GAI和CRI的关系
Color Rendering: A T ale of T wo Metrics Mark S. Rea, * Jean P . Freyssinier-Nova Lighting Research Center, Rensselaer Polytechnic Institute, Troy, NY 12180 Received 2 June 2007; revised 24 August 2007; accepted 5 September 2007 Abstract: It was the best measure of color rendering, it was the worst measure of color rendering. Color render- ing index (CRI) is the most common metric used by the lighting industry to represent the color rendering proper- ties of electric light sources. CRI was intended to charac- terize how ‘‘true’’ or ‘‘natural’’ objects appeared when illuminated by a light source, but was never intended to, for example, represent how well object colors could be differentiated under a light source, another important as- pect of color rendering. Data presented here demonstrate that CRI in conjunction with another measure of color rendering, gamut area index (GAI), is useful at predicting subjective judgments of how ‘‘natural’’ objects appear as well as how ‘‘vivid’’ objects appear, and how well one can discriminate between subtle differences in hue. Nei- ther measure by itself, however, is sufficient for meeting all of the expectations of a light source for providing good color rendering under all viewing conditions. It remains for future research to determine if just two met- rics are sufficient to assure good color rendering from a light source and whether these two metrics (CRI and GAI) are the best for such purpose. In the meantime, CRI and GAI should be used jointly in recommendations as practical, useful, and mutually reinforcing measures of color rendering. The data presented here also demon- strate that total irradiance is important for good color rendering. /C2112008 Wiley Periodicals, Inc. Col Res Appl, 33, 192– 202, 2008; Published online in Wiley InterScience (www.interscience. wiley.com). DOI 10.1002/col.20399 Key words: color rendering; index; gamut area; full spectrum; fluorescent; light emitting diodes; phosphor; correlated color temperature; cool; warm; white; natural- ness; vividness; discrimination; Farnsworth INTRODUCTION P.J. Bouma in 1947 described the aspects of daylight that make it an ideal source of illumination 1: ‘‘It (daylight) displays (1) a great variety of colors, (2) makes it easy to distinguish slight shades of color, and (3) the colors of objects around us obviously look natural.’’ His description is, in fact, an articulation of the important features of color rendering for any light source. In general, a source with good color rendering properties, daylight or electric, should reveal a full range of colors (i.e., hue, lightness, and saturation), should enable good color discrimination between objects of similar spectral reflectance (e.g., pur- ple from maroon), and should not distort colors (e.g., should not overenhance the red appearance of hamburger in a meat case). In the context of Bouma’s early description, color ren- dering should be evaluated in several ways because there are several different aspects of color rendering. More strongly stated, color rendering should not be measured by any single metric. Nevertheless, we presently have only one recognized measure of color rendering in the lighting industry, color rendering index (CRI),* developed in the early 1960s through a collaborative effort among interested scientists and manufacturers. 1,3 These thought leaders developed CRI as a measure of how ‘‘true’’ objects were rendered by electric light sources. 1,3 They assumed that daylight and incandescent light should be reference illuminants in a system of color rendering because these were the most familiar sources (at least for those people in the United States) and should, thereby, render objects ‘‘truly’’ or ‘‘naturally.’’ The general CRI of all electric light sources was defined in terms of the net shift in color space of eight standard reflectances rela- tive to the coordinates of these standard reflectances under the reference light source of the same correlated color temperature (CCT). The general CRI scale was normal- ized to 100 for reference illuminants and set at 50 for a warm white halophosphor fluorescent lamp. *Correspondence to: Mark S. Rea (e-mail: ream@rpi.edu). Contract grant sponsor: US Environmental Protection Agency; Peter Banwell served as project manager. VVC 2008 Wiley Periodicals, Inc. *For the purposes of this report, CRI is used synonymously with gen- eral CRI. 2 192 COLOR research and application Their thinking was that because daylight and incandes- cent light were the most familiar light sources, the larger the shift within the color space of the standard reflectan- ces under the test light source, relative to the chromaticity under the reference source, the less ‘‘true’’ or ‘‘natural’’ objects would appear under that source. Importantly too, these thought leaders recognized that among the limits of CRI, any single measure was incomplete in describing the color rendering properties of a light source. Judd, one of the authors of CRI, developed a ‘‘flattery index’’ in response to the inherent limits of CRI to characterize color rendering completely. 1,3 Judd believed that CRI could characterize how ‘‘true’’ or ‘‘natural’’ objects appeared, but could not characterize how ‘‘flattering’’ or ‘‘vivid’’ objects might appear, particularly the redness and saturation of human skin. 4 Another important aspect of color rendering identified by Bouma, but only sporadically considered, 5–9 is the ability of a light source to reveal subtle differences in spectral reflectance (e.g., purples from maroons). Thorn- ton promoted the concept of gamut area as a measure of ‘‘color discrimination.’’ This concept has not been embraced by the lighting industry 10 but has been the foundation of the display industry because it is important for the accurate reproduction of saturated colors. Interest- ingly, gamut area is also the basis for modern triphosphor fluorescent lamps developed first by Thornton when he was at Westinghouse. 11–13 With the ra…