[argyllcms] Re: New user questions on display calibration

  • From: Graeme Gill <graeme@xxxxxxxxxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx
  • Date: Tue, 18 Aug 2009 17:45:43 +1000

Eric Fishwick, CLSA wrote:

-         Is my understanding of what dispcal does correct, if so how
does the icc profile add to this

dispcal creates calibration curves that are used to change the behavior
of the display. profiling characterizes the resulting behavior. The
calibration is typically embedded within the profile for convenience,
so that the display will be put in the correct state to match the profile.
See <http://www.argyllcms.com/doc/calvschar.htm>

-         I used the -as switch (mainly because www.
<http://www.marcelpatek.com> marcelpatek.com gave it as example) to
generate a "shaper+matrix" profile. How does this differ from any other

It's rather technical, but it selects a particular profile model.
Different models make different trade offs. A shaper/matrix model
will be smoother and more exactly invertible than a CLUT based
model, but will only accurately model a device that behaves
additively (ie. the primary color channels don't interact).
A CLUT base profile is more flexible in accommodating arbitrary
device behavior, but being table based, adds approximation type

-         The Argyll docs imply that the .cal file (from the
calibration) is embedded in the .icm file and is redundant, am I reading
this correctly?

Yes. It usually gets placed in the 'vcgt' tag.

-         For non-colour matched apps and the desktop am I right in
thinking that only the calibration part applies ie I now have gamma and
grey scale tracking correct but other colours might be off...


-         And that for icc aware apps all colours should be corrected
for the inaccuracies of the display (but corrected to what reference - I
shoot raw?)

Device color profiles allow transforming between colorspaces. To do this,
two device profiles are needed, so the "reference" is the source profile
used. This will usually be the profile the images are tagged with, or
colorspace they are assumed to be in.

-         Is Lightroom 2 such an icc aware application (I assume yes but
can't find this anywhere in the LR help files).

Sorry, I don't have detailed knowledge of Lightroom.

One last question when I get around to profiling the printer.
Colprof has an option to include a source icc, it has 2 switches and the
help states:

 -s src.icc <file:///F:/Argyll_V1.0.4/doc/colprof.html#s#s>    Apply
gamut mapping to perceptual B2A table for given source
 -S src.icc <file:///F:/Argyll_V1.0.4/doc/colprof.html#S#S>    Apply
gamut mapping to perceptual and saturation B2A table

Trouble is, I haven't a clue what this means. Which if any is
appropriate? Or is this irrelevant and I should just calibrate the
printer without a source icc profile. The printer is a Canon i9950
inkjet, my understanding is that its gamut is greater than the screens.

The source profile provided is used to setup the gamut mapping
incorporated in the profiles perceptual and saturation B2A
tables (Profile Connection Space to device tables).

Without a definition of the source gamut, the gamut mapping can't
be formulated. So the answer is for you to provide a source profile
that is the one you will be using when you make use of the
printer profile as a destination, or one that is similar to,
or typical of the source profile you will be using.

For instance, if the images you wish to print are in (say)
sRGB colorspace, then you would provide the sRGB profile as
the parameter to the -s or -S options.

My purpose in this is to have what I see on screen in Lightroom (hich
doesn;t soft proof) as close to what I get printed as possible.

It's never really possible, since the devices have different gamuts
and viewing conditions. The point though is to end up with the
best possible reproduction, and the intent is a way of instructing
the system what you mean by "best".

Graeme Gill.

Other related posts: