Fingerprinting Suggestions

[meddington]

Conclusion: On press, the 3/C gray patch and live image color do not have a useful or meaningful connection.

Just to illustrate my earlier point, here's a sample of CMYK values taken directly out of Gordo's picture (converted to Coated Gracol 2006 in Photoshop using re. col.), along with a 3/C gray patch (offset of center in the image), with the top half having a 3% increase in magenta TVI. Its most difficult to detect the difference in the deep red and green patches, as the magenta percentages move further away from 50%, where the 3% TVI increase is at its highest, but its still detectable. The eye is most sensitive to the 3/c gray patch, but the difference is easily detectable to the other hue in my eyes. The delta E76 values range from 1 to 2.82 (averaging 2.28 with a standard deviation of 0.67), while the delta E2000 values, more closely aligned with human vision, range from .52 to 3.08 (ave: 2.23, stdev: 0.80). The 3/c gray patch gave the highest dE2000 value, followed by the live-image-gray (3.01), and the flesh tone immediately to the 3/c gray patch's left (2.81). So, this is hardly a formal study, but for what its worth, I feel it shows the 3/c gray patch has a meaningful relationship with these other live image colors.

 

[gordo]

Just to illustrate my earlier point, here's a sample of CMYK values taken directly out of Gordo's picture (converted to Coated Gracol 2006 in Photoshop using re. col.), along with a 3/C gray patch (offset of center in the image), with the top half having a 3% increase in magenta TVI. Its most difficult to detect the difference in the deep red and green patches, as the magenta percentages move further away from 50%, where the 3% TVI increase is at its highest, but its still detectable. The eye is most sensitive to the 3/c gray patch, but the difference is easily detectable to the other hue in my eyes. The delta E76 values range from 1 to 2.82 (averaging 2.28 with a standard deviation of 0.67), while the delta E2000 values, more closely aligned with human vision, range from .52 to 3.08 (ave: 2.23, stdev: 0.80). The 3/c gray patch gave the highest dE2000 value, followed by the live-image-gray (3.01), and the flesh tone immediately to the 3/c gray patch's left (2.81). So, this is hardly a formal study, but for what its worth, I feel it shows the 3/c gray patch has a meaningful relationship with these other live image colors.

First off, I agree with you that large solid fields of color will always show a more pronounced visual change when subjected to varying SID or TVI compared with typical live images. However even with that being said, your seeing a strongish shift (3.08 dE2000) in the 3/C gray in the sample patches you measured would not predict or indicate any meaningful color shift in the live color patches. I.e. some colors shifted by .52 while others by 3.01 and others somewhere between. That, to me, suggests that the 3/C gray can move all over the place while the live image area colors may, or may not shift, depending on content.
Which, IMHO, supports my contention.

best, gordo

 

[meddington]

First off, I agree with you that large solid fields of color will always show a more pronounced visual change when subjected to varying SID or TVI compared with typical live images. However even with that being said, your seeing a strongish shift (3.08 dE2000) in the 3/C gray in the sample patches you measured would not predict or indicate any meaningful color shift in the live color patches. I.e. some colors shifted by .52 while others by 3.01 and others somewhere between. That, to me, suggests that the 3/C gray can move all over the place while the live image area colors may, or may not shift, depending on content.
Which, IMHO, supports my contention.

best, gordo

You mean I spent all that time taking spot readings off of your mug and you still disagree? Fine, be that way. But by your definition, any metric used in the color bar will have the same disconnect, no? Beyond that, though live image colors may not homogeneously move in tandem to the 3/C gray, it would be fairly indicative that the imagery is well balanced when the 3/C gray is balanced, and its the quickest way to indicate an imbalance.

 

[gordo]

You mean I spent all that time taking spot readings off of your mug and you still disagree? Fine, be that way. But by your definition, any metric used in the color bar will have the same disconnect, no? Beyond that, though live image colors may not homogeneously move in tandem to the 3/C gray, it would be fairly indicative that the imagery is well balanced when the 3/C gray is balanced, and its the quickest way to indicate an imbalance.

I am grateful to you for taking the time to do the readings - especially since it ended up confirming what I was saying ) I'd really appreciate it if you sent me the Delta E numbers that you got (in left to right order) for each of the colors as I can use them on a project I'm working on.

On a press, with production work, the 3/C gray being balanced in the color bar is not fairly indicative that the imagery is well balanced. It may do so, but only coincidently. And because, relative to the live image area, it is so easily shifted by insignificant changes in the process it's also not an effective way to indicate an imbalance.

That being said, gray balance is important when initially setting up a press, screening, ink selection, etc.

best, gordo

 

[Deleted member 16349]

That being said, gray balance is important when initially setting up a press, screening, ink selection, etc.

best, gordo

To satisfy the goal of reproducing an image from some target, would not the capability of reproducing the total image correctly be a prerequisite for the application of any so called "gray balancing" to be affective?

The image itself does not have any gray balance. The image is a final version of what some person has felt that the image should look like. It then becomes the target to be reproduced.

I would agree that if one could reproduce the image, correctly at every point then yes, a hue or cast shift of the image could be described as being out of gray balance but if the image can not be reproduced at every point, the idea of having this faulty reproduction in gray balance is misleading. One part might be reproducing the target colour while another part does not. Shift the gray balance then the part that was not reproducing the colour now does.

I think the preoccupation with talking about the importance of gray balance is misplaced. The attention should be with the capability to reproduce the target image and with the consistency of running the printed image so it does not shift.

And this is related to the original topic of fingerprinting a press. If the press is not capable of consistently printing at any point in the printed image and is not capable of consistent running, how can you get a good and valid fingerprint? Oh shoot, I forgot again. Press manufacturers don't think that rethinking the press design is needed. Damn, I kept forgetting that.

 

[meddington]

I am grateful to you for taking the time to do the readings - especially since it ended up confirming what I was saying )

Interesting how two people can look at the same data and draw different conclusions. I don’t see this as a confirmation at all. We agree that imagery can be resistant to deviation based on its make up, but not that there is no meaningful connection. We are talking about an image that has been presented with magenta bias, which is represented in both the image data and the 3/c gray. The “strength” of this bias is to a degree dependent on imagery (how much or little magenta is present in any particular pixel in relation to where the bias is introduced...midtone in this case), and the sensitivity of human perception, but the magenta bias shifts the image both numerically and visually in same direction as the 3/c gray for all colors with magenta values from 1-99%. In the case of the 0.52 de2000 sample, magenta is at the high 3/4 end of the tone scale and only undergoes a 1% change. Observe the gamut plot of your imagery in the attached example which shows the original image colors (shown in blue) and the magenta biased colors (red), which are without significant exception shifted toward the magenta side of the CIElab plot.

I'd really appreciate it if you sent me the Delta E numbers that you got (in left to right order) for each of the colors as I can use them on a project I'm working on.

I will certainly do that offline as now I'm really interested in what you're up to.

On a press, with production work, the 3/C gray being balanced in the color bar is not fairly indicative that the imagery is well balanced.

It’s indicative that the printing units...the individual TVI values that the press was initially adjusted toward, are in balance, and if the separations are based off appropriate press characterization data (also gray balanced) the image will yeild predictable color...provided other metrics are also conforming (solid Lab/SID, etc).

It may do so, but only coincidently. And because, relative to the live image area, it is so easily shifted by insignificant changes in the process it's also not an effective way to indicate an imbalance.

It has been my experienceobserving many different presses that changes to the colorbar are generally reflected in the image area to a similar degree and vice versa. As I stated earlier, a casted print is a casted print. You’ll often see casting in the 3/c gray patch more readily, but this isn’t problematic as if the print and proof are matching, there’s no need to react.

 

[gordo]

@ Erik

You and James Dyson (of the vacuums) have much in common

I think you are quite right in what you say about gray balance and image reproduction. However, for setting up a press, gray balance does has value because it is part of setting up and aligning base inks, TVI, SIDs, ink trapping etc. It is part of setting up a print characteristic. Once that's done, it's value lies in other areas than daily production.

best, gordon p

 

[meddington]

Hi Erik

...if one could reproduce the image, correctly at every point then yes, a hue or cast shift of the image could be described as being out of gray balance but if the image can not be reproduced at every point, the idea of having this faulty reproduction in gray balance is misleading.

ICC Profiles can go quite a long way in describing color reproduction at "every" point. So long as the images are prepped correctly, this generally isn't problematic. Not saying there's no room for improvement, but what do you envision as superior?

I think the preoccupation with talking about the importance of gray balance is misplaced. The attention should be with the capability to reproduce the target image and with the consistency of running the printed image so it does not shift.

Totally agree. Consistency is more important than any particular arbitrary target, gray balance or otherwise. That said, even a consistent device can benefit from a "standard" target to aim at to facilitate a common visual result from press to press & process to process. For example, inkjet proofers can be impressively consistent, but you have to make them look alike.

And this is related to the original topic of fingerprinting a press. If the press is not capable of consistently printing at any point in the printed image and is not capable of consistent running, how can you get a good and valid fingerprint? Oh shoot, I forgot again. Press manufacturers don't think that rethinking the press design is needed. Damn, I kept forgetting that.

It is possible to determine the capability of a particular press and operate it within a window that achieves acceptable, sellable, even high quality results, but your point is well taken.

 

[meddington]

Hi again Erik,

Re-reading my response to your post, its clear that I did not fully address the issue you raised, so I’ll try again...

The image itself does not have any gray balance. The image is a final version of what some person has felt that the image should look like. It then becomes the target to be reproduced.

This is correct. Adjusting “gray” on photography will alter it from its original to a possibly unintended state. You may effectively reduce any overall cast (intended or not), but this will not adjust all colors, as there is no definition of “correctness”. However, assuming the image is the target to be reproduced, and that the image is converted to its appropriate output intent, gray balance now has meaning, and can be a target to correct for deviation...to a degree. For example, tone increase deviation in 1 or more colors can impart a cast, and adjusting towards a balanced gray can correct for this issue. Now if there is a change in ink hue, the definition of a balanced gray has now changed and there will be differences that gray balance cannot account for by itself. To fully correct the image, you'd need a to characterize this new condition These differences might be significant, but the eye is most sensitive to gray, so adjustments towards a balanced gray may visually “improve” the image, but not fully correct for it.

I think the preoccupation with talking about the importance of gray balance is misplaced. The attention should be with the capability to reproduce the target image and with the consistency of running the printed image so it does not shift.

Yes, for the purpose of this thread, the discussion should indeed include capability and consistency, in addition to a target to aim toward, and process control techniques.

 

[Deleted member 16349]

@ Erik

You and James Dyson (of the vacuums) have much in common

best, gordon p

Yes, I know. Hot air.

 

[dimitri]

Assuming that the same ink film is applied across the cylinder, isn't the whole difference in this discussion that the color bar has no GCR in it?

 

[gordo]

Yes, I know. Hot air.

Noooo. Dyson saw that the bags that collected dust in vacuum cleaners greatly reduced their efficiency because the bags increasingly reduced airflow as they filled with dust and debris. So he came up with an alternative that eliminated the dust bag and maintained vacuum efficiency. He approached the big brand vacuum manufacturers - non of which were interested in his ideas (perhaps because they were protecting a $500 million a year in dust bag sales?). He eventually went into business for himself and has sold some $6 billion worth of product since 1993.

best, gordon

 

[gordo]

Assuming that the same ink film is applied across the cylinder, isn't the whole difference in this discussion that the color bar has no GCR in it?

Yes.

Prior to DTP, the methods used for image separations meant that the 3/C patch in the color bar had a much closer relationship with the live image area. However, since the mid 90s and with PShop's default being GCR, as well as with the increased use of extreme GCR reseparations the connection, IMHO, has long been broken.

gordon p

 

[Deleted member 16349]

Hi again Erik,

Yes, for the purpose of this thread, the discussion should indeed include capability and consistency, in addition to a target to aim toward, and process control techniques.

Just to start with I would like to say that I understand that people are now trying to cope with exsting technology and have to do what works best for them. On the other hand, I am an engineer and I am trying to think of solutions to problems and therefore very specific understanding of the constaints is very important to know in order to come up with practical solutions. Nature is not forgiving and if one chooses to ignor what physically is happening, one will be punished with poor results.

Capability is very important and is something that can be measured. Statistical capability has to do with a process's variability and whether that variability is small enough to satisfy the performance goal of the process. The performance goal is determined by setting a tolerance on some measurable parameter.

There are two kinds of capablities that printing should be concerned about.
One is the capability to run a press consistently.
The other is the capability to obtain predicatble colour at any point in an image.

Basically from a statistical point of view, printing is not statistically capable with either consistency or predictability of obtaining point colour.

Statistical capability is basically defined as:
abs value of [the average measured value - the target value] + 3 standard deviations must be less than the tolerance.

As an example Let's say that the density tolerance is +/- 0.05pts.
If one has a target density of 1.30 but the average measured value is 1.32 the difference is 0.03.

Then for statistical capability the standard deviation would need to be (0.05 - 0.02)/3 = (0.03)/3 = 0.01

This is in terms of density and not even measured colour. The same kind of technique can be used to measure the capability of predictable colour at points in the image from job to job.

This is just an example to give you an idea that if someone from outside your industry came and measured your process, they would say that the process is not capable.

I have commented often about the ink feed problem that leads to the lack of capability in consistency of printed density, so let's not go there.

Let's talk about the consistency of predictable colour at any point in the image. If you had an ink jet printer that printed two images. One at the top of the page and one at the bottom of the page. Let's say that the image at the top of the page is always the same image but the image at the bottom of the page is different.

Because ink jet printers are reasonably capable of consistency and predictability of printing colour at any point of the sheet, one would expect that the image at the top would alway look the same and be independent of the image at the bottom. But if you had an ink jet printer that printed the image at the top differently each time you printed a different image at the bottom, you would have a big problem with predictability. How could you profile such a printer with any accuracy.

Well this is the situation with most offset presses. The image on the plate affects how the image on the plate is printed. In the extreme, one can have mechanical ghosting, where the image has taken ink from the form rollers in a way that it affects the inking of the plate in another location.

You also have presses that print a different density at the top of the sheet than at the bottom of the sheet. There is also starvation ghosting where the operator can not obtain a even density across the sheet due to the lack of ability of the roller train to provide an even ink film when certain images are printed.

When you have these kinds of weaknesses in the design of the press, how can one expect to have predictable characteristics that are required for accurate profiles etc. You can't.

A press does not print colour but is supposed to print with consistent ink film conditions on form rollers that result in consistent and predictable printing of solids and screens. The process of printing these ink films, even if they are consistent are not linear and are not independent. This reality means that trying to predict the colour outcome with curves applied to screen data is mathematically not valid.

If you had a press that did print consistently and predictably at all points in the image, then I would agree that an approach like ICC is right. My only concern here is in the number of points to characterise how the press prints. Some years ago I was looking at an inkjet printer and playing around with how many measured points were required to be able to map out the whole gamut in an accurate way. It took about 9000 points which were close enough to each other to make an interation between them that could come close to predicting a colour between the points. That's a lot of points. Automate that kind of method and the accuracy increases as the number of points are increased.

So in all of this discussion and even in Gordon's view that gray balance is needed at the beginning to set up the press, I just do not see it that way. I think in terms of capability of the process to do what one wants to do. I have no idea how one mathematically describes gray balance. I have not seen it described mathematically except in the targets of a few screen combinations. Trivial info.

The comments regarding the faults in the press may sound too grave but much can be done to improve both the capability of consistency and predictabiltiy. How does one change the mindset of an industry from coping with problems to solving problems?

 

[danremaley]

Assuming that the same ink film is applied across the cylinder, isn't the whole difference in this discussion that the color bar has no GCR in it?

Right the color bar has no GCR, in fact some printers still run the color bar 'linear' and the work curved! In this case Gordo is correct it has NO reference to the work. But people, being what they are, still do it.
The reality is that the whole gray balance issue is predetermined in Photoshop when you make the RGB/CMYK conversion the entire image is based on a grayscale that has more cyan than mag/yelo due to the imperfections of the ink, this is true everywhere, throughout the tone scale.
There may not be a 50C/40M/40Y in the image but their are relationships to this neutral value. The reason GREAT printers use this gray patch is that it's the most sensitive to change. Look at a CMC color tolerance chart, these elliptical shapes show the what the observer will accept (i'll attach one). Note that the SID that we measure has the largest shape! Therefore we are measuring the widest tolerance! Note where neutral gray resides - yes in the center with the smallest tolerance!
Also, when changes are made to SID the areas of change are 40-50-60% the de-saturated (10-20-30%) and saturated (70-80-90%) areas don't change.
That's why GCR is such a great process - a 50C/40M/40Y becomes 50%Black with 100% GCR. If it were 50% GCR the numbers may be 25C/20M/20Y/25k making it 4-color but much easier to print! By the way, Photoshops "Medium" default GCR only amounts to a true 15-20% black replacement not 50%!
In my seminars I teach the pressroom that gray, purple and brown are the hardest colors to match, why? Because their values are near the midtone.
Now the images may not be in gray balance, there are no more scanner operators BUT the proof should be in gray balance and therefore the press should match it.
Call me with any questions. . .or "google me".
System Brunner is aware of all this, he called it a 'picture classification" numbered 0,1,2,3 this related how much change (variation in dot gain) a a picture would accept.
0 = will accept no change - a neutral gray patch or a purple (50C/50M)
1= will accept a +/- 2%
2= will accept a +/- 4%
3= will accept a +/- 6% - a red fire truck (80Y/90M)
So if we target the toughest color all the others will be fine. . .see attached pdf.
Dan Remaley
412.889.7643

 

[gordo]

[SNIP}
A press does not print colour but is supposed to print with consistent ink film conditions on form rollers that result in consistent and predictable printing of solids and screens. The process of printing these ink films, even if they are consistent are not linear and are not independent. This reality means that trying to predict the colour outcome with curves applied to screen data is mathematically not valid.
[SNIP]
So in all of this discussion and even in Gordon's view that gray balance is needed at the beginning to set up the press, I just do not see it that way. I think in terms of capability of the process to do what one wants to do. I have no idea how one mathematically describes gray balance. I have not seen it described mathematically except in the targets of a few screen combinations. Trivial info.

The comments regarding the faults in the press may sound too grave but much can be done to improve both the capability of consistency and predictabiltiy. How does one change the mindset of an industry from coping with problems to solving problems?

Amen.

The press operator's job is not to make color but to lay down a consistent film of ink at a specific specific thickness as well as have the ink films adhere to each other so as to create effective light filters. The tools are the press and inks which have been designed (well or poorly) to do this.

To be clear, when I say that gray balance is needed at the beginning to set up a press, I mean initial set up - not set up prior to actual production runs. I believe that it is an important aspect of selecting the right ink, tone curves etc. because gray balance is part of the industry specifications. Also, gray balance is NOT considered when inks are formulated - so part of initial setting up is to identify the appropriate ink series that will be used for production.

I agree that there is a problem with the mindset of many printers, there is often a magical thinking approach to solving problems using a combination of facts, anecdotal information, history, guesswork, all mixed in with a bit of experience. The basic research either hasn't been done, or has been done decades ago when processes and equipment are not as they are today.

As DanRemaley sums it up: "In this case Gordo is correct it has NO reference to the work. But people, being what they are, still do it."

best, gordon p

 

[meddington]

Yes.

Prior to DTP, the methods used for image separations meant that the 3/C patch in the color bar had a much closer relationship with the live image area. However, since the mid 90s and with PShop's default being GCR, as well as with the increased use of extreme GCR reseparations the connection, IMHO, has long been broken.

gordon p

If taking Photoshop's default ICC profiles (SWOP/Gracol/Fogra) into consideration, GCR is rather moderate, only adding about 10 - 15% black with CMY not far removed from 50C, 40MY at all, and quite frankly, a deviation of 3% magenta has nearly as profound effect on these separations as it does on the 3/c gray. Doesn't really sound "broke" to me.

Extreme GCR will certainly create separations less susceptible to deviation, but the there is always the possibility of imagery that contains elements sensitive to deviation similar to the 3/c gray, so I wouldn't completely discount it based on the premise presented.

Statistical capability is basically defined as:
abs value of [the average measured value - the target value] + 3 standard deviations must be less than the tolerance.

As an example Let's say that the density tolerance is +/- 0.05pts.
If one has a target density of 1.30 but the average measured value is 1.32 the difference is 0.03.

Then for statistical capability the standard deviation would need to be (0.05 - 0.02)/3 = (0.03)/3 = 0.01

This is in terms of density and not even measured colour. The same kind of technique can be used to measure the capability of predictable colour at points in the image from job to job.

This is just an example to give you an idea that if someone from outside your industry came and measured your process, they would say that the process is not capable.

I don't want to argue your excellent points, but in regards to the example above, a density tolerance of +/-0.05 pts, while historically accepted, doesn't in and of itself have a direct connection to the attributes of print that really matter to the buyer...what it looks like. For example, with a target magenta density of 1.30, the upper and lower limits are 1.25 and 1.35, and if you calculate the delta E differences between the upper or lower limit, and the target CIELab value, we may only be looking at a spread of 0.50 delta E76, and/or a TVI shift that is negligible (if detectable). So while a device might not be capable of delivering ink film/density within legacy tolerances, it might certainly be able to do so enough to deliver visually consistent results. I'm not saying that tight tolerances aren't desirable or that consistency and capability or not important, but at what point do they fail to add additional value? Similar to proofing systems that can iterate the average result below well 1 delta E...could you really see the difference between that and a proof at 1.5 delta E

 

[Deleted member 16349]

I don't want to argue your excellent points, but in regards to the example above, a density tolerance of +/-0.05 pts, while historically accepted, doesn't in and of itself have a direct connection to the attributes of print that really matter to the buyer...what it looks like. For example, with a target magenta density of 1.30, the upper and lower limits are 1.25 and 1.35, and if you calculate the delta E differences between the upper or lower limit, and the target CIELab value, we may only be looking at a spread of 0.50 delta E76, and/or a TVI shift that is negligible (if detectable).

I agree that density is not colour but it is a process control variable. At the end of my example I did comment that it was not even colour.

The colour difference of the solid density variation does not change that much but I would suggest that the TVI does more than you suggest. It really depends on the strength of the inks and paper etc.

But the real problem comes when printing a screen that has two or more inks. Say you are printing mid tones of Cyan and Magenta. The existing tolerance allows a density variation of +/- 0.05 pts. If during the run you print the Cyan to the upper limit and the Magenta to the lower limit and later in the run you print the Cyan to the lower limit and Magenta to the upper limit, I would argue that the colour will be quite different. My own limited experience and technical papers have shown this to be an issue.

What this says to me is that the density tolerance of +/- 0.05 pts is NOT capable of reproducing colours without visible variation. It means that the tolerance has to be even tighter to obtain consistent printed colour. To get to tighter tolerances, one needs to change the process so it is capable.

The +/- 0.05 tolerance is not tight. It is very loose. It represents about +/- 8% of change in ink film thickness.

I totally agree that the final goal is to get tighter control of variations in what the customer actually buys, which is colour.

Let me also add that printing is not just CMYK. Printing is done for packaging and now for electonics and different inks are used such as metalic inks. This can lead to a situation where one can not use density values to monitor the process. Positive ink feed not only makes the process more capable but also can eliminated the need to monitor the process during the run with measurements.

 

[meddington]

The colour difference of the solid density does not change that much but I would suggest the the TVI does more that you suggest.

Possibly, but maybe not. Tvi based on the murray-davies equation is reliant of solid, screen and paper density. If density deviated +/- .05 at the solid, and we assume a density change of half that amount at midtone (which might be generous) wed likely be looking at a total tvi spread of 3% between the upper and lower limits. Not to say that color wouldnt change, just that tvi might not indicate it as significant.

But the real problem comes when printing a screen that has two or more inks. Say you are printing mid tones of Cyan and Magenta. The existing tolerance allows a density variation of +/- 0.05 pts. If during the run you print the Cyan to the upper limit and the Magenta to the lower limit and later in the run you print the Cyan to the lower limit and Magenta to the upper limit, I would argue that the colour will be quite different. My own limited experience and technical papers have shown this to be an issue.

This depends. A solid ink tolerance by itself isn't enough. A tolerance at midtone, either tvi or gray balance (NPDC), is necessary, and I would argue that NPDC will have more a meaningful correlation to the image than individual tvi values. 1. Tvi is a metric relying on the solid ink density, and as density changes, deviations in tonality can be masked (density of sold and screen both moving up in tandom resulting in minimal tvi shift when larger tonality shifts may exist), as a result, Tvi doesn't directly describe color or tonality, and 2. Individual tvi cannot take wet trap into consideration.


What this says to me is that the density tolerance of +/- 0.05 pts is NOT capable of reproducing colours without visible variation..[/QUOTE]

Again, it depends. As you said, what goes on in the screened area with two or more inks is more important, which is why I advocate the use of the 3/c gray as a monotoring metric...part of a complete process control breakfast.

To be sure though, a +/- .05 tolerance is certainly tighter than a delta E value of 5.


And I totally agree with your philosophy of addressing root cause issues rather than symptomatic issues. Increasing process capability can only help and would certainly lead to reductions of these issues.

 

[Hanson]

Let me also add that printing is not just CMYK. Printing is done for packaging and now for electonics and different inks are used such as metalic inks. This can lead to a situation where one can not use density values to monitor the process. Positive ink feed not only makes the process more capable but also can eliminated the need to monitor the process during the run with measurements.

"Positive ink feed" means " waterless printing", right?