Calibration and profiling of wide format solvent printer

[m***@public.gmane.org]

Hi! I'm going to calibrate and profile wide format solvent printer: Roland XJ-740.RIP software i'm using (VersaWorks) has ability to calibrate printer... but i want to disable it (by setting linear curves) and use Argyll all the way through (i want profile with calibration) This is my calibration & profiling procedure:=== Calibration ===1. targen -v -d4 -s50 -e4 -f0 Roland_XJ-740_calibration2. printtarg -v -ii1 -R13 -pA4 Roland_XJ-740_calibration3. printing "calibration test chart"4. chartread -H Roland_XJ-740_calibration5. printcal -v -p -D"Roland_XJ-740 calibration" -i Roland_XJ-740_calibration === Profiling ===6. targen -v -d4 -G -g11 -s11 -f1323 Roland_XJ-7407. printtarg -v -ii1 -R13 -pA4 -K Roland_XJ-740_calibration.cal Roland_XJ-7408. printing "profiling test chart"9. chartread -H Roland_XJ-74010. colprof -v -D"Roland_XJ-740" -qm -S sRGB.icm -cmt -dpp Roland_XJ-74011. applycal -v Roland_XJ-740_calibration.cal Roland_XJ-740.icm Roland_XJ-740_calibrated.icm Hence i have a couple of questions:A) Are above steps right? Maybe You have some comments or advices...:)B) Do I have to additionally apply calibration file to the created profile (step 11) or     "-K Roland_XJ-740_calibration.cal" in printtarg (step 7) sufficient?    In other words... by using -K option is calibration file passed to colprof     or it is used only while printing "profiling test chart" (step 8)?C) Is colprof smart enough to use ink limits from the calibration file or I have to give them to colprof explicitly?D) Is setting up ink limits at RIP required (for example total ink limit: 250%)?     or limits contained in icc profile are sufficient (while RIP individual limits are set to 100%, total ink limit: 400%)? Best regardsMichał

[m***@public.gmane.org]

Hi!
 
I'm going to calibrate and profile wide format solvent printer: Roland XJ-740.
RIP software i'm using (VersaWorks) has ability to calibrate printer... but 
i want to disable it (by setting linear curves) and use Argyll all the way through (i want profile with calibration)
 
This is my calibration & profiling procedure:
=== Calibration ===
1. targen -v -d4 -s50 -e4 -f0 Roland_XJ-740_calibration
2. printtarg -v -ii1 -R13 -pA4 Roland_XJ-740_calibration
3. printing "calibration test chart"
4. chartread -H Roland_XJ-740_calibration
5. printcal -v -p -D"Roland_XJ-740 calibration" -i Roland_XJ-740_calibration
 
=== Profiling ===
6. targen -v -d4 -G -g11 -s11 -f1323 Roland_XJ-740
7. printtarg -v -ii1 -R13 -pA4 -K Roland_XJ-740_calibration.cal Roland_XJ-740
8. printing "profiling test chart"
9. chartread -H Roland_XJ-740
10. colprof -v -D"Roland_XJ-740" -qm -S sRGB.icm -cmt -dpp Roland_XJ-740
11. applycal -v Roland_XJ-740_calibration.cal Roland_XJ-740.icm Roland_XJ-740_calibrated.icm
 
Hence i have a couple of questions:
A) Are above steps right? Maybe You have some comments or advices...
B) Do I have to additionaly apply calibration file to the created profile (step 11) or 
   "-K Roland_XJ-740_calibration.cal" in printtarg (step 7) sufficient?
   In other words... by using -K option is calibration file passed to colprof 
   or it is used only while printing "profiling test chart" (step 8)?
C) Is colprof smart enough to use ink limits from the calibration file or I have to give them to colprof explicitly?
D) Is setting up ink limits at RIP required (for example total ink limit: 250%)? 
   or limits contained in icc profile are sufficient (while RIP individual limits are set to 100%, total ink limit: 400%)?
 
Best regards
Michał

[m***@public.gmane.org]

Hi!
 
I'm going to calibrate and profile wide format solvent printer: Roland XJ-740.
RIP software i'm using (VersaWorks) has ability to calibrate printer... but 
i want to disable it (by setting linear curves) and use Argyll all the way through (i want profile with calibration)
 
This is my calibration & profiling procedure:
=== Calibration ===
1. targen -v -d4 -s50 -e4 -f0 Roland_XJ-740_calibration
2. printtarg -v -ii1 -R13 -pA4 Roland_XJ-740_calibration
3. printing "calibration test chart"
4. chartread -H Roland_XJ-740_calibration
5. printcal -v -p -D"Roland_XJ-740 calibration" -i Roland_XJ-740_calibration
 
=== Profiling ===
6. targen -v -d4 -G -g11 -s11 -f1323 Roland_XJ-740
7. printtarg -v -ii1 -R13 -pA4 -K Roland_XJ-740_calibration.cal Roland_XJ-740
8. printing "profiling test chart"
9. chartread -H Roland_XJ-740
10. colprof -v -D"Roland_XJ-740" -qm -S sRGB.icm -cmt -dpp Roland_XJ-740
11. applycal -v Roland_XJ-740_calibration.cal Roland_XJ-740.icm Roland_XJ-740_calibrated.icm
 
Hence i have a couple of questions:
A) Are above steps right? Maybe You have some comments or advices...
B) Do I have to additionaly apply calibration file to the created profile (step 11) or 
   "-K Roland_XJ-740_calibration.cal" in printtarg (step 7) sufficient?
   In other words... by using -K option is calibration file passed to colprof 
   or it is used only while printing "profiling test chart" (step 8)?
C) Is colprof smart enough to use ink limits from the calibration file or I have to give them to colprof explicitly?
D) Is setting up ink limits at RIP required (for example total ink limit: 250%)? 
   or limits contained in icc profile are sufficient (while RIP individual limits are set to 100%, total ink limit: 400%)?
 
Best regards
Michał

[Nikolay Pokhilchenko]

Post by m***@public.gmane.org
Hi!
I'm going to calibrate and profile wide format solvent printer: Roland XJ-740.
RIP software i'm using (VersaWorks) has ability to calibrate printer... but 
i want to disable it (by setting linear curves) and use Argyll all the way through (i want profile with calibration) I'd recommend You to find the way using internal RIP calibration curves because of quantization errors while applying the calibration externally.

It will be better if the image at the RIP input will be "normal" and the calibration curves will apply inside the RIP

Post by m***@public.gmane.org
=== Calibration ===
1. targen -v -d4 -s50 -e4 -f0 Roland_XJ-740_calibration
2. printtarg -v -ii1 -R13 -pA4 Roland_XJ-740_calibration
3. printing "calibration test chart"
4. chartread -H Roland_XJ-740_calibration
5. printcal -v -p -D"Roland_XJ-740 calibration" -i Roland_XJ-740_calibration This seems OK but I prefer 128 patches per channel. This is not a problem for a wide format media :). More patches will improve immunity to printing and measuring disturbances.
=== Profiling ===
6. targen -v -d4 -G -g11 -s11 -f1323 Roland_XJ-740 I don't recommend using -s and -g parameters.

For better ink limit accounting in targen (if it works at all) I'd recommend use some preliminary profile with calibration applied (embedded in the profile by applycal utility). If preliminary profile isn't correct for certain calibration curves disable optimization by -A0 parameter addition.

Post by m***@public.gmane.org
7. printtarg -v -ii1 -R13 -pA4 -K Roland_XJ-740_calibration.cal Roland_XJ-740 You need -K with cal. file if You do not apply the calibration in RIP. But You need to use the -I parameter (capital "i") with calibration in case of applying the calibration in the RIP. The -I helps to limit the inks in colprof correctly.
8. printing "profiling test chart"
9. chartread -H Roland_XJ-740
10. colprof -v -D"Roland_XJ-740" -qm -S sRGB.icm -cmt -dpp Roland_XJ-740
11. applycal -v Roland_XJ-740_calibration.cal Roland_XJ-740.icm Roland_XJ-740_calibrated.icm If You apply the calibration in RIP, You needn't step No.11. Your production profile will be without the calibration embedded. But You need embed the calibration into profile for using it with targen for next re-profiling for better ink limiting at the target generation stage.
C) Is colprof smart enough to use ink limits from the calibration file or I have to give them to colprof explicitly? If You supply the calibration at No.7 stage the colprof should use ink limiting with calibration accounting.
D) Is setting up ink limits at RIP required (for example total ink limit: 250%)? 
   or limits contained in icc profile are sufficient (while RIP individual limits are set to 100%, total ink limit: 400%)? It depends on actual channel limits needed. If You didn't use internal RIP calibration capability You needn't limit the inks in the RIP. But if below 100% channel limits are desired it may be better to set them into the RIP (the RIP may scale 100% input to the limit entered). But in case of applying of channel limits in the RIP the scaling of the calibration curves in *.cal files needed for correct TIL accounting IL in ArgyllCMS. I.e. when 70% channel limit in RIP is entered, the colprof should "know" that 100% on the input of calibration curve means 70% actual ink quantity.

Best regards,
НОкПлай ППхОльчеМкП

[m***@public.gmane.org]

Nikolay, thank you very much for the tips.

Post by m***@public.gmane.org

Post by m***@public.gmane.org
7. printtarg -v -ii1 -R13 -pA4 -K Roland_XJ-740_calibration.cal Roland_XJ-740

You need -K with cal. file if You do not apply the calibration in RIP.
But You need to use the -I parameter (capital "i") with > calibration in case
of applying the calibration in the RIP. The -I helps to limit the inks in colprof correctly.

A) Let's assume that i've done calibration in RIP and it is now in calibrated state,
do i still need "-I calibration_file" parameter to help Argyll to estimate ink limits?
B) Is this method (-I option) equal to providing TIL an IL as parameters to colprof?
C) What Agryll's tool will print ink limits values if i wanted to use them in RIP? colprof?

Post by m***@public.gmane.org
It depends on actual channel
limits needed. If You didn't use internal RIP calibration capability You
needn't limit the inks in the RIP. But if below 100% channel limits are
desired it may be better to set them into the RIP (the RIP may scale 100%
input to the limit entered). But in case of applying of channel limits in the
RIP the scaling of the calibration curves in *.cal files needed for correct
TIL accounting IL in ArgyllCMS. I.e. when 70% channel limit in RIP is entered,
the colprof should "know" that 100% on the input of calibration curve means
70% actual ink quantity.

D) If i would skip Argyll's calibration altogether and do calibration in RIP,
what tool do i have to provide ink limits to? targen or colprof?

E) What i've read, TIL provided to targen (-l option) should be 10% more than TIL provided to colprof? Am i right?

F) Does colprof manage to make good profile without need to provide black generation parameters (-k option) to it?

I have some idea, correct me if i'm wrong.
What i want to do is to use Agryll's calibration capabilities and obtain TIL and maybe IL values (without inks curves, only limits)
then enter them to RIP and make calibration there.
If TIL turns out to be 250%, where do i have to use this value while profiling? targen? colprof?

Best Regards
Michał

[Nikolay Pokhilchenko]

Post by m***@public.gmane.org
Nikolay, thank you very much for the tips.

Post by m***@public.gmane.org

Post by m***@public.gmane.org
7. printtarg -v -ii1 -R13 -pA4 -K Roland_XJ-740_calibration.cal Roland_XJ-740

You need -K with cal. file if You do not apply the calibration in RIP.
But You need to use the -I parameter (capital "i") with > calibration in case
of applying the calibration in the RIP. The -I helps to limit the inks in colprof correctly.

A) Let's assume that i've done calibration in RIP and it is now in calibrated state,
do i still need "-I calibration_file" parameter to help Argyll to estimate ink limits? Yes. In case the calibration is applied in the RIP, the curves and channel limits (if any) should be passed to the colprof for accurate total ink estimation during profile computation. The -I parameter includes the calibration curves with theirs channel limits into *.ti2 file at the printtarg stage. But the patch values of the target remains "uncalibrated". Next You print the target through the RIP with it's internal calibrations and channel limits applied (I suggest to not apply the total ink limit in the RIP, apply only channel limits).

After reading the targets You'll got *.ti3 file with uncalibrated patch CMYK values and the calibration curves. The profiling will be performed for calibrated device. The calibration data helps the profiler "sense" actual inks amount.
You should pay an attention to the RIP behavior in case of channel ink limits definition. Will the curve loaded into the RIP be scaled to the defined ink limit? Or it will be clipped at ink limit value? In any case the RIP calibration plus limit behavior should be represented in the curves which are supplied to the printtarg (the colprof finally).

Post by m***@public.gmane.org
B) Is this method (-I option) equal to providing TIL an IL as parameters to colprof? They are not equivalent. It's just a part of the process with calibration envolved. There is no explicit channel limiting in the ArgyllCMS except the limiting by calibration curves. So if You need per-channel limiting You need to include (and apply in RIP or CMS) some kind of calibration file with maximum channel value(s) below 100%.

As for TIL: the TIL is just the number which instruct the targen how to generate the target and instruct the colprof which limit it have to achieve deepest dark colors. The TIL can be overrided at the profiling stage; But for the best results TIL of targets should be some higher than TIL of profiling. Both of the TILs will be more relevant to the print if the targen and colprof will be supplied with the calibration curves. As soon as You have channel limits applied in the calibration there is no practical need to limit the channels at the profiling stage. IMO that is why the colprof have no individual channel limits.
From my experience the targen and colprof don't hold the limits very tight. The last is better, but not very sharp. You may get up to 20% (if I've remembered correctly) overflow with targen and up to 5..12% with colprof. For first tries I'd recommend to check the targets for TIL. Generate the target with -K prittarg parameter to check the TIL in the spreadsheet.

Post by m***@public.gmane.org
C) What Agryll's tool will print ink limits values if i wanted to use them in RIP? colprof? What have You meant "print"? IMO You should define channel limits yourself by means of printcal and TIL by some TIL test images.

Not every time whole TIL which media can accept is needed. For some media the black/dark color saturation starts before the inks overfill the media. You can check this by inverse lookup large target through current media profile. You'll see many of CMYK values and will be able to compute practical TIL by maximums on a inverse lookup spreadsheet, Use xicclu for that.
At the beginning I recommend You apply the maximum TIL the media can carry without inks run out.

Post by m***@public.gmane.org

Post by m***@public.gmane.org
It depends on actual channel
limits needed. If You didn't use internal RIP calibration capability You
needn't limit the inks in the RIP. But if below 100% channel limits are
desired it may be better to set them into the RIP (the RIP may scale 100%
input to the limit entered). But in case of applying of channel limits in the
RIP the scaling of the calibration curves in *.cal files needed for correct
TIL accounting IL in ArgyllCMS. I.e. when 70% channel limit in RIP is entered,
the colprof should "know" that 100% on the input of calibration curve means
70% actual ink quantity.

D) If i would skip Argyll's calibration altogether and do calibration in RIP,
   what tool do i have to provide ink limits to? targen or colprof? If You apply any type of calibration external to the ArgyllCMS You'll be not able to provide the CMS with actual ink quantity other way than supplying the calibration curves in *.cal files. When the curves are applied in the RIP, actual volume on the media will differ from percentage in the target image. So the targen and colprof will have no "idea" what actual ink quantity is without of relevant calibration file.

If You know the curves and channel limits that the RIP applies to the image, You are able to make the *.cal file by spreadcheet editor.

Post by m***@public.gmane.org
E) What i've read, TIL provided to targen (-l option) should be 10% more than TIL provided to colprof? Am i right? See upper. This is desirable but in some cases not mandatory. Depends on the inks behavior and device response at the extreme ink densities.
F) Does colprof manage to make good profile without need to provide black generation parameters (-k option) to it? In generally no. You can play with the parameters yourself when have got real device data. The choice of -k (or even -K) parameters is may be most laborious part of the process in some (rater often) cases. Depends on process.
What i want to do is to use Agryll's calibration capabilities and obtain TIL and maybe IL values (without inks curves, only limits)
then enter them to RIP and make calibration there. You may do that. But in case of linear limiting in the RIP you need provide the CMS by synthetic linear calibration data with channel maximums at the RIP limits. This is because ArgyllCMS have no tool to individual channel limiting other than printcal with the calibration curves file. I've wrote about upper.
If TIL turns out to be 250%, where do i have to use this value while profiling? targen? colprof? If TIL=250% was chosen with some practical margin I'd recommend to print the target with 260% actual limit. This way the colprof will "know" how the device behaves at the 250% TIL boundary. You may even decrease the TIL while see on the black generation curve and checking the total profile gamut with the iccgamut.

That's all I can say to You about wide format, RIPs, calibration and TIL. You should labor the details yourself.
I'm not the "last instance". I can mistaken. Hope the others and Graeme will correct me in case.

[m***@public.gmane.org]

Nikolay, thank you for your time and comprehensive answers.
Because of my low color management skills, there are few things that are unclear to me:
A) should I do RIP software calibration first, then do Argyll calibration & profiling procedure on top of "RIP calibrated/linearised
printer" or in reverse order - do preliminary calibration & profiling procedure with Argyll on uncalibrated printer (with
calibration on RIP disabled)?

B) If i will go the "preliminary profile" route you've recommended, do i have to do Argyll calibration with preliminary profile and
again with target/big profile? or may calibration stage be done once (with 128 patches per colorant)?
It seems that calibration & profiling is a feedback loop process...

C) From what you've written, i understand that Argyll does't provide the way to calculate/estimate individual ink limits,
the only way to do it is in RIP software using special test chart?

This is RIP calibration section in software i'm using:
Loading Image...

So general process should look like this?:
1. Calibration using RIP software - estimating ILs and curves (You've recommended to not setting TIL)
2. Creating preliminary calibration & profiling with Argyll (with overestimated/rough ink limits)
3. Estimating TIL and ILs with more accuracy from preliminary profile
5. Creating main calibration & profiling with Argyll (using preliminary profile and estimated TIL, calibration curves)

Greetings
Michał

[Nikolay Pokhilchenko]

Post by m***@public.gmane.org
A) should I do RIP software calibration first, then do Argyll calibration & profiling procedure on top of "RIP calibrated/linearised
   printer" or in reverse order - do preliminary calibration & profiling procedure with Argyll on uncalibrated printer (with
   calibration on RIP disabled)? You may start right from ArgyllCMS printcal. If You're calibrate the process in the RIP I'd not recommend You to envolve ArgyllCMS over existing RIP calibration. It's hard to compute total calibration behaviour while calibrating already calibrated device, when two calibrations will persist in series with different nubmers of points.

Just perform ArgyllCMS calibration, convert it into the RIP internal calibration data. After that You may convert actual RIP calibration curve (lover number of points) back into Argyll *.cal. But if the curve entered in the RIP from ArgyllCMS *.cal data is very similar to the original printcal curve You may leave printcal curve as is for using with targen/colprof. Do not forget to check ink limit behavior (I see an error in Your curve example from the link with PNG).

Post by m***@public.gmane.org
B) If i will go the "preliminary profile" route you've recommended, do i have to do Argyll calibration with preliminary profile and
   again with target/big profile? or may calibration stage be done once (with 128 patches per colorant)?
   It seems that calibration & profiling is a feedback loop process... Calibration may be performed once based only on single channel ramps. You needn't use profiling chart for that. The aim of the calibration to provide most linear and repeatable device response. You may recalibrate the device (using original reference calibration data from first pass) with just CMYK ramps and preserve profile color tolerance without reprofiling. Someone tried to linearize the device by several passes of printcal. There is a difference between only one pass and several pass calibration curves. But the difference isn't significant. So You can safely stay with one linearization pass. I recommend preserve several percent of ink level below 100% for future limit increasing if increase of saturation will needed on later recalibration. See the printcal manual.
C) From what you've written, i understand that Argyll does't provide the way to calculate/estimate individual ink limits,
   the only way to do it is in RIP software using special test chart? Yes. But You can use printcal plots for estimation of individual channel limits. And may use reulting profile for checking the highest TIL usable in certain profile.
http://s28.postimg.org/g9eqag6bh/RIP_calibration.png I think there is an error (or not optimal use of the RIP features). IMO if You've selected 75% channel limit there is no another limit needed in the curve itself. I see on the picture 50.1%*75%=38.25% ink limit for Cyan. Is it correct? I think it's better to setup 38% channel ink limit and draw the curve till 100% at 100% input. But for correct ink limiting You need to provide the ArgyllCMS with the curve 0..38% for Cyan because Argyll have no another instances of individual channel limiting.
1. Calibration using RIP software - estimating ILs and curves (You've recommended to not setting TIL) Not using TIL but estimate TIL basing on RIP tests for future using and resonable target limiting. If You're using calibration by RIP software do not use ArfyllCMS printcal.

Alternatively do not use RIP calibration, only use individual channel ink limiting and use printcal. Then move calibration data from printcal into the RIP internal curves. Print only single channel ramps for calibration.

Post by m***@public.gmane.org
2. Creating preliminary calibration & profiling with Argyll (with overestimated/rough ink limits) At this stage just perform good calbraton. It will may become first and final. After loading the calibration in the RIP provide appropriate *.cal file to the targen. Generate the target with high but reasonable TIL. Print the target with calibration in the RIP. Check for TIL. If ink leaks prevent the target from correct reading by instrument lower TIL in the targen or alternatively remove the patches with excessive TIL from *.ti1 target file.
3. Estimating TIL and ILs with more accuracy from preliminary profile Yes. But anyway use some higher TIL than will be in the final profile.
5. Creating main calibration & profiling with Argyll (using preliminary profile and estimated TIL, calibration curves) Don't recalibrate. Leave the calibration from step 2 as the reference and change it only if have badly missed the individual channel limits.

Hope it helps.

[m***@public.gmane.org]

These are my current RIP calibration settings:
Loading Image...
When i change Single Color Limit field to from 75% to 100% curve scales but values in table don't change:
Loading Image...
Should I change Single Color Limit field for all channels to 100%?
I've also done Argyll calibration on raw/uncalibrated printer (RIP calibration not activated) with 50 (i'll do more later) patches per channel and -p1.37 (because of dot gain):
1. targen -v -d4 -p1.37 -s50 -e4 -f0 Roland_XJ-740_calibration
2. printtarg -v -ii1 -R13 -pA4 Roland_XJ-740_calibration
3. chartread -H Roland_XJ-740_calibration
4. printcal -v -p -D"Roland_XJ-740_calibration" -i Roland_XJ-740_calibration
Roland_XJ-740_calibration.zip:
http://www.datafilehost.com/d/53428758
greetings!
Michał

[Nikolay Pokhilchenko]

This post might be inappropriate. Click to display it.

[Graeme Gill]

Post by Nikolay Pokhilchenko
From my experience the targen and colprof don't hold the
limits very tight. The last is better, but not very sharp. You may get up to 20% (if
I've remembered correctly) overflow with targen and up to 5..12% with colprof.

Hi,
I'd be interested in any such examples, as I'm not aware of either tool
overflowing the set ink limits by much. Colprof does limit B2A only node values -
it doesn't attempt to evaluate or limit the total of interpolated combinations of
device ink values. This should only become a problem if the separation is very
non-smooth, in which case you have more severe problems than ink limits.

Graeme Gill.

[m***@public.gmane.org]

So.... now i'm trying to get better understanding of whole process and sort it out in my head:)
While browsing the list i've found some tips about estimating TIL using gamut volume change
(by inceasing ink limit and observing gamut volume change), is it good method or there is a new/better one?

Besides I've recently noticed that my RIP uses some unknown "default CMYK profile" when it is in uncalibrated state.
I don't know what exactly this profile is representing and i'm not sure if it won't somehow distort values sent to printer,
so i've got the idea of replacing it (for the purpose of calibration)
with some "simple linear profile" (if something like this exists) which won't distort values sent to printer.
Is Agryll capable of creating this kind of "linear profile"?

Michał

[m***@public.gmane.org]

So.... now i'm trying to get better understanding of whole process and sort
it out in my head:)
While browsing the list i've found some tips about estimating TIL using
gamut volume change
(by inceasing ink limit and observing gamut volume change), is it good
method or there is a new/better one?

Besides I've recently noticed that my RIP uses some unknown "default CMYK
profile" when it is in uncalibrated state.
I don't know what exactly this profile is representing and i'm not sure if
it won't somehow distort values sent to printer,
so i've got the idea of replacing it (for the purpose of calibration)
with some "simple linear profile" (if something like this exists) which
won't distort values sent to printer.
Is Agryll capable of creating this kind of "linear profile"?

Michał

[Graeme Gill]

Post by m***@public.gmane.org
While browsing the list i've found some tips about estimating TIL using
gamut volume change
(by inceasing ink limit and observing gamut volume change), is it good
method or there is a new/better one?

Hi,

That's one way, but there is more than one criteria. Argyll's printcal
judges the per channel limits on diminishing returns, but it's up
to you to judge if the page is getting too wet, affecting resolution,
uniformity, or drying time. Some suitable test charts would assist
in judging this - fine white lines on 100% ink backgrounds is a critical
test for resolution. Similar considerations apply for the full CMYK
combination, and looking at the 3D gamut boundary can help to judge
diminishing returns there as well.

Post by m***@public.gmane.org
Besides I've recently noticed that my RIP uses some unknown "default CMYK
profile" when it is in uncalibrated state.
I don't know what exactly this profile is representing and i'm not sure if
it won't somehow distort values sent to printer,
so i've got the idea of replacing it (for the purpose of calibration)
with some "simple linear profile" (if something like this exists) which
won't distort values sent to printer.

If you're doing true profiling, there is no profile to replace - the CMYK x per
channel calibration curves in, is what you get on the page.

Post by m***@public.gmane.org
Is Argyll capable of creating this kind of "linear profile"?

There is no such animal. A device profile converts PCS (CIE) to/from
device values. They are different colorspaces (that's the point of it),
so there is no such thing as "straight through" or "linear" device profile.

Graeme Gill.

[m***@public.gmane.org]

Sorry about the mess:/

[Michał]

Graeme,
i think i've stated it unclear (or You've already got it right and i didn't understand).

In my RIP each media type has a field where i have to choose an icc profile (it is mandatory), even when i don't have one yet (it's just going to be created in profiling process).
So, when i create new, fresh media type i have to choose some profile (i choose Default CMYK), then i do calibration & profiling and at the end i select newly created profile as media's default (instead of "default CMYK" profile).
The question are:

1. is "default CMYK" necessary?

2. if yes: is it generic/vast enough not to limit/spoil calibration & profiling... it is additional layer of transformation... can i check somehow if its gamut contains medias' gamuts?

3. is there any good generic/synthetic profile that RIP's "default CMYK" can be replaced to?

4. or i

[Graeme Gill]

Post by Michał
In my RIP each media type has a field where i have to choose an icc profile (it is
mandatory), even when i don't have one yet (it's just going to be created in profiling
process). So, when i create new, fresh media type i have to choose some profile (i
choose Default CMYK), then i do calibration & profiling and at the end i select newly
created profile as media's default (instead of "default CMYK" profile).

Hi,
my point still stands. If the selected profile changes the test chart,
then you aren't able to profile the device in that manner. To profile, you
must bypass the color management machinery you are trying to setup. The manufacturer
has to have a method for doing this, or they wouldn't be able to profile it themselves.

Graeme Gill.

[Graeme Gill]

This post might be inappropriate. Click to display it.

[m***@public.gmane.org]

Hi,

Let there be a workflow (without native calibration capabilities):
1. Creating PrinterA calibration file (with printcal)
2. Creating PrinterA preliminary profile with calibration (from step 1) applied to it.
3. Using PrinterA_preliminary.icm as targen -c parameter when creating main profile.
4. Next step is printtarg.... do I have to specify -K or -I parameter with calibration file from step 1?

Greetings
Michał

[Nikolay Pokhilchenko]

Let there be a workflow (without native calibration capabilities): Let it be!
1. Creating PrinterA calibration file (with printcal)
2. Creating PrinterA preliminary profile with calibration (from step 1) applied to it. If you're not involving RIP calibration capabilities you should apply the calibration curves onto the profiling target image. So use printarg parameter which include the calibration AND apply it on target image.
3. Using PrinterA_preliminary.icm as targen -c parameter when creating main profile. Yes. Just before that include the calbration into the profile by applycal utility. Or check that the calibration was included in the profile from *.ti3 file. I didn't remember how ArgyllCMS will do exactly, but the calibration inside the preliminary profile is needed for targen to correct ink limiting.
4. Next step is printtarg.... do I have to specify -K or -I parameter with calibration file from step 1? The same as I've wrote for step two. Include and apply calibration with printtarg.

[Nikolay Pokhilchenko]

An advice: do not use 8 bit per channel bitmaps nor for profiling nor for production printing if the calibration curves are steep because of quantizing. 8 bit is OK for calibrated device and for traditional RGB abstract color spaces. But as soon as you apply the calibration on image the last become "uncalibrated" and in case of high degree of curvature you may got banding on gradients and loss of details.
It's ok using PostScript at profiling stage because it's precise (the channel values passed from argyllcms in floating point format). But for real image printing you may need convert the sources into 16 bit before converting by cctiff. If your RIP don't supporting 16 bit or have a speed constraints with 16 bit per channel CMYK images you may convert 16-bit "uncalibrated" image into 8 bit with apllying some kind of dither.
For example an ImageMagic command line:

convert -verbose -ordered-dither o4x4,256 -depth 8 -compress LZW Input_from_Argyll.tif Output_to_RIP.tif

[Nikolay Pokhilchenko]

Post by Nikolay Pokhilchenko
An advice: do not use 8 bit per channel bitmaps nor for profiling nor for production printing if the calibration curves are steep because of quantizing. 8 bit is OK for calibrated device and for traditional RGB abstract color spaces. But as soon as you apply the calibration on image the last become "uncalibrated" and in case of high degree of curvature you may got banding on gradients and loss of details.
It's ok using PostScript at profiling stage because it's precise (the channel values passed from argyllcms in floating point format). But for real image printing you may need convert the sources into 16 bit before converting by cctiff. If your RIP doesn't support 16 bit or have a speed constraints with 16 bit per channel CMYK images you may convert 16-bit "uncalibrated" image into 8 bit with apllying some kind of dither.
convert -verbose -ordered-dither o4x4,256 -depth 8 -compress LZW Input_from_Argyll.tif Output_to_RIP.tif

Hello Graeme!
As written abowe I ask a dither feature(s) for cctiff:
* dither by default when source TIFF image have 16 bit depth and destination is JPEG;
* an option to force 8 bit output with 16 bit input. Default dithering for that;
* an option to disable dither (for two cases abowe);
* an option to enable dithering with 16 bit source and destination both (may be in precision mode only if useless in standard).

Nikolay Pokhilchenko.

[Graeme Gill]

Hi,
I'll make a note of it - I'm not sure when I will have time
to implement it though.

Graeme.

[m***@public.gmane.org]

Hi Nikolay I roughly understand the difference between -I and -K options and I realize that RIP calibration is better than applying it to profile....but one thing is bothering me. Isn't it redundant/harmful to supply the same calibration data twice into the main profile workflow: - first time: indirectly via preliminary profile as targen -c parameter   (If preliminary profile has calibration applied then targen -c with -A > 0 generates color domain    already corrected by calibration in some degree)  - second time: directly via printtarg -K/-I parameter Michał

[m***@public.gmane.org]

Hi Nikolay
 
I roughly understand the difference between -I and -K options 
and I realize that RIP calibration is better than applying it to profile....
but one thing is bothering me. 
Isn't it redundant/harmful to supply the same calibration data twice into the main profile workflow:
 - first time: indirectly via preliminary profile as targen -c parameter
   (If preliminary profile has calibration applied then targen -c with -A > 0
generates color domain already corrected by calibration in some degree) 
 - second time: directly via printtarg -K/-I parameter
 
Michał

[m***@public.gmane.org]

Hi all,

I've finally found some time to setup printer into uncalibrated state and managed to disable all color management options.
Printed charts look very different now compared to those attached to my previous posts...

My workflow was:
1. targen -v -d4 -p2.73 -s200 -f0 Roland_XJ-740_calibration
2. printtarg -v -ii1 -R13 -pA4 Roland_XJ-740_calibration
3. chartread -H Roland_XJ-740_calibration
4. printcal -v -p -D"Roland_XJ-740 calibration" -i Roland_XJ-740_calibration
5. targen -v -d4 -G -f1323 Roland_XJ-740
6. printtarg -v -ii1 -R13 -pA4 -K Roland_XJ-740_calibration.cal Roland_XJ-740
7. chartread -H Roland_XJ-740
8. colprof -v -D"Roland_XJ-740" -qm -S sRGB.icm -P -cmt -dpp Roland_XJ-740
9. applycal -v Roland_XJ-740_calibration.cal Roland_XJ-740.icm Roland_XJ-740_calibrated.icm

Roland XJ-740.zip (Please use Slow Download button):
http://www.fileswap.com/dl/PtJ68qWUD0/

1. I want to estimate TIL and individual ink limits from calibration and profile,
so which profile should I use with iccgamut/xicclu: with calibration applied or without it?

3. When I view printcal's charts, I'm somehow worried about black calibration curve,
black response isn't straight line after calibration correction applied, is it ok?.

4. Nikolay, You suggested me to put calibration curves into a spreadsheet and convert them
into RIP usable form, enter them into RIP, enter limits and then do profiling.
I've took data from .cal file and reflected individual curves by y=x line.
May You take a look if something like following spreadsheet will do?
Gnueric spreadsheet (You need to enable "Time Series Analysis Functions" plugin in Gnumeric => Tools => Plug-ins)
http://www.fileswap.com/dl/Wu84arqPKu/

Best Regards
Michał

[Nikolay Pokhilchenko]

Post by m***@public.gmane.org
Hi all,
I've finally found some time to setup printer into uncalibrated state and managed to disable all color management options.
Printed charts look very different now compared to those attached to my previous posts...
1. targen -v -d4 -p2.73 -s200 -f0 Roland_XJ-740_calibration
2. printtarg -v -ii1 -R13 -pA4 Roland_XJ-740_calibration
3. chartread -H Roland_XJ-740_calibration

printcal -v -i -p -x0 80 -x1 65 -x2 80 -x3 65 Roland_XJ-740_calibration

I.e. Cyan=Yellow=80%, Magenta=Black=65%. It's obvious for me that your pirnting resolution and drop sizing gives are somewhat excessive for this media. If it's possible you may change lower resolution to decrease the density by about 20% and increase the printing speed.

Post by m***@public.gmane.org
5. targen -v -d4 -G -f1323 Roland_XJ-740
6. printtarg -v -ii1 -R13 -pA4 -K Roland_XJ-740_calibration.cal Roland_XJ-74 I thing there should be ink overruns. At least the wery wet patches without channel and total ink limiting. It's OK for the start. But you may finish with less ink quantities.
7. chartread -H Roland_XJ-740
8. colprof -v -D"Roland_XJ-740" -qm -S sRGB.icm -P -cmt -dpp Roland_XJ-740 You skipped quite inportant stage - the black generation parameter optimization which are needed for better results. If the profile isn't intended for practical print the -S and -P parameters are useless and consumes a lot of computation time. You may check the gamut shape by "iccgamut -s". It's faster than computing of gamut mapping and corresponding B2A tables.
9. applycal -v Roland_XJ-740_calibration.cal Roland_XJ-740.icm Roland_XJ-740_calibrated.icm Now you have a profile with calibration which may be the start poing for production profile. The calibration included will help the targen to limit the total inks. But with such steep calibration curves (because of channel limiting absence) the TIL computation may be not consistent. You need to try.
1. I want to estimate TIL and individual ink limits from calibration and profile,
   so which profile should I use with iccgamut/xicclu: with calibration applied or without it? You need apply the TIL value at the targen stage. The targen uses the calibration curves (I hope uses) and deals with actual TIL. So for TIL tests you need the profile with calibration. It seems to be right because the TIL by the profile with calibration data is lower as it shoul actually be.
3. When I view printcal's charts, I'm somehow worried about black calibration curve,
   black response isn't straight line after calibration correction applied, is it ok? It's OK. But you need to play the black generation parameters hard (-kp or even -Kp) to obtain the best black behaviour in sight of gradient smoothness, ink consumption and black in the lights.
4. Nikolay, You suggested me to put calibration curves into a spreadsheet and convert them
   into RIP usable form, enter them into RIP, enter limits and then do profiling.
   I've took data from .cal file and reflected individual curves by y=x line.
   May You take a look if something like following spreadsheet will do?
   Gnueric spreadsheet (You need to enable "Time Series Analysis Functions" plugin in Gnumeric => Tools => Plug-ins)
http://www.fileswap.com/dl/Wu84arqPKu/

I haven't Gnumetric on my computers yet.

[m***@public.gmane.org]

Hi!

At this stage you've needed to limit individual channels. Such low delta E vs delta % are usable with precaultion.
printcal -v -i -p -x0 80 -x1 65 -x2 80 -x3 65 Roland_XJ-740_calibration
I.e. Cyan=Yellow=80%, Magenta=Black=65%.

Nikolay how did you determine these limits? from printcal charts?

It's obvious for me that your pirnting resolution and drop sizing gives are somewhat excessive for this media.
If it's possible you may change lower resolution to decrease the density by about 20% and increase the printing speed.

Yes, you are right, without ink limiting, prints are very wet, spacers on the calibration/profiling charts are hardly visible.

If I would transfer printcal calibration curves and limits to RIP software,
then print Argyll calibration chart with RIP calibration enabled,
might I verify RIP calibration with printcal -e?

Best regards
Michał