The color rendering index (CRI) is a value which expresses the degree of reproducing color of a sample to a reference illuminant. In various fields of application in video and still imaging it is necessary to reproduce colored targets. The color of an object is determined by three properties: the illuminant, the object’s reflectance and the observer. Regarding to the ability of color reproduction of different illuminants, we want to explain the meaning of the color rendering index in this TechNote.
The attempt of the CRI is to describe this illuminant’s color reproduction ability in just one value to make it easily comparable to others. The task is to show how good the test light source fits to the known reference light source and how far colors look the same under these two light sources.
General Color Rendering Index
The general color index (CIE, 1965) is defined by the following formular:
Where d1...8 are the distances in the u-v chromaticity diagram (scaled) of defined colors between the reference illuminant and the sample illuminant. These 8 representing colors have the same spectral reflectance as the Munsell colors 7.5R6/4, 5Y6/4, 5GY6/8, 2.5G6/6, 10BG6/4, 5PB6/8, 2.5P6/8 and 10P6/8.
As a reference illuminant for correlated color temperatures below 5000K a Planckian radiator source is used, above 5000K D-illuminants will be used instead. The correlated color temperature is “[...] defined as the temperature of the Planckian radiator that produces light most closely matching the particular source.”¹ The Ra can be up to 100 (d1...8 = 0) which means a perfect reproduction of these colors. But other colors don’t have to be visually equal under these light sources, which is a point of criticism.
Special Color Rendering Index
Beside the general CRI there’s a special CRI which can be evaluated for individual colors. The formula for the special CRI is:
In this formula di is the distance (in U*, V*, W*) between the reference color illuminated by the nearest illuminant and the sample illuminant.
In general the CRI is an additional information about the used light source and its ability of reproducing color without knowing its specific spectrum. Imagine two illuminants with nearly the same color temperature of about 3200 K, for example tungsten and a warm white fluorescent light, but with different spectrums and abilities to separate colors and as a result a different CRI.
Criticism
There are several points which can be problematic with the CRI for example that only 8 color samples were compared, the used colorspace is obsolete, it can’t be used for light sources that do not have a correlated color temperature and the CRI is not correlated to human vision because all distances are weighted equally in the color space. It’s a challenging task to characterize the light source with just one value, especially if you want to compare the light of these sources by obtaining the color appearance of an object.
The CRI can be helpful but has some disadvantages, because it can not cover how colors can be differentiated under a light source.²
Alternative approach via light spectrum
Other approaches try to characterize a light source by comparing directly the spectrum of the sample illuminant to a target light source³. For this spectral band method the power (mW) ratio of serveral spectral bands (for example from 400-455, 455-510, 510-540 etc.) will be calculated and the excess over tolerances will be summed up to get a figure of merit. This comparison of the spectra will give an idea of the color reproduction ability but is more complex.
_____________________________
¹ R.W.G. Hunt „The Reproduction of Colour”, Sixth Edition 2004, Wiley-IS-&T in imaging science and technology, p. 155
² an other approach: Mark S. Rea*,Jean P. Freyssinier-Nova; Color rendering: A tale of two metrics; Article first published online: 8 APR 2008; DOI: 10.1002/col.20399
³ R.W.G. Hunt „The Reproduction of Colour”, Sixth Edition 2004, Wiley-IS-&T in imaging science and technology, p. 158; Crawford, 1963 a and b; British Standard 950: 1967
ck