[argyllcms] Re: hplip printer driver and color management
- From: "Hal V. Engel" <hvengel@xxxxxxxxxxx>
- To: argyllcms@xxxxxxxxxxxxx
- Date: Mon, 8 Sep 2008 17:58:16 -0700
On Monday 08 September 2008 04:33:30 pm Graeme Gill wrote:
> Hal V. Engel wrote:
> > I think the main consideration is that the total ink limit used for
> > creating your targets and profiles should be slightly higher than the
> > total ink limit setting used in gutenprint.
>
> Hmm. I would have said the opposite!
>
> Generally there is no hard limit for ink levels in a printer, but
> at some point there will start to be negative effects such as
> clogging up of screen patterns, excessive drying time, smudging,
> or even ink running off the page. For Xerographic printers, there
> can be more catastrophic results such as paper or mechanism jams,
> excessive toner floating about and blocking sensors etc., but there
> is still a progression from "not as black/vivid as it could be" to
> "too much!".
>
> But my experience is that the device behavior being fed into the
> profile software should have a margin over the final chosen ink limit,
> otherwise the device behavior isn't being characterized sufficiently
> for accurate results, and is in fact being extrapolated. Usually
> the extrapolation isn't that good. [ You can explore this by creating
> a test chart with a low ink limit, and then looking at the resulting
> profile a values beyond that limit. ]
>
> So ideally any ink limit being imposed by the print driver, or by the
> test chart should exceed the final chosen ink limit imposed by
> colprof or collink, and my rule of thumb is 10%, but it should
> really be as large as is practical.
OK it appears that I didn't correctly understand the documentation for targen.
Thank you for correcting me. If I understand this it would indicate that the
ink limit set in gutenprint and targen should be the same or nearly so and
should be high enough that it is just to the point where there are signs of
over inking. And that I should use a slightly lower limit in colprof to pull
the actual limits back below the point of over inking or about 10% below the
setting used in gutenprint and targen.
>
> In a fully flexible print driver the settings would be approached in
> stages. At the first level would be the per channel upper limit and gross
> linearisation, making sure that the limit errs on the side of not limiting
> the gamut.
In gutenprint you would use the density adjustments to change the amount of
ink applied on a per channel basis (IE. to change the "per channel upper
limit"). I found that I could arrive at a good setting for density by
printing and measuring a ramp of darker tone patches for each channel. Then I
increased the density until I saw that L* for the darkest patch had either
stopped increasing or had become nearly the same as the next lighter patch
(the Y channel did both of these at the same point on my printer) or there
were signs of over inking (all the other channels did this on my printer).
For CMY this was only about 5% more than the default setting but for black it
was about 90% greater. My goal was to get as much gamut and dynamic range as
possible without being too concerned about the overall balance of the channels
since this would be controlled by the profile.
Again there are currently no tools for linearization in gutenprint. So
setting the "per channel upper limit" is about all you can do with this part.
>
> At the second level there would be a per channel calibration and finely
> tuned per channel upper limit.
These are currently difficult to do in gutenprint. The next stuff is targen
and colprof related.
>
> At the third level there would be characterizing the device, and the
> total ink limit chosen for the test chart and/or print driver would
> be chosen such that grossly bad effects such as ink run/jams are
> avoided, but erring on the side of not limiting the possible
> gamut.
>
> At the final level would be the finely chosen total limit imposed in
> the profile creation (B2A table/device link table).
>
> > ink limit (as set in gutenprint). In other words the gutenprint ink
> > limts are to prevent the printer from using too much ink and the ink
> > limits used in targen and colprof are to create profiles that use the
> > full range available for the printer. Graeme is that correct?
>
> I wouldn't have put it quite that way. "Using too much ink" is not
> a physical limit, but is a user choice, and so I wouldn't say
> that this is the reason for setting a gutenprint limit. Instead,
> a print driver limit should really be reflecting the practical
> limits of the device and protecting it from harm. Preference
> type choices such as running costs, screen clogging, drying time
> or achieved gamut should really be set at the point of maximum
> flexibility, which is after characterization and during profile/link
> creation.
>
> Graeme Gill.
I guess I could have been clearer since by "Using too much ink" I was
referening to things like over inking (bronzing and running ink). In other
words I was talking about the physical limits of the device and paper not
things like operating costs. Again I should have been clearer.
In a private email to me Robert Krawitz suggested that dark redish brown
patches were the most sensitive to over inking for most ink jet printers. I
did confirmed that this was the case for my printer. So you likely don't
need to look at a broad range of patches to see if you need to cut back on the
ink limit since the dark redish brown patches will show signs of bronzing
before most other patches will.
Hal
Other related posts:
