Re: creating target for litho prints
Re: creating target for litho prints
- Subject: Re: creating target for litho prints
- From: Klaus Karcher <email@hidden>
- Date: Sun, 17 Jan 2010 21:17:57 +0100
Hi Stefan, Hi Edmund,
edmund ronald wrote:
Klaus is our resident list expert on this topic :)
You might be right as Stefan apparently is talking about a Cruse scanner.
I'm engaged in improving the color fidelity of Cruse scanners for more
than two years now and came up with a bundle of measures.
The overall results are amazing:
- average Delta E 2000 reduced by 70% to 80%
- max. Delta E 2000 reduced by considerably more than 50%
compared to conventionally profiled and configured scanners.
This statements refer to freshly profiled systems. Typically the
improvements are even higher in practice as one of my measures consist
in balancing both short-term and long-term variations (e.g. due to lamp
aging and system drift). While a conventionally profiled scanner is
getting worse from scan to scan after profiling, my workflow caters for
sustained accuracy, repeatability and consistent results over different
lighting modes.
The problem Stephan describes is well known to me and largely solved by
the measures mentioned. In well-defined areas of application we can get
average DE00's significantly < 1.
There are two early adopters who use the workflow I developed with great
success. Both of them affirm that they don't want to do without it anymore.
At the moment I'm working with Cruse on the integration of my measures
into the scanner software and the collocation of bundles adjusted to the
customers needs (software updates, hardware modifications and profiling
services).
Stefan (and everyone else interested) is invited to contact me off-list
to discuss further details.
On Sun, Jan 17, 2010 at 11:06 AM, Stefan Ohlsson <email@hidden> wrote:
I'm trying to create a scanner profile for use in a large format scanner.
They will use it to scan old maps and big litho prints. I have used a Hutch
chart to make this profile. When I check this profile the result is OK. But
when we scan the maps they get a magenta cast.
I came across this issue many times (not only with Cruse Scanners).
There are two main reasons:
- the system doesn't fulfill the Luther condition (i.e the effective
spectral sensitivities of the scanner are no linear combinations of
those for the Standard Observer)
- the spectral properties of your profiling target doesn't represent the
spectral properties of your originals good enough. Your profile would
give good results if the pigments of your originals would match the
pigments of your profiling target. As this is typically not the case,
you get systematic errors.
I counteract this problems i.a. with improved
lighting/filter-combinations and better training sets. My typical
training set consists of more than 1200 patches. Many of them are
pinpointed to the metamerism issues of the sensor. The training set is
made up of many different pigments. Typically I use 10 different targets
to create a scanner profile. Only 3 of them are commercially available).
Furthermore I use methods to weight and balance the training set in
order to adjust it to the properties of typical originals.
Needles to mention that such operations are barely realizable with
conventional, off-the-shelf profiling tools. I use ArgyllCMS and a
statistical computing environment to create the profiles. Argyll
provides excellent profiling algorithms and the flexibility I need.
I've checked the scanner light's spectral distribution and it is quite
spiky. The scanner maker says that they use these fluorescent tubes because
of the high light output and that we have to slow down the scanner if we
would change the tubes to those with a more even spectral distribution. As
my customer will scan thousands of maps, longer scanner times aren't
acceptable.
Cruse is right: unfortunately there is a negative correlation between
spectral smoothness and luminous flux (or energy-efficiency). We have
tested countless filter/tube-combinations and there is always a tradeoff
between scan time, noise and color fidelity -- but I'm pretty sure that
we can provide you (or your customer) with the configuration that best
fits your needs.
I'm thinking of creating a scanner target with colour patches from some
discarded prints. Any ideas how to do this? How many patches do I have to
use to create an acceptable scanner profile?
see above.
Best
Stefan Ohlsson
Klaus
_______________________________________________
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