http://www.luminous-landscape.com/essays/fancy-graphics2.shtml
Of course not everyone is happy about the result, wich can and will
be
disputet. Still its the first shootout of this caracter I am aware
of,
its done with a great deal of effort, and the PFP comes through
with
flying colors :=)
Just something to hold on to when the bashing gets too rough out
there.
:-)
I agree. The bottom line was, the PFP print of his test image did
as well
as the most expensive competition; and the review didn't even
mention the new
B&W profiling feature, which is unique to PFP.
This was a welcome followup to the original article, which was
negative
on PFP because the profiles lacked perfect smoothness on the
outside of the
gamut shape when the profile was viewed through Gamutvision. We
knew that
profiles built with other, much more expensive packages would
show the same
kinds of issues to one degree or another, and that if ALL
packages were
compared this way, and if actual prints were compared as well, it
would become
apparent that looking at the OUTER surface of the printer gamut
shape wasn't
the right way to judge the profiles.
Looking at his results more closely, there are some other
interesting things
to note about the gamut shapes as well.
It's unfortunate that the gamut plots shown in the combined
"Adobe RGB(1998) to
six profiles" aren't -larger-, because to be quite frank,
it's not possible
to look at ANY of these very closely to see how much rippling or
bumpiness
there is on the outer surface of the shape.
Looking at the fully saturated colors displayed next to each
gamut shape
(these are plots in HSL space, where the hue, H, goes from 0 to
360, the
full range, from left to right; saturation, S, is max at 100% for
all colors;
and lightness starts at 0% and increases to 100% from bottom to
top) - you
can see interesting effects in every one of them; there are
"ripples" or
flattened areas or even unexpected shifts in color in various
places for all
of the profiles.
In the HBc profile, there's a large lighter blot inside the blue
shape;
there's a very non-smooth transition from blue to magenta into
red, going
to the right; there are discontinuities in the green, and other
"ripples"
in the transitions across hues; these are obvious even at this
small scale
and would be more apparent if the reader was given a larger
version of this
to look at.
The XR4 profile is interesting in that it flattens out/blocks up
all of the
saturated greens as lightness increases (going up vertically on
the plot),
rather than giving the smoother taper into a point (triangular
effect) that
most of the other profiles provide. In that sense, the HP profile
looks the
most like the XR4 profile. I also see ripples in the transition
from blue
into magenta. Fully saturated cyans in this profile are almost
nonexistent,
you can see this from the narrowness of the band.
The PFP gamut shape is the LARGEST of all six, but there's no
mention of
this. In particular, look how it extends down into the darkest
blue.
The ski slope plots below that all have dips and bumps. Not very
useful
for drawing conclusions.
Etc etc, I could spend pages going over these comparisons, but
the bottom
line is: I don't see any correlation between Gamutvision plots
and the
prints that got the most points in the final comparison.
PFP came in third; and it came in -ahead- of the JK3 profile,
which was built
by Jan Klabacka using EyeOne Match3, which is part of the much
more expensive
EyeOne Photo package; and measurements were taken with the
much more expensive EyeOne spectrophotometer. The JK3 profile, in
an attempt
to provide the best possible results, was built using a larger
number of
patches. Let's make this clear: according to the review, the JK3
profile
was built using measurements from an EyeOne; using not only 918
patches
(nine hundred and eighteen patches); but an ADDITIONAL 459 low
saturation
patches (four hundred fifty nine); for a total of 918 + 459 =
1377 patches.
One thousand, three hundred seventy seven patches.
1377 patches. EyeOne spectrophotometer. EyeOne Match3 software
from Gretag.
30 out of 50 points based on the fern print, total score:
41.
729 patches, PrintFIX PRO, Datacolor 1005
spectrocolorimeter.
40 out of 50 points based on the print, total score: 45
729 patches, XRite Pulse and software, (Pulse is a discontinued
product now)
45 out of 50 points based on the print, total score: 68
PFP is by FAR the least expensive of these solutions, and prints
from all
three of these systems are comparable. Based on the prints, PFP
-beats-
the EyeOne software that is using almost TWICE as many patches,
and which
is taking measurements with a much more expensive
spectrophotometer.
(So anyone who wants to bash the PFP spectrocolorimeter vs. the
EyeOne
and XRite spectrophotometers because it's a
"spectrocolorimeter" should
pay close attention to this).
***
The non-print points are not meaningful to look at in the context
of
this evaluation, and reading his final chart tells us why.
Add up the first 5 rows and PFP is given only 5 points, which
would appear
to put it dead-last against the HP profile (11 points), the HBc
profile
(6 points), the XR4 profile (23 points!), JK-ACMS (19!) and JK3
(11).
Based on this sort of technical analysis, PFP would make by far
the worst
prints of the group; that is, if this analysis was meaningful and
could be
used as a basis to predict print quality.
But instead, what do we see, in the PRINT results?
The HP profile ("twice" as good as the PFP on the
GamutVision analysis) comes
in dead-last, with 0 points for the print.
The JK3 profile (also "twice" as good as PFP on the
GamutVision) comes in
2nd-to-last, with 30 points for the print.
The PFP profile, which GamutVision implies should be the worst,
gets 40 print
points; 5 subjective points behind the XR4 print, (which came in
FIRST PLACE
by a huge margin in the GamutVision areas).
That tells the story. The XR4 profile, which rates far higher in
all of the
GamutVision categories; not only doesn't come in first in the
print
test; it barely (if at all) beats the PFP profile, which came in
last in the
GamutVision categories.
So much for the value of GamutVision analysis of these profiles.
These
results show that kind of analysis is NOT meaningful when it
comes to
real world prints.
Best regards,
--
David Miller
Senior Software Developer, Digital Color Solutions
ColorVision