Printer Calibration/Verification

[Eddie Carle]

Hello. I've got an Epson Stylus Photo R2880 that I'm trying to get
working on my Gentoo system with Gutenprint 5.2.9 and ArgyllCMS 1.4.0.
This whole process is very new to me and I'm not quite sure I'm doing
things correctly. To start off, the colour was waaayyy off. I started
googling and found out about ArgyllCMS and sourced a ColorMunki. I went
through the whole "profiling" procedure, used the ICC file and the
colour was still way off. Upon further research, I decided that perhaps
I needed to go through the calibration process before profiling so that
is where I am and still not luck. The issue is that I calibrate, print a
new test chart, verify and the verification always fails miserably. Here
are the steps I am taking. Perhaps someone can shed some light?

Generate the target:
targen -v -d2 -s50 -e3 -f0 calibration

Create the TIFF image for my 13 inch roll paper:
printtarg -v -iCM -p269x330 -T600 calibration

Print the image in Gutenprint PhotoPrint ensuring that all colour
management is fully disabled.

Hookup the ColorMunki and scan the resulting print:
chartread -v -H -T0.4 calibration

Now for the calibration:
printcal -v -i calibration

Move the calibration.cal over and clear some stuff out
mv calibration.cal calibration_old.cal
rm calibration.ti2 calibration.ti3

Now for a verification print:
printtarg -K calibration_old.cal -v -iCM -p269x330.2 -T600 \
calibration

This is of course printed again in PhotoPrint with all colour management
disabled

Then I scan in again:
chartread -v -H -T0.4 calibration

Now I try to verify the calibration via
printcal -v -e calibration_old calibration

which of course fails telling me
Average white = XYZ 0.690350 0.749210 0.721547, D50 Lab
89.355977 -6.815261 -9.623604
Verify results:
Channel 0 has DE avg 6.5, rms 8.5, max 16.8
Verify results:
Channel 1 has DE avg 6.9, rms 9.1, max 22.1
Verify results:
Channel 2 has DE avg 20.4, rms 22.6, max 36.4
Verification FAILED

Then I try re-calibrating via
printcal -v -r calibration_old calibration

and start from the beginning but the problem never sorts itself out. It
doesn't seem to matter how many times I re-calibrate. Does anybody have
some insight they could share?

[BC Rider]

I now have a way (ColorMunki) to measure black density so I experimented with changing the "ink density" setting in the Epson RGB driver (on Epson 3880 with Epson Premium Luster as test paper).

The ink density can be changed +/- 50%. I changed it +20% and printed RGB=0, 2, 4 test patches. I then measured with Spotread. Dmax improved from L*=5.7 down to L*=3.9. This seems significant!

Since this is all new to me I have some questions:

1) What are the issues/pitfalls with increasing ink density in the RGB driver? What am I about to learn that everyone else already knows! lol

2) How does I determine the optimum density? I imagine issues with ink drying/stabilization, pooling, graininess from uneven drying/absorption. I also imagine the curve shape changing and even maybe reversing at some point (with too much ink due to the above problems).

3) I would like to view an overall curve for black (and maybe the RGB primaries too). What is the easiest way to do this (Argyll or otherwise)? I would make a curve with default setting and then compare with various increased ink densities looking for issues and the point of diminishing returns.


Finally, FYI Epson claims the Epson sp3880..."produces a black density up to 2.55 with an L* value of 2.9*. I definitely don't get those values with the stock settings...but perhaps with increased ink density it is possible. Maybe wider gamut too! :)

All input appreciated! Thanks.

[Alan Goldhammer]

I've used the image that Scott Martin developed to test both paper settings
and adjustments -
http://www.on-sight.com/2008/04/04/how-to-determine-the-optimal-media-select
ion-for-any-paper/ there are tradeoffs given that you might lose sharpness
because of ink pooling.



Alan



From: argyllcms-bounce-***@public.gmane.org [mailto:argyllcms-bounce-***@public.gmane.org]
On Behalf Of BC Rider
Sent: Monday, November 18, 2013 2:55 PM
To: argyllcms-***@public.gmane.org
Subject: [argyllcms] Changing "ink density" in Epson driver prior to
profiling



I now have a way (ColorMunki) to measure black density so I experimented
with changing the "ink density" setting in the Epson RGB driver (on Epson
3880 with Epson Premium Luster as test paper).

The ink density can be changed +/- 50%. I changed it +20% and printed
RGB=0, 2, 4 test patches. I then measured with Spotread. Dmax improved
from L*=5.7 down to L*=3.9. This seems significant!

Since this is all new to me I have some questions:

1) What are the issues/pitfalls with increasing ink density in the RGB
driver? What am I about to learn that everyone else already knows! lol

2) How does I determine the optimum density? I imagine issues with ink
drying/stabilization, pooling, graininess from uneven drying/absorption. I
also imagine the curve shape changing and even maybe reversing at some point
(with too much ink due to the above problems).

3) I would like to view an overall curve for black (and maybe the RGB
primaries too). What is the easiest way to do this (Argyll or otherwise)?
I would make a curve with default setting and then compare with various
increased ink densities looking for issues and the point of diminishing
returns.


Finally, FYI Epson claims the Epson sp3880..."produces a black density up
to 2.55 with an L* value of 2.9*. I definitely don't get those values with
the stock settings...but perhaps with increased ink density it is possible.
Maybe wider gamut too! :)

All input appreciated! Thanks.

[Michael Darling]

(1) Alan pointed out the big issue, sharpness. Also, of course, cost.

On canvas, I often see what you describe as graininess, from ink pooling in
the valleys, as I increase density. Looks horrible. Easy to spot on a
color target print, especially in the darker colors.

If you print alternating 1 pixel thick black/white lines, you'll find as
you increase ink density, you will likely lose the ability to see the
individual lines. This illustrates an issue for you to consider, and do
test prints of using an actual print you want to make. Largely depends on
the specific type of image you're printing. Fine detail might not matter.

To do such a test, you should make a TIFF of 360ppi or 720ppi, whichever
you use to print. (If you aren't readily familiar with these two numbers,
google "epson 360ppi 720ppi" for some very worthwhile reading. Jeff Schewe
has some nice writeups on the issue as well.)

(2) I would say there is no "optimum density". There's tradeoffs you have
to balance, but you can always have multiple ways to print based on the
content of the image.

(3) Make a print target of 256 patches, RGB 0/0/0 to 255/255/255. Print
using no color management (using Adobe Color Print Utility - free program)
to see what you're looking for, without introducing the variance of a
profile or rendering intents. You can then measure the target, and make a
chart in Excel.

I’ve used the image that Scott Martin developed to test both paper
settings and adjustments -
http://www.on-sight.com/2008/04/04/how-to-determine-the-optimal-media-selection-for-any-paper/
there are tradeoffs given that you might lose sharpness because of ink
pooling.
Alan
*Sent:* Monday, November 18, 2013 2:55 PM
*Subject:* [argyllcms] Changing "ink density" in Epson driver prior to
profiling
I now have a way (ColorMunki) to measure black density so I experimented
with changing the "ink density" setting in the Epson RGB driver (on Epson
3880 with Epson Premium Luster as test paper).
The ink density can be changed +/- 50%. I changed it +20% and
printed RGB=0, 2, 4 test patches. I then measured with
Spotread. Dmax improved from L*=5.7 down to L*=3.9. This seems
significant!
1) What are the issues/pitfalls with increasing ink density in the RGB
driver? What am I about to learn that everyone else already knows! lol
2) How does I determine the optimum density? I imagine issues with ink
drying/stabilization, pooling, graininess from uneven drying/absorption.
I also imagine the curve shape changing and even maybe reversing at some
point (with too much ink due to the above problems).
3) I would like to view an overall curve for black (and maybe the RGB
primaries too). What is the easiest way to do this (Argyll or
otherwise)? I would make a curve with default setting and then compare
with various increased ink densities looking for issues and the point of
diminishing returns.
Finally, FYI Epson claims the Epson sp3880..."produces a black density up
to 2.55 with an L* value of 2.9*. I definitely don't get those values
with the stock settings...but perhaps with increased ink density it is
possible. Maybe wider gamut too! :)
All input appreciated! Thanks.

[Graeme Gill]

This post might be inappropriate. Click to display it.

[BC Rider]

Post by Graeme Gill
One of the problems with an RGB driver is that you don't have the tools to
tackle these problems properly.

I was readingyour Printcal tools and beginning to get that picture. At this point I don't know what +/- 20% on the Epson RGB driver equates to...maybe it is a big or small change. In any case, visual inspection, a few measurements and reprofiling seems about all I can do without going to a RIP (which exceeds my interest level). In any case, will Printcal give me the basic curves I'm looking for? The advantage is automatic target generation, measurement and graphing! As a test, I tried: targen -d2 -e0 - s50 -f0 testfilefakeread Epson_premium_luster_profile.icm testfileprintcal -p -i -d testfile It produced graphs for the rgb "primaries" that seem useable. But I could not get Printcal to produce a black/grey curve (ie. based on targen -g50). Is that possible?

Post by Graeme Gill
The test chart pointed to by Alan Goldhammer seems a good start.
Yes, looks brilliant. Might be all I need. I've downloaded it. Thanks Alan. Thanks for the comments everyone. I've gone from knowing nothing about changing the ink density to knowing just enough to be dangerous! lol

[Michael Darling]

P.S. Keep in mind there are rare occasions when it helps to use a negative
print density, if the material you are printing on just can't handle the
ink load associated with the media type you're using.

Post by BC Rider

Post by Graeme Gill
One of the problems with an RGB driver is that you don't have the tools

to

Post by Graeme Gill
tackle these problems properly.

I was readingyour Printcal tools and beginning to get that picture. At
this point I don't know what +/- 20% on the Epson RGB driver equates
to...maybe it is a big or small change. In any case, visual inspection, a
few measurements and reprofiling seems about all I can do without going to
a RIP (which exceeds my interest level).
In any case, will Printcal give me the basic curves I'm looking for?
The advantage is automatic target generation, measurement and graphing!
targen -d2 -e0 - s50 -f0 testfile
fakeread Epson_premium_luster_profile.icm testfile
printcal -p -i -d testfile
It produced graphs for the rgb "primaries" that seem useable. But I
could not get Printcal to produce a black/grey curve (ie. based on targen
-g50). Is that possible?

Post by Graeme Gill
The test chart pointed to by Alan Goldhammer seems a good start.

Yes, looks brilliant. Might be all I need. I've downloaded it. Thanks
Alan.
Thanks for the comments everyone. I've gone from knowing nothing about
changing the ink density to knowing just enough to be dangerous! lol

[Ernst Dinkla]

Post by Graeme Gill
Hi,

Post by BC Rider
1) What are the issues/pitfalls with increasing ink density in the RGB driver? What
am I about to learn that everyone else already knows! lol

Ink running off the page is the most obvious. Other issues would
be a change in the ink channel curve shape. But because you are
dealing with an RGB emulation on top of it, you can probably expect
the overall RGB response to change to some degree as well.

Post by BC Rider
2) How does I determine the optimum density? I imagine issues with ink
drying/stabilization, pooling, graininess from uneven drying/absorption. I also
imagine the curve shape changing and even maybe reversing at some point (with too much
ink due to the above problems).

One of the problems with an RGB driver is that you don't have the tools to
tackle these problems properly. For good control, you need to be able to
set and correct each physical ink channel. But I guess if your change is small,
you might get away with simply profiling it in RGB space.

Post by BC Rider
3) I would like to view an overall curve for black (and maybe the RGB primaries too).
What is the easiest way to do this (Argyll or otherwise)? I would make a curve with
default setting and then compare with various increased ink densities looking for
issues and the point of diminishing returns.

The test chart pointed to by Alan Goldhammer seems a good start.
Graeme Gill.

The ink density settings in OEM printer drivers are often gamma
controlled and not comparable to ink limit settings in RIPs. While there
can be gain in Dmax (more often not) the bleeding/dot gain/detail loss
can happen in other parts of the tone range.
Selecting another higher ink load media preset and adjusting the density
setting to lower values can be as good or better. HP offered quite some
information on the ink load of different media presets for the Z3100 and
Z3200. I did not see the same in Canon and Epson documents.

For the QuadTone RIP I made some greyscale targets that show detail loss
at more spots along the tone range. Can be used for a normal driver too.
http://www.pigment-print.com/Quad%20QTR/Index.html

[BC Rider]

Post by Ernst Dinkla
For the QuadTone RIP I made some greyscale targets that show detail loss
at more spots along the tone range. Can be used for a normal driver too.

All very interesting.

Last night I printed test charts at 0%, 20% and 40% ink density using the Onsight image Alan mentioned. Visually, the 20% had more contrast with stronger colors, the 40% even more so. Placed beside the 40% print the original 0% print looked washed out.

The 0% and 20% had none of the problems or issues we have been discussing, both were perfectly fine. The 40% has some very slight mottling in a couple spots of deep black.

Black Dmax went from L* 5.2 down to L* 2.7 and then further to L* 2.6 with the 40% version. The grey steps were all visible except in the 40% where I couldn't separate the last two steps.

Note: after 24 hours the ink seems to have settled at 4.9 and 2.9 and 2.9 respectively. Contrary to what X-rite claims with their Munki software, 10 minutes is *NOT* sufficient time to allow for inks to stabilize!

This morning I printed again, this time with the test target Ernst was kind enough to share. All lines were visible in all patches for all three prints (0, 20, 40% ink density). No apparent lost of sharpness or resolution.

Eventually I will make profiles for the 0% and 20% ink density and compare. But at this point it looks like the +20% is a winner from this set. Perhaps a lesser increase will provide most of the benefit. More testing is needed.

Since the patches on the left side are becoming more vibrant, I don't think the ink changes are due to gamma. One wonders what the media type selections really do vis-à-vis the ink density setting. I could find nothing describing Epson drivers in this regard.

In any case, based on these test, optimizing the media and ink settings in the RGB driver should be the first step in the printer profiling process! Unfortunately it seems to be a lot of printing and subjective "eyeballing" for RGB drivers. There are a lot of possible combinations of media type and ink density. It would be nice to automate and quantify this process somehow!

[edmund ronald]

One use for the density setting is when you have a non-Epson paper to print
on, and you can then use an existing setting as a base, with a modded
density.

Buckling, blurriness etc are some of the effects if you dump too much ink
on paper; it used to be that Epson's default settings were too "heavy" for
default use, but I think that is in the past.

Edmund

Post by BC Rider

Post by Ernst Dinkla
For the QuadTone RIP I made some greyscale targets that show detail loss
at more spots along the tone range. Can be used for a normal driver too.

All very interesting.
Last night I printed test charts at 0%, 20% and 40% ink density using the
Onsight image Alan mentioned. Visually, the 20% had more contrast with
stronger colors, the 40% even more so. Placed beside the 40% print the
original 0% print looked washed out.
The 0% and 20% had none of the problems or issues we have been
discussing, both were perfectly fine. The 40% has some very slight
mottling in a couple spots of deep black.
Black Dmax went from L* 5.2 down to L* 2.7 and then further to L* 2.6 with
the 40% version. The grey steps were all visible except in the 40% where
I couldn't separate the last two steps.
Note: after 24 hours the ink seems to have settled at 4.9 and 2.9 and 2.9
respectively. Contrary to what X-rite claims with their Munki software,
10 minutes is *NOT* sufficient time to allow for inks to stabilize!
This morning I printed again, this time with the test target Ernst was
kind enough to share. All lines were visible in all patches for all three
prints (0, 20, 40% ink density). No apparent lost of sharpness or
resolution.
Eventually I will make profiles for the 0% and 20% ink density and
compare. But at this point it looks like the +20% is a winner from this
set. Perhaps a lesser increase will provide most of the benefit. More
testing is needed.
Since the patches on the left side are becoming more vibrant, I don't
think the ink changes are due to gamma. One wonders what the media type
selections really do vis-à-vis the ink density setting. I could find
nothing describing Epson drivers in this regard.
In any case, based on these test, optimizing the media and ink settings in
the RGB driver should be the first step in the printer profiling process!
Unfortunately it seems to be a lot of printing and subjective "eyeballing"
for RGB drivers. There are a lot of possible combinations of media type
and ink density. It would be nice to automate and quantify this process
somehow!

[BC Rider]

Date: Tue, 19 Nov 2013 14:12:22 +0100
Subject: [argyllcms] Re: Changing "ink density" in Epson driver prior to profiling
Selecting another higher ink load media preset and adjusting the density
setting to lower values can be as good or better.

I've searched and not found anything from Epson that provides a roadmap to their media presets. I've been reading about RIPs to gain an understanding of the issues. Frankly, given 8 inks it all looks terribly complicated to get right and there must be a great deal of "secret sauce" in the Epson drivers. There are three photo paper media choices in the Epson driver (Luster, Glossy and Semi-Gloss) plus a separate Exhibition Fiber setting. There are three fine art paper choices, plus three more matte paper choices, plus separate Hot Press and Cold Press choices. Presumably each one has its own ink limits, total limits, black generation strategies, ink transitions from lighter to darker inks, N=8 color mixing strategies etc. All hidden in the driver by Epson and unknown to us. Some have suggested the order Epson lists the paper has some significance (and it may very well) but I haven't seen it documented. Also some variables may be based more on the ink than the paper. Others may change with media but in a manner easily compensated by profiling. So I wonder how much of significance *actually* changes with media type. How do people predict the behavior of different media presets? Or for that matter, how does one even know which media preset produces a higher or lower ink load without printing test images for every possibility? For Epson papers I generally assume they "got it right" so only (minor) changes to the ink density setting is worthwhile. But for 3rd party papers and especially fine art/canvas more radical experimentation is probably warranted.

[Graeme Gill]

Post by Eddie Carle
I've got an Epson Stylus Photo R2880 that I'm trying to get
working on my Gentoo system with Gutenprint 5.2.9 and ArgyllCMS 1.4.0.

Hi,
unfortunately I know little about Gutenprint or how it
deals with color management. Perhaps others with more experience
in this area (Edmund ?) could chime in. You might also consider
posting you question to the OpenICC mailing list, rather than
here, if it comes to the intricacies of Gutenprint.

Post by Eddie Carle
This whole process is very new to me and I'm not quite sure I'm doing
things correctly. To start off, the colour was waaayyy off. I started
googling and found out about ArgyllCMS and sourced a ColorMunki. I went
through the whole "profiling" procedure, used the ICC file and the
colour was still way off. Upon further research, I decided that perhaps
I needed to go through the calibration process before profiling so that
is where I am and still not luck.

If you are dealing with raw inkjet printer, then printer setup
and calibration have to be the first steps.

Basic setup is:

Resolution and screening type.

Crossovers (for forming light + dark ink into a single
channel)

Per channel gross ink limiting.

then you can do calibration:

Per channel fine ink limiting + linearisation/dot gain correction

then you can profile it.

Post by Eddie Carle
The issue is that I calibrate, print a
new test chart, verify and the verification always fails miserably. Here
are the steps I am taking. Perhaps someone can shed some light?
targen -v -d2 -s50 -e3 -f0 calibration
printtarg -v -iCM -p269x330 -T600 calibration
Print the image in Gutenprint PhotoPrint ensuring that all colour
management is fully disabled.
chartread -v -H -T0.4 calibration
printcal -v -i calibration
Move the calibration.cal over and clear some stuff out
mv calibration.cal calibration_old.cal
rm calibration.ti2 calibration.ti3
printtarg -K calibration_old.cal -v -iCM -p269x330.2 -T600 \
calibration
This is of course printed again in PhotoPrint with all colour management
disabled
chartread -v -H -T0.4 calibration
Now I try to verify the calibration via
printcal -v -e calibration_old calibration
which of course fails telling me
Average white = XYZ 0.690350 0.749210 0.721547, D50 Lab
89.355977 -6.815261 -9.623604
Channel 0 has DE avg 6.5, rms 8.5, max 16.8
Channel 1 has DE avg 6.9, rms 9.1, max 22.1
Channel 2 has DE avg 20.4, rms 22.6, max 36.4
Verification FAILED
and start from the beginning but the problem never sorts itself out. It
doesn't seem to matter how many times I re-calibrate. Does anybody have
some insight they could share?

The basic procedure looks OK. Can you point me to the files though, so
I can have a look at them ?

Graeme Gill.