Grey balance test

[gordo]

Some scans of the press test I did on a Goss Colorliner coldset newspaper press to see (among other things) the impact of a solid ink density shift on the 3/C grey balance patch vs the live image area.

The page setup was Double Truck so everything was on the same plate and units.



The left side had seps done using PShop's default settings which uses medium GCR and represents, in my experience, the majority of separations going out to printers.
The right page had those same files run through ink optimization software which uses intelligent maximum GCR. This is pretty typical of most newspaper workflows today.
(The grey balance patches on the optimized side have had GCR applied to them - we couldn't prevent the software from doing its thing on that page).
The photo of the football player is a greyscale image converted to CMYK.

Start point.
The press operator was instructed to ignore the images and just run SIDs to SNAP standards - even across the form.



Then he was instructed to increase only the Magenta to .07 over the SNAP standard.



That increase in SID would be the absolute maximum that SNAP would allow - the normal acceptable variation is +/- .05.
There is an obvious shift in color in the 3/C grey balance patch but no noticeable shift in the live image area of either the standard or optimized pages.

Then he was instructed to increase the density of the Magenta to the max - which in this case was plus .20 points. This is a density that would never happen in actual production.



The 3/C grey patch became seriously biased. The live image area with standard separations took on a noticeable shift in color, however the optimized images were virtually unchanged.

My conclusion.

The 3/C grey balance patch has no value in production presswork vs just measuring solid ink density. The 3/C grey patch reports shifts that do not reflect what is happening in the live image area. The disconnect is even greater if images have gone through an ink optimization process.

The press operators were amazed in that they could actually see how little impact their SID moves had on the final presswork - as long as they hit the nominal SID standards within the accepted tolerance.

This exercise is one that I strongly recommend you do if you are at all interested in understanding QC targets such as the 3/C grey patch in production presswork.

best, gordo

 

[gordo]

This is a photo of the press that I ran the test on (couldn't include in the original post due to forum restrictions):



The head press operator checking the run.



The press guys use manual densitometers. I took this snap on a separate test in preparation to implementing FM screening.



best, gordo

 

[Deleted member 16349]

The left side had seps done using PShop's default settings which uses medium GCR and represents, in my experience, the majority of separations going out to printers.

The right page had those same files run through ink optimization software which uses intelligent maximum GCR. This is pretty typical of most newspaper workflows today.

Interesting stuff.

Question. In your opinion, did the medium GCR print look better than the maximum GCR printed version?

I am wondering that if it did, then being able to tightly control the density variation (+/- 0.03) might result in better looking printed images without the need for too much GCR.

Just curious.

 

[gig0]

Damn, that press looks like it was encased in rich mahogany. Sweet!

 

[gordo]

Interesting stuff.

Question. In your opinion, did the medium GCR print look better than the maximum GCR printed version?

I am wondering that if it did, then being able to tightly control the density variation (+/- 0.03) might result in better looking printed images without the need for too much GCR.

Just curious.

In this case the ink optimization (heavy GCR) and image optimization are applied to all images together. So the processed images look way better than the images separated normally. Of course image optimization is not applied to the 4/C greyscale image so it's a good indicator of the impact of SID shifts.
Ideally, a proper test should be done by a more sophisticated organization and its results published.
Nice to get confirmation (again) of my position on this subject.

best, gordo

 

[rich apollo]

In this case the ink optimization (heavy GCR) and image optimization are applied to all images together. So the processed images look way better than the images separated normally. Of course image optimization is not applied to the 4/C greyscale image so it's a good indicator of the impact of SID shifts.
Ideally, a proper test should be done by a more sophisticated organization and its results published.
Nice to get confirmation (again) of my position on this subject.

best, gordo

Gordo,
Your conclusion here is a little like saying that a canary in a coal mine was of no use because the birds died before the miners. A 3C patch is supposed to give you feedback before you have problems. It also gives you a visual tool by which you can monitor press performance. You can see the change in the gray patch whereas you cannot see the density changes in the solid.

In your pics I see the image with the Canadian flag going all over the place. The sky goes purple, the shadows take on a cast, the white area takes on a tint, the building in the background goes pink.

You definitely did demonstrate the increased stability that can come from image optimization (GCR).

Gordo, I think you and Erik would enjoy speaking with Andrew Yan. His company Dragon Tone has some interesting things going on concerning press control and ink feed via gray-balance.

 

[Lukas Engqvist]

What surprises me is that the sky does not seem to shift so much (in the optimised image) … but we have no bright blues… and is i understand correct it is just testing magenta from desired value to way over. Would the shifts be greater if the deviation was to dip the SID rather than push it? Also curious as to what the top images in the test form reveal

 

[Stephen Marsh]

Gordo, can you please clarify which exact profile or method you are referring to when you state that the default Photoshop CMYK profile uses Medium GCR?

Back in Photoshop 4 (not CS4), the old proprietary colour table method used Medium GCR as a default. In v5 (not CS5), ICC profiles were an option, however the old legacy settings were still favoured by many. In v6 (not CS6), the colour settings were changed to what is basically available today, favouring ICC profiles (with the old Adobe Custom CMYK table method hidden away as an on the fly configurable ICC profile).

Photoshop now uses presets in colour settings, that default colour profiles and colour management policies etc (in CS6, these are 3 distinct regions, Europe, Japan and North American). It took Adobe a while to ship a default setting for newspapers after the colour settings were changed back in v6.

Do you mean that the industry standard specifications that Adobe have created custom ICC profiles for use a GCR similar to the old legacy “Medium GCR in Adobe Custom CMYK”, or that the default CMYK in Photoshop is Custom CMYK using a Medium GCR setting (Meduim GCR is pretty much an arbitrary value which would vary depending upon the profile creation software, however we could presume that it is somewhere in the middle between minimum and maximum GCR values for any given ICC package).


Stephen Marsh

 

[Deleted member 16349]

Would the shifts be greater if the deviation was to dip the SID rather than push it?

All different combinations would be of interest to test.

Issues about tolerances have bothered me for a long time. There has been a tendency to use +/- 0.05 density points as a practical tolerance. Even in some sources, they are talking about +/- 0.10 as a tolerance. I think these are too wide.

For me, the issue has to be seen in the most demanding print conditions. What tolerance is required to maintain the printed image so that it does not visually change much.

When looking at screens, I think one should not just look at one screen but two screen combinations of the CMY screens and maybe three combinations when the K screen is used.

It is important to see how much variation is seen when one density goes down and another goes up to their tolerance limits. This would be the condition where the greatest variation is seen.

I suspect that the tolerances to maintain the screen parts of an image will be much tighter than is normally accepted. I suspect one needs a tolerance of about +/-0.03. Maybe less.

Also an image is not just screens but can be composed of screens, overprints, and solids. In the most demanding condition, all of these would be in an image and if one wanted to maintain consistency in a run, they would all be required to be maintained.

There is no doubt that using GCR works to reduce the visible variation of the printed screens but in any effort to obtain the kind of consistency one would like in a process, an engineering approach requires the solution to the most demanding conditions to be able to say that the process is in control within the more important tolerance of visual variation.

This is why I am still a firm believer in the requirement to obtain very tight density tolerance control capabilities, which can only come via correcting the ink feed problem at the fundamental level.

This then leads to predictability not only of density on press but to the predictability of colour when one is dealing with a more consistent process. One can not get predictable results with a process that is inconsistent. Consistency is a prerequisite.

Consistency does not come from setting standards but from making the process more capable.

 

[gordo]

Gordo, can you please clarify which exact profile or method you are referring to when you state that the default Photoshop CMYK profile uses Medium GCR?

Back in Photoshop 4 (not CS4), the old proprietary colour table method used Medium GCR as a default. In v5 (not CS5), ICC profiles were an option, however the old legacy settings were still favoured by many. In v6 (not CS6), the colour settings were changed to what is basically available today, favouring ICC profiles (with the old Adobe Custom CMYK table method hidden away as an on the fly configurable ICC profile).

PShop is set to its default "North America General Purpose 2" That is used by the newspaper editorial and, as far as I can tell, also by the agencies that supply ads to the paper. Everyone seems to be using PShop 5.

Do you mean that the industry standard specifications that Adobe have created custom ICC profiles for use a GCR similar to the old legacy “Medium GCR in Adobe Custom CMYK”, or that the default CMYK in Photoshop is Custom CMYK using a Medium GCR setting (Meduim GCR is pretty much an arbitrary value which would vary depending upon the profile creation software, however we could presume that it is somewhere in the middle between minimum and maximum GCR values for any given ICC package).

My understanding is that if PShop is left to its default settings then the profile that's used to convert RGB to CMYK uses "medium" GCR.

The results are consistent with tests that I, and the color engineers at Creo/Kodak ran to validate some of the claims made for Staccato FM screening before we went to market. In those days it was CS3's default settings. On press we brought C,Y, and K up to target densities - then, once the press was stable, we brought M up to density and then way over. Measurements were taken at each .05 density point from below to over target density. Unfortunately I no longer have access to those test results.

I have yet to meet a designer or ad agency that has changed their PShop's default color settings. All incoming art, in this workflow, is run through Binuscan to optimize the images, apply proprietary heavy GCR, and dot gain compensation curves.

My goal was to try and instil the notion of "running to the numbers" as they currently do everything by eye and do not include any measurable targets in the presswork. Also, I'm trying to help the press operators get a "feel" for how the imagery reacts to their SID moves - hopefully to avoid their wasting time and materials with making SID moves to "fix" the color.

We'll be running another press test early in the new year. This will be to see how Auraia DM screening reacts to variation in solid ink densities (the last test was done using 110 lpi AM). If you'd like to provide an image that has been separated using different profiles then please send them to me and I'll try to include them. Contact me at pritchardgordon (@ ) gmail (dot) com

best, gordo

 

[gordo]

Gordo,
Your conclusion here is a little like saying that a canary in a coal mine was of no use because the birds died before the miners. A 3C patch is supposed to give you feedback before you have problems. It also gives you a visual tool by which you can monitor press performance. You can see the change in the gray patch whereas you cannot see the density changes in the solid.

You are offering an opinion - but you're not backing it up with data, so, IMHO, it's just repeating an old wive's tale. This industry has a tendency to do that a great deal.
My point is that the 3/C patch react to SID changes that are not worth the press operator compensating for. There is a natural variation in SIDs on every press that I've every met with. That variation will show up in the 3/C patch and be rightly ignored since it does not affect the live image area - and most certainly if the images have been optimized (heavy GCR). I used to believe the grey balance patch story, at least until I actually worked in a printshop and was able to monitor presswork on a daily basis.

In your pics I see the image with the Canadian flag going all over the place. The sky goes purple, the shadows take on a cast, the white area takes on a tint, the building in the background goes pink.

In the unoptimized image, yes. However, we went way outside of industry tolerance to make that happen.


gordo

 

[Lukas Engqvist]

Wow "they currently do everything by eye" …did you talk of the effect that bright orange machinery has on judging colour? It was my first reaction, but then since it was measured colour I thought I wouldn't make a point of it. I wonder what the manufacturer was thinking when making the machine that colour.

 

[gordo]

Wow "they currently do everything by eye" …did you talk of the effect that bright orange machinery has on judging colour? It was my first reaction, but then since it was measured colour I thought I wouldn't make a point of it. I wonder what the manufacturer was thinking when making the machine that colour.

They don't check color at the press. Instead they take the newspaper section into the control room, judge color there and make SID moves accordingly. So the fact that the press is an intense orange isn't really an issue. The fellow at the press is doing a quick check for scumming, set off, registration, etc. before taking it into the control room.

In my experience people tend to continue to do what they've always done - unless something major changes their perception. Even then, unless there is a potential for a real positive impact on themselves they're unlikely to make a change. Often, even if there is a personal benefit they'll continue doing what they've always done. Perhaps it's just comfortable to do so.

One of the things that I'm trying to do is to chart a path for how to quantify, in economic terms, how making certain changes in production can have a positive impact on the business.* Economic terms are good because they are easy to understand and are not subjective. So they are easy to argue for and difficult to argue against. E.g. If a $5,000 purchase has a payback in three months, then every three months going forward returns $5,000 to increase margins or use for other investments. Management (decision makers) can usually understand an economics argument. It's more difficult to convince front line employees (especially unionized ones) as they are typically disconnected from the economics of the business and/or their job descriptions/responsibilities are very narrowly defined. Finding a value proposition to make a change for those folks can be a real challenge.

Thoughts you may want to share and discuss with your students.

best, gordo

* My mantra is "Turn qualitative thinking into quantitative thinking. Then turn quantitative thinking into economic thinking."

 

[Lukas Engqvist]

Ok then I get that then press colour isn't an issue. Those printers I've been to that do the judging at press, some have brightly coloured presses and no measuring equipment.

 

[meddington]

You are offering an opinion - but you're not backing it up with data, so, IMHO, it's just repeating an old wive's tale.

I think that's a misinterpretation of Rich's point.

My point is that the 3/C patch react to SID changes that are not worth the press operator compensating for.

You could just as easily conclude that the measured density deviations from the SID targets are not worth the press operator compensating for. The 3/c gray patch offers a visual (and measurable, but lets stick with visual) "alarm" that the press SIDS are deviating from the norm. The SIDs themselves are not so much visual, but the measurable result of them should be just as "alarming" to the press operator, should he/she be monitoring it. In this case, the visual result of the 3/c patches are no more or less of an "alarm bell" than the measurable SID. Both show a gross deviation from the target, but neither of them are in this case very indicative of the visual result of the live image area. This is a testament to the ink optimization process, but I don't see that it discounts the validity of a 3/c patch for a visual analysis any more than it discounts the measured density of the SIDs. If anything it discounts them both equally.

 

[jstotz]

Gordo, I think that your point that the 3-color gray is too sensitive to density changes has some merit. Although I also see the validity in an early warning that is more sensitive than the page content. I'm curious how you feel about it's use to indicate that the TVI is out of whack even if the SIDs are correct? Indicating that the press is not printing the same way it was when the curves were made. Is that sort of problem more worth the operator's attention?

 

[Alois Senefelder]

Alarm Bells

Alarm Bells

Gentlemen,

I agree with Meddington's comment , on the National UK Paper I worked for in the 1980s, we used the 3/C grey target as Alarm
Bells

We controlled the press print run using SIDs and this was in the days before CTP , the colour reproduction using

Crosfield Scanners and GCR.


PDFs -- that I hope you find of interest.


Regards, Alois

 

[tmiller_iluvprinting]

Great discussion!

 

[tmiller_iluvprinting]

We'll be running another press test early in the new year. This will be to see how Auraia DM screening reacts to variation in solid ink densities (the last test was done using 110 lpi AM).

If it is allowed, please post the results of the Auraia DM screening test, or your observations. I have researched Auraia-II DM screening and it would be great to have a first hand account if it truly delivers on the claims its creators make. Thanks Gordo!
Regards,
Todd

 

[gordo]

[snip] the National UK Paper I worked for in the 1980s, we used the 3/C grey target as Alarm Bells

People do lots of things - that does not mean they are doing the right thing. For decades doctors used to promote smoking cigarettes as a health benefit - that doesn't mean that was right.

PDFs -- that I hope you find of interest.

Print Attributes #1287 GATF Analysis of Print Attributes test form user guide does not list grey balance in the attributes.

Print Attributes #3289 The paragraph describing grey balance only refers to it in relation to the live image area and not to the colour bar or 3/C grey balance patch. It says nothing about using the 3/C patch as a process control metric. So it's not really related to the discussion.

best, gordo