Interesting reading from:
(broken link removed to http://www.energy.ca.gov/2012publications/CEC-400-2012-016/CEC-400-2012-016-SF.pdf):
Color Rendering
Color rendering is a measure of how true the colors of objects look under a light source, as compared to how they look under a reference light source of the same (or similar) color temperature. The two most common reference sources are an incandescent lamp at 2700K and a typical spectrum of daylight at 6500K (known as “D65”).
The most common metric of color rendering is the Color Rendering Index (CRI). A CRI of 100 indicates that a light source renders a particular palette of eight colors in exactly the same way as the reference light source to which it is being compared. A light source with a CRI significantly less than 100 typically makes one or more of the eight standard palette colors appear more gray (less saturated) than the reference source, although a low CRI can also arise because a light source renders certain colors as being too saturated. An incandescent lamp is considered the ”reference” light source for all lamps with color temperatures less than 4000K, so by definition incandescent lamps achieve a CRI of 100 because they are being compared to themselves.
It is important to realize that color rendering is not necessarily a measure of “naturalness” or “saturation” or “quality”. The CRI metric was developed for industrial purposes, when color checking was carried out by human observers, to prevent observers confusing one color with
another (“metamerism”). The CRI metric was not optimized to describe viewers’ subjective preferences for color balance or saturation. Other metrics have been and continue to be explored to quantify viewers’ subjective preferences for different light sources and for the colors of rendered objects, though none has replaced CRI as the primary metric. In particular, the National Institute of Standards and Technology (NIST) is continuing work on the Color Quality Scale (CQS), though at the time of writing there is no specific schedule for development of a new metric.
Despite the various minor shortcomings of the CRI metric that have been noted by researchers, it is a reliable and meaningful scale toward the top end (that is, at values approaching 100 CRI). This is because the only way to achieve such high values is by using a nearly continuous spectrum, by using multiple phosphors or LED chips to give almost‐equal weighting to all spectral colors. Natural daylight also has a continuous spectrum. Therefore, a light source that approaches 100 CRI will be perceived by consumers as being colorful and natural, as well as rendering colors accurately.
The images in Figure 4 demonstrate the difference between low CRI and high CRI. In the picture on the left, which represents a 50 CRI light source, the pink, red, and orange pencils are hard to distinguish from one another. In the image on the right, representative of a 90 CRI light source, the colors are much more easily distinguished. (These images are for illustration only and should not be taken to accurately represent the actual appearance of objects.)
Figure 4: Comparison of Color Rendering Index of Different Light Sources
Source: Lighting Matters’ LED Blog, lightingmatters.com.au/blog/ledlight-quality-cri/
In addition to quoting CRI, LED manufacturers have begun to quote the performance of their products in terms of how they render a ninth color (“R9”—a saturated red color). This is, in part, because a high R9 value indicates improved rendition of important common materials (skin tones, earth tones, woods and vibrant red colors). It is also because manufacturers wish to distinguish “good” LEDs from “bad” LEDs, and from typical triphosphor T8 lamps, which do not render the R9 color well. R9 performance is defined on the same scale as the CRI, although values are often much lower, or even negative, because manufacturers have historically not optimized their lamps for R9 performance. Typically the R9 value for T8 lamps is 10‐20, whereas LEDs are currently available that achieve over 50.
Because CFL program evaluations did not identify color as a key factor, improved color performance was not demanded or encouraged and was therefore abandoned in favor of lower cost. Innovation and new investment in color were hard to justify and largely not pursued by mainstream manufacturers.
However, color rendering has become a more widely discussed lamp attribute, and some manufacturers in the blossoming LED lamp industry are already striving for improved CRI in their products. As of May 30, 2012, the U. S. Department of Energy’s SSL Lighting Facts product database contained 123 replacement lamp products with a CRI of 90 or better. Because the LED market already includes a large number of products with high CRI, and because color quality has been identified as a key opportunity for improvement, the specification includes a minimum requirement for color rendering index.
