Digital BW, The Print

I've printed several samples this morning to further explore the flat
95-100% print separation issue.  Consistent with what Roy noted, I've found
QTR Create ICC 3.2.1 does not always work with un-even increments.  On the
other hand, if there are 1% increments from 100 - 95%, then 5% to 75% and
then 10% steps on the rest of the scale, the print is about the same as if
there are 5% increments all the way.  In both cases there are steps in the
95 - 100% where the "lighter" input step is printed darker than the adjacent
"darker" input step.  On the other hand, they are all so close that this
"posterization" could be random printing artifacts and X-Rite reading
errors, and there is no visible separation or "posterization."  

Of more relevance, I'm not generating enough 95% - 100% separation with the
Create ICC approach to have visual separation in that region.  For example,
looking just at the case where the input text file has 5% steps over the
entire scale, the 100% - 95% separation or difference is Lab L 0.46.  This
is with the C86 and MIS EZ inks on EEM.  (PS CS2, Print with Preview, Let PS
Determine Colors, Perceptual, and Black Point Compensation on)  The driver
with no ICC loaded has a 100% - 95% separation of L 2.39, with every 1% step
being measurably separated.  For comparison, the 2400 with Eboni loaded, in
the ABW mode with neutral and normal tone settings (and no ICC loaded in the
PwP) has a 100% - 95% separation of L 2.30.

The gammas generated by these printing approaches is different, and that, of
course, would have some effect on the 100% - 95% separation.  Looking at the
50% density as a relative measure of gamma, the 2400 (EEM, ABW, neutral,
normal tone) and C86 EZ (Epson Color Controls gamma 2.2, Contrast -10) have
essentially the same L 58 50% density.  The C86 with the ICC has a 50%
density of L 54.

My main concern is that a visually un-separated 95 - 100% "wastes" limited
grayscale by essentially clipping the darkest 5%.  It is true, of course,
that with 16 bit files this is much more tolerable than is was with 8 bits
and a barely sufficient 256 total steps.

I have noticed that Epson drivers in printers aimed only at the color market
often if not typically clipped the darkest 5% also, as did some samples of
B&W output that were printed by Image Print on a 9600. 

Additionally, without a custom dot gain curve, my monitor (profiled with
SpyderPro 2) also has no visual 100% - 95% separation.  However, I'd thought
in an earlier thread many said that they do have such separation.  So, I'd
assumed my modestly priced CRT and video board were largely responsible for
the lack of deep shadow separation.  

On LCD displays I have on other computers in my house, still modestly
priced, the 95% - 100% steps are separated.  

In short, my "wasting grayscale" concerns aside, the ICC approach I'm
getting -- i.e., clipped darkest 5% -- matches my old CRT monitor quite
well.

On the other hand, the 2400 ABW and C86 prints (close to the old standards
I'd used from the Piezo days forward) match my LCDs better.  If matching the
monitor is the issue, I'm not sure matching an older CRT is a worthwhile
goal.  I suspect LCDs more like my newer, entry level Princeton displays are
more typical of the larger market I'm most interested in facilitating.

What kind of 100% - 95% print separation (Lab L differences) are others
seeing with the Create ICC approach?  Is it just my system?

Paul
www.PaulRoark.com