Dan Margulis Applied Color Theory - What Makes a Contract Proof?

Date: Mon, 04 Nov 2002 08:16:21 -0600
From: Al MacDonald
Subject: contract proof question

We are photographers, not pre-press gurus. For years, color houses have told us that we need to make "contract" proofs to realistically predict color on press. Back in the days prior to a digital workflow this was done from the B/w separation negs and 4 color match prints. For the past several years the Kodak Approval (or other similar product) has been the benchmark. We are told that dye sub prints and Fuji Pictro prints are not "contract" quality and can never truly predict press outcome.

Last week, when color on a job seemed out of balance on an Approval proof, I had a pre-press house tell me that "color varies from one Approval machine to another".

If that's true, what makes a "contract proof" a contract proof, and how can any of them accurately represent on press color? What makes them any better than a profiled dye sub print?
--
Al MacDonald
Shaughnessy MacDonald, Inc.
Elk Grove Village, IL 60007
http//www.shaumac.com


Date: Mon, 04 Nov 2002 09:19:59 -0600
From: Henry Segalini
Subject: Re: contract proof question

The integrity and ability of the printing company. We make contract proofs (Fuji) and our obligation is to match the proof on the press. We will sometimes make the proofs on job stock.

Our proofing systems, presses, and inks have been coordinated to make this possible.

Another printer might not be able to match our proofs because they have different presses, inks, etc.


Date: Mon, 04 Nov 2002 08:48:09 -0800
From: Rosemary Hill
Subject: Re: contract proof question

IMHO, the kind of attitude that your prepress house has represents a total abnegation of responsibility. I have run into it many times, as well. Instead of working to make the system as reliable as proofs from film, many printers and prepress houses are just using the easy way out: saying that proofs from digital files are not as exact. It is just a different process. An Approval proof should be calibrated to represent what you can expect on press. I have a printer who actually *does* match their proofs, but they've told me that it took them a long time to get their system properly calibrated. I would suggest having the printer do your contract proof in the future. It's harder to weasel out of matching the proof on press if they've done the proof themselves, although I've had a printer do it before.

Best,
Rosemary

Al MacDonald wrote:

> Last week, when color on a job seemed out of balance on an Approval proof, I
> had a pre-press house tell me that "color varies from one Approval machine
> to another".


Date: Mon, 4 Nov 2002 09:54:22 -0700
From: Chris Murphy
Subject: Re: contract proof question

On Monday, November 4, 2002, at 07:16 AM, Al MacDonald wrote:

> We are photographers, not pre-press gurus. For years, color houses have told
> us that we need to make "contract" proofs to realistically predict color on
> press. Back in the days prior to a digital workflow this was done from the
> B&W separation negs and 4 color match prints. For the past several years the
> Kodak Approval (or other similar product) has been the benchmark. We are
> told that dye sub prints and Fuji Pictro prints are not "contract" quality
> and can never truly predict press outcome.

Well for that matter neither can Matchprints. Fuji PictroProofs have been extremely popular wherever I've seen them installed. But then, so have been Matchprints, and I really don't like Matchprints. Ultimately, what is contract quality is a proof that is AGREED upon to be contract quality by the print buyer and supplier. If it's crayons, then so be it.

The only thing that truly predicts press outcome is a press proof, and even that's not 100%, but pretty close to it. Digital proofing based on actual press behavior is probably the most practical means of predicting actual press output, but it's still challenging because it requires rigorous process control, and a really good profile for the press, and a good profile for the proofing device.

> Last week, when color on a job seemed out of balance on an Approval proof, I
> had a pre-press house tell me that "color varies from one Approval machine
> to another".

They can because the user can tweak the output behavior of the approval to better match their specific press condition.

> If that's true, what makes a "contract proof" a contract proof, and how can
> any of them accurately represent on press color? What makes them any better
> than a profiled dye sub print?

You want to buy your contract proof from your printer, or from a service bureau that your printer agrees to working with. If you just get Matchprints from anyone and expect the printer to match them, he may try but ultimately he's not in a position to guarantee a match on anything he didn't make because that's not where the industry is at. The industry has largely rejected standards based printing conditions, so this is what happens as a result.

Chris Murphy
Color Remedies (tm)
Boulder, CO
303-415-9932


Date: Mon, 4 Nov 2002 08:58:55 -0800 (PST)
From: Mike Bevans
Subject: Re: contract proof question

IMHO, big words like "abnegation" are hard to read on Monday mornings.

But I do agree, if I understand properly. It seems since photography has gone digital, everyone is pretty much starting over. Just as photographers are using new tools, prepress is learning and retooling, too.

I have recently conducted several print tests with several types of outputs at sevral locations. Results varied.

What did not vary was an eagerness of the prepress houses to work with me to solve any of the problems that came up. Everyone wants to get paid. The more we work together to solve repro issues the closer we get to our checks.

Photographers and prepress can do a lot to work together to get good reproductions, and everyone learns something new. Beyond all the tech issues, the most helpful aspect of production is a personal cooperative relationship with the people who are printing your work.

-Mike Bevans
www.tribecalabs.com


Date: Mon, 04 Nov 2002 12:04:47 -0500
From: Jim Rich
Subject: Re: contract proof question

On 11/4/02 10:19 AM, Henry Segalini wrote:

> The integrity and ability of the printing company. We make contract proofs
> (Fuji) and our obligation is to match the proof on the press. We will
> sometimes make the proofs on job stock.
>
> Our proofing systems, presses, and inks have been coordinated to make this
> possible.
>
> Another printer might not be able to match our proofs because they have
> different presses, inks, etc.

As Henry points out in his post his shops digital contract proofs are just that, a contract between him and his clients.

As for Approval proof drifting, sure it happens, but a good printer / pre-press service should be in control of that and if they are not then... perhaps it is time to find another service provider.

Also,

Another reason that businesses like to use the Kodak Approval is that those devices provide halftone dots in the proof where as a dye sub or ink jet printers don1t provide halftone dot options. If you need to have people in the process see dots in a pre-press proof (via your digital workflow) then a product like the Kodak Approval is a must, but if you don't care about dots then other printer products can create visual results that look like contract proofs. Having stated that in an off-handed way. There is more to setting up a digital contract proofing system than buying an inkjet printer. There is usually a rip, profiling tools, and a learning curve.

Jim Rich


Date: Mon, 4 Nov 2002 10:37:15 -0700
From: Chris Murphy
Subject: Re: contract proof question

Jim Rich wrote:

> As for Approval proof drifting, sure it happens, but a good printer /
> pre-press service should be in control of that and if they are not then...
> perhaps it is time to find another service provider.

It may not just be a matter of drift. There are a number of different pigment sheets for simulating different inkset on the Approval in addition to curves adjustments so that dot gain other than the standard dot gain can be simulated.

This is hardly different than a Matchprint which allows for different exposure times to simulate higher or lower dot gain (within the limits of the laminate being used). With just materials alone there are something like 32 different kinds of Matchprints. If you include exposure into the equation there are potentially hundreds. While skilled pressmen can whack out their press in order to match some crazy random Matchprint that they didn't make, this is not an ideal scenario. I regularly run into shops that have problems matching Matchprints, or Waterproofs, or Color Arts. It's all dependent on actual press behavior. Sometimes the design of the Matchprint is not mean to simulate the behavior of a particular press condition and trying to force the issue really isn't going to get you anywhere. The proof is simply the wrong proof for the job.

Chris Murphy
Color Remedies (tm)
Boulder, CO
303-415-9932


Date: Mon, 4 Nov 2002 12:43:39 -0500
From: Dan Remaley
Subject: RE: contract proof question

If you go to SWOP.org you can download a SWOP certified specification (how a proof should be made to match) SWOP printing. Each proofing system is measured in Lab - also density and dot gain. They also recommend a GATF proof Comparator on every proof for compliance. The Pictro Proof is compliant and uses GRAY BALANCE for calibration - I love it!
Dan Remaley/GATF
412.741.6860x450


Date: Tue, 5 Nov 2002 10:21:25 -0500
From: Scott Olswold
Subject: RE: contract proof question

Al,

Whoa-ho! You've asked a question that opens a fine can of worms.

A contract proof can, in truth, be any output that the printing house accepts as a benchmark for the press-run (accommodating, of course, a press check). But that has to be determined beforehand. A recent project that I sent out came back as Sherpa proofs for color checks prior to hitting the press. When I asked if the Sherpa system was calibrated/characterized/referencing the press? The answer was "No, we're too busy to do that often...but the printshop knows how to handle them." Surprisingly (OK, no, not really--I've worked really hard to make it close), the output from my color laser was very close to the Sherpa print--and the print job. By extension, that means that my laser output would have qualified as a contract proof.

On another production floor, the only want Matchprint or Rainbow proofs. In the end, its an agreement between you and the printer.

Scott Olswold


Date: Tue, 5 Nov 2002 13:43:18 -0800
From: Ray Maxwell
Subject: RE: contract proof question

What is a contract proof?

It is an agreement between the print buyer and the printer. If the press sheets match the proof then the buyer is obligated to pay the bill.

The only problem with this is...What is a Match?

Imagine that I ask a machinist to build a round rod 6 inches long and 1 inch in diameter. He won't start the job until I specify a tolerance. He has to know these tolerances in order to know what percesion the buyer expects. His price will be different depending on the tolerance. If I want it accurate to plus or minus 1 micron the price will be higher than if I can live with plus or minus 0.001 of an inch.

In press tests I have conducted I have measured a rapid press variations, every four sheets, that are greater than 0.02D. There are longer time variables on press that are greater than this.

If we are going to make printing an accurate manufacturing process we need an objective way to specify color and to tolerance it. In addition we need some standard printing conditions as well as ways to specify custom printing conditions.

Dan, you said that SWOP has measured Lab values. If I understand correctly SWOP has not officalilly made TR001, which lists Lab values, a part of SWOP.

Currently SWOP makes an "approved" press run every few years and then asks the vendors to match it. The match is determined visually by a committee of experts. It has been shown that the more recent press runs do not "match" the run that was measured in TR001. Currently this standard is moving slightly and is not specified in an objective way.

I understand that GRACOL is working on producing objective printing standards. I don't know where they are at this time.

I would love it if the industry used objective and toleranced standards.

Ray

THE OPIONS EXPRESSED IN THIS EMAIL ARE THOSE OF RAY MAXWELL AND DO NOT
REPRESENT THE OPIONIONS OF THE COMPANY HE WORKS FOR.

Creo
Ray Maxwell | Senior Color Systems Engineer, Inkjet Printing
4225 Kincaid Street | Phone (604) 451-2700 ext. 2004
Burnaby, B.C.
Canada V5G 4P5
IMAGINE CREATE BELIEVE


Date: Tue, 5 Nov 2002 15:24:39 -0700
From: Chris Murphy
Subject: Re: contract proof question

Scott Olswold wrote:

> A contract proof can, in truth, be any output that the printing house
> accepts as a benchmark for the press-run (accommodating, of course, a press check).

and

> Surprisingly (OK, no, not really--I've worked really hard to make it close),
> the output from my color laser was very close to the Sherpa print--and the
> print job. By extension, that means that my laser output would have
> qualified as a contract proof.

No, and definitely no. A contract proof, by definition, requires two parties to agree that it is a contract proof. In the first case, you are pointing only to agreement of the printer, and in the second case you're pointing only to agreement from your point of view. Take the color laser to the print shop for the next job and tell them to accept it as a contract proof, with no other proofs, with all of the legal implication thereof and I'm very sure they will say no. I've personally witnessed density variations on color laser printers over as little as two hours that caused them to have double the density and half the density as two hours previous. This is not an acceptable contract proofing mechanism for most workflows, unless a lot of process control effort is put into it, and even then I rarely see something like the laser printer based Matchprint being used.

Chris


Date: Tue, 5 Nov 2002 15:26:31 -0700
From: Chris Murphy
Subject: Re: contract proof question

Ray Maxwell wrote:

> I would love it if the industry used objective and toleranced standards.

Heretic! Don't you know that standards are nothing short of communism?! It's a printer's right to have completely unpredictable press behavior, and plenty regularly choose to assert this right.

Chris Murphy
Color Remedies (tm)
Boulder, CO
303-415-9932


Date: Tue, 5 Nov 2002 18:55:05 EST
From: Dan Margulis
Subject: Re: contract proof question

Ray writes, with Chris cheering him on,

> Imagine that I ask a machinist to build a round rod 6 inches long and 1
> inch in diameter. He won't start the job until I specify a tolerance. He
> has to know these tolerances in order to know what percesion the buyer
> expects. His price will be different depending on the tolerance. If I
> want it accurate to plus or minus 1 micron the price will be higher than
> if I can live with plus or minus 0.001 of an inch.

There is universal agreement not just on what a micron, or an inch, or a centimeter, means, but that any of these are appropriate and unambiguous units of measure. We have no such standard measure with respect to color. To the extent there is one, it's the preposterous Delta-E, which weights its components in absurd ways that completely misapprehend how humans perceive color. But even if somebody came up with a better measurement system there would still be cases where the measurement would say that things are in spec where a human would say they aren't, and vice versa.

Second, if a machine tells us that a certain rod is .037 inches longer than a second rod, we can, with some effort, arrange to verify the measurement with our own eyes. And all three of us will agree not just with each other but with the machine.

But if a machine tells us that a certain color is, by formula, acceptably close to another, there's no way for us to verify that it's measuring the two colors properly, and every reason to suspect that it isn't. All three of us are capable of discerning accuracy better than the machine can, but the problem is that we may not agree with one another. So we are stuck with the choice of using a system that gives repeatable results that are meaningless, or a system that gives meaningful results that can't be used.

That color printing can somehow be governed by the same rules as precision metalwork is an idea that has derailed many a calibrationist over the years. It sounds so logical, but the devil's in the details.

Dan Margulis


Date: Tue, 5 Nov 2002 21:52:33 -0700
From: Chris Murphy
Subject: Re: Re: contract proof question

Dan Margulis wrote:

> There is universal agreement not just on what a micron, or an inch, or a
> centimeter, means, but that any of these are appropriate and unambiguous
> units of measure. We have no such standard measure with respect to color. To
> the extent there is one, it's the preposterous Delta-E, which weights its
> components in absurd ways that completely misapprehend how humans perceive
> color.

Actually, the unambiguous measurement of color would be CIE XYZ, CIE LAB, or CIE LUV. Those are standards.

I'm not aware of any Delta-E formula being a standard (even though it very well may be), and in any event it has nothing to do with standard measurement of color. It has to do with predicting the perceived difference between two colors. There is more than one delta E as well. Delta E 76, I believe was the first and perhaps the least accurate. I forget what percent of the time it's reasonably accurate in prediction but I think it's around 30%. Not great. Delta E 94 is quite a bit better, as is Delta E cmc. Supposedly there is a Delta E 2000 around somewhere but I haven't confirmed if it's been finalized, or if it works any better than the others previously mentioned.

> But even if somebody came up with a better measurement system there
> would still be cases where the measurement would say that things are > in spec
> where a human would say they aren't, and vice versa.

Until there is means of incorporating color appearance models (including luminosity and ambient lighting, something we don't have now) as well as the effect of screening algorithms, that's correct.

> But if a machine tells us that a certain color is, by formula, acceptably
> close to another, there's no way for us to verify that it's measuring the two
> colors properly, and every reason to suspect that it isn't.

Devices don't really measure color. They can only measure the stimulus. If multiple instruments agree on the stimulus then the measurement is likely correct. That the resulting measurement may not agree with what our brain tells us is dilemma of colorimetry being different than color: the difference between the stimulus (what is measured, also called colorimetry) versus the human response to that stimulus (color).

> All three of us
> are capable of discerning accuracy better than the machine can, but the
> problem is that we may not agree with one another. So we are stuck with the
> choice of using a system that gives repeatable results that are meaningless,
> or a system that gives meaningful results that can't be used.

Meaningful or meaningless depends on context, and the opinions of the right people (people who matter). Colorimetry is not inherently meaningless. There are ways to use common sense and training to form livable compromises between the use of colorimetry and color.

Chris Murphy
Color Remedies (tm)
Boulder, CO
303-415-9932


Date: Wed, 6 Nov 2002 05:49:00 -0700
From: Chris Murphy
Subject: Re: Re: contract proof question

Oh, and one more thing, Dan:

To really get us back on track, we're both off top because this isn't a color measurement, color management, color model, color perception, or even a question of color at all. Standards are about file and proof portability. There are printing companies around the world (including a few in the U.S.) who make standards work in their environment, and there are printing standards - ISO 12647-1 through something or other, 9 I think. But even the politically correct GRACoL and SWOP specifications, while not really standards, are appropriate guidelines that if better followed would significantly improve file and proof portability. I think this is increasingly needed because digital proofing allows every printer to optimize their press however they want, and therefore an uncommon printing process.

Chris Murphy
Color Remedies (tm)
Boulder, CO
303-415-9932


Date: Wed, 6 Nov 2002 08:39:23 -0500
From: Scott Olswold
Subject: RE: contract proof question

Chris,

You're going to flag me on semantics<g>?

OK, so I didn't explicitly call out the designer/photographer half of the first equation (implicitly, yes). And I didn't specifically say that the print shop was going to accept my color laser as a contract proof--I said "would" when I should have said "could"--and simply on the basis that the colors were visually and metrically similar (measured with a DTP41). I was being loose with my structure. I apologize.

I think that several of us on the list have all basically said the same thing in not as many words: Any given composite print can be used as a contract proof as long as both the print shop and the client agree. You said yourself that this could be crayon, as long as both parties say "yes, that's our contract proof."

Oh, and I have a laser here that's a ton more stable than your observed device. I've been using the ColorMetrix software for the past three months to measure my CLC 1150 with only small drift (easily corrected with a linearization).

Scott Olswold


Date: Wed, 6 Nov 2002 08:26:35 -0500
From: Dan Remaley
Subject: RE: contract proof question

Ray, you are correct, the Lab values (listed at swop.org) are only to 'show' a visual match to a SWOP certified proof. It's a place to start!

The .02 density shift on press that you mentioned isn't any different than the 'standard' deviation of +/- 2% midtone gain every press has. The 'secret' is to use GCR in scanning to eliminate the color shifts associated with the +/- 2% deviation = a more stable process. My conviction is to print to gray balance and use GCR to 'help' the press natural deviation.
Dan Remaley/GATF
412.741.6860x450

> From: Ray Maxwell

> > In press tests I have conducted I have measured a rapid press
> variations,
> > every four sheets, that are greater than 0.02D. There are longer time
> > variables on press that are greater than this.


Date: Wed, 6 Nov 2002 10:35:15 -0800
From: Ray Maxwell
Subject: RE: Re: contract proof question

Dan Wrote:
> ----------
> So we are stuck with the
> choice of using a system that gives repeatable results that are
> meaningless,
> or a system that gives meaningful results that can't be used.

Dan,

You elude to viewer metameterism...different people will not agree on when two colors match. This can differ by as much as delta E of five in tests I have conducted.

I would choose repeatable results. I don't agree that consistency is meaningless.

In a poll of our customers the number one requirement for a proofing system was repeatability. Consistency was more important that an accurate match. They felt that it was more important that the proof be stable and consistent than that it be an exact match to the press.

Calibration with instruments can make a proofer consistent.

Profiling and use of a device link can get a close match.

Auto correction of the device link and manual editing can get an extremely good match. I agree that the final tweak is best done by a human eye, however, I suggest that an instrument can do a very good job of keeping a proofer consistent once it has been characterized by a hand tweaked profile.

I also think that colormetric measurements are a good idea for proofing standards. I hope that TR001 or the measuement of a new run will become part of SWOP and GRACOL. I think this will happen as instruments become better and less expensive.

Ray

> Creo
> Ray Maxwell | Senior Color Systems Engineer, Inkjet Printing
> 4225 Kincaid Street | Phone (604) 451-2700 ext. 2004
> Burnaby, B.C.
> Canada V5G 4P5
> IMAGINE CREATE BELIEVE


Date: Wed, 6 Nov 2002 18:56:10 -0500
From: David McDowell
Subject: Re: contract proof question

Ray wrote:
"I hope that TR001 or the measurement of a new run will become part of SWOP "

It is. On page 22 the SWOP 2000 book says "ANSI/CGATS TR 001 documents the colorimetric characterization of the CMYK to CIELAB relationship for print conditions that are used to produce SWOP certified press proofs. SWOP specifies the use of ANSI/CGATS TR 001 characterization for any color-managed applications (e.g., SWOP certified off-press proofing, remote proofing, etc.)."

That is as strong an endorsement as one needs.

David Q. McDowell
Standards Consultant
51 Parkwood Lane
Penfield, NY 14526
Tel: 585-383-1706, Fax: 585-385-3828


Date: Wed, 06 Nov 2002 18:26:56 -0600
From: Al MacDonald
Subject: Re: contract proof question

Many thanks to all for the replies to my original "contract proof" question.

As photographers we used to give transparencies to our clients and when viewed in a proper booth the client knew what color they had. End of job for us!

Now with digital capture the client wants a random proof of a CMYK file to see color before they place our files in their final layout documents. They use several different printers, both sheet fed and web. So, since the print shop has not been selected at time of photo the printer can't make the proof as many of you have suggested.

But, the client still wants an accurate hard proof that represents on press behaviour of the file.

Dan Remaley at GATF has suggested placing a small proof comparator file onto every output. The comparator patches can then be measured on the output to see if the proofing device is set to known standards and that mid tone gray balance is neutral. At the very least we will know if the proofing device is introducing a color cast. Makes sense to me!
Al
--
Al MacDonald
Shaughnessy MacDonald, Inc.
http//www.shaumac.com
--------


Date: Wed, 6 Nov 2002 17:34:00 -0800
From: Ray Maxwell
Subject: RE: Re: contract proof question

David McDowell wrote:

> It is. On page 22 the SWOP 2000 book says
> "ANSI/CGATS TR 001 documents the colorimetric characterization of the CMYK
> to CIELAB relationship for
> print conditions that are used to produce SWOP certified press proofs.
> SWOP specifies the use of
> ANSI/CGATS TR 001 characterization for any color-managed applications
> (e.g., SWOP certified off-press
> proofing, remote proofing, etc.)."

I agree that is want the SWOP document states, however, it is my understanding that if a vendor takes a proofer into be certified by SWOP it is done by visual means only. No instruments are used and no reference is made to TR001.

If this has changed I would be delighted.

Good to hear from you Dave,

Ray

> Creo
> Ray Maxwell | Senior Color Systems Engineer, Inkjet Printing
> 4225 Kincaid Street | Phone (604) 451-2700 ext. 2004
> Burnaby, B.C.
> Canada V5G 4P5
> IMAGINE CREATE BELIEVE


Date: Thu, 7 Nov 2002 07:45:14 -0500
From: David McDowell
Subject: Re: contract proof question

Ray wrote:

"I agree that is want the SWOP document states, however, it is my understanding that if a vendor takes a proofer into be certified by SWOP it is done by visual means only. No instruments are used and no reference is made to TR001.

If this has changed I would be delighted"

My understanding is that it is a combination of visual evaluation and instrumental evaluation.and comparison to TR001

It is clear that IF the delta-E between the IT8.7/3 target measured on the proof and on the reference were zero they would match. However, that is highly unlikely.

If they do not match, we do not know how to weight the various contributions to error in a complex image. All the work that the CIE has done is focused on the comparison of a color match of relatively large areas of uniform color. Paint and textiles.

CIE TC8-02 has the task of trying to develop a measure of color differences in complex images but is still trying to develop a test methodology.

We simply do not know how to meaningfully define a tolerance. People have looked at histograms of the delta-Es and tried to relate the mean, the shape, the 90th percentile, the 99th percentile etc to subjective judgements but there is no clear answer. The biggest problem is that there are too few image sets for which we have both measurement data and psychometric ranking.

David Q. McDowell
Standards Consultant
51 Parkwood Lane
Penfield, NY 14526
Tel: 585-383-1706, Fax: 585-385-3828


Date: Thu, 7 Nov 2002 12:50:04 -0500
From: John Rawlins
Subject: color proofing

Question in regards to the contract proofing topic:

Can we all agree (or disagree) that there is no type of proof that EXACTLY matches a printed press sheet?

If so, then what we are really doing is to define HOW CLOSE they can be made to match, and at what limit of expense we are willing to go to get there. I submit that every photographer, design artist, client, and printer have their own definition of what a match is, and what cost limits they are willing to spend. The "contract" portion comes when they agree in advance with the printer what those expectations and limits are.

John Rawlins


Date: Thu, 07 Nov 2002 11:35:10 -0800
From: Joel Marion
Subject: Re: contract proof question

We get press-proofs.

Thanks,
Joel Marion
Art Director
Impact Photo Graphics
916.939.9333 x17
916.939.9334 fax


Date: Thu, 07 Nov 2002 21:08:00 -0000
From: Ron Scratch
Subject: Re: color proofing

John wrote:

> Can we all agree (or disagree) that there is no type of proof that EXACTLY matches a printed press sheet?

I definitely agree. I have never seen a digital, analog or any other proofing system that matches a printed press sheet. However, I think the real job of a "Proof" is to predict the reproduction of an ink set along with it's printing process. My clients and I both deal with Gravure, Offset Web and Offset sheetfed printing processes. I cannot begin to count how many different individual press conditions are out there that I supply files for. I do have an understanding with the Pressrooms I deal with that I will shoot for the SWOP standard when providing "contract" proofs. In turn my client has a relationship with the Printer, to achive a "Reasonable Match" between the digital proof and the press sheet. In the end I get a client sign off on my digital proof saying that we agree that this is the ideal color in which we would like to achive on press. Then the client uses that proof and ultimatley signs off on an actual press sheet saying- The pressman achived a reasonable match to the digital proof supplied.

The only thing I run into. Is that SWOP is percieved to be an Ethereal standard. Meaning that - saying it meets SWOP doesn't always absolve anyone from taking the "blame" when a resonable match isn't achived.

Ron Scratch


Date: Thu, 7 Nov 2002 16:04:19 -0500
From: Scott Olswold
Subject: RE: color proofing

John,

At the risk of agreeing without any technical justification behind it, I'll submit a "Yes" that's how it really works.

Scott Olswold


Date: Thu, 7 Nov 2002 15:17:40 -0800
From: Ray Maxwell
Subject: RE: Re: contract proof question

David wrote:

> We simply do not know how to meaningfully define a tolerance. People have
> looked at histograms of the delta-Es and tried to relate the mean, the
> shape, the 90th percentile, the 99th percentile etc to subjective
> judgments but there is no clear answer. The biggest problem is that there
> are too few image sets for which we have both measurement data and
> psychometric ranking.

Thanks for the explanation David,

I will stick my neck out and make a naive suggestion. What if we choose a test target (it8.7/3, ECI2002) and then specified a separate tolerance for each patch with a separate tolerance for each L*, a*, b* value. I know this sounds cumbersome, but with computers it would not be that hard to build a program that would check results against this model and give a pass fail or suggested moves to correct the patch. One could then tune a profile until the match was within the model.

I know this does not allow for appearance model deviations. But it would be a start in the right direction.

Ray

> Creo
> Ray Maxwell | Senior Color Systems Engineer, Inkjet Printing
> 4225 Kincaid Street | Phone (604) 451-2700 ext. 2004
> Burnaby, B.C.
> Canada V5G 4P5
> IMAGINE CREATE BELIEVE


Date: Fri, 8 Nov 2002 10:39:04 EST
From: Dan Margulis
Subject: RE: contract proof question

Ray writes,

>>In a poll of our customers the number one requirement for a proofing system was repeatability. Consistency was more important that an accurate match. They felt that it was more important that the proof be stable and consistent than that it be an exact match to the press.>>

Your customers are exactly right. This is the way it should be. However, the message that you were replying to was not about proofing systems, but about presses. The same considerations don't apply there.

>>I agree that the final tweak is best done by a human eye, however, I suggest that an instrument can do a very good job of keeping a proofer consistent once it has been characterized by a hand tweaked profile.>>

The instrument can keep the proofer consistent regardless of what calibration method has been used, even if "consistent" means "consistently bad." But vastly fewer factors influence the performance of a proof than is the case with a press. On a reasonable proofing system one can set up targets, measure them with a lumihumitron or whatever, and if they don't vary from day to day performance won't vary much either. On presses, this isn't true--one can have results that measure the same and look entirely different.

The whole discussion goes hand in hand with the thread about what contract proof the printer should accept. We have broad agreement in the group that the printer isn't obligated to accept proofs that he doesn't have confidence he can match. We also have agreement in the group that it would be nice if there were a way to make proofs that printers would trust. The way to achieve this is just what you're saying--more consistent, more predictable proofing. More consistent, more predictable presses would be nice too, but we've lived without them for a good long time. They aren't the problem here.

Dan Margulis


Date: Fri, 8 Nov 2002 10:38:24 EST
From: Dan Margulis
Subject: Re: Re: contract proof question

Chris writes,

>>To really get us back on track, we're both off top because this isn't a color measurement, color management, color model, color perception, or even a question of color at all.>>

It's none of those things in the sense that we are talking about the idea of standards abstractly; but it's all of those things in that people who don't understand them wind up with impossible ideas of what the standards should be.

>>But even the politically correct GRACoL and SWOP specifications, while not really standards, are appropriate guidelines that if better followed would significantly improve file and proof portability.>>

They are indeed useful guidelines, which is why almost all printers at least pay lip service to them, and have for many, many years. However, both of these have very large leeways built in. As time has gone on and presses have become easier to control, some of the specs have tightened up, as well they should.

>>I think this is increasingly needed because digital proofing allows every printer to optimize their press however they want, and therefore an uncommon printing process.>>

Printers don't want to have wildly different printing conditions than their competitors.

Process control has improved in the last few years, but still, any given printer probably has significantly different performance on the front and the back of the sheet, and on the sides as opposed to the center of the form, not to mention significant variations from minute to minute during the run. Proofing systems can be made extremely consistent. People who preach that presses can be made just as consistent are just setting themselves up for a fall.

Dan Margulis


Date: Fri, 8 Nov 2002 10:39:44 EST
From: Dan Margulis
Subject: RE: contract proof question

David McDowell writes,

> We simply do not know how to meaningfully define a tolerance. People
> have looked at histograms of the delta-Es and tried to relate the mean, the
> shape, the 90th percentile, the 99th percentile etc to subjective
> judgements but there is no clear answer. The biggest problem is that
> there are too few image sets for which we have both measurement data and
> psychometric ranking.

No, the biggest problem is that the exercise is roughly akin to analyzing an image by means of its histogram but first applying a Gaussian blur filter at a radius of 15 pixels. Randomness is introduced by this absurdly primitive delta-E concept which savagely overweights variation in the B and underweights variation in the L. Furthermore, it doesn't take account of age variation in the viewing population. That is, even assuming somebody worked out a replacement for delta-E that made more sense, you'd have to have different formulas for persons over and under 40, because of very well-known vision changes pertaining to the aging process.

But, the bottom line is as you say: we don't know how to define tolerances at this point.

Dan Margulis


Date: Fri, 08 Nov 2002 16:37:55 +0000
From: David Deranian
Subject: RE: contract proof question

Hello Everyone-

Just to throw in a thought here on the subject of calibration and tolerance.

While I'm not really qualified to speak to the available technology that is specifically used to measure color or the various processes used to produce consistent proofs for color matching, I believe Dan is right, and that no such measuring standard will or should ever exist.

However, there are existing theories and technologies that could be utilized to get you damn close. Tweaking of the human eye will always (in my opinion) be the final word.

Accepting the notion (for the moment) that color perception will always be subjective no matter how accurate we are able to measure it and apply a specific number to it, wouldnÕt it make sense to use an average consensus technology of real life images to help standardize calibration.

For instance, Nikon cameras have for years used an intelligent metering system that analyzes the light coming through the lens and performs a near instant comparison of thousands of images stored digitally on a chip. Then it derives the correct exposure based on where in light is brightest, darkest and does a pretty good job of averaging the exposure to give you a uniform exposure across the entire frame. Of course the photographer still has the last word, in that the cameraÕs ÒsuggestionÓ can be altered or a manual override can be performed by the human.

It would seem to me that based on the subjective nature of color calibration, that rather than thinking in terms of ÒmeasurementÓ, we should be thinking in terms of a Òcomparison databaseÓ of thousands (maybe millions, who knows?) of real life images that would get our machines real close, but still leave us the capability to tweak using our eyes as we do now.

Just a thought. If any technology like what I've described already exists, I'd be interested in hearing from anyone our there that has had experience with it.

-David Deranian
--
Digital Arts & Sciences, Inc.
Communication Arts for
the Digital Age


Date: Fri, 8 Nov 2002 09:55:11 -0800
From: Al Ferrari
Subject: Re: Re: contract proof question

Dan Margulis writes:

>Process control has improved in the last few years, but still, any given
>printer probably has significantly different performance on the front and the
>back of the sheet, and on the sides as opposed to the center of the form, not
>to mention significant variations from minute to minute during the run.
>Proofing systems can be made extremely consistent. People who preach that
>presses can be made just as consistent are just setting themselves up for a
>fall.

And in another message Dan also writes:

>More consistent, more predictable presses would be nice too, but we've lived
>without them for a good long time. They aren't the problem here.

Dan,

At the risk of being relegated to the dustbin of the calibrationists, I would draw your attention to some very exiting and promising developments in offset printing as embodied in the Genius 52 and the 74 Karat printing presses by the German manufacturer KBA. More significant than the claimed improvements in process control in these machines are the new approaches to delivering ink to the printing plate which do away with the actual sources of many of the problems in lithographic printing to date. These innovations directly impact all of the performance issues that you refer to above. This is a new paradigm, not mere incremental improvements in process control. I will venture a guess that you have not had your images printed on some of these new presses.

Those interested can check out the links below.

http://www.karat-digital-press.com/74karat/index.html

http://www.kba-print.de/en/produkte/bogenoffsetdruck/genius52/index.html


Date: Fri, 8 Nov 2002 15:57:17 EST
From: Dan Margulis
Subject: Re: contract proof questions

Al writes,

>>At the risk of being relegated to the dustbin of the calibrationists, I would draw your attention to some very exiting and promising developments in offset printing as embodied in the Genius 52 and the 74 Karat printing presses by the German manufacturer KBA. More significant than the claimed improvements in process control in these machines are the new approaches to delivering ink to the printing plate which do away with the actual sources of many of the problems in lithographic printing to date. These innovations directly impact all of the performance issues that you refer to above. This is a new paradigm, not mere incremental improvements in process control. I will venture a guess that you have not had your images printed on some of these new presses.>>

That guess would be wholly inaccurate. If you've ever seen sample sheets from this manufacturer you were probably looking at images that I provided.

The approach they have taken makes eminent sense for a product that is aimed at relatively short runs. It doesn't speak to the problems that we have been discussing in the thread.

Dan Margulis


Date: Fri, 8 Nov 2002 16:31:40 -0500
From: Jim Ray
Subject: RE: Re: contract proof questions

Dan:

I am surprised that you identify the inking system design as one most suitable for short runs. I'm no printer, but from what I know of the printing process from 12 years of press check on moderately long runs (450000 copies, 96 page catalog), it sure looks like a step in the right direction. I understand that it does not address the issues that started this discussion. Perhaps the issue is the definition of relatively short runs? Please provide a bit more info. I don't doubt you, I'd just like to understand more about the apparent limitations of the technology.

Thanks,

Jim Ray


Date: Fri, 8 Nov 2002 23:18:20 EST
From: Dan Margulis
Subject: Re: contract proof question

Jim Ray writes,

>>I am surprised that you identify the inking system design as one most suitable for short runs. I'm no printer, but from what I know of the printing process from 12 years of press check on moderately long runs (450000 copies, 96 page catalog), it sure looks like a step in the right direction. I understand that it does not address the issues that started this discussion. Perhaps the issue is the definition of relatively short runs? Please provide a bit more info. I don't doubt you, I'd just like to understand more about the apparent limitations of the technology.>>

The question is one of how to match the proof. The press being spoken of is cheap enough to makeready that quality-conscious customers who wish to print on it will also proof on it. Therefore, the problem is to match something that has previously been printed on the same press and the same stock. The solution this company has come up with, of continuous machine adjustment on the fly, is indeed the best way to do it. If you have a bunch of test swatches from the proofing run, if you can make the swatches match on the live job then the images themselves will match.

When the problem is to match somebody else's proof that's made by a totally different process on a different substrate, this is a poor solution. Whether or not the test swatches match has very little to do with whether the images match. There conceivably are uses for measurements in achieving the match but they shouldn't be taken too seriously. All I was saying was that people who make a religion of measurement are generally relying on equipment that can't tell whether pictures match to decide whether pictures match, and on a formula that is clueless as to whether colors actually match to determine whether colors actually match.

Dan Margulis

Adobe Photoshop training classes are taught in the US by Sterling Ledet & Associates, Inc.