[Gimp-print-devel] More high end stuff

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

I found a few things going wrong that led to excessive grain on the
870:

1) The code to decide what dot size to use was looking at the density
   rather than just the color to be printed.  This is probably good at
   very high densities, but it's not so good in regions where there's
   a small amount of one color (read: cyan) in a field of something
   else (read: yellow).  The upshot was that large cyan dots (and
   possibly even small dark cyan dots) were being printed in magenta
   or yellow with even a slight mix of cyan.

2) Similarly (but not identically), the black code was just making a
   binary on/off decision, and either printed the full value of the
   black or nothing.  Again, that led to large black dots.  I've
   changed that to simply slide between k_lower and k_upper, so that
   small amounts of black ink are printed as small dots rather than
   fewer large dots.

3) The adaptive dithers were omitting all ink in some regions that
   they should have been printing.  This is quite visible if you know
   what to look for.

4) The k_lower value was too high in variable dot size printing, now
   that I fixed (2).

5) The black equivalent values in the Epson code were too high (1.5).
   That resulted in too little color ink being printed, which resulted
   in grainy dark grays.  Setting it to 1.0 helps, although it doesn't
   entirely solve the problem.

6) I eliminated use of the largest dot size of the light ink.  This is
   a bit tricky.  It was causing problems (fairly sharp transitions)
   in darker regions of mixed colors, because the transition band
   between light and dark inks was too narrow.  I experimented with a
   few things, and this seems to be the best compromise.

The yellow cast seems to be entirely gone.  There might be a VERY
slight blue or cyan cast to the grays right now, less than one point.

I haven't hooked up the EX yet to try it.  Nor, for that matter, have
I tried printing a real image yet, just my test shot.  I'm going to
push the PDI target through overnight (doubtless at the cost of a few
dollars worth of ink) at 1440x720 highest quality (which seems to want
me to set the density to 1.7 rather than 1.5, probably because of
imperfect head alignment -- good ghod, this can be an expensive hobby)
to see what happens.

Yawn.

Jean-Marc, my apologies.  I know I promised I wouldn't touch this for
a while, but I couldn't help myself...

-- 
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