Beyond CRI: A practical and predictive two-metric approach to color rendering for retail lighting

Transcription

1 IES NY Issues in Color Seminar February 26, 2011 Beyond CRI: A practical and predictive two-metric approach to color rendering for retail lighting Jean Paul Freyssinier and Mark S. Rea Lighting Research Center, Rensselaer Polytechnic Institute Troy, New York, U.S.A. sponsored by 1

2 Acknowledgments NYC IES organizers of Issues in Color Seminar, especially Jason Livingston, Wendy Luedtke,, Dan Rogers, and Meg Smith LRC faculty, staff, and students Sponsors of ASSIST Program 2

3 Specifiers views on color characteristics 3

4 Architectural Lighting Goals Retail lighting has several important goals: Attract attention, generate interest, create a comfortable atmosphere, integrate with store identity, be flexible, facilitate sales, help evaluate products, etc. Color rendering and the color of illumination are two key factors supporting retail lighting. However, presently accepted metrics used to describe color properties often are not predictive of human color perception. 4

5 Specifiers views on color characteristics 5

6 Basic issues Different dimensions of color rendering Naturalness, hue discrimination, vividness, preference, color naming accuracy, color harmony CRI used exclusively by lighting industry Known limitations Poor predictor of preference and color naming accuracy Implications for energy savings and standards 6

7 Basic issues Color rendering is about sources not about what is being illuminated, therefore color rendering can have different meanings. Bouma (1947): suggested that daylight is the best source because can tell differences among colors (hue, lightness, and saturation) can discriminate among subtle shades can make objects appear natural Judd (1967) Flattery index to complement CRI Thornton (1972) Gamut Area for good color discrimination 7

8 Color rendering index (CRI) Measure of the degree of color shift objects undergo when illuminated by the light source as compared with the color of those same objects when illuminated by a reference source of comparable color temperature. y CIE 1931 Chromaticity Diagram with Planktain Locus CRI = 90 CRI = x IESNA Lighting Handbook, 9 th Ed. (2000) CRI = 50 8

9 Calculation of CRI Uses CIE 1964 UCS to calculate a color difference based on chromaticity change Differences in chromaticity between the test and reference sources are removed by an adaptive color shift calculation General CRI Mean color shift of 8 test color samples Special CRI Color shift of any of the 14 test-color samples Reference Illuminants For CCT < 5000 K, a Planckian radiator is used For CCT > 5000 K, one of a series of radiant power distributions of phases of CIE daylight is used (e.g., D65) 9

10 CRI test color samples y CIE CIE Chromaticity Diagram with Planktain Locus Locus CIE 1931 Chromaticity Diagram with Planckian Locus x 10

11 Adaptive color shift y CIE CIE 1931 Chromaticity Diagram with Planktain Planckian Locus Locus x 11

12 CRI formulae Color difference formula E 1 * 2 * 2 * 2 L u v 2 Individual and special color chip rendering R E i i General Color Rendering Index (CRI Ra ) 8 1 R a R i

13 Color Gamut Area y CIE CIE Chromaticity Diagram with Planktain Locus Locus CIE 1931 Chromaticity Diagram with Planckian Locus x 13

14 Color Gamut Area Area of the polygon created by the 8 test color samples when illuminated by the light source of interest. Gamut area of an equal energy spectrum (EES) = Gamut area index normalizes the GA of the EES to 100. Other light sources are scaled accordingly. In general, the larger the gamut area, the more saturated the color samples are and the easier it is to discriminate between them. u Daylight MH Fluor. Inc. v HPS Adapted from Human Factors in Lighting - 2nd Edition Boyce, P. R London; New York: Taylor & Francis. 14

15 ASSIST recommends Guide to Specifying Color Properties of Light Sources for Retail Lighting Recommendation Color rendering: 110 Use of color rendering index AND gamut area index to increase the chances a light source will provide good color rendering for most retail applications 80 CRI and 80 GAI 100 Gamut area index Color rendering index 15

16 Experimental approach: A priori test of two metrics Test directly the hypothesis that CRI and GAI complement each other High CRI and high GAI will be preferred Subjective evaluation Multicolor scene Subjective responses to vividness, naturalness and overall acceptability for a retail context Asked subjects to rank the hues that influenced their decisions the most Evaluation of one source at a time 16

17 Apparatus Display of fresh fruits and vegetables Color chart 17

18 Apparatus Six light sources: combination of incandescent, phosphor and colored LEDs One light level: 355 lx horizontal illuminance K K CW6 CW WW6 WW6 CW7 CW7 WW7 WW7 GAI GAI CW5 CW5 WW5 WW CRI CRI 18

19 Task Eighteen volunteers who had participated in previous color rendering studies answered four questions per light source 19

20 Subjective ratings of naturalness Light sources with both a high CRI and a high GAI (WW7 and CW7) rendered the display more naturally Warm Cool 20

21 Subjective ratings of vividness Although WW7 and CW7 rendered the display vividly, they did not rendered the display most vividly GAI is a better indicator of subjective ratings of vividness, but clearly high levels of GAI do not ensure high ratings of naturalness ness Warm Cool 21

22 Subjective ratings Cool Naturalness Vividness Cool Warm Warm 22

23 Acceptance by subjects Cool Naturalness Acceptance Cool Warm Warm 23

24 Acceptance by subjects Light sources with both a high CRI and a high GAI (WW7 and CW7) were rated as acceptable more times than the other two light sources in each CCT. Judgments of acceptability seem to mirror those of naturalness, not vividness. Warm Cool 24

25 Hue ranking Warm colors were the most prominent in informing the observer s s opinions in terms of naturalness, vividness, and acceptability of the light sources No difference in the rankings between the cool and warm CCTs Not clear how rankings would be different if more blues and purples had been included in the display 25

26 Hue ranking Rea and Freyssinier (2008) found that when assessing warm hues, observers prefer illumination from warm light sources than from cool light sources Consistent with those findings, since the warm hues dominated observer judgments, scene vividness was always greater with the warm light sources than with the cool light sources Warm Cool Overall, this light source makes this scene look VIVID Warm (avg) 2.05 Cool (avg) Neither Agree Strongly agree Warm (avg) Cool (avg) 26

27 Discussion These results are a direct a priori test of the two-metric hypothesis Data from this and past research indicate that using CRI and GAI together can ensure good color rendering Remember: color rendering is about the source without regard for what exactly is being illuminated need a measure of well enough for architectural lighting. CRI does not meet that expectation CRI + GAI seems to meet that expectation Simple to use and calculate Meets the goals of Bouma (1947), Judd (1967), and Thornton (1972) As a reminder, light level is important for good color rendering. 27

28 Commercial Light Sources with 80 CRI and 80 GAI 100 This table is not intended to be a comprehensive compilation but includes sample light sources in all families (i.e., incandescent, fluorescent, high intensity discharge, and SSL). The inclusion or mention of any specific brand or product in this table is for illustrative purposes only and does not constitute an endorsement by ASSIST or the Lighting Research Center. 28

29 Commercial Light Sources with 80 CRI and 80 GAI Gamut area index Color rendering index 29

30 Summary This two-metric recommendation is practical maintains the use of CRI gamut area index is easily calculated with the same color test samples as CRI predictive, overall subjective assessments of vividness, naturalness and acceptability and has been validated by several human factors studies. 30

31 Summary Several experimental results support the two- metric approach; i.e., using CRI and GAI together can ensure good color rendering For natural objects and prints (pictures) For small viewing boxes and real size displays For warm (3000 K) and cool (5000 K) light sources Both with phosphor based and narrow band sources For low (5-10 fc) ) and high (30-50 fc) ) light levels For general diffuse illumination and for accent lighting 31

32 ASSIST recommends: Retail Lighting Documents available for free download: Short Guide to Light and Color in Retail Merchandising Guide to Specifying Color Properties of Light Sources for Retail Lighting 32

33 Thank you! We gratefully acknowledge the ASSIST sponsors for supporting the program Acuity Brands Lighting, Amerlux, Bridgelux, China Solid State Lighting Alliance, Cree, Everlight Electronics Co., Ltd., Federal Aviation Administration, GE Lighting Solutions, ITRI Industrial Technology Research Institute, Intematix, LG Electronics, LG Innotek, Lighting Science Group, Lite-On, NeoPac Lighting, New York State Energy Research and Development Authority (NYSERDA), OSRAM SYLVANIA/OSRAM Opto Semiconductors, Permlight, Philips Lighting, Sharp Laboratories of America, Seoul Semiconductor, United States Environmental Protection Agency, WAC Lighting Acknowledgements Peter Banwell, EPA Mark Fairchild, MSCL Yoshi Ohno, NIST Wendy Davis, NIST Andrew Bierman, LRC John Bullough, LRC Mariana Figueiro, LRC N. Narendran, LRC Leora Radetsky, LRC 33