software for building dot gain curves?

mihas

Member
The description is incredibly (to me) unclear and includes film in the equation.
Film does not matter. It is important that the difference Print characterisnic curve (reference) minus Print characterisnic curve (sample) is not equal to the difference Plate characteristic curve (reference) minus Plate characteristic curve (sample).
 

gordo

Well-known member
Yes, automatically. Only one iteration! Two iteration maximum in MetaDimension only and the second badly.
While the point administration allows a lot of iterations, mathematically perfect. The difference between the independent calculation of the size of the point, and automatically in the RIP is the number of iterations.
I translated now Compensation Calculator http://rudtp.pp.ru/dgcor/en.php but my English is bad. Colleagues help me edit!
It is registered algorithm of modern rip, Creo/Kodak including. You can calculate the point itself to here. But it is indisputable - you can calculate automatically in the RIP, but only one iteration! Modern RIPs also uses interpolation other than linear, and it is very difficult to replicate in Excel.
For example, offset, the first iteration[/URL]. We want to do better halftone black. Second iteration will help. In TVI or dot gain can not count right second iteration, but surely in the amount of dot.
I'm afraid that language is a real barrier. Try contacting one of these people who might help you with translation:

Grigori Sapunkov
Excourse
http://www.excourse.com/
grigori @ excourse (dot) com

Andrey Demyanenko,
Tex-design Russia
andrewtd @ mail (dot) ru

Alexander Schneider,
PrintWeek Russia
www.alexschneider.ru
info @ alexschneider (dot) ru
 

lazzz

Member
Gordon, i'm not a native speaker too, but let me please check if i got you right:
So we have to print test sheet with fine control strip (like 25-33 fields for every color), then we compare current measured tones with target ones. For example if target curve should have 68% at 50% field, but it has less. Then we can find that let's say 55% field has the requested 68% tone. So we can just replace in RIP 50% dot with 55% and get the right tones this way. Did you mean that?

But what if we can't find exact tone values in our measurements? Maybe we have just 64 and 71. Then i think we need interpolation of data, like Michael said. As i think, the difference between your and Michael approaches is that you use built-in interpolation algorithms of RIP software and Michael uses changeable and finely adjustable ones. I haven't compared yet by myself, but i believe that both methods should have very close results. I'll try to find and upload my curves "before" and "after".

(Off-topic) Alexander has very good english and overall knowledge of printing industry. I've met him, when i was in the very beginning of my career and he had his own printing company called Ares. He tried to promote wide-gamut printing (Hexachrome) and everything in their company looked pretty smart and correct for me. Despite one thing: he constantly repeated that printing is dead and internet rules. So finally he left that business. Now i have printing business myself, but remember what he told and still learn from him. Thank you, i'll know where he is now.
 

gordo

Well-known member
Gordon, i'm not a native speaker too, but let me please check if i got you right:
So we have to print test sheet with fine control strip (like 25-33 fields for every color), then we compare current measured tones with target ones. For example if target curve should have 68% at 50% field, but it has less. Then we can find that let's say 55% field has the requested 68% tone. So we can just replace in RIP 50% dot with 55% and get the right tones this way. Did you mean that?


Correct.


But what if we can't find exact tone values in our measurements? Maybe we have just 64 and 71. Then i think we need interpolation of data, like Michael said. As i think, the difference between your and Michael approaches is that you use built-in interpolation algorithms of RIP software and Michael uses changeable and finely adjustable ones. I haven't compared yet by myself, but i believe that both methods should have very close results. I'll try to find and upload my curves "before" and "after".


There is a tyranny in numbers. They are absolutes. But printing, and often measuring, is not like that. So you do what ISO does. You smooth the curves.

When taking multiple measurements with a densitometer/spectrodensitometer, there is always a slight margin for error. Possible reasons for error include human error or variation in the tint patch. Errors are indicated by slight raises or dips in a curve. Since a curve must be smooth for the output tints to change gradually and evenly, you must smooth the curve by adjusting the curve. To many data points (tone points) and you get a bumpy curve with possible shadestepping/tone bumps in the printing. Not enough data points (tone points) and the curve will not correctly represent what you are trying to match.

So the typical tone values that are measured are:

100
95
90
80
70
60
50
40
30
20
10
5

0



(Off-topic) Alexander has very good english and overall knowledge of printing industry. I've met him, when i was in the very beginning of my career and he had his own printing company called Ares. He tried to promote wide-gamut printing (Hexachrome) and everything in their company looked pretty smart and correct for me. Despite one thing: he constantly repeated that printing is dead and internet rules. So finally he left that business. Now i have printing business myself, but remember what he told and still learn from him. Thank you, i'll know where he is now.
Yes, I thought he would be a good choice to help you because he has the technical background and his English is very good. I hope he will be willing to help you.
 

Erik Nikkanen

Well-known member
It is obvious that people agonize over these dot gain curves. It is understandable but I have come to the view that it is not really so important. It does make sense if the goal of printing is to print dots that have a specific resulting dot value but that is not what the real goal of printing is. The real goal of printing is to print a predictable colour.

So, OK there is an interest to print the right values of the CMYK screens. These curves used are only showing how the individual screens will print. They don't show how those screens will print at the same time. It is hoped and I say hoped because the method is not so defined, that the printing of the corrected curves of CMY will result in a gray when printed at the same time. Something like that. G7 seems to try to at least make the curve compensations hit the gray values, which is in principle a bit better. It is better but it is again not capable of predicting a printed colour.

Another issue is that basically often printing does not print the CMY screens in the combinations that represent a mid to dark grays. Black is used instead to represent those grays. So basically all the effort to obtain these tone curves does not always show up in the print.

As I said, the goal of printing in my view is to print predictable colours. Colours require all kinds of combinations of the CMYK screens. For example, one might have a 20% M screen printed with an 80% C screen. Most likely there will be overprinting of parts of the dots. So what is the dot gain of an overprint dot or a partially overprint dot? Basically you have no idea what it is but it is important because it will affect colour. Does the dot spread differently? Is there wet trapping issues? Can the models that are used to try to predict colour account for all these issues? Probably not.

It is a messy problem and trying to obtain highly accurate dot gain curves will not solve this mess. And this mess does not include the mess provided by the variation one has during the run or the variation of how the plate is inked around the cylinder that affects dot gains. Worrying about highly accurate dot gain curves is a waste of time. These curves are changing quite a bit while the press runs and the amount of change is probably much greater than the small errors one might worry about in the curves themselves. The smoothing of the curves is as good a method as needed considering the amount of variation.

Just my view on the issue.
 

gordo

Well-known member
It is obvious that people agonize over these dot gain curves. It is understandable but I have come to the view that it is not really so important. It does make sense if the goal of printing is to print dots that have a specific resulting dot value but that is not what the real goal of printing is. The real goal of printing is to print a predictable colour.
Dot gain is not a target. Tone reproduction is/should be. Tone reproduction is not color.
The real goal of printing is not to print a predictable colour. There's a great deal of printing that has no color. IMHO the real goal of printing is to reliably reproduce a signed off proof.

So, OK there is an interest to print the right values of the CMYK screens. These curves used are only showing how the individual screens will print.
Correct. In doing so the can be an effective diagnostic tool for the mechanical condition of the press.
They don't show how those screens will print at the same time.
Correct. They are tone curves not color curves.
It is hoped and I say hoped because the method is not so defined, that the printing of the corrected curves of CMY will result in a gray when printed at the same time. Something like that. G7 seems to try to at least make the curve compensations hit the gray values, which is in principle a bit better. It is better but it is again not capable of predicting a printed colour.
G7 bends tone curves to achieve grey balance through the tone scale. It can be argued that this is the inappropriate use of tone curves. Grey balance was an ISO target before G7 - was it not achieved before G7? Is grey balance the right target? what about brown balance i.e. Isometric curves?
Another issue is that basically often printing does not print the CMY screens in the combinations that represent a mid to dark grays. Black is used instead to represent those grays. So basically all the effort to obtain these tone curves does not always show up in the print.
I think you're referring to UCR? Tone curves have nothing to do with UCR or GCR for that matter.
As I said, the goal of printing in my view is to print predictable colours. Colours require all kinds of combinations of the CMYK screens. For example, one might have a 20% M screen printed with an 80% C screen. Most likely there will be overprinting of parts of the dots. So what is the dot gain of an overprint dot or a partially overprint dot? Basically you have no idea what it is but it is important because it will affect colour. Does the dot spread differently? Is there wet trapping issues? Can the models that are used to try to predict colour account for all these issues? Probably not.
Dot gain, tone curves have nothing to do with color, overprints, etc,.
It is a messy problem and trying to obtain highly accurate dot gain curves will not solve this mess.
Yes it's a mess. But dot gain curves are not related to the mess.
And this mess does not include the mess provided by the variation one has during the run or the variation of how the plate is inked around the cylinder that affects dot gains. Worrying about highly accurate dot gain curves is a waste of time. These curves are changing quite a bit while the press runs and the amount of change is probably much greater than the small errors one might worry about in the curves themselves.
I agree that worrying about highly accurate dot gain curves is a waste of time which is why I think the OP has made a technical mountain out of a molehill.
The smoothing of the curves is as good a method as needed considering the amount of variation.
Yup - that's why it's done.
The big issue IMHO is that there appears to be a huge disconnect between the folks creating the standards and specifications (ISO, GRACoL et al) and the folks that are challenged to implement them. The result is confusion or disassociation.
 

Erik Nikkanen

Well-known member
Dot gain is not a target. Tone reproduction is/should be. Tone reproduction is not color.
The real goal of printing is not to print a predictable colour. There's a great deal of printing that has no color. IMHO the real goal of printing is to reliably reproduce a signed off proof.
Thanks for the correction. I again mixed up dot gain and tone concepts. Related but not the same.

I would think that it would be easier to reproduce a signed off proof if there was the capability to reproduce colour but in this industry I might be wrong. :) You have lots of experience in that area so no comment from me.

Correct. In doing so the can be an effective diagnostic tool for the mechanical condition of the press.
Yes, tone values are needed for process control but IMO not for colour management in the future.


I think you're referring to UCR? Tone curves have nothing to do with UCR or GCR for that matter.
Yes I agree. I only brought up the idea because it appears that there is a great interest to pin down screen values that one does not actually use much in the process. Especially with the G7 gray line, which is basically the L line in the colour gamut plot, much effort is made to find this gray line but it is not used in the process because of UCR or GCR. So it is a target that is not used much.

Outside that target line is a vast volume of colour. Even Don Hutcheson has stated in the early days, that obtaining the gray line in G7 does not guarantee that the colours will be right in the rest of the colour space. He was quite honest about that then. I suspect he still holds this view.

So one has a method where one does lots of work to find information that is not used and that information does not guarantee predictability of colour in the rest of the colour space. I am sure it helps but it is not a capable system.


Yes it's a mess. But dot gain curves are not related to the mess.
True they are not related to the mess, but they are also not needed to obtain predictable colour. I guess you can argue that they are needed to obtain the signed proof by satisfying some standard but that is another issue. :)

I agree that worrying about highly accurate dot gain curves is a waste of time which is why I think the OP has made a technical mountain out of a molehill.

Yup - that's why it's done.
The big issue IMHO is that there appears to be a huge disconnect between the folks creating the standards and specifications (ISO, GRACoL et al) and the folks that are challenged to implement them. The result is confusion or disassociation.
The confusion starts from the top. It seems impossible for the technical leaders in the industry to think in different patterns. It is "group think" on a grand scale. These problems are not that difficult but the technical community continues to make it difficult because they don't have the ability to look at the problem and see what is actually needed and what might result in a more direct, easy and reliable method. But consultants love this mess. It is good for business and the more mess the better. :)
 
Last edited:

mihas

Member
Yes, but they are in this business so they know the correct terms. Alexander in particular is very good at English.
Grigori and Andrey responded and helped with translation, Alexander not responded. So except me helped three Russian speakers. All are encouraged to refer to native English speakers for help. All errors that they found I corrected. I do not cease to hope for help in this forum. English version of the print and colorimetric tools here:
http://rudtp.pp.ru/spectralcalc_en.php
http://rudtp.pp.ru/spectral_translator.js
http://rudtp.pp.ru/dgcor/en.php
 

gordo

Well-known member
The confusion starts from the top. It seems impossible for the technical leaders in the industry to think in different patterns. It is "group think" on a grand scale. These problems are not that difficult but the technical community continues to make it difficult because they don't have the ability to look at the problem and see what is actually needed and what might result in a more direct, easy and reliable method. But consultants love this mess. It is good for business and the more mess the better. :)
The standards and specification organizations are populated by vendors and consultants. What the industry receives as a result is quite predictable.
 

beanz

Member
so, i tried color port and exporting the lab datas to excel, then used an lab dot gain calculator, however the values i obtained were off by around 3% when compared to the measurements i got using xrite exact. so i ended up just making the measurements using my exact.

remember guys, what i want is a solution to make step dot gain readings in a fast manner (maybe through scanning a patch similar to a media wedge). just to remind everyone of my topic in particular :)
 

lazzz

Member
Beanz, then look into eci grayconL color bar. It provides good averall process control: solids, trappings, gray balance and 10-step TVI control. It can be measured at one pass. You may use color toolbox for fast measurement and assesment.
 

jpfulton248

Well-known member
I calibrated our platemaker to counteract dot gain using a super old X-Rite Densitometer from ebay, some formulas I found online and an excel spreadsheet. After making the adjustments according to those calculations we were amazed at the improvement. We did some final "eyeball tweaks" for certain values and that was it.
 

Automatically Autonomous Automation

Automatically Autonomous Automation
Although the autonomous car is not quite ready, a lights out print operation is something you can do right now if you have a comprehensive Print MIS (Management Information System). The advantages can put money on your bottom line. So what’s your next step? Link to Article

   
Top