Meta Standards for Color Rendering Metrics. Implications for Sample Spectral Sets

(1)

(2)

(3)

Color Concepts Are Confusing!

• Two strong views of color and light sources:

Two strong views of color and light sources:

–

Color matters a lot

Color matters little

–

Color matters little

• Both statements are correct, sometimes:

–

Some aspects of color do matter a lot to people

–

Some aspects of color don’t matter all that much

(4)

An important case: “Flesh Tone”

An important case: Flesh Tone

0.8 1.0 ₃₀ skin and matched photo under Planckian 4500K illumination skin under matched RGB illumination matched photo under matched RGB illuminant Planckian 4100K Colour Matched RGB 0.4 0.6 Re fl e ct an ce

untanned Caucasion forearm skin 20

(5)

An important case: “Flesh Tone”

0.8 1.0 ₃₀ skin and matched photo under Planckian 4500K illumination skin under matched RGB illumination matched photo under matched RGB illuminant Planckian 4100K Colour Matched RGB

An important case: Flesh Tone

0.4 0.6 Re fl e ct an ce

(6)

An important case: “Flesh Tone”

An important case: Flesh Tone

(7)

An important case: “Flesh Tone”

An important case: Flesh Tone

(8)

Key Ideas

• Different spectra can appear the same color

• But may render colors of surfaces differently

• We don’t adapt to poor color rendering

• _{This causes some people discomfort}

(9)

(10)

TC 1‐69 meeting July 13 2011: The TC now plans two reports:

Report 1:

• Recommends an update for the current CIE CRI, including:

• An updated calculation engine;

• From Ronnier Luo’s collaborative group);

• fixing “outdatedness”;

fixing outdatedness ;

• An updated sample set;

• Eliminates spectral response non‐uniformity;

A

i

l

d “

l”

ibl

(11)

TC 1‐69 meeting July 13 2011: The TC now plans two reports:

Report 1:

• Recommends an update for the current CIE CRI, including:

• An updated calculation engine;

• From Ronnier Luo’s collaborative group);

• fixing “outdatedness”;

fixing outdatedness ;

• An updated sample set;

• Eliminates spectral response non‐uniformity;

A

i

l

d “

l”

ibl

• As simple and “real” as possible.

Report 2:

• Recommends that the CRI not be viewed as a colour preference index;

• Surveys literature supporting this general view;

(12)

TC 1‐69 meeting July 13 2011: The TC now plans two reports:

Report 1:

• Recommends an update for the current CIE CRI, including:

• An updated calculation engine;

• From Ronnier Luo’s collaborative group);

• fixing “outdatedness”;

fixing outdatedness ;

• An updated sample set;

• Eliminates spectral response non‐uniformity;

A

i

l

d “

l”

ibl

• As simple and “real” as possible.

Report 2:

• Recommends that the CRI not be viewed as a colour preference index;

• Surveys literature supporting this general view;

• Highlights the CQS (there are differing view on how to do so);

Highlights the CQS (there are differing view on how to do so);

(13)

(14)

200

160

180

200 d

e

(/pm

)

100

120

140 e

Ma

gn

it

u

d

40

60

80 n

Cu

rv

at

u

re

0

20

400

450

500

550

600

650

700 Mea

Wavelength (nm)

(15)

200

160

180

200 e

(/pm)

100

120

140 e

Ma

gn

it

ud

40

60

80 an

Cu

rv

atur

e

1000 Non‐hybrid Leeds Samples

0

20

400

450

500

550

600

650

700 Me

a

1000 Hybrid Leeds Samples

Wavelength (nm)

(16)

0.70 0.80 0.90 1.00

0.8

1.0

0.40 0.50 0.60 0.70

0.4

0.6

0.10 0.20 0.30

0 0

0.2

0.00 400 450 500 550 600 650 700

0.0

380

430

480

530

580

630

680

730

780 Hybrid Spectra

Raw Spectra

Hybrid Spectra

(17)

F

f R

t ti

S

t

(18)

(19)

(20)

(21)

(22)

(23)

(24)

(25)

Conclusions

• The HL17 set is proposed for CRI sample set

• This reduces CRI error from 5 points to <1

This reduces CRI error from 5 points to <1

• _{The CRI will then work well with LEDs}

W k i

ill

d d

l

f

(26)