Hi Howard,
That's a worthwhile set of tests to try and understand color management.
The results are all reasonable and tell you something about your setup.
I think if you can understand how these cases come about you'll have a
much better feeling for what color management is all about.
I conclude that there is something special about the QTR-Gray Lab
profile as it appears to be the only grayscale profile which does not
change the K values in Photoshop when files are converted to it.
The reason for this is that in Color Settings you have QTR-Gray Lab as your
default
gray working space. What this means is that any untagged file will be
treated as Gray Lab.
So untagged, assign Gray Lab an convert-to Gray Lab are all identical --
both K values
and the corresponding L values don't change. Screen is the same, prints are
the same.
But when you used GG 2.2 you get two different effects. Assign GG2.2 leaves
K's the
same but L's will change (the GG2.2 is a new "meaning" for the K values).
Convert-to GG2.2
changes the K values but preserves the L values. The Screen and Prints work
differently
though. The screen is ALWAYS handled with color management so effectively
it always
shows the L values of a file. Convert preserved the L values so the screen
stayed the
same, assign changed them so the screen changed.
Printing on the other hand works a little different as you've seen. The
Print depends on whether
you used color management in printing (i.e. the print profile). You
aren't: so effectively you
are printing the K values regardless of the L value meaning. The Assigned
version with
the unchanged K values printed the same whereas the converted version with
changed
K values printed differently.
Roy
On Dec 12, 2007 2:55 PM, Howard Shaw <glassman@...> wrote:
> Ernst Dinkla wrote:
> > ...
> > Where would the same 21 steps of the wedge land in print
> > density if that wedge goes through the profile first ?
> >
> I did a little testing of this tonight. I printed 5 21-step wedge files
> - untagged, QTR - Gray LAB assigned, QTR - Gray LAB converted, GG2.2
> assigned and GG2.2 converted. (I chose QTR - Gray LAB as I habitually
> edit & print my images with this profile.) They were printed using a
> linearised QTR curve for a quadtone MIS ink set up.
>
> In 4 out of 5 cases, the exception being the file converted to GG2.2,
> both the K values (0,5..100) in Photoshop and the density measurements
> remained constant and they were all perfectly linearised according to
> the Density Tables in the QTR documentation.
>
> Converting the file to GG2.2 actually changed the K values in Photoshop
> (although there was no visible difference) as follows:
>
> 0 0
> 5 6
> 10 12
> 15 18
> 20 23
> 25 28
> 30 34
> 35 39
> 40 44
> 45 49
> 50 54
> 55 59
> 60 63
> 65 67
> 70 72
> 75 76
> 80 80
> 85 84
> 90 87
> 95 91
> 100 100
>
> The resulting printed step wedge was not linear but appeared to reflect
> the values as shown above - ie. a markedly large gap between 95 & 100%
> and all the steps below 75% were darker than the equivalent untagged
> file. I would think (although I didn't test it) that manually changing
> the steps back to 0,5..100 etc in photoshop would have corrected the
> linearisation although they would then of course have been visually
> different on screen to the steps in the untagged file.
>
> I conclude that there is something special about the QTR-Gray Lab
> profile as it appears to be the only grayscale profile which does not
> change the K values in Photoshop when files are converted to it.
>
> To create the assigned profile files I opened the untagged file in
> Photoshop CS3 and then, when prompted opted to assign the appropriate
> profile. For the converted files I opened the untagged file, opted to
> 'Leave as is' and then clicked Edit->Convert to Profile, selected the
> profile, Perceptual Intent with both BPC & Dither ticked.
>
> My method of checking the linearisations was to deduct paper white
> (0.04) from all the readings such that a measured DMax of 1.72 became an
> adjusted 1.68 and then comapred the readings to the 'ideal' density
> tables for DMax of 1.68. My suspicion is that this part of my
> methodology is flawed and that a better way of doing this is suggested
> by your 85%/last 15% comment.
>
> I'm not sure if this was a worthwhile exercise but at least I know that
> for my own workflow, what I'm seeing on screen, the numbers in the file
> and the resulting printed densities are all as I would expect.
>
> I would welcome any comments or criticisms of these results and the
> methods used to obtain them. With apologies for dragging the thread in a
> different direction and thanks to Ernst & Roy for their illuminating
> comments.
>
> Howard
>
> > There could be density tables made for several
> > Dmax-Paperwhite ranges that way. Maybe just down to earth
> > density ranges of certain papers and the Dmax possible on
> > them with a given inkset as obtained in practice but I'm
> > sure there are ways to create them more mathematically.
> >
> > The linearised tables are nice for control when there's
> > control on printer linearisation or one has to find out what
> > setup has the best linearised output if there's no control.
> > Good to have control at say 85% of the job to get a printer
> > in line but why not add what that last 15% does. I too have
> > some difficulty in the way that perceptual tone separation
> > is expressed in densities but if it works for the tables of
> > a linearised tone range I do not see that it is worse for
> > what shall we call it .......?
> > It may be more useful for some to have also tables that show
> > the output after all steps as in many cases there's a black
> > box in between, not in QTR of course.
> >
>
>
>
>
>
> Yahoo! Groups Links
>
>
>
>
[Non-text portions of this message have been removed]