PERCEPTIE VAN DE HELDERHEID VAN LICHTBRONNEN

Peter Hanselaer

ESAT/Light&Lighting Laboratory, KU Leuven, BE

Ruimte en Licht FestivalEindhoven, 17 november 2016

Light&Lighting LaboratoryThe team

2

New Light Sources

Lighting Optical design

AppearancePerception

MeasurementFacilities

Light&Lighting LaboratoryTopics

3

Introduction: colour and numbers

Stimulus

Light source

object , ( )eE λ λ

, ( )eL λ λ

4

ColorimetryBasic Elements

10

Reflectance Spectral radiance

2°

,

,

,

( ). ( )

( ). ( )

( ). ( )

e

e

e

X k L x

Y k L y

Z k L z

λ

λ

λ

λ λ

λ λ

λ λ

=

=

=

∑

∑

∑

5

ColorimetryTristimulus values

Colour Matching Functions

11

SpectralRadiance

From spectral radiance to tristimulus values (X,Y,Z)

6

ColorimetryColour coordinates

+( , )x y L ′ ′ +( , )u v L

11

From tristimulus values (X,Y,Z) to colour coordinates

7

ColorimetryColour coordinates

11

Disadvantages of colour coordinates

• Not uniform with respect to colour differences• Not directly related to perceptional quantities: hue,

saturation, brightness• Not considering background and surround

8

ColorimetryRelated stimuli

(X,Y,Z) or (x,y,L) is the typical colorimetric output for a single stimulusVisual perception is however strongly determined by the characteristics of stimulus and environment: related colours

12

9

ColorimetryColour Appearance Model

Colour appearance models CAM generate perceptualcorrelates of stimuli in relation to background and surround: brightness (lightness), hue, colourfulness (chroma, saturation)

CIECAM02 is an accepted CAM, mimicking some processesin the retina

14

ColorimetryCIECAM02

CIECAM02Model

SampleX,Y,Z

Correlates of:• redness/ greenness, a

• yellowness/ blueness, b

• brightness, Q• lightness, J• colourfulness, M• chroma, C• saturation, s• hue angle, h• hue composition, H

White

Xw,Yw,Zw

Surround

LA

Background

Avg, dim, dark

Chromaticadaptation

Cone response

Surroundinduction

10

*

Please stare at the central asteriskColorimetryCIECAM02

11

Now stare at the central asterisk, then at the right-hand asterisk

* *

12

ColourCompression and adaptation

13

1/2

.n

a n nIg

I Iρ =

+

13

L= 60

J = 70

L=60

J = 73

L= 60

J = 80

ColorimetryEffect of Background

14

14

Perception of light sourcesTwo categories

15

Dark background

Luminousbackground

CAM available?

Perception of light sourcesState-of-the-art

16

Dark background

5 years ago: three Colour Appearance Models

• CAM97u (Hunt, 1997)• ATD01(Guth, 2002)• CAMFu (Fu, 2012)• LEq (CIE, 2011)• LEq Nayatani (2 versions, 1997)

Low correspondence to our experimental data for the brightness of coloured stimuli due to an underestimation of

the Helmholtz-Kohlrausch effect

H-K: brightness perception increases with saturation

Perception of light sourcesState-of-the-art

17

Luminous background

No model available!

• Traffic signals• Display (smartphone, billboards)• Glare

Perception of light sourcesResearch questions

18

Part 1: can we improve the actualCAMs for a dark background?

Part 2: can we develop a new CAM for a luminous background?

Part 1: experimentalStimuli

• RGBW module, 10°FOV• 105 + 52 stimuli• 6 cd/m2 < L10 < 297 cd/m2

(photopic/glare)• Dark background• Chromaticity range:

19

Selection• 9 + 11 observers, aged 21 to 32• Laypersons

Q&A• Brightness compared to reference (temporal juxtaposition)• Amount of white versus colour in %• Relative amount of two unique hues• Magnitude estimation

Intra- and inter-observer variability• CV Hue: 10 to 11 %• CV Brightness: 14 to 20 %• CV Amount of white: 30 to 44%

Part 1: experimentalObservers

20

Reference brightness: value of 500 = no brightness

21

Part 1: experimentalPsychophysics

Brightness?22

Part 1: experimentalPsychophysics

Brightness 50 -> answer23

Part 1: experimentalPsychophysics

24

Part 1: experimentalHelmholtz-Kohlrausch

60 cd/m²60 cd/m²60 cd/m²60 cd/m² 218 cd/m²218 cd/m²218 cd/m²218 cd/m²

CAM97u

Building a new model: CAM15u

25

26

CAM15u: building the model Step 1: absolute cone excitations

830

10 e, 10

390

830

10 e, 10

390

830

10 e, 10

390

( ) ( )d

( ) m

666.7

782.3

1444

( )d

( ).6 s ( )d

λ

λ

λ

ρ λ λ λ

γ λ λ λ

β λ λ λ

=

=

=

∫

∫

∫

L l

L

L

• Absolute spectral radiance• CIE 2006 cone fundamentals (10°)• Normalization constants

27

• Power law• Power value: determined from observer

brightness data of neutral stimuli

10

1

1/3

1/30

1/0

1 3

ρ ργ γβ β

=

=

=

c

c

c

CAM15u: building the model Step 2: cone compression

28

CAM15u: building the model Step 3: neural signals

Achromatic signal

Red-green balance

Yellow-blue balance

c c c

1222

23.

0ρ γ β = + +

A

cc c

12( )

11 111

βρ γ= − +a

c c c( 2 )0.117 ρ γ β= + −b

29

CAM15u: building the model Step 4: hue, colourfulness, brightness, saturation, amount of white

1180Hue h tan ( / )

π−= b a

2 2Colourfulness 135.5 2= × +M a ba

b

h

0.561Brightness 2.559= + ×Q A M

Saturation = Ms

Q

2.68

100Amount of white

2.291=

+ ×W

s

30

New validation data set of 52 mixed chromatic andachromatic stimuli; magnitude estimation

Brightness Hue Amount of white

R² rS CV R² rS CV R² rS CVCAM15u 0.87 0.94 7 0.99 1.00 5 0.76 0.84 32CAM97u 0.36 0.57 16 0.99 0.99 5 - - -CAMFu 0.22 0.41 26 0.99 0.99 5 - - -

CAM15u: performance Hue, brightness, amount of white

31

CAM15u Application

European Standard for VMS (CEN, 2014) prescribes hue dependent maximum luminance values in dark viewing conditions (circles)

Perception of light sources State-of-the-art

32

Dark background

CAM �

Luminous background

CAM �

Perception of light sourcesResearch questions

33

Part 1: can we improve the actualCAMs for a dark background?

Part 2: can we develop a new CAM for a luminous background?

• 6 hues x 5 = 30 stimuli, 10° FOV• L10 = 50 cd/m2 (positive contrast)• Background: L10 = 28 cd/m2

CCT= 3880 K70° FOV

• 12 + 10 observers, aged 20 to 50• Graphical response sheet

Part 2: experimentalStimuli

34

35

Part 2: resultsH-K effect

Helmholtz-Kohlrausch is even important when a luminousbackground is present

36

Part 2: resultsBackground effect

Impact of the (neutral) background on the brightnessperception (red stimulus, 125 cd/m2)

Negativecontrast

37

Development of a new CAM for self-luminousstimuli and background

• First preliminary results (one fixed stimulus)• Helmholtz-Kohlrausch effect is still important• Positive and negative contrast: impact on

brightness• Modelling: in progress

Part 2Conclusions

Brightness 50 -> answer38

CAM15u References

Withouck, M., Smet, K., Ryckaert, W., Hanselaer, P. (2015). Experimental driven modelling of the color appearance of unrelated self-luminous stimuli: CAM15u. Optics Express, 23 (9), 12045-12064.

Withouck, M., Smet, K., Hanselaer, P. (2015). Brightness prediction of different sized unrelated self-luminous stimuli. Optics Express, 23 (10), 13455-13466.

Ready-to-use code available:

www.github.com/ksmet1977/CAM15u

Peter.Hanselaer@kuleuven.be

Thank you!Questions?