Digital BW, The Print

Jeff,

You wrote:

>Can someone explain why the "visual densities" of the original
>quadtone ink set (as printed on paper) are bunched at the dark end of
>the grayscale (4-45, 4-50, 4-75 are all very dark)?  Seems to me that
>if the visual densities were more evenly spread out over the grayscale
>spectrum that we could produce more evenly and smoothly transitioned
>RGB partitions that use all the inks.  ...

To judge ink densities, I print the 21-step test file using what I call my
"Color Test" curve.  (The driver settings I use for this include Matte HW
paper type and No Color Adjustment.) This basically pumps out 100% of each
ink at a different part of the test strip.  Then I scan the test strip and
do a levels on it.  So, black is 100% and white is 0% when measured with the
eyedropper in Photoshop (if the image is in grayscale mode).

Scanner settings, etc. can affect results, so I use the tool just to compare
inks -- and mix new brews.

Piezo gives results as follows: K = 100, C = 84, M = 38, Y = 27%.

When I worked up the MIS VM inkset, I didn't just clone Piezo, even though I
was already using the Piezo variable-tone formula that just accepted the
Piezo densities.  Rather, I tried a number of density combinations and found
that, under the loupe, as the middle ink got closer to the black ink, the
dots at about 30% on the 21-step test file print increased.  As the middle
gray got closer to the light gray, the dots at about 70% increased.  The
optimum density that minimized the dots was so close to Piezo, that I ended
up right where Piezo is.  That is, for a 3 ink mix, the 100, 84, 38% spread
was about optimum.  Since I was interested in making a standard that would
have competitive sources of ink readily available, it also made sense to
have the MIS VM inkset as close as possible to the Piezo and variable-Piezo
that I was using.  (Since the 38% give dotless highlights on the 1160, the
lighter ink was simply irrelevant -- that position became the toner position
on quad printers.)

The MIS standard quads are as follows on my Color Test:
K = 100, "75" = 96, "50" = 79, "25" = 24%.

I don't know the reasons or history of why MIS chose this density
distribution.  In my view the gap between 24% and 79% is too much for my
purposes.  On the other hand, I expect it produces a more linear g/s density
distribution than the Piezo/MIS VM & FS distribution when it is simply
printed through the Epson driver with no curves or other quad workflow.  It
no doubt also produces dotless highlights in even a 3000 when a partitioning
workflow is used.

Paul
http://www.PaulRoark.com