[argyllcms] Re: New to CMS - looking for advice on device to use with argyllcms
- From: Klaus Karcher <lists@xxxxxxxxxxxxxxxxx>
- To: argyllcms@xxxxxxxxxxxxx
- Date: Mon, 04 Jan 2010 20:57:44 +0100
Ben Goren wrote:
For photography, how often would you personally have an i1 Pro calibrated?
hmm ... it's hard to say ... I think it depends on many factors (dust,
storage conditions, frequency and magnitude of rapid temperature and
humidity changes, mechanical shocks, total hours of lamp operation,
quality of initial calibration, transport conditions, ...)
I've made some round robin tests and found for example an seriously
inaccurate i1 pro that was less then three moths old. The owner asserted
that it was always handled with care. XRite replaced it free of charge.
Some other eye-ones I've been testing were still in the right ballpark
after four or five years of frequent usage without factory recalibration.
I think it's a good idea to compare instruments among each others every
few moths if possible. If there are no other instruments to compare
with, it would be helpful to have some steady, durable reference
samples. Unfortunately (calibrated) BRCA tiles are very expensive.
Therefore I'm thinking about using suitable ceramic tiles, rugged paint
samples, plastic chips or laminated prints for that purpose. I think
suitable samples can be used both as long-term reference and as tools
for accuracy estimations. When one measures them frequently with
different instruments, they can be used to detect "black sheep" and to
get a better idea of the accuracy achievable in practice. Until now I
used proofs, spot color fan decks or scanner targets as reference
samples, but they are not durable enough for long-term tests.
One simple test that doesn't require a second instrument can be
performed with the white reference tile of the instrument: calibrate the
instrument and then measure the white reference several hundred times
(for some minutes) without interruption. If the measurement results
drift off significantly, it's an indication that the lamp has to be
replaced.
Danny Rich presented an IMHO rather alarming paper on the effect of
spectrocolorimeter reproducibility on print production workflows at the
last CGIV:
<http://www.imaging.org/ScriptContent/store/epub.cfm?abstrid=38731>:
"Reports by Wyble and Rich [...] have indicated that many, portable,
hand-held spectrocolorimeters have extremely good precision"
(Please note: /precision/ does not mean /accuracy/!)
"but different instruments produced consistently
different results on standard materials [...] They
reported differences as large at 6 CIELAB units with averages
between 1 and 3 CIELAB units."
I think that instrument variations in this order of magnitude are still
acceptable in many cases when absolute accuracy doesn't play a decisive
role (exceptions are e.g. tasks like fine art reproduction), but they
are a serious problem e.g. in matters of proof system calibration and
verification, (cross-company) quality assurance, corporate design
compliance and so on.
As factory re-certifications are expensive and lengthy I've been
thinking about collaborative round-robin tests to verify the integrity
and accuracy of spectrometers. The principle is very simple: Instead of
sending out the instruments to the factory (and waiting donkey's years
until they return ;-), robust probes circulate among the participants
and the measurement results get published and compared among themselves.
Every participant can check his instrument against the others
anonymously this way.
In the scope of scientific work and some industry segments (e.g. paint-
and paper industry) such tests are already common practice and there are
established service providers that co-ordinate and evaluate the tests
(see e.g. <http://www.collaborativetesting.com/>).
But I a not aware of any comparable service for the hand-held
instruments common in pre-press and print shops (eye-one, SpectroEye,
...). Those users seem to have rather little interest in instrument
accuracy and verification (at least according to the feedback I got so
far). Apparently time is not yet ripe for it and most users seem still
to be completely unaware of the problem.
Klaus
Other related posts:
