Abstract
The CIE defines ‘colour rendering’ as the effect of a light source on the colour appearance of objects relative to how they appear under a reference illuminant. To quantify the ‘colour rendering’, the CIE has adopted a General Colour Rendering Index (CRI), which is a measure of the average shift of a set of standard coloured samples for test and reference illuminants. Several studies have shown that CRI does not satisfactorily characterise the colour rendering of light sources such as LEDs which have non-continuous spectra with narrow peaks. Although CRI is almost the only criterion used by lighting professionals and the lighting industry, there are currently about a dozen proposals to update, complement or even replace it. Most of them are modifications of the CRI but others, such as the Memory Colour Quality metric (MEQM), have focused on more subjective aspects of lighting colour quality. This metric is based on the importance of familiar objects for colour perception in the natural world, and in this sense is based on the perceived similarity between the apparent colour of an object and its memory colour. So far, there has been no agreement as to which, if any, of these proposed metrics should be adopted by the CIE.
It is my opinion that a decision on a new ‘colour rendering’ metric is needed quickly, for three reasons. First, solid-state lighting is growing rapidly and involves numerous small companies. Such companies rarely have a professional with enough knowledge of colour rendering to judge the relative merits of several different metrics. Second, the challenge in creating LEDs for use as light sources is to provide the highest possible luminous efficacy while achieving the best colour rendering possible. At the moment, the emphasis is on maximising lumens per watt. If we also want to have good colour rendering, a colour rendering metric to act as a target is necessary. Third, LEDs allow great freedom in spectral design so many different forms of white light can be produced. A metric capable of quantifying how well a given spectrum will render colours is essential.
Unfortunately, choosing one metric from all those suggested is not easy. This is because there are at least three aspects of colour rendering that are relevant to light source selection for application: fidelity, discrimination and preference. While in some situations one of these aspects will be dominant over the others, in most applications it is necessary to satisfy all three aspects together. What this means is that the proposed metrics need to be evaluated for their ability to predict the effect of a light spectrum on all three aspects of colour rendering. The metric that shows the most consistent separation between light sources on all three aspects should be adopted by the CIE, lighting professionals and the lighting industry. Unless some authoritative guidance on the best ‘colour rendering’ metric is provided soon, there is a risk that the evident gap between lighting research and lighting practice will widen further.