Great promise, few options:
Can advances in color science shift
the market?
Tony Esposito
January 29, 2020
Michael Royer
Lorne Whitehead
2
Optimizing Illumination
illumination
electricity
human impact
watt
well-being
well-being “joy per joule”
watt
n’s Joy per Joule
illumination
human impact
Optimizing Illumination’s Joy per Joule
quality quantity
comfort performance
illumination
human impact
quality quantity
comfort performance
illumination
human impact
quality
comfort performance
quantity
illumination
human impact
Over-Emphasizing the Easily-Measured
quality
comfort performance
quantity
- a lingering problem from the 20th century -
Over-Emphasizing the Easily-Measured- a lingering problem from the 20th century -
CIE photopic luminosity function V(λ)
Over-Emphasizing the Easily-Measured- a lingering problem from the 20th century -
20
50
2
5
.001Illuminance (fc)
9
Location Quantity Spectrum Timing Control
10
Location Quantity Spectrum Timing Control
11
SPD
Chromaticity
Relative Photoreceptor
Stimulation
LER
Color Rendition
CCT
Duv
S/P
M/P
Fidelity
Shifts
White-ness
Match
Match
Hue
Chroma
Pupil
Circa-dian
LPW
Visual Perf.
12
SPD
Chromaticity
Relative Photoreceptor
Stimulation
LER
Color Rendition
CCT
Duv
S/P
M/P
Fidelity
Shifts
White-ness
Match
Match
Hue
Chroma
Pupil
Circa-dian
LPW
Visual Perf.
13
14
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
v'
u'
0.0000.0050.0100.0150.0200.0250.0300.0350.040
380 480 580 680 780
Spec
tral
Pow
er (W
/nm
)
Wavelength (nm)
RedLimeAmberGreenCyanBlueIndigo
15
16
RfRg
Rcs,h1Rt
CRIR9
M/PS/P
LER
68 – 9479 – 120-19% – 26%88 – 9643 – 95-120 – 980.6 – 1.11.4 – 2.2286 – 352
0.0000.0020.0040.0060.0080.0100.0120.0140.0160.018
380 430 480 530 580 630 680 730 780
Spec
tral
Pow
er (W
/nm
)
Wavelength (nm)
2700 K3000 K3500 K4000 K5000 K6500 K
Nominal CCTNormalized for Equal Lumens
17
Status Quo: Given a CCT and minimum color fidelity, maximize lumens/watt.
18
S/P = M/P = CRI = R9 =
LER = LPW ≈
S/P = M/P = CRI = R9 =
LER = LPW ≈
1.40.7100991515
1.10.482-836080
S/P = M/P = CRI = R9 =
LER = LPW ≈
1.20.58112310130
19
What will it take to changethe status quo?
New metrics?Induce demand / does anyone care?Material R&D?Patent barriers?Cost?Reevaluation of Tradeoffs?
How Vision Senses Surface Colors
How Vision Senses Surface Colors
How Vision Senses Surface Colors
A Surface’s Color Relates to its Spectral Reflectance Function
Wavelength
Ref
lect
ance
“IncreasingLightness”
How Vision Senses Surface Colors
Lightness:
How Vision Senses Surface Colors
A Surface’s Color Relates to its Spectral Reflectance Function
Wavelength
Ref
lect
ance
“Blue to Orange”
“Changing Average
Slope”
Blue
Orange
Blue
Orange
How Vision Senses Surface Colors
A Surface’s Color Relates to its Spectral Reflectance Function
Wavelength
Ref
lect
ance
“Purple to Green”
“Changing AverageCurvature”
Purple
Green
Purple
Green
How Vision Senses Surface Colors
33
“white” light
“gray” surface
orange reflection!
Poor Rendering is Due to Spectral Interference !
FACTS ABOUT COLOR RENDERING AND SOME COMMON MISCONCEPTIONS
1. The IES/CIE Color Fidelity Index Is a number less than or equal to 100, but it is not a percentage.
2. If two objects look the same under daylight, they may not match under a poor color rendering source.
3. Lamps with the same color rendering score may have very different SPDs and cause different shifts.
FACTS ABOUT COLOR RENDERING AND SOME COMMON MISCONCEPTIONS
4. People cannot adapt to poor color rendering
5. Color rendering cannot be judged by looking directly at a light source.
6. Light source color rendering is a completely different property than light source color.
Rf
∆Rf
Consider An Important Design Choice:
Most often, most designers select Design AYet usually, Design B is better for occupants
Design A acceptable acceptable 90 80Design B same as A same as A 80 90
Installed Cost
Power Consumed
Light Quantity
Light Quality
Sheet1
Installed CostPower ConsumedLight QuantityLight Quality
Design Aacceptableacceptable9080
Design Bsame as Asame as A8090
A look to the future:
Excellent Color Rendering (nearly) Everywhere
Requiring: Excellent Metrics for Color Rendering
45
ANSI/IES TM-30-18
46
What is it?
47
“Effect of an illuminant on the color appearance of objects by conscious or subconscious comparison with their color appearance under a reference illuminant.”
- International Commission on Illumination (CIE)
Color rendition is relative
Review: color rendering, defined
48
Test Source
Reference Source
Δ1 Δ2 Δ3 Δ4 Δ5 Δ6 Δ7 Δ8
Color rendition is relative
49
Perfect Fidelity
Limitation of considering only color fidelity
50
CRI = 80 CRI = 80
Increase Saturation
DecreaseSaturation
Limitation of considering only color fidelity
51
CRI = 80
CRI = 80
Positive Hue Shift
Negative Hue Shift
Increase Saturation
DecreaseSaturation
Limitation of considering only color fidelity
52
Increase Saturation
DecreaseSaturation
Positive Hue Shift
Negative Hue Shift
Constant Fidelity (CRI)
Limitation of considering only color fidelity
53
EQ EQ EQ EQ
Rf,h1100
Rf,h180
Rf,h160
Rf,h180
Rf,h160
(Color appearances not accurate. For educational purposes only)
Limitation of considering only color fidelity
Rcs,h10%
Rcs,h1+8%
Rcs,h1+15%
Rcs,h1-8%
Rcs,h1-15%
54
How many metrics are contained within TM-30?
55
At least 149(1) Average Color Fidelity, Rf(1) Average Gamut Area, Rg(16) Local Chroma Shift, Rcs,hj(16) Local Hue Shift Rh1,hj(16) Local Color Fidelity, Rf,hj(99) Sample Color Fidelity, Rf,CESi
How many metrics are contained within TM-30?
56
How many should be minimally considered?
57
How many should be minimally considered?
58
Average Fidelity, Rf: indicates the average deviation from the source’s reference illuminant. It only indicates the magnitude of the difference, not the direction.
Average Gamut Area, Rg: indicates the average increase or decrease in saturation relative to the source’s reference illuminant.
Local Color Fidelity Hue Bin 1 (Red), Rf,h1: indicates the specific deviation, relative to the source’s reference illuminant, for red hues.
Local Chroma Shift, Rcs,h1: indicates the specific increase or decrease in saturation, relative to the source’s reference illuminant, for red hues.
e.g. “CRI”
e.g. CQS QgGAI
e.g. R9
e.g. None
The main 4 metrics
59
Metric Quizzzzzzzzzzzz
60
What can we say about this point?
Rf = 87, Rg = 109
Rf = 87
Rg = 109
This light source causes distortions relative to ref source (don’t know direction of distortion)
This light source, ON AVERAGE, increases the saturation of objects (relative to ref source). We don’t know which hues are being enhanced.
TM-30 Metric Quiz
61
Relative to the ref source, this light source is:
• Increasing the average saturation of RED hues (Rcs,h1 = + 11%)
• Slightly increasing the saturation of green and yellow-green hues (Rcs,h6 = +4%)
• Rf,h1 < 100
What can we say about this point?
Rf = 87, Rg = 109
TM-30 Metric Quiz
62
63
What is it?
64
MeasureAverage Color fidelity
Red Fidelity (hue bin 1)
Gamut area
Gamut shape/Specific color shifts
(specifically red)
SymbolRf
Rf,h1Rg
CVGRcs,h1
CharacteristicFidelity
Preference(object and skin)
Vividness
IES TM-30 ANNEX E
65
Design IntentThe desired effect of color rendition on the illuminated environment.
Preference (P) Vividness (V) Fidelity (F)
Prio
rity
Leve
lTh
e ba
lanc
e be
twee
n al
low
ing
trade
offs
and
incr
easi
ng th
e lik
elih
ood
of m
eetin
g th
e de
sign
inte
nt. 1 P1
Rf ≥ 78
V1
Rg ≥ 118
F1 Rf ≥ 95Rg ≥ 95 Rcs,h1 ≥ 15%
-1% ≤ Rcs,h1 ≤ 15%
2 P2
Rf ≥ 75
V2
Rg ≥ 110
F2
Rf ≥ 90
Rg ≥ 95 Rcs,h1 ≥ 6% Rf,h1 ≥ 90
-7% ≤ Rcs,h1 ≤ 15%
3 P3
Rf ≥ 70
V3
Rg ≥ 100
F3
Rf ≥ 85
Rg ≥ 88 Rcs,h1 ≥ 0% Rf,h1 ≥ 85
-12% ≤ Rcs,h1 ≤ 18%
IES TM-30 ANNEX E
66
Design IntentThe desired effect of color rendition on the illuminated environment.
Preference (P) Vividness (V) Fidelity (F)
Prio
rity
Leve
lTh
e ba
lanc
e be
twee
n al
low
ing
trade
offs
and
incr
easi
ng th
e lik
elih
ood
of m
eetin
g th
e de
sign
inte
nt. 1 P1
Rf ≥ 78
V1
Rg ≥ 118
F1 Rf ≥ 95Rg ≥ 95 Rcs,h1 ≥ 15%
-1% ≤ Rcs,h1 ≤ 15%
2 P2
Rf ≥ 75
V2
Rg ≥ 110
F2
Rf ≥ 90
Rg ≥ 95 Rcs,h1 ≥ 6% Rf,h1 ≥ 90
-7% ≤ Rcs,h1 ≤ 15%
3 P3
Rf ≥ 70
V3
Rg ≥ 100
F3
Rf ≥ 85
Rg ≥ 88 Rcs,h1 ≥ 0% Rf,h1 ≥ 85
-12% ≤ Rcs,h1 ≤ 18%
Assumptions: 200-700 lux, polychromatic environment, single chromaticity
IES TM-30 ANNEX E
67
Design IntentThe desired effect of color rendition on the illuminated environment.
Preference (P) Vividness (V) Fidelity (F)
Prio
rity
Leve
lTh
e ba
lanc
e be
twee
n al
low
ing
trade
offs
and
incr
easi
ng th
e lik
elih
ood
of m
eetin
g th
e de
sign
inte
nt. 1 P1
Rf ≥ 78
V1
Rg ≥ 118
F1 Rf ≥ 95Rg ≥ 95 Rcs,h1 ≥ 15%
-1% ≤ Rcs,h1 ≤ 15%
2 P2
Rf ≥ 75
V2
Rg ≥ 110
F2
Rf ≥ 90
Rg ≥ 95 Rcs,h1 ≥ 6% Rf,h1 ≥ 90
-7% ≤ Rcs,h1 ≤ 15%
3 P3
Rf ≥ 70
V3
Rg ≥ 100
F3
Rf ≥ 85
Rg ≥ 88 Rcs,h1 ≥ 0% Rf,h1 ≥ 85
-12% ≤ Rcs,h1 ≤ 18%
Assumptions: 200-700 lux, polychromatic environment, single chromaticity
IES TM-30 ANNEX E
68
Design IntentThe desired effect of color rendition on the illuminated environment.
Preference (P) Vividness (V) Fidelity (F)
Prio
rity
Leve
lTh
e ba
lanc
e be
twee
n al
low
ing
trade
offs
and
incr
easi
ng th
e lik
elih
ood
of m
eetin
g th
e de
sign
inte
nt. 1 P1
Rf ≥ 78
V1
Rg ≥ 118
F1 Rf ≥ 95Rg ≥ 95 Rcs,h1 ≥ 15%
-1% ≤ Rcs,h1 ≤ 15%
2 P2
Rf ≥ 75
V2
Rg ≥ 110
F2
Rf ≥ 90
Rg ≥ 95 Rcs,h1 ≥ 6% Rf,h1 ≥ 90
-7% ≤ Rcs,h1 ≤ 15%
3 P3
Rf ≥ 70
V3
Rg ≥ 100
F3
Rf ≥ 85
Rg ≥ 88 Rcs,h1 ≥ 0% Rf,h1 ≥ 85
-12% ≤ Rcs,h1 ≤ 18%
Assumptions: 200-700 lux, polychromatic environment, single chromaticity
IES TM-30 ANNEX E
69
Design IntentThe desired effect of color rendition on the illuminated environment.
Preference (P) Vividness (V) Fidelity (F)
Prio
rity
Leve
lTh
e ba
lanc
e be
twee
n al
low
ing
trade
offs
and
incr
easi
ng th
e lik
elih
ood
of m
eetin
g th
e de
sign
inte
nt. 1 P1
Rf ≥ 78
V1
Rg ≥ 118
F1 Rf ≥ 95Rg ≥ 95 Rcs,h1 ≥ 15%
-1% ≤ Rcs,h1 ≤ 15%
2 P2
Rf ≥ 75
V2
Rg ≥ 110
F2
Rf ≥ 90
Rg ≥ 95 Rcs,h1 ≥ 6% Rf,h1 ≥ 90
-7% ≤ Rcs,h1 ≤ 15%
3 P3
Rf ≥ 70
V3
Rg ≥ 100
F3
Rf ≥ 85
Rg ≥ 88 Rcs,h1 ≥ 0% Rf,h1 ≥ 85
-12% ≤ Rcs,h1 ≤ 18%
Assumptions: 200-700 lux, polychromatic environment, single chromaticity
IES TM-30 ANNEX E
70
Important Notes• The Assumptions are important!
• 200 – 700 Lux• Polychromatic Environment• Single Chromaticity
• PVF categories ARE NOT INDEPENDENT. P1V1F1 is not possible
• ANNEX E does not specifically consider the rendition of skin tones
IES TM-30 ANNEX E
71
So What?
72
F1
F2
F3
All Theoretical SPDsTheoretical SPDs with Ra ≥ 80 and R9 ≥ 0Theoretical SPDs with Ra ≥ 90 and R9 ≥ 50
Commercially-Available LED SPDs
IES TM-30 ANNEX E: Color Fidelity
73
P1*P2*P3*
IES TM-30 ANNEX E: Color Preference
74
Summary, TM-30 and ANNEX E• IES TM-30 is a unified system of objective measures that capture color difference
• Specifically, TM-30 provides measures beyond average fidelity and average gamut area that provide information about hue-SPECIFIC color shifts
• IES TM-30 ANNEX E was developed to facilitate the ease of use of TM-30 and to help specifiers flex color rendition as a parameter in their designs
• And, potentially, as a way for manufacturers to differentiate their products
• TM-30 helps illustrate the tradeoff between luminous efficacy and red rendition, and shows that many commercially available sources desaturate red colors
• TM-30 can be used in an optimization routine to produce sources with a spectrum that is more purposefully tuned to the intended application
75
Metameric Uncertainty
https://www.astroml.org/book_figures/chapter3/fig_bivariate_gaussian.html
76
MeasureAverage Color fidelity
Red Fidelity (hue bin 1)
Gamut area
Gamut shape/Specific color shifts
(specifically red)
SymbolRf
Rf,h1Rg
CVGRcs,h1
CharacteristicFidelity
Preference(object and skin)
Vividness
Metameric Uncertainty
77
MeasureAverage Color fidelity
Red Fidelity (hue bin 1)
Gamut area
Gamut shape/Specific color shifts
(specifically red)
Light Source Error Score
Metameric Uncertainty Index
SymbolRf
Rf,h1Rg
CVGRcs,h1
RdRt
CharacteristicFidelity
Preference(object and skin)
Vividness
Color Discrimination
Metameric Mismatch
IES TM-30
Metameric Uncertainty
78
What is metamerism?Metameric SourcesSources that have different SPDs, but the same chromaticity.They have the same appearance when viewed directly.
Metameric ObjectsSources that have different “Practical Colors” (SPD*SRD), but the same chromaticity.They have the same appearance when viewed directly.
79
Why is it so uncertain?
80
Who cares?
81
Metameric Mismatch
https://www.xrite.com/blog/two-essential-components-color-programhttps://store.nike.com/us/en_us/product/air-force-1-high-essential-id/
https://www.chick-fil-a.com/menu-items/coca-cola
https://www.socialbakers.com/statistics/twitter/profiles/detail/71026122-mcdonaldshttps://apps.apple.com/us/app/t-mobile/id561625752
https://twitter.com/starbuckshttps://www.facebook.com/
82
Metameric Mismatch
“The sensitivity of the strips on clinical urine specimens may vary depending upon several factors, such as the variability of color perception, specific gravity, pH value, and the lighting conditions when the strips are read visually.”
83
We want to know,
What is the likelihood that a light source will cause a mismatch between objects that were previously metameric (with respect to a given illuminant)
84
Metameric Uncertainty Index, Rt
85
Matching greenunder 5000 Kdaylight
Wavelength (nm)
R R R
Matching blueunder 5000 Kdaylight
Matching redunder 5000 Kdaylight
Wavelength (nm) Wavelength (nm)
SRDs that are metameric with respect to D50
86
SRDs that are metameric with respect to D50
87
SRDs that are metameric with respect to D50
88
SRDs redistribute when the source is changed
89
SRDs redistribute when the source is changed
91
SRDs redistribute when the source is changed
92
SRDs redistribute when the source is changed
93
Actual color shift is dissected into two pieces
“Actual Color Shift”
94
“Base Color Shift”
“Metameric Color Shift” “Actual Color Shift”
Actual color shift is dissected into two pieces
95
Metameric Uncertainty Index, Rt
100 Rt
96
Metameric Uncertainty Index, Rt
High Rt
97
Metameric Uncertainty Index, Rt
Low Rt
98
Summary, Rt• Light source-induced color shifts divided into two components:
• base color shift, following a regular pattern
• metameric color shift, which is random
• These can be quantified using the 99 IES TM-30 samples and a vector field model for estimating base color shift
• Metameric color shift intensity is new useful information
• It can be converted into Rt, the Metameric Uncertainty Index
• This can be useful in situations where metamerism matters
• Next steps for Rt include psychophysical experimentation. Does Rt actually measure what we think it does?
99
Great promise, few options: Can advances in color science shift the market?
100
S/P = M/P = CRI = R9 =
LER = LPW ≈
S/P = M/P = CRI = R9 =
LER = LPW ≈
1.40.7100991515
1.10.482-836080
S/P = M/P = CRI = R9 =
LER = LPW ≈
1.20.58112310130
101Royer M. Evaluating tradeoffs between energy efficiency and color rendition, OSA Continuum 2, 2308-2327 (2019)
102
CRI = R9 =
LER =
9572290
CRI = R9 =
LER =
1009915
CRI = R9 =
LER =
9798260
103
CRI = R9 =
LER =
9296315
CRI = R9 =
LER =
771513
CRI = R9 =
LER =
8646250
104
CRI = R9 =
LER =
9124380
CRI = R9 =
LER =
7943398
CRI = R9 =
LER =
8321311
105
Lightening Round:What’s Next In Color Science
• Updating Chromaticity• Observer Variability• Color Discrimination• Color Rendition Variability• Far Red• Light Level Variation• Tuning for Visually Impaired
106
Last Word• What’s your vision for the future?• What’s wrong with the status quo?
Great promise, few options: Can advances in color science shift the market?Optimizing Illumination's Joy per JouleOver-Emphasizing the Easily-MeasuredStatus QuoWhat will it take to change the status quo?How Vision Senses Surface ColorsFACTS ABOUT COLOR RENDERINGAND SOME COMMON MISCONCEPTIONSConsider An Important Design Choice:UntitledANSI/IES TM-30-18What is it?Review: color rendering, definedColor rendition is relativeLimitation of considering only color fidelityHow many metrics are contained within TM-30?How many should be minimally considered?The main 4 metricsMetric QuizzzzzzzzzzzzTM-30 Metric QuizAnnex EWhat is it?IES TM-30 ANNEX ESo What?IES TM-30 ANNEX E: Color FidelityIES TM-30 ANNEX E: Color PreferenceSummary, TM-30 and ANNEX EMetameric UncertaintyWhat is metamerism?Why is it so uncertain?Who cares?Metameric MismatchWe want to know,Metameric Uncertainty Index, RtSRDs that are metameric with respect to D50SRDs redistribute when the source is changedActual color shift is dissected into two piecesMetameric Uncertainty Index, RtSummary, RtGreat promise, few options:Lightening Round:What’s Next In Color ScienceLast Word