CIE 228

Abstract

Introduction This report presents the calculation of a self-luminous neutral scale as a complement to the original use of CIE lightness, L*, as defined in the CIELAB and CIELUV colour spaces ( ISO/CIE, 2008; ISO/CIE, 2016), for reflective surfaces. The CIE colour-matching function y(λ) was defined to be the same as the spectral luminous efficiency for photopic vision, V(λ), (Wyszecki and Stiles, 1982). V(λ) was defined based on self-luminous flicker photometry.  According to Wyszecki and Stiles (p.157), in photometry “the Y-tristimulus value in the CIE 1931 system becomes the luminance”.  In applications to objectcolour stimuli, in “most practical situations, only relative spectral radiant power distributions of the given light sources are required”.  Hence, the “Y-tristimulus value of the object-color stimulus,…, defines the luminance-factor of the object color stimulus”.  For Wyszecki and Stiles most practical situations involved reflective objects of colorimetry and isolated coloured lights, not self-luminous electronic displays with high luminance, high resolution and high contrast.  The luminance of light reflected from an object depends upon the reflectance of the object and the luminous intensity of the light from the radiant source that illuminates the object.  The luminous intensity of the source was measured as the Y tristimulus value of magnesium oxide or barium sulphate illuminated by that source.  These highly-reflective materials were later idealized as a perfect diffuse reflector such that all of the energy incident upon the material will be reflected back into the hemisphere above the perfect diffuse reflector.  When the CIELAB and CIELUV colour differences were developed (Robertson, 1977), Semmelroth suggested a pseudo-adaptation transform dividing the luminance of an object, Y, by the luminance of the perfect diffuse reflector (at first designated Y0 and later designated Yn as it is today).  A perfect diffuse reflector would have a reflectance of Y/Yn = 100 % and thus a maximum L* magnitude of 100.  Other tristimulus values, X and Z, are normalized to Y/Yn = 100.  Wyszecki and Stiles (1982) state (pp. 167, 168): “The tristimulus values Xn, Yn and Zn are those of the nominally white stimulus.… Under these conditions, Xn, Yn and Zn are the tristimulus values of the standard illuminant with Yn equal to 100.  ” The point is that, depending upon whether your application requires only relative information or the actual luminance of the visual target , Yn can be thought of as unitless, defined as 100, or alternatively as a maximum luminance of a particular radiant source or reflective object (Pointer, 2017; Fairman, 2017).  The self-luminous neutral scale recommended in this report is in the latter category (requiring actual luminances measured in cd.m−2), whereas standard current practice (media-relative colorimetry) is in the former. The use of CIE lightness, L*, in colour-difference calculations for self-luminous electronic displays is based on a recommendation by Carter and Carter (1983), in which the tristimulus value of the white point, Yn, is set equal to the numerical value of luminance obtained with maximum intensity on all display primaries, and the medium white point is completed by setting the other tristimulus values Xn and Zn to CIE standard illuminant D65, relative to Yn.  This is different from the case of reflective media or isolated coloured lights, where Yn is proportional to the luminance of a perfect diffuse reflector illuminated by a radiant source.  Since the tristimulus values Xn, Yn and Zn represent the medium white point, the calculation of L* is said to be “media relative” (for the definition of "media-relative colorimetry" see Annex A).  In mediarelative colorimetry, L* represents the lightness relative to the range of luminances available from a specific self-luminous medium; this has become common practice (Oleari, 2016; Tooms, 2016).