[argyllcms] Re: Accounting for dot gain in press simulation
- From: Craig Ringer <craig@xxxxxxxxxxxxxxxxxxxxx>
- To: argyllcms@xxxxxxxxxxxxx
- Date: Wed, 02 Mar 2011 13:19:21 +0800
Graeme (and Alastair),
Thanks _very_ much for taking the time to respond. I'll be trying out
the suggested process for proofing using Argyll, and now have a better
idea of what I should expect from the press profile's performance.
I might see if I can write some of this up for the next poor press user
who comes along; I'll post a summary link once I'm done nutting things
out with the press operators.
Rest of reply follows inline.
On 24/02/11 15:23, Graeme Gill wrote:
>> The very short version: Can ICC profiles account for dot gain? If so,
>> can Argyll help create a profile that does?
>
> Yes, ICC profiles account for the characteristics of the device.
OK, that's what I suspected. After all, we're comparing input colour
values to output colour values, and don't really care what particular
mechanism created the output colour values. I wasn't sure there wasn't
some perceptual or measurement quirk at work, though.
> Soft proof ? I thought Soft Proof meant a display based proof.
> I would tend to call a laser printer print a "Hard copy proof".
I think the folks here refer to "soft" proofs as anything not done on
the final output device. A "hard" proof is one run off the printing
press. I've picked up that language, but as they use a lot of weird
language here I wouldn't be surprised if it was a bit out there.
> I'm not sure what you mean by that. The way to do a proof is to take
> the Press CMYK (contone separation) and feed it into a link
> made between the press profile and the laser printer profile.
> Typically this would be an absolute colorimetric link for
> background simulation.
OK, I can test that.
>> The press folks say the dramatic colour differences in select areas is a
>> dot gain issue caused by the inks bleeding out into the paper. That's an
>> issue I'm familiar with, but haven't seen having this dramatic an effect
>> before.
>
> That sounds like a press-man's sort of explanation :-)
> It doesn't make any sense from a profiling point of view though.
> If that's what the press does with a certain CMYK input, then
> that's what the instrument will measure, and that's how the
> profile will represent its behaviour.
Yeah, I though it smelled for much the same reasons, too, but I don't
feel confident enough to call them on it. They're the kind of press
folks who think that "colour management" is something dangerous and
unpredictable that should be turned off at every stage of the workflow.
They won't accept PDF/X-1a or PDF/X-3 jobs from us, and haven't heard of
JDF.
> (This is assuming you have managed to
> profile the press - a non-trivial task typically, given the
> cost of running it up for a test chart, and getting the press
> to behave in a similar manner to "normal".)
They have a reasonably modern press with low run-up costs and low waste,
so it's not too hard to get them to do small test jobs.
The press folks (finally) supplied me with a real profile last year,
after one of their biggest clients threatened to go elsewhere unless
they made one. Prior to that they were insisting that a totally generic
profile that didn't even have paper colour information was "accurate"
and "made by measuring press output". That despite the fact that I could
create a byte-for-byte identical profile with even the same title in
Photoshop by choosing web uncoated, medium GCR and 30% dot gain!
Just prior to getting a half-decent profile from them I'd successfully
convinced them to run some swatch sheets off for me. I produced a very
rough profile from those using Argyll, but it had some weird abberations
and wasn't usable in production. I suspect the problem was that the
output was so far from what Argyll had expected that I had to turn the
spot read tolerance way up, so I probably also got bad reads accepted as
valid as a results.
What I want to do now is use their profile for characterisation data
when creating a new target, so I can refine their profile with new
swatch data. First, though, I need them to get their press to produce
the same output from run to run, or I'll be wasting my time measuring
swatches that won't match production.
> If the current press separation profile wasn't made from actually measuring
> the press (or alternatively, if the press hasn't or can't be adjusted
> to match the separation profile), then you will certainly
> have problems with proofing, since you don't have a technical
> means of looking up how the press behaves.
It was, but I suspect that the way the press is being run now isn't as
similar to how it was run for the profile as it should be. In fact, the
profile isn't even for the same press, as they've upgraded their
equipment recently and haven't supplied a new profile.
In any case, it seems they're not using sensor-feedback controlled ink
flow. They apparently prime the press setup with CIP3 data from the RIP,
but they play with the ink flow by hand as the press runs, using a
combination of manual solid colour densitometer (not spectro) readings
and "by eye" judgement. Needless to say, this doesn't exactly help
produce repeatable and predictable output.
I'm trying to convince them to use greybar / colour bar
measurement-driven press control for our jobs now, but I'm not sure they
even have the sensors and software to do it.
> so 1000 patches usually
> gave a reasonable profile. If you can do 3000 and it makes any sense
> (ie. the print run variation doesn't swamp other inaccuracies),
> then by all means do so.
OK, thanks for the confirmation there.
>> Just to complicate things a little more, the paper stock is off-white
>> (creamy), though at least it's without any FWA. The existing profile was
>> produced using that stock, so at least we're not lying to the CMS and
>> pretending to be printing on white stock.
>
> Nothing unusual about that. If you were using Argyll to create a proofing
> device link, then I would recommend you investigate the collink -w J,a,b
> option to fine tune the paper color emulation by eye.
OK, that's a very handy tip. Thanks.
--
Craig Ringer
Other related posts:
