1. Colour electrochromism and Thomas Bangert thomas.bangert@qmul.ac.uk

2. part 1: The Colour Model The Munsell Colour Model
actual mapping to human vision
A colour catalog vs a colour model

3. color catalog vs colour model
catalog requires selection of colours based on perceptual matching
partial colour model codes spectrum as systematic mixing of wavelengths
true colour model codes the color of spectrum
X=100,Y=100,Z=0
Yellow

4. Colour as information
a theory of information processing.

5. Colour Reproduction true colour code + specs of viewer = image
colour defined by code
viewer can be group or individual
display decides how to create colour from code
gives perceptual predictability
Yellow
Orange +
bluish-red
Magenta

6. The Standard Observer from colour matching studies
CIE1931 xy chromaticity diagram primaries at: nm, 546.1nm, 700nm
The XYZ sensor response
Y is defined as luminance difference from Y is the colour information The Math:
… 2-d as z is redundant

7. Understanding CIE chromaticity
x and y show difference from Y
Best understood as a failed colour circle
White in center
Saturated / monochromatic wavelengths on the periphery
Everything in between is a mix of white and the colour
Circular colour models are the holy grail of colour theory … so far no one has succeeded!

8. But does the CIE model work?
Does it match?
Problem #1: ‘negative primaries’
Problem #2: no definition of colour

9. Colour Sensor response to monochromatic light
Human
Bird
4 sensors
Equidistant on spectrum
What are these sensors used for?
What information is needed?
my answer is: Wavelength

10. How to calculate wavelength with 2 poor quality luminance sensors.1 .
a shift of Δ
from a known reference point . 8 G R . 6 . 4 . 2 .
λ-Δ λ
λ+Δ
Wavelength
Roughly speaking:

11. the ideal light stimulus
Monochromacy: The reason we see rainbows is because the human visual system works with single wavelength light -- monochromatic light
monochromatic stimulus
This is the underlying paradigm!
Allows wavelength to be measured in relation to reference.

12. Problem: natural light is not ideal
Light stimulus might not activate reference sensor fully.
Light stimulus might not be fully monochromatic.
ie. there might be white mixed in

13. Solution:
Then reference sensor can be normalized
A 3rd sensor is used to measure equiluminance.
Which is subtracted.
Equiluminance & Normalization – essential to finding wavelength, can also called saturation and lightness

14. a 4 sensor design 2 opponent pairs only 1 of each pair can be active
min sensor is equiluminance

15. Human Retina only has 3 sensors! What to do?
We add an emulation layer.
Hardware has 3 physical sensors but emulates 4 sensors
No maths … just a diagram!

16. Testing Colour Opponent model
What we should see
What we do see
There is Red in our Blue – the problem of Purple

17. Pigment Absorption Data of human cone sensors
Red > Green

18. Dual Opponency with Circularity
an ideal model using 2 sensor pairs

19. a circular colour model
We divide colour coding and colour reproduction:
Coding no need to link to specific observer – ideal observer not linked specifically to human vision
Display decides how best to present colour to observer – making colour anomalies fit

20. Part 2 – Reproducing Colour
Part 1 – Coding Colour
fully circular
universal ideal observer
Part 2 – Reproducing Colour
takes knowledge about observer and optimizes/distorts to the individual/group
improved or natural reproduction modes

21. Coding Natural Colour Problem #1: Real world is not monochromatic
Spectrum of a common yellow flower

22. Colour coding … for dual channel opponency
Problem # 1
easy to solve
we simply assume monochromacy
when stimuli are not monochromatic opponent channels simply subtract to 0
green, yellow and red are active
r-g = 0
leaving only yellow
b = 0
stimuli equivalent to monochromatic

23. Opponent Coding Only primaries are true colours
all other colours are intermediary
… and can be generated by proportions of primaries!

24. Accurate colour reproduction … for humans
Problem # 2
Any colour may be displayed by a combination of 2 primaries
but the location of primaries can vary between individuals
and intermediary locations can be distorted

25. Accurate colour reproduction … tuned to the individual
primaries must be mapped for the individual
mid-points must be mapped
Provides an individual colour profile … a map of the primaries and intermediary points.
467
517
573
644
545
503
603

26. tunable primaries 573 644 517 467 W a v e l e n g t h ( n m ) 1 . 2
Yellow
644 1 .
517 . 8
467
Red . 6
Green . 4
Blue . 2 . - .2 3 5 4 4 5 5 55 60 65
700 W a v e l e n g t h ( n m )

27. Part 3 testing the theory
is it sound?
is it useful?
does human vision use it?
Is there empirical evidence to support paradigm + theory?
note: a theoretical model about information is the information itself!

28. Apparatus monochromator light source
equal light across visible spectrum

29. the stimuli

30. Transition Colour Matching
generate subject selectable monochromatic stimuli
subject selects colour
perceptual primaries are calculated

31. Results no leading questions -- only “blue”
4 primaries (pure colours) naturally resolve to blue, green, yellow and red
primaries are equidistant
transitions worked for all subjects
Most subjects see peripheral colours
red in blue
40% could see “magenta” – blue in red
potential problem: people treat purple as if it were primary
some colour blind people can’t see purple

32. histogram of results

33. results from mapping colour vision

34. Application
natural colour reproduction

35. Luminance: High Dynamic Range
Current display technology: – 100 cd/m2
(currently pushed up to 500, but designed for 100 cd/m2)
DICOM GSDF: – 4,000 cd/m2
(defined for grayscale medical imaging only)
Natural environment: – 10,000 cd/m2

36. Coding HDR
… using an absolute lumiance code rather than a relative code
HDR is here now … using multiple exposure!

37. the colour of infra-red (650-750nm)
not the stereo-type
but true infra-red – high wavelength light
remove the filter from a digital camera & it will work in the infra-red
Images in the infra-red produced by enthusiasts now!
What is the colour when you go beyond red?

38. related work: Dolby

39. Examples of real world colour?
Colours are often computed, not measured!

40. … an extreme example
What is the colour?

41. Part 4 – Building Real Colour Displays
Colour coded based on ideal model
colour model based on perception of colour not retinal sensors
Colour display tuned to the individual

42. Electrochromism similar to rechargeable battery
electrolyte is source of ions
electrical potential pushes ions into electrochromic layer
ions cause oxidation/reduction and the absorption/colour changes
reverse the polarity and the ions are forced out and the reaction reverses
colour change is caused by some wavelengths being absorbed
the basic application is a smart window

43. EcoPix
The aim of EcoPix is to turn electrochromism into useful display technology
€1,000,000 EU funded project
main partners are METU in Turkey, CENTI in Portugal and QMUL
full colour (subtractive RGB)
displays are intended to be exterior, billboard size with natural illumination

44. work so far Samples we are receiving from METU: PEDOT:PSS Blue/Cyan
single layer devices using ITO as electrodes
roughly 35x35mm

45. Blue/Cyan without white balance / xenon light source
light from xenon short – similar to sunlight
samples absorb about 50% of light
coloured state absorbs further 50% of light

46. Blue/Cyan Transmittance (white balanced)
potential progressively applied

47. Transparent / Coloured States

48. Transparent / Coloured state transition
frame rate: 25 fps
state change takes about 2 frames
80ms
very low current
order of 100µa for state change in current samples
stable for approx. 30s

49. Lab Facilities Dedicated Colour Lab to deliver QMUL share of EcoPix.
Standard 35mm photographic slide projection
50x50mm slides
xenon light (used in cinema)
room size front and rear projection screen
Precision programmable power supply

50. measurement by rear projection
Sensor

51. Questions?
References
Poynton, C. A. (1995). “Poynton’s Color FAQ”, electronic preprint.
Bangert, Thomas (2008). “TriangleVision: A Toy Visual System”, ICANN 2008.
Goldsmith, Timothy H. (July 2006). “What birds see”. Scientific American: 69–75.
Neitz, Jay; Neitz, Maureen. (August 2008). “Colour Vision: The Wonder of Hue”. Current Biology 18(16): R700-r702.