Color Appearance Models:

CIECAM02 and Beyond

IS&T/SID 12th Color Imaging Conference

Tutorial T1A, 11/9/04, 8:00-10:00AM

Mark D. Fairchild <[email protected], www.cis.rit.edu/fairchild>

RIT Munsell Color Science Laboratory <www.cis.rit.edu/mcsl>

Outline

Color Appearance Phenomena

Chromatic Adaptation
Structure of Color Appearance Models
CIECAM02
Image Appearance: iCAM

Color PDF of notes at <www.cis.rit.edu/fairchild/PDFs/CIC2004.pdf>

 Color Appearance
Phenomena
If two stimuli do not match in color appearance when (XYZ)1 = (XYZ)2,
then some aspect of the viewing conditions differs.

Various color-appearance phenomena describe relationships between

changes in viewing conditions and changes in appearance.

Bezold-Brücke Hue Shift

Abney Effect
Helmholtz-Kohlrausch Effect
Hunt Effect
Simultaneous Contrast
Crispening
Helson-Judd Effect
Stevens Effect
Bartleson-Breneman Equations
Chromatic Adaptation
Color Constancy
Memory Color
Object Recognition

Simultaneous Contrast
The background in which a stimulus is
presented influences the apparent color of the
stimulus.
Stimulus

Indicates lateral interactions and adaptation.

Background Stimulus Color-

Background Change
Appearance Change
Darker Lighter

Lighter Darker

Red Green

Green Red

Yellow Blue

Blue Yellow
Simultaneous Contrast
Example
(a)

(b)

Josef Albers
 Complex Spatial
Interactions

Hunt Effect
Corresponding chromaticities across indicated
relative changes in luminance (Hypothetical Data)
For a constant chromaticity, perceived
0.6
colorfulness increases with luminance.

0.5 As luminance increases, stimuli of

lower colorimetric purity are required
1
10 to match a given reference stimulus.
y
0.4 100
1000
10000

10000 1000 100 10 1

0.3
Indicates nonlinearities in visual
processing.
0.2
0.2 0.3 0.4 0.5 0.6
x
 Stevens Effect

1
Perceived lightness contrast increases with
increasing adapting luminance.

As adapting luminance increases:

dark colors look darker and
light colors look lighter.
L (Dark)

L (63 dB) Indicates luminance-dependent nonlinearities.

L (79 dB)

L (97 dB)
0.1
0.1 1
Y/Yn

Stevens & Hunt Effects

0.1 cd/m2 1.0 cd/m2 10 cd/m2

100 cd/m2 1000 cd/m2 10,000 cd/m2

 Bartleson-Breneman

Perceived lightness as a function of relative luminance

for various surround relative luminances
100
90
Apparent contrast in complex stimuli (i.e.
80
images) increases with increasing surround
luminance.
70
60
Decreased surround luminance increases the
50
brightness of all image colors, but the effect is
40
greater for dark colors.
30 Lightness (Average)

20 Lightness (Dim)

10 Lightness (Dark) Indicates a differential contrast effect (white-

0 point resetting).
0 10 20 30 40 50 60 70 80 90 100
Relative Luminance, Y

Surround Effect Demo

 Adaptation
Light Adaptation:
Decrease in visual sensitivity with increases in luminance.
(Automatic Exposure Control)

Dark Adaptation:
Increase in visual sensitivity with decreases in luminance.
(Automatic Exposure Control)

Chromatic Adaptation:
Independent sensitivity regulation of the mechanisms of color vision.
(Automatic Color Balance)
 Local Adaptation

Linear Mapping Perceptual Mapping

Chromatic Adaptation

1.25

The three cone types, LMS, are capable of

1
independent sensitivity regulation. (Adaptation
occurs in higher-level mechanisms as well.)
0.75

Magnitudes of chromatic responses are

0.5
dependent on the state of adaptation (local,
spatial, temporal). Afterimages provide evidence.
0.25

0
400 450 500 550 600 650 700
Wavelength (nm)
 Color Constancy
(Discounting)
0.54
Adapting Chromaticities
Achromatic Chromaticities
Incandescent We perceive the colors of objects to remain
0.52
<-- Hands
unchanged across large changes in
<-- No Hands illumination color.
•Not True
v' 0.50 •Chromatic Adaptation
•Poor Color Memory
•Cognitive Discounting-the-Illuminant
0.48
Daylight

<-- Hands
<-- No Hands
0.46
0.18 0.20 0.22 0.24 0.26

u'
Diamonds

With Tips
 On Backgrounds

Both Tips & Backgrounds!

 Purves “Brown”

Chromatic Adaptation
Modeling
Chromatic Adaptation:
Largely independent sensitivity regulation of the (three) mechanisms of color vision.

1.25

0.75

0.5

0.25

0
400 450 500 550 600 650 700
Wavelength (nm)
 Chromatic Adaptation
Models
Model: X1Y1Z1
L a = f(L, Lwhite ,...) 3x3
M a = f(M, M white ,...) L1M1S1
S a = f(S, S white ,...)
VC1
LaMaSa
Transform (CAT): VC2
XYZ 2 = f(XYZ1 , XYZwhite ,...)
L2M2S2
3x3
X2Y2Z2

Chromatic Adaptation Model

Output
Raw “Radiance” Images

Adapted “Perceptual” Images

 Chromatic Adaptation
Transform Output
Raw D65 “Radiance” Image

Raw A “Radiance” Image

A Image Transformed to
Corresponding D65 Appearance

Analysis of Chromatic
Adaptation Models
ADVANTAGES: DISADVANTAGES:
•Corresponding Colors •No Appearance Attributes
userdict
/mypsb
currentpoint
/newXScale
/newYScale
/psb
/pse /mypse
/mypsb •Thus,
{} store
/md
/psb newWidth
newHeight
/pse
pop
/newHeight
/picOriginY
load {}
store
def
storeColor
Cricket
known{/CricketAdjust
load/newWidth
store
/mypse
614
340
exch
div
div Reproductions
Software!
exch
def
picOriginY
/pse
def
def/picOriginX
picOriginX
load
true def
def}{/CricketAdjust
sub sub
def
exch
pop
defdef (e.g., Lightness, Chroma, Hue)
false def}ifelse

•Simpler •Can't Edit, Gamut Map, etc.

Since chromatic adaptation

models provide only nominal
scales, one could take all
viewing conditions into
account properly and never
know what color a stimulus is.

But, a chromatic adaptation

transform could be used as
input to index into a color-
order system to specify
appearance.
 Cone Excitations
1
How are they found? S M L
•Copunctal Points of Dichromats w/CMFs 0.75

Relative sensitivity
•Chromatic Adaptation w/Model
•Selective Retinal Conditioning/Thresholds
0.5
•Retinal Pigment Absorption Measurements
•Genetics
0.25

Why are they important?

0
•Produce the 1st-Stage Color Signals
380 480 580 680 780
•Subjected to Actions of Adaptation Mechanisms Wavelength, nm

XYZ-to-LMS

2
z
1

S M L 1.5
0.75
Relative sensitivity

Tristimulus values

L 0.400 0.708 !0.081 X y

x
0.5 1
M = !0.226 1.165 0.046 Y
0.25
S 0.000 0.000 0.918 Z 0.5

0
380 480 580 680 780
Wavelength, nm 0
380 480 580 680 780
Wavelength, nm
 Johannes von Kries
Johannes von Kries
"Father of Chromatic-Adaptation Models"

"If some day it becomes possible to

recognize and to distinguish in an objective
way the various effects of light by direct
observation of the retina, people will
perhaps recall with pitying smiles the efforts
of previous decades which undertook to
seek an understanding of the same
phenomena by such lengthy detours."

von Kries Hypothesis

"This can be conceived in the sense that the individual
components present in the organ of vision are
completely independent of one another and each is
fatigued or adapted exclusively according to its own
function."
-von Kries, 1902

La = kLL
M a = k MM
Sa = k S S

Von Kries thought of this “proportionality law” as an

extension of Grassmann’s Laws to span two viewing
conditions.
Modern “von Kries” Model
kL = 1L or 1 L
max white
La = L L Ma = M M Sa = S S
kM = 1M or 1 M max max max
max white

kS = 1 or 1
S max S white

Corresponding Colors (CAT):

! L1 # ! M1 # ! S1 #
L2 = L M2 = M max $ M max S2 = S
" Lmax $ max
1
2
" 1
2
" S max $ max 1
2

Matrix Form: XYZ Corresponding Colors (CAT):

1 0 0 1 0 0
La L max L X2 L max 2
0 0 Lmax 1
X1
Ma = 0 1 0 M Y 2 = M !1 0 M max 0 0 1 0 M Y1
M max 2 M max1
Sa 0 0 1 S Z2 0 0 S max 0 0 1 Z1
S max 2
S max 1

The Next Line...

“But if the real physiological equipment

is considered, on which the processes
are based, it is permissible to doubt
whether things are so simple.”
-von Kries, 1902
Some Evolution of CATs

Nayatani et al. Nonlinear Model

Fairchild (1991) & (1994) Models
Bradford Model
CIELAB & CIELUV
CAT02

Back to von Kries

Fairchild (2001)
Linear CATs can Perform Like Bradford

Optimization on Matrix (not LMS)

Relationship to “Spectral Sharpening”

Calabria and Fairchild (2001)

Herding CATs

Insignificant Differences

CAT02 in CIECAM02
Simple von Kies (100 years later!) non-LMS Matrix
 Linear CATs
Simple von Kries Model
“Optimized” XYZ-to-RGB Transform

X2 Rw 2 0 0 1/Rw1 0 0 X1
"1
Y2 = MCAT 0 Gw 2 0 0 1/Gw1 0 MCAT Y1
Z2 0 0 Bw 2 0 0 1/Bw1 Z1

!
MCAT Defines the Transform

Picking a CAT
Various Techniques used to Derive M CAT

Calabria & Fairchild (2001)

“Herding CATs”

Only LMS Significantly Different

TC8-01 Compromised on M CAT02

Basis of CIECAM02

0.7328 0.4296 "0.1624

MCAT 02 = "0.7036 1.6975 0.0061
0.0030 0.0136 0.9834

!
 What About Appearance?
Chromatic-adaptation models provide nominal scales for color appearance.

Two stimuli in their relative viewing conditions match each other.

BUT what color are they??

We need color-appearance models to get interval and ratio scales of:

Lightness,
Brightness,
Hue,
Chroma, and
Colorfulness.

Color Appearance Models

A Color Appearance Model provides mathematical formulae
to transform physical measurements of the stimulus and
viewing environment into correlates of perceptual attributes
of color (e.g., lightness, chroma, hue, etc.).

Nominal Ordinal Interval Ratio

unequal intervals
Ratio Scale

equal intervals
Perceptual

equal intervals

? ?
no intervals no zero no zero zero
no zero

e.g., numbers e.g., e.g., e.g., length

Physical Ordinal Scale on football hardness temperature
players °F
 Flow Chart
Measure Physical Stimuli
!(")

Stimulus, Background,
Surround, etc.

Calculate Tristimulus Values

XYZ (LMS)

Stimulus, Background,
Surround, etc.

Calculate Correlates of
Perceptual Attributes

Lightness, Brightness, Chroma,

Colorfulness, Hue, etc.

Structure of CAMs
1.25

Chromatic Adaptation Transform 0.75

(to Implicit or Explicit Reference Conditions)

0.5
Corresponding Colors
0.25

0
400 450 500 550 600 650 700
Wavelength (nm)

Color Space Construction

Cone Responses
Opponent Responses
Appearance Correlates
 CIELAB as an Example

CIELAB Does:
•Model Chromatic Adaptation
•Model Response Compression
•Include Correlates for Lightness, Chroma, Hue
•Include Useful Color Difference Measure

CIELAB Doesn't:
•Predict Luminance Dependent Effects
•Predict Background or Surround Effects
•Have an Accurate Adaptation Transform

CIELAB as a CAM

L* = 100 (white)
yellowish
b*

Chromatic Adaptation -a* a*

reddish
greenish
X/Xn, Y/Yn, Z/Zn
-b*
bluish

Opponent Processes L* = 0 (black)

X-Y
Y-Z Lightness L

Yellowness +b

Uniform Spacing
Constants 116, 500, 200 Greenness –a Redness +a

Cube Root
Blueness –b
 CIELAB Equations

L* = 116f(Y / Y n ) ! 16
a* = 500[ f(X / X n ) ! f(Y / Y n )]
b* = 200[ f(Y / Y n ) ! f(Z / Z n )]

f(") = (")1/3 " > 0.008856

f(") = 7.787(") + 16/ 116 " # 0.008856

CIELAB Lightness
L* = 116f(Y / Y n ) ! 16

f(") = (")1/3 " > 0.008856

f(") = 7.787(") + 16/ 116 " # 0.008856

10
9
8
7
6
Munsell
Value

5
4
3
2
1
0
0 1 2 3 4 5 6 7 8 9 10
L*/10
 CIELAB Chroma
1/ 2
C* = (a * 2 +b *2 )

Increasing Chroma,
C*

Neutrals Have Zero Chroma, C* = 0.0

Saturation in CIELAB
Due to the lack of a related chromaticity diagram,
saturation is not officially defined in CIELAB.

However recalling the definitions of chroma (colorfulness/

brightness of white), lightness (brightness/brightness of
white), and saturation (colorfulness/brightness).
L* Constant Constant
Chroma Saturation

Constant
Lightness
Saturation = C*/L*

C*
 CIELAB Hue
90°
+b*

h°
ab

" b *$ +a*
h ab = tan !1 180° 0°
# a *% -a*

-b*

Relative Hue Scale — Where's Red? 270°

Approximate Hue Angles of NCS Unique Hues

R — 24°
Y — 90°
G — 162°
B — 246°

Note: The number of discriminable hue steps is not equal between each of the unique hues.

CIELAB Performance
100
Value 3

50

b* 0

-50 100
Value 5

-100 50
-150 -100 -50 0 50 100 150
a*
b* 0

-50
125
Value 7
-100
-150 -100 -50 0 50 100 150
75
a*

b* 25

-25

-75
-150 -100 -50 0 50 100 150
a*
 Why Not Just CIELAB?
Positive Aspects:
•Accounts for Chromatic Adaptation
•Works Well for Near-Daylight Illuminants
(also Medium Gray Background & Surround and Moderate Luminance Levels)
•
Negative Aspects:
•Does Not Account for Changes in:
Background
Surround
Luminance
Cognition
•Cannot Predict Brightness & Colorfulness
•"Wrong" von Kries Transform Works Poorly for Large Changes From Daylight
•Constant-Hue Predictions could be Improved
(especially Blue)

CIELAB Makes a Good, Simple Baseline for Comparison

Beyond CIELAB
• More Accurate Adaptation Transform

• Luminance Dependencies

• Surround Dependencies

• Brightness & Colorfulness

• Hue Linearity

(Hunt, Nayatani, RLAB, LLAB, CIECAM97s, CIECAM02)

 Extending CIELAB

Main Limitation is “Wrong” von Kries

Can be Replaced with More Accurate CAT
CIELAB Under Daylight a Very Good Color
Space

CIELAB plus CAT Concept

Stimulus & Viewing Conditions
Colorimetry

Accurate CAT (e.g.,

CAT02)

Corresponding Colors (XYZ) and

Reference Illuminant (XnYnZn)

CIELAB Equations

CIELAB Lightness, Chroma, Hue

(L*C*h) to Describe Appearance
CIELAB plus CAT Example
Step 1: Obtain colorimetric data for stimulus and
viewing conditions.

Step 2: Use CAT02 to compute corresponding colors

for CIE Illuminant D65 (and 1000 lux).

Step 3: Compute CIELAB coordinates using

corresponding colors from step 2 and D65 white.

Step 4: Use CIELAB L*C*h as appearance correlates.

CIECAM02
 Need for CIECAM02

• Vienna Experts Symposium (1996)

• Industrial Demand

• Uniformity of Practice (like CIELAB)

History
• Task Assigned to TC1-34 (1996)

• CIECAM97s Completed May 1997 !!

• Several Suggestions for Improvements

• TC8-01 Tasked with Suggesting Revisions (1998)

• CIECAM02 Published Nov. 2002

Where Did CIECAM97s Come
From?
Examples of Model Pedigree Include:

•Bradford Chromatic-Adaptation Transform (Lam, 1985; Luo, 1997)

•Different Exponent on Short-Wavelength (Nayatani et al., 1982)

•Partial Adaptation Factors (Fairchild, 1996; Nayatani, 1997)

•Cone Responsivities (Estevez; see Hunt and Pointer, 1985)

•Hyperbolic Response Function (Seim and Valberg, 1986)

•R-G and Y-B Scales (Hunt, 1994; Nayatani, 1995)

•Surround Effects (Bartleson and Breneman, 1967)

•No Negative Lightness Predictions (Nayatani, 1995, Fairchild, 1996)

•Chroma Scale (Hunt, 1994)

CIECAM97s & CIECAM97c

• Comprehensive version that includes a wide
range of visual phenomena.

• Simplified version (fully compatible) that is

adequate for practical applications.

• CIECAM97s Exists (May, 1997)

• CIECAM97c Does Not (No Apparent Demand?)

Changes Considered (TC8-01)
• Correction of Surround Anomaly in Nc
• Adjustment of J for Zero Luminance
• Linear Adaptation Transform (Simple Inversion)
• Continuously Variable Surround Compensation
• Reduce Expansion of Chroma Scale for Near Neutrals
• Define Rectangular Coordinates
• References and Summary (submitted to CR&A/TC8-01)
<www.cis.rit.edu/fairchild/PDFs/CIECAM97sRev.pdf>

CIECAM02
• Revision of CIECAM97s

• Simplified and Improved

• Just Published (CIC10, 2002 … CIE Pub. 159:2004)

• No “s” since there is no CIECAM02c

 Inputs
LA: Adapting Field Luminance in cd/m2
(often 20% of the luminance of white)

XYZ: Relative Tristimulus Values of the Sample

XwYwZw: Relative Tristimulus Values of the White

Yb: Relative Luminance of the Background

D: Specifies the Degree of Adaptation:

D = 1.0, (Complete Adaptation or Discounting)
D = 0.0, (No Adaptation)
D in Between, (Various Degrees of Incomplete Adaptation)

Changes from CIECAM97s:

None

Stimulus

• Uniform Patch of About 2° Angular Subtense

• Single Pixel in an Image?

• Required Measurements
Relative XYZ (CIE 2°)
 Background

• Area Immediately Adjacent to Stimulus Out to About 10°

• Average for Images? (20% Gray)

• Required Measurements
Relative Y (called Yb)

Surround

• Remainder of Visual Field Outside Background

• Required Measurements
Often Categorical
Average (>20%)
Dim (0-20%)
Dark (0%)

Changes from CIECAM97s:

Continuous Surround Compensation
 Adapting Stimulus

• Stimulus that Sets the State of Adaptation

• 0.2xWhite (Gray)
• Scene Average (Spatially Weighted?)

• Required Measurements
Absolute XYZ (CIE 2°, cd/m2)
or
Relative XYZ and LA (cd/m2)

Parameter Decision Table

c: Impact of Surround
Nc: Chromatic Induction Factor
F: Factor for Degree of Adaptation

Viewing Condition c Nc F
Average Surround 0.69 1.0 1.0
Dim Surround 0.59 0.9 0.9
Dark Surround 0.525 0.8 0.8

Changes from CIECAM97s:

Removal of 2 Conditions (Large Samples & Cut-Sheet)
FLL Removed (Always 1.0)
Correction of Nc (Now Monotonic)
Change in F & Nc for Dark (from 0.9)
Continuously-Variable Surround

Changes from CIECAM97s:

New Feature

Discounting

• Is the stimulus viewed as an illuminated object

(Discounting) or as self-luminous (No Discounting)?

• Required Measurements
) # 1 & #% "La"42&( ,
Yes: D=1.0 D = F+1" % (e 92 .
$ '

No: Use Equation +* $ 3.6 ' .-

No Adaptation: D=0.0

Changes from CIECAM97s:

Simplified Equation
Real-World Discounting
Outline of Model Structure

• Chromatic Adaptation Transform

(to Implicit Ill. E Reference Conditions)
Corresponding Colors

• Color Space Construction

Cone Responses
Opponent Responses
Appearance Correlates

Chromatic Adaptation
 Chromatic Adaptation
Transform: CAT02

• von Kries Normalization

• Now Linear (normal von Kries)
• “Sharpened” “Cone” Reponses (Optimized)
• Generally Good Performance
(Not Different from CIECAM97s)

Transform to RGB
Responses
" R% "X %
$ ' $ '
G
$ ' = M CAT 02 $Y '
$#B'& $# Z '&
" 0.7328 0.4296 (0.1624% " 1.0961 (0.2789 0.1827%
$ ' (1 $ '
MCAT 02 = $(0.7036 1.6975 0.0061 ' MB = $ 0.4544 0.4735 0.0721'
#$ 0.0030 0.0136 0.9834 '& $#(0.0096 (0.0057 1.0153'&

!
!

Changes from CIECAM97s:

Simplified Transform (No /Y)
New Optimized Matrix
 “Sharpened” “Cone”
Responses
2
1.200
R(H) R(B)
HPE L
G(H) G(B)
1.75 HPE M
B(H) B(B)
HPE S
1.000
1.5 CAT02 R
CAT02 G
1.25 CAT02 B
0.800
1

Relative Response
0.75 0.600

0.5

0.400
0.25

0
0.200

-0.25
400 450 500 550 600 650 700
Wavelength (nm)
0.000
360 410 460 510 560 610 660 710 760

-0.200

Wavelength (nm)

Changes from CIECAM97s:

RGB(CAT02) Slightly Different from RGB(B)

Adaptation Transform

Rc = [YW ( D /Rw ) + (1" D)] R

Gc = [YW ( D /Gw ) + (1" D)]G

! [
Bc = YW ( D /Bw ) + (1" D) B ]
!

!
Changes from CIECAM97s:
YW Added (for cases it is not 100)
Nonlinearity on B Removed
 Color Space

• Based on Structure within Hunt Model & CIECAM97s

• Enhancements Based on Various Tests, etc.

• Hyperbolic Nonlinearity

• Color Difference Signals

• Appearance Correlates

Intermediate Parameters
Some numbers for further computations...

FL: Luminance Level Adaptation Factor

n: Background Induction Factor
Nbb and Ncb: Brightness and Chromatic Background Factors
z: Base Exponential Nonlinearity
k = 1/( 5L A + 1)
2
(
FL = 0.2k 4 (5L A ) + 0.1 1 ! k 4 ) (5L A )1/ 3
n = Yb / Yw
N bb = N cb = 0.725(1/ n) 0.2

z = 1.48 + n1/ 2
Changes from CIECAM97s:
Slight Change in z Equation

"
Adapted Cone Responses
"R'% " Rc %
$ ' (1 $ '
$G'' = M H MCAT 02 $Gc '
$#B''& $# Bc '&
" 0.38971 0.68898 (0.07868 % "1.9102 (1. 1121 0.2019%
$ ' (1 $ '
M H = $ (0.22981 1.18340 0. 04641 ' MH = $0.3710 0.6291 0.00 '
$# 0.00 0.00 1.00 '& $# 0.00 0.00 1.00 '&
!

0.42
400( FL R' /100)
R' a = + 0.1
[(F R' /100)
L
0.42
+ 27.13 ]
0.42
400( FL G' /100)
G'a = + 0.1
[(F G' /100)
L
0.42
+ 27.13 ]
! 0.42
400( FL B' /100)
B'a = + 0.1
[(F B' /100)
L
0.42
+ 27.13 ]
!

Changes from CIECAM97s:

! No Y Multiplication Required Before Transform
Modified Nonlinearity (Square-Root Behavior over Larger Range)

Opponent Responses

A = [2R' a +G'a +(1/20)B'a !0.305] N bb

a = R' a !12G' a /11 + B' a / 11

"
b = ( 1/9 )(R ' a + G' a !2B'a )

Changes from CIECAM97s:

Adjusted Constant in A (Perfect Black)
Appearance Correlates
• Brightness, Lightness
• Colorfulness, Chroma, Saturation
• Hue

• Built Up to Fit Experimental Data

• Need 5 of 6 to Fully Describe Appearance

Hue

The degree to which a stimulus can be described

as similar to or different from stimuli that are
described as red, green, blue, and yellow.

!1 1) # " & ,
h = tan (b / a) e = +cos% h + 2( + 3.8.
4 * $ 180 ' -

Red: h = 20.14, e = 0.8, H = 0 or 400,

Yellow: h = 90.00, ! e = 0.7, H = 100,
Green: h = 164.25, e = 1.0, H = 200,
Blue: h = 237.53, e = 1.2. H = 300

Changes from CIECAM97s:

e Now Defined Analytically
 Lightness
The brightness of a stimulus relative to the
brightness of a stimulus that appears
white under similar viewing situations.

cz
J =100 (A/A w )

Changes from CIECAM97s:

None

Brightness

The perceived quantity of light emanating from a stimulus.

0.5
Q = ( 4 /c )( J /100) ( Aw + 4 )FL0.25

Changes from CIECAM97s:

New Constants, FL Added, Simplified
 Chroma
The colorfulness of a stimulus relative to
the brightness of a stimulus that appears
white under similar viewing conditions.

1/ 2
(50000 /13)N c N cb e( a 2 + b 2 )
t=
R'a +G'a +(21/20)B' a

0.5 n 0.73
! C = t 0.9 ( J /100) (1.64 " 0.29 )

! Changes from CIECAM97s:

t simplified form of former s (constants in new e formula)
C Simplified and Modified (Munsell, Low Chromas)

Colorfulness

The perceived quantity of hue content

(difference from gray) in a stimulus.

Colorfulness increases with luminance.

M = CFL0.25

! Changes from CIECAM97s:

0.25 instead of 0.15
 Saturation

The colorfulness of a stimulus relative to its own brightness.

s = 100 M Q

!
Changes from CIECAM97s:
Simple, Logical Definition
Data Now Available

Chroma/Saturation
Lightness
Lightness

Chroma Saturation
 Definitions in “Equations”

Chroma = (Colorfulness)/(Brightness of White)

Saturation = (Colorfulness)/(Brightness)

Lightness = (Brightness)/(Brightness of White)

Saturation
= (Chroma)/(Lightness)
= [(Colorfulness)/(Brightness of White)][(Brightness of White)/(Brightness)]
=(Colorfulness)/(Brightness)

How Many Terms?

Any color perception can be described completely by its:
•Brightness
•Lightness
•Colorfulness
•Chroma
•Hue
and only one of brightness or colorfulness is required to derive the others.

In general, the relative appearance attributes are adequate for object

colors in typical viewing environments:
•Lightness
•Chroma
•Hue
Saturation is often redundant.
 Lightness/Chroma vs.
Brightness/Colorfulness
When predicting color matches across different
viewing conditions, Lightness-Chroma matches
are not identical to Brightness-Colorfulness
matches. See Nayatani et al. (1990).

For related colors, and typical conditions,

Lightness-Chroma matching (and therefore
reproduction) is the only practical choice.

Reproduction at Higher
Luminance

Lightness/Chroma

Brightness/Colorfulness

Original (50 cd/m2)

Reproductions (5000 cd/m2)

 Reproduction at Lower
Luminance

Lightness/Chroma

Brightness/Colorfulness

Original (5000 cd/m )

2

Reproductions (50 cd/m2)

Color Space (Rectangular)

• CIECAM02 is Expressed in Cylindrical Coordinates, JCh

• Coordinate Transformation Required for Rectangular Plots

bC = Csin(h)

aC = Ccos(h)

• Alternative Combinations (QMh, Qsh, Jsh)

Changes from CIECAM97s:
Now Explicitly Defined
 Inverse Model
• Necessary to Reproduce Colors

• Application / ICC Flow Chart

Profile
Original Connection Reproduction
Space

CAM CAM

Inversion

• CIECAM97s is Not Quite Analytically Invertible

15 Steps with 1 Approximation

• CIECAM02 is Invertible
Linear CAT Fixes It!

Eqs. in CIE Report; See <www.colour.org/tc8-01/>.

 Image Appearance
Modeling

What is an Image Appearance

Model?
Image appearance models extend color appearance
models to include spatial vision, temporal vision,
and image difference/quality properties.

They account for more complex changes in visual

response in a more automated manner.
What are Some of the Missing
Links?
Spatial Vision (Filtering & Adaptation)
Scene Interpretation
Computational Surround Effects
Color/Image Difference Metrics
Image Processing Efficiencies

Other Spatial Models

S-CIELAB (Zhang & Wandell)
CVDM (Feng et al.)
Sarnoff Model (Lubin et al.)
Spatial ATD (Granger)
MOM (Pattanaik et al.)
Modular Image Difference (Johnson et al.)
 Meet iCAM

iCAM — image Color Appearance Model

A simple framework for color appearance, spatial vision effects, image
difference (quality), image processing, and temporal effects (eventually).

Image Appearance & Quality

• IQ (Thresholds & Magnitudes)
• Combine with Color Appearance

• Get “Image Appearance”

 Moving Image
Appearance & Quality
• Temporal Adaptation & Filtering

Pointwise iCAM
 Spatial iCAM

iCAM Performance
Examples
Chromatic Adaptation Transform
(CAT)
Color Appearance Scales
Constant Hue Lines
Simultaneous Contrast
Chroma Crispening
Hue Spreading
HDR Tone Mapping
Image Difference (Quality)
Basic Appearance Attributes

Chromatic Adaptation Transform (CAT)

Identical to CIECAM02

Color Appearance Scales

Similar to Munsell / CIECAM02 (limited)

Constant Hue Lines

Best Available (IPT)
Facilitates Gamut Mapping

iCAM Simultaneous
Contrast

Original Image iCAM Lightness

iCAM Chroma Crispening

Original Image iCAM Chroma

http://www.hpl.hp.com/personal/Nathan_Moroney/

iCAM Spreading

Original Image iCAM Hue

iCAM High-Dynamic-Range
Tone Mapping

Earlier-Model Results

<www.debevec.org>

Image Difference Process

Reproduction 1 Reproduction 2

Mean !E*ab 2.5 Mean !E*ab 1.25

Mean !"m 0.5 Mean !"m 1.5

Spatial Filtering, Local Attention, Local & Global Contrast, CIE Color Difference
iCAM Image Difference
(Image Quality)
1.2

Model Prediction
0.8

!Im 0.6

0.4

0.2

0
-5 -4 -3 -2 -1 0 1 2
Perceived Difference

Image Difference Prediction (Sharpness Data)

iCAM Image Difference

(Image Quality)
(a)
14

12

10
Model Prediction

!Im
6

0
-4.00 -3.00 -2.00 -1.00 0.00 1.00 2.00
Perceived Contrast

Image Difference Prediction (Contrast Data)

 Spatial iCAM

Detailed references to each step and

coded examples on the internet.

Open-Source Science

<www.cis.rit.edu/mcsl/iCAM>

Image
Rendering
Examples

<www.debevec.org>
 Conclusions
Ingredients
Color Appearance Model
Spatial Adaptation & Filtering Models
Temporal Adaptation & Filtering Models
Image Difference Metrics
Results
Still & Video Rendering Algorithms
Still & Video Quality Metrics
 Free Code

<www.cis.rit.edu/mcsl/iCAM/>
Mathematica, Matlab, IDL, C++, etc.
Updates.

Summary

Color Appearance Phenomena

Chromatic Adaptation
Structure of Color Appearance Models
CIECAM02
Image Appearance: iCAM
 Reading List & Errata Sheet
M.D. Fairchild, Color Appearance Models, Addison-
Reading List from SIMC 703, Color Appearance, attached.
Wesley, Reading, Mass. (1998)

Watch for the 2nd Ed. in late 2004.

<www.cis.rit.edu/fairchild/CAM.html>
ROCHESTER INSTITUTE OF TECHNOLOGY
Munsell Color Science Laboratory

SIMC 703 Color Appearance

READING LIST:

Course Text
M.D. Fairchild, Color Appearance Models, Addison-Wesley, Reading, MA (1998).

Basic and Advanced Colorimetry

W.D. Wright, 50 years of the 1931 CIE standard observer for colorimetry, AIC
Color 81, Paper A3 (1981).
G. Wyszecki, Current developments in colorimetry, AIC Colour 73, 21-51 (1973).

Color Appearance Terminology

A. Hard and L. Sivik, NCS—Natural Color System: A Swedish standard for color
notation, Color Res. Appl. 6, 129-138 (1981).
S.M. Newhall, Preliminary report of the O.S.A. subcommittee on the spacing of
the Munsell colors, J. Opt. Soc. Am. 30, 617-645 (1940).
C.J. Bartleson, Brown, Color Res. Appl. 1, 181-191 (1976).

Color Appearance Phenomena

D.M. Purdy, Spectral hue as a function of intensity, Am. J. Psych, 43, 541-559
(1931).
M.D. Fairchild and E. Pirrotta, Predicting the lightness of chromatic object colors
using CIELAB, Color Res. Appl. 16, 385-393 (1991).
R.W.G. Hunt, Light and dark adaptation and the perception of color, J. Opt. Soc.
Am. 42, 190-199 (1952).
H. Helson, Fundamental problems in color vision. I. The principle governing
changes in hue, saturation, and lightness of non-selective samples in
chromatic illumination, J. Exp. Psych. 23, 439-477(1938).
J.C. Stevens and S.S. Stevens, Brightness functions: Effects of adaptation, J. Opt.
Soc. Am. 53, 375-385 (1963).
M.D. Fairchild, Considering the surround in device-independent color imaging,
Color Res. Appl. 20, 352-263 (1995).

1
Viewing Conditions
OSA, Psychological concepts: Perceptual and affective aspects of color, Chapter 5
in The Science of Color, Optical Society of America, Washington, 145-171 (1963).

Chromatic Adaptation
J. von Kries, Chromatic adaptation, Festschrift der Albrecht-Ludwig-Universität,
(Fribourg) (1902) (Translation: D.L. MacAdam, Sources of Color Science, MIT
Press, Cambridge, (1970)).
H. Helson, D.B. Judd, and M.H. Warren, Object color changes from daylight to
incandescent filament illumination, Illum. Eng. 47, 221-233 (1952)
W.D. Wright, Why and how chromatic adaptation has been studied, Color Res.
Appl. 6, 147-152 (1981).
M.D. Fairchild, Chromatic adaptation to image displays, TAGA 2, 803-824 (1992).
J. Neitz, J. Carroll, Y. Yamauchi, M. Neitz, and D.R. Williams, Color perception is
mediated by a plastic neural mechanism that is adjustable in adults, Neuron
35, 1-20 (2002).

Color Appearance Modeling

Y. Nayatani, K. Takahama, H. Sobagaki, and K. Hashimoto, Color-appearance
model and chromatic adaptation transform, Color Res. Appl. 15, 210-221 (1990).
R.W.G. Hunt, Revised colour-appearance model for related and unrelated
colours, Color Res. Appl. 16, 146-165 (1991).
R.W.G. Hunt, An improved predictor of colourfulness in a model of colour
vision, Color Res. Appl. 19, 23-26 (1994).
M.D. Fairchild and R.S. Berns, Image color appearance specification through
extension of CIELAB, Color Res. Appl. 18, 178-190 (1993).
M.D. Fairchild, Refinement of the RLAB color space, Color Res. Appl. 21, 338-346
(1996).
CIE, The CIE 1997 Interim Colour Appearance Model (Simple Version),
CIECAM97s, CIE Pub. 131 (1998).
M.D. Fairchild, A revision of CIECAM97s for Practical Applications, Color Res.
Appl. 26, 418-427 (2001).
N. Moroney, M.D. Fairchild, R.W.G. Hunt, C. Li, M.R. Luo and T. Newman The
CIECAM02 Color Appearance Model, IS&T/SID CIC 10, Scottsdale, 23-27
(2002).

2
Testing Color Appearance Models
M.R. Luo, A.A. Clarke, P.A. Rhodes, A. Schappo, S.A.R. Scrivner, and C.J. Tait,
Quantifying colour appearance. Part I. LUTCHI colour appearance data, Color
Res. Appl. 16, 166-180 (1991).
R.W.G. Hunt and M.R. Luo, Evaluation of a model of colour vision by magnitude
scalings: Discussion of collected results, Color Res. Appl. 19, 27-33 (1994).
L. Mori, H. Sobagaki, H. Komatsubara and K. Ikeda, Field trials on CIE chromatic
adaptation formula, Proceedings of the CIE 22nd Session, 55-58 (1991).
P.J. Alessi, CIE guidelines for coordinated research on evaluation of colour
appearance models for reflection print and self-luminous display
comparisons, Color Res. Appl. 19, 48-58 (1994).
K.M. Braun and M.D. Fairchild, Testing five color appearance models for
changes in viewing conditions, Color Res. and Appl. 22, 165-174 (1997).
C.J Li, M.R. Luo, R.W.G. Hunt, N. Moroney, M.D. Fairchild, and T. Newman, The
performance of CIECAM02, IS&T/SID CIC 10, Scottsdale, 28-32 (2002).

Spatial Models
E.M. Granger, Uniform color space as a function of spatial frequency, SPIE 1913,
449-461 (1993).
X. Zhang and B.A. Wandell, A spatial extension of CIELAB for digital color
image reproduction, SID Digest. 19, 27-33 (1996).
X. Zhang and B.A. Wandell, Color image fidelity metrics evaluated using image
distortion maps, Signal Processing 70, 201-214 (1998).
G.M. Johnson and M.D. Fairchild, A top down description of S-CIELAB and
CIEDE2000, Color Res. and Appl. 28, in press (2003).
M.D. Fairchild, and G.M. Johnson, Meet iCAM: A next-generation color
appearance model, IS&T/SID CIC 10, Scottsdale, 33-38 (2002).