Color Seminar · Mark Shaw, Hewlett Packard Company 15 • CIE Color Matching Functions (1931 2º...
Transcript of Color Seminar · Mark Shaw, Hewlett Packard Company 15 • CIE Color Matching Functions (1931 2º...
Color Seminar
Mark ShawHewlett Packard [email protected] @ p
SENSITOMETRY AND THEHUMAN VISUAL SYSTEM
• Basics
– Spectral Properties• Visible spectrum• Wavelength range ~ 400-700nm
– Reflectance• Light hits an object, then the eye• Perception is relative to a white referencep
– Radiance• Emissive device• Not illuminant dependant
3Mark Shaw, Hewlett Packard Company
• What does reflectance factor look like?
4Mark Shaw, Hewlett Packard Company
• How do we measure color?
– Spectral Analysis
– Tri-ChromaticTri Chromatic
5Mark Shaw, Hewlett Packard Company
• The Human Visual System
– The eye is an integrator
– Rods vs Coneshttp://www.colorado.edu/intphys/Class/
• Scotopic - Night vision, only Rods• Mesopic - Dusk vision, both Rods and Cones• Photopic - Day vision, only Cones
http://www.colorado.edu/intphys/Class/IPHY3730/image/figure6e.jpg
6Mark Shaw, Hewlett Packard Company
• Not all things are made equal
– Foveal Density
– 20 L Cones10 M C– 10 M Cones
– 1 S Cones
http://www.beercolor.com/color_basics1_files/image005.jpg
7Mark Shaw, Hewlett Packard Company
8Mark Shaw, Hewlett Packard Company
Color Constancy Demo
9Mark Shaw, Hewlett Packard Company
10Mark Shaw, Hewlett Packard Company
Look at the black square
11Mark Shaw, Hewlett Packard Company
12Mark Shaw, Hewlett Packard Company
COLOR THEORY
• Color Theory
– Trichromatic Theory Young and Helmholtz ( 1800’s )
– Opponent Theory Herring ( Late 1800’s )
www.handprint.com/HP/WCL/color2.html
14Mark Shaw, Hewlett Packard Company
• Color Matching
– How do we get a color match?
• Subjectively
• Quantatively
– Every observer is slightly different
• Work of Wright and Guild (1920’s)• Achromatic mixing experiment• Negative lobes occurred because the third
primary was needed in order to make aprimary was needed in order to make a match, that primary essentially acts as asubtraction
15Mark Shaw, Hewlett Packard Company
• CIE Color Matching Functions (1931 2º Observer and 1964 10º Observer )
2
1931 2 Degree CMFs
z2
1964 10 Degree CMFs
z
1
1.5
stim
ulus
Val
ue
xy1
1.5
stim
ulus
Val
ue
xy
0.5
Tris
0.5Tr
is
400 450 500 550 600 650 7000
Wavelength ()400 450 500 550 600 650 700
0
Wavelength ()
16Mark Shaw, Hewlett Packard Company
700
700
dxX400 dyY
400
700
dzZ 700
400
Reflectance Radiance
= S() R() = L() or E()
S is the illuminant spectraR is the reflectance factor of the object
L is the Luminosity, also calledE ( Emittance )
17Mark Shaw, Hewlett Packard Company
• A colorimetric match if …XX0000
21
dxdx 700
4002
700
4001
1
dydy
YY
700
2
700
1
21
2
the tristimulus values match under
400400
ZZ 21 specific conditions.
dzdz
ZZ
700
4002
700
4001
21
18Mark Shaw, Hewlett Packard Company
• Metamerism
– What does it mean if …
but
dxRSdxRS 700
40021
700
40011
dxRSdxRS 700
40022
700
40012
we have a metameric match under S1 but not under S2
19Mark Shaw, Hewlett Packard Company
• Observer Metamerism
– When two colors match for one observer but not for another
– Color blindness is one example
• Illuminant Metamerism
– Two samples match under one light source, but not another
20Mark Shaw, Hewlett Packard Company
http://tqc.eu/images/products/150x150/colorbox_lightsources_animation.gif
21Mark Shaw, Hewlett Packard Company
COLOR APPEARANCE
• Color Appearance Modeling
– Absolute
• Hue• Brightness• Brightness• Colorfulness
– Relative
• Hue• Lightness
Ch• Chroma
23Mark Shaw, Hewlett Packard Company
– CIE 1976 ( L*a*b* ) Color Space - aka CIELAB
31
16116*YYL
31
31
500*nn
n
YY
XXa
Y
Opponent
dxSX n
700
700
400
31
31
200*nn Z
ZYYb
Opponent Signals
dzSZ
dySY
n
n
700
400
400
400
Xn Yn Zn are the tristimulusvalues of the white point
24Mark Shaw, Hewlett Packard Company
– Converting Cartesian co-ordinates into a Polar space
*
2*2**
**
b
baC
LL
*1tan*
abhab
– Calculating Color Differences
2*2*2* baLE
Color differences are not normally distributed
76 baLE
25Mark Shaw, Hewlett Packard Company
COLOR MODELING
• How does one map colors between devices?
– First one must build colorimetric models of the source and destination devices
The range of colors the device can create is called the “Color Gamut”The range of colors the device can create is called the Color Gamut
– Then one must map all colors from the source gamut to the destination gamut
27Mark Shaw, Hewlett Packard Company
• How does one build a model of a device?
28Mark Shaw, Hewlett Packard Company
• Comparing Device Gamuts
The process of mapping colors from one device to another is called Gamut Mapping
29Mark Shaw, Hewlett Packard Company
• Color Conversion Workflow
Device Color Model
Color Appearance
Model
Color Appearance
Model
Device Color Model
Gamut Mapping
ConnectionSpace
30Mark Shaw, Hewlett Packard Company
SUMMARYSUMMARY
– The Human Visual System is an integrator
A l i t i t h l h ld if th i i diti i– A colorimetric match only holds if the viewing conditions remain unchanged
– Metamerism is a very important concepty p p
• Most current imaging systems rely upon the a metameric match
• Spectral matches would be ideal, but are currently prevented by technological y y glimitations
– Color appearance models enable the communication of color in meaningful termsg
– Color Modeling enables one to communicate color consistently
32Mark Shaw, Hewlett Packard Company