QTR-Quadtone RIP

Hi Roy,

After playing with the curve creation tools on a Mac with an Eye-One
Pro, I have
the following observations/questions.  I wish I can get some kind
insightful
answers/comments from you?

All the efforts has been for the Hahnemuhle Photo Rag 308 paper on a new
Epson
3800 with the Ultrachrome K3 inks (Matte Black, of course).

1. Using the Epson's Advanced Black and White (ABW) mode (1440dpi,
"Normal",
    warm toned a little bit), the linearity was not good (compressed
highlights
    and shadows, expecially the shadows).  However, the D-Max was 1.624:

File: /Users/leping/Desktop/3800
Linearization/ABW-RL_1440_Normal-out.txt
Step   Dens  Lab   A    B
   0.00 0.040 96.47 0.40 0.35 -                             |b
L  +
   5.00 0.077 93.34 0.58 0.53 -                             | b
L   +
  10.00 0.132 88.79 0.68 0.81 -                             | ab
L      +
  15.00 0.198 83.61 0.91 1.03 -                             |  ab
L         +
  20.00 0.260 79.05 0.96 1.37 -                             |  a b
L            +
  25.00 0.320 74.76 0.96 1.54 -                             |  a  b      
L               +
  30.00 0.391 69.91 1.17 1.19 -                             |   b      L
+
  35.00 0.464 65.26 0.78 1.87 -                             |  a   b L
+
  40.00 0.542 60.50 0.85 2.14 -                             |  a  L b
+
  45.00 0.618 56.19 0.94 2.32 -                             |  a     b
+
  50.00 0.704 51.57 0.80 2.49 -                             L  a     b
+
  55.00 0.800 46.76 0.81 2.06 -                           L |  a    b
+
  60.00 0.895 42.38 0.74 1.86 -                        L    | a    b
+
  65.00 1.024 36.87 1.03 1.72 -                     L       |   a b
+
  70.00 1.133 32.61 0.93 1.51 -                  L          |  a  b
+
  75.00 1.257 28.20 0.77 1.39 -               L             |  a b
+
  80.00 1.370 24.54 0.47 0.90 -             L               |a b
+
  85.00 1.475 21.40 0.12 1.10 -           L                 a   b
+
  90.00 1.550 19.30 0.17 1.15 -          L                  a   b
+
  95.00 1.602 17.93 0.43 1.25 -         L                   |a   b
+
100.00 1.624 17.34 0.59 1.61 -         L                   | a   b
+

2. With a blend (created by Richard Lohman to match his excellent
    Hextone prints off a modified Epson 3000) of your released 3800
    curves for the paper (to achieve a similar warmth, a split of
    the cool, neutral, and the sepis curves), the linearity was good,
    but the D-Max was only 1.552:

QTR-Linearize-Data version 2.5.1.0
File: /Users/leping/Desktop/3800 Linearization/QTR-1440-RL_Split-out.txt
Step   Dens  Lab   A    B
   0.00 0.037 96.69 0.41 0.35 -                             |b
L +
   5.00 0.086 92.62 0.61 0.34 -                             |ba
L    +
  10.00 0.134 88.72 0.79 0.36 -                             |b a
L      +
  15.00 0.184 84.73 0.95 0.41 -                             |b a
L         +
  20.00 0.237 80.69 1.09 0.42 -                             |b  a
L           +
  25.00 0.290 76.85 1.26 0.47 -                             |b   a
L             +
  30.00 0.356 72.28 1.49 0.60 -                             | b  a      
L                +
  35.00 0.414 68.41 1.49 0.67 -                             | b  a     L
+
  40.00 0.478 64.35 1.65 0.83 -                             |  b  a L
+
  45.00 0.542 60.53 1.71 1.01 -                             |   b a
+
  50.00 0.619 56.11 1.73 1.11 -                             |  Lb a
+
  55.00 0.697 51.91 1.98 1.28 -                             |L   b a
+
  60.00 0.771 48.18 1.80 1.58 -                           L |     ba
+
  65.00 0.863 43.84 1.51 1.72 -                         L   |     b
+
  70.00 0.943 40.27 1.67 1.64 -                       L     |     b
+
  75.00 1.036 36.36 1.91 1.92 -                    L        |      b
+
  80.00 1.135 32.53 2.11 2.15 -                  L          |       b
+
  85.00 1.267 27.86 2.22 2.28 -               L             |       ab
+
  90.00 1.360 24.86 2.10 2.33 -             L               |       ab
+
  95.00 1.451 22.09 1.88 2.06 -            L                |      ab
+
100.00 1.552 19.24 1.50 1.86 -          L                  |     ab
+

3. Next, I did the ink limit and separation tests, using both
    your "inkseparation.psd" pattern as we as Daniel (dchow1961)'s
    modified Ink_Limit_Pattern_Page from the list's Files/Tools
    section.  The measurements were done both visually and with
    the Eye-One Pro.

    What I found is, for all the 3800 inks, more ink always produces
    higher densities, although the shoulder location can be clearly
    identified.  The following is an example of the 3800 K (Matte
    Black):

QTR-Linearize-Data version 2.5.1.0
File: /Users/leping/Desktop/3800 Linearization/K-InkLimit-out.txt
Step   Dens  Lab   A    B
   0.00 0.043 96.21 0.51 0.66 -                             | b
L  +
   5.00 0.413 68.48 1.22 2.91 -                             |   a      b
+
  10.00 0.700 51.79 1.60 4.08 -                             |L    a
b             +
  15.00 0.854 44.23 1.61 4.11 -                         L   |     a
b             +
  20.00 1.027 36.75 1.52 3.82 -                     L       |     a
b              +
  25.00 1.205 30.02 1.33 3.23 -                 L           |    a      b
+
  30.00 1.273 27.67 1.16 2.59 -               L             |   a     b
+
  35.00 1.372 24.48 1.06 2.39 -             L               |   a    b
+
  40.00 1.445 22.26 0.94 2.17 -            L                |  a    b
+
  45.00 1.514 20.28 0.89 2.17 -           L                 |  a    b
+
  50.00 1.556 19.13 0.82 2.06 -          L                  |  a    b
+
  55.00 1.579 18.53 0.77 1.81 -          L                  |  a   b
+
  60.00 1.599 18.01 0.76 1.83 -         L                   |  a   b
+
  65.00 1.609 17.73 0.65 1.75 -         L                   | a    b
+
  70.00 1.610 17.70 0.69 1.85 -         L                   | a    b
+
  75.00 1.623 17.39 0.65 1.77 -         L                   | a    b
+
  80.00 1.624 17.36 0.67 1.88 -         L                   | a    b
+
  85.00 1.624 17.34 0.61 1.87 -         L                   | a    b
+
  90.00 1.625 17.33 0.64 1.80 -         L                   | a    b
+
  95.00 1.630 17.20 0.63 1.95 -         L                   | a    b
+
100.00 1.630 17.19 0.64 1.79 -         L                   | a    b
+

    so that, as the ABW driver results showed, the Ultrachrome K3
    Matte Black ink is able to produce a D-Max of 1.63.

    And I found, visually the ink limits are:

    K: 60 / C: 85 / M: 85 / Y: 60 / LC: 90 / LM: 90 / LB: 80 / LLB: 85

    If this is true, then my first question is, is the ink limit
    settings (45/50) in your released 3800 Photo Rag curves too
    low, contributing to the reduced D-Max using these curves?

    I went to check your released Photo Rag curve ink limit
    numbers for the 4800/7800/9800 UC inks, and can see the
    limits were all 100.  If the 3800 shares the same K3 ink
    with its bigger brothers, why the settings should be so
    different?

    And, since the lowest L values achieved with the curve
    settings can be seen from the linearization array (the last
    number on the line), after converting to density, we can
    see with the 100% ink limits you were able to attain
    D-Max of 1.70 (Coolse) / 1.69 (Cool and Sepia) / 1.65 (Warm)
    with your 4800 curves!

    Nevertheless, I noticed the Coolse and Cool curve description
    file headers are different from the other two, with the
    comment line that read "for 4800 with ultrachrome k3 inks
    and MIS Eboni in the K slot".  This is quite confusing, but
    since I tend to think the MIS Eboni Black ink would produce
    D-Max values deeper than 1.70 (Richard got 1.76), and you
    released the curves as a family in the folder named for
    UC ink the comment is just a leftover from other calibrations?

    Then, without changing the ink limits, I tried to increase the
    BOOST_K value from 55/60 to 85.  The results showed immediately
    an increased D-Max of 1.598.

    This encouraged me to create a new set of 3800 curves from
    copying yours and changing the default ink limits to 55 and
    the BOOST_K to 100.  I could have tried something like a
    15% GRAY_OVERLAP as well but decided to do one thing a time.
    I always let prints to dry for at least 10 hours before any
    measurement.  I had no problems with linearization at all,
    and the post-linearization results, from the altered 55/100
    settings, and still with the Richard's curve spliting formula,
    the result was:

File: /Users/leping/Desktop/3800
Linearization/Boost100-Linearized-RL-Split-out.txt
Step   Dens  Lab   A    B
   0.00 0.051 95.57 0.40 0.37 -                             |b
L  +
   5.00 0.095 91.81 0.63 0.35 -                             |ba
L    +
  10.00 0.145 87.83 0.82 0.38 -                             |b a
L       +
  15.00 0.192 84.08 1.03 0.40 -                             |b  a
L         +
  20.00 0.248 79.86 1.19 0.43 -                             |b  a
L            +
  25.00 0.297 76.39 1.39 0.48 -                             |b   a
L              +
  30.00 0.348 72.81 1.65 0.59 -                             | b   a     
L                +
  35.00 0.414 68.39 1.67 0.78 -                             |  b  a    L
+
  40.00 0.472 64.72 1.83 1.00 -                             |  b   aL
+
  45.00 0.534 61.02 1.96 1.19 -                             |   b La
+
  50.00 0.605 56.88 2.04 1.33 -                             |   Lb  a
+
  55.00 0.674 53.17 2.35 1.55 -                             |L    b  a
+
  60.00 0.751 49.17 2.21 1.83 -                            L|      ba
+
  65.00 0.842 44.80 1.96 2.17 -                         L   |      ab
+
  70.00 0.916 41.43 2.10 2.18 -                       L     |       b
+
  75.00 1.017 37.13 2.17 2.45 -                     L       |       ab
+
  80.00 1.114 33.34 2.27 2.67 -                   L         |        ab
+
  85.00 1.221 29.45 2.25 2.76 -                L            |        a b
+
  90.00 1.319 26.16 2.14 2.68 -              L              |       a b
+
  95.00 1.434 22.58 1.73 2.30 -            L                |     a  b
+
100.00 1.587 18.32 0.93 2.22 -         L                   |  a    b
+

    so the D-Max increased to 1.587 from the 1.552 with the
    released curves, but still lower than the possible value of
    1.63 from the ink limit calibration.

    Thus, my next question is, should I try to make the default
    ink limit to 100%, and put in a little GRAY_OVERLAP next, in
    the hope to further increase the D-Max on the Photo Rag?

4. The last questing is the most puzzling.  From my ink limit
    calibration, the (minimum) L values at 100% ink ware 17.19 (K),
    30.17 (LK), and 17.19 (LLK).  After interpolation (directly, or
    using kevbell007's GRAY_VAL_Interpolation_for_3_inks spreadsheet,
    also in the group's Files/Tool folder), the relative density of
    the 3800 UC inks should be [100/24.9/7.0].

    However, in all your released 3800/2400 curves, such ratios has
    been [100/33/12], which is way off my measurements.  Then I went
    to your 4800 curves and found the ratios there were 100/24/9,
    that is much closer to my numbers.

    What happens if I use the [100/25/7] results to create new curves?
    I have to say I did try to fool the QTR software to install the
    4800 curved for the QTR3800 printer (by changing the folder name),
    but the results were very bad with heavy color casts so the
    direct substitution is not a solution.

Thank you so much for your patients (and all who is following me
this far) reading this long post.

Yours very truly,
Leping
San Mateo, California
www.lepingzha.com <http://www.lepingzha.com>




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

Hi Leping.... I tried some split-tones with the curves supplied for
silver rag.  Like you I noticed a real decrease in dmax from the ABW
prints I'd been making (from say 2.4ish to about 2).  Like you I
opened the curves and noticed the Boost K value seemed low.  So I
increased it to 100 and printed some more tests wedges.  Then I
realized that you can't get a truly cool curve if you have the K value
that high, because the K ink is intrinsically warm.

Ed

--- In QuadtoneRIP@yahoogroups.com, "Leping Zha" <leping@...> wrote:
>
> 
> Hi Roy,
> 
> 
snip

> 
>     What happens if I use the [100/25/7] results to create new curves?
>     I have to say I did try to fool the QTR software to install the
>     4800 curved for the QTR3800 printer (by changing the folder name),
>     but the results were very bad with heavy color casts so the
>     direct substitution is not a solution.
> 
> Thank you so much for your patients (and all who is following me
> this far) reading this long post.
> 
> Yours very truly,
> Leping
> San Mateo, California
> www.lepingzha.com <http://www.lepingzha.com>
> 
> 
> 
> 
> [Non-text portions of this message have been removed]
>

Ed,
I have done the same thing.  My impression is that boost K only 
impact the very deep shadow.  When checking the curves, I only see a 
change in the area above about 85% and have not noted the cool curves 
becoming much warmer.  It is interesting that we are all having 
problems getting good dmax from the premade curves.  My results in 
terms of d max are very similiar to yours.  I am using Innova ultra 
smooth glossy.
Leping,  Are you interested in sharing Richard Lohman's curves or 
setting?  I am familiar with his work, and really like what he is 
doing.  
Thanks,
Peter


--- In QuadtoneRIP@yahoogroups.com, "edrudolpho" <erudolph@...> wrote:
>
> Hi Leping.... I tried some split-tones with the curves supplied for
> silver rag.  Like you I noticed a real decrease in dmax from the ABW
> prints I'd been making (from say 2.4ish to about 2).  Like you I
> opened the curves and noticed the Boost K value seemed low.  So I
> increased it to 100 and printed some more tests wedges.  Then I
> realized that you can't get a truly cool curve if you have the K 
value
> that high, because the K ink is intrinsically warm.
> 
> Ed
> 
> --- In QuadtoneRIP@yahoogroups.com, "Leping Zha" <leping@> wrote:
> >
> > 
> > Hi Roy,
> > 
> > 
> snip
> 
> > 
> >     What happens if I use the [100/25/7] results to create new 
curves?
> >     I have to say I did try to fool the QTR software to install 
the
> >     4800 curved for the QTR3800 printer (by changing the folder 
name),

> >     but the results were very bad with heavy color casts so the
> >     direct substitution is not a solution.
> > 
> > Thank you so much for your patients (and all who is following me
> > this far) reading this long post.
> > 
> > Yours very truly,
> > Leping
> > San Mateo, California
> > www.lepingzha.com <http://www.lepingzha.com>
> > 
> > 
> > 
> > 
> > [Non-text portions of this message have been removed]
> >
>

Hello,
I wondered if it would be possible for a new feature to be added to QTR - the ability to save more than one custom paper size. For example, I would like to have a couple of sizes - half 13'x19' and half letter - but can only have one custom size.

thanks,
Richard
--
http://smallfield.vze.com
http://photos.smallfield.vze.com (Photos web site)
http://warkworth.vze.com/ (Warkworth photo essay)
http://picasaweb.google.com/rsmallfield/ (Recent work) 

   "Stay away from that jazz man, Lisa. Nothing personal, I just fear the
   unfamiliar."
   --Marge Simpson

You can have many custom paper sizes in OS X with QTR.  Is this a  
limitation with the PC?

On Apr 28, 2007, at 4:40 PM, Richard Smallfield wrote:

> Hello,
> I wondered if it would be possible for a new feature to be added to  
> QTR - the ability to save more than one custom paper size. For  
> example, I would like to have a couple of sizes - half 13'x19' and  
> half letter - but can only have one custom size.
>
> thanks,
> Richard
> --
> http://smallfield.vze.com
> http://photos.smallfield.vze.com (Photos web site)
> http://warkworth.vze.com/ (Warkworth photo essay)
> http://picasaweb.google.com/rsmallfield/ (Recent work)
>
>    "Stay away from that jazz man, Lisa. Nothing personal, I just  
> fear the
>    unfamiliar."
>    --Marge Simpson

1. I tried to repost to take care of the ill-reformatted charts,
   but obviously did not succeed.  Sorry -- multiple identical
   posts also went into the daily digest.  Anyone really
   interested in these charts please email me for the original
   text file.

2. After several days adjusting, waiting for ink to dry, and
   remeasuring, I realized Roy's curve settings are basically
   what we can get.  By raising the K-ink limits to 55-65, and
   the K-BOOST to 100 I did raise the D-Max a little bit, but
   not too much.  And more than 75 did not work.  Overlapping
   the inks did not help either (within the Eye-One's error
   bars).

3. Also I found the K/LK/LLK ink ratio Roy used make more sense
   than the lower ratios from my measurements.  Using my ratios
   resulted in less linear gray scales (with no linearization).
   Corrections always work the best when there is not too much
   to correct.

Thanks for all who responded publicly and privately,
Leping

The real question to me now is, what's the advantage to use
Roy's QTR-GRAY/RGB-MATTE/PHOTO profiles?

The QTR profiles came with a Gamma close to 2.5.  So when
Photoshop convert the images, the dark shadows are compressed,
potentially a source of banding for 8-bit files.

We can make QTR curve calibrations with or without using these
profiles (i.e., let Photoshop handles color with the profile
specified as printer color space, or "No Color Management").
We just need to make different curves for each of the methods. 
If we make custom profiles as Roy suggested, since we can only
specify one as the Printer Profile, it is more convenient to make
such custom profiles with "No Color Management" (and the blend
of the curves we want to use without the linearization line),
so that we do not need double convertions (to QTR-GRAY first and
the custom profile later).

I understand we need to standarize the QTR input (so that images
tagged with Gamma 1.8 and 2.2, or what ever, will all print
correctly, since they are always converted to the standard), but,
as a virtual printer, can the QTR's characters be efficiently
modeled in the linearization in the curves, so using the
QTR-GRAY curves is not always necessary, when we make and use
our own custom profiles, and always convert the sourse image
profiles to such and feed directly to the RIP?

Some printing process (such as Richard Lohmann's Hextone) has
a standard input requirement of Gamma 1.8.  My question is
that, with all the QTR's default curves currently designed to
work with the approximately Gamma 2.5 input (in the standard
QTR-GRAY profiles that all the images are supposed to be converted
to before reaching the QTR engine), is the workflow essential
or we can by-pass these QTR-GRAY curves and to use our own
profiles in a single step?

I do understand part of Roy's efforts was to address the
visual difference of matte and gloss prints with these
separate profiles.

Thanks again,
Leping

Sorry again -- the higher Gamma profiles compress highlights not
shadows.  The QTR-GRAY-LAB/MATTE/PHOTO profiles deveotes more
bits to the shadows (than the Gray Gamma 2.2).

--- In QuadtoneRIP@yahoogroups.com, "Leping Zha" <leping@...> wrote:

>
> The real question to me now is, what's the advantage to use
> Roy's QTR-GRAY/RGB-MATTE/PHOTO profiles?
> 
> The QTR profiles came with a Gamma close to 2.5.  So when
> Photoshop convert the images, the dark shadows are compressed,
> potentially a source of banding for 8-bit files.

There have been a couple of comments about the 3800 profiles on matte 
paper.
The shadow regions always present compromises for dMax, color accuracy,
and shadow separation and linearity.   All these things tend to 
conflict with each other.   It
appears that tradeoffs I made may be a little different than the Epson 
ABW mode.

Roy

On Wednesday, May 2, 2007, at 05:22  PM, Leping Zha wrote:

> 1. I tried to repost to take care of the ill-reformatted charts,
>    but obviously did not succeed.  Sorry -- multiple identical
>    posts also went into the daily digest.  Anyone really
>    interested in these charts please email me for the original
>    text file.
>
> 2. After several days adjusting, waiting for ink to dry, and
>    remeasuring, I realized Roy's curve settings are basically
>    what we can get.  By raising the K-ink limits to 55-65, and
>    the K-BOOST to 100 I did raise the D-Max a little bit, but
>    not too much.  And more than 75 did not work.  Overlapping
>    the inks did not help either (within the Eye-One's error
>    bars).
>
> 3. Also I found the K/LK/LLK ink ratio Roy used make more sense
>    than the lower ratios from my measurements.  Using my ratios
>    resulted in less linear gray scales (with no linearization).
>    Corrections always work the best when there is not too much
>    to correct.
>
> Thanks for all who responded publicly and privately,
> Leping
>
>
-
Roy Harrington
roy@...
Black & White Photo Gallery
http://www.harrington.com