Digital BW, The Print

I have been trying hard to understand the QTR ICC tool. I thought I 
got it but... no!!!

When the stepwedge is printed linearised and profile converted I got 
values far lighter than when non profil-converted :		
		
DEN	Linear	Linear+converted
0	96,45	96,14
5	92,9	93,59
10	89,24	90,35
15	85,27	86,92
20	81,34	83,99
25	77,4	80,57
30	73,61	77,03
35	70,04	73,91
40	66,63	70,39
45	62,3	66,99
50	58,03	63,17
55	53,54	58,76
60	50,04	55,38
65	45,97	51,15
70	41,94	46,6
75	37,92	42,89
80	34,08	38,01
85	30,44	31,62
90	25,85	26,12
95	21,73	21,11
100	17,7	17,89

Basically it looks like it applies a higher gamma...

I just took the QTR stepwedge, opened it with no color managed 
conversion (CS then assumes it will have the default grey profile 
e.g. 15% dot in my color settings), converted to QTR ICC and print it 
in QTR. 

Anyway, when visualizing the kTRC tag of the QTR profile, the curve 
does look like it maps destination values lighter than the source 
ones in a nice below the 45° line curve.

Last to make it easier : kTRC seems (from ICC spec.) to be build from 
XYZ. I don't have PM, so I just past Lab values in a PM measurement 
file I got. PM XYZ look pretty strange with double digit figures 
while XYZ range from 0 to 1.99... ?

So 2 questions :

1. Do I need to also input XYZ in PM file to feed QTR create ICC and 
if so how do I convert "real" XYZ to the PM XYZ format ?
2. Is that normal QTR ICC lightens a file ? 

Any help would be more than welcome...

Olivier

Olivier

Roy should step in here but (1) the ICC spec discusses how the kTRC data
should be scaled for media white point-relative colorimetry and (2) in the
documents section of www.color.org you will also find a discussion paper on
Adobe black point compensation.  Roy has done both the black and white point
scaling.


> From: odesmais <odesmais@...>
> Reply-To: <DigitalBlackandWhiteThePrint@yahoogroups.com>
> Date: Sun, 07 Aug 2005 12:13:25 -0000
> To: <DigitalBlackandWhiteThePrint@yahoogroups.com>
> Subject: [Digital BW] QTR ICC still in the dark...
> 
> I have been trying hard to understand the QTR ICC tool. I thought I
> got it but... no!!!
> 
> When the stepwedge is printed linearised and profile converted I got
> values far lighter than when non profil-converted :  
> 
> DEN Linear Linear+converted
> 0 96,45 96,14
> 5 92,9 93,59
> 10 89,24 90,35
> 15 85,27 86,92
> 20 81,34 83,99
> 25 77,4 80,57
> 30 73,61 77,03
> 35 70,04 73,91
> 40 66,63 70,39
> 45 62,3 66,99
> 50 58,03 63,17
> 55 53,54 58,76
> 60 50,04 55,38
> 65 45,97 51,15
> 70 41,94 46,6
> 75 37,92 42,89
> 80 34,08 38,01
> 85 30,44 31,62
> 90 25,85 26,12
> 95 21,73 21,11
> 100 17,7 17,89
> 
> Basically it looks like it applies a higher gamma...
> 
> I just took the QTR stepwedge, opened it with no color managed
> conversion (CS then assumes it will have the default grey profile
> e.g. 15% dot in my color settings), converted to QTR ICC and print it
> in QTR. 
> 
> Anyway, when visualizing the kTRC tag of the QTR profile, the curve
> does look like it maps destination values lighter than the source
> ones in a nice below the 45° line curve.

The kTRC should look convex/bowed upwards.  So for a given input (vertical
axis) a lesser (than 1:1) pixel value is read from the curve and hence, for
the most part, the image is printed darker than with the simple linear
output.
> 
> Last to make it easier : kTRC seems (from ICC spec.) to be build from
> XYZ. 

XYZ_Y or L* with scaling done in XYZ_Y

>I don't have PM, so I just past Lab values in a PM measurement
> file I got. PM XYZ look pretty strange with double digit figures
> while XYZ range from 0 to 1.99... ?
> 
> So 2 questions :
> 
> 1. Do I need to also input XYZ in PM file to feed QTR create ICC and
> if so how do I convert "real" XYZ to the PM XYZ format ?

I am sorry but I don't follow this at all.  You should be feeding QTR Create
ICC an Eye One Measure Tool file simply by dragging and dropping it.  You
don't need to change anything.


> 2. Is that normal QTR ICC lightens a file ?

His does not seem correct.  Workflow:

1. Build a QTR ink descriptor file to generate a greyscale (curve).
2. Linearize the greyscale by measuring a printed step wedge and inserting
the LINEARIZE function into the ink descriptor file.
3.  Print another step wedge using this newly linearized ink curve and use
MeasureTool to measure it.
4.  Drag the MeasureTool output file to QTR Create ICC.  The output ICC
profile is named the same as the MeasureTool file but its use-name is
QTR-[MeasureTool output file name]

I would expect the opposite of the results above.

Steve,

First I need to apologize : I have double posted the below in here and 
QTR group which is not very ethical... In fact I just realised as you 
below mentionned that Roy would be of help in this.

I did look at the kTRC explanation and also the color.org explanation 
of BPC, apparently this is an issue the organisation is wanting to 
tackle, I don't know if this is being implemented in v4 or not, but by 
experince not all soft (profilers, don't know about editing 
softs) "understand" v4. There's also a lack of standardised perceptual 
intents (seems all profilers implement a proprietary algorithm) which 
is also something they are looking at, and to make it even more 
confusing there's now a debate about CIE Lab as a PCS (it does not see 
the same for XYZ)due to hue shifting thus non-uniformity... Also this 
is simply and unfortunately coming from booking reading and web 
browsing : I try to know but seldomly understand.   

Anyway, in fact what happens is that since my PS color settings is 
(default, I never bothered to set it properly)grey : dot 15% I work 
with too "light" a working space and since the stepwedge is not tagged 
PS assumes the dot 15% as a source space thus printing light after 
converting.

I'd like to measure the actual output of the QTR-ICC profiling but I 
feel it's a lost search. When printing the raw stepwedge in QTR, datas 
are fed irrespective of color space, so the raw L* value are dependent 
on printer-ink-paper and does not reflect...anything else than the 
linearised stepwedge printed on ONE printer with ONE ink set and ONE 
paper.  May be there's a way to translate the linear L* measurements 
into a gamma and set it as a grey working space to perform the 
conversion with the QTR ICC profile to evaluate, but I don't know it. I 
asked Roy if 1.8 would make sense. If not I leave it here since anyway 
it will not improve the workflow. I'll set color settings grey : gamma 
2.2 (RVB and the windows environment) to minimize conversions and that 
will do.

BPC remains a mystery : I use to do the soft-proofing with 
perceptual+BPC+white paper (which also turns on Blakc ink). However it 
might be that you need/can soft proof not turning BPC but Black ink 
simulation (which is done when white paper is checked)... to be 
followed since it's only my understanding.

I'm sure I'm tiring many people with some lack of competence/knowledge 
and not-so-important issues... So, thanks Steve for your great 
willingness to help, I do appreciate a lot.

Olivier


--- In DigitalBlackandWhiteThePrint@yahoogroups.com, Steve Kale 
<stevekale@b...> wrote:
> Olivier
> 
> Roy should step in here but (1) the ICC spec discusses how the kTRC 
data
> should be scaled for media white point-relative colorimetry and (2) 
in the
> documents section of www.color.org you will also find a discussion 
paper on
> Adobe black point compensation.  Roy has done both the black and 
white point

> scaling.
> 
>

Olivier

No problem.  I have not joined the QTR group yet (in general I am against
fragmentation of this list - it is powerful due to its breadth).  I am not
so far in front of you on this stuff so sorry if I am not definitive at
times.


> From: odesmais <odesmais@...>
> Reply-To: <DigitalBlackandWhiteThePrint@yahoogroups.com>
> Date: Mon, 08 Aug 2005 11:51:39 -0000
> To: <DigitalBlackandWhiteThePrint@yahoogroups.com>
> Subject: Re: [Digital BW] QTR ICC still in the dark...
> 
> Steve,
> 
> First I need to apologize : I have double posted the below in here and
> QTR group which is not very ethical... In fact I just realised as you
> below mentionned that Roy would be of help in this.
> 
> I did look at the kTRC explanation and also the color.org explanation
> of BPC, apparently this is an issue the organisation is wanting to
> tackle, I don't know if this is being implemented in v4 or not, but by
> experince not all soft (profilers, don't know about editing
> softs) "understand" v4. There's also a lack of standardised perceptual
> intents (seems all profilers implement a proprietary algorithm) which
> is also something they are looking at, and to make it even more
> confusing there's now a debate about CIE Lab as a PCS (it does not see
> the same for XYZ)due to hue shifting thus non-uniformity... Also this
> is simply and unfortunately coming from booking reading and web
> browsing : I try to know but seldomly understand.
> 
> Anyway, in fact what happens is that since my PS color settings is
> (default, I never bothered to set it properly)grey : dot 15% I work
> with too "light" a working space and since the stepwedge is not tagged
> PS assumes the dot 15% as a source space thus printing light after
> converting.

I use Gray Gamma 2.2 but this should not make a difference as the raw pixel
values are sent.  What you are measuring with the ICC profile is simply the
response behaviour associated with a particular pixel value.  When you do
the conversion to the ICC profile created by QTR Create ICC then the pixel
values are altered so that the shades of grey produced are the same in the
new space.  (I am sure you know this!!)

> 
> I'd like to measure the actual output of the QTR-ICC profiling but I
> feel it's a lost search. When printing the raw stepwedge in QTR, datas
> are fed irrespective of color space, so the raw L* value are dependent
> on printer-ink-paper and does not reflect...anything else than the
> linearised stepwedge printed on ONE printer with ONE ink set and ONE
> paper.  May be there's a way to translate the linear L* measurements
> into a gamma and set it as a grey working space to perform the
> conversion with the QTR ICC profile to evaluate, but I don't know it. I
> asked Roy if 1.8 would make sense. If not I leave it here since anyway
> it will not improve the workflow. I'll set color settings grey : gamma
> 2.2 (RVB and the windows environment) to minimize conversions and that
> will do.

In a Same as Source workflow your workspace makes a difference.  A
particular (visual) shade of grey has a particular (and different) pixel
value associated with it in each space.  So the same grey sends a different
number to the printer from different spaces.  This is why Black Only users,
for example, have fiddled with or tried various workspaces to improve their
print to screen match.  (QTR Grey Lab was an attempt to more closely align
working and print spaces for QTR users when it was a Same as Source
workflow.)  I use GG 2.2 simply because it is closer to the gamma embedded
in ProPhoto RGB or Adobe RGB and hence perhaps eases the space mapping when
doing conversions in and out of colour spaces (eg at initial B&W conversion
or when sharpening with Photokit Sharpener).

But one of the advantages of moving to a "colour" managed workflow (ok QTR
ICC profiles aren't colour profiles but you get my point) is that your
workspace no longer matters.  Because you have profiled the luminance that
the printer produces when receiving pixel value x you can simply do a
conversion to this space before printing so that the right file values are
sent.


> 
> BPC remains a mystery : I use to do the soft-proofing with
> perceptual+BPC+white paper (which also turns on Blakc ink).


I don't think it is toooo much of a mystery.  Scale values in XYZ_Y to
adjust for the weaker black ie make the range black point relative.

>However it 
> might be that you need/can soft proof not turning BPC but Black ink
> simulation (which is done when white paper is checked)... to be
> followed since it's only my understanding.

I think because Roy has done the BPC in the profile it throws Adobe PS out
of whack when it comes to soft proofing.  It sees perfect black and so
Simulate Ink Black etc is greyed out.


> 
> I'm sure I'm tiring many people with some lack of competence/knowledge
> and not-so-important issues... So, thanks Steve for your great
> willingness to help, I do appreciate a lot.
> 

Never give up trying to understand!! (especially if it is not working for
you)  I am puzzled by the results you posted and can't help but think the
answer is very simple....I would start by doing a methodical check of each
step:  is the Linearize function in the QTR ink descriptor file, is this the
same ink curve you are using, are you doing the conversion in PS and then
using QTR the same as before etc etc.

Steve

Steve,

I'm too against fragmenting, but I feel QTR has much more into it to 
just be debated on the DBWP group : not that I disregards this group, 
but IMHO QTR is not (yet?) well documented enough despite Tom Moore's 
tutorial. So it needs additional early and experimented users' 
feedback which would not necessarily appeal the DBWP group being too 
much specific like precisely the grey ICC.

So I personally favor following up on both in search for experiences 
(in fact I also read but not post on piezo group and a couple of 
others). At the end of the day, this is a question of who 
participates/responds and where to post to debate with these 
people... 

What I tend to avoid is to mail Roy directly out of the group (though 
I did once) since despite his support and prompt reply and having 
felt frustrated at the little information I found on the group, I 
think it's far better to share Roy's reply having it public.

The lastest ressource I'm looking for is an ICC one : color.org is 
not so "accessible", and I found nothing on Yahoo groups. I'm 
currently following sf.net lcms this is a profiler development 
project, lots of info but really hard to follow, I saw Roy posting 
there too.

As always thanks for your replies and help.

Olivier


--- In DigitalBlackandWhiteThePrint@yahoogroups.com, Steve Kale 
<stevekale@b...> wrote:
> Olivier
> 
> No problem.  I have not joined the QTR group yet (in general I am 
against
> fragmentation of this list - it is powerful due to its breadth).  I 
am not
> so far in front of you on this stuff so sorry if I am not 
definitive at
> times.
>

--- In DigitalBlackandWhiteThePrint@yahoogroups.com, "odesmais" <odesmais@y...> 
wrote:
> Steve,
> 
> First I need to apologize : I have double posted the below in here and 
> QTR group which is not very ethical... In fact I just realised as you 
> below mentionned that Roy would be of help in this.
> 
> I did look at the kTRC explanation and also the color.org explanation 
> of BPC, apparently this is an issue the organisation is wanting to 
> tackle, I don't know if this is being implemented in v4 or not, but by 
> experince not all soft (profilers, don't know about editing 
> softs) "understand" v4. There's also a lack of standardised perceptual 
> intents (seems all profilers implement a proprietary algorithm) which 
> is also something they are looking at, and to make it even more 
> confusing there's now a debate about CIE Lab as a PCS (it does not see 
> the same for XYZ)due to hue shifting thus non-uniformity... Also this 
> is simply and unfortunately coming from booking reading and web 
> browsing : I try to know but seldomly understand.   
> 
> Anyway, in fact what happens is that since my PS color settings is 
> (default, I never bothered to set it properly)grey : dot 15% I work 
> with too "light" a working space and since the stepwedge is not tagged 
> PS assumes the dot 15% as a source space thus printing light after 
> converting.

It's not really that dotgain 15% is "too" light.  It's just lighter than some of
the other grayspaces.   Profiles/grayspaces just define what the numbers
in the file mean.  Think of a ruler.  It can be labeled in cm or inches.
Either works just fine.  CM are shorter than IN but labels like "too short" or 
"too long" don't make much sense.  Convert to Profile is like changing
12 in to 30 cm -- same length different units.  Assign is like stamping the
cm or in label on the ruler.

> 
> I'd like to measure the actual output of the QTR-ICC profiling but I 
> feel it's a lost search. 

I think maybe this question is a source of confusion.  The QTR-ICC profile
is basically just a conversion table.  It allows the CMS to convert from the
source profile to the printer output.  So I don't know what "actual output 
of the QTR-ICC" really means.

   When printing the raw stepwedge in QTR, datas 
> are fed irrespective of color space, so the raw L* value are dependent 
> on printer-ink-paper and does not reflect...anything else than the 
> linearised stepwedge printed on ONE printer with ONE ink set and ONE 
> paper.  

Yes, that's right.

   May be there's a way to translate the linear L* measurements 
> into a gamma 

At lot tend to be read into the word "gamma".  All it is, is the exponent in
a mathematical function.  It's a different math function than L*.

    and set it as a grey working space to perform the 
> conversion with the QTR ICC profile to evaluate, but I don't know it. I 
> asked Roy if 1.8 would make sense. If not I leave it here since anyway 
> it will not improve the workflow. I'll set color settings grey : gamma 
> 2.2 (RVB and the windows environment) to minimize conversions and that 
> will do.

If you do both gray and RGB (RVB) I'd stick with workspaces that have the
same luminance curve i.e. gamma.  GG 2.2 corresponds to Adobe RGB.
GG 1.8 corresponds with Apple RGB or Colormatch RGB.

> 
> BPC remains a mystery : I use to do the soft-proofing with 
> perceptual+BPC+white paper (which also turns on Blakc ink). However it 
> might be that you need/can soft proof not turning BPC but Black ink 
> simulation (which is done when white paper is checked)... to be 
> followed since it's only my understanding.

Softproofing with color is a lot more important than with B&W.  In fact I rarely
use it anymore.  Since there are no gamut issues with B&W mappings are
very simple white-to-white and black-to-black.   The important part is just
that all the intermediate values are linearly expanded or contracted using
Y luminance values (from XYZ).   The only exception to this is that you
may want to reduce the black on the screen (i.e. reduce the dynamic range)
so that it's closer to the DR of matte paper.  I think BPC is mainly an issue
for color where a black could be a warm black or a cool black.  But it can also
be used here for reducing the dyn range.  You can easily see this by using
the QTR Gray Matte Paper profile -- this allows selecting BPC or not.

Roy

> 
> I'm sure I'm tiring many people with some lack of competence/knowledge 
> and not-so-important issues... So, thanks Steve for your great 
> willingness to help, I do appreciate a lot.
> 
> Olivier
> 
> 
> --- In DigitalBlackandWhiteThePrint@yahoogroups.com, Steve Kale 
> <stevekale@b...> wrote:
> > Olivier
> > 
> > Roy should step in here but (1) the ICC spec discusses how the kTRC 
> data
> > should be scaled for media white point-relative colorimetry and (2) 
> in the
> > documents section of www.color.org you will also find a discussion 
> paper on
> > Adobe black point compensation.  Roy has done both the black and 
> white point
> > scaling.
> > 
> >

Roy,

Sorry, my wording and conception of ICC are quite approximative.

So yes whether grey dot 15,20%, gamma1.8, 2.2... the profile will 
still do its job of converting from source to destination. My initial 
purpose was just to visualise the QTR ICC in action. Because I don't 
understand why when I look at the generic Matte Profile the kTRC is 
(far) above the 45° while my QTR ICC is (far) below ? And this should 
have nothing to do with working spaces : all is done in QTR with no 
CMS. Have done it wrong ? kTRC curve reflects sourceL to destinationL 
right ? so my ICC would actually darken the print, while the generic 
Matte lightens it ???

I personally can't really visualise the print without the soft 
proofing. At least as of now it reflects the paper white (in L) 
calling sometimes for addtional contrast. Besides, as a matter of 
fact I played a bit today with the Matte Profile and could see it 
does add something to preview prints with and without BPC also to 
determine a wanted contrast.  
So how do you recommend soft proofing as of now : perceptual, white 
paper on(that also shows Black ink) and forget BPC that can not (yet)
be used ?

Thanks.

Olivier

--- In DigitalBlackandWhiteThePrint@yahoogroups.com, "odesmais" <odesmais@y...> 
wrote:
> Roy,
> 
> Sorry, my wording and conception of ICC are quite approximative.
> 
> So yes whether grey dot 15,20%, gamma1.8, 2.2... the profile will 
> still do its job of converting from source to destination. My initial 
> purpose was just to visualise the QTR ICC in action. Because I don't 
> understand why when I look at the generic Matte Profile the kTRC is 
> (far) above the 45° while my QTR ICC is (far) below ? And this should 
> have nothing to do with working spaces : all is done in QTR with no 
> CMS. Have done it wrong ? kTRC curve reflects sourceL to destinationL 
> right ? so my ICC would actually darken the print, while the generic 
> Matte lightens it ???

I'm beginning to think the discussion so far has been too theoretical.  This
sounds like there's something amiss in the workflow -- i.e. the
practical side.   You should send me details of what you are doing so
I can understand what's happening.

> 
> I personally can't really visualise the print without the soft 
> proofing. At least as of now it reflects the paper white (in L) 

You ought to be able to get the same view in regular mode that you
are finding useful in the softproof.

Roy

> calling sometimes for addtional contrast. Besides, as a matter of 
> fact I played a bit today with the Matte Profile and could see it 
> does add something to preview prints with and without BPC also to 
> determine a wanted contrast.  
> So how do you recommend soft proofing as of now : perceptual, white 
> paper on(that also shows Black ink) and forget BPC that can not (yet)
> be used ?
> 
> Thanks.
> 
> Olivier

> From: Roy Harrington <roy@...>
:
>> Roy,
>> 
>> Sorry, my wording and conception of ICC are quite approximative.
>> 
>> So yes whether grey dot 15,20%, gamma1.8, 2.2... the profile will
>> still do its job of converting from source to destination. My initial
>> purpose was just to visualise the QTR ICC in action. Because I don't
>> understand why when I look at the generic Matte Profile the kTRC is
>> (far) above the 45° while my QTR ICC is (far) below ? And this should
>> have nothing to do with working spaces : all is done in QTR with no
>> CMS. Have done it wrong ? kTRC curve reflects sourceL to destinationL
>> right ? so my ICC would actually darken the print, while the generic
>> Matte lightens it ???
> 
> I'm beginning to think the discussion so far has been too theoretical.  This
> sounds like there's something amiss in the workflow -- i.e. the
> practical side.   You should send me details of what you are doing so
> I can understand what's happening.
> 

Yes.  Oliver - the situation should be the opposite of what you describe
above.  I am very surprised that you are getting a concave kTRC when the PCS
is CIELab.  A concave kTRC darkens the print vs linear L*.

Agh!!  Sorry I should have said:

"A conVEX kTRC darkens the print vs linear L*."

> From: Steve Kale <stevekale@...>
> Reply-To: <DigitalBlackandWhiteThePrint@yahoogroups.com>
> Date: Tue, 09 Aug 2005 10:19:24 +0100
> To: <DigitalBlackandWhiteThePrint@yahoogroups.com>
> Subject: Re: [Digital BW] QTR ICC  in use
> 
> 
> 
> 
>> From: Roy Harrington <roy@harrington.com>
> :
>>> Roy,
>>> 
>>> Sorry, my wording and conception of ICC are quite approximative.
>>> 
>>> So yes whether grey dot 15,20%, gamma1.8, 2.2... the profile will
>>> still do its job of converting from source to destination. My initial
>>> purpose was just to visualise the QTR ICC in action. Because I don't
>>> understand why when I look at the generic Matte Profile the kTRC is
>>> (far) above the 45° while my QTR ICC is (far) below ? And this should
>>> have nothing to do with working spaces : all is done in QTR with no
>>> CMS. Have done it wrong ? kTRC curve reflects sourceL to destinationL
>>> right ? so my ICC would actually darken the print, while the generic
>>> Matte lightens it ???
>> 
>> I'm beginning to think the discussion so far has been too theoretical.  This
>> sounds like there's something amiss in the workflow -- i.e. the
>> practical side.   You should send me details of what you are doing so
>> I can understand what's happening.
>> 
> 
> Yes.  Oliver - the situation should be the opposite of what you describe
> above.  I am very surprised that you are getting a concave kTRC when the PCS
> is CIELab.  A concave kTRC darkens the print vs linear L*.
> 
> 
> 
> 
> 
> Please visit the Group Homepage to check the Files, and other resources as
> they are often being updated.
> 
> http://groups.yahoo.com/group/DigitalBlackandWhiteThePrint
> 
> If you wish to receive no emails or just a daily digest, or you wish to
> unsubscribe, please edit your Membership preferences by visiting this same
> page.
> 
> Please follow these basic guidelines:
> - As threads develop, trim off excess portions of earlier messages to keep
> them short.
> - Good manners are required at all time. No personal attacks or flames.
> Hostile, aggressive or argumentative users may be removed from the membership
> without notice.
> - Keep your posts and threads related to the group topic of digital B&W
> printing. Users who persistently make off-topic posts may be removed from the
> membership.
> - By posting on this forum you agree to abide by the group rules and
> guidelines, and to abide by the actions and decisions of the group Owner and
> Moderators. See ³Group Topic, Rules and Guidelines² in the Files section:
> http://groups.yahoo.com/group/DigitalBlackandWhiteThePrint/files/
> 
> BY PARTICIPATING IN AND/OR POSTING MESSAGES TO THE DIGITAL BW, THE PRINT
> YAHOO! GROUP YOU EXPRESSLY UNDERSTAND AND AGREE THAT THE ³OWNER² AND
> ³MODERATORS² OF DIGITAL BW, THE PRINT YAHOO GROUP SHALL NOT BE LIABLE TO YOU
> FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, CONSEQUENTIAL OR EXEMPLARY
> DAMAGES, INCLUDING BUT NOT LIMITED TO, DAMAGES FOR LOSS OF PROFITS, GOODWILL,
> USE, DATA OR OTHER INTANGIBLE LOSSES (EVEN IF THE  ³OWNER² AND ³MODERATORS² OF
> DIGITAL BW, THE PRINT YAHOO GROUP HAVE BEEN ADVISED OF THE POSSIBILITY OF SUCH
> DAMAGES), RESULTING FROM: (i) THE USE OR THE INABILITY TO USE THE DIGITAL BW,
> THE PRINT YAHOO GROUP; (ii) UNAUTHORIZED ACCESS TO OR ALTERATION OF YOUR
> TRANSMISSIONS OR DATA; (iii) STATEMENTS OR CONDUCT OF ANY THIRD PARTY ON THE
> DIGITAL BW, THE PRINT YAHOO GROUP; OR (iv) ANY OTHER MATTER RELATING TO THE
> DIGITAL BW, THE PRINT YAHOO GROUP.
>  
> Yahoo! Groups Links
> 
> 
> 
>  
> 
> 
>

Roy, Steve,

Thanks both of you for your help. I agree this shoud go more 
practical. So my system :

Windows XP
Epson 1280
PiezoBW ICC CIS (K, Dark grey, 2 same grey, 2 same light grey, the 
tubing and correspondances KCMY/LC/LM with the Piezo inks have of 
course been re-checked)
DTP 20 X-rite Spectro
Colorshop X for color/gamut/profile/measures viewing
Paper : EEM 

The curve is :

PRINTER=Quad1290
CURVE_NAME=EEM2
GRAPH_CURVE=YES
N_OF_INKS=6
DEFAULT_INK_LIMIT=65
BOOST_K=75
LIMIT_K=
LIMIT_C=
LIMIT_M=50
LIMIT_Y=50
LIMIT_LC=
LIMIT_LM=
N_OF_GRAY_PARTS=4
GRAY_INK_1=K
GRAY_VAL_1=100
GRAY_INK_2=C
GRAY_VAL_2=39
GRAY_INK_3=M
GRAY_VAL_3=12
GRAY_INK_4=Y
GRAY_VAL_4=5
GRAY_HIGHLIGHT=4
GRAY_SHADOW=8
GRAY_OVERLAP=5
GRAY_GAMMA=1
GRAY_CURVE=
N_OF_TONER_PARTS=0
TONER_HIGHLIGHT=10
TONER_SHADOW=10
TONER_GAMMA=1
TONER_CURVE=
N_OF_TONER_2_PARTS=0
TONER_2_HIGHLIGHT=10
TONER_2_SHADOW=10
TONER_2_GAMMA=1
TONER_2_CURVE=
N_OF_UNUSED=0
COPY_CURVE_LC=M
COPY_CURVE_LM=Y
LINEARIZE="96,37 81,17 70,98 62,61 54,44 48,72 45,2 41,69 38,03 34,46 
32,23 30,35 29,03 27,75 25,28 22,65 20,87 19,72 18,61 18,08 17,75 "

L* measures once linearised :

L A B
96,45 -0,20 -1,20
92,90 -0,06 0,03
89,24 0,12 1,26
85,27 0,31 2,21
81,34 0,56 3,00
77,40 0,68 3,63
73,61 0,76 4,15
70,04 0,84 4,32
66,63 1,01 4,34
62,30 1,13 4,52
58,03 1,25 4,48
53,54 1,42 4,28
50,04 1,39 4,27
45,97 1,44 3,81
41,94 1,35 3,41
37,92 1,29 3,10
34,08 1,02 2,91
30,44 0,86 2,10
25,85 0,78 1,77
21,73 0,85 1,67
17,70 0,73 1,24

PM5 file I feed Create-ICC with :

LGOROWLENGTH 1
CREATED "7/21/2003" # Time: 12:57
INSTRUMENTATION "eye-one"
MEASUREMENT_SOURCE ""Illumination=D50 ObserverAngle=2°
WhiteBase=Absolute Filter=No""
KEYWORD "SampleID"
KEYWORD "SAMPLE_NAME"
NUMBER_OF_FIELDS 9
BEGIN_DATA_FORMAT
SampleID SAMPLE_NAME GRAY XYZ_X XYZ_Y XYZ_Z
LAB_L LAB_A LAB_B
END_DATA_FORMAT
NUMBER_OF_SETS 21
BEGIN_DATA
1 A1 0.00 90.30 93.00 83.24 96,45 -0,20 -1,20
2 B1 5.00 81.48 83.75 73.66 92,90 -0,06 0,03
3 C1 10.00 72.64 74.52 64.14 89,24 0,12 1,26
4 D1 15.00 64.02 65.53 55.03 85,27 0,31 2,21
5 E1 20.00 56.26 57.46 47.10 81,34 0,56 3,00
6 F1 25.00 50.08 51.05 41.01 77,40 0,68 3,63
7 G1 30.00 43.86 44.60 34.99 73,61 0,76 4,15
8 H1 35.00 37.74 38.29 29.29 70,04 0,84 4,32
9 I1 40.00 32.45 32.85 24.73 66,63 1,01 4,34
10 J1 45.00 27.61 27.90 20.62 62,30 1,13 4,52
11 K1 50.00 23.84 24.05 17.57 58,03 1,25 4,48
12 L1 55.00 20.16 20.31 14.71 53,54 1,42 4,28
13 M1 60.00 16.91 17.04 12.35 50,04 1,39 4,27
14 N1 65.00 13.62 13.72 9.98 45,97 1,44 3,81
15 O1 70.00 11.01 11.10 8.10 41,94 1,35 3,41
16 P1 75.00 8.96 9.04 6.59 37,92 1,29 3,10
17 Q1 80.00 7.10 7.16 5.22 34,08 1,02 2,91
18 R1 85.00 5.56 5.61 4.11 30,44 0,86 2,10
19 S1 90.00 4.32 4.36 3.18 25,85 0,78 1,77
20 T1 95.00 2.92 2.95 2.21 21,73 0,85 1,67
21 U1 100.00 2.04 2.08 1.56 17,70 0,73 1,24
END_DATA

As you can see only the Lab (not the xyz, just a bit lazy to do the 
copy/paste manipulation also for xyz but it does not seem to be 
necessary)have been modified from a file I found (BTW on one of the 
groups).

Now L* measures :

- G2,2 is the QTR stepwedge opened in PS converted to grey working 
space gamma2,2 (all values read from 0-100 in 5 even steps)and 
printed on EEM.
- G2,2 is QTR stepwedge opened in PS converted to grey working space 
gamma2,2 and converted again with the QTR ICC created (values from 0-
100 is various steps after the QTR conversion as expected)and printed 
on EEM. 

G 2,2	G 2,2_Profil

96,16	96,31
92,75	92,82
88,97	87,89
84,98	83,7
81,03	79,67
76,85	75,25
72,72	70,42
69,22	66,39
65,85	62,03
61,83	57,92
57,86	54,21
54,04	49,45
50,62	46,73
46,93	41,67
42,84	36,07
38,59	30,65
34,4	26,86
30,44	23,35
25,46	19,83
21,31	18,43
17,93	17,5

My problem is :
1. I expected the compression resulting of the conversion should be 
in the form of a "perfect" curve but if you draw the G2.2profiled L* 
curve (excel)you will see "bumpings" (I guess this is the way you say 
it in English). G2,2 is a straight line, while G2,2_Profil is neither 
a straight line (fine) nor a perfect curve in a kind of gamma shape 
one.
2. Why the Generic Matte Profile and the QTR ICC created one don't 
have the kTRC curve in the same part of the 45° line (GMprofile 
above, mine below?).In IMHO both should be in the upper part 
(lighter) because if Lblack is no pure black the all values should be 
lighten "perceptual". Besides GMprofile kTRC is a perfect curve again 
in a fine gamma shape look, while the one I get is an bumped bowl 
shape (what my 3-years-old daughter would draw as a smile !!!). It 
doesn't matter really nut this is where I feel it demonstrate a 
problem.
3. I get significant blockages in the 90-100 once the profile is 
applied so that I did not even bother running the Piezo test for the 
lightest 10-5% and darkest 90-95%.

At the time I'm writting it I'm just getting a bad doubt : should the 
PM5 file first be feeding the "QTR-Linearize-data" and the after get 
the "whatever-out.txt" dropped to "QTR-create-ICC" ? 

I'm FOR SURE doing something wrong but can't identify... 

I can e-mail my profile or whatever is needed if necessary.
Sorry for my wording, it's not too easy to express these things.
Thanks again.

Olivier

Sorry I missed a description.

--- In DigitalBlackandWhiteThePrint@yahoogroups.com, "odesmais" 
<odesmais@y...> wrote:
> Roy, Steve,
> 
> Thanks both of you for your help. I agree this shoud go more 
> practical. So my system :
> 
> Windows XP
> Epson 1280
> PiezoBW ICC CIS (K, Dark grey, 2 same grey, 2 same light grey, the 
> tubing and correspondances KCMY/LC/LM with the Piezo inks have of 
> course been re-checked)
> DTP 20 X-rite Spectro
> Colorshop X for color/gamut/profile/measures viewing
> Paper : EEM 
> 
> The curve is :
> 
> PRINTER=Quad1290
> CURVE_NAME=EEM2
> GRAPH_CURVE=YES
> N_OF_INKS=6
> DEFAULT_INK_LIMIT=65
> BOOST_K=75
> LIMIT_K=
> LIMIT_C=
> LIMIT_M=50
> LIMIT_Y=50
> LIMIT_LC=
> LIMIT_LM=
> N_OF_GRAY_PARTS=4
> GRAY_INK_1=K
> GRAY_VAL_1=100
> GRAY_INK_2=C
> GRAY_VAL_2=39
> GRAY_INK_3=M
> GRAY_VAL_3=12
> GRAY_INK_4=Y
> GRAY_VAL_4=5
> GRAY_HIGHLIGHT=4
> GRAY_SHADOW=8
> GRAY_OVERLAP=5
> GRAY_GAMMA=1
> GRAY_CURVE=
> N_OF_TONER_PARTS=0
> TONER_HIGHLIGHT=10
> TONER_SHADOW=10
> TONER_GAMMA=1
> TONER_CURVE=
> N_OF_TONER_2_PARTS=0
> TONER_2_HIGHLIGHT=10
> TONER_2_SHADOW=10
> TONER_2_GAMMA=1
> TONER_2_CURVE=
> N_OF_UNUSED=0
> COPY_CURVE_LC=M
> COPY_CURVE_LM=Y
> LINEARIZE="96,37 81,17 70,98 62,61 54,44 48,72 45,2 41,69 38,03 
34,46 
> 32,23 30,35 29,03 27,75 25,28 22,65 20,87 19,72 18,61 18,08 17,75 "
> 
> L* measures once linearised :
> 
> L A B
> 96,45 -0,20 -1,20
> 92,90 -0,06 0,03
> 89,24 0,12 1,26
> 85,27 0,31 2,21
> 81,34 0,56 3,00
> 77,40 0,68 3,63
> 73,61 0,76 4,15
> 70,04 0,84 4,32
> 66,63 1,01 4,34
> 62,30 1,13 4,52
> 58,03 1,25 4,48
> 53,54 1,42 4,28
> 50,04 1,39 4,27
> 45,97 1,44 3,81
> 41,94 1,35 3,41
> 37,92 1,29 3,10
> 34,08 1,02 2,91
> 30,44 0,86 2,10
> 25,85 0,78 1,77
> 21,73 0,85 1,67
> 17,70 0,73 1,24
> 
> PM5 file I feed Create-ICC with :
> 
> LGOROWLENGTH 1
> CREATED "7/21/2003" # Time: 12:57
> INSTRUMENTATION "eye-one"
> MEASUREMENT_SOURCE ""Illumination=D50 ObserverAngle=2°
> WhiteBase=Absolute Filter=No""
> KEYWORD "SampleID"
> KEYWORD "SAMPLE_NAME"
> NUMBER_OF_FIELDS 9
> BEGIN_DATA_FORMAT
> SampleID SAMPLE_NAME GRAY XYZ_X XYZ_Y XYZ_Z
> LAB_L LAB_A LAB_B
> END_DATA_FORMAT
> NUMBER_OF_SETS 21
> BEGIN_DATA
> 1 A1 0.00 90.30 93.00 83.24 96,45 -0,20 -1,20
> 2 B1 5.00 81.48 83.75 73.66 92,90 -0,06 0,03
> 3 C1 10.00 72.64 74.52 64.14 89,24 0,12 1,26
> 4 D1 15.00 64.02 65.53 55.03 85,27 0,31 2,21
> 5 E1 20.00 56.26 57.46 47.10 81,34 0,56 3,00
> 6 F1 25.00 50.08 51.05 41.01 77,40 0,68 3,63
> 7 G1 30.00 43.86 44.60 34.99 73,61 0,76 4,15
> 8 H1 35.00 37.74 38.29 29.29 70,04 0,84 4,32
> 9 I1 40.00 32.45 32.85 24.73 66,63 1,01 4,34
> 10 J1 45.00 27.61 27.90 20.62 62,30 1,13 4,52
> 11 K1 50.00 23.84 24.05 17.57 58,03 1,25 4,48
> 12 L1 55.00 20.16 20.31 14.71 53,54 1,42 4,28
> 13 M1 60.00 16.91 17.04 12.35 50,04 1,39 4,27
> 14 N1 65.00 13.62 13.72 9.98 45,97 1,44 3,81
> 15 O1 70.00 11.01 11.10 8.10 41,94 1,35 3,41
> 16 P1 75.00 8.96 9.04 6.59 37,92 1,29 3,10
> 17 Q1 80.00 7.10 7.16 5.22 34,08 1,02 2,91
> 18 R1 85.00 5.56 5.61 4.11 30,44 0,86 2,10
> 19 S1 90.00 4.32 4.36 3.18 25,85 0,78 1,77
> 20 T1 95.00 2.92 2.95 2.21 21,73 0,85 1,67
> 21 U1 100.00 2.04 2.08 1.56 17,70 0,73 1,24
> END_DATA
> 
> As you can see only the Lab (not the xyz, just a bit lazy to do the 
> copy/paste manipulation also for xyz but it does not seem to be 
> necessary)have been modified from a file I found (BTW on one of the 
> groups).
> 
> Now L* measures :
> 
> - G2,2 is the QTR stepwedge opened in PS converted to grey working 
> space gamma2,2 (all values read from 0-100 in 5 even steps)and 
> printed on EEM.
> - G2,2_Profil is QTR stepwedge opened in PS converted to grey 
working space 
> gamma2,2 and converted again with the QTR ICC created (values from 
0-
> 100 is various steps after the QTR conversion as expected)and 
printed 
> on EEM. 
> 
> G 2,2	G 2,2_Profil
> 
> 96,16	96,31
> 92,75	92,82
> 88,97	87,89
> 84,98	83,7
> 81,03	79,67
> 76,85	75,25
> 72,72	70,42
> 69,22	66,39
> 65,85	62,03
> 61,83	57,92
> 57,86	54,21
> 54,04	49,45
> 50,62	46,73
> 46,93	41,67
> 42,84	36,07
> 38,59	30,65
> 34,4	26,86
> 30,44	23,35
> 25,46	19,83
> 21,31	18,43
> 17,93	17,5
> 
> My problem is :
> 1. I expected the compression resulting of the conversion should be 
> in the form of a "perfect" curve but if you draw the G2.2_profiled 
L* 
> curve (excel)you will see "bumpings" (I guess this is the way you 
say 
> it in English). G2,2 is a straight line, while G2,2_Profil is 
neither 
> a straight line (fine) nor a perfect curve in a kind of gamma shape 
> one.
> 2. Why the Generic Matte Profile and the QTR ICC created one don't 
> have the kTRC curve in the same part of the 45° line (GMprofile 
> above, mine below?).In IMHO both should be in the upper part 
> (lighter) because if Lblack is no pure black the all values should 
be 
> lighten "perceptual". Besides GMprofile kTRC is a perfect curve 
again 
> in a fine gamma shape look, while the one I get is an bumped bowl 
> shape (what my 3-years-old daughter would draw as a smile !!!). It 
> doesn't matter really nut this is where I feel it demonstrate a 
> problem.
> 3. I get significant blockages in the 90-100 once the profile is 
> applied so that I did not even bother running the Piezo test for 
the 
> lightest 10-5% and darkest 90-95%.
> 
> At the time I'm writting it I'm just getting a bad doubt : should 
the 
> PM5 file first be feeding the "QTR-Linearize-data" and the after 
get 

> the "whatever-out.txt" dropped to "QTR-create-ICC" ? 
> 
> I'm FOR SURE doing something wrong but can't identify... 
> 
> I can e-mail my profile or whatever is needed if necessary.
> Sorry for my wording, it's not too easy to express these things.
> Thanks again.
> 
> Olivier

> From: odesmais <odesmais@...>

> 
> My problem is :

> 1. I expected the compression resulting of the conversion should be
> in the form of a "perfect" curve but if you draw the G2.2profiled L*
> curve (excel)you will see "bumpings" (I guess this is the way you say
> it in English). G2,2 is a straight line, while G2,2_Profil is neither
> a straight line (fine) nor a perfect curve in a kind of gamma shape
> one.

I will let Roy answer this as he is conversant with the white point and
black point scaling.

> 2. Why the Generic Matte Profile and the QTR ICC created one don't
> have the kTRC curve in the same part of the 45° line (GMprofile
> above, mine below?).In IMHO both should be in the upper part
> (lighter) because if Lblack is no pure black the all values should be
> lighten "perceptual". Besides GMprofile kTRC is a perfect curve again
> in a fine gamma shape look, while the one I get is an bumped bowl
> shape (what my 3-years-old daughter would draw as a smile !!!). It
> doesn't matter really nut this is where I feel it demonstrate a
> problem.

The generic ICC profiles have XYZ as the PCS and hence are concave curves -
you are looking at XYZ_Y and hence it has curvature.  QTR Create ICC uses
Lab as the PCS.  If there were no BPC being done and your linearized
greyscale was perfectly linear then the kTRC would be a straight line
(whereas if the PCS were XYZ it would still be curved like the generics).
Roy will have to comment on the effect of the BPC he is doing.

> 3. I get significant blockages in the 90-100 once the profile is
> applied so that I did not even bother running the Piezo test for the
> lightest 10-5% and darkest 90-95%.

The compression of shadows you are seeing is similar to what I have
observed.  Again Roy will need to comment on the BPC.
> 
> At the time I'm writting it I'm just getting a bad doubt : should the
> PM5 file first be feeding the "QTR-Linearize-data" and the after get
> the "whatever-out.txt" dropped to "QTR-create-ICC" ?

I am not sure what you mean by the above but you should first print a
unlinearized greyscale and measure it with MeasureTool and use the output to
run QTR-Linearize-Data.  Then add the linearize function to the ink curve
descriptor file.  Print a greyscale again with the now linearized curve and
re-measure with MeasureTool.  Use this MT output to run QTR Create ICC.

Steve

Hi Olivier,

This is very helpful.  
From what I see you are very close to the right result.

First issue is comma vs point in numbers.  The Lab values that you entered into
the pseudo data file have commas instead of points.  I think this is what is
causing the little wiggles in the curve -- the numbers are being truncated.  I can
fix this in the next version.

I've taken your values and they make a reasonable icc profile.  Both Steve and
I are on Macs so the program that plots the icc's is different.  From your
description I think your program swaps the x and y axis so what's curves down
for you is curved up for us.  The other difference is that icc's can have two
different PCS's XYZ or Lab.  Either can be used but the curves are very different.
The Gray Matte Paper one is probably based on XYZ and the one you created
is Lab which makes them hard to compare.  (I've made both kinds the Lab ones
are more obvious to look at since they start straight.)

I think you have a decent icc file as it is (except for the wiggles).  Your data
shows a darker stepwedge as expected.  No doubt you've plotted the two sets
of Lab data (G2.2 and G2.2profile).   The difference shouldn't look like a gamma
curve, in fact it'll look at lot like the icc xTRC curve I think you are seeing.  This
is basically the Perceptual Intent mapping.  It's a straightline expansion in Y units
but viewed in Lab units it darkens all and bunches the blacks together somewhat.
(BTW, notice that the K=50 step reads L=54 which is exactly the right for GG2.2)

As you mentioned the black end gets pretty blocked up.  This is due also to the
fact that you are using GG2.2.  This alone has fairly blocked shadows (see the
reduced separation on your screen) and when force that onto matte paper you 
get hit again.   But the idea is to match the print with the screen so this ought
to give you as close a match as possible.

I think you are to the point where you should try a real picture, edited in PS
to what you want on the screen (embedded profile doesn't matter).  Convert to
the GrayICC that you made and print with QTRgui.  See how well the print
matches the screen.

Roy


--- In DigitalBlackandWhiteThePrint@yahoogroups.com, "odesmais" <odesmais@y...> 
wrote:

> Roy, Steve,
> 
> Thanks both of you for your help. I agree this shoud go more 
> practical. So my system :
> 
> Windows XP
> Epson 1280
> PiezoBW ICC CIS (K, Dark grey, 2 same grey, 2 same light grey, the 
> tubing and correspondances KCMY/LC/LM with the Piezo inks have of 
> course been re-checked)
> DTP 20 X-rite Spectro
> Colorshop X for color/gamut/profile/measures viewing
> Paper : EEM 
> 
> The curve is :
> 
> PRINTER=Quad1290
> CURVE_NAME=EEM2
> GRAPH_CURVE=YES
> N_OF_INKS=6
> DEFAULT_INK_LIMIT=65
> BOOST_K=75
> LIMIT_K=
> LIMIT_C=
> LIMIT_M=50
> LIMIT_Y=50
> LIMIT_LC=
> LIMIT_LM=
> N_OF_GRAY_PARTS=4
> GRAY_INK_1=K
> GRAY_VAL_1=100
> GRAY_INK_2=C
> GRAY_VAL_2=39
> GRAY_INK_3=M
> GRAY_VAL_3=12
> GRAY_INK_4=Y
> GRAY_VAL_4=5
> GRAY_HIGHLIGHT=4
> GRAY_SHADOW=8
> GRAY_OVERLAP=5
> GRAY_GAMMA=1
> GRAY_CURVE=
> N_OF_TONER_PARTS=0
> TONER_HIGHLIGHT=10
> TONER_SHADOW=10
> TONER_GAMMA=1
> TONER_CURVE=
> N_OF_TONER_2_PARTS=0
> TONER_2_HIGHLIGHT=10
> TONER_2_SHADOW=10
> TONER_2_GAMMA=1
> TONER_2_CURVE=
> N_OF_UNUSED=0
> COPY_CURVE_LC=M
> COPY_CURVE_LM=Y
> LINEARIZE="96,37 81,17 70,98 62,61 54,44 48,72 45,2 41,69 38,03 34,46 
> 32,23 30,35 29,03 27,75 25,28 22,65 20,87 19,72 18,61 18,08 17,75 "
> 
> L* measures once linearised :
> 
> L A B
> 96,45 -0,20 -1,20
> 92,90 -0,06 0,03
> 89,24 0,12 1,26
> 85,27 0,31 2,21
> 81,34 0,56 3,00
> 77,40 0,68 3,63
> 73,61 0,76 4,15
> 70,04 0,84 4,32
> 66,63 1,01 4,34
> 62,30 1,13 4,52
> 58,03 1,25 4,48
> 53,54 1,42 4,28
> 50,04 1,39 4,27
> 45,97 1,44 3,81
> 41,94 1,35 3,41
> 37,92 1,29 3,10
> 34,08 1,02 2,91
> 30,44 0,86 2,10
> 25,85 0,78 1,77
> 21,73 0,85 1,67
> 17,70 0,73 1,24
> 
> PM5 file I feed Create-ICC with :
> 
> LGOROWLENGTH 1
> CREATED "7/21/2003" # Time: 12:57
> INSTRUMENTATION "eye-one"
> MEASUREMENT_SOURCE ""Illumination=D50 ObserverAngle=2°
> WhiteBase=Absolute Filter=No""
> KEYWORD "SampleID"
> KEYWORD "SAMPLE_NAME"
> NUMBER_OF_FIELDS 9
> BEGIN_DATA_FORMAT
> SampleID SAMPLE_NAME GRAY XYZ_X XYZ_Y XYZ_Z
> LAB_L LAB_A LAB_B
> END_DATA_FORMAT
> NUMBER_OF_SETS 21
> BEGIN_DATA
> 1 A1 0.00 90.30 93.00 83.24 96,45 -0,20 -1,20
> 2 B1 5.00 81.48 83.75 73.66 92,90 -0,06 0,03
> 3 C1 10.00 72.64 74.52 64.14 89,24 0,12 1,26
> 4 D1 15.00 64.02 65.53 55.03 85,27 0,31 2,21
> 5 E1 20.00 56.26 57.46 47.10 81,34 0,56 3,00
> 6 F1 25.00 50.08 51.05 41.01 77,40 0,68 3,63
> 7 G1 30.00 43.86 44.60 34.99 73,61 0,76 4,15
> 8 H1 35.00 37.74 38.29 29.29 70,04 0,84 4,32
> 9 I1 40.00 32.45 32.85 24.73 66,63 1,01 4,34
> 10 J1 45.00 27.61 27.90 20.62 62,30 1,13 4,52
> 11 K1 50.00 23.84 24.05 17.57 58,03 1,25 4,48
> 12 L1 55.00 20.16 20.31 14.71 53,54 1,42 4,28
> 13 M1 60.00 16.91 17.04 12.35 50,04 1,39 4,27
> 14 N1 65.00 13.62 13.72 9.98 45,97 1,44 3,81
> 15 O1 70.00 11.01 11.10 8.10 41,94 1,35 3,41
> 16 P1 75.00 8.96 9.04 6.59 37,92 1,29 3,10
> 17 Q1 80.00 7.10 7.16 5.22 34,08 1,02 2,91
> 18 R1 85.00 5.56 5.61 4.11 30,44 0,86 2,10
> 19 S1 90.00 4.32 4.36 3.18 25,85 0,78 1,77
> 20 T1 95.00 2.92 2.95 2.21 21,73 0,85 1,67
> 21 U1 100.00 2.04 2.08 1.56 17,70 0,73 1,24
> END_DATA
> 
> As you can see only the Lab (not the xyz, just a bit lazy to do the 
> copy/paste manipulation also for xyz but it does not seem to be 
> necessary)have been modified from a file I found (BTW on one of the 
> groups).
> 
> Now L* measures :
> 
> - G2,2 is the QTR stepwedge opened in PS converted to grey working 
> space gamma2,2 (all values read from 0-100 in 5 even steps)and 
> printed on EEM.
> - G2,2 is QTR stepwedge opened in PS converted to grey working space 
> gamma2,2 and converted again with the QTR ICC created (values from 0-
> 100 is various steps after the QTR conversion as expected)and printed 
> on EEM. 
> 
> G 2,2	G 2,2_Profil
> 
> 96,16	96,31
> 92,75	92,82
> 88,97	87,89
> 84,98	83,7
> 81,03	79,67
> 76,85	75,25
> 72,72	70,42
> 69,22	66,39
> 65,85	62,03
> 61,83	57,92
> 57,86	54,21
> 54,04	49,45
> 50,62	46,73
> 46,93	41,67
> 42,84	36,07
> 38,59	30,65
> 34,4	26,86
> 30,44	23,35
> 25,46	19,83
> 21,31	18,43
> 17,93	17,5
> 
> My problem is :
> 1. I expected the compression resulting of the conversion should be 
> in the form of a "perfect" curve but if you draw the G2.2profiled L* 
> curve (excel)you will see "bumpings" (I guess this is the way you say 
> it in English). G2,2 is a straight line, while G2,2_Profil is neither 
> a straight line (fine) nor a perfect curve in a kind of gamma shape 
> one.
> 2. Why the Generic Matte Profile and the QTR ICC created one don't 
> have the kTRC curve in the same part of the 45° line (GMprofile 
> above, mine below?).In IMHO both should be in the upper part 
> (lighter) because if Lblack is no pure black the all values should be 
> lighten "perceptual". Besides GMprofile kTRC is a perfect curve again 
> in a fine gamma shape look, while the one I get is an bumped bowl 
> shape (what my 3-years-old daughter would draw as a smile !!!). It 
> doesn't matter really nut this is where I feel it demonstrate a 
> problem.
> 3. I get significant blockages in the 90-100 once the profile is 
> applied so that I did not even bother running the Piezo test for the 
> lightest 10-5% and darkest 90-95%.
> 
> At the time I'm writting it I'm just getting a bad doubt : should the 
> PM5 file first be feeding the "QTR-Linearize-data" and the after get 
> the "whatever-out.txt" dropped to "QTR-create-ICC" ? 
> 
> I'm FOR SURE doing something wrong but can't identify... 
> 
> I can e-mail my profile or whatever is needed if necessary.
> Sorry for my wording, it's not too easy to express these things.
> Thanks again.
> 
> Olivier

Roy

Isn't this a little misleading in that the black point will be quite
(significantly) different (print vs display)?

Cheers

Steve

> From: Roy Harrington <roy@...>

> 
> I think you are to the point where you should try a real picture, edited in PS
> to what you want on the screen (embedded profile doesn't matter).  Convert to
> the GrayICC that you made and print with QTRgui.  See how well the print
> matches the screen.

Roy,

Thanks for the reply. 

First, I believe I can turn comma vs point probably by simply 
swapping Colorshop X back to US vs French. If not I'll do it manually.

Where I don't follow you is on the assessing the profile is 
acceptable. Yes the G2,2_Profil samples make a similar-to-kTRC 
curve : this precisely is why I suspected my whole profile(ing) was 
wrong, I was expecting a curve similar to the GMprofile one in the 
sense of being perfectly curvilinear since the L* measures feeding 
the droplet are in a linear description and there's no color. A mere 
compression in my expectations should have provided the perfect curve 
I see on the GMprofile in a kind of gamma function. 

Perceptual intent is compressing the gamut, and in our case the L* 
only, so why isn't it a kind of (curvi)linear line and not the messy 
one I got? Has it got anyting to do with the BPC handling ?


Yes G2,2 is a "dark" space, but any source space does not really 
matter as long as you keep it as source for both prints (w/ and w/o 
the profile)for the evaluating (as you said embedded profile does not 
really matter): as you can see at G2,2 the measured L* are the one I 
get when printing just the linearised stepwedge. The "proportional" 
effect (darkening of the print) thru the ICC was what I intended to 
evaluate and did prove unsastisfactory (IMHO). 

Yes, I started producing one or two prints though not yet with the 
purpose of really evaluating precisely the prints (I was just bored 
with the spectro and all the datas...). Well as expected blocked 
shadows... but I admit I did not edit at all (and even not for the 
blacks). And I'd say it matched the screen, but again this is not an 
evaluation, I did not carefully looked at the blacks and the 
highlights...

When I work color I mostly use rel/col. rendering unless I have a lot 
of out-of-gamut colors and if needed edit one or two tones so I keep 
the saturation (there's no harsh and arbiratory compression of the 
source gamut). Even for KCMY (vs grey inks) B&W I was getting 
acceptable results (albeit metamerism and dominantes with some 
papers). Wouldn't it make (any) sense to use the same for 
monochromatic B&W ? 

I must sound like winding, but I was expecting the ICC to possibly 
compress "evenly" the values and use it straight to print... I feel I 
got a CM workflow but at the detriment of the grayscale smoothness.

Anyway, you have a great soft Roy and offer a great support. If you 
grant me additional advises they will be more than welcome.

Thanks a lot.

Olivier

Steve,

Yes it will, but personally if I have the white (paper one) well 
plotted, I feel I can manage a different black : after some prints 
you can predict more or less ; and print black vs monitor black is 
always an issue that can't really be tackled. All one needs is some 
kind of predictability.

I've answered Roy, and frankly I'm not comfortable with the profile I 
created : to edit is fine but the black region is the most critical 
and difficult one. Besides since the profile has lots of bumpings 
even with the "best" of my editing ability I only expect average 
results.

This is why I suggested col/rel rendering which IMO will keep the 
smonthess vs percep. and ease the editing (unless wrong in my little 
understanding of QTR and ICC).

Olivier

--- In DigitalBlackandWhiteThePrint@yahoogroups.com, Steve Kale 
<stevekale@b...> wrote:

> Roy
> 
> Isn't this a little misleading in that the black point will be quite
> (significantly) different (print vs display)?
> 
> Cheers
> 
> Steve
> 
>

Olivier, can you post the text file that comes from exporting the 
data from the Density Scratch pad in ColorshopX? It will make a text 
file, and if you could just cut and paste the data into a message. I 
want to try something in ColorshopX

I think you will find there is no difference between relcol and perceptual
in a B&W world...A greyscale ICC profile has only one intent.  It is
required to be media relative with respect to white point.  The only choice
is BPC which you would definitely want else you simply clip the blacks.

> From: odesmais <odesmais@...>

> 
> This is why I suggested col/rel rendering which IMO will keep the
> smonthess vs percep. and ease the editing (unless wrong in my little
> understanding of QTR and ICC).

--- In DigitalBlackandWhiteThePrint@yahoogroups.com, "odesmais" <odesmais@y...> 
wrote:
> Roy,
> 
> Thanks for the reply. 
> 
> First, I believe I can turn comma vs point probably by simply 
> swapping Colorshop X back to US vs French. If not I'll do it manually.
> 
> Where I don't follow you is on the assessing the profile is 
> acceptable. Yes the G2,2_Profil samples make a similar-to-kTRC 
> curve : this precisely is why I suspected my whole profile(ing) was 
> wrong, I was expecting a curve similar to the GMprofile one in the 
> sense of being perfectly curvilinear since the L* measures feeding 
> the droplet are in a linear description and there's no color. A mere 
> compression in my expectations should have provided the perfect curve 
> I see on the GMprofile in a kind of gamma function. 
> 
> Perceptual intent is compressing the gamut, and in our case the L* 
> only, so why isn't it a kind of (curvi)linear line and not the messy 
> one I got? 

Well the standard QTR linearization does a straightline of L values from
paper white to dmax black.  At first I thought that was the "right" way to
to scale L (0...100) to L (17...96).  But if you do this the prints always look
lighter that what you expect.  After a fair amount to reading I found that the
more correct way to scale is using the Y luminance values.  If you were to
plot the before and after Y values the scaling is linear -- i.e. straight-line.
It's hard to visualize this especially since the values are not separated evenly.
The weird shape in the L graph is not at all intuitive but the math just works 
out this way.

Has it got anyting to do with the BPC handling ?

Sort of because the linear Y scaling is exactly what PS does for BPC.

> 
> Yes G2,2 is a "dark" space, but any source space does not really 
> matter as long as you keep it as source for both prints (w/ and w/o 
> the profile)for the evaluating (as you said embedded profile does not 

The only important thing -- especially on the PC -- is you have to be very
deliberate about your Assigning and Converting profiles.  Sometimes its
explicit and sometimes implicit.

> really matter): as you can see at G2,2 the measured L* are the one I 
> get when printing just the linearised stepwedge. The "proportional" 
> effect (darkening of the print) thru the ICC was what I intended to 
> evaluate and did prove unsastisfactory (IMHO). 
> 
> Yes, I started producing one or two prints though not yet with the 
> purpose of really evaluating precisely the prints (I was just bored 
> with the spectro and all the datas...). Well as expected blocked 
> shadows... but I admit I did not edit at all (and even not for the 
> blacks). And I'd say it matched the screen, but again this is not an 
> evaluation, I did not carefully looked at the blacks and the 
> highlights...

"Matching the screen" is really the ONLY goal.  Color management doesn't
really address things such as shadow separation.  It's entirely based on profiles
that characterize the different devices and conversions to "best" (not perfect)
give the same feel of the image.

> 
> When I work color I mostly use rel/col. rendering unless I have a lot 
> of out-of-gamut colors and if needed edit one or two tones so I keep 
> the saturation (there's no harsh and arbiratory compression of the 
> source gamut). Even for KCMY (vs grey inks) B&W I was getting 
> acceptable results (albeit metamerism and dominantes with some 
> papers). Wouldn't it make (any) sense to use the same for 
> monochromatic B&W ? 

I believe that that is what you are getting -- its just that with color you can't
see what's going on behind the scenes.  If you can do a good color
profile it might be interesting to see and measure the results of a gray
stepwedge there.

> 
> I must sound like winding, but I was expecting the ICC to possibly 
> compress "evenly" the values and use it straight to print... I feel I 
> got a CM workflow but at the detriment of the grayscale smoothness.

I think you are focusing on the look of the curves too much.  You really need
to go by seeing your profiled screen and your prints.  Try it both ways and see
if you can see the difference in screen/print matching.

Roy

> 
> Anyway, you have a great soft Roy and offer a great support. If you 
> grant me additional advises they will be more than welcome.
> 
> Thanks a lot.
> 
> Olivier

--- In DigitalBlackandWhiteThePrint@yahoogroups.com, "dfaprinting" 
<dfaprinting@y...> wrote:
> Olivier, can you post the text file that comes from exporting the 
> data from the Density Scratch pad in ColorshopX? It will make a text 
> file, and if you could just cut and paste the data into a message. I 
> want to try something in ColorshopX

Very easily. Just let me know which one of the density measures you 
want, there tends to be a "few" files now :

1. raw
2. raw linearised
3. raw linearised G2.2 converted
4. linearised G2.2 converted and profile converted

What's your idea ?

I'll be away from Friday to Monday, if I get your reply before I'll 
post the measures, if not then after Monday.

Olivier

Roy, Steve,

Ok then I'll stop data focussing... Just adjust my data in PM5 file 
to get dot vs comma.
Now how do you advise to soft proofing : from say the grey file G2.2. 
I should go of course display->soft-proof (this is a translation of a 
French PS) and go perceptual BPC on, and white paper on (which also 
turns black in on unless wrong) in the small soft proofing setting 
window ? Edit from there and ultimately convert to the QTR ICC 
profile before the printing.

Does this make sense ?

Also, I tried to edit the dark end of the curve in PS : this is 
really a nasty piece of work, the 90-100% is hell !!! Is there any 
way to deblock shadows except harsh editing ? The only other thing I 
found was to assign my grey G2.2 file with an dot15% profile, convert 
it to QTR ICC and then apply a curve remapping 50 at 50 (well 
plotting 38 at 50 in a PS curve)and 60 at 78 . I haven't printed yet 
to see real results. But does this sound stupid ?

Thanks for your help.

Olivier

> From: odesmais <odesmais@...>

> Now how do you advise to soft proofing : from say the grey file G2.2.
> I should go of course display->soft-proof (this is a translation of a
> French PS) and go perceptual BPC on, and white paper on

Yes

>(which also 
> turns black in on unless wrong)

No it doesn't.  Take a look.

>in the small soft proofing setting
> window ? Edit from there and ultimately convert to the QTR ICC
> profile before the printing.

Personally I would work up your image WITHOUT the proof and have that be
your master - only converting or adjusting the image for the paper profile
just prior to printing (if at all).  To be honest the soft proof isn't going
to tell you much.  By far the largest issue is the different black point
(when printing to matte paper) and the QTR ICC profile will not allow you to
soft proof for that as we have discussed.  Take a look for yourself.  If you
do want a rough soft proof of the lighter black take a peak with the generic
QTR matte profile. 


> 
> Does this make sense ?
> 
> Also, I tried to edit the dark end of the curve in PS

What "curve"?


>: this is 
> really a nasty piece of work, the 90-100% is hell !!! Is there any
> way to deblock shadows except harsh editing ? The only other thing I
> found was to assign my grey G2.2 file with an dot15% profile, convert
> it to QTR ICC and then apply a curve remapping 50 at 50 (well
> plotting 38 at 50 in a PS curve)and 60 at 78 . I haven't printed yet
> to see real results. But does this sound stupid ?


I would echo Roy's comments that you should not get too wound up in the L*
figures.  For any ACTUAL image, you can open up the shadows with a curve if
need be.  I do though think it would be nicer to be able to see on screen
the lower black and the effect of the BPC in order to check if such
adjustment is necessary.  This is something Roy and I have debated.  He
quite rightly points out that BPC has to be done - it is just a matter of
whether the ICC profile does it or PS.  At the end of the day, though, the
best soft proof is a print - except that it is not dynamic and real-time
reflection as you make edits and adjustments.

Steve,

Yes you're right : I'm now in front of my PC and the black isn't turn 
on when you do paper white (I actually tried with the generic matte 
that I had on another PC and that has the black, not the media white so 
I thought the actual created ICC had both but no black at the end of 
the day). 

Anyway, the thing is when I soft-proof like discussed (working space 
2.2 but it does not really matter) : I have/see NOTHING except media 
white changing ??? So indeed the soft-proof tells me nothing ... at 
all !!! Is that supposed to be like that ?

Olivier