Assessing color reproduction tolerances in commercial print workflow
-
Upload
giordano-beretta -
Category
Business
-
view
640 -
download
0
description
Transcript of Assessing color reproduction tolerances in commercial print workflow
Assessing color reproduction tolerances in commercialprint workflow
G. Beretta1 E. Hoarau1 S. Kothari1 I-J. Lin2 J. Zeng1
1Hewlett-Packard Laboratories, 1501 Page Mill Rd., Palo Alto, California
2MulticoreWare Inc., 1375 S. DeAnza Blvd., Cupertino, California
Color Imaging XVII:Displaying, Processing, Hardcopy, and Applications
San Francisco Airport, 24 – 26 January 2012
Giordano Beretta (HP Labs) Color Imaging XVII EI 2012 paper 8292-32 1 / 12
The nightmare of ICC profiles
I the ICC profile standard has no provisions for workflowI oftentimes embedded profiles are incorrectI oftentimes profiles are of poor qualityI workflow system must be able to swap out profilesI want to target good enough quality for given price point (SLA)I select relevant papers:
I Lammens, “Night of the living color: horror scenarios in colormanagement land,” this conference 1998
I Phillips et al., “Comparing image quality of print-on-demand books andphotobooks from web-based vendors,” JEI 2010
I Beretta et al., “ICC profiles: are we better off without them?” thisconference 2011
I Falkenstern et al., “Adaptively selecting a printer color workflow,” NIP2011
Giordano Beretta (HP Labs) Color Imaging XVII EI 2012 paper 8292-32 2 / 12
Assessing print quality
I average ∆E values are not usefulI a useful metric must allow specific assessmentI Peter Zolliker et al. from EMPA have proposed several more useful
methodsI is there a simpler method?
Giordano Beretta (HP Labs) Color Imaging XVII EI 2012 paper 8292-32 3 / 12
Color fidelity vs. color integrity
I in practice, color fidelity is not that criticalI Phillips et al., “Comparing image quality of print-on-demand books and
photobooks from web-based vendors,” JEI 2010I print does not contain sets of unrelated colors but color palettesI goal: preserve the integrity of color palettes
Example (catalog selling denim garments)
One page might have jeans in original, washed, and stone-washed denim,while a second page might have corresponding jackets. The specific hue ofeach garment is less important than being able to match jackets to jeans whilepreserving the rank of the processes
Giordano Beretta (HP Labs) Color Imaging XVII EI 2012 paper 8292-32 4 / 12
Farnsworth-Munsell 100-hue test
I in the case of humans, color discrimination can be assessed with theFarnsworth-Munsell 100-hue test
I can we use this test for color reproduction systems?
Giordano Beretta (HP Labs) Color Imaging XVII EI 2012 paper 8292-32 5 / 12
Scoring transpositions
Giordano Beretta (HP Labs) Color Imaging XVII EI 2012 paper 8292-32 6 / 12
Graphical evaluation
3
4
5
6
7
8
9
10
11
12
13
85 1
R
2 3 4 5 6
7 8
9 10
YR
11 12
13
14
15
16
17
18
Y
19
20
21
22
23
24
25
26
27
GY
28
29
30
31
32
33 34
35 36
G
37 38
39 40 41 42 43 44 45 46
BG
47 48 49 50 51
52 53
54
B
55
56
57
58
59
60
61
PB
62
63
64
65
66
67
68
69
70
71
P
72
73
74
75
76 77
78
RP
79 80
81 82 83 84
3
4
5
6
7
8
9
10
11
12
13
85 1
R
2 3 4 5 6
7 8
9 10
YR
11 12
13
14
15
16
17
18
Y
19
20
21
22
23
24
25
26
27
GY
28
29
30
31
32
33 34
35 36
G
37 38
39 40 41 42 43 44 45 46
BG
47 48 49 50 51
52 53
54
B
55
56
57
58
59
60
61
PB
62
63
64
65
66
67
68
69
70
71
P
72
73
74
75
76 77
78
RP
79 80
81 82 83 84
Giordano Beretta (HP Labs) Color Imaging XVII EI 2012 paper 8292-32 7 / 12
Practical application
I FM-100 assesses entire hue circleI we need to assess only the critical color rangesI include one or more scales in the crop marginI normalize scores to FM-100I examples of parameters to assess:
I ICC profile plausibility & correctnessI ICC profile reputation & precisionI rendering intentI gamut mapping algorithmI black skeletonI halftoningI inks, process vs. Hexachrome; spot colorsI media
Giordano Beretta (HP Labs) Color Imaging XVII EI 2012 paper 8292-32 8 / 12
Example
A uniform color scale in CIELAB from (80, 15, 2) to (85, 20, 45)
14 16 18 20
80
82
84
86
a*
L*
14 16 18 200
20
40
a*
b*
The non-uniformities are due to rounding the coordinates to one decimal digit
Giordano Beretta (HP Labs) Color Imaging XVII EI 2012 paper 8292-32 9 / 12
Reproduction on office laser printer
14 16 18 20
76
80
84
88
a*
L*
14 16 18 20
0
20
40
a*b*
∆E94 = 5.6
Giordano Beretta (HP Labs) Color Imaging XVII EI 2012 paper 8292-32 10 / 12
Comparison of 3 devices
76 80 84
76
80
84
L*
L* d
evic
e
16 18 20 22 24
16
18
20
22
24
a*
a* d
evic
e0 20 40
0
20
40
b*
b* d
evic
e
∆E94:I green device: 5.6I red device: 1.2I blue device: 0.7
Giordano Beretta (HP Labs) Color Imaging XVII EI 2012 paper 8292-32 11 / 12
Physical vs. simulated
add test strip
RIP
measure test strip
score color scale
add test strip
RIP
extract color scale
predict colorimetry
score color scale
physical simulated
Giordano Beretta (HP Labs) Color Imaging XVII EI 2012 paper 8292-32 12 / 12