Digital BW, The Print

I'm trying to use QTR's create-icc-rgb to linearize my workflow for
UT7 inks. I'm following Paul Roark's recipe for combining a linearization
with his Photoshop curves (http://www.paulroark.com/BW-Info/
Embedding_Photoshop_Curves_in_ICCs.pdf).

It seems to almost work. The problem is that the resulting ICC profile
is not at all linear at the high density end, and my shadows
get blocked up. I have tried both 21 step and 51 step targets, and
I have the same problem with each. I'm using a PrintFIX Pro spectrometer
and its measure tool to generate the input for create-icc.

Any idea what's going on? Has anyone seen this problem before?

--Daniel

Hi Daniel,

>I'm trying to use QTR's create-icc-rgb to linearize my workflow for
>UT7 inks. I'm following Paul Roark's recipe for combining a linearization
>with his Photoshop curves (http://www.paulroark.com/BW-Info/
>Embedding_Photoshop_Curves_in_ICCs.pdf).

>It seems to almost work. The problem is that the resulting ICC profile
>is not at all linear at the high density end, and my shadows
>get blocked up. ... I'm using a PrintFIX Pro spectrometer
>and its measure tool to generate the input for create-icc.

You're probably noticing the difference between how a "color managed"
workflow prints the image and how the more linear standard used for the old
curves handled the image.

The ICC reads your working gray space and uses those standards.  I, for
example, now use the standard Gray Gamma 2.2 as my gray working space.
Unlike the strictly linear printing of the popular rips and my old curves,
Gray Gamma 2.2 compresses the shadow tones significantly.  As such, your
images will print darker.  The good news is that, with a calibrated monitor,
the prints should match the monitor better even if they do not match the
prints that you used to print with the curves and the non-color managed
workflow.

There are several ways to coordinate the view and print.  Since I prefer to
use the standard GG 2.2, I use a layer to correct for a printing approach
that does not read my working space. (I'll send one off list.)

Paul
www.PaulRoark.com

Hi, Paul! 

I have the same problem as Daniel with blocked shadows when printing 
with icc that have been created from your curves as per embedding 
Photoshop curves instructions.  The inkset is UT7, printer Epson 2100 
and I am using the PrintFix Pro spectro. Lates version of Quadtone 
RIP. 

First I will describe my icc creation process to make sure that I am 
doing everything correctly.

1) When printing Paul's Grey Gamma 2.2 21-step wedge for profile 
creation, I open it in PS, change mode to RGB (in my case Adobe RGB), 
apply selected curve (in my case EEM carbon) and print it with color 
management off (both in PS and in printer driver).  I get these LAB 
values

96.47	0.83	-0.52
91.77	1.20	1.26
87.66	1.50	2.93
84.51	1.75	4.06
81.10	1.94	5.10
78.27	2.15	5.73
74.96	2.32	6.31
71.08	2.52	6.89
67.74	2.65	7.21
64.32	2.73	7.40
61.00	2.85	7.47
58.30	2.87	7.40
54.48	2.85	7.31
50.07	2.78	6.91
45.65	2.56	6.11
41.53	2.39	5.05
37.45	2.14	3.56
32.18	1.69	2.11
26.90	1.28	1.14
20.37	0.97	0.92
16.27	0.46	0.25 

I create the ICC profile from the measurement TXT file and the RGG-
RAW-16.RAW.

Then I print the same GG 2.2 21-step wedge from PS using the newly 
created profile (using perceptual with BPC) - in this case I leave 
the file in greyscale, because the profile takes care of "converting" 
it to RGB.  The shadows get blocked up and the result is much LESS 
linear using ICC than it would be using simple curves approach. These 
are the measurments.

96.54	0.72	-0.74
92.15	1.01	0.87
88.30	1.34	2.48
84.22	1.63	3.98
80.09	1.92	5.15
75.99	2.17	6.06
71.70	2.40	6.76
67.20	2.61	7.19
63.24	2.70	7.39
58.82	2.81	7.41
54.24	2.78	7.24
49.49	2.66	6.70
45.01	2.45	5.96
39.97	2.14	4.40
35.57	1.84	2.92
30.55	1.36	1.64
26.69	1.12	1.10
22.54	1.05	1.10
17.45	0.65	0.55
16.16	0.43	0.27
16.10	0.50	0.19

I have two questions really

1) Perhaps I am doing something wrong?

2) If I am doing it coreectly and the above results are to be 
expected, then I must confess that I do not quite understand the 
approach that you suggested to Daniel by using a correction layer, 
nor do I understand what you suggest to be the cause of such blocked 
shadoows. Could you explain it a bit more and perhaps send the 
correction layer to my address off list.  

The way I see it it is not a "view to print matching" problem, 
because in my case a file with known and "correct" values (21-step 
wedge) prints wrongly though ICC. 


Best regards

Didzis

PS
Unrelated issue I have is that my Windows XP appears to have problems 
recognizing these icc. First, I cannot install the profile 
using "install profile" - I get an error message about invalid color 
profile. I copy the icc manually into windows icc directory - but 
even then Photoshop only recognizes the icc only after I "repair" the 
file name/internal name mismatch error that these profiles have.  



> Hi Daniel,
> 
> You're probably noticing the difference between how a "color 
managed"
> workflow prints the image and how the more linear standard used for 
the old
> curves handled the image.
> 
> The ICC reads your working gray space and uses those standards.  I, 
for
> example, now use the standard Gray Gamma 2.2 as my gray working 
space.
> Unlike the strictly linear printing of the popular rips and my old 
curves,
> Gray Gamma 2.2 compresses the shadow tones significantly.  As such, 
your
> images will print darker.  The good news is that, with a calibrated 
monitor,
> the prints should match the monitor better even if they do not 
match the
> prints that you used to print with the curves and the non-color 
managed
> workflow.
> 
> There are several ways to coordinate the view and print.  Since I 
prefer to
> use the standard GG 2.2, I use a layer to correct for a printing 
approach

> that does not read my working space. (I'll send one off list.)
> 
> Paul

Hi Didzis,


>... problem ... with blocked shadows when printing 
>with icc that have been created from your curves as 
>per embedding Photoshop curves instructions. 
>The inkset is UT7, printer Epson 2100 
>and I am using the PrintFix Pro spectro. Latest version 
>of Quadtone RIP. 

> ... my icc creation process ...

Looks OK.



>1) When printing Paul's ... curve ...

With my curves on the AdobeRGB version of the file, your 50% test patch has
a Lab L of 61. That's a bit light.  A strictly linear response, which is
close to what the curves were originally made for, would be half way between
the paper white and 100% black.  That point is L = 56. (Note that my old
target was density = 0.61, which would be about L 56.6.)

>I create the ICC profile ...

>Then I print the same GG 2.2 21-step wedge from PS using 
>the newly created profile ...
>The shadows get blocked up and the result is much 
>LESS linear using ICC than it would be using simple curves ...

Yes, Gray Gamma 2.2 is not linear and the old curves target was more linear.
But the ICC should -- if the monitor is calibrated and the systems are
working right -- make a print that matches the monitor better than the old
curves.  Most of us used some method of overriding the usual gray spaces
such as a custom dot gain curve or the like to better match the monitor to
the prints.

With the ICC, your 50% patch is L = 54.  I think that's about right.

In fact, with the ICC the ramp is very linear to 90%.  Then the problem is
that 90% to 100% has very little separation.  The 50% to 100% usually has
very little separation in Gray Gamma 2.2.  The 90% to 95% is, however, less
than I would expect.  I'm guessing that the Create ICC RGB program is simply
working with too little information about what is going on in the deep
shadows.  It's probably assuming a straight line response between the 95%
and 100% input points when, in fact, the output gets much darker sooner than
the program assumes. 

I'm not sure if the Create ICC programs will accept 51 step inputs, but most
of QTR will.  That might get enough information to the program to correct
the problem.  Be sure to print the target vertically, as I've found it more
accurate that way.  


>... I do not quite understand the approach that you 
> suggested to Daniel by using a correction layer, 

If a file was edited for a linear output and is then printed in a Gray Gamma
2.2 space, the shadows will be crushed.  A correction layer is just a way to
use older files (edited for linear spaces) in GG 2.2 (or vice versa) without
having to re-edit the files.  I wish all our workflows, spaces, etc. were
consistent, but they just are not.  

If the curves work for you and match your monitor, then by all means stay
with that workflow.  I think, however, for most people it would be easier to
stay with a standard gray working space like GG 2.2 and use a workflow that,
at least ideally, should match the print to a calibrated monitor without the
need for any special layers or soft proofing procedures.

>Unrelated issue I have is that my Windows XP appears to 
>have problems recognizing these icc.

True, only Photoshop and Elements print previews recognize these grayscale
ICCs.

> I copy the icc manually into windows icc directory - but 
>even then Photoshop only recognizes the icc only after I 
>"repair" the file name/internal name mismatch error that 
>these profiles have.

Is this the extension problem with the old GTE ISP?  That is, have you been
able to download the ICC with the *.icc extension in tact, yet you still get
this error?  If the latter, I'm not sure what the problem is.

I've changed the link in the UT7 readme files now to go to the new ISP.
Hopefully the extension problem will be corrected now.  To download use
normal procedures -- right click, save as -- not extension correction shoud
be needed.

Paul
www.PaulRoark.com

Hi, Paul!

Thank you very much for your explanation. It is getting a bit 
clearer for me :).

First of all I have just realized that although I have downloaded 
and instaled the most recent QTR version, somehow the QTR-Create-ICC 
RGB had remained from an earlier version 2.4.2.0, (the latest one 
being 2.5.1.0)  I will see if I can clean the installation and have 
the latest version installed.  Perhaps using the new version my 
shadow problem goes away - although the release notes do not refer 
to any improvements in QTR-Create-ICC RGB.  
 
> With my curves on the AdobeRGB version of the file, your 50% test 
patch has
> a Lab L of 61. That's a bit light.  A strictly linear response, 
which is
> close to what the curves were originally made for, would be half 
way between
> the paper white and 100% black.  That point is L = 56. (Note that 
my old
> target was density = 0.61, which would be about L 56.6.)

I guess this means that if I stay with the curves approach, I would 
ideally need to adjust the curves to make the response more linear, 
right?

> Yes, Gray Gamma 2.2 is not linear and the old curves target was 
more linear.
> But the ICC should -- if the monitor is calibrated and the systems 
are
> working right -- make a print that matches the monitor better than 
the old
> curves.  

I have a calibrated monitor with spyder at 6500K and Gamma 2.2 and I 
would say that I get a better match with the old curves approach.  
The shadows in the prints are just compressed too much.  But I 
digress :).


> 
> With the ICC, your 50% patch is L = 54.  I think that's about 
right.

Yes, this I confirmed by printing real images - I think there is 
improvement in the midtones in prints made with ICC. It is only in 
90-100% range that the problem starts. 


> In fact, with the ICC the ramp is very linear to 90%.  Then the 
problem is
> that 90% to 100% has very little separation.  The 50% to 100% 
usually has
> very little separation in Gray Gamma 2.2.  The 90% to 95% is, 
however, less
> than I would expect.  I'm guessing that the Create ICC RGB program 
is simply
> working with too little information about what is going on in the 
deep
> shadows.  It's probably assuming a straight line response between 
the 95%
> and 100% input points when, in fact, the output gets much darker 
sooner than
> the program assumes. 
> 
> I'm not sure if the Create ICC programs will accept 51 step 
inputs, but most
> of QTR will.  That might get enough information to the program to 
correct
> the problem.  Be sure to print the target vertically, as I've 
found it more
> accurate that way.  
 
I will give the 51-step wedge a try.  Can I use the one that is 
included with the Quadtone RIP Eye-One folder?


> A correction layer is just a way to
> use older files (edited for linear spaces) in GG 2.2 (or vice 
versa) without
> having to re-edit the files.  I wish all our workflows, spaces, 
etc. were
> consistent, but they just are not.  

> If the curves work for you and match your monitor, then by all 
means stay
> with that workflow.  

I will give the correction layer a try. If it works I will have to 
see if and how it could fit in the workflow.  The reason that I am 
interested in icc approach in the first place is simplification of 
the process - using icc instead of curves would make printing so 
much easier from Qimage (that I have just "discovered") or 
eventually maybe even from Lightroom when the printing end of that 
application improves. 


Thanks for all the help so far. I will report on my further progress 
with this.

Best regards

Didzis

Didzis,

>... Perhaps using the new version my shadow problem goes away ...

Probably not.


>... if I stay with the curves approach, I would 
>ideally need to adjust the curves to make the response 
>more linear, right?

Well, that was my very early assumption -- that people would tweak the
curves as needed to make them more "custom" profiles instead of "canned"
profiles.  However, few actually do this.  That is part of the attempt to
move to a more automated systems using Create ICC with the curves.  However,
I see that it may be causing other problems.

All of this has caused me to move have some approaches that are easier --
like the C86 "EZ" inkset, the 220 UT_R2, and now the 1800 3-MK 100% Eboni
carbon.  These mono inksets that allow one to re-linearize with QTR (or
Create ICC) are more reliable and less subject to problems.  The 3-MK is
particularly reliable because there is no mixing by MIS, there are no
cross-overs, and the approach has virtually no sensitivity to ink batch
variances.  In fact, the same underlying QTR curves set is used for every
paper.  So, bottom line, I recognize that the older inksets often involved
too much complexity and sensitivity to variables.



>> ... the ICC should -- if the monitor is calibrated and the 
>> systems are working right -- make a print that matches the 
>> monitor better than the old curves. 

>I have a calibrated monitor with spyder at 6500K 

(cold, try 5000)

>and Gamma 2.2

(probably not relevant)

> and I would say that I get a better match with the old 
> curves approach. 

That's odd.  Gray Gamma 2.2 usually compresses the 95 - 100%, whereas my old
curves had good separation between these points.

>The shadows in the prints are just compressed too much. 

Yes, I agree with that.

The truth is, I suspect perfectionists may still make their own dot gain
curves or viewing layers -- or they have much better equipment than I do.  I
do not find the "calibration" systems I've used perfect by any means,
particularly for B&W.  But they are really needed for color.  When you think
about it, in B&W we use 21 steps as our standard test strip, many use much
larger ones.  With color these systems need to calibrate a 3 dimensional
space.  To have 21-steps in each color, that might require 21 to the third
power -- 9261.  I'm too lazy to read a chart close to this large without a
strip reader, but I suspect the pros with top equipment do.

> 
> It is only in 90-100% range that the problem starts. 

Yes, and Gray Gamma 2.2 usually compresses this zone, although not as much
as your results in the 95-100% range.  So, I'm not sure even a "perfect" ICC
is going to give you a great match between your monitor and print.
Something else is going on. 

Do you use an LCD monitor that is very bright?  I found on some of my
family's laptops that print make with my curves matched them better (when
they were plugged in) than the ICC output did.

>I will give the 51-step wedge a try. Can I use the one that is 
>included with the Quadtone RIP Eye-One folder?

It's worth a try.  I recommend printing vertically and in the middle of the
page.  Try to avoid any area that is going to be touched with a roller.


Good luck.

Paul
www.PaulRoark.com

Paul!

> 
> All of this has caused me to move have some approaches that are 
easier --
> like the C86 "EZ" inkset, the 220 UT_R2, and now the 1800 3-MK 100% 
Eboni
> carbon.  

I am aware of these approaches, however I cannot really justify 
changing a printer (I only print for myself).  So I need something 
that fits in my 2100.  As I said I am quite happy with the curves 
approach, I just thought that icc's would make printing from various 
applications easier.

Even if do not succeed with the UT7 and ICC, I can give the UT3D a 
shot using the PrintFix Pro.



> Do you use an LCD monitor that is very bright?  I found on some of 
my
> family's laptops that print make with my curves matched them better 
(when
> they were plugged in) than the ICC output did.


Yes it is an LCD NEC 1990SXI, but I have turned the backlight down 
and calibrated it to 100 cd m (but my ambient light is also very 
low).  


Thanks for your help.  I will let you know how I fare.

Didzis

Hi, Paul!

> >I will give the 51-step wedge a try. Can I use the one that is 
> >included with the Quadtone RIP Eye-One folder?
> 
> It's worth a try.  I recommend printing vertically and in the 
middle of the
> page.  Try to avoid any area that is going to be touched with a 
roller.
> 

The 51-step wedge might just be the answer.  I used the one that is 
in the eye-one folder and I believe the resulting ICC is better in 
terms of dark tones separation. I made an ICC from the carbon curve. 
Then I printed your grey gamma 2.2 21-step wedge using this ICC.

Here are the figures.

96.73	0.94	-0.44
92.77	1.21	1.14
88.23	1.56	2.86
83.90	1.88	4.38
80.10	2.07	5.42
75.68	2.35	6.34
71.21	2.54	6.91
66.93	2.69	7.26
62.71	2.78	7.53
58.27	2.92	7.50
53.77	2.90	7.36
48.58	2.74	6.74
43.84	2.50	5.84
39.09	2.20	4.19
34.31	1.92	2.63
29.58	1.56	1.46
25.83	1.16	1.10
22.95	1.17	1.13
19.54	0.83	0.88
17.02	0.72	0.38
16.35	0.60	0.18

Looks better to me. The separation between 95 and 100% is not big, 
but it is much better than before, and also the whole 90-100% range 
is better separated now. What do you think? Are these kinds of 
figures to be expected in the 90-100% range? Visually the new step 
wedge has better shadow separation, howver I still have a hard time 
distinguishing 95% and 100%.  

For the sake of completnesss I here include also the measurement of 
the 51-step wedge (with carbon curve applied), from which the ICC was 
made.

96.73	0.95	-0.41
94.84	0.98	0.23
92.62	1.07	1.05
91.22	1.26	1.61
89.68	1.39	2.24
87.99	1.50	2.87
86.86	1.68	3.35
85.26	1.73	3.94
84.32	1.86	4.24
82.84	1.94	4.59
80.87	2.03	5.11
79.76	2.11	5.41
78.60	2.20	5.64
77.70	2.18	5.83
76.57	2.32	6.09
74.96	2.40	6.36
73.52	2.46	6.61
71.99	2.58	6.85
70.50	2.63	6.98
68.56	2.56	7.13
67.52	2.72	7.23
66.01	2.72	7.20
64.51	2.81	7.47
63.43	2.85	7.39
62.31	2.87	7.35
60.96	2.84	7.40
59.80	2.89	7.38
58.79	2.90	7.35
57.39	2.93	7.35
56.13	2.94	7.27
54.02	2.90	7.12
52.55	2.86	6.97
50.89	2.85	6.73
49.32	2.73	6.39
47.54	2.62	6.12
46.04	2.59	5.78
44.44	2.51	5.45
42.99	2.44	5.22
41.14	2.32	4.53
39.38	2.30	3.92
37.70	2.27	3.52
36.05	2.11	3.05
33.98	1.92	2.52
31.49	1.78	1.93
29.19	1.55	1.43
27.40	1.50	1.24
23.83	1.28	1.21
20.88	1.16	0.92
18.07	0.85	0.66
16.88	0.63	0.42
16.48	0.53	0.26

Best regards

Didzis

Hi Didzis,

 

>The 51-step wedge might just be the answer.

> the resulting ICC is better in terms of dark tones separation. . 

 

I was just pondering the easiest way to accomplish this goal.  The problem
is, I suspect, that the 95 - 100%  black ramps are not very linear.  The
responses of the ink-paper combinations become very non-linear there.  But
our 21-step curves just don't have the information in them that is needed.
This makes Create ICC just assume something - probably linear or a smooth
exponential curve - and that assumed ink response is just not close enough
to what we're actually getting. 

 

Good 51-step test strips give more data points there.  Some of the high end
rips have even more test patches.

 

The very large test strips, however, are not the easy to use with desktop
printers and spectros that are not strip readers.  Roller marks and other
problems distort the information.  I found that vertical printing helps.
Unfortunately, a letter size sheet is not enough room for a straight line of
51 test patches large enough to read.  If there are 2 columns, then one has
to watch out for a difference introduces by a roller. 

 

I use a 21-step but then also do a 90-100% 1% test strip.  In fact, I do it
as part of setting up a printer.  It helps me nail the dmax.  

See http://www.paulroark.com/BW-Info/Fine%20Tuning%20the%20Dmax.pdf  

 

I've found the highlights and even midtones (with the workflows I mostly use
now) are usually not a problem.  There I think the 5% steps of the 21-step
test strip are enough. 

 

I'm not sure where the "sweet spot" is in terms of time and effort,
cost-benefit with respect to the darkest 5%.  If the dmax is "correctly" set
by the original profile, or the ink matches the existing driver assumptions
(testing the 260 at the moment), then it may be that the 21-steps is enough.
However, I think that some of the older profiles with steep cross-overs and
curves may be sensitive to printer and inkset variations.  

 

Paul

www.PaulRoark.com <http://www.paulroark.com/>  

 

 

 

 

  _____  

From: DigitalBlackandWhiteThePrint@yahoogroups.com
[mailto:DigitalBlackandWhiteThePrint@yahoogroups.com] On Behalf Of dizpark
Sent: Sunday, September 09, 2007 1:57 PM
To: DigitalBlackandWhiteThePrint@yahoogroups.com
Subject: Re: [Digital BW] create-icc problem

 

Hi, Paul!

> >I will give the 51-step wedge a try. Can I use the one that is 
> >included with the Quadtone RIP Eye-One folder?
> 
> It's worth a try. I recommend printing vertically and in the 
middle of the
> page. Try to avoid any area that is going to be touched with a 
roller.
> 

The 51-step wedge might just be the answer. I used the one that is 
in the eye-one folder and I believe the resulting ICC is better in 
terms of dark tones separation. I made an ICC from the carbon curve. 
Then I printed your grey gamma 2.2 21-step wedge using this ICC.

Here are the figures.

96.73 0.94 -0.44
92.77 1.21 1.14
88.23 1.56 2.86
83.90 1.88 4.38
80.10 2.07 5.42
75.68 2.35 6.34
71.21 2.54 6.91
66.93 2.69 7.26
62.71 2.78 7.53
58.27 2.92 7.50
53.77 2.90 7.36
48.58 2.74 6.74
43.84 2.50 5.84
39.09 2.20 4.19
34.31 1.92 2.63
29.58 1.56 1.46
25.83 1.16 1.10
22.95 1.17 1.13
19.54 0.83 0.88
17.02 0.72 0.38
16.35 0.60 0.18

Looks better to me. The separation between 95 and 100% is not big, 
but it is much better than before, and also the whole 90-100% range 
is better separated now. What do you think? Are these kinds of 
figures to be expected in the 90-100% range? Visually the new step 
wedge has better shadow separation, howver I still have a hard time 
distinguishing 95% and 100%. 

For the sake of completnesss I here include also the measurement of 
the 51-step wedge (with carbon curve applied), from which the ICC was 
made.

96.73 0.95 -0.41
94.84 0.98 0.23
92.62 1.07 1.05
91.22 1.26 1.61
89.68 1.39 2.24
87.99 1.50 2.87
86.86 1.68 3.35
85.26 1.73 3.94
84.32 1.86 4.24
82.84 1.94 4.59
80.87 2.03 5.11
79.76 2.11 5.41
78.60 2.20 5.64
77.70 2.18 5.83
76.57 2.32 6.09
74.96 2.40 6.36
73.52 2.46 6.61
71.99 2.58 6.85
70.50 2.63 6.98
68.56 2.56 7.13
67.52 2.72 7.23
66.01 2.72 7.20
64.51 2.81 7.47
63.43 2.85 7.39
62.31 2.87 7.35
60.96 2.84 7.40
59.80 2.89 7.38
58.79 2.90 7.35
57.39 2.93 7.35
56.13 2.94 7.27
54.02 2.90 7.12
52.55 2.86 6.97
50.89 2.85 6.73
49.32 2.73 6.39
47.54 2.62 6.12
46.04 2.59 5.78
44.44 2.51 5.45
42.99 2.44 5.22
41.14 2.32 4.53
39.38 2.30 3.92
37.70 2.27 3.52
36.05 2.11 3.05
33.98 1.92 2.52
31.49 1.78 1.93
29.19 1.55 1.43
27.40 1.50 1.24
23.83 1.28 1.21
20.88 1.16 0.92
18.07 0.85 0.66
16.88 0.63 0.42
16.48 0.53 0.26

Best regards

Didzis

 



[Non-text portions of this message have been removed]