Automatic GCR software

[michaelejahn]

When I first looked into the idea of same quality, less ink use - I worked for a gravure printer.

While GCR was not some magic wand, we did discover that we could re-separate pages (back then, we actually scanned film) and come up a new set of separations that printed better. This was way back in the 1980s - before ICC profiles and CMMs, before Photoshop profiles, before PDF.

So, absolutely you can re-separate an entire PDF - or even just the images in a PDF - so they would still look great, but somehow, use more black ink in the shadow areas (where hues may not shift when you use a different recipe of C, M Y and K to make the same color.

Now, the tricky bit is in the saturated hues. It has always been my experience that as hues become darker, that using Adobe Photoshop to generate GCR tends to muddy things up - so, it is less about using GCR, and more about using the right approach.

IQColour has some nice side by side samples you might want to take a peek at;

IQ Colour, Inc.

enjoy !

Michael Jahn
Jahn & Associates

[Louis Dery]

There are pro's and con's to using a standardized ICC profile.

Pro's: Level playing field. Most users are either going to have a specific "house profile" or will adopt a standardized profile. No "tweaking" of profiles.

Con's: Can't "tweak" profiles.

Personally, I would like to see a standard set of profiles used. But as long as everyone agrees on a test pattern, set of conditions, make the results public (everyone gets fair use rights), then I'm all for doing what ever I can do to get this rolling.

Matt,
Understand what you mean. Just set the rules and let me know when you are ready!
1, 2, 3, Go!

Regards,

Louis Dery
TGLC inc.
TGLC Color Management and Color Control Solutions: PerfX Softwares, Training, Implementation, Certification

[mattbeals]

Well not being a color expert I would hesitate to define a set of criteria. So having said that I'll take a shot at it

I do think that the test patterns need to be standard patterns. In terms of layout of the patterns I would think there would be four permutations. One, the pattern/image separated using the defined and agreed upon ICC profile with no "Optimizations". Second, Black plate of the "Optimized" image. Third, C, M & Y plates. Fourth, a composite of the "Optimized" image. Something we're all familiar with maybe? Ole Morie, musicians, etc. Doesn't matter really, otherwise we end up with a mess of who thinks what's best. Maybe GATF/Fuji/Kodak has a suitable pattern. Again, something everyone has access to.

A single defined stock. Maybe someone at West Coast, Unisource, Xpedx or whom ever want's to volunteer some paper to "get some press" (bad joke... I'm sorry... Kind of...). Or each print facility can be responsible for ordering the agreed upon stock.

A agreed upon grouping of print providers. Maybe a regional thing, bi-costal, international? Any reason this can't be with FOGRA for our international friends too?

A defined scope of press conditions as I said earlier.

A defined set of parameters such as line screen, dot shape, conventional, stochastic or hybrid.

What ever the criteria are they need to reflect what the "average" pressroom conditions would be. Something that everyone can hit so they can see how practical (or impractical) such a solution would be for them. None of this "tweak this, tweak that" crud that "high end" (who *isn't* a high end printer). Run it to the numbers just like any pressman would do anywhere. Real world, real results that really mean something to the broadest possible gamut (again, bad joke) of print providers.

Then once it's all done we make the comparisons available as a PDF and printed samples to all participating vendors with fair use rights for the participants. No "secret sauce", no "voodoo", everything is in the public view. Damn... Am I going Socialist???

In all seriousness though, someone far smarter and more familiar with this than I am will have to define a set of rules that all participants will ratify and adhere to. Otherwise it's just not fair to the end user in my opinion when evaluating such products. Personally I don't see how any one vendor will be "better than the rest". It is such a subjective evaluation. Obviously there will be flaws we will see just as there will be "shining examples". But at least this way the end user can decide for themselves which solution provides the most appealing result to them so that they can intelligently evaluate the options. Good, bad or indifferent results will be reported openly. So if you don't think you can cut it (the ink, again with the bad jokes) with the other solutions, don't play.

Vendors I can think of off-hand: Alwan, TGLC, GMG, CGS, OneVision? Maybe after a nice single malt I will have some more ideas.

[juergenroesch]

About the proposed shootout - I know I threw that one out there and suddenly everyone seems to be excited.

Let's see if we can get it rolling. It's not the first time I do that (usually with printers or prepress about to compare 2 or 3 products they already look at), so I just throw out a few items that I found important:

- images need to be properly prepared for the print standard that's to be tested. This means no SWOP separated images proofed on a relative colorimetric SWOP setup and then printed sheetfed (i.e. GRACoL).

as test images I recommend the reference suite from bvdm (Bundesverband Druck und Medien, Germany; roughly translated: German Printing Association) which you can see at www.roman16.com They are available for purchase (very inexpensive if you ask me) and are of absolutely stunning quality.
People buying them will get a box with 2 DVD containing the images at an incredible 700dpi (so you can resize to whatever you need without loss due to sampling) in 16bit ECI-RGB, 8bit ECI-RGB and 8bit ISOcoated v2 (Fogra39L). Profiles are embedded and also open on the disks. Also along comes a printed magazine which contains the printed results to 39L with reference color values and Lab numbers for what those values are to read on a print. Have a look at it - you'll like it. While each image has an individual color theme, all contain skin colors and the same neutral gray table top, making them perfect for comparisons.

So the older SWOP and olenomoire images are in my opinion outdated and have the added problem that some have been separated or created in times when separations were made for dirty inks on low grade paper with conventional film and dot gains approaching 25%. Paper, ink and CTP technology has improved then by leaps and bounds, so why keep using images that are not up for it.

I also like using a test element from GMG (GMG_CMYK_smoothCheck_V1.tif) which is incredibly helpful in analyzing flaws in an engine. To test it, open it in PhotoShop, convert it to any other CMYK and then look at what happened to the channels. What you're now looking at could scare someone off any reseparation for life. So don't be shocked!
(Never turn the file to JPG - the jpeg algorithms would make it lose many of its functions)

The task of a good GCR tool is to maintain the primaries and secondaries as they don't contain any 3rd color which can be substituted, but also to not touch the black where it is on its own (you don't want K-only images to turn 4c) and then to gradually substitute redundent CMY with K. This should maintain the relativity of colors in the color space - or in understandable terms: keep blends as blends and not turn them into rollercoaster rides.

Now we found that some tools do not follow these rules and can introduce K into C+M also they shift colors in the gamut a bit where it is hardly noticable in order to get more replacable CMY values. You can clearly identify such behavoior with the attached file. Visually on a proof you normally would not or hardly see it. On press it might remain hidden as well. But like skeleton black printing such prints would have troubles if densities of the one or other channel shift - and if registration is just a bit out grays can start to shimmer and look iridescent. Eactly what you'd not expect from a GCR.

Here a list of items you need on the page:
- vibrant CMYK images with CMY content (skeleton black)
- CMYK images around the neutrals (isn't there one image with gray rolls of fabric and red velvet on a chair?)
- true K-only picture
- primary and secondary only image (just reds, greens, blues - no photo, just a made up TIFF - which should not see alteration or color shifts)
- vector artwork side by side with same colored imagery. You see today very often background boxes color matched to images - how well is that retained? Place a "dirty colors" color bar as TIFF on the sheet - then create in your layout software boxes using the same colors and line them up - no point in using your regular press bar, but both the IDEAlliance and FOGRA strips contain a few tertiary colors, but you can just as well make up your own
- any artwork color elements using standard print effects like overprint, knockout, trap, color text... whatever - basically image elements which will test the ability of the used tool to retain the ability of the software to keep the document structure
- for good measure throw in a 400% Siemens-Star (that's the one with alternating color sectors, often used for registration and film resolution testing) and see how it works before and after
- color text elements

Once the file has processed through any software bring the result back into your workflow and see if you can still work with it or if the internal structure has altered and you now can no longer set trapping.

Therefore using only a single TIFF on a page would be a test content 10 years ago, now the tools have to work with the workflow.

If a color accurate proof (for the print standard to which is printed) got supplied the converted file should display no alteration to the visual appearance of the new proof.


When the test is finally put on press have the press operator grade the ease of make-ready. How fast can he achieve a press to proof match? How stable is the document throughout the run of a fixed amount of sheets - compared to the original version of the document and to the other competitors in the test? Does the press operator have to readjust? How did he see the need to readjust?
(Surprisingly the best tool might suffer the highest density fluctuations, but without the press operator noticing it - in other words: such a document would be very forgiving)

Finally how does the job respond to intentional density modifications? When you run the K up - or if you have to boost for a client the "reds" (= give M+Y a push)?

A viewing booth comparison likely might reveal less than the experiences the press operator made. But still there might be surprises as well.


-----------

on a side note: if you try to stick to standards, you create a more level playing field. If you require custom profiles to be used, then some custom profiles might have already built-in flaws, so you'd also test the ability of those tools to edit or fix these flaws in the profiles first, dividing the field further. This might be a real-life requirement, but I'm not sure if I'd dare to throw this variable into the mix here as well.

But why not add it as a separate test, where participants get a home-made press profile and are to build a good reseparation profile for that setting. Just processing the aforementioned smoothCheck file through each tool with the profile they just built (in some products the profile gets created on-the-fly in others it needs to be externally built) and then comparing the channel results in PhotoShop should display strengths and weaknesses.

I would not recommend proofing that file through the ICC profile for a simple reason: if the profile is flawed, then a separation might build those flaws right into the separation which when proofed through that same flawed profile will appear to be perfect because the notches of each line up like a key in a lock. As these flaws are most likely the result of a snapshot of the press at one given - not optimal - time, this same notch pattern might not work on the press ever again. Therefore you're saver checking in PhotoShop or run the job actually on press - but don't trust the proof (as this will accurately stick to that singular snapshot)


Did I miss something?

[kaiserwilhelm]

Did someone mention that the file should be some complicated thing - it, live transparency?
Not that the test is being done for me, but shouldn't we try to make sure that each piece of software can handle APPE pdfs? How does it help to find out that software company A can save me 21 percent ink (and look the best) if they flatten my pdf before they perform their function?
BTW, I love the idea of a real life test. Kudos to Matt.
Kaiser

[mattbeals]

Your right, it would be great if a test form could include such intricacies.

[Hopkins Printing]

to Jon Morgan:
color problems should be fixed in prepress. Attempting color management on press is waste bins full of make-ready paper. And it's time consuming. Successful printers streamline their processes by capturing and tackling color problems in prepress, so the press operator only has to mount the plates and "press the print button".
Because the gray balance is much more pre-determined the press operator has usually less balancing to do and gets acceptable results with less than half the make-ready sheets of a normal run. Probably only 20% of the sheets of a run where you want him do the color management on press.

Hi juergenroesch. My apologies for the confusion. I was not suggesting that color problems be fixed at press. I was just pointing out that if GCR is applied too strongly thus leaving only the bare bones minimum CMY, it may not leave much latitude for pressmen to push the CMY if they would need to.

Cheers,
Jon Morgan :-)
Hopkins Printing

[juergenroesch]

Hi juergenroesch. My apologies for the confusion. I was not suggesting that color problems be fixed at press. I was just pointing out that if GCR is applied too strongly thus leaving only the bare bones minimum CMY, it may not leave much latitude for pressmen to push the CMY if they would need to.

Cheers,
Jon Morgan :-)
Hopkins Printing

Hi Jon,

actually it might surprise you, but the pressman now precisely can PUSH CMY if they want to, e.g. to achieve a punchier red WITHOUT having to worry too much about the overall tonality or gray balance. Because these 2 are now save on the K plate.

What they WILL LOOSE is the ability to mess around with the gray balance though. Not much of a loss in the big picture if you want my honest opinion on that one.

Also repeat prints [someone dared to call them "bulletproof"...] become easier as the gray balance and tonality is what print customer pick up on first ...


As for my earlier response to your post: we all know from contributing to forums like this that if something isn't precisely worded, it can be easily misunderstood - or understood in a way the author didn't have in mind.

[pnkearns]

Hello all,

I have read the posts on this thread and would like your help to settle a debate I'm having on when GCR/UCR can be applied, as well as better understand how to calculate the economics of using GCR/UCR technologies.

Applying GCR Debate
I'm a newbie to the print world. My friend is a 20 year prepress veteran. My friend's experience is primarily in programs like PhotoShop, InDesign and Quark, but has experience in print workflows.

When I read about GCR/UCR tedhnologies, it appears to me the GCR/UCR formulas are generic to anything using CYMK inks. My friend insists that GCR/UCR is unique to images being converted from RGB to CYMK color space.

"Retain black (GCR/UCR) is something that happens when the RGB image passes through L*A*B where the black channel that did not previously exist is created before it goes into CMYK. You cannot apply retain black to an image that is already CMYK. To do that, you would have to convert the image from CMYK back to RGB, then BACK through L*A*B to CMYK again. "

I cracked open my Harlequin 8 RIP and ColorPro manuals, and found GCR/UCR in the RGB image output rendering intent. This would imply my friend is right. An Onyx 7 (JAWS based) RIP manual has "ICC Setup" and "Advanced Black Generation" setting options for CYMK image, RGB images, CYMK vector and RGB vector. This would imply my friend is wrong. I spent time on Google and the examples of GCR/UCR always match my friend's view, associating the GCR/UCR process to when RGB images are converted to CYMK image for printing. Finally, the examples at IQ Color, Inc., posted in this thread earlier by Michael Jahn are all RGB --> CYMK, even though he implies he has done some CYMK --> CYMK work. So my friend may be right or wrong.

Can GCR/UCR be done on an existing CYMK image to reduce ink?
Can GCR/UCR be done on either RGB or CYMK vector data?

If there are examples or a write-up on doing this, I would appreciate knowing.

Economics of Using GCR/UCR Technologies
In calculating the ink savings of GCR/UCR on an individual print job, it seems you would:
a. Calculate percentage of print job made up of RGB images (or objects that can use GCR/UCR)
b. Calculate percentage of ink saving for each image
c. Sum up the deltas of (b)
d. Subtract other overhead production factors such as any increase in job setup time, and add from items as reduced drying time.
Is this a reasonable summary or am I missing a major factor?
Has anyone seen a good tool that will estimate this automatically?

Thank you for your time. I look forward to any response.

Regards,
Patrick Kearns
patrick@niyi.org
Niyi.org (a future graphic arts online community web site)

[gordo]

RE: Can GCR/UCR be done on an existing CYMK image to reduce ink?
Can GCR/UCR be done on either RGB or CYMK vector data?

Yes to both. Typically the reseparation is accomplished with a device link profile; CMYK direct to CMYK

The amount of ink savings is dependent on the original type of CMYK separation (which varies, e.g. could already be a GCR separation ), its TAC, vs the amount of GCR applied in the reseparation and its tack vs the image content and the total image content of the page or press form.
There is an excellent little, free, utility here: TGLC - PerfX Device Link� - ICC DeviceLink Profile - Ink Saving - CMYK Conversion - ICC Device Link Profile that can give you an idea of savings using your own images. Some press ink key presetting applications can also give you that information based on the image bitmaps. Even PShop can give you an idea by comparing two versions of the same image.
You should also have an idea as to how much ink, in pounds, you are currently consuming on average. That will give you the cost you have been paying, which in turn will give you an idea of what savings you can expect going forward. There's quite a range of % savings that vendors are claiming, so I would go conservative, i.e. what if we saved 5%, 10%, 15%, 20%, or 25%? A % of about 15% savings is realistic for an ROI - you'll likely do better.
Other benefits of GCR can include; faster makeready, less color variation when SIDs vary, less web growth, less visibility of misregistration, faster drying, more uniform color response on press since all separations have been brought to a common denominator.
Potential downsides:
Heavy GCR heavily depends on the quality of your black printer/press unit. Your presses may not be capable of actually using heavy GCR without some kind of technical optimization. You may even have to change ink sequence.
Reseparating CMYK images may allow your customers to hold you liable if they do not like the presswork color.
Possible introduction of subject moiré.
Dirty/grainy pastels.
Dull/weak shadows (loss of contrast)

There's a very wide variation in prices, features, and implementation options with reseparating.
Some things (among many) to consider when evaluating the offerings available:
Is the process accomplished automatically within the workflow?
Is the process accomplished outside of the workflow in a separate box?
How are single color and two color images handled (i.e. does a vignette of 100-0% cyan maintain its integrity)
Vector vs raster handling.
Is the device link profile based on your presswork or on a generic industry standard (e.g. GRACoL)?
Can you tweak/customize the conversion or is it fixed by the vendor?
At what point in the workflow does reseparation occur - early or late or is the choice yours.
Is the solution complete from one vendor or do you need to put the parts from different vendors together yourself.
What press-side support is provided by the vendor?
If you search around the forum you'll find more info including a test image you can use to help evaluate the vendor you are considering.

best, gordo