--- In DigitalBlackandWhiteThePrint@yahoogroups.com, "Paul
Roark" <paul.roark@v...> wrote:
Hi Paul,
> Roy,
>
> You wrote:
>
...
>
> >A guy trying my system with a 1280 had lots of the same
> >problems.
>
> One issue with hextone printers is the built-in midtone
cross-overs that
> have to be dealt with.
My gut feeling is that it "ought" to be a builtin freeby
not something to be "dealt with".
>
> >I did some Q-tip swab tests of the various inks
>
> The only way to be sure about densities is to use the printer.
Loading
> black carts and comparing the black-only ("BO") densities may
be the best
> way for the lighter inks. I also use "color ink" printing and my
"color
> test" file (like a quad purge pattern) for comparisons. However,
there is
> no way to isolate the light cyan and magenta except by putting it
in the
> black cart. (BO printing gives a darker 100% black than "color
ink"
> printing.)
>
> > I'm
> >beginning to think that the light inks are way too dark for
> >what the driver software is expecting. The swab tests give:
> >K=161, C=142, c=129, M=59, m=58, Y=63. These tests
> >aren't real accurate but illustrate the issue. Curiously the
> >two M's (the sepia toner) have almost the same density
>
> If the light magenta is the same as the full strength magenta,
there is a
> problem. The light ink is a 50% greater dilution of the
pigments.
>
> >but
> >vary in color quite a bit: M is a reddish orange, m is a much
> >browner and less color saturated.
>
> The M should be more saturated. The density readings are
probably the
> sampling method problem.
Maybe I'm using the wrong words, but the color is different.
It doesn't LOOK like a simple 50% dilution with clear. I'd
guess its a 50% dilution with light gray. Remember the guy
recently who had a bad 1280 vmsepia cart -- here's his jpg:
http://photolab.elmer.uaf.edu/tests/Purge6sepia.jpg
To me patch 6 which is m is a different "color" than patches
3 and 4. Obviously, these patches have a potential mixing
of inks which can make comparison hard but if the inks
were the same color the patches would be too. My swab test
shows these same differences but more pronounced.
>
> >I then tried printing out 21steps off only C and only M with
> >both the Epson driver and with the gimp-print driver. The
> >Epson driver is somewhat tolerant of these densities but
> >for cyan steps I get 100%=83 and 50%=59 and for magenta
> >I get 100%=59 and 50%=49 so most of the ramp up
> >happens in the first 50% and its pretty flat above that.
>
> Do you have Epson OEM ink comparative numbers?
No, I don't have much of any outside info. I know the q-tip
is very approximate. I'm just not really setup to do a lot
of ink testing -- and since I don't know what values to look
for I don't know what good I'd be doing.
>
> With the measuring method you're using, it's not clear what the
density
> differences should be. The 50% reading will be a combination
of the light
> and dark inks.
Aside from the test using a q-tip swab, everything I based
my comments on is using the 21step wedge with both
drivers and using the wedge with individual colors i.e. a
cyan wedge, a magenta wedge, and a yellow wedge. With
gimp-print I can just select one of the inks. In all cases I'm
comparing the 50% step with the 100% step (max) using a
densitometer. With the 4 ink printers, the 50% step is
approximately 40% of the max density. This also
matches pretty closely to what the final gray step wedges
should be 50%=64 and 100%=160. However if I use
the 6 ink printer with vms ink and Epson drivers, I get the
50% cyan to be about 70% of the max and the 50% magenta
as high as 80% of the max. But the yellow stays down
near 40%.
This is a different way of measuring and understanding
the light/dark transitions, but just looking at the output is
the way most of the profiling is done. I think this is very
much related to the recent discussions of linearizing the
output for inks before trying to do partitioning.
>
> The optimum relative densities will, no doubt, vary according to
the driver,
> but I assume Epson sets a de facto standard. The best way to
find the
> relative densities would be to put samples of Epson C & c, M &
m, in black
> carts and measure the BO densities.
>
> With the VM inkset, it was set up way before the system was
that
> sophisticated. With the quads I was working with, I only had
one ink per
> color. So, it was not an issue for me.
>
> When the vm system was taken to the hextone system, the
inks were based on
> an estimate of the 3rd party pigment relationships. It worked,
so mo more
> was done. Once curves are written, backward compatibility
rules.
I understand the history and I'm not in any way trying to get
you or anyone else to change anything. You've mentioned
about starting with a new slate for things in the future and
I just thought it would be useful for future reference.
>
> With my 7500, I found the 7000 driver and the vm relationships
worked very
> well.
>
> >With gimp-print things are quite a lot worse, its very
> >intolerant of wrong densities,
>
> Or "different" densities. I'm not sure what is "right" or "wrong"
with the
> various drivers.
>
> >... But,
> >seeing how I have the source code I was able to reset the
> >assumed densities in the software.
>
> Can you tell from the software what the assumed densities
were?
All I mean by right or wrong is whether the actual ink differences
match what the software thinks is the ink difference.
The gimp-print software has weightings for the different
inks and dotsizes. I assume all the calibration was done
with OEM inks but I really don't know how extensive or
accurate it was done. With that caveat, these are the values
it came with: light cyan = .27 and light magenta = .35
From what I can figure out , this means c is .27 as dark as
C -- or in other words nearly 4 dots of c is about as dark as
1 dot of C. I don't know why magenta is different. To get
a smooth 21 step I changed light cyan to .56 so only
2 dots of c = 1 dot of C. These two numbers are used for
all the non-variable dotsize 6 ink printers -- 7500 being one
of them.
>
> > After this I was able to
> >get great step wedges -- smooth from 0 to 100%.
>
> >So I don't where this should all lead. Changing ink formulas
> >would obviously be a big hassle for all the existing curves.
> >I'm thinking of remixing a lighter light cyan for my next load
> >of the cartridges but I'm doing my own curves.
>
> For the dual-quad, vm-s system I made for the 7500, I started
with a clean
> piece of paper. Unfortunately, the densities I chose appeared
to be part of
> the problems I ran into with the inkset. I might start with a
clean piece
> of paper if I migrate to a different driver (RIP or whatever).
>
> For now, however, I'm going to work around the existing vm de
facto
> standard. It works OK for at least most systems.
>
> Paul
> http://www.PaulRoark.com
Like I said I just thought you might find this useful or maybe
not.
Roy
http://www.harrington.com