QTR-Quadtone RIP

Hello,

my name is Gottlieb and I'm new in this group.

I'm trying to create some grayscale ICC profiles for the Epson 3800
ABW driver using the QTR-Create-ICC.exe and Datacolor Spyder3Print
(also knwon as PrintFix PRO with the Datacolor 1005 device).

On my glossy proof paper the Epson 3800 ABW (dark setting) is almost
linear as it should be.

The generated Grayscale ICC Profile should only correct the L* curve
only marginally. But instead it seems that the calibration does the
contrary.

But may be the caibration does all right and I make an error in
reasoning or calibrating.

The way I created the grayscale ICC profiles was the following:

1) Open the file "Step-21-gray.tif" in Photoshop CS2 without CMS

2) Print it with no CMS and the Epson 3800 ABW driver set to the
setting "Dark"

3) Let the print dry over night

4) Read the 21 Patches with Spyder3Print and save it to an Create-ICC
formated file.

5) Drag the file to QTR-Create-ICC.exe

6) Install the Profile

7) Open "Step-21-gray.tif" in Photoshop CS2 without CMS

8) Print it with CMS managed by Photoshop, the QTR-ICC pofile and the
Epson 3800 ABW driver set to the setting "Dark" with the same media
settings

9) Read the 21 Patches with Spyder3Print and save the L* vaules to an
file.

10) Import the L* Values from the file of step 4 and 9 into an
spreadsheet an compare them

The resulting graph and the L* values can be found here:

http://www.ironsys.de/pub/yahoo_qtr_group/profile-graph-1.gif

(The color profile was generated with Spyder3Print)

I hope someone can help me.

Thanks,
 Gottlieb

Hi Gottlieb,
This is a good set of questions and I think leads to a good introduction on
color management.
The idea of color management -- i.e. using display and printer ICC profiles
-- is to match
L values from the original file.  This is different than making L values
linear.  The
stepwedges have K values that are linear but depending on the profile the L
values
may or may not be linear.  When a file such as the steps you're using has no
embedded profile
it inherits the gray working space defined in Color Settings.  Not knowing
your settings
makes it hard to evaluate your results.

So what you see on the screen depends on the Color Settings -- vary the
workspace and you'll see.
Printing however depends on the selections on the Print page -- No Color
Management means
the Color Settings have no effect on the result.  A Print Profile means that
the workspace
is involved in the output.  The idea being that you want the display and the
print to match.

The idea in making the ICC profile with (1) and (2) is to have a "known"
value sent to the driver,
hence the linear K values are sent to the driver.  But in using the profile
in (7) and (8) the CMS
is trying to match the L values of the input in the print by changing the K
values sent to the
driver.  (Note in (7) "without CMS" isn't really true because the working
space will be used).

So it's hard to say whether the new densities you got are "better".  The
most important criteria
is which stepwedge print matches (i.e. looks) the closest to what you see on
the screen.  I suspect that
the random stepwedge is not a very good image to evaluate this.  Possibly
the "21step" file with
the steps in order is better or maybe a real image.

It's a bit hard to accurately measure this because the idea is a
"perceptual" intent not a straight-line.
But since you using a photo paper with a very dark dMax you can get pretty
close to the straight-line
test by using the QTR-Gray Lab profile for the gray working space in Color
Settings.  This still
isn't perfect since the mapping of L=0 of the file to L=3.xx of the print
adds some non-linearity.

Roy

On Jan 6, 2008 2:57 PM, g.bossert <g.bossert@...> wrote:

> Hello,
>
> my name is Gottlieb and I'm new in this group.
>
> I'm trying to create some grayscale ICC profiles for the Epson 3800
> ABW driver using the QTR-Create-ICC.exe and Datacolor Spyder3Print
> (also knwon as PrintFix PRO with the Datacolor 1005 device).
>
> On my glossy proof paper the Epson 3800 ABW (dark setting) is almost
> linear as it should be.
>
> The generated Grayscale ICC Profile should only correct the L* curve
> only marginally. But instead it seems that the calibration does the
> contrary.
>
> But may be the caibration does all right and I make an error in
> reasoning or calibrating.
>
> The way I created the grayscale ICC profiles was the following:
>
> 1) Open the file "Step-21-gray.tif" in Photoshop CS2 without CMS
>
> 2) Print it with no CMS and the Epson 3800 ABW driver set to the
> setting "Dark"
>
> 3) Let the print dry over night
>
> 4) Read the 21 Patches with Spyder3Print and save it to an Create-ICC
> formated file.
>
> 5) Drag the file to QTR-Create-ICC.exe
>
> 6) Install the Profile
>
> 7) Open "Step-21-gray.tif" in Photoshop CS2 without CMS
>
> 8) Print it with CMS managed by Photoshop, the QTR-ICC pofile and the
> Epson 3800 ABW driver set to the setting "Dark" with the same media
> settings
>
> 9) Read the 21 Patches with Spyder3Print and save the L* vaules to an
> file.
>
> 10) Import the L* Values from the file of step 4 and 9 into an
> spreadsheet an compare them
>
> The resulting graph and the L* values can be found here:
>
> http://www.ironsys.de/pub/yahoo_qtr_group/profile-graph-1.gif
>
> (The color profile was generated with Spyder3Print)
>
> I hope someone can help me.
>
> Thanks,
>  Gottlieb
>
>
>


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

> 
> The resulting graph and the L* values can be found here:
> 
> http://www.ironsys.de/pub/yahoo_qtr_group/profile-graph-1.gif
> 
> (The color profile was generated with Spyder3Print)
> 
> I hope someone can help me.
> 
> Thanks,
>  Gottlieb

Gottlieb,

The "curve" you also included in the graph is based on ?



-- 
Met vriendelijke groeten, Ernst


|  Dinkla Grafische Techniek  |
|     www.pigment-print.com    |
|             ( unvollendet )            |

--- In QuadtoneRIP@yahoogroups.com, "g.bossert" <g.bossert@...> wrote:
>
> Hello,
> 
> my name is Gottlieb and I'm new in this group.
> 
> I'm trying to create some grayscale ICC profiles for the Epson 3800
> ABW driver using the QTR-Create-ICC.exe and Datacolor Spyder3Print
> (also knwon as PrintFix PRO with the Datacolor 1005 device).
> 
>
Hi, Gottlieb,

I am way out of my depth here so if you are wanting technical info I
cannot help you, but if you are wanting solutions, Eric Chan may be
able to assist
http://people.csail.mit.edu/ericchan/dp/Epson3800/abwprofiles.html

Regards,

Colin

-- 
Before photography, I looked - now, I see.

www.colinconwayphotography.com

I build the curve with the algorithm described by Eric Chan on
http://people.csail.mit.edu/ericchan/dp/Epson3800/bwcurves.html


--- In QuadtoneRIP@yahoogroups.com, Ernst Dinkla <E.Dinkla@...> wrote:

>
> 
> > 
> > The resulting graph and the L* values can be found here:
> > 
> > http://www.ironsys.de/pub/yahoo_qtr_group/profile-graph-1.gif
> > 
> > (The color profile was generated with Spyder3Print)
> > 
> > I hope someone can help me.
> > 
> > Thanks,
> >  Gottlieb
> 
> Gottlieb,
> 
> The "curve" you also included in the graph is based on ?
> 
> 
> 
> -- 
> Met vriendelijke groeten, Ernst
> 
> 
> |  Dinkla Grafische Techniek  |
> |     www.pigment-print.com    |
> |             ( unvollendet )            |
>

Thank you Roy. That's the key. Sitting around and waiting for my train
I thought about the problem and found out, that the non-linearity
looks like some sort of gamma applied to the cuve.
Thanks for your excellent explanation. Now I am back on the right way.
Your introduction helped me a lot to understand how it works.

My next step will be to profile my papers for QTR itself. But this
won't happen until my next vacation. My first testprints with the
preprofiled EEM look greath and even more neutral then Epson ABW.

$50 for your greath RIP and help are on the way.

Gottlieb


--- In QuadtoneRIP@yahoogroups.com, "Roy Harrington" <roy@...> wrote:
>
> Hi Gottlieb,
> This is a good set of questions and I think leads to a good
introduction on
> color management.
> The idea of color management -- i.e. using display and printer ICC
profiles
> -- is to match
> L values from the original file.  This is different than making L values
> linear.  The
> stepwedges have K values that are linear but depending on the
profile the L
> values
> may or may not be linear.  When a file such as the steps you're
using has no
> embedded profile
> it inherits the gray working space defined in Color Settings.  Not
knowing
> your settings
> makes it hard to evaluate your results.
> 
> So what you see on the screen depends on the Color Settings -- vary the
> workspace and you'll see.
> Printing however depends on the selections on the Print page -- No Color
> Management means
> the Color Settings have no effect on the result.  A Print Profile
means that
> the workspace
> is involved in the output.  The idea being that you want the display
and the
> print to match.
> 
> The idea in making the ICC profile with (1) and (2) is to have a "known"
> value sent to the driver,
> hence the linear K values are sent to the driver.  But in using the
profile
> in (7) and (8) the CMS
> is trying to match the L values of the input in the print by
changing the K
> values sent to the
> driver.  (Note in (7) "without CMS" isn't really true because the
working
> space will be used).
> 
> So it's hard to say whether the new densities you got are "better".  The
> most important criteria
> is which stepwedge print matches (i.e. looks) the closest to what
you see on
> the screen.  I suspect that
> the random stepwedge is not a very good image to evaluate this. 
Possibly
> the "21step" file with
> the steps in order is better or maybe a real image.
> 
> It's a bit hard to accurately measure this because the idea is a
> "perceptual" intent not a straight-line.
> But since you using a photo paper with a very dark dMax you can get
pretty
> close to the straight-line
> test by using the QTR-Gray Lab profile for the gray working space in
Color
> Settings.  This still
> isn't perfect since the mapping of L=0 of the file to L=3.xx of the
print

> adds some non-linearity.
> 
> Roy
> 
> On Jan 6, 2008 2:57 PM, g.bossert <g.bossert@...> wrote:
> 
> > Hello,
> >
> > my name is Gottlieb and I'm new in this group.
> >
> > I'm trying to create some grayscale ICC profiles for the Epson 3800
> > ABW driver using the QTR-Create-ICC.exe and Datacolor Spyder3Print
> > (also knwon as PrintFix PRO with the Datacolor 1005 device).
> >
> > On my glossy proof paper the Epson 3800 ABW (dark setting) is almost
> > linear as it should be.
> >
> > The generated Grayscale ICC Profile should only correct the L* curve
> > only marginally. But instead it seems that the calibration does the
> > contrary.
> >
> > But may be the caibration does all right and I make an error in
> > reasoning or calibrating.
> >
> > The way I created the grayscale ICC profiles was the following:
> >
> > 1) Open the file "Step-21-gray.tif" in Photoshop CS2 without CMS
> >
> > 2) Print it with no CMS and the Epson 3800 ABW driver set to the
> > setting "Dark"
> >
> > 3) Let the print dry over night
> >
> > 4) Read the 21 Patches with Spyder3Print and save it to an Create-ICC
> > formated file.
> >
> > 5) Drag the file to QTR-Create-ICC.exe
> >
> > 6) Install the Profile
> >
> > 7) Open "Step-21-gray.tif" in Photoshop CS2 without CMS
> >
> > 8) Print it with CMS managed by Photoshop, the QTR-ICC pofile and the
> > Epson 3800 ABW driver set to the setting "Dark" with the same media
> > settings
> >
> > 9) Read the 21 Patches with Spyder3Print and save the L* vaules to an
> > file.
> >
> > 10) Import the L* Values from the file of step 4 and 9 into an
> > spreadsheet an compare them
> >
> > The resulting graph and the L* values can be found here:
> >
> > http://www.ironsys.de/pub/yahoo_qtr_group/profile-graph-1.gif
> >
> > (The color profile was generated with Spyder3Print)
> >
> > I hope someone can help me.
> >
> > Thanks,
> >  Gottlieb
> >
> >
> >
> 
> 
> [Non-text portions of this message have been removed]
>

g.bossert wrote:
> I build the curve with the algorithm described by Eric Chan on
> http://people.csail.mit.edu/ericchan/dp/Epson3800/bwcurves.html

OK,  the conversion to the QTR printer profile does about 
the same to the tonal range of the image.
The linearity of the printer isn't changed but the tone 
range of the image is adapted in the conversion to that 
profile for what is possible in tone range on that paper. 
But I see you got that idea already.


-- 
Met vriendelijke groeten, Ernst


|  Dinkla Grafische Techniek  |
|     www.pigment-print.com    |
|             ( unvollendet )            |

> OK,  the conversion to the QTR printer profile does about 
> the same to the tonal range of the image.
> The linearity of the printer isn't changed but the tone 
> range of the image is adapted in the conversion to that 
> profile for what is possible in tone range on that paper. 
> But I see you got that idea already.

Fine, but the way it's done I'm not so sure. 
I'm having a desperate time trying to properly output MPS K7 on glossy 
media. 
I somehow managed (finally) a proper linearization using the 51 
stepwedge, but when the profile is created and used with percep. w/ BPC-
on in CS3 to convert the piezography.psd target, the 
100/99/98/97/96/95% patches get clustered together in a 
100/99/99/99/98/98% ramp given a L*14.5 Dmax. What this leads to is a 
blocked shadow area at the end of the greyscale.
Perceptual intent compresses the tonal range but here the compression 
is very agressive. 
Olivier

On Jan 8, 2008 2:04 AM, Olivier <odesmais@...> wrote:

> > OK,  the conversion to the QTR printer profile does about
> > the same to the tonal range of the image.
> > The linearity of the printer isn't changed but the tone
> > range of the image is adapted in the conversion to that
> > profile for what is possible in tone range on that paper.
> > But I see you got that idea already.
>
> Fine, but the way it's done I'm not so sure.
> I'm having a desperate time trying to properly output MPS K7 on glossy
> media.
> I somehow managed (finally) a proper linearization using the 51
> stepwedge, but when the profile is created and used with percep. w/ BPC-
> on in CS3 to convert the piezography.psd target, the
> 100/99/98/97/96/95% patches get clustered together in a
> 100/99/99/99/98/98% ramp given a L*14.5 Dmax. What this leads to is a
> blocked shadow area at the end of the greyscale.
> Perceptual intent compresses the tonal range but here the compression
> is very agressive.
> Olivier
>
>
Olivier and others,

With a weak dMax, yes the shadows are compressed somewhat -- that's
inevitable.

But your mentioning only the K values of the ramp you are using misses the
point that it's the L values that are important.  Color Management is all
about L-values.
If you happen to have a GrayGamma 2.2 stepwedge then K=100 gives L=0 but
K=95 gives L=1.
So the difference between K=100 to K=95 is already very small -- only 1 unit
of L.
It's the stepwedge itself that is very compressed in the shadow area.
(don't fall into the trap:  "but it's untagged and I said no color
management".   If you look
at L values or you Convert-to-Profile, remember the file ALWAYS inherits the
working space)

"Thinking in color management" requires a switch in concept from thinking K
(or RGB) values
to Lab values.  This really is the crux of the issue with CM discussions.
 It's a "light-bulb" moment.

Virtually all stepwedges are evenly spaced in K values but not necessarily
even in L values.  So any CM discussion of values MUST MUST MUST include the
L values
or the profile that is being used so one can convert.

Roy


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

> Virtually all stepwedges are evenly spaced in K values but not 
necessarily
> even in L values.  So any CM discussion of values MUST MUST MUST 
include the
> L values
> or the profile that is being used so one can convert.
> 

Absolutely Roy. 
But as I said, MPS K7 on Innova F-Type R1800 output is a struggle to 
say the least. 
The canned profile doesn't work for me : it over-inks and a 1mm-thick 
layer of ink floats on the paper at the 2880 recommended resolution. So 
I'm left with the need to produce my own curve set. While I'm not 
competent at partionning and understanding density, I'm fairly ok with 
QTR usage.
I went for curve creations both with a standard proceeding (small 
overlaps) and a canned-curve comparable one (i.e. long overlaps). In 
both cases I also tracked any potential over-inking. I used the 
51stepwedge to get the output linearized and I got quite a good result 
till there.
21stepwedge and Create-icc gets me the profile. 
The final print is just terrible : over-dark and no shadow separation.
I only spent a couple of days on MPS and this is where I stand so more 
testing are needed, but I was expecting better at the initial stage. 
My assumption is that despite no over-inking at the dark end of the 
greyscale and fine linearization, the partitioning is not right. Now I 
don't understand the so many inks usage in the shadows in the canned 
profile and I don't know how to do that on my own.
Olivier

Hi Everybody Im a newbie in here, 
Sorry if my english is not very good, im also have question about QTR Create ICC/ICC RGB. I 
love this feature because finally im able to Soft Proofing Epson ABW.
If im not wrong (please correct me if im wrong), the function of QTR create ICC is to Linearize 
data and then make icc Profile based on linearize data. How if i want to make ICC profile 
without Linearized the data , so i can softproofing Before and after Linearize result in 
Photoshop? if it not posibble, There is another method for doing it?

Thanks Everybody,

chiaro2112 wrote:
> Hi Everybody Im a newbie in here, 
> Sorry if my english is not very good, im also have question about QTR Create ICC/ICC RGB. I 
> love this feature because finally im able to Soft Proofing Epson ABW.
> If im not wrong (please correct me if im wrong), the function of QTR create ICC is to Linearize 
> data and then make icc Profile based on linearize data. How if i want to make ICC profile 
> without Linearized the data , so i can softproofing Before and after Linearize result in 
> Photoshop? if it not posibble, There is another method for doing it?
> 
> Thanks Everybody,


No, the QTR create ICC is to represent your printer output 
in PS and adapt your image tone range to the tone range 
possible between the Dmax and paper white of the paper you 
made the profile for. You have to make or find a linear 
output of your printer first.

Find an ABW driver setting that will deliver a linear 
output. You can test that with the 21 or 51 step greyscale 
targets. You have to check whether the Dmax doesn't run back 
on the 100% patch compared to a lower one as that sometimes 
happens. If the driver allows some ink limiting like my HP 
Z3100 you can use that to bring the highest Dmax on the 100% 
patch. Otherwise you could use the gamma function of 
lighter/darker in the driver to make the output more linear, 
a clumsy approach but there's no alternative. You have to 
keep that setting then when printing later on.

When the printer's output is linear you can make the profile 
with the same target.

For consistency in time you can not go back to a new 
linearisation of the printer output like possible with QTR 
or the internal calibration of the HP Z3100.  There's 
Colorbase for calibrating some Epson wide formats but you 
didn't mention the model.

If it is difficult to get the printer linearised you better 
use the QTRip itself. A profile based on a non linear 
printer output doesn't really help to get your shadows 
correct etc. In that case you probably have more help from a 
correction curve in PS if you want to stay with ABW.


-- 
Met vriendelijke groeten,  Ernst


|  Dinkla Grafische Techniek  |
|     www.pigment-print.com    |
|             ( unvollendet )            |

--- In QuadtoneRIP@yahoogroups.com, Ernst Dinkla <E.Dinkla@...> wrote:
>
> chiaro2112 wrote:
> > Hi Everybody Im a newbie in here, 
> > Sorry if my english is not very good, im also have question about QTR Create ICC/ICC 
RGB. I 
> > love this feature because finally im able to Soft Proofing Epson ABW.
> > If im not wrong (please correct me if im wrong), the function of QTR create ICC is to 
Linearize 
> > data and then make icc Profile based on linearize data. How if i want to make ICC 
profile 
> > without Linearized the data , so i can softproofing Before and after Linearize result in 
> > Photoshop? if it not posibble, There is another method for doing it?
> > 
> > Thanks Everybody,
> 
> 
> No, the QTR create ICC is to represent your printer output 
> in PS and adapt your image tone range to the tone range 
> possible between the Dmax and paper white of the paper you 
> made the profile for. You have to make or find a linear 
> output of your printer first.
> 
> Find an ABW driver setting that will deliver a linear 
> output. You can test that with the 21 or 51 step greyscale 
> targets. You have to check whether the Dmax doesn't run back 
> on the 100% patch compared to a lower one as that sometimes 
> happens. If the driver allows some ink limiting like my HP 
> Z3100 you can use that to bring the highest Dmax on the 100% 
> patch. Otherwise you could use the gamma function of 
> lighter/darker in the driver to make the output more linear, 
> a clumsy approach but there's no alternative. You have to 
> keep that setting then when printing later on.
> 
> When the printer's output is linear you can make the profile 
> with the same target.
> 
> For consistency in time you can not go back to a new 
> linearisation of the printer output like possible with QTR 
> or the internal calibration of the HP Z3100.  There's 
> Colorbase for calibrating some Epson wide formats but you 
> didn't mention the model.
> 
> If it is difficult to get the printer linearised you better 
> use the QTRip itself. A profile based on a non linear 
> printer output doesn't really help to get your shadows 
> correct etc. In that case you probably have more help from a 
> correction curve in PS if you want to stay with ABW.
> 
> 
> -- 
> Met vriendelijke groeten,  Ernst
> 
> 
> |  Dinkla Grafische Techniek  |
> |     www.pigment-print.com    |
> |             ( unvollendet )            |
>

Hi Ernst
Thanks For your detail Explanation, its very helpful info for me to find good starting point. 
I just experiment with epson ABW setting in my Epson Premium Glossy Paper and Ilford 
Smooth Pearl Paper. It seems difficult to find Linear result with ABW setting (from Light to 
darkest setting) in Epson Premium Glossy Paper. It always introduce Banding in Highlight 
area but the Dmax doesn't run back on the 100% patch . I believe Im already choosed the 
proper media setting for that. From 21 & 51 gray Target that i Printed out I see Obvious 
Color Shift in 25%( from 21 gray Target) & 24-28% (from 51 gray target).Before 25% and 
Toward 100% the patch are acceptable neutral. There is any Possibility to correct this 
(Color Shift) in Epson ABW setting?
Its very strange because it more easier to find Linear result with my Ilford Smooth pearl 
paper with Epson Semi Gloss setting and ABW dark setting. I think you're right about 
epson COLOR Base and maybe my printer behave differently than it used to be. Im already 
try download it but im just successfully got the Win XP version, The website (epson UK) 
didn't respond when i tried to choose Mac Os Version.
When you mention about Correction Curve in PS, are you mean Curve Adjustment layer or 
custom curve in dot gain setting?
Oh ya, i use Epson 4800 and it has color density setting is it same with ink limit feature in 
your hp driver?

Thanks Ernst and everybody
Davi Arzika

chiaro2112 wrote:

> Thanks For your detail Explanation, its very helpful info for me to find good starting point. 
> I just experiment with epson ABW setting in my Epson Premium Glossy Paper and Ilford 
> Smooth Pearl Paper. It seems difficult to find Linear result with ABW setting (from Light to 
> darkest setting) in Epson Premium Glossy Paper. It always introduce Banding in Highlight 
> area but the Dmax doesn't run back on the 100% patch . I believe Im already choosed the 
> proper media setting for that. From 21 & 51 gray Target that i Printed out I see Obvious 
> Color Shift in 25%( from 21 gray Target) & 24-28% (from 51 gray target).Before 25% and 
> Toward 100% the patch are acceptable neutral. There is any Possibility to correct this 
> (Color Shift) in Epson ABW setting?

Sorry, I am not that familiar with the Epson ABW driver. 
There must be a manual for the right settings (CM included) 
and what media preset to use for the paper. The Epson ABW 
still uses color in its ABW mode so shifts are possible I guess.

> Its very strange because it more easier to find Linear result with my Ilford Smooth pearl 
> paper with Epson Semi Gloss setting and ABW dark setting. I think you're right about 
> epson COLOR Base and maybe my printer behave differently than it used to be. Im already 
> try download it but im just successfully got the Win XP version, The website (epson UK) 
> didn't respond when i tried to choose Mac Os Version.


> When you mention about Correction Curve in PS, are you mean Curve Adjustment layer or 
> custom curve in dot gain setting?

Any choice I think. Just making a curve and save it for a 
conversion. Or an action with a curve. You apply it before 
printing but you don't save it with the file as there will 
be another printer or method in the future.

> Oh ya, i use Epson 4800 and it has color density setting is it same with ink limit feature in 
> your hp driver?

Don't think so, the color density setting will be more of a 
gamma shift that influences the mid tones heavier. The 
inklimit in the HP Z3100 is not in the driver but in a 
separate utility where you can make custom media presets 
based on existing ones to use in the driver. Drying time, 
paper thickness, ink limitation 100%-80% (it actually 
reduces the ink amount lineary),  and when made it will 
calibrate that media preset/paper immediately and if desired 
make a custom (color) profile after that. In the driver 
itself there are density settings like in Epson's ABW.

-- 
Met vriendelijke groeten, Ernst


|  Dinkla Grafische Techniek  |
|     www.pigment-print.com    |
|             ( unvollendet )            |