[Gimp-print-devel] Epson stuff

[Robert L K. <rl...@al...>]

I fixed a major bug in the vertical positioning stuff.  Everyone who
prints in color should grab this one.

I also retuned the ink constants for the 870.  I really think that a
ratio of .25 for light/dark works much better (the transition is less
abrupt), and I think the dot sizes are also a bit smaller than we had
them set for before (I changed them to 1.0/.667/.5).  It also got rid
of most of the yellow cast, which suggested that the cyan and magenta
had been too light before (i. e. the light inks were getting credit
for more cyan/magenta than they were really supplying).

Also, I bumped down the k_lower setting for variable dot size
printers.  I figure it's OK to use black at lower levels, since the
dots are smaller, and the results seem to bear this out.

The settings I'm now using:

Print mode: 1440x720 softweave
Dither mode: random Floyd-Steinberg
Image type: photo
Gamma: 1.2
Density: 1.5
Saturation: 1.1
Red: 101
Blue: 101

If anything, the remaining yellow cast in the light-medium grays
suggests that this may not be quite far enough.

1440x720 softweave works much better than 1440x720 enhanced.  That
surprises me.  What's also surprising is that 720 high quality has
some gross flaws.

I'm seeing slight, but noticeable, banding, at 5 lines/mm (120
lines/in., or 6 vertical spaces), with 1440x720 softweave.  What's
more interesting is that there is very slight vertical banding visible
under an 8x loupe, at about 15 lines/mm (my loupe has a scale), or 400
lines/in (close to 360 dpi).  This is much more noticeable (and
coarser) at 720 high quality.  I don't know what's happening here.
This too goes away at 1440 enhanced.

Finally, there's a noticeable red cast in some dark grays (9-11 on the
gray scale target in the PDI test image).  Perhaps the correct
light/dark ratio is different for for cyan and magenta.  Perhaps it's
in part due to the slight red and blue adjustment I made, and the
elevated saturation.

It's a bit too bright right now; the dimples in the golf balls are
somewhat blown out.

It probably goes without saying, but this effort makes the EX look
crude by comparison.  It won't stand up to inspection by a loupe, but
with few exceptions it will stand up to visual inspection.

I printed the image at 480 dpi, which was the most I could do on a
piece of letter size paper.  720 dpi would probably be a better choice
for this kind of test.

On another note, I've taken a quick look at the rinkj code, but I
haven't done anything with it yet.    The abstractions are quite
different from those used by gimp-print, and there's going to be a lot
of fiddling required to make it all play together.  Currently, rinkj
doesn't support variable dot size printers and 6-color inks, which
we'll ultimately need to do.

-- 
Robert Krawitz <rl...@al...>      http://www.tiac.net/users/rlk/

Tall Clubs International  --  http://www.tall.org/ or 1-888-IM-TALL-2
Member of the League for Programming Freedom -- mail lp...@uu...
Project lead for The Gimp Print --  http://gimp-print.sourceforge.net

"Linux doesn't dictate how I work, I dictate how Linux works."
--Eric Crampton