Yoshi Ohno. ssl.iea-4e.org. SSL Annex IC 2017 Task Leader (CIE President, NIST Fellow) National Institute of Standards and Technology, USA

Transcription

1 ssl.iea-4e.org Yoshi Ohno SSL Annex IC 2017 Task Leader (CIE President, NIST Fellow) National Institute of Standards and Technology, USA SSL Annex Conference; 23 November

2 Color Quality of Lighting Color Quality Chromaticity (white light) Color Rendering Color fidelity (CRI) Color preference CCT, Duv Chromaticity coordinates (x,y), (u,v ) SSL Annex Conference; 23 November

3 Standards for White Light Chromaticity for SSL IEC for Fluorescent Lamps (5 step MacAdam ellipses) ANSI C Chromaticity Specifications for Solid State Lighting Products 1 st version 2008 (Ohno, Technical coordinator) SSL Annex Conference; 23 November

4 ANSI C Duv defined in ANSI C SSL Annex Conference; 23 November

5 Specifications expressed in (x, y) or (u,v ) (u, v ) diagram is recommended SSL Annex Conference; 23 November

6 Annex B, C for Smaller Tolerances Annex B. 4-step quadrangles Annex C. 4-step u v circles SSL Annex Conference; 23 November

7 CIE TN 001 Chromaticity Difference Specification Published in July 2014 u v circle is recommended to replace MacAdam ellipses 5-step u v circles for IEC Requires 3 parameters No standards for interpolation Research by 1 subject u v circle: a circle with radius r on CIE (u,v ) diagram. n-step u v circle: u v circle with a radius n x ( This corresponds to n-step MacAdam ellipses.) CIE TN 001: Freely available on CIE website: Google TN 001 SSL Annex Conference; 23 November

8 Preference to below blackbody An example of a neodymium lamp Above blackbody CRI R a = 77 Below blackbody R a score does not correspond well with visual evaluation. SSL Annex Conference; 23 November

9 Experiment on perceived white point by LRC (2013) Recent Studies Experiment on preferred lighting by NRC, Canada (2013) NIST vision experiment in 2013 & 2015 M. S. Rea,* J. P. Freyssinier, White Lighting, CR&A, 30-2, 82-92, Dikel et al, Preferred Chromaticity of Color- Tunable LED Lighting, LEUKOS, 10:2, , DOI: / (2013). Ohno, Y., Fein, M., Vision Experiment on Acceptable and Preferred White Light Chromaticity for Lighting, CIE x039:2014, pp (2014). SSL Annex Conference; 23 November

10 NIST Spectrally Tunable Lighting Facility Energy Star Webinar March

11 NIST Experiments on Preferred and Acceptable level of Duv (2013, 2015) 18 subject (2013), 21 subjects (2015). Targets: (1) mixed fruits and vegetables and the whole room, (2) skin tone. 23 pairs of light presented. Subjects adapted to each Duv point. Asked which light looks more natural Duv Ohno, Y., Fein, M., Vision Experiment on Acceptable and Preferred White Light Chromaticity for Lighting, CIE x039:2014, pp (2014). Ohno, Y., Oh, Semin., Vision Experiment II on White Light Chromaticity for Lighting, CIE x042:2016, pp (2016) SSL Annex Conference; 23 November

12 ANCI C revision published, July, 2017 Flexible Duv Specification Basic Specification Extended Specification 12 SSL Annex Conference; 23 November

13 Some Products already available Wouter Soer, presentation at DOE SSL R&D Workshop Jan SSL Annex Conference; 23 November

14 Color Quality of Lighting Color Quality Chromaticity (white light) Color Rendering Color fidelity (CRI) Color preference CCT, Duv Chromaticity coordinates (x,y), (u,v ) SSL Annex Conference; 23 November

15 Measure for Color Quality of Light Sources for General Lighting Color quality is the most critical characteristic for lighting quality. Color Rendering Index CRI (CIE 13.3:1995) with 1974 formula, has been used over 40 years. SSL Annex Conference; 23 November

16 Problems of CRI with SSL Sources CRI worked well for traditional lamps (FLs). LEDs (SSL) emerged. CRI (R a ) values do not always correlate well with visual evaluation. CRI is used in products specifications, many government programs, energy regulations, etc. General trade-off relationship, CRI vs lm/w. (CRI is an energy-related issue) Many proposals to improve CRI (e.g., CQS). SSL Annex Conference; 23 November

17 CIE 224: CIE 2017 ColourFidelity Index for Accurate Scientific Use Published April 2017, from TC1-90. First CIE publication on a metric of color rendering since 1974 (CIE 13.2). CIE 13.3 (1995) is only editorial revision. This is based on Fidelity Index R f in IES TM-30, color space and all formulae updated. New design tool for various applications Wavelength (nm) The colour fidelity index R f is not a replacement of the CRI (R a ) for the purpose of rating and specification of products nor for regulatory or other minimum performance requirements Reflectance factor SSL Annex Conference; 23 November

18 Color Fidelity Test source Same CCT [K] Reference illuminant Planckian (CCT<5000 K) Standard Daylight (CCT > 5000 K) CRI R a, R 9,.. CIE 2017 R f TM-30 R f Color Fidelity: How accurate the object colors appear compared to the reference illuminant SSL Annex Conference; 23 November

19 Two Aspects in the Problems of CRI 1. Scientific inaccuracies as a color fidelity metric Outdated color space (W*U*V*) Outdated chromatic adaptation transform Too small number (8) of test samples Color fidelity is a fundamental measure of color quality 2. Problem of CRI being a color fidelity metric Perception of color quality is different from color fidelity Consider color preference The users of lighting expect CRI to work as an overall color quality measure. Addressed by CIE 224 Not yet addressed SSL Annex Conference; 23 November

20 CRI Ra=94 Ra=78 Which light looks better? AB SSL Annex Conference; 23 November

21 Why perception differs from the CRI scores? CRI Ra=94 CRI Ra=78 Looks better (for most people) Test light (R a =78) Reference (Planckian) Increased color contrast is preferred. CRI penalizes good preferred light sources. SSL Annex Conference; 23 November

22 CIELAB 2014 Vision Experiment at NIST on preferred chroma saturation level 20 subjects 2700 K, 3500 K, 5000 K D uv =0, (3500 K only) Illuminance 300 lx C * ab = C * ab = a* Most saturated b* Most de-saturated SSL Annex Conference; 23 November

23 Results of 2014 experiment Mixed fruits & v. Subjects' preference 100% 80% 60% 40% 20% Fruits/Vegetable Skin tone Red fruits Green fruits De-saturation Average of all CCTs CRI Ra Over-saturation Skin tone Red fruits/vegetables 0% Chroma difference C * ab (CQS red sample) Don t raise the bar for CRI too high. Use additional metric to improve color quality Green fruits/vegetables Y. Ohno, M. Fein, C. Miller, Vision Experiment on Chroma Saturation for Color Quality Preference, CIE 216 :2015, pp (2015) SSL Annex Conference; 23 November

24 IES TM-30 Concept of a Two-Metric System (2015) C=8 C=4 Gamut R g NIST experiment lights Image from Michael Royer Fidelity R f Image from Michael Royer. SSL Annex Conference; 23 November

25 Gamut area is not sufficient Different gamut shapes with same gamut area IES R g =100 CIE R f =77 IES R g =100 CIE R f =77 This looks better Perceived color quality of these two lights are significantly different. SSL Annex Conference; 23 November

26 2016 NIST experiment on color saturation preference in different hues 11 different gamut shapes 19 subjects evaluate preference in light pairs (Numbers) are C * ab from the reference. Y. Ohno, S. Oh, Y. Kwak, Vision Experiment on Chroma Saturation Preference in Different Hues, Proc. CIE 2017 Midterm Jeju, Korea (2017) SSL Annex Conference; 23 November

27 Summary Overall Color Quality Color Fidelity (CIE 224) CRI Color perception/ preference <Fidelity> still used for general products CRI R a, R 9 specifications, energy CIE 2017 R f regulations, etc. IES TM-30 R f. used as design tools, research <Preference/perception> TM-30 Rg GAI CQS and other metrics Replace CRI in the future CIE is developing a standard preference/perception metric TC1-91 New TC SSL Annex Conference; 23 November

28 Thank You Any Questions Contact: Yoshi Ohno SSL Annex Conference; 23 November