I have printed a CUPS test page with the DeskJet 840C CUPS driver of
GIMP-Print 4.1.4 into a file and then "pcl-unprint"ed it. I have used
the settings 600x600 dpi, "Solid Tones", "Adaptive Hybrid". In the
resulting image all was OK, no light zones at the border. So these light
zones/missing pixels can only be produced by the printer hardware. So it
is proven now, that the printer firmware does an "ink reduction" and one
must find out how to turn it off, suppress it, or work around it.
300x300 dpi with the HP DeskJet 840C driver (also "Solid Tones",
"Adaptive Hybrid") gives the big barcode which I told of in an earlier
mail also when "pcl-unprint"ing an appropriate output file. So the big
barcodes of 600x300 dpi and 300x300 dpi are really a bug of the CUPS
driver of GIMP-Print 4.1.4. With GIMP-Print 4.0.4 I didn't have any
problems with the low resolutions. You can also do further tests on this
problem because "pcl-unprint"ing does not require the appropriate
Now I did a test to work around the "ink reduction" of the HP DeskJet
840C. I am still using the CUPS driver of GIMP-Print 4.1.4, 600x600 dpi
(the only resolution which works), "Solid Tones" (to keep all symmetric,
there is no GIMP-Print-internal calibration), "Fast" dithering (the
strongest "ink reduction" effect). I have done a colour calibration with
"cups-calibrate". The first step of the calibration shows three rows of
16 squares each, one row bl;ack, second yellow, and third red. The 16
squares are intended to be a linear intensity ramp from 25 % intensity
on the right to 100 % intensity on the left in steps of 5 %. On my
calibration printout all from 25 % to 75 % is OK, it looks like a linear
ramp, but due to the "ink reduction" the squares for 80 % have more or
less the colour intensity of the 40-% squares., the squares for 90 %
have the intensity of the 45-% squares, but the squares for 100 % look
OK. They show a really black black and also intensive red and yellow,
they all three are more intensive than the squares for 75 % (typed: A).
So the "ink reduction" seems to reduce by 50 % and not to act on colours
with 100 % of intensity. It seems to take place on an intensity range
starting from an intensity value between 76 % and 80 % and stopping at
99 %. I don't know for what such an ink reduction can be good for.
To try to get a printout without "ink reduction" effects I simply chose
75 % for the intensities of black, yellow, and red. In the second step I
chose a gamma of 1.3 (typed: 12) and in the last +10 (typed: 5) for red,
0 (typed: 7) for green, and +12.5 (typed 54) for blue. This gave the
profile line for the PPD file
*cupsColorProfile -/-: "0.750 1.300 1.000 0.000 0.000 -0.250 1.000
0.000 0.000 -0.200 1.000"
which I simply saved (puts the profile into /etc/cups/lpoptions). Now I
printed the CUPS test page from the command line (not with "Test
printer" in KUPS, because this ignores /etc/cups/lpoptions) and it came
out without any light borders in the colour wheel, but with a somewhat
gray black (because I reduced the intensity of all inks and so when on
the dithering 100 % intensity comes out the calibration makes 75 % out
of it). The Mandrake logo looks very nice after the calibration, too.
Now I tried to print a photo by sending the attached JPG file directly
to the CUPS queue of the calibrated DeskJet 840C, but in the dark area
on the lower left there are still "ink reduction" effects. This gave me
the impression that the colour calibration perhaps does not work on JPG,
but only an PostScript files. So I loaded the photo with GIMP and set up
the printing to PostScript level 2 on the CUPS queue of the calibrated
printer (no "-o raw" in printing command to use the calibrated CUPS
driver). This gave a too light output (probably because all intensities
are reduced to 75 %), but it does not show any "ink reduction" effect.
I hope this information helps to find a way to make the output of HP
Till Kamppeter wrote:
> Perhaps it is really some kind of ink reduction by the firmware of the
> printer, because all which I described in my e-mail happens only on HP
> and not on Epson, Canon, or Lexmark. The gray of the text I mentioned is
> not really gray, it is very grainy, so it is a black with missing dots.
> So it is very evident that HP does an ink reduction in the firmware. To
> check whether GIMP-Print is OK or not, I will print into a file and look
> with pcl-unprint on it. So the problem is how to adapt GIMP-Print to
> take this into account. When density adjustment does not solve it, one
> should perhaps find out which models do ink-reduction and rate all of
> them as "Partially" working on http://www.linuxprinting.org until HP publishes
> its planned drivers for them. For now I will recommend to the people to
> stay away from non-PostScript HP inkjets. The best choice is Epson and
> the Lexmark Z52 is also much better than any HP.
> Robert L Krawitz wrote:
> > From: Till Kamppeter <till.kamppeter@...>
> > Date: Sun, 18 Feb 2001 03:30:56 +0100
> > The missing dot phenomena which is reported here for the HP DeskJet
> > 350C seems to be the same as what I called "overflow effects" on my
> > HP DeskJet 840C. Ordered dithering does not solve the problem.
> > I'm beginning to doubt this theory that it's due to overflow in the
> > dithering code; I've seen no other signs of overflow there, and I've
> > printed enough at 100% density (using low resolution modes with the
> > Epson driver) that I don't think there's a problem there. I don't
> > know if you've tried adjusting the density (either up or down) to see
> > if it improves (or even changes) anything, but it's worth trying.
> > Grant made some comment that an HP person told him that the printer
> > actually does ink reduction itself. If that's the case, it's possible
> > that it interacts badly with our dithering algorithms; maybe it's been
> > designed to work well with the dithering that HP's driver does and
> > they didn't bother generalizing it. That's just a guess; I've never
> > seen any of the output in question.
> > I notice the issue of pale blacks; apparently it needs a higher black
> > density than color ink density. We've discussed this before.
> > All problems of the transition between pure and mixed black seem to be
> > solved. Visible problems with black in photos are due to the dithering
> > problems in the high intensities of black.
> > Well waitaminnit. You just said that black prints gray...that doesn't
> > sound solved???