Re: VIGC study on spectrophotometers
Re: VIGC study on spectrophotometers
- Subject: Re: VIGC study on spectrophotometers
- From: "dpascale" <email@hidden>
- Date: Wed, 17 Sep 2008 10:40:03 -0400
Hello Paul.
Scary?
No, just a reminder that we work with real-world tools in a real-world
environment.
Just take the i1 maximum inter-instrument spec of 1 DeltaE(CIE94); this is
equivalent, on average, to about 2 DeltaE*ab(CIELab) for small color
differences.
So, just between two i1, for a single measurement, a Delta-E*ab difference
of 2 is possible (and within specs!). Add to this a variation due to
calibration drift and the relative absolute accuracy of the i1 vs the
accuracy of another brand/model of instrument (Note: Absolute accuracy is
never given for the instruments used in the graphics field), plus the
different type of illumination in the instruments, and you easily get 3.77
difference (VIGC data) for a single patch. On average, as VIGC measured, the
difference is 1.56 when using only 13 patches, which is quite good for such
a small number of patches.
I recently compared the measurements on targets with 50and 176 patches;
these targets are used by a RIP for linearization and quality control
purposes.
I compared the data for the patches from the RIP, as made through the
printer's spectro (an i1 built-in a HPZ3100), with (manual) measurements
done with an i1-Pro.
I also compared the measurements, of another set of patches, made with an
ICColor and the i1-Pro.
i1-Z3100 vs i1-Pro (50 patches)
CIELAB: Avg=2.1 (Std-Dev=0.90); 95th perc.=3.3; Max=4.2
CIE94: Avg=1.4 (Std-Dev=0.54); 95th perc.=2.1; Max=2.4
CIEDE2000: Avg=1.2 (Std-Dev=0.53); 95th perc.=2.1; Max=2.4
i1-Z3100 vs i1-Pro (176 patches)
CIELAB: Avg=1.8 (Std-Dev=0.64); 95th perc.=2.8; Max=3.6
CIE94: Avg=1.3 (Std-Dev=0.45); 95th perc.=2.0; Max=2.8
CIEDE2000: Avg=1.2 (Std-Dev=0.45); ; 95th perc.=1.9; Max=2.9
i1-ICColor vs i1-Pro (176 patches)
CIELAB: Avg=1.4; 95th perc.=3.3; Max=4.0
CIE94: Avg=0.9; 95th perc.=1.8; Max=2.7
CIEDE2000: Avg=0.8; 95th perc.=1.4; Max=1.8
Note 1: The i1-Pro measurements were all made on an ISO compliant black
backing.
Note 2: The Z3100 uses a black backing.
Note 3: The i1 uses a tungsten lamp, the Z3100 spectro uses LEDs.
Note 4: The ICColor uses a white backing. Accordingly, it was measured that
the i1-Pro measurements, made on a black backing, have an average Delta-L*
of minus 0.75 (-0.75) compared with the ICColor measurements. This is
included in the Delta-E* results above. A similar comparison betwen the
ICColor and the i1-Pro with the white-backed i1 measurements was not done.
The RIP built-in optimization routine was used to optimize the profile; this
routine recursively prints a chart, measures it (through the printer's
spectro), and modifies the profile, with an average of less than 1 (CIELAB)
and a max difference below 4 on an IT8.7/4 target after a few iterations.
The optimization has nothing to do with the comparison between spectros, per
se, since a calibrated profile is not required for such a comparison (as
long as we measure the same patches). However, this is a common procedure
for RIP users and we often see such optimization numbers given as proof of a
printer's accuracy.
Of course, measuring an IT8.7/4 target with my i1-Pro gave me an average
error larger than what I was given by the RIP optimization procedure (the
i1-Pro avg error is 2.3 CIELAB compared to less than 1 for the RIP; Note:
the i1-Pro avg is 1.5 for CIEDE2000), and that is a good example of why we
must give some allowance for inter-instrument and inter-model accuracy
differences when making a calibration with one instrument and checking it
with another. Again, there is nothing new in my last statement, except, that
we (me included) often forget to take it into account when looking at a
bunch of measurements given to us by a customer or colleague, especialy when
striving for fractions of Delta-E accuracies.
Where is does bother me is that the use of a limited number of patches for
quality control is very common (i.e. wedges and other targets), and when
measurements are done on the primaries, any error in these will throw a
large range of colors out of tolerance. As a minimum, inter-instrument
agreements should be optimized for these few colors.
Danny Pascale
email@hidden
www.babelcolor.com
----- Original Message -----
From: "Paul Sherfield" <email@hidden>
To: <email@hidden>
Sent: Wednesday, September 17, 2008 4:25 AM
Subject: Re: VIGC study on spectrophotometers
Hi
Follow this link to an interesting piece on spectros and accuracy, all a
little scary, from VIGC the Flemish Innovation Centre for Graphic Arts.
http://www.graphicbrain.com/spectrophotometer-nightmare/
Regards
Paul Sherfield
The Missing Horse Consultancy Ltd
Telephone: 01442 871752
Mobile: 07899 906385
Save Paper - Do you really need to print this e-mail?
http://www.missinghorsecons.co.uk
Apple Solutions Expert-Print & Publishing
Member - UK TC130 Technical Advisory Group (ISO 12647 Printing Standards)
Member - BPIF Technical Standards Committee
This email and any attachments may be confidential and are intended solely
for the use of addressee. If you are not the intended recipient of this
message, any disclosure, copying or other distribution is strictly
prohibited and may be unlawful.
If you have received this message in error, please notify the sender and
delete all copies from your system. Email may be susceptible to data
corruption, interception and unauthorised amendment.
We do not accept liability for any such corruption, interception or
amendment or the consequences thereof. We do not accept liability for any
action or inaction by the recipient as a result of advice or information
contained within this email.
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Colorsync-users mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Colorsync-users mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden