Poll - Do you linearize your plates?

[John W]

Gordo,
You are correct that we are kind of using press curves on top of a linear curve but we're simply arguing semantics. We use these on-top-of curves at the RIP to correct for items that are not specifically for the press but for issues such as legacy work (where film was used and now we must match the old result via ctp), linear (new jobs were the basic linear curve is sufficient ie: no curve added on top), substrate curves (for example, if the client changes from virgin board to recycled board we add an on-top-of "scoop" curve to attempt to match the previous result only now printed on recycled board), G7 (when the client demands G7 specifications). We also have on-top-of curves for lamination effect such as on credit cards since the lamination of the credit cards is so hot that printed colours darken and shift as they get "cooked" during lamination. I do not consider these extra curves, press curves, but you might.
John W

 

[destryer]

Linearizing plates

Linearizing plates

We have 2 CTP devices for newspaper production . We average the measurements and create a single linear curve, followed by a press curve.

Like Sandpiper said we also are a newspaper with 2 CTP devices and use the same media in both devices but the lasers are in different places in their lifecycles. We linearize them to a known curve to make them identical. Adding a linearization curve removes the plate equipment processors, and media from the press gain and color management issues if prepress conditions change. We also average our 4 presses curves and all our inkjets to make them match each other. Not sure how that makes it more complicated, it gives us more control in each step of the workflow to deal with a problem in a specific area.
Monday morning exposure test to see where the lasers and processors are, tweak the curve and we are off and running.

 

[gordo]

Gordo,
You are correct that we are kind of using press curves on top of a linear curve but we're simply arguing semantics. We use these on-top-of curves at the RIP to correct for items that are not specifically for the press but for issues such as legacy work (where film was used and now we must match the old result via ctp), linear (new jobs were the basic linear curve is sufficient ie: no curve added on top), substrate curves (for example, if the client changes from virgin board to recycled board we add an on-top-of "scoop" curve to attempt to match the previous result only now printed on recycled board), G7 (when the client demands G7 specifications). We also have on-top-of curves for lamination effect such as on credit cards since the lamination of the credit cards is so hot that printed colours darken and shift as they get "cooked" during lamination. I do not consider these extra curves, press curves, but you might.
John W

Not semantics. If you only use one curve - whatever that curve is - even if its to linearize the plate, then IMHO it's a press curve.
If you use two curves, one to linearize the plate and then another applied after the plate linearization curve to achieve the presswork tonality that you want, then IMHO you are using two curves; a plate curve and a press curve.

My argument (explained over several posts) is that applying two curves - a plate curve followed by a press curve is a redundant waste of time for 99.99% of cases. Even though it appears that some 70% of printers do that. IMHO it is an unthought through holdover from a film workflow.

best, gordon p

 

[gordo]

OK, I'll take a different tack in this thread....

For the 70% of printers that linearize their plates before applying a press curve, how much variation in plate imaging are you experiencing?
Put another way - when, why, and how often do you have to modify the plate curve to restore linearity? Is it only when you change plate batches? Is it related to processor chemistry age? Or...?

thanks for your input, gordon p

 

[maxon]

Gordo,

I am also an advocate of linearizing the plate before going on press. One thing that struck me is the entire thread seems to be centered around plates other than violet photopolymer. Ozkan was the only one to put a finger on this :

Ozkan Hangisi : especially if the customer is using Photopolymer plates (negative), linearization is a must. Because without any plate curve, 50% area becomes 60%-65% with ideal expose power and it may harm smoothness of degrade on printed sheet.

Negative photopolymer emulsion types have a significant dotgain which I strongly believe needs to be addressed before going on press. In uncalibrated mode depending on the actual imaging device being used halftones above 90-ish percent may fill in, readings from a press sheet are impossible. You will immediately say the same applies for a plate, don't you see a full solid beyond 90, isn't it the same thing ? Yes and no. In short my theory is :
1. you have more chances to better control a process by using short feedback loops rather than encompassing multiple process steps in one go. It's often risky to rely on taking readings from a press sheet and use them to build a calibration curve in the rip because of the so many variables involved.
2. A simple matter of who is responsible of what - the finished product of the prepress dept. is the plate. By using plate linearization, prepress people have means to demonstrate to pressmen the plates are being exposed and processed correctly based on colorbars saying I've prepared good quality plates for you to use, from this point onwards it's your job to put the right colors on paper. People tend to keep things simple.
3. Prepress studios selling exposed plates to a variety of printing shops will always find difficult to build reliable press curves into their rip as they don't have much knowledge or control of what's happening in so many pressrooms. The only reasonable way to control the quality is by delivering linearized plates to their customers.
4. A practical reason, true for violet photopolymer technology at least - one has to prepare a linearization curve for each set of variables i.e. plate make/brand, laser exposure value and imaging, dot shape, screening frequency and device resolution. This alone would result in having perhaps 20 or more curves just to linearize the plate output. In most cases prepress people will reduce it by using one specific dot shape for all jobs, one specific type of plate and trying to limit the addressable resolutions and required lpi to minimum required. Compare this to the press curve strategy having 15 times more variables to consider, how many curves one has to create ? Let's put it this way, there are just a few printing houses able to entirely standardize their process, the vast majority still use their offset machines as color correction devices relying on pressmen to tweak the settings. It's a jungle out there. The cost factor is important, it's virtually impossible to fully control the pressroom variables these days, who can afford to purchase the exact same specific paper and ink. Given the economy situation most will choose whatever ink, fountain additives, paper, blankets a.s.o. are cheaper and/or available, even pressmen. Considering the above how safe is the single curve approach ?

For the 70% of printers that linearize their plates before applying a press curve, how much variation in plate imaging are you experiencing?

None. Once set, these curves are/should not be touched for years, at least in the violet photopolymer world that I know. Prepress people only have to ensure the laser intensity is set according to plate sensitivity/batch and the processor machine is in reasonably good shape i.e. brushes, cleanliness and replenish system working as expected. They simply need to make sure all hardware works within the safe zone, within specifications and the developer is changed at the specific time intervals. Chemistry is always ready for use regardless of make/brand afaik, they can't go wrong with that. The colorbars are regularly checked with a platereader to prove the plate is linear. They prefer not to care what's happening on the press floor. I have myself tested the linearization curves over and over again and found no need to change them.

 

[Bret Hesler]

We have been tracking our (thermal) plates for the last 3 months with the chemistry changed about every three weeks. Using an ICPlate2 , measuring a 300LS, round dot, unlinearized patch, all the readings are within 1% of the average (standard deviation of .37%).

In my experience, the square dot, 175LS patch in some control targets does not correlate well to what is happening with higher line screens and stochastic. I had two identical platesetters, next to each other with the same plate lot loaded, and had 0.5% difference between them with the 175LS Square dot plate wedge. Thought that was acceptable, until I imaged a plate with 200LS, 240LS, 283LS and 340LS round dots and measured them. With each step up in LS, the difference between the platesetters grew, ending up with 3% at 340LS.

Bret

 

[gordo]

@ Bret: "Using an ICPlate2 , measuring a 300LS, round dot, unlinearized patch, all the readings are within 1% of the average (standard deviation of .37%)" That's great. Readings like that at 300 lpi indicate very consistent plate imaging. If the lpi was lower, and more typical, then the consistency plate to plate would be even better. If it wasn't so consistent then it might be argued that applying a plate curve before a press curve might have validity. However, with the consistency you're achieving (which I don't think is very unusual) - there is no need to linearize first.

@maxon I'll try and answer your points.

1) You wrote "It's often risky to rely on taking readings from a press sheet and use them to build a calibration curve in the rip because of the so many variables involved." I don't agree. Building calibration curves for the RIP based on reading press sheets is standard practice. It is essential to methods such as G7. It works very well. That being said - if you do not have a reliable, consistent, printing process then you cannot use curves to fix press room troubles.

2) I totally disagree. It is not the role of prepress to simply throw linear plates into the press room and expect the press room to "put the right colors on paper." It is the responsibility of prepress to work with the press room so that the plates are optimized for the press in order to deliver the color that prepress has defined with the proof. I agree that "people tend to keep things simple" which is why I advocate a one curve approach (press curve) rather than a two curve approach (plate curve followed by a press curve).

3) Well, prepress studios selling exposed plates to a variety of printing shops need to talk to their customers about what their specific customer needs are. If the customer requires a linear plate then that's fine - apply a linear press curve to the uncalibrated plate and be done with it. If the customer needs are for curved plates then do the appropriate tests, build a press curve and apply it to the uncalibrated plate and be done with it.
Controlling the quality is NOT delivering linearized plates to their customers. Controlling the quality IS delivering consistently imaged plates to their customers - curved or not.

4) I'm not quite sure what you're trying to say.
Most shops use one plate technology and one screening method. Therefore, at the minimum, one curve is needed - a press curve. What you seem to be saying is that they should make this more complicated by also using a linearization curve? Most shops using one plate technology and one screening method will have three press curves applied to the plate: one for gloss coated, one for uncoated, and possibly one for matte coated. If the shop wants to save money, it is much, much, less expensive to use curves to optimize the plates for the press room than it is to use a press as a color correction device.
I agree that the cost factor is important - that is why it is important to understand and standardize the process rather than arbitrarily, apply a linearization curve.

You say: "Once set, these curves are/should not be touched for years, at least in the violet photopolymer world that I know." That's great. It speaks to the stability and consistency of the plate imaging. It also eliminates one of the arguments for applying a linearizing curve before applying a press curve.
With a plate that has only had a press curve applied, then QC is very simple, you can measure the color bar and the tone percent for each patch should have the value that the press curve has requested.

When you say: "They [prepress] prefer not to care what's happening on the press floor. I have myself tested the linearization curves over and over again and found no need to change them." You are describing a very poorly managed shop that is not maximizing its profit potential, nor is it likely to meet print buyers expectation.
Most printers that I know cannot afford to operate that way.

BTW, because you only apply one curve to the plate - a linearizing curve - then I consider that you agree with me since your linearizing curve is basically a press curve and you have a one curve workflow. Not necessarily the best press curve - but it is a press curve none the less.
The other option is to apply a plate curve to linearize the plate and then apply a press curve on top of that initial linearizing curve. That two curve option is what I disagree with.

best, gordon p

 

[Bret Hesler]

... - there is no need to linearize first. ...

Oh yeah, forgot to mention, I am part of the "don't linearize first" crowd. However, I do see some validity in the situation where you have two different imaging devices. That was the case at my previous employer, but we just modified the curve built on the two identical machines to work on the less-used, different machine.

Bret

 

[mglouis]

We use multiple platesetters and they all print differently in their native states - especially between thermal and violet systems. Because we linearize the plates, we can use the same press curves for all platesetters. If we did not linearize plates, each platesetter would have its own unique translation of each set of press curves.

 

[gordo]

We use multiple platesetters and they all print differently in their native states - especially between thermal and violet systems. Because we linearize the plates, we can use the same press curves for all platesetters. If we did not linearize plates, each platesetter would have its own unique translation of each set of press curves.

If they all print differently in their native states doesn't that mean that you need a unique linearization curve for each platesetter before you can apply your single press curve?

If that's correct then you are no further ahead and have one more curve than you need in your workflow.

best, gordon p

 

[RickS]

If they all print differently in their native states doesn't that mean that you need a unique linearization curve for each platesetter before you can apply your single press curve?

If that's correct then you are no further ahead and have one more curve than you need in your workflow.

best, gordon p

Wouldn't that mean more press tests and that expense and down time? No press time with linearization. Seems way ahead for me.
Rick

 

[meddington]

Wouldn't that mean more press tests and that expense and down time?

One press test should be sufficient. The goal would be to ensure the same plate dot is sent to the press from both CTP devices, and in my opinion, ideally using as few correction factors as possible. So augmenting the press curve for the differences between CTP-a and CTP-b should be sufficient, and allows for 1 correction for each device. Perhaps its more intuitive to utilize linearization curves on both devices, and one press curve for both, but I think the potential for banding is increased.

 

[Rick@IPCO]

Just throwing my 2 cents in... We are fortunate in that we have a quality, consistent platesetter with the correct wattage setting for the type of plates we use (also, high quality and consistent. I won't mention brand names here because that's not the point and and I don't want this to look like a product endorsement ad.) Because of these factors, and, I'm guessing, some luck, our linear plates are actually linear; a 50% imaged dot equals 50% and so on, with a 1-1.5% variance here and there at other percentages. We also only have one platesetter, which, pertaining to this particular argument, makes things easier. We have several different presses and use a variety of types and grades of stock on each press, which means we have more than a couple different sets of press curves. Like I said, we are lucky in that our platesetter is linear. If for some reason our platesetter stopped being so and drifted to be 3 or 4% out and the wattage couldn't be adjusted to bring it back in line, I would much rather make one linearization curve to bring it back in than change all of my press curves to make up for it. The curve is a setting that is applied once in my workflow system and left alone after that, it's not something that I have to do manually every time, so it's not extra work when we are making plates. And if we did have two platesetters imaging slightly different, I would want them both starting from the same known state: linear. Why would I want to have to adjust two sets of many press curves (one for each platesetter) when I can just make a linearization curve and leave all of my press curves alone? Let's say we need to replace our platesetter with a newer one (and we will someday). What if it doesn't image exactly the same as our old one? Hmmm, should I apply a linearization curve or change all of our press curves? Or what if it doesn't require a linearization curve but our old one did towards the end of its life cycle? My point is that if I'm using press curves, I would rather have them always start at the same starting point if I can manage it.

 

[mglouis]

Hello Gordo,

You are correct. It is a pleasure to banter with you. I have been a fan of your postings for many years.

As you know we linearize plates, then apply press curves on top of this condition. At this point in time we do not have a process for updating press curves based on CtP drift but this is subject to change. I imagine this process would involve offline calculations, perhaps in Excel, because our Rips and desktop software lack the facility to to do this as far as I am aware.

We manage 14 unique press curve sets: uncoated and gloss for each of seven press groups in three cities, each of which has their own CtP devices. 4 of these press sets output to 6 mil plates and 10 to 12 mil. We manage 7 unique plate media curves: CityA has 2 devices and outputs 6 and 12 mil plates; CityB has one device for 6 and 12 mil; City3 outputs only 12 mil.

Only two cities send plate data. They each store all curves sets, effectively duplicating the press curve count to 28 (14 unique X's 2) and the Ctp linearization curve count to 14 (7 unique Xs 2). Remote plating is performed via 1-bit tiff.

Under the status quo, we are able to train people without previous graphics arts experience to maintain linear states for all platesetters. The first plate of each day is measured and the actual dot area is compared to the expected values, which vary depending on the press curve applied to the plate. If the deltas are greater than 1% in highlights or 2% anywhere else, we scrap the plate, linearize the device, than begin making plates for the day. All seven CtP curves are checked once per day. Drift is nearly non-existent on some devices and rare on the others.

Under a natural-ctp-state workflow, on a morning when a CtP shift is detected, rather than updating two plate curves (1 unique X's 2 cities), I would need to update either 4 or 10 press curves (8 or 20 after duplication across two cities), depending on whether or not the plate in hand is 6 mil or 12 mil. If we were to do this, we would still be applying another curve to to the previous curve to make the new curve. The Rip might only have one curve, but there would indeed be a second curve which was applied on my calculator.

My head spins thinking about implementing a one curve workflow which is really a many curve workflow. Each time I would modify a press curve, I would have to make four curves rather than one, even if the CtP devices never drifted -- one for each platesetter. I agree with others that in theory making press curves on a raw plate is better in terms of quality (simpler math) but for me it does not warrant the extra stress of maintaining them.

I think the platesetter linearization curve workflow will continue to be simpler and less expensive to manage as our business scales up.

- Matt Louis

 

[gordo]

[SNIP]Because of these factors, and, I'm guessing, some luck, our linear plates are actually linear; a 50% imaged dot equals 50% and so on, with a 1-1.5% variance here and there at other percentages. We also only have one platesetter, which, pertaining to this particular argument, makes things easier. We have several different presses and use a variety of types and grades of stock on each press, which means we have more than a couple different sets of press curves. [SNIP]we are lucky in that our platesetter is linear. If for some reason our platesetter stopped being so and drifted to be 3 or 4% out and the wattage couldn't be adjusted to bring it back in line, I would much rather make one linearization curve to bring it back in than change all of my press curves to make up for it.[SNIP]And if we did have two platesetters imaging slightly different, I would want them both starting from the same known state: linear. Why would I want to have to adjust two sets of many press curves (one for each platesetter) when I can just make a linearization curve and leave all of my press curves alone? Let's say we need to replace our platesetter with a newer one (and we will someday). What if it doesn't image exactly the same as our old one? Hmmm, should I apply a linearization curve or change all of our press curves? Or what if it doesn't require a linearization curve but our old one did towards the end of its life cycle? My point is that if I'm using press curves, I would rather have them always start at the same starting point if I can manage it.

You are neither lucky or unlucky that your plate/CTP combination is inherently linear since it doesn't matter. Some combinations simply are linear. As a result you are inadvertently only using press curves without first applying a linearizing curve. AFAIK, no engineer would apply a plate curve to compensate for a laser plate imaging issue. When a system is installed the engineer sets it up to create a robust image on the plate that is free from artifacts like laser swath banding. They don't set it up to be linear, even if their set up results in a linear plate.

If you have two platesetters with two different native states, then in a linearize first workflow you need two different linearizing curves then one press curve per print target (most shops only use one press curve, some have two, and very few have more than three). If you didn't use a linearizing curve first then you would only need one press curve for each device so that plates coming off either device will print the same. Put another way the two press curves are equivalent to using two linearizing curves except that there is no need to add a press curve. So what you would do is use press curves the same way that you are currently using them. The curves would just be slightly different than the ones that you are currently using.

The plate always starts at the same point. Whether it happens to be linear or not. It starts with a characteristic response curve. As long as that characteristic curve is consistent then you can apply press curves to create the press response that you want.

If the plate is not consistent then linearizing it is applying a curve. However you cannot build curves for an inconsistent device. You'd have to measure each plate and build a new compensation curve every time you went to image a plate, which wouldn't make sense.

best, gordon

 

[gordo]

Hello Gordo,

You are correct. It is a pleasure to banter with you. I have been a fan of your postings for many years.

I wish you'd have stopped right there! LOL

Actually, in your specific case, unless I'm missing something, I can understand why you would linearize first and then apply a press curve. As they say, you may be the exception that proves the rule.

But, for the vast majority of printshops, using one curve only rather than piggy-backing two curves makes the most practical sense.

best, gordon p

 

[Doctor_Blade]

If I am using a thermal CTP, I may bypass linearization. If I am using a violet CTP then I must linearize my plate.

 

[gordo]

If I am using a thermal CTP, I may bypass linearization. If I am using a violet CTP then I must linearize my plate.

And the reason why?
Can you elaborate?

gordon p

 

[Slyoder]

This has been a very interesting thread. We are installing a new CTP device in a couple of weeks and I had been tossing around the idea of creating a linearization/plate curve first before I recreate my press curves.

My old device did not have a a linearization/plate curve applied, just my press curves. I'm thinking had I used a linearization/plate curve on my old device I could now create one on my new device, add my existing press curves, and be ready to go. It sounds easy and good but is this maybe not feasible in real life?

I can see the point of not having a curve on top of a curve and how this could be detrimental to qualtiy. And I realize that the final dot on the plate (after it's been influenced by all the curves) is what really counts. Maybe I'm missing something here but when talking about running a couple of different devices or in my situation switching devices it seems to me it would be much less work to get and keep the devices all in a linear state (50%=50%) with a linearization/plate curve and then have only one press curve per substrate that would be applied on top of the plate curve of each different device.

When I think of doing a linearization/plate curve I think of measuring only one channel, let's say for example 10 patches = 10 measurements. When creating a press curve I think of creating a curve for each color 10 patches each = 40 measurements.

Gordo, I am not at all trying to dispute your views, I just thought maybe there's something more involved here that I'm not seeing.

Steve

 

[gordo]

[SNIP] We are installing a new CTP device in a couple of weeks and I had been tossing around the idea of creating a linearization/plate curve first before I recreate my press curves.

My old device did not have a a linearization/plate curve applied, just my press curves. I'm thinking had I used a linearization/plate curve on my old device I could now create one on my new device, add my existing press curves, and be ready to go.[SNIP]

I can see the point of not having a curve on top of a curve and how this could be detrimental to qualtiy. And I realize that the final dot on the plate (after it's been influenced by all the curves) is what really counts. Maybe I'm missing something here but when talking about running a couple of different devices or in my situation switching devices it seems to me it would be much less work to get and keep the devices all in a linear state (50%=50%) with a linearization/plate curve and then have only one press curve per substrate that would be applied on top of the plate curve of each different device.

When I think of doing a linearization/plate curve I think of measuring only one channel, let's say for example 10 patches = 10 measurements. When creating a press curve I think of creating a curve for each color 10 patches each = 40 measurements.

Because your old device did not have a a linearization/plate curve applied, just press curves (bravo), you just need to make a new press curve that makes the tones on your new CtP/plates match the tones of your old CtP/plates - no pre linearizing required. If you had used a linearization/plate curve on your old device you would have to create a new one on your new device, then add your existing press curves. So, in this case prelinearizing and applying only a press curve amount to the same amount of work since in both cases you need to determine the natural uncalibrated state of the device in order to build a linearizing curve or to create a press curve. The only difference would be that if you after you do a linearizing curve you then have to apply a press curve - i.e. you add complexity without any benefit.

The same thing holds true if you have multiple devices with difference output responses. Linearizing adds an unnecessary step - the more devices the more unnecessary steps.


If you need different curves for each color then your workflow choices are:

1) Create a prelinearizing curve. Create and apply press curves for each color.
(Linearize = 10 patches = 10 measurements. Then press curve for each color 10 patches each = 40 measurements. 50 measurements total)

or

2) Create and press curves for each color.
(Press curve for each color 10 patches each = 40 measurements total.)

Typically though you only use one press curve for all colors per print condition - whether you prelinearize or not.


Make sense?

best, gordon p