Professional Documents
Culture Documents
Agenda
1.Aim and motivation of the Forum
2.Test programme & Findings
1.Characterising
PART I 2.Proofing
3.Spectral Prediction
4.Separation of images
5.Separation of spot colour
Solutions from:
AGFA, Alwan,
Huber supplied
CGS, ColorGate,
spot samples
ColorLogic,
ESKO, GMG,
Heidelberg & manroland
Kodak printed
testforms
11 partners
Dr. Andreas Kraushaar | kraushaar@fogra.org !3
Fogra Multicolor Forum 2018, 2018-10-04
2. Test programme
Testform A Testform B 4 Spot colours
0,5
Print run ID
Testform A
Testform C
2. Testform setup
Separations:
2.1 Characterising
¬Based on a provided spectral reference (ECG V4) with 4200 patches
¬More information on ECG-V4 in the Kodak presentation (Ron Ellis)
¬Creating an ICC-profile (or another data transform)
Findings:
Avg_∆E76 Avg_∆E00 Max_∆E76 Max_∆E00 95% Quantile 95% Quantile
∆E76 ∆E00
AGFA* 0.4 0.3 5.6 6.9 2.3 1.4
AGFA 0.9 0.6 13.5 10.9 4.0 2.9
ColorLogic 0.4 0.3 4.6 4.0 1.3 1.0
KODAK 0.8 0.5 6.9 5.3 2.0 1.5
Heidelberg 2.7 1.8 16.2 18.3 7.1 5.1
Alwan 1.0 0.6 9.5 6.1 2.7 1.9
ColorGate 0.4 0.3 7.3 5.2 1.6 1.2
GMG (MXN profile) 0.0 0.0 2.6 2.3 0.0 0.0
CGS 7.3 5.1 32.8 30.3 16.5 12.9
CGS* 1.0 0.6 6.9 6.0 2.5 1.6
Dr. Andreas Kraushaar | kraushaar@fogra.org *re-submission, see problem statement during partner presentation !8
Fogra Multicolor Forum 2018, 2018-10-04
40 40 40 0 0 0 0
70 70 70 0 0 0 0
100 100 100 0 0 0
40 40 40 40 0 0 0
ISO
Evalua&on Agfa Alwan CGS ColorGATE
12647-7:2016
A-2 A-3 A-2 A-3 A-2 A-3 A-2 A-3
ECG-V4
4200 samples
Reference
ECG-V4
4200 samples
Test
45° 15° 75° 0° 15° 75° 45°
ECG-V4
4200 samples
Reference ECG-V4
4200 samples
Prediction
+
test-print
?
4200 CMYKOGV tone values
ECG-V4
4200 samples
Test
ID 95% Quantile
95%
Avg_∆E76 Avg_∆E00 Max_∆E76 Max_∆E00 Quantile
∆E76
∆E00
¬ So you can not get better (than lying between all patches)
For instance:
0 0 0 0 100 0 0
L* a* b* ∆E00
62.9 60.4 80.3 0,3
63.7 58.6 79.2 0,6
63.1 60.6 79.6 0,4
mean 63.2 59.9 79.7 MCDM = 0.4
Dr. Andreas Kraushaar| kraushaar@fogra.org !14
Fogra Multicolor Forum 2018, 2018-10-04
C M Y K O G Blue
….
Dr. Andreas Kraushaar| kraushaar@fogra.org !17
*re-submission, see problem statement during partner presentation
Fogra Multicolor Forum 2018, 2018-10-04
C M Y K O G Blue
….
+
CxF/X-4b
different
ues that 2.5 separation of spot colours: graininess
E00_std, After analyzing the results, it was determined that dE00
an, and std was the best of the six metrics to measure image qua-
Fogra method
an and lity. The following six categories based on dE00 std were
Image Quality Based on Graininess
ted and thus created for reading distance applications (40cm.) Fogra
ated by AIM
The aim of this experiment was to evaluate image quality Mean Rank vs avg dEOO_std
0,90
based on measurements of graininess and determine the 0,80
t ~·f
0 S AS 0,25 ................. Completely smooth with no visible grain best metric on which to judge samples.
0, 70
'B 0,60
R1 .:.Q,8808
~ I 0,5 0
0 ,40
·•
f
0,25 < B S 0,5 ................... Mostly smooth with a very slight grain PROCESS
First, six media wedges of various noise quality were
0,30
0,20
0,10 '
.
0,00
scanned at 300 dpi, converted to 8-bit LAB files, and 000 1 00 200 300 4,00 ,00 600 700
0,5 < C S 0,75 .................................................. Noticable grain evaluated for graininess in MATLAB at three different
Mean Rank
F > 1,25 ...... Textured appearance with highly pronounced grain used for analysis. An example of a report generated by
MATLAB is displayed below: 0 S AS 0,25 ................. Completely smooth with no visible grain
__
0,25 < B S 0,5 ................... Mostly smooth with a very slight grain
...,,sradR11•t..»1u--.s ,~ira.•• 021021 0,5 < C S 0,75 .................................................. Noticable grain
__
a'flfflllll'ld~Til•O .WIM 1"9fflll1Jtl!l01>• 04J.161
.,.... - ---r- - _.,.... -,------,
•
1
"
.,
i!. t tt
, , P 6oc,, ."lo1,•
t,, °·J~ai,o~ 11 'I t•~r/oo /t l'a 0 \ L II te:-
~ä~ a.:;;~I°ytr~ ,~a:aC09~ao"6Cq-0cr~~a11 .a•~a·ooO 0': g.001t~~ "0
0
'~g:: The second part of the experiment consisted of measu-
0
7
' 11 •o '.lO .i
ge qua- JK2
JK3
6,00
5,00
6,00
5,00
6,00
5,00
6,00
4,00
6,00
5,00
6,00
4,80 -
-'-
. -----~-----~
....--
,,
' "
JK4 4,00 4,00 4,00 5,00 4,00 4,20
elow: JK5 3,00 3,00 3,00 3,00 3,00 3,00
,--~==---r--::;;;;;::=:-
..._"c'_•_"=· The data from these two experiments were then ana-
Based on the results of the experiments, dE00 std was de-
termined to be the best metric on which to judge image
5 Avg
1,20
'·i:t====c==~
u - -- ~-~--~ T~
.,........,,
--~----'-
•• - _j
,. •
lyzed in Excel using a basic linear regression model in
an attempt to determine the best metric for measuring
image quality. An example of the resulting graph for
Mean Rank vs dE00 std is shown:
quality, and six categories were created. The samples used
in the experiments are presented in the following pages
of this report. Further research must be conducted to va-
lidate these results.
.............,.u,·~-...«• • l ,11..Q
6,00
4,80 -
-'-
. -----~-----~
.... --
,,
' "
4,20
Dr. Andreas Kraushaar | kraushaar@fogra.org !25
3,00
Fogra Multicolor Forum 2018, 2018-10-04
∆E00STD ∆E00STD
ID ∆E00STD ∆E00STD
3. Partner presentations
at the Fogra table you can see (under D50):
Offset prints:
Reference prints Testform A: KCMYVOG
Reference prints Testform A: KCMYOGBlue
Reference prints Testform C: KCMYVOG
Proofs:
Submitted partner proofs of Testform A: KCMYVOG
Submitted partner proofs of Testform B: KCMYOGBlue
Submitted partner proofs of Testform C: KCMYVOG
Submitted partner proofs of Testform C: KCMYOGBlue
Spots
Spot colour samples (Testform C)
with reference in the centre