Re: PS Preview
Re: PS Preview
- Subject: Re: PS Preview
- From: Roger Breton <email@hidden>
- Date: Sun, 28 Jan 2007 23:18:25 -0500
Kevin,
I maybe wrong. But...
> By merely assigning a CMYK image any profile that is not the profile which
> created it we introduce conversion errors because we lose the ICC managed
> 'realness' of those numbers. Here is proof of concept.
Hmmh. I'll read on.
> 1) Create a new file in Photoshop and assign the profile of an in-house,
> non-color managed output device such as a Kodak XP Approval.
OK. I create a new CMYK document in Photoshop to which I assign my faithful
Kodak XP Approval ICC profile. Therefore, anything I see onscreen should
reflect (relatively or absolutely) the appearance of CMYK colors as rendered
by the XP.
> 2) Create four separate 100% patches of CMYK in the digital file.
A 100C patch, a 100M patch, a 100Y patch and a 100K patch. Or have I got
that wrong? Sorry to be so dense if I have that wrong :(
> 3) Print a proof and place in a color booth next to the monitor.
You mean, send those CMYK patches to the Approval, as is, unmanaged. Done.
> As counter intiutive as it may be, we easily see that the CMYK patches on the
> proof do not match the patches on monitor even though we assigned the correct
> profile to the image.
Not only it would be counter intuitive but it would be against the
underlying principles of ICC color management, Kevin. I know monitors to
proof is an art in itself but on strict colorimetric grounds, the CMYK
patches on the proof DO match the patches on the monitor. Yes, a good
monitor and good profiling software are required to achieve this but this is
not science fiction.
> Measuring the proof with a spectro will yield even more
> interesting results, but let me move on to the point.
I lost you. See my post of Thursday or Friday where I was describing a
situation where the two devices, proof and LCD, matched each other to a very
low deltaE degree, albeit with a slight overall appearance mismatch.
> This test shows that we should think of ICC as a separation technology. It
> creates the separation that is appropriate for the specified output
> device/media but that doesn't mean ICC has a 100% accurate idea of what those
> CMYK numbers mean!
Well, I'm sure some people on the List will argue the contrary, Kevin.
Since, by definition, the CMYK characterization *is* the 100% "accurate idea
of what those CMYK numbers mean" -- otherwise, we're all lost. Why do you
think we measure those numbers on the press substrate (or Approval) in the
first place? Take SWOP11 as an example: not everyone will always use it for
separation, true (I have not seen how good or bad the results will be on
press, yet), but anyone is welcome to use the freely distributed dataset off
IDEAlliance website anytime (the data is finalized, btw) and use it for
proofing. Will everyone come up with wrong colors on their proofs? The new
SWOP rules state that certified SWOP proofing is no longer a visual thing
but a rigourous deltaE matching exercise. So, evidently, the numbers
attached to the CMYK numbers, by virtue of the ICC output profile, have to
carry meaningful visual interpretation.
Maybe I have not following what you were saying. And in that case, I'm sorry
to lead you and the list astray.
> The only thing ICC knows (so long as it has the proper
> profile) is what combination of CMYK numbers it would have chosen to create a
> color representation that best matches color X that was in a well behaved
> color space.
Again, an ICC output profile is bi-directional, by definition. It has a
table in for decoding the appearance of color from CMYK percentages, and, as
you say, it has a table (three, to be exact) to be used for separations.
> ICC was designed to go device independent forward but not CMYK backwards.
In which way, Kevin? Sorry to be so dense again.
> I understand people 'softproof' CMYK backwards all the time and they are
> mostly happy with the results.
Aha!
> Those people who are becoming not happy could
> see an improvement by separating appropriately to the intended output
> device/media combination if possible!
True. Then again, we could take Anthony Sanna's advice: just CMYK, you know
;-)
> -- Kevin Muldoon
Regards,
Roger Breton | Laval, Canada | email@hidden
http://pages.infinit.net/graxx
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Colorsync-users mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden