<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
<title>Argyll Usage Scenarios</title>
<meta http-equiv="content-type" content="text/html;
charset=ISO-8859-1">
</head>
<body>
<h2><u>Typical usage Scenarios and Examples</u></h2>
Choose a task from the list below. For more details on alternative
options, follow the links to the individual tools being used.<br>
<br>
Note that by default it is assumed that ICC profile have the file
extension <span style="font-weight: bold;">.icm</span>, but that on
Apple OS X and Unix/Linux platforms, the <span style="font-weight:
bold;">.icc</span> extension is expected and should be used.<br>
<h4><a href="#PM1">Profiling Displays</a></h4>
<h4> <a href="#PM1a">Checking you can access your
display<br>
</a></h4>
<h4> <a href="#PM1b">Adjusting and Calibrating a
displays</a></h4>
<h4> <a href="#PM1c">Adjusting, calibrating and
profiling in one step<br>
</a><span style="font-weight: bold;"></span><span
style="font-weight: bold;"></span><span style="text-decoration:
underline;"></span></h4>
<h4> <a href="#PM2">Creating display test values</a></h4>
<h4> <a href="#PM3">Taking readings from a
display</a></h4>
<h4> <a href="#PM4">Creating a display profile</a></h4>
<h4> <span style="text-decoration: underline;"></span><a
href="#PM5">Installing a display profile</a></h4>
<h4> <span style="text-decoration: underline;"></span><a
href="#PM6">Expert tips when measuring displays</a></h4>
<h4> <span style="text-decoration: underline;"></span><a
href="#PM7">Calibrating and profiling a display that doesn't
have VideoLUT access.</a></h4>
<h4><br>
<a href="#PS1">Profiling Scanners and other input devices such as
cameras<br>
</a></h4>
<h4> <a href="#PS2">Types of test charts</a></h4>
<h4> <a href="#PS3">Taking readings from a
scanner</a></h4>
<h4> <a href="#PS4">Creating a scanner profile</a></h4>
<h4><br>
<a href="#PP1">Profiling Printers</a></h4>
<h4> <a href="#PP2">Creating a print profile
test chart</a></h4>
<h4> <a href="Scenarios.html#PP2b">Printing a
print profile test chart</a></h4>
<h4> <a href="#PP3">Reading a print test chart
using an instrument</a></h4>
<h4> <a href="#PP4">Reading a print test chart
using a scanner</a></h4>
<h4> </h4>
<h4> <a href="#PP5">Creating a printer profile<br>
</a></h4>
<h4> <a href="#PP6">Choosing a black generation
curve</a></h4>
<br>
<h4><a href="Scenarios.html#PC1">Calibrating Printers</a></h4>
<h4> <a href="Scenarios.html#PC2">Calibrated
print workflows</a></h4>
<h4> <a href="Scenarios.html#PC3">Creating a
print calibration test chart</a></h4>
<h4> </h4>
<h4> <a href="Scenarios.html#PC4">Creating a
printer calibration<br>
</a></h4>
<h4> <a href="Scenarios.html#PC5">Using a printer
calibration</a></h4>
<h4> <a href="#PC6">How profile ink limits are
handled when calibration is being used<br>
</a></h4>
<h4><br>
<a href="#LP1">Linking Profiles</a></h4>
<h4><br>
<a href="#TR1">Transforming colorspaces of raster files</a></h4>
<br>
<hr style="width: 100%; height: 2px;"><br>
<h3><a name="PM1"></a>Profiling Displays</h3>
Argyll supports adjusting, calibrating and profiling of displays
using one of a number of instruments - see <a
href="instruments.html">instruments</a> for a current list.
Adjustment and calibration are prior steps to profiling, in which
the display is adjusted using it's screen controls, and then
per channel lookup tables are created to make it meet a well behaved
response of the desired type. The process following that of
creating a display profile is then similar to that of all other
output devices :- first a set of device colorspace test values needs
to be created to exercise the display, then these values need to be
displayed, while taking measurements of the resulting colors using
the instrument. Finally, the device value/measured color values need
to be converted into an ICC profile.<br>
<br>
<h3><a name="PM1a"></a>Checking you can access your display<br>
</h3>
You might first want to check that you are accessing and can
calibrate your display. You can do this using the <a
href="dispwin.html">dispwin</a><span style="font-weight: bold;"></span>
tool<span style="font-weight: bold;">.</span> If you just run <span
style="font-weight: bold;">dispwin</span> it will create a test
window and run through a series of test colors before checking that
the VideoLUT can be accessed by the display. If you invoke the usage
for <span style="font-weight: bold;">dispwin</span> (by giving it
an unrecognized option, e.g. <span style="font-weight: bold;">-?</span>)
then it will show a list of available displays next to the <span
style="font-weight: bold;"><span style="font-weight: bold;">-d</span></span>
flag. Make sure that you are accessing the display you intend to
calibrate and profile, and that the VideoLUT is effective (the <span
style="font-weight: bold;">-r</span> flag can be used to just run
the VideoLUT test). You can also try clearing the VideoLUTs using
the <span style="font-weight: bold;">-c</span> flag, and loading a
deliberately strange looking calibration <span style="font-weight:
bold;">strange.cal</span> that is provided in the Argyll <span
style="font-weight: bold;">ref</span> directory.<br>
<br>
Note that calibrating and/or profiling <span style="font-weight:
bold;">remote</span> displays is possible using X11 or a web
browser (see <span style="font-weight: bold;">-d</span> option of
dispcal and dispread), or by using some external program to send
test colors to a display (see <span style="font-weight: bold;">-C</span>
and <span style="font-weight: bold;">-M</span> options of dispcal
and dispread), but you may want to refer to <a href="#PM7">Calibrating
and profiling a display that doesn't have VideoLUT access</a>.<br>
<br>
<h3><a name="PM1b"></a>Adjusting and Calibrating Displays</h3>
Please read <a href="calvschar.html">What's the difference between
Calibration and Characterization ?</a> if you are unclear as to
the difference .<br>
<br>
The first step is to decide what the target should be for adjustment
and calibration. This boils down to three things: The desired
brightness, the desired white point, and the desired response curve.
The native brightness and white points of a display may be different
to the desired characteristics for some purposes. For instance, for
graphic arts use, it might be desirable to run with a warmer white
point of about 5000 degrees Kelvin, rather than the default display
white point of 6500 to 9000 Kelvin. Some LCD displays are too bright
to compare to printed material under available lighting, so it might
be desirable to reduce the maximum brightness.<br>
<br>
You can run <a href="dispcal.html#r">dispcal -r</a> to check on how
your display is currently set up. (you may have to run this as <span
style="text-decoration: underline; color: rgb(204, 51, 204);">dispcal
-yl
-r</span> for an LCD display, or <span style="text-decoration:
underline; color: rgb(204, 51, 204);">dispcal -yc -r</span> for a
CRT display with most of the colorimeter instruments. If so, this
will apply to all of the following examples.)<br>
<br>
Once this is done, <a href="dispcal.html">dispcal</a> can be run to
guide you through the display adjustments, and then calibrate it. By
default, the brightness and white point will be kept the same as the
devices natural brightness and white point. The default response
curve is a gamma of 2.4, except for Apple OS X systems prior to 10.6
where a gamma of 1.8 is the default. 2.4 is close to that of
many monitors, and close to that of the sRGB colorspace. <br>
<br>
A typical calibration that leaves the brightness and white point
alone, might be:<br>
<br>
<a href="dispcal.html">dispcal</a> -v TargetA<br>
<br>
which will result in a "TargetA.cal" calibration file, that can then
be used during the profiling stage.<br>
<br>
If the absolutely native response of the display is desired during
profiling, then calibration should be skipped, and the linear.cal
file from the "ref" directory used instead as the argument to the -k
flag of <span style="font-weight: bold;">dispread</span>.<br>
<br>
<b>Dispcal</b> will display a test window in the middle of the
screen, and issue a series of instructions about placing the
instrument on the display. You may need to make sure that the
display cursor is not in the test window, and it may also be
necessary to disable any screensaver and powersavers before starting
the process, although both <span style="font-weight: bold;">dispcal</span>
and <span style="font-weight: bold;">dispread</span> will attempt
to do this for you. It's also highly desirable on CRT's, to clear
your screen of any white or bright background images or windows
(running your shell window with white text on a black background
helps a lot here.), or at least keep any bright areas away from the
test window, and be careful not to change anything on the display
while the readings are taken. Lots of bright images or windows can
affect the ability to measure the black point accurately, and
changing images on the display can cause inconsistency in the
readings, and leading to poor results.<span
style="font-weight: bold;"></span> LCD displays seem to be less
influenced by what else is on the screen.<br>
<br>
If <span style="font-weight: bold;">dispcal</span> is run without
arguments, it will provide a usage screen. The <span
style="font-weight: bold;">-c</span> parameter allows selecting a
communication port for an instrument, or selecting the instrument
you want to use, and the <a href="dispcal.html#d"><span
style="font-weight: bold;">-d</span></a> option allows selecting
a target display on a multi-display system. On some multi-monitor
systems, it may not be possible to independently calibrate and
profile each display if they appear as one single screen to the
operating system, or if it is not possible to set separate video
lookup tables for each display. You can change the position and size
of the test window using the <a href="dispcal.html#P"><span
style="font-weight: bold;">-P</span></a> parameter. You can
determine how best to arrange the test window, as well as whether
each display has separate video lookup capability, by experimenting
with the <a href="dispwin.html">dispwin</a> tool. <br>
<br>
For a more detailed discussion on interactively adjusting the
display controls using <span style="font-weight: bold;">dispcal</span>,
see <a href="dispcal.html#Adjustment">dispcal-adjustment</a>. Once
you have adjusted and calibrated your display, you can move on to
the next step.<br>
<br>
When you have calibrated and profiled your display, you can keep it
calibrated using the <a href="dispcal.html#u">dispcal -u</a>
option.<br>
<br>
<h4><a name="PM1c"></a>Adjusting, calibrating and profiling in one
step.</h4>
If a simple matrix/shaper display profile is all that is desired, <span
style="font-weight: bold;">dispcal</span> can be used to do this,
permitting display adjustment, calibration and profiling all in one
operation. This is done by using the <span style="font-weight:
bold;"><span style="font-weight: bold;">dispcal </span>-o</span>
flag:<br>
<br>
<a href="dispcal.html">dispcal</a> <a href="dispcal.html#v">-v</a>
<a href="dispcal.html#o">-o</a> <a href="dispcal.html#p1">TargetA</a><br>
<br>
This will create both a TargetA.cal file, but also a TargetA.icm
file. See <a href="dispcal.html#o">-o</a> and <a
href="dispcal.html#O">-O</a> for other variations.<br>
<br>
For more flexibility in creating a display profile, the separate
steps of creating characterization test values using <span
style="font-weight: bold;">targen</span>, reading them from the
display using <span style="font-weight: bold;">dispread</span>, and
then creating a profile using <span style="font-weight: bold;">colprof</span>
are used. The following steps illustrate this:<br>
<h4><a name="PM2"></a>Profiling in several steps: Creating display
test values</h4>
If the <span style="font-weight: bold;">dispcal</span> has not been
used to create a display profile at the same time as adjustment and
calibration, then it can be used to create a suitable set of
calibration curves as the first step, or the calibration step can be
omitted, and the display cansimply be profiled.<br>
<br>
The first step in profiling any output device, is to create a set of
device colorspace test values. The important parameters needed are:
<br>
<ul>
<li>What colorspace does the device use ?</li>
<li>How many test patches do I want to use ?</li>
<li>What information do I already have about how the device
behaves ?</li>
</ul>
For a display device, the colorspace will be RGB. The number
of test patches will depend somewhat on what quality profile you
want to make, what type of profile you want to make, and how long
you are prepared to wait when testing the display.<br>
At a minimum, a few hundred values are needed. A matrix/shaper type
of profile can get by with fewer test values, while a LUT based
profile will give better results if more test values are used. A
typical number might be 200-600 or so values, while 1000-2000 is not
an unreasonable number for a high quality characterization of a
display.<br>
<br>
To assist the choice of test patch values, it can help to have a
rough idea of how the device behaves. This could be in the form of
an ICC profile of a similar device, or a lower quality, or previous
profile for that particular device. If one were going to make a very
high quality LUT based profile, then it might be worthwhile to make
up a smaller, preliminary shaper/matrix profile using a few hundred
test points, before embarking on testing the device with several
thousand.<br>
<br>
Lets say that we ultimately want to make a profile for the device
"DisplayA", the simplest approach is to make a set of test values
that is independent of the characteristics of the particular device:<br>
<br>
<a href="targen.html">targen</a> <a href="targen.html#v">-v</a>
<a href="targen.html#d">-d3</a> <a href="targen.html#f">-f500</a>
<a href="targen.html#p1">DisplayA</a><br>
<br>
If there is a preliminary or previous profile called "OldDisplay"
available, and we want to try creating a "pre-conditioned" set of
test values that will more efficiently sample the device response,
then the following would achieve this:<br>
<u><br>
</u><a href="targen.html"> targen</a> <a href="targen.html#v">-v</a>
<a href="targen.html#d">-d3</a> <a href="targen.html#f">-f500</a>
<a href="targen.html#c">-cOldDisplay.icm</a> <a
href="targen.html#p1">DisplayA</a><br>
<br>
The output of <b>targen</b> will be the file DisplayA.ti1,
containing the device space test values, as well as expected CIE
values used for chart recognition purposes.<br>
<br>
<h4><a name="PM3"></a>Profiling in several steps: Taking readings
from a display</h4>
First it is necessary to connect your measurement instrument to your
computer, and check which communication port it is connected to. In
the following example, it is assumed that the instrument is
connected to the default port 1, which is either the first USB
instrument found, or serial port found. Invoking dispread so as to
display the usage information (by using a flag -? or --) will list
the identified serial and USB ports, and their labels.<br>
<br>
<a href="dispread.html">dispread</a> <a href="dispread.html#v">-v</a>
<a href="dispread.html#p1">DisplayA</a><br>
<br>
If we created a calibration for the display using <a
href="dispcal.html">dispcal</a>, then we will want to use this
when we take the display readings (e.g. TargetA.cal from the
calibration example)..<br>
<br>
<a href="dispread.html">dispread</a> <a href="dispread.html#v">-v</a>
<a href="dispread.html#k">-k TargetA.cal</a> <a
href="dispread.html#p1">DisplayA</a><br>
<br>
<b>dispread</b> will display a test window in the middle of the
screen, and issue a series of instructions about placing the
instrument on the display. You may need to make sure that the
display cursor is not in the test window, and it may also be
necessary to disable any screensaver before starting the process.
Exactly the same facilities are provided to select alternate
displays using the <span style="font-weight: bold;">-d</span>
parameter, and an alternate location and size for the test window
using the <span style="font-weight: bold;">-P</span> parameter as
with <span style="font-weight: bold;">dispcal</span>.<br>
<h4><a name="PM4"></a>Profiling in several steps: Creating a display
profile</h4>
There are two basic choices of profile type for a display, a
shaper/matrix profile, or a LUT based profile. They have different
tradeoffs. A shaper/matrix profile will work well on a well behaved
display, that is one that behaves in an additive color manner, will
give very smooth looking results, and needs fewer test points to
create. A LUT based profile on the other hand, will model any
display behaviour more accurately, and can accommodate gamut mapping
and different intent tables. Often it can show some unevenness and
contouring in the results though.<br>
<br>
To create a matrix/shaper profile, the following suffices:<br>
<br>
<a href="colprof.html">colprof</a> <a href="colprof.html#v">-v</a>
<a href="colprof.html#E">-D"Display A"</a> <a href="colprof.html#q">-qm</a>
<a href="colprof.html#a">-as</a> <a href="colprof.html#p1">DisplayA</a><br>
<br>
For a LUT based profile, where gamut mapping is desired, then a
source profile will need to be provided to define the source gamut.
For instance, if the display profile was likely to be linked to a
CMYK printing source profile, say "swop.icm" or "fogra39l.icm", then
the following would suffice:<br>
<br>
<a href="colprof.html">colprof</a> <a href="colprof.html#v">-v</a>
<a href="colprof.html#E">-D"Display A"</a> <a href="colprof.html#q">-qm</a>
<a href="colprof.html#S">-S</a><a href="colprof.html#S">
fogra39l.icm</a> <a href="colprof.html#c">-cpp</a> <a
href="colprof.html#d">-dmt</a> <a href="colprof.html#p1">DisplayA</a><br>
<br>
Make sure you check the delta E report at the end of the profile
creation, to see if the profile is behaving reasonably.<br>
If a calibration file was used with <a href="dispread.html">dispread</a>,
then it will be converted to a vcgt tag in the profile, so that the
operating system or other system color tools load the lookup curves
into the display hardware, when the profile is used.<br>
<h4><a name="PM5"></a>Installing a display profile</h4>
<a href="dispwin.html">dispwin</a> provides a convenient way of
installing a profile as the default system profile for the chosen
display:<br>
<br>
<a href="dispwin.html">dispwin</a> <a href="dispwin.html#I">-I</a>
<a href="dispwin.html#p1">DisplayA.icm</a><br>
<br>
This also sets the display to the calibration contained in the
profile. If you want to try out a calibration before installing the
profile, using dispwin without the <span style="font-weight: bold;">-I</span>
option will load a calibration (ICC profile or .cal file) into the
current display.<br>
<br>
Some systems will automatically set the display to the calibration
contained in the installed profile (ie. OS X), while on other
systems (ie. MSWindows and Linux/X11) it is necessary to use some
tool to do this. On MSWindows XP you could install the
optional <span style="font-weight: bold;">Microsoft Color Control Panel Applet for Windows XP</span>
available for download from Microsoft to do this, but <span
style="font-weight: bold;">NOTE</span> however that it seems to
have a <span style="font-weight: bold;">bug</span>, in that it
sometimes associates the profiles with the <span
style="font-weight: bold;">wrong monitor</span> entry. Other
display calibration tools will often install a similar tool, so
beware of there being multiple, competing programs. [ Commonly these
will be in your Start->Programs->Startup folder. ]<br>
On Microsoft Vista, you need to use dispwin -L or some other tool to
load the installed profiles calibration at startup.<br>
<br>
To use dispwin to load the installed profiles calibration to the
display, use<br>
<br>
<a href="dispwin.html">dispwin</a> <a href="dispwin.html#L">-L</a><br>
<br>
As per usual, you can select the appropriate display using the <a
href="dispwin.html#d">-d</a> flag.<br>
<br>
This can be automated on MSWindows and X11/Linux by adding this
command to an appropriate startup script.<br>
More system specific details, including how to create such startup
scripts are <a href="dispprofloc.html">here</a>. <br>
<br>
If you are using Microsoft <span style="font-weight: bold;">Vista</span>,
there is a known <span style="font-weight: bold;">bug</span> in
Vista that resets the calibration every time a fade-in effect is
executed, which happens if you lock and unlock the computer, resume
from sleep or hibernate, or User Access Control is activated. Using
<a href="dispwin.html">dispwin</a> <a href="dispwin.html#L">-L</a>
may not restore the calibration, because Vista filters out setting
(what it thinks) is a calibration that is already loaded. Use <a
href="dispwin.html">dispwin</a> <a href="dispwin.html#c">-c</a> <a
href="dispwin.html#L">-L</a><span style="font-family: monospace;"></span>
as a workaround, as this will first clear the calibration, then
re-load the current calibration.<br>
<br>
On X11/Linux systems, you could try adding <a href="dispwin.html">dispwin</a>
<a href="dispwin.html#L">-L</a> to your <span style="font-weight:
bold;">~/.config/autostart</span> file, so that your window
manager automatically sets calibration when it starts. If you are
running XRandR 1.2, you might consider running the experimental <a
href="dispwin.html#D">dispwin -E</a> in the background, as in its
"daemon" mode it will update the profile and calibration in response
to any changes in the the connected display.<br>
<br>
<h4><a name="PM6"></a>Expert tips when measuring displays:<br>
</h4>
Sometimes it can be difficult to get good quality, consistent and
visually relevant readings from displays, due to various practical
considerations with regard to instruments and the displays
themselves. Argyll's tools have some extra options that may assist
in overcoming these problems.<br>
<br>
If you are using an Eye-One Pro or ColorMunki spectrometer, then you
may wish to use the <a href="dispcal.html#H">high resolution
spectral mode</a> (<span style="font-weight: bold;">-H</span>).
This may be better at capturing the often narrow wavelength peaks
that are typical of display primary colors.<br>
<br>
Another option that can be used with the Eye-One Pro or ColorMunki
spectrometer is the <a href="dispcal.html#V">adaptive measurement
mode</a> (<span style="font-weight: bold;">-V</span>). By default
a fixed measurement integration time is used, as this will give the
most consistent results, but for displays with high contrast ratio's
and deep blacks, the integration time may be too short to give
adequate precision. The adaptive measurement mode increases
integration time when measuring dark colors (which will increase the
overall calibration or profiling time), and is capable of achieving
higher precision for these dark measurements.<br>
<br>
All instruments depend on silicon sensors, and such sensors generate
a temperature dependant level of noise ("dark noise") that is
factored out of the measurements by a dark or black instrument
calibration. The spectrometers in particular need this calibration
before commencing each set of measurements. Often an instrument will
warm up as it sits on a display, and this warming up can cause the
dark noise to increase, leading to inaccuracies in dark patch
measurements. The longer the measurement takes, the worse this
problem is likely to be. One way of addressing this is to
"acclimatise" the instrument before commencing measurements by
placing it on the screen in a powered up state, and leaving it for
some time. (Some people leave it for up to an hour to acclimatise.).
Another approach is to try and <a href="dispcal.html#I">compensate
for dark calibration changes</a> (<span style="font-weight: bold;">-Ib</span>)
by doing on the fly calibrations during the measurements, based on
the assumption that the black level of the display itself won't
change significantly. <br>
<br>
Some displays take a long time to settle down and stabilise. The is
often the case with LCD (Liquid Crystal) displays that use
fluorescent back lights, and these sorts of displays can change in
brightness significantly with changes in temperature. One way of
addressing this is to make sure that the display is given adequate
time to warm up before measurements. Another approach is to try and
<a href="dispcal.html#I">compensate for display white level</a>
(<span style="font-weight: bold;">-Iw</span>) changes by doing on
the fly calibrations during the measurements. Instrument black level
drift and display white level drift can be combined (<span
style="font-weight: bold;">-Ibw</span>).<br>
<br>
Colorimeter instruments make use of physical color filters that
approximate the standard observer spectral sensitivity curves.
Because these filters are not perfectly accurate, the manufacturer
calibrates the instrument for typical displays, which is why you
have to make a selection between CRT (Cathode Ray Tube) and LCD
(Liquid Crystal Display) modes. If you are measuring a display that
has primary colorants that differ significantly from those typical
displays, (ie. you have a Wide Gamut Display), then you may
get disappointing results with a Colorimeter. One way of addressing
this problem is to use a <a href="File_Formats.html#.ccmx">Colorimeter
Correction Matrix</a>. These are specific to a particular
Colorimeter and Display make and model combination, although a
matrix for a different but similar type of display may give better
results than none at all. A list of contributed <span
style="font-weight: bold;">ccmx</span> files is <a
href="ccmxs.html">here</a>.<br>
<br>
<h4><a name="PM7"></a>Calibrating and profiling a display that
doesn't have VideoLUT access.</h4>
<p>In some situation there is no access to a displays VideoLUT
hardware, and this hardware is what is usually used to implement
display calibration. This could be because the display is being
accessed via a web server, or because the driver or windowing
system doesn't support VideoLUT access.<br>
</p>
<p>There are two basic options in this situation:<br>
</p>
<p> 1) Don't attempt to calibrate, just profile the display.<br>
2) Calibrate, but incorporate the calibration in some other
way in the workflow.<br>
</p>
<p>The first case requires nothing special - just skip calibration
(see the previous section <a href="#PM7">Profiling in several
steps: Creating display test values</a>).</p>
<p> In the second case, there are three choices:<br>
</p>
<p> 2a) Use dispcal to create a calibration and a quick profile
that incorporates the calibration into the profile.<br>
2b) Use dispcal to create the calibration, then dispread and
colprof to create a profile, and then incorporate the calibration
into the profile using applycal.<br>
2c) Use dispcal to create the calibration, then dispread and
colprof to create a profile, and then apply the calibration after
the profile in a cctiff workflow.<br>
</p>
<p>The first case requires nothing special, use dispcal in a normal
fashioned with the <span style="font-weight: bold;">-o</span>
option to generate a quick profile.The profile created will <span
style="text-decoration: underline;">not</span> contain a 'vcgt'
tag, but instead will have the calibration curves incorporated
into the profile itself. If calibration parameters are chosen that
change the displays white point or brightness, then this will
result in a slightly unusual profile that has a white point that
does not correspond with device R=G=B=1.0. Some systems may not
cope properly with this type of profile, and a general shift in
white point through such a profile can create an odd looking
display if it is applied to images but not to other elements on
the display say as GUI decoration elements or other application
windows.<br>
</p>
<p>In the second case, the calibration file created using dispcal
should be provided to dispread using the <span
style="font-weight: bold;">-K</span> flag:<br>
</p>
<p><a href="dispread.html">dispread</a> <a href="dispread.html#v">-v</a>
<a href="dispread.html#K">-K TargetA.cal</a> <a
href="dispread.html#p1">DisplayA</a></p>
<p><span style="font-weight: bold;"></span>Create the profile as
usual using colprof. but note that colprof will ignore the
calibration, and that no 'vcgt' tag will be added to the profile.<br>
You can then use <a href="applycal.html">applycal </a>to combine
the calibration into the profile. Note that the resulting profile
will be slightly unusual, since the profile is not made completely
consistent with the effects of the calibration, and the device
R=G=B=1.0 probably not longer corresponds with the PCS white or
the white point.<br>
</p>
In the third case, the same procedure as above is used to create a
profile, but the calibration is applied in a raster transformation
workflow explicitly, e.g.:<br>
<br>
<a href="cctiff.html">cctiff</a> <a
href="cctiff.html#p1">SourceProfile.icm</a> <a
href="cctiff.html#p1">DisplayA.icm</a> <a href="cctiff.html#p2">DisplayA.cal</a>
<a href="cctiff.html#p3">infile.tif</a> <a href="cctiff.html#p4">outfile.tif</a><br>
or<br>
<a href="cctiff.html">cctiff</a> <a
href="cctiff.html#p1">SourceProfile.icm</a> <a
href="cctiff.html#p1">DisplayA.icm</a> <a href="cctiff.html#p2">DisplayA.cal</a>
<a href="cctiff.html#p3">infile.jpg</a> <a href="cctiff.html#p4">outfile.jpg</a><br>
<span style="font-weight: bold;"></span><br>
<hr size="2" width="100%">
<h3><a name="PS1"></a>Profiling Scanners and other input devices
such as cameras<br>
</h3>
Because a scanner or camera is an input device, it is necessary to
go about profiling it in quite a different way to an output device.
To profile it, a test chart is needed to exercise the input device
response, to which the CIE values for each test patch is known.
Generally standard reflection or transparency test charts are used
for this purpose.<br>
<h4><a name="PS2"></a>Types of test charts</h4>
The most common and popular test chart for scanner profiling is the
IT8.7/2 chart. This is a standard format chart generally reproduced
on photographic film, containing about 264 test patches.<br>
An accessible and affordable source of such targets is Wolf Faust a
<a href="http://www.targets.coloraid.de/">www.coloraid.de</a>.<br>
Another source is LaserSoft <a
href="http://www.silverfast.com/show/it8/en.html">www.silverfast.com.</a><br>
The Kodak Q-60 Color Input Target is also a typical example:<br>
<br>
<img src="Q60.jpg" alt="Kodak Q60 chart image" height="141"
width="200"> <br>
<br>
A very simple chart that is widely available is the Macbeth
ColorChecker chart, although it contains only 24 patches and
therefore is probably not ideal for creating profiles:<br>
<img alt="ColorChecker 24 patch" src="colorchecker.jpg"
style="width: 112px; height: 78px;"><br>
<br>
Other popular charts are the X-Rite/GretagMacbeth ColorChecker DC
and <a href="http://www.xrite.com/product_overview.aspx?ID=938">ColorChecker
SG</a> charts:<br>
<br>
<img src="DC.jpg" alt="GretagMacbeth ColorChecker DC chart"
height="122" width="200"> <img alt="ColorChecker SG" src="SG.jpg"
style="width: 174px; height: 122px;"><br>
<br>
The GretagMacbeth Eye-One Pro Scan Target 1.4 can also be used:<br>
<br>
<img alt="Eye-One Scan Target 1.4" src="i1scan14.jpg" style="border:
2px solid ; width: 200px; height: 140px;"><br>
<br>
Also supported is the <a href="http://www.hutchcolor.com/hct.htm">HutchColor
HCT</a> :<br>
<br>
<img alt="HutchColor HCT" src="HCT.jpg" style="width: 182px; height:
140px;"><br>
<br>
<br>
and <a
href="http://www.christophe-metairie-photographie.com/eng%20digital%20target.html">Christophe
Métairie's Digital TargeT 003</a> and <a
href="http://www.christophe-metairie-photographie.com/eng%20digital%20target.html">Christophe
Métairie's Digital Target - 3</a> :<br>
<br>
<img alt="CMP_DT_003" src="CMP_DT_003.jpg" style="width: 186px;
height: 141px;"> <img style="width: 203px; height: 140px;"
alt="CMP_Digital_Target-3" src="CMP_Digital_Target-3.jpg"><br>
<br>
and the <a href="http://www.silverfast.com/show/dc-targets/en.html">LaserSoft
Imaging DCPro Target</a>:<br>
<br>
<img style="width: 153px; height: 122px;" alt="LaserSoft DCPro
Target" src="LSDC.jpg"><br>
<br>
The Datacolor <a
href="http://spyder.datacolor.com/product-cb-spydercheckr.php">SpyderCheckr</a>:<br>
<br>
<img style=" width: 146px; height: 109px;" alt="Datacolor
SpyderCheckr" src="SpyderChecker.jpg"><br>
<br>
<h4><a name="PS3"></a>Taking readings from a scanner or camera<br>
</h4>
The test chart you are using needs to be placed on the scanner, and
the scanner needs to be configured to a suitable state, and restored
to that same state when used subsequently with the resulting
profile. For a camera, the chart needs to be lit in a controlled and
even manner using the light source that will be used for subsequent
photographs, and should be shot so as to minimise any geometric
distortion, although the <a href="scanin.html#p">scanin -p</a> flag
may be used to compensate for some degree of distortion. As with any
color profiling task, it is important to setup a known and
repeatable image processing flow, to ensure that the resulting
profile will be usable.<br>
<br>
The chart should be captured and saved to a TIFF format file. I will
assume the resulting file is called scanner.tif. The raster file
need only be roughly cropped so as to contain the test chart
(including the charts edges).<br>
<br>
The second step is to extract the RGB values from the scanner.tif
file, and match then to the reference CIE values. To locate the
patch values in the scan, the <b>scanin</b> tool needs to be given
a template <a href="File_Formats.html#.cht">.cht</a> file that
describes the features of the chart, and how the test patches are
labeled. Also needed is a file containing the CIE values for each of
the patches in the chart, which is typically supplied with the
chart, available from the manufacturers web site, or has been
measured using a spectrometer.<br>
<br>
<div style="margin-left: 40px;">For an IT8.7/2 chart, this is the <span
style="font-weight: bold;">ref/</span><b>it8.cht</b> file
supplied with Argyll, and the manufacturer will will supply
an individual or batch average file along with the chart
containing this information, or downloadable from their web site.<br>
NOTE that the reference file for the IT8.7/2 chart supplied with <span
style="font-weight: bold;">Monaco EZcolor</span> can be
obtained by unzipping the .mrf file. (You may have to make a copy
of the file with a .zip extension to do this.)<br>
<br>
For the ColorChecker 24 patch chart, the <span
style="font-weight: bold;">ref/ColorChecker.cht</span> file
should be used, and there is also a <span style="font-weight:
bold;">ref/ColorChecker.cie</span> file provided that is based
on the manufacturers reference values for the chart. You can also
create your own reference file using an instrument and chartread,
making use of the chart reference file <span style="font-weight:
bold;">ref/ColorChecker.ti2</span>:<br>
<a href="chartread.html">chartread</a> -n -a
ColorChecker.ti2<br>
Note that due to the small number of patches, a profile created
from such a chart is not likely to be very detailed.<br>
<br>
For the ColorChecker DC chart, the <span style="font-weight:
bold;">ref/ColorCheckerDC.cht</span> file should be used, and
there will be a ColorCheckerDC reference file supplied by
X-Rite/GretagMacbeth with the chart.<br>
<br>
The ColorChecker SG is relatively expensive, but is preferred by
many people because (like the ColorChecker and ColorCheckerDC) its
colors are composed of multiple different pigments, giving it
reflective spectra that are more representative of the real world,
unlike many other charts that are created out of combination of 3
or 4 colorants.<br>
A limited CIE reference file is available from X-Rite <a
href="http://www.xrite.com/documents/apps/public/digital_colorchecker_sg_l_a_b.txt">here</a>,
but it is not in the usual CGATS format. To convert it to a CIE
reference file useful for <span style="font-weight: bold;">scanin</span>,
you will need to edit the X-Rite file using a <span
style="text-decoration: underline;">plain text</span> editor,
first deleting everything before the line starting with "A1" and
everything after "N10", then prepending <a href="SG_header.txt">this
header</a>, and appending <a href="SG_footer.txt">this footer</a>,
making sure there are no blank lines inserted in the process.<br>
If you do happen to have access to a more comprehensive instrument
measurement of the ColorChecker SG, or you have measured it
yourself using a color instrument,<br>
then you <span style="text-decoration: underline;">may</span>
need to convert the reference information from spectral <span
style="font-weight: bold;">ColorCheckerSG.txt</span> file to CIE
value <span style="font-weight: bold;">ColorCheckerSG.cie</span>
reference file, follow the following steps:<br>
<a href="txt2ti3.html">txt2ti3</a>
ColorCheckerSG.txt ColorCheckerSG<br>
<a href="spec2cie.html">spec2cie</a>
ColorCheckerSG.ti3 ColorCheckerSG.cie<br>
<br>
For the Eye-One Pro Scan Target 1.4 chart, the <span
style="font-weight: bold;"><span style="font-weight: bold;">ref/</span>i1_RGB_Scan_1.4.cht</span>
file should be used, and as there is no reference file
accompanying this chart, the chart needs to be read with an
instrument (usually the Eye-One Pro). This can be done using
chartread, making use of the chart reference file <span
style="font-weight: bold;">ref/i1_RGB_Scan_1.4.ti2</span>:<br>
<a href="chartread.html">chartread</a> -n -a
i1_RGB_Scan_1.4<br>
and then rename the resulting <span style="font-weight: bold;">i1_RGB_Scan_1.4.ti3</span>
file to <span style="font-weight: bold;">i1_RGB_Scan_1.4.cie</span><br>
<span style="font-weight: bold;"></span><br>
For the HutchColor HCT chart, the <span style="font-weight:
bold;"><span style="font-weight: bold;">ref/</span>Hutchcolor.cht</span>
file should be used, and the reference <span style="font-weight:
bold;">.txt</span> file downloaded from the HutchColor website.<br>
<br>
For the Christophe Métairie's Digital TargeT 003 chart with
285 patches, the <span style="font-weight: bold;"><span
style="font-weight: bold;">ref/</span>CMP_DT_003.cht</span>
file should be used, and the cie reference <span
style="font-weight: bold;"></span>files come with the chart.<br>
<br>
For the Christophe Métairie's Digital Target-3 chart with
570 patches, the <span style="font-weight: bold;">ref/CMP_Digital_Target-3.cht</span>
file should be used, and the cie reference <span
style="font-weight: bold;"></span>files come with the chart.<br>
<br>
For the LaserSoft DCPro chart, the <span style="font-weight:
bold;">ref/LaserSoftDCPro.cht</span> file should be used, and
reference <span style="font-weight: bold;">.txt</span> file
downloaded from the <a
href="http://www.silverfast.com/it8calibration/">Silverfast
website</a>.<br>
<br>
For the Datacolor SpyderCheckr, the <span style="font-weight:
bold;">ref/SpyderChecker.cht</span> file should be used, and a
reference <span style="font-weight: bold;">ref/SpyderChecker.cie
</span>file made from measuring a sample chart is also available.
Alternately you could create your own reference file by
transcribing the <a
href="http://spyder.datacolor.com/images/photo_checkr_colordatainfo.jpg">values</a>
on the Datacolor website. <br>
</div>
<br>
For any other type of chart, a chart recognition template file will
need to be created (this is beyond the scope of the current
documentation, although see the <a href="cht_format.html">.cht_format
documentation</a>).<br>
<br>
To create the scanner .ti3 file, run the <b>scanin</b> tool as
follows (assuming an IT8 chart is being used):<br>
<br>
<a href="scanin.html"> scanin</a> -v scanner.tif It8.cht It8ref.txt<br>
<br>
"It8ref.txt" or "It8ref.cie" is assumed to be the name of the CIE
reference file supplied by the chart manufacturer. The resulting
file will be named "<b>scanner.ti3</b>".<br>
<br>
<span style="font-weight: bold;">scanin</span> will process 16 bit
per component .tiff files, which (if the scanner is capable of
creating such files), may improve the quality of the profile.
<br>
<br>
If you have any doubts about the correctness of the chart
recognition, or the subsequent profile's delta E report is unusual,
then use the scanin diagnostic flags <a href="scanin.html#d">-dipn</a>
and examine the <span style="font-weight: bold;">diag.tif</span>
diagnostic file, to make sure that the patches are identified and
aligned correctly. If you have problems getting good automatic
alignment, then consider doing a manual alignment by locating the
fiducial marks on your scan, and feeding them into scanin <a
href="scanin.html#F">-F</a> parameters. The fiducial marks should
be listed in a clockwise direction starting at the top left.<br>
<h4><a name="PS4"></a>Creating a scanner or camera input profile</h4>
Similar to a display profile, an input profile can be either a
shaper/matrix or LUT based profile. Well behaved input devices will
probably give the best results with a shaper/matrix profile, and
this may also be the best choice if your test chart has a small or
unevenly distributed set of test patchs (ie. the IT8.7.2). If a
shaper/matrix profile is a poor fit, consider using a LUT type
profile.<br>
<br>
When creating a LUT type profile, there is the choice of XYZ or
L*a*b* PCS (Device independent, Profile Connection Space). Often for
input devices, it is better to choose the XYZ PCS, as this may be a
better fit given that input devices are usually close to being
linearly additive in behaviour.<br>
<br>
If the purpose of the input profile is to use it as a substitute for
a colorimeter, then the <b>-u</b> flag should be used to avoid
clipping values above the white point. Unless the shaper/matrix type
profile is a very good fit, it is probably advisable to use a LUT
type profile in this situation.<br>
<br>
To create a matrix/shaper profile, the following suffices:<br>
<br>
<a href="colprof.html">colprof</a> <a href="colprof.html#v">-v</a>
<a href="colprof.html#E">-D"Scanner</a> <a href="colprof.html#E">A"</a>
<a href="colprof.html#q">-qm</a> <a href="colprof.html#a">-as</a> <a
href="colprof.html#p1">scanner</a><br>
<br>
For an XYZ PCS LUT based profile then the following would be used:<br>
<br>
<a href="colprof.html">colprof</a> <a href="colprof.html#v">-v</a>
<a href="colprof.html#E">-D"Scanner A"</a> <a href="colprof.html#q">-qm</a>
<a href="colprof.html#a">-ax</a> <a href="colprof.html#p1">scanner</a><br>
<br>
For the purposes of a poor mans colorimeter, the following would
generally be used:<br>
<br>
<a href="colprof.html">colprof</a> <a href="colprof.html#v">-v</a>
<a href="colprof.html#E">-D"Scanner A"</a> <a href="colprof.html#q">-qm</a>
<a href="colprof.html#a">-ax</a> <a href="colprof.html#u">-u</a> <a
href="colprof.html#p1">scanner</a><br>
<br>
Make sure you check the delta E report at the end of the profile
creation, to see if the profile is behaving reasonably.<br>
<br>
<br>
If profiling a <span style="font-weight: bold;">camera</span> in <span
style="font-weight: bold;">RAW</span> mode, then there may be some
advantage in creating a pure matrix only profile, in which it is
assumed that the camera response is completely linear. This may
reduce extrapolation artefacts. If setting the white point will be
done in some application, then it may also be an advantage to use
the <span style="font-weight: bold;">-u</span> flag and avoid
setting the white point to that of the profile chart:<br>
<br>
<a href="colprof.html">colprof</a> <a href="colprof.html#v">-v</a>
<a href="colprof.html#E">-D"Camera"</a> <a href="colprof.html#q">-qm</a>
<a href="colprof.html#a">-am</a> <a href="colprof.html#u">-u</a> <a
href="colprof.html#p1">scanner</a><br>
<br>
<br>
<hr size="2" width="100%">
<h3><a name="PP1"></a>Profiling Printers<br>
</h3>
The overall process is to create a set of device measurement target
values, print them out, measure them, and then create an ICC profile
from the measurements. If the printer is an RGB based printer, then
the process is only slightly more complicated than profiling a
display. If the printer is CMYK based, then some additional
parameters are required to set the total ink limit (TAC) and
black generation curve.<br>
<h4><a name="PP2"></a>Creating a print profile test chart</h4>
The first step in profiling any output device, is to create a set of
device colorspace test values. The important parameters needed are:<br>
<ul>
<li>What colorspace does the device use ?</li>
<li>How many test patches do I want to use/what paper size do I
want to use ?</li>
<li>What instrument am I going to use to read the patches ?<br>
</li>
<li>If it is a CMYK device, what is the total ink limit ?<br>
</li>
<li>What information do I already have about how the device
behaves ?</li>
</ul>
Most printers running through simple drivers will appear as if they
are RGB devices. In fact there is no such thing as a real RGB
printer, since printers use white media and the colorant must
subtract from the light reflected on it to create color, but the
printer itself turns the incoming RGB into the native print
colorspace, so for this reason we will tell targen to use the "Print
RGB" colorspace, so that it knows that it's really a subtractive
media. Other drivers will drive a printer more directly, and will
expect a CMYK profile. [Currently Argyll is not capable of creating
an ICC profile for devices with more colorants than CMYK. When this
capability is introduced, it will by creating an additional
separation profile which then allows the printer to be treated as a
CMY or CMYK printer.] One way of telling what sort of profile is
expected for your device is to examine an existing profile for that
device using <a href="http://www.argyllcms.com/doc/iccdump.html">iccdump</a>.<br>
<br>
The number of test patches will depend somewhat on what quality
profile you want to make, how well behaved the printer is, as well
as the effort needed to read the number of test values. Generally it
is convenient to fill a certain paper size with the maximum number
of test values that will fit.<br>
<br>
At a minimum, for an "RGB" device, a few hundred values are needed
(400-1000). For high quality CMYK profiles, 1000-3000 is not an
unreasonable number of patches.<br>
<br>
To assist the determination of test patch values, it can help to
have a rough idea of how the device behaves, so that the device test
point locations can be pre-conditioned. This could be in the form of
an ICC profile of a similar device, or a lower quality, or previous
profile for that particular device. If one were going to make a very
high quality Lut based profile, then it might be worthwhile to make
up a smaller, preliminary shaper/matrix profile using a few hundred
test points, before embarking on testing the device with several
thousand.<br>
<br>
The documentation for the <a
href="http://www.argyllcms.com/doc/targen.html">targen</a> tool
lists a <a href="http://www.argyllcms.com/doc/targen.html#Table">table</a>
of paper sizes and number of patches for typical situations.<br>
<br>
For a CMYK device, a total ink limit usually needs to be specified.
Sometimes a device will have a maximum total ink limit set by its
manufacturer or operator, and some CMYK systems (such as chemical
proofing systems) don't have any limit. Typical printing devices
such as Xerographic printers, inkjet printers and printing presses
will have a limit. The exact procedure for determining an ink limit
is outside the scope of this document, but one way of going about
this might be to generate some small (say a few hundred patches)
with targen & pritntarg with different total ink limits, and
printing them out, making the ink limit as large as possible without
striking problems that are caused by too much ink.<br>
<br>
Generally one wants to use the maximum possible amount of ink to
maximize the gamut available on the device. For most CMYK devices,
an ink limit between 200 and 400 is usual, but and ink limit of 250%
or over is generally desirable for reasonably dense blacks and dark
saturated colors. And ink limit of less than 200% will begin to
compromise the fully saturated gamut, as secondary colors (ie
combinations of any two primary colorants) will not be able to reach
full strength.<br>
<br>
Once an ink limit is used in printing the characterization test
chart for a device, it becomes a critical parameter in knowing what
the characterized gamut of the device is. If after printing the test
chart, a greater ink limit were to be used, the the software would
effectively be extrapolating the device behaviour at total ink
levels beyond that used in the test chart, leading to inaccuracies.<br>
<br>
Generally in Argyll, the ink limit is established when creating the
test chart values, and then carried through the profile making
process automatically. Once the profile has been made however, the
ink limit is no longer recorded, and you, the user, will have to
keep track of it if the ICC profile is used in any program than
needs to know the usable gamut of the device.<br>
<br>
<br>
Lets consider two devices in our examples, "PrinterA" which is an
"RGB" device, and "PrinterB" which is CMYK, and has a target ink
limit of 250%. <br>
<br>
The simplest approach is to make a set of test values that is
independent of the characteristics of the particular device:<br>
<br>
<a href="targen.html">targen</a> <a href="targen.html#v">-v</a>
<a href="targen.html#d">-d2</a> <a href="targen.html#f">-f1053</a>
<a href="targen.html#p1">PrinterA</a><br>
<br>
<a href="targen.html">targen</a> <a href="targen.html#v">-v</a>
<a href="targen.html#d">-d4</a> <a href="targen.html#l">-l260</a>
<a href="targen.html#f">-f1053</a> <a href="targen.html#p1">PrinterB</a><br>
<br>
The number of patches chosen here happens to be right for an A4
paper size being read using a Spectroscan instrument. See the <a
href="targen.html#Table">table</a> in the <a
href="targen.html">targen</a> documentation for some other
suggested numbers.<br>
<br>
If there is a preliminary or previous profile called "OldPrinterA"
available, and we want to try creating a "pre-conditioned" set of
test values that will more efficiently sample the device response,
then the following would achieve this:<u><br>
</u><br>
<a href="targen.html">targen</a> <a href="targen.html#v">-v</a>
<a href="targen.html#d">-d2</a> <a href="targen.html#f">-f1053</a>
<a href="targen.html#c">-c OldPrinterA</a> <a
href="targen.html#p1">PrinterA</a><br>
<br>
<a href="targen.html">targen</a> <a href="targen.html#v">-v</a>
<a href="targen.html#d">-d4</a> <a href="targen.html#l">-l260</a>
<a href="targen.html#f">-f1053</a> <a href="targen.html#c">-c
OldPrinterB</a> <a href="targen.html#p1">PrinterB</a><br>
<a href="targen.html#p1"></a><br>
<br>
The output of <b>targen</b> will be the file PrinterA.ti1 and
PrinterB.ti1 respectively, containing the device space test values,
as well as expected CIE values used for chart recognition purposes.<br>
<br>
<h4><a name="PP2b"></a>Printing a print profile test chart<br>
<br>
</h4>
The next step is turn the test values in to a PostScript or TIFF
raster test file that can printed on the device. The basic
information that needs to be supplied is the type of instrument that
will be used to read the patches, as well as the paper size it is to
be formatted for.<br>
<br>
For an X-Rite DTP41, the following would be typical:<br>
<br>
<a href="printtarg.html">printtarg</a> <a href="printtarg.html#v">-v</a>
<a href="printtarg.html#i">-i41</a> <a href="printtarg.html#p">-pA4</a>
<a href="printtarg.html#p1">PrinterA</a><br>
<br>
For a Gretag Eye-One Pro, the following would be typical:<br>
<br>
<a href="printtarg.html">printtarg</a> <a href="printtarg.html#v">-v</a>
<a href="printtarg.html#i">-ii1</a> <a href="printtarg.html#p">-pA4</a>
<a href="printtarg.html#p1">PrinterA</a><br>
<br>
For using with a scanner as a colorimeter, the Gretag Spectroscan
layout is suitable, but the <a href="printtarg.html#s">-s</a> flag
should be used so as to generate a layout suitable for scan
recognition, as well as generating the scan recognition template
files. (You probably want to use less patches with <span
style="font-weight: bold;">targen</span>, when using the <span
style="font-weight: bold;">printtarg -s</span> flag, e.g. 1026
patches for an A4R page, etc.) The following would be typical:<br>
<br>
<a href="printtarg.html">printtarg</a> <a href="printtarg.html#v">-v</a>
<a href="printtarg.html#s">-s</a> <a href="printtarg.html#i">-iSS</a>
<a href="printtarg.html#p">-pA4R</a> <a href="printtarg.html#p1">PrinterA</a><br>
<span style="font-weight: bold;"><br>
printtarg</span> reads the PrinterA.ti1 file, creates a
PrinterA.ti2 file containing the layout information as well as the
device values and expected CIE values, as well as a PrinterA.ps file
containing the test chart. If the <span style="font-weight: bold;">-s</span>
flag is used, one or more PrinterA.cht files is created to allow the
<a href="scanin.html">scanin</a> program to recognize the chart.<br>
<br>
To create TIFF raster files rather than PostScript, use the <a
href="printtarg.html#t"><span style="font-weight: bold;">-t</span></a>
flag.<br>
<br>
<span style="font-weight: bold;">GSview</span> is a good program to
use to check what the PostScript file will look like, without
actually printing it out. You could also use <span
style="font-weight: bold;">Photoshop</span> or <span
style="font-weight: bold;">ImageMagick</span> for this purpose.<br>
<br>
The last step is to print the chart out.<br>
<br>
Using a suitable PostScript or raster file printing program,
downloader, print the chart. If you are not using a TIFF test chart,
and you do not have a PostScript capable printer, then an
interpreter like GhostScript or even Photoshop could be used to
rasterize the file into something that can be printed. Note that it
is important that the PostScript interpreter or TIFF printing
application and printer configuration is setup for a device
profiling run, and that any sort of color conversion of color
correction be turned off so that the device values in the PostScript
or TIFF file are sent directly to the device. If the device has a
calibration system, then it would be usual to have setup and
calibrated the device before starting the profiling run, and to
apply calibration to the chart values. If Photoshop was to be used,
then either the chart needs to be a single page, or separate .eps or
.tiff files for each page should be used, so that they can be
converted and printed one at a time (see the <a
href="printtarg.html#e">-e</a> and <a href="printtarg.html#t">-t</a>
flags).<br>
<br>
<h4><a name="PP3"></a>Reading a print test chart using an instrument</h4>
Once the test chart has been printed, the color of the patches needs
to be read using a suitable instrument.<br>
<br>
Several different instruments are currently supported, some that
need to be used patch by patch, some read a strip at a time, and
some read a sheet at a time. See <a href="instruments.html">instruments</a>
for a current list.<br>
<br>
The instrument needs to be connected to your computer before running
the <a href="chartread.html">chartread</a> command. Both serial
port and USB connected Instruments are supported. A serial port to
USB adapter might have to be used if your computer doesn't have any
serial ports, and you have a serial interface connected instrument.<br>
<br>
If you run <a href="chartread.html">chartread</a> so as to print
out its usage message (ie. by using a <span style="font-weight:
bold;">-?</span> or <span style="font-weight: bold;">--</span>
flags), then it will list any identified serial ports or USB
connected instruments, and their corresponding number for the <a
href="chartread.html#c">-c</a> option. By default, <a
href="chartread.html">chartread</a> will try to connect to the
first available USB instrument, or an instrument on the first serial
port.<br>
<br>
The only arguments required is to specify the basename of the .ti2
file. If a non-default serial port is to be used, then the <span
style="font-weight: bold;">-c</span> option would also be
specified.<br>
<br>
e.g. for a Spectroscan on the second port:<br>
<br>
<a href="chartread.html">chartread</a> <a href="chartread.html#c">-c2</a>
<a href="chartread.html#p1">PrinterA</a><br>
<br>
For a DTP41 to the default serial port:<br>
<br>
<a href="chartread.html">chartread</a><a href="chartread.html#i"></a>
<a href="chartread.html#p1">PrinterA</a><br>
<br>
<span style="font-weight: bold;">chartread</span> will interactively
prompt you through the process of reading each sheet or strip. See <a
href="chartread.html">chartread</a> for more details on the
responses for each type of instrument. Continue with <a
href="Scenarios.html#PP5">Creating a printer profile</a>.<br>
<br>
<h4><a name="PP4"></a>Reading a print test chart using a scanner or
camera<br>
</h4>
<br>
Argyll supports using a scanner or even a camera as a substitute for
a colorimeter. While a scanner or camera is no replacement for a
color measurement instrument, it may give acceptable results in some
situations, and may give better results than a generic profile for a
printing device.<br>
<br>
The main limitation of the scanner-as-colorimeter approach are:<br>
<br>
* The scanner dynamic range and/or precision may not match the
printers or what is required for a good profile.<br>
* The spectral interaction of the scanner test chart and printer
test chart with the scanner spectral response can cause color
errors.<br>
* Spectral differences caused by different black amounts in the
print test chart can cause color errors. <br>
* The scanner reference chart gamut may be much smaller than the
printers gamut, making the scanner profile too inaccurate to be
useful. <br>
<br>
As well as some of the above, a camera may not be suitable if it
automatically adjusts exposure or white point when taking a picture,
and this behavior cannot be disabled.<br>
<br>
The end result is often a profile that has a noticeable color cast,
compared to a profile created using a colorimeter or spectrometer.<br>
<br>
<br>
It is assumed that you have created a scanner or camera profile
following the <a
href="http://www.argyllcms.com/doc/Scenarios.html#PS1">procedure</a>
outline above. For best possible results it is advisable to both
profile the scanner or camera, and use it in scanning the printed
test chart, in as "raw" mode as possible (i.e. using 16 bits per
component images, if the scanner or camera is capable of doing so;
not setting white or black points, using a fixed exposure etc.). It
is generally advisable to create a LUT type input profile, and use
the <a href="http://www.argyllcms.com/doc/colprof.html#u">-u</a>
flag to avoid clipping scanned value whiter than the input
calibration chart.<br>
<br>
Scan or photograph your printer chart (or charts) on the scanner or
camera previously profiled. <big><span style="font-weight: bold;">The
scanner or camera must be configured and used exactly the same
as it was when it was profiled.</span></big><br>
<br>
I will assume the resulting scan/photo input file is called <span
style="font-weight: bold;">PrinterB.tif</span> (or <span
style="font-weight: bold;">PrinterB1.tif</span>, <span
style="font-weight: bold;">PrinterB2.tif</span> etc. in the case
of multiple charts). As with profiling the scanner or camera, the
raster file need only be roughly cropped so as to contain the test
chart.<br>
<br>
The scanner recognition files created when <span
style="font-weight: bold;">printtarg</span> was run is assumed to
be called <span style="font-weight: bold;">PrinterB.cht</span>.
Using the scanner profile created previously (assumed to be called <span
style="font-weight: bold;">scanner.icm</span>), the printer test
chart scan patches are converted to CIE values using the <span
style="font-weight: bold;">scanin</span> tool:<br>
<br>
<a href="scanin.html">scanin</a> <a href="scanin.html#v">-v</a> <a
href="scanin.html#c">-c</a> <a href="scanin.html#cp1">PrinterB.tif</a>
<a href="scanin.html#cp2">PrinterB.cht</a> <a
href="scanin.html#cp3">scanner.icm</a> <a href="scanin.html#cp4">PrinterB</a><br>
<br>
If there were multiple test chart pages, the results would be
accumulated page by page using the <a href="scanin.html#ca">-ca</a>
option, ie., if there were 3 pages:<br>
<br>
<a href="scanin.html">scanin</a> <a href="scanin.html#v">-v</a> <a
href="scanin.html#c">-c</a> <a href="scanin.html#cp1">PrinterB1.tif</a>
<a href="scanin.html#cp2">PrinterB1.cht</a> <a
href="scanin.html#cp3">scanner.icm</a> <a href="scanin.html#cp4">PrinterB</a><br>
<a href="scanin.html">scanin</a> <a href="scanin.html#v">-v</a> <a
href="scanin.html#ca">-ca</a> <a href="scanin.html#cp1">PrinterB2.tif</a>
<a href="scanin.html#cp2">PrinterB2.cht</a> <a
href="scanin.html#cp3">scanner.icm</a> <a href="scanin.html#cp4">PrinterB</a><br>
<a href="scanin.html">scanin</a> <a href="scanin.html#v">-v</a> <a
href="scanin.html#ca">-ca</a> <a href="scanin.html#cp1">PrinterB3.tif</a>
<a href="scanin.html#cp2">PrinterB3.cht</a> <a
href="scanin.html#cp3">scanner.icm</a> <a href="scanin.html#cp4">PrinterB</a><br>
<br>
Now that the <span style="font-weight: bold;">PrinterB.ti3</span>
data has been obtained, the profile continue in the next section
with <span style="font-weight: bold;">Creating a printer profile</span>.<br>
<br>
If you have any doubts about the correctness of the chart
recognition, or the subsequent profile's delta E report is unusual,
then use the scanin diagnostic flags <a href="scanin.html#d">-dipn</a>
and examine the <span style="font-weight: bold;">diag.tif</span>
diagnostic file.<br>
<h4><a name="PP5"></a>Creating a printer profile<br>
</h4>
Creating an RGB based printing profile is very similar to creating a
display device profile. For a CMYK printer, some additional
information is needed to set the black generation.<br>
<br>
Where the resulting profile will be used conventionally (ie. using <a
href="collink.html">collink</a> <a href="collink.html#s">-s</a>,
or <a href="cctiff.html">cctiff</a> or most other "dumb" CMMs) it
is important to specify that gamut mapping should be computed for
the output (B2A) perceptual and saturation tables. This is done by
specifying a device profile as the parameter to the <a
href="colprof.html">colprof</a> <a href="colprof.html#S">-S</a>
flag. When you intend to create a "general use" profile, it can be a
good technique to specify the source gamut as the opposite type of
profile to that being created, i.e. if a printer profile is being
created, specify a display profile (e.g. sRGB) as the source gamut.
If a display profile is being created, then specify a printer
profile as the source (e.g. Figra, SWOP etc.). When linking to
the profile you have created this way as the output profile, then
use perceptual intent if the source is the opposite type, and
relative colorimetric if it is the same type.<br>
<br>
"Opposite type of profile" refers to the native gamut of the device,
and what its fundamental nature is, additive or subtractive. An
emissive display will have additive primaries (R, G & B), while
a reflective print, will have subtractive primaries (C, M, Y &
possibly others), irrespective of what colorspace the printer is
driven in (a printer might present an RGB interface, but internally
this will be converted to CMY, and it will have a CMY type of
gamut). Because of the complimentary nature of additive and
subtractive device primary colorants, these types of devices have
the most different gamuts, and hence need the most gamut mapping to
convert from one colorspace to the other.<br>
<br>
If you are creating a profile for a specific purpose, intending to
link it to a specific input profile, then you will get the best
results by specifying that source profile as the source gamut.<br>
<br>
If a profile is only going to be used as an input profile, or is
going to be used with a "smart" CMM (e.g. <a href="collink.html">collink</a>
<a href="collink.html#g">-g</a> or <a href="collink.html#G">-G</a>),
then
it can save considerable processing time and space if the -b flag is
used, and the -S flag not used.<br>
<br>
For an RGB printer intended to print RGB originals, the following
might be a typical profile usage:<br>
<br>
<a href="colprof.html">colprof</a> <a href="colprof.html#v">-v</a>
<a href="colprof.html#E">-D"Printer A"</a> <a href="colprof.html#q">-qm</a>
<a href="colprof.html#S">-S</a><a href="colprof.html#S"> sRGB.icm</a>
<a href="colprof.html#c">-cmt</a> <a href="colprof.html#d">-dpp</a>
<a href="colprof.html#p1">PrinterA</a><br>
<br>
or if you intent to print from Fogra, SWOP or other standard CMYK
style originals:<br>
<br>
<a href="colprof.html">colprof</a> <a href="colprof.html#v">-v</a>
<a href="colprof.html#E">-D"Printer A"</a> <a href="colprof.html#q">-qm</a>
<a href="colprof.html#S">-S</a><a href="colprof.html#S">
fogra39l.icm</a> <a href="colprof.html#c">-cmt</a> <a
href="colprof.html#d">-dpp</a> <a href="colprof.html#p1">PrinterA</a><br>
<br>
If you know what colorspace your originals are in, use that as the
argument to <span style="font-weight: bold;">-S</span>.<br>
<br>
<h4><a name="PP6"></a>Choosing a black generation curve (and other
CMYK printer options)<br>
</h4>
For a CMYK printer, it would be normal to specify the type of black
generation, either as something simple, or as a specific curve. The
documentation in <a href="colprof.html#k">colprof</a> for the
details of the options.<span style="font-weight: bold;"><br>
<br>
Note</span> that making a good choice of black generation curve
can affect things such as: how robust neutrals are given printer
drift or changes in viewing lighting, how visible screening is, and
how smooth looking the B2A conversion is.<br>
<br>
For instance, maximizing the level of K will mean that the neutral
colors are composed of greater amounts of Black ink, and black ink
retains its neutral appearance irrespective of printer behavior or
the spectrum of the illuminant used to view the print. On the other
hand, output which is dominantly from one of the color channels will
tend to emphasize the screening pattern and any unevenness (banding
etc.) of that channel, and the black channel in particular has the
highest visibility. So in practice, some balance between the levels
of the four channels is probably best, with more K if the screening
is fine and a robust neutral balance is important, or less K if the
screening is more visible and neutral balance is less critical. The
levels of K at the edges of the gamut of the device will be fixed by
the nature of the ink combinations that maximize the gamut (ie.
typically zero ink for light chromatic colors, some combination for
dark colors, and a high level of black for very dark near neutrals),
and it is also usually important to set a curve that smoothly
transitions to the K values at the gamut edges. Dramatic changes in
K imply equally dramatic changes in CMY, and these abrupt
transitions will reveal the limited precision and detail that can be
captured in a lookup table based profile, often resulting in a
"bumpy" looking output.<br>
<br>
If you want to experiment with the various black generation
parameters, then it might be a good idea to create a preliminary
profile (using <a href="colprof.html#q">-ql</a> <a
href="colprof.html#b">-b</a> <a href="colprof.html#ni">-no</a>, <a
href="colprof.html#no">-ni</a> and no <a href="colprof.html#S">-S</a>),
and then used <a href="xicclu.html#g">xicclu</a> to explore the
effect of the parameters.<br>
<br>
For instance, say we have our CMYK .ti3 file <span
style="font-weight: bold;">PrinterB.ti3</span>. First we make a
preliminary profile called <span style="font-weight: bold;">PrinterBt</span>:<br>
<br>
copy PrinterB.ti3 PrinterBt.ti3 (Use
"cp" on Linux or OSX of course.)<br>
<a href="colprof.html">colprof</a> <a href="colprof.html#v">-v</a>
<a href="colprof.html#q">-qm</a> <a href="colprof.html#b">-b</a> <a
href="colprof.html#c">-cmt</a> <a href="colprof.html#d">-dpp</a>
<a href="colprof.html#p1">PrinterBt</a><br>
<br>
Then see what the minimum black level down the neutral axis can be.
Note that we need to also set any ink limits we've decided on as
well (coloprof defaulting to 10% less than the value recorded in the
.ti3 file). In this example the test chart has a 300% total ink
limit, and we've decided to use 290%:<br>
<br>
<a href="xicclu.html">xicclu</a> <a href="xicclu.html#g">-g</a> <a
href="xicclu.html#k">-kz</a> <a href="xicclu.html#l">-l290</a> <a
href="xicclu.html#f">-fif</a> <a href="xicclu.html#i">-ir</a> <a
href="xicclu.html#p1">PrinterBt.icm</a><br>
<br>
Which might be a graph something like this:<br>
<br>
<img alt="Graph of CMYK neutral axis with minimum K"
src="Kgraph1.jpg" style="width: 250px; height: 250px;"><br>
<br>
Note how the minimum black is zero up to 93% of the
white->black L* curve, and then jumps up to 87%. This is because
we've reached the total ink limit, and K then has to be substituted
for CMY, to keep the total under the total ink limit.<br>
<br>
Then let's see what the maximum black level down the neutral axis
can be:<br>
<br>
<a href="xicclu.html">xicclu</a> <a href="xicclu.html#g">-g</a> <a
href="xicclu.html#k">-kx</a> <a href="xicclu.html#l">-l290</a> <a
href="xicclu.html#f">-fif</a> <a href="xicclu.html#i">-ir</a> <a
href="xicclu.html#p1">PrinterBt.icm</a><br>
<br>
Which might be a graph something like this:<br>
<br>
<img alt="Graph of CMYK neutral axis with maximum K"
src="Kgraph2.jpg" style="width: 250px; height: 250px;"><br>
<br>
Note how the CMY values are fairly low up to 93% of the
white->black L* curve (the low levels of CMY are helping set the
neutral color), and then they jump up. This is because we've reach
the point where black on it's own, isn't as dark as the color that
can be achieved using CMY and K. Because the K has a dominant effect
on the hue of the black, the levels of CMY are often fairly volatile
in this region.<br>
<br>
Any K curve we specify must lie between the black curves of the
above two graphs.<br>
<br>
Let's say we'd like to chose a moderate black curve, one that aims
for about equal levels of CMY and K. We should also aim for it to be
fairly smooth, since this will minimize visual artefacts caused by
the limited fidelity that profile LUT tables are able to represent
inside the profile.<br>
<br>
<img style="width: 340px; height: 258px;" alt="-k parameters"
src="Kparams.jpg"><br>
<br>
<br>
For minimum discontinuities we should aim for the curve to finish at
the point it has to reach to satisfy the total ink limit at 87%
curve and 93% black. For a first try we can simply set a straight
line to that point: <br>
<br>
<a href="xicclu.html">xicclu</a> <a href="xicclu.html#g">-g</a> <a
href="xicclu.html#k">-kp 0 0 .93 .87 1.0</a> <a
href="xicclu.html#l">-l290</a> <a href="xicclu.html#f">-fif</a> <a
href="xicclu.html#i">-ir</a> <a href="xicclu.html#p1">PrinterBt.icm</a><br>
<br>
<img alt="Graph of CMYK neutral axis with kp 0 0 1.0 1.0 1.0 -l290"
src="Kgraph3.jpg" style="width: 250px; height: 250px;"><br>
<br>
The black "curve" hits the 93%/87% mark well, but is a bit too far
above CMY, so we'll try making the black curve concave:<br>
<br>
<a href="xicclu.html">xicclu</a> <a href="xicclu.html#g">-g</a> <a
href="xicclu.html#k">-kp </a><a href="xicclu.html#k">0 0 .93 .87
0.65</a> <a href="xicclu.html#l">-l290</a> <a
href="xicclu.html#f">-fif</a> <a href="xicclu.html#i">-ir</a> <a
href="xicclu.html#p1">PrinterBt.icm</a><br>
<br>
<img alt="Graph of CMYK neutral axis with -kp 0 .05 1 .9 1 -l290"
src="Kgraph4.jpg" style="width: 250px; height: 249px;"><br>
<br>
This looks just about perfect, so the the curve parameters can now
be used to generate our real profile:<br>
<br>
<a href="colprof.html">colprof</a> <a href="colprof.html#v">-v</a>
<a href="colprof.html#E">-D"Printer B"</a> <a href="colprof.html#q">-qm</a>
<a href="colprof.html#k">-kp </a><a href="xicclu.html#k">0 0 .93
.87 0.65</a> <a href="colprof.html#S">-S</a><a
href="colprof.html#S"> sRGB.icm</a> <a href="colprof.html#c">-cmt</a>
<a href="colprof.html#d">-dpp</a> <a href="colprof.html#p1">PrinterB</a><br>
<br>
and the resulting B2A table black curve can be checked using xicclu:<br>
<br>
<a href="xicclu.html">xicclu</a> <a href="xicclu.html#g">-g</a> <a
href="xicclu.html#f">-fb</a> <a href="xicclu.html#i">-ir</a> <a
href="xicclu.html#p1">PrinterB.icm</a><br>
<br>
<img style="width: 250px; height: 250px;" alt="sadsadas"
src="Kgraph5.jpg"><br>
<br>
<br>
<hr style="margin-left: 0px; margin-right: auto; width: 20%; height:
2px;"><br>
<span style="font-weight: bold;">Examples of other inkings:<br>
<br>
</span>A smoothed zero black inking:<br>
<br>
<a href="xicclu.html">xicclu</a> <a href="xicclu.html#g">-g</a> <a
href="xicclu.html#k">-kp </a><a href="xicclu.html#k">0 .7 .93 .87
1.0</a> <a href="xicclu.html#l">-l290</a> <a
href="xicclu.html#f">-fif</a> <a href="xicclu.html#i">-ir</a> <a
href="xicclu.html#p1">PrinterBt.icm</a><br>
<br>
<img style="width: 250px; height: 250px;" alt="sadsadas"
src="Kgraph6.jpg"><br>
<br>
A low black inking:<br>
<br>
<a href="xicclu.html">xicclu</a> <a href="xicclu.html#g">-g</a> <a
href="xicclu.html#k">-kp </a><a href="xicclu.html#k">0 0 .93 .87
0.15</a> <a href="xicclu.html#l">-l290</a> <a
href="xicclu.html#f">-fif</a> <a href="xicclu.html#i">-ir</a> <a
href="xicclu.html#p1">PrinterBt.icm</a><br>
<br>
<img style="width: 250px; height: 250px;" alt="sadsadas"
src="Kgraph7.jpg"><br>
<br>
<br>
A high black inking:<br>
<br>
<a href="xicclu.html">xicclu</a> <a href="xicclu.html#g">-g</a> <a
href="xicclu.html#k">-kp </a><a href="xicclu.html#k">0 0 .93 .87
1.2</a> <a href="xicclu.html#l">-l290</a> <a
href="xicclu.html#f">-fif</a> <a href="xicclu.html#i">-ir</a> <a
href="xicclu.html#p1">PrinterBt.icm</a><br>
<br>
<img style="width: 250px; height: 250px;" alt="sadsadas"
src="Kgraph8.jpg"><br>
<br>
<span style="font-weight: bold;"></span>
<h4>Overriding the ink limit<br>
</h4>
Normally the total ink limit will be read from the <span
style="font-weight: bold;">PrinterB.ti3</span> file, and will be
set at a level 10% lower than the number used in creating the test
chart values using <a href="targen.html#l">targen -l</a>. If you
want to override this with a lower limit, then use the <a
href="colprof.html#l">-l flag</a>.<br>
<br>
<a href="colprof.html">colprof</a> <a href="colprof.html#v">-v</a>
<a href="colprof.html#E">-D"Printer B"</a> <a href="colprof.html#q">-qm</a>
<a href="colprof.html#S">-S</a><a href="colprof.html#S"> sRGB.icm</a>
<a href="colprof.html#c">-cmt</a> <a href="colprof.html#d">-dpp</a>
<a href="colprof.html#k">-kr</a> <a href="xicclu.html#l">-l290</a>
<a href="colprof.html#p1">PrinterB</a><br>
<br>
Make sure you check the delta E report at the end of the profile
creation, to see if the profile is behaving reasonably.<br>
<br>
One way of checking that your ink limit is not too high, is to use "<span
style="font-weight: bold;">xicc -fif -ia</span>" to check, by
setting different ink limits using the <span style="font-weight:
bold;">-l</span> option, feeding Lab = 0 0 0 into it, and checking
the resulting black point. Starting with the ink limit used
with <span style="font-weight: bold;">targen</span> for the test
chart, reduce it until the black point starts to be affected. If it
is immediately affected by any reduction in the ink limit, then the
black point may be improved by increasing the ink limit used to
generate the test chart and then re-print and re-measuring it,
assuming other aspects such as wetness, smudging, spreading or
drying time are not an issue.<br>
<br>
<hr style="width: 100%; height: 2px;"><br>
<h3><a name="PC1"></a>Calibrating Printers<br>
</h3>
<span style="font-weight: bold;">Profiling</span> creates a
description of how a device behaves, while <span
style="font-weight: bold;">calibration</span> on the other hand is
intended to <span style="text-decoration: underline;">change</span>
how a device behaves. Argyll has the ability to create per-channel
device space calibration curves for print devices, that can then be
used to improve the behavior of of the device, making a subsequent
profile fit the device more easily and also allow day to day
correction of device drift without resorting to a full re-profile.<br>
<br>
<span style="font-weight: bold;">NOTE:</span> Because calibration
adds yet another layer to the way color is processed, it is
recommended that it not be attempted until the normal profiling
workflow is established, understood and verified.<br>
<h4><a name="PC2"></a>Calibrated print workflows</h4>
There are two main workflows that printer calibration curves can be
applied to:<br>
<br>
<span style="text-decoration: underline;">Workflow <span
style="font-weight: bold;">with</span> native calibration
capability</span>:<br>
<br>
Firstly the printer itself may have the capability of using per
channel calibration curves. In this situation, the calibration
process will be largely independent of profiling. Firstly the
printer is configured to have both its color management and
calibration disabled (the latter perhaps achieved by loading linear
calibration curves), and a print calibration test chart that
consists of per channel color wedges is printed. The calibration
chart is read and the resulting .ti3 file converted into calibration
curves by processing it using <span style="font-weight: bold;">printcal</span>.
The calibration is then installed into the printer. Subsequent
profiling will be performed on the <span style="text-decoration:
underline;">calibrated</span> printer (ie. the profile test chart
will have the calibration curves applied to it by the printer, and
the resulting ICC profile will represent the behavior of the
calibrated printer.)<br>
<br>
<span style="text-decoration: underline;">Workflow <span
style="font-weight: bold;">without</span> native calibration
capability</span>:<br>
<br>
The second workflow is one in which the printer has no calibration
capability itself. In this situation, the calibration process will
have to be applied using the ICC color management tools, so careful
coordination with profiling is needed. Firstly the printer is
configured to have its color management disabled, and a print
calibration test chart that consists of per channel color wedges is
printed. The calibration chart is converted into calibration curves
by reading it and then processing the resultant .ti3 using <span
style="font-weight: bold;">printcal</span>,. During the subsequent
<span style="text-decoration: underline;">profiling</span>, the
calibration curves will need to be applied to the profile test chart
in the process of using <span style="font-weight: bold;">printtarg</span>.
Once the the profile has been created, then in subsequent printing
the calibration curves will need to be applied to an image being
printed either explicitly when using <span style="font-weight:
bold;">cctiff</span> to apply color profiles <span
style="text-decoration: underline;">and</span> calibration, <span
style="font-weight: bold;">OR</span> by creating a version of the
profile that has had the calibration curves incorporated into it
using the <span style="font-weight: bold;">applycal</span> tool.
The latter is useful when some CMM (color management module) other
than <span style="font-weight: bold;">cctiff </span>is being used.<br>
<br>
Once calibration aim targets for a particular device and mode
(screening, paper etc.) have been established, then the printer can
be re-calibrated at any time to bring its per channel behavior back
into line if it drifts, and the new calibration curves can be
installed into the printer, or re-incorporated into the profile.
<h4><a name="PC3"></a>Creating a print calibration test chart</h4>
The first step is to create a print calibration test chart. Since
calibration only creates per-channel curves, only single channel
step wedges are required for the chart. The main choice is the
number of steps in each wedge. For simple fast calibrations perhaps
as few as 20 steps per channel may be enough, but for a better
quality of calibration something like 50 or more steps would be a
better choice.<br>
<br>
Let's consider two devices in our examples, "PrinterA" which is an
"RGB" printer device, and "PrinterB" which is CMYK. In fact there is
no such thing as a real RGB printer, since printers use white media
and the colorant must subtract from the light reflected on it to
create color, but the printer itself turns the incoming RGB into the
native print colorspace, so for this reason we are careful to tell
targen to use the "Print RGB" colorspace, so that it knows to create
step wedges from media white to full colorant values.<br>
<br>
For instance, to create a 50 steps per channel calibration test
chart for our RGB and CMYK devices, the following would be
sufficient:<br>
<br>
<a href="targen.html">targen</a> <a href="targen.html#v">-v</a>
<a href="targen.html#d">-d2</a> <a href="targen.html#s">-s50</a>
<a href="targen.html#e">-e3</a> <a href="targen.html#f">-f0</a> <a
href="targen.html#p1">PrinterA_c</a><br>
<br>
<a href="targen.html">targen</a> <a href="targen.html#v">-v</a>
<a href="targen.html#d">-d4</a> <a href="targen.html#s">-s50</a>
<a href="targen.html#e">-e4</a> <a href="targen.html#f">-f0</a> <a
href="targen.html#p1">PrinterB_c</a><br>
<a href="targen.html#p1"></a><br>
For an outline of how to then print and read the resulting test
chart, see <a href="Scenarios.html#PP2b">Printing a print
profile test chart</a>, and <a href="Scenarios.html#PP3">Reading
a print test chart using an instrument</a>. Note that the printer
must be in an un-profiled and un-calibrated mode when doing this
print. Having done this, there will be a PrinterA.ti3 or
PrinterB.ti3 file containing the step wedge calibration chart
readings.<br>
<br>
<span style="font-weight: bold;">NOTE</span> that if you are
calibrating a raw printer driver, and there is considerable dot
gain, then you may want to use the <a href="targen.html#p">-p</a>
parameter to adjust the test chart point distribution to spread them
more evenly in perceptual space, giving more accurate control over
the calibration. Typically this will be a value greater than one for
a device that has dot gain, e.g. values of 1.5, 2.0 or 2.5 might be
good places to start. You can do a preliminary calibration and use
the verbose output of printcal to recommend a suitable value for <span
style="font-weight: bold;">-p</span>.<br>
<h4><a name="PC4"></a>Creating a printer calibration<br>
</h4>
The <a href="printcal.html">printcal</a> tool turns a calibration
chart <a href="File_Formats.html#.ti3">.ti3</a> file into a <a
href="File_Formats.html#.cal">.cal</a> file. It has three main
operating modes:- Initial calibration, Re-Calibration, and
Verification. (A fourth mode, "Imitation" is very like Initial
Calibration, but is used for establishing a calibration target that
a similar printer can attempt to imitate.)<br>
<br>
The distinction between Initial Calibration and Re-Calibration is
that in the initial calibration we establish the "aim points" or
response we want out of the printer after calibration. There are
three basic parameters to set this for each channel: Maximum level,
minimum level, and curve shape.<br>
<br>
By default the maximum level will be set using a heuristic which
attempts to pick the point when there is diminishing returns for
applying more colorant. This can be overridden using the <span
style="font-weight: bold;">-x# percent</span> option, where <span
style="font-weight: bold;">#</span> represents the choice of
channel this will be applied to. The parameter is the percentage of
device maximum. <br>
<br>
The minimum level defaults to 0, but can be overridden using the <span
style="font-weight: bold;">-n# deltaE</span> option. A minimum of
0 means that zero colorant will correspond to the natural media
color, but it may be desirable to set a non-pure media color using
calibration for the purposes of emulating some other media. The
parameter is in Delta E units.<br>
<br>
The curve shape defaults to being perceptually uniform, which means
that even steps of calibrated device value result in perceptually
even color steps. In some situations it may be desirable to alter
this curve (for instance when non color managed output needs to be
sent to the calibrated printer), and a simple curve shape target can
be set using the <span style="font-weight: bold;">-t# percent</span>
parameter. This affects the output value at 50% input value, and
represents the percentage of perceptual output. By default it is 50%
perceptual output for 50% device input.<br>
<br>
Once a device has been calibrated, it can be re-calibrated to the
same aim target.<br>
<br>
Verification uses a calibration test chart printed through the
calibration, and compares the achieved response to the aim target.<br>
<br>
The simplest possible way of creating the <span style="font-weight:
bold;">PrinterA.cal</span> file is:<br>
<br>
<a href="printcal.html">printcal</a> <a
href="printcal.html#i">-i</a> <a href="colprof.html#p2">PrinterA_c</a><br>
<br>
For more detailed information, you can add the <span
style="font-weight: bold;">-v</span> and <span
style="font-weight: bold;">-p</span> flags:<br>
<br>
<a href="printcal.html">printcal</a> <a
href="printcal.html#v">-v</a> <a href="printcal.html#p">-p</a> <a
href="printcal.html#i">-i</a> <a href="colprof.html#p2">PrinterB_c</a><br>
<br>
(You will need to select the plot window and hit a key to advance
past each plot).<br>
<br>
For re-calibration, the name of the previous calibration file will
need to be supplied, and a new calibration<br>
file will be created:<br>
<br>
<a href="printcal.html">printcal</a> <a
href="printcal.html#v">-v</a> <a href="printcal.html#p">-p</a> <a
href="printcal.html#r">-r</a> <a href="colprof.html#p1">PrinterB_c_old</a>
<a href="colprof.html#p2">PrinterB_c_new</a><br>
<br>
Various aim points are normally set automatically by <span
style="font-weight: bold;">printcal</span>, but these can be
overridden using the <a href="colprof.html#x">-x</a>, <a
href="colprof.html#n">-n</a> and <a href="colprof.html#t">-t</a>
options. e.g. say we wanted to set the maximum ink for Cyan to 80%
and Black to 95%, we might use:<br>
<br>
<a href="printcal.html">printcal</a> <a
href="printcal.html#v">-v</a> <a href="printcal.html#p">-p</a> <a
href="printcal.html#i">-i</a> <a href="colprof.html#x">-xc 80</a>
<a href="colprof.html#x">-xk 95</a> <a href="colprof.html#p2">PrinterB_c</a><br>
<br>
<a href="colprof.html#p2"></a>
<h4><a name="PC5"></a>Using a printer calibration</h4>
The resulting calibration curves can be used with the following
other Argyll tools:<br>
<br>
<a href="printtarg.html#K">printtarg</a>
To
apply
calibration
to
a
profile
test
chart,
and/or to have it included in .ti3 file.<br>
<a href="cctiff.html#p2">cctiff</a>
To
apply
color
management
and
calibration
to
an
image file.<br>
<a href="applycal.html#p1">applycal</a>
To incorporate calibration into an ICC profile.<br>
<a href="chartread.html#I">chartread</a>
To
override
the
calibration
assumed
when
reading
a
profile chart.<br>
<br>
<br>
In a workflow <span style="font-weight: bold;">with</span> native
calibration capability, the calibration curves would be used with
printarg during subsequent <span style="font-weight: bold;">profiling</span>
so that any ink limit calculations will reflect final device values,
while not otherwise using the calibration within the ICC workflow:<br>
<br>
<a href="printtarg.html">printtarg</a> <a
href="printtarg.html#v">-v</a> <a href="printtarg.html#i">-ii1</a>
<a href="printtarg.html#p">-pA4</a> <a href="printtarg.html#I">-I
PrinterA_c.cal</a> <a href="printtarg.html#p1">PrinterA</a><br>
<br>
This will cause the .ti2 and resulting .ti3 and ICC profiles to
contain the calibration curves, allowing all the tools to be able to
compute final device value ink limits. The calibration curves must
also of course be installed into the printer. The means to do this
is currently outside the scope of Argyll (ie. either the print
system needs to be able to understand Argyll CAL format files, or
some tool will be needed to convert Argyll CAL files into the
printer calibration format).<br>
<br>
<br>
In a workflow <span style="font-weight: bold;">without</span>
native calibration capability, the calibration curves would be used
with printarg to <span style="text-decoration: underline;">apply</span>
the calibration to the test patch samples during subsequent <span
style="font-weight: bold;">profiling</span>, as well as embedding
it in the resulting .ti3 to allow all the tools to be able to
compute final device value ink limits:<br>
<br>
<a href="printtarg.html">printtarg</a> <a
href="printtarg.html#v">-v</a> <a href="printtarg.html#i">-ii1</a>
<a href="printtarg.html#p">-pA4</a> <a href="printtarg.html#K">-K
PrinterA_c.cal</a> <a href="printtarg.html#p1">PrinterA</a><br>
<a href="cctiff.html#p4"></a><br>
To apply calibration to an ICC profile, so that a calibration
unaware CMM can be used:<br>
<br>
<a href="applycal.html">applycal</a> <a
href="applycal.html#p1">PrinterA.cal</a> <a
href="applycal.html#p2">PrinterA.icm</a> <a
href="applycal.html#p3">PrinterA_cal.icm</a><br>
<br>
To apply color management and calibration to a raster image:<br>
<br>
<a href="cctiff.html">cctiff</a> <a
href="cctiff.html#p1">Source2Destination.icm</a> <a
href="cctiff.html#p2">PrinterA_c.cal</a> <a href="cctiff.html#p3">infile.tif</a>
<a href="cctiff.html#p4">outfile.tif</a><br>
or<br>
<a href="cctiff.html">cctiff</a> <a
href="cctiff.html#p1">Source2Destination.icm</a> <a
href="cctiff.html#p2">PrinterA_c.cal</a> <a href="cctiff.html#p3">infile.jpg</a>
<a href="cctiff.html#p4">outfile.jpg</a><br>
<br>
<br>
Another useful tool is <a href="synthcal.html">synthcal</a>, that
allows creating linear or synthetic calibration files for disabling
calibration or testing.<br>
Similarly, <a href="fakeread.html">fakeread</a> also supports
applying calibration curves and embedding them in the resulting .ti3
file<br>
<h4><a name="PC6"></a>How profile ink limits are handled when
calibration is being used.</h4>
Even though the profiling process is carried out on top of the
linearized device, and the profiling is generally unaware of the
underlying non-linearized device values, an exception is made in the
calculation of ink limits during profiling. This is made possible by
including the calibration curves in the profile charts .ti2 and
subsequent .ti3 file and resulting ICC profile <span
style="font-weight: bold;">'targ'</span> text tag, by way of the <span
style="font-weight: bold;">printtarg</span> <span
style="font-weight: bold;">-I</span> or <span style="font-weight:
bold;">-K</span> options. This is done on the assumption that the
physical quantity of ink is what's important in setting the ink
limit, and that the underlying non-linearized device values
represent such a physical quantity.<br>
<br>
<br>
<hr size="2" width="100%">
<h3><a name="LP1"></a>Linking Profiles</h3>
Two device profiles can be linked together to create a device link
profile, than encapsulates a particular device to device transform.
Often this step is not necessary, as many systems and tools will
link two device profiles "on the fly", but creating a device link
profile gives you the option of using "smart CMM" techniques, such
as true gamut mapping, improved inverse transform accuracy, tailored
black generation and ink limiting.<br>
<br>
The overall process is to link the input space and output space
profiles using <a href="collink.html">collink</a>, creating a
device to device link profile. The device to device link profile can
then be used by cctiff (or other ICC device profile capable tools),
to color correct a raster files.<br>
<br>
Three examples will be given here, showing the three different modes
than <span style="font-weight: bold;">collink</span> supports.<br>
<br>
In <a href="collink.html#s">simple mode</a>, the two profiles are
linked together in a similar fashion to other <span
style="font-weight: bold;">CMMs</span> simply using the forward
and backwards color transforms defined by the profiles. Any gamut
mapping is determined by the content of the tables within the two
profiles, together with the particular intent chosen. Typically the
same intent will be used for both the source and destination
profile:<br>
<br>
<a href="collink.html">collink</a> <a href="collink.html#v">-v</a>
<a href="collink.html#q">-qm</a> <a href="collink.html#s">-s</a> <a
href="collink.html#si">-ip</a> <a href="collink.html#so">-op</a>
<a href="collink.html#p1">SouceProfile.icm</a> <a
href="collink.html#p2">DestinationProfile.icm</a> <a
href="collink.html#p3">Source2Destination.icm</a><br>
<br>
<br>
In <a href="collink.html#g">gamut mapping mode</a>, the
pre-computed intent mappings inside the profiles are not used, but
instead the gamut mapping between source and destination is tailored
to the specific gamuts of the two profiles, and the intent parameter
supplied to <span style="font-weight: bold;">collink</span>.
Additionally, source and destination viewing conditions should be
provided, to allow the color appearance space conversion to work as
intended. The colorimetric B2A table in the destination profile is
used, and this will determine any black generation and ink limiting:<br>
<br>
<a href="collink.html">collink</a> <a href="collink.html#v">-v</a>
<a href="collink.html#q">-qm</a> <a href="collink.html#g">-g</a> <a
href="collink.html#si">-ip</a> <a href="collink.html#c">-cmt</a>
<a href="collink.html#d">-dpp</a> <a href="collink.html#p1">MonitorSouceProfile.icm</a>
<a href="collink.html#p2">DestinationProfile.icm</a> <a
href="collink.html#p3">Source2Destination.icm</a><br>
<br>
<br>
In <a href="collink.html#G">inverse output table gamut mapping mode</a>,
the pre-computed intent mappings inside the profiles are not used,
but instead the gamut mapping between source and destination is
tailored to the specific gamuts of the two profiles, and the intent
parameter supplied to <span style="font-weight: bold;">collink</span>.
In addition, the B2A table is <span style="font-weight: bold;">not</span>
used in the destination profile, but the A2B table is instead
inverted, leading to improved transform accuracy, and in CMYK
devices, allowing the ink limiting and black generation parameters
to be set:<br>
<br>
For a CLUT table based RGB printer destination profile, the
following would be appropriate:<br>
<br>
<a href="collink.html">collink</a> <a href="collink.html#v">-v</a>
<a href="collink.html#q">-qm</a> <a href="collink.html#G">-G</a> <a
href="collink.html#si">-ip</a> <a href="collink.html#c">-cmt</a>
<a href="collink.html#d">-dpp</a> <a href="collink.html#p1">MonitorSouceProfile.icm</a>
<a href="collink.html#p2">RGBDestinationProfile.icm</a> <a
href="collink.html#p3">Source2Destination.icm</a><br>
<br>
For a CMYK profile, the total ink limit needs to be specified (a
typical value being 10% less than the value used in creating the
device test chart), and the type of black generation also needs to
be specified:<br>
<br>
<a href="collink.html">collink</a> <a href="collink.html#v">-v</a>
<a href="collink.html#q">-qm</a> <a href="collink.html#G">-G</a> <a
href="collink.html#si">-ip</a> <a href="collink.html#c">-cmt</a>
<a href="collink.html#d">-dpp</a> <a href="collink.html#l">-l250</a>
<a href="collink.html#k">-kr</a> <a href="collink.html#p1">MonitorSouceProfile.icm</a>
<a href="collink.html#p2">CMYKDestinationProfile.icm</a> <a
href="collink.html#p3">Source2Destination.icm</a><br>
<br>
Note that you should set the source (<a href="collink.html#c">-c</a>)
and destination (<a href="collink.html#d">-d</a>) viewing conditions
for the type of device the profile represents, and the conditions
under which it will be viewed.<br>
<br>
<h3><a name="LP2"></a>Soft Proofing Link</h3>
Often it is desirable to get an idea what a particular devices
output will look like using a different device. Typically this might
be trying to evaluate print output using a display. Often it is
sufficient to use an absolute or relative colorimetric transform
from the print device space to the display space, but while these
provide a colorimetric preview of the result, they do not take into
account the subjective appearance differences due to the different
device conditions. It can therefore be useful to create a soft proof
appearance transform using collink:<br>
<br>
<a href="collink.html">collink</a> <a href="collink.html#v">-v</a>
<a href="collink.html#q">-qm</a> <a href="collink.html#G">-G</a> <a
href="collink.html#si">-ila</a> <a href="collink.html#c">-cpp</a>
<a href="collink.html#d">-dmt</a> <a href="collink.html#l">-t250</a> <a
href="collink.html#k"></a><a href="collink.html#p1">CMYKDestinationProfile.icm</a>
<a href="collink.html#p2">MonitorProfile.icm</a> <a
href="collink.html#p3">SoftProof.icm</a><br>
<br>
We use the Luminance matched appearance intent, to preserve the
subjective apperance of the target device, which takes into account
the viewing conditions and assumes adaptation to the differences in
the luminence range, but otherwise not attempting to compress or
change the gamut.<br>
<hr size="2" width="100%"><br>
<h3><a name="TR1"></a>Transforming colorspaces of raster files</h3>
Although a device profile or device link profile may be useful with
other programs and systems, Argyll provides the tool <a
href="cctiff.html">cctiff</a> for directly applying a device to
device transform to a <a href="File_Formats.html#TIFF">TIFF</a>
or <a href="File_Formats.html#JPEG">JPEG</a> raster file. The
cctiff tool is capable of linking an arbitrary sequence of device
profiles, device links, abstract profiles and calibration curves.
Each device profile can be preceded by the <span
style="font-weight: bold;">-i</span> option to indicate the intent
that should be used. Both 8 and 16 bit per component files can be
handled, and up to 8 color channels. The color transform is
optimized to perform the overall transformation rapidly.<br>
<br>
If a device link is to be used, the following is a typical example:<br>
<br>
<a href="cctiff.html">cctiff</a> <a href="cctiff.html#p1">Source2Destination.icm</a>
<a href="cctiff.html#p3">infile.tif</a> <a href="cctiff.html#p4">outfile.tif</a><br>
or<br>
<a href="cctiff.html">cctiff</a> <a href="cctiff.html#p1">Source2Destination.icm</a>
<a href="cctiff.html#p3">infile.jpg</a> <a href="cctiff.html#p4">outfile.jpg</a><br>
<br>
<i><br>
</i>If a source and destination profile are to be used, the
following would be a typical example:<br>
<br>
<a href="cctiff.html"> cctiff</a> <a href="cctiff.html#i">-ip</a>
<a href="cctiff.html#p1i">SourceProfile.icm</a> <a
href="cctiff.html#i">-ip</a> <a href="cctiff.html#p1o">DestinationProfile.icm</a>
<a href="cctiff.html#p3">infile.tif</a> <a href="cctiff.html#p4">outfile.tif</a><br>
or<br>
<a href="cctiff.html"> cctiff</a> <a href="cctiff.html#i">-ip</a>
<a href="cctiff.html#p1i">SourceProfile.icm</a> <a
href="cctiff.html#i">-ip</a> <a href="cctiff.html#p1o">DestinationProfile.icm</a>
<a href="cctiff.html#p3">infile.jpg</a> <a href="cctiff.html#p4">outfile.jpg</a><br>
<br>
<br>
<hr size="2" width="100%"><br>
<br>
<br>
<br>
<br>
<br>
<br>
<br>
<br>
<br>
<br>
<br>
<br>
<br>
<br>
<br>
<br>
<br>
<br>
</body>
</html>
