[argyllcms] Re: Of ink limiting and maximizing gamut
- From: Devadas Varma <devadas@xxxxxxxxxx>
- To: <argyllcms@xxxxxxxxxxxxx>
- Date: Wed, 28 Jan 2009 20:50:38 -0800
I have been using argyll for many many years, but never posted to this
mailing list (always read it), so this is my first one:)
I use StudioPrint Rip and the fact that it has a huge amount of flexibility
caused me to spend (waste) a lot of paper on linearization. Finally I think
I found some thing that works for me (at least with StudioPrint, but maybe
applicable to other situations also).
First, I tried density based linearization (the only one supported by
StudioPrint version that I have). On papers that have relatively low ink
limits (Crane Museo or Kodak Professional Luster), this caused problems,
because the density did not reverse and the ink limit was very high. By the
way, I use Epson 4000, but I don't use OEM inks, I use MIS K4 inks - which
are actually for Epson 4800!
Then I used Chroma based ink limiting (c in Lch or a combination of a and b
in L*a*b), and this worked much better. To my surprise, the ink limits were
lower than I was used to in Density based setup. But this was still high ink
limits. I kept the ink limits fairly high (but lower than density based,
especially Cyan did reverse chroma).
Then I tried limiting Chroma to numbers suggested by ColrBurst RIP (which
uses Chroma based limiting, for eaxmple for Epson K3 inks, they suggest for
premium Luster, C=1.8 Chroma=70, M=1.75 Chroma=80, Y=1.1 Chroma=100, K=1.8
Chroma=85 - it is interesting to note that these were not necessarily peak
chromas especially for Yellow and left reasonable amount of headroom), and
also later based on values recommended for CMYK SWOP in a document called
the G7 method (that someone posted to this mailing list). These values
resulted in much lower density levels (C=1.45, M=1.45, Y=1.0,K = 1.7).
First, I thought this will give me much lower Dmax and Gamut, but it was
only marginally lower, and much cleaner (less muddy shadows, clean
separation of 98% and 100%, better overall linearization). I eventually
ended up using slightly higher K density values,but still much lower than I
did with earlier density based ink limiting. To my surprise, I actually
found that Dmax was actually higher. I don't know if this is specific to
StudioPrint (I use the Prefer Small Dots rastering and also set the ink
limits carefully for a smooth crossover on my Epson 4000 from light inks).
Anyway, after much experimentation, I did find something that work for me,
at least for the papers I was having trouble with (Crane Museo and Kodak
Luster, EpsonPGPP was much easier), and I didn't sacrifice much gamut if any
at all, and the paper appeared dry right after printing,with no ink overflow
etc.
This is what my workflow for profiling a new or difficult paper is now:
1. Use the standard StudioPrint ink limiting method. But I check for any
chroma saturation or reversal. As a target, I set the density limits as low
as above mentioned values, and increase it if the linearization curve looks
like it has lots of head room but in general staying conservative.
2. I then print out a gray wedge and measure L*a*b of each of the 21 patches
(I use the QuadToneRip wedges and read it in MeasureTool just because it is
easier), and look at the deviation from nominal grav values for all patches.
Ignore small deviations. If there is large departures from gray, I use table
6.4 in the G7 document someone posted as a link before (page 22), to
increase or decrease specific ink limits, so I get close to neutral gray for
mid-tones (highlights and shadows probably won't track mid-tones). The sole
purpose of this step is to get close to grayscale tracking by adjusting ink
limits, so even without profiling, the grayscale is close. I am just
assuming this will make it easier on profiling. In one case I had to
increase cyan limit from 1.45 to 1.75 and increase yellow to 1.15 density to
achieve this.
3. I print out an ink limit chart from studioprint (or any available on the
internet, it is just a mix of various CMYK inks, the StudioPrint one has
"teeth" in them so you cansee any ink pooling or flooding easily). I note
this down. I don't set the RIP ink limit to this value though, this is for
ink limiting in the profile. Usually,because my individual ink limits are
already lower than normal, the two ink combinations are already pretty good
(R,G,B).
4. Set the ink limit in profile chart to the value found above. Normally I
prefer maximum black generation, but lately I have gotten increased Dmax and
less banding inshadows with a GCR3 in ProfileMaker (or equivalent ramp in
Arygyll - I always got better grayscale tracking in Argyll for some reason
than PM).
Well, this procedure has worked for me quite well, and I use less ink now,
and visibly also, the lower ink limits don't seem to have affected any
visible gamut (I also looked in ColorThink, hardly any degradation in
Gamut). This maybe characteristic of the paper though, as Museo Silver seems
to have a higher Gamut,, but this also worked pretty well for Kodak
Professional Luster and Gloss.
Anyway, in short I am a convert to lower individual ink limits now. I have
not seen any appreciable lowering of Gamut.
On 1/28/09 5:43 PM, "Alastair M. Robinson"
<profiling@xxxxxxxxxxxxxxxxxxxxxxx> wrote:
> Hi :)
>
> Martin Weberg wrote:
>
>> I prefer plotting a and b together looking through the L plane, which
>> I think better illustrates how CMY would effect the final gamut. If
>> the color bends (shift hue) really fast it's often really obvious that
>> adding further ink will not expand the gamut.
>
> Yup, I shall try that in due course. I suspect that will be case for
> Cyan in the graphs plotted I linked to - which would make sense since
> that seems to be the channel which most easily saturates the paper.
> I need to be prepared to handle an individual ink limit for each channel.
>
>> Then I prefer to limit
>> right at the start of that bend or just a little beyond (subjective? I
>> know :-)).
>
> Whatever works, I guess :)
>
> Thanks again for the input
>
> All the best,
> --
> Alastair M. Robinson
>
Other related posts:
