Home' FLEXO Magazine : January 2016 Contents YOU CAN’T JUDGE A CHART BY ITS COLORS
Both of these test charts contain randomized patches. The first dis-
cernable difference is that the 7-color ICC profile chart seems to have
randomized patches throughout color space, while the 7-color 4x4
profile chart seems to have the randomized patches clustered together
by color sector. If this were the only difference, one might favor the
7-color ICC profile chart under the premise, “The more random; the
better.” However, an equally valid view is that with almost all pro-
duction jobs, colors tend to be clustered together. Because test charts
are supposed to simulate production, having randomized patches
clustered by color sector may be better.
In fact, there are far greater differences between these two charts
than clustering of randomized patches. However, it’s a fact of the
human brain that randomization tends to hide true systematic
differences. To “see” the differences, we need to put these patches
in some type of order. When the patches are ordered, it is clear that
the 7-color ICC profile target contains 1,976 “relevant table points”
while the 7-color 4x4 profile target contains 5,512 “relevant table
points”—almost three times as many. Given software capable of
working with either profile format, the patch set with the most rele-
vant table points is the one with the potential for the most accurate
color. As such, the 7-color 4x4 target is far superior to the 7-color
ICC profile target. Understanding why a color profile chart with
more relevant table points is superior to a chart with less relevant
table points starts with understanding the concept of table points
and the concept of relevant.
THE CONCEPT OF TABLE POINTS
When adjusting tone using dot gain adjustment curves, the concept
of “table points” does not apply. Each color is adjusted independently
of every other color. If the test target had five, single color sample
points of cyan and 25 single color sample points of magenta, a logical
dot gain adjustment strategy would use all 30 points (the sum of both
sample point totals).
When adjusting color using profiles, the concept of table points enters
the discussion. Perhaps the easiest way to understand the concept of a
“table” is to consider a 2-color printing process; say cyan and magenta
(we’ll consider 4-color through 7-color later). A 2-color process forms a
two dimensional (2-D) table. A 2-D table is convenient to show graphi-
cally because computer screens and magazine pages are themselves 2-D.
Three-color tables can also be illustrated graphically, but, being three
dimensional (3-D), are harder to draw (4-color and beyond tables are
impossible to show graphically, but are equally valid mathematically).
A 2-color table contains combinations of the two colors. The cyan x
magenta box (Figure 3) is a five step table. It can be used to interpolate
the L*a*b* or spectral values for any combination of cyan and ma-
genta. The more steps in the table, the greater the potential accuracy.
The magenta box in Figure 4 is not a table—even though it is cleverly
arranged to look like a table. It cannot be used to interpolate L*a*b*
or spectral values for
combinations of the
two colors used in the
printing process, as it
contains information on
If Figure 3 and Figure
4 were included in
a patch set, it would
include 50 patches, but
only 25 table points.
Such a patch set could
be said to be 50 percent
efficient (25 ÷ 50 = 50
percent). There are
more efficient ways to
use 50 measurement points. The goal is to have as many table points
as possible, relative to the total number of patches. For this 2-color
printing process example, the 7x7 patch set in Figure 5 is 100 percent
efficient. Each patch in the target is used in a table. Given the space
to print 50 measurement points, this 49 point patch set would enable
more accurate color than the previous 50 point patch set.
THE CONCEPT OF RELEVANT
The concept of “relevant” says that only color combinations to be used
in a production job should be included in a patch set or test chart. In
4-color (CMYK) applications, there was never much thought paid to
the concept or relevance. Having adequate sampling was thought to
be easy to achieve with four colors (a separate topic is that we now
know far better patch sets can be made for CMYK applications than
today’s standard IT8.7 -4 patch set—see the sidebar on page 48).
But even if we concede that today’s CMYK targets have adequate
sampling, to have the same sampling “power” with every combination
of seven colors instead of four would require a patch set that is exactly
1,000 times the size of the 4-color patch set (107 ÷ 104
= 103 = 1,000).
There is no press large enough to print such a target, nor would any
production printer agree to extend measurement time from one hour
to 1,000 hours (the equivalent of 25 full, 40 hour work weeks) to
Figure 3 & Figure 4: The cyan x magenta 5x5 box (left) is a two dimensional
table with five table points; the magenta box (right) is not a table.
Figure 5: This 7x7 step table would produce
more accurate color results than the five step
JANUARY 2016 | FLEXO 43
Links Archive December 2015 February 2016 Navigation Previous Page Next Page