparing the two media for which the gamut mapping was intended. Furthermore it is not clear from the description of the experiment what images were used, especially what their characteristics were in terms of chroma range, which might also influence the results. Nonetheless, the paper gives a good overview of a range of possible mapping techniques with some useful suggestions as to compression ratios.
 

Last updated: 17 August 1999 by Jan Morovic