1

May 5–7, 2015

Lightfair

How New Methods for Evaluating Color Rendering will Affect You

Michael Royer, PNNL

2

Color Fidelity: A Partial Picture

Reference Image

3

The CIE Color Rendering Index (CRI)

What CRI conveys:

• Average color fidelity/color shift (but with limitations) – Outdated color science

– Few samples

– Only pastel Munsell samples

– Can be selectively optimized

What CRI doesn’t convey:

• Direction/type of color shifts

• Increases or decreases in chroma

• Information about specific hue regions

• Human preference

• Color discrimination potential

• Difference in color for any specific object

• How one source will make things look compared to another

4

One Index is Not Enough. But how many are needed?

And what should they be?

5

A Path Forward: IES Color Metrics Task Group

• Technical Memorandum (i.e., calculation procedure), TM-30

• Currently in final stage of balloting

• It draws from a variety of other color perception research

• Based objective/mathematical approach

• Two-metric system (fidelity [Rf] and gamut [Rg]) for quantifying average color rendition

• 99 color evaluation samples (CES) with color space and wavelength uniformity – Excellent correlation to much larger sets of color samples

• Improved color science

• Many additional tools (e.g., color distortion icon) for better understanding rendition of specific hues.

TM-30 IS NOT YET A FINAL DOCUMENT!

6

The CIE Color Rendering Index (CRI)

What CRI conveys:

• Average color fidelity/color shift (but with limitations) – Outdate color science

– Few samples

– Only Pastel Munsell Samples

– Can be selectively optimized

What CRI doesn’t convey:

• Direction/type of color shifts

• Increases or decreases in chroma

• Information about specific hue regions

• Human preference

• Color discrimination potential

• Difference in color for any specific object

• How one source will make things look compared to another

7

Summary of Notable Features

Calculation Engine

• 99 color evaluation samples – Real objects

– Color space uniformity

– Wavelength uniformity

• Calculations in CAM02-UCS – CIECAM02 chromatic

adaptation

– Most uniform color space

– Stability over CCT

• Mixed reference (4500 K to 5500 K) eliminates discontinuity

Outputs

• Average Color Fidelity: Rf

• Average Color Gamut: Rg

• Rf Versus Rg Plot

• Color Distortion Icon / Color Saturation Icon

• Color Fidelity by Hue Angle (16 Groups)

• Chroma Change by Hue Angle (16 Groups)

• Skin Tone Color Fidelity

• Color Fidelity by Sample

8

A Two-Metric System for Average Characterization

0 < Rf ≤ 100

When Rf > 60:

60 < Rg ≤ 140

9

Hue shift

Rf Versus Rg Plot

Increasing

saturation

Decreasing

saturation

9

10

What does it all mean…

For Specifiers?

For Manufacturers?

For Researchers?

For Energy Efficiency Programs/Utilities?

11

Specifiers

1. Learn about the new metrics – Presentations, Journal Articles, TM-30 (Pending), DG-01 (Pending)

– What they do and don’t characterize; limitations

2. Examine values for existing sources, understand your preferences in applications where you commonly specify

3. Use additional information for better specifications

• Case study: Museums

12

Comparing Rf and Rg

60

70

80

90

100

110

120

130

140

50 60 70 80 90 100

Gam

ut

Ind

ex, R

g

Fidelity Index, Rf

HalogenNeodymium IncandescentF40T12WW/RS/EWF32T8TL830CDM 830Hybrid LEDRAGB LEDPhosphor LED

13

Color Distortion Icon

Rf = 83 | Rg = 112 Rf = 83 | Rg = 99

Increased Saturation

Decreased Saturation

Hue Shift

13

14

Manufacturers

1. Learn about the new metrics – Presentations, Journal Articles, TM-30 (Pending), DG-01 (Pending)

– What they do and don’t characterize; limitations

2. How do your products measure up against others?

3. Does your product have distinguishing features that are marketable?

4. Use tools for future optimizations – Balance between efficiency and color rendition

• Case study: Neodymium Lamps

15

Communicating Performance

(Live Calculation Tool Demo)

16

Beyond Averages

Rf = 82

Rg = 95

LER = 342

Rf = 82

Rg = 95

LER = 318

0

0.5

1

1.5

2

2.5

380 430 480 530 580 630 680 730 780

Rad

ian

t P

ow

er

(no

rmal

ize

d t

o Y

=10

0)

l (nm)

0

0.5

1

1.5

2

2.5

380 430 480 530 580 630 680 730 780

Rad

ian

t P

ow

er

(no

rmal

ize

d t

o Y

=10

0)

l (nm)

17

Beyond Averages

50

55

60

65

70

75

80

85

90

95

100

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

Rf b

y H

ue

Hue Bin

-20%

-15%

-10%

-5%

0%

5%

10%

15%

20%

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

Ch

rom

a Sh

ift

By

Hu

e

Hue Bin

50

55

60

65

70

75

80

85

90

95

100

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

Rf b

y H

ue

Hue Bin

-20%

-15%

-10%

-5%

0%

5%

10%

15%

20%

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

Ch

rom

a Sh

ift

By

Hu

e

Hue Bin

18

EE Programs/Regulators

1. Learn about the new metrics – Presentations, Journal Articles, TM-30 (Pending), DG-01 (Pending)

– What they do and don’t characterize; limitations

2. How do existing thresholds for color rendering relate?

3. Should all sources that previously qualified still do so? Should the numerical value stay the same?

• Case study: Triphosphor Fluorescent

19

Comparing Rf and Ra

~16 point spread in

Rf scores at Ra =

~80 70

75

80

85

90

95

100

70 75 80 85 90 95 100

Rf

CRI Ra

FilamentDaylight ModelsNarrowband FluorescentBroadband FluorescentHIDHybrid LEDColor Mixed LEDPhosphor LED

20

Researchers

1. Learn about the new metrics – Presentations, Journal Articles, TM-30 (Pending), DG-01 (Pending)

– What they do and don’t characterize; limitations

2. New information can improve characterization (but averages shouldn’t always be used to establish correlations)

3. What new investigations are warranted?

• Case study: Preference by Application

21

WHY ADOPT?

• With two average numbers and other visualization tools, TM-30 provides a more complete characterization of color rendition than a fidelity metric alone (e.g., CRI).

• With a greater number of samples, the values are harder to selectively optimize and should provide a better representation of average color rendering.

• The method can help manufacturers optimize spectral designs, as well as accurately weigh tradeoffs between color rendition and other characteristics.

• The color rendition characteristics of products can more easily be differentiated (and marketed).

• The method can help specifiers and purchasers to select products that are more appropriate for their needs.

• Thresholds, design guidance, and preconceptions will need to be re-evaluated.

TM-30 IS NOT YET A FINAL DOCUMENT!