Digital BW, The Print

I've been experimenting with QTR recently, using the generic K3 setup in a
220 (K, LK, LLK, LC, LM, & Y - which is working great and probably will in
any Epson hextone or better).  I was making a "neutral" profile for Premier
Art Matte BW.  Just when I thought I had a very smooth tonal distribution
with my current standard 21-step test file, I tried the 51-step file that
used to be distributed with the QTR package.  The tones were suddenly much
more inconsistent, with sudden jumps of up to 0.7 lab unit in Lab A & B.
They were not only inconsistent internally with the 51-step test strip, but
also with the previous 21-step test strip.  This, of course, got me
wondering what was going on and how consistent my profiling approach was.

 

The bad news is that with the same paper batch, the dmax was varying from
1.71 to 1.56, and that Lab A & B readings in the midtones on the same piece
of paper printed and read at the same time, even after overnight dry down,
can vary by 0.5 units.

 

The good news is that on the same piece of paper, the Lab L values were
close enough for consistent linearization as long as the paper was
thoroughly dry and a test strip with the 100% black patch at the left side
of the paper was avoided.   The test strip I've been using recently (see
jpeg version at
http://home1.gte.net/res09aij/21-Step_0-100_L16_Un-Tagged.jpg -- save this
as a tiff and it's very useable) avoids most of the dmax inconsistency
problems by having the black end of the scale on the right.  The most
consistent and highest dmaxs are with the test strip printed vertically on
the paper and away from either edge.

 

While the ColorVision PFP spectro has some inconsistencies, over the course
of the experiments I was recently doing they appeared to be small relative
to what I'm talking about.  Multiple readings of the same test strips were
relatively consistent.  

 

The apparent variables that affected mostly the Lab A & B values during my
recent experiments appeared to include where the test patches were on the
paper and the order of the patches - what was printed before the patch being
read.  

 

The most bazaar patterns were seen with the old QTR 51-step test strip,
which has 17 columns with 3 rows.  The order of the patches is from light to
dark down each column, then over to the next row.  When printed vertically
there was a very obvious saw-tooth pattern of 3 readings of the Lab A & B
values going against the larger, gentle change in the values.  When the
51-step test strip was printed horizontally, the adjacent values appeared to
smooth out, except for an abrupt change in one part of the paper. See
http://www.paulroark.com/BW-Info/51-Step-Lab-A-B.jpg 

 

It made me think that the paper coating might have "vertical stripes" or
inconsistencies in the coating that account for some of what I was seeing.
As such, the most consistent test strips may be those that are printed
vertically.  One could print multiple ones in different parts of the paper
and average them to reduce the un-wanted noise, but it's questionable if
it's needed or worth the time (not).

 

The bottom line is that I'll probably ignore the 51-step test strip (which
Roy no longer appears to include in the QTR package for good reasons), not
use test strips with the black patch on the left edge, and print my final
test strip for linearization vertically on the paper.  Additionally, I'm not
going to worry much about 0.5 changes in Lab A & B values.  If the curves
are smooth and generally the shape I want, many of the apparently abrupt
changes in the spectral A & B readings may be artifacts that have nothing to
do with the profiling curves.

 

Paul

www.PaulRoark.com <http://www.paulroark.com/>  

 



[Non-text portions of this message have been removed]