QTR-Quadtone RIP

Has anyone bothered to compare the 2880 .quad K7 curves that Jon Cone
supplies for a number of paper & printer combinations, with curves
produced by the QTR curves creation tool?  Jon claims that his produce
the ultimate smoothness.  Certainly his procedure for making the
curves sounds pretty exhausting.  My question is, how much difference
does it make in practice?

Ferdinand

Ferdinand

If you're asking about 'numerical' characteristics, inks in the Cone 
curves overlap a lot more than in the QTR generated ones. The 
auto-generation method that QTR uses means that you cannot have more 
than three inks overlapping at once. With a quadtone this works fine but 
when there are 7 inks this method is limiting.

You should be the judge as to whether you prefer the end results.

If you are looking for a way to make Cone-style curves with QTR I posted 
details of a method here:

http://www.howardshaw.org/docs/k7-qtr/k7curves.htm

You could also use the point list table available in qtr instead of the 
.acv files to generate similar curves.

regards
Howard
--
www.howardshaw.org

ferdinand_paris wrote:

 > Has anyone bothered to compare the 2880 .quad K7 curves that Jon Cone
 > supplies for a number of paper & printer combinations, with curves
 > produced by the QTR curves creation tool?  Jon claims that his produce
 > the ultimate smoothness.  Certainly his procedure for making the
 > curves sounds pretty exhausting.  My question is, how much difference
 > does it make in practice?
 >
 > Ferdinand
 >

Thanks Howard

The more I know, it seems the more I don't know.

What I actually meant was: can you see a significant difference in prints?

However my question now is:  how do you use such curves?  It wasn't
clear from your URL.  Is there a guide somewhere?  I can work it out
up to a point.  I can see where to load the curves into the Curve
Creator.  

I guess the question is:  where to go from there?  Is it much the same
as normal - i.e. setting the overall limit, partitioning the inks,
checking if black boost is needed, linearising and creating a profile?
 Should I be setting ink specific limits?  What about "Grey Curve",
"Toner Curve 1", "Toner Curve 2"?  Are there safe values that I could
use here?

Thanks for sharing this with me.  Any further tips would be appreciated.

Ferdinand


--- In QuadtoneRIP@yahoogroups.com, Howard Shaw <glassman@...> wrote:
>
> Ferdinand
> 
> If you're asking about 'numerical' characteristics, inks in the Cone 
> curves overlap a lot more than in the QTR generated ones. The 
> auto-generation method that QTR uses means that you cannot have more 
> than three inks overlapping at once. With a quadtone this works fine
but 
> when there are 7 inks this method is limiting.
> 
> You should be the judge as to whether you prefer the end results.
> 
> If you are looking for a way to make Cone-style curves with QTR I
posted 
> details of a method here:
> 
> http://www.howardshaw.org/docs/k7-qtr/k7curves.htm
> 
> You could also use the point list table available in qtr instead of the 
> .acv files to generate similar curves.
> 
> regards
> Howard
> --
> www.howardshaw.org
> 
> ferdinand_paris wrote:
>  > Has anyone bothered to compare the 2880 .quad K7 curves that Jon Cone
>  > supplies for a number of paper & printer combinations, with curves
>  > produced by the QTR curves creation tool?  Jon claims that his
produce

>  > the ultimate smoothness.  Certainly his procedure for making the
>  > curves sounds pretty exhausting.  My question is, how much difference
>  > does it make in practice?
>  >
>  > Ferdinand
>  >
>

Ferdinand

When you use Load Curve you are pretty much on your own - you can use 
limits but the shape of the loaded curve will also determine the limit. 
With the example curves (.acv's) the limit would be set to 100. None of 
the gray curve or toner curves facilities are needed as you can make all 
those adjustments with the shape of the loaded curve.

However, using that method can produce bad results unless you are you 
prepared to undertake a lot of testing.

Roy's suggestion of taking a preexisting curve for a similar paper (in 
particular find one with similar ink limit characteristics) and applying 
an icc profile to it is a very good one which will likely save you a lot 
of time.

regards
Howard

ferdinand_paris wrote:

> Thanks Howard
> 
> The more I know, it seems the more I don't know.
> 
> What I actually meant was: can you see a significant difference in prints?
> 
> However my question now is:  how do you use such curves?  It wasn't
> clear from your URL.  Is there a guide somewhere?  I can work it out
> up to a point.  I can see where to load the curves into the Curve
> Creator.  
> 
> I guess the question is:  where to go from there?  Is it much the same
> as normal - i.e. setting the overall limit, partitioning the inks,
> checking if black boost is needed, linearising and creating a profile?
>  Should I be setting ink specific limits?  What about "Grey Curve",
> "Toner Curve 1", "Toner Curve 2"?  Are there safe values that I could
> use here?
> 
> Thanks for sharing this with me.  Any further tips would be appreciated.
> 
> Ferdinand
>

--- In QuadtoneRIP@yahoogroups.com, Howard Shaw <glassman@...> wrote:
> However, using that method can produce bad results unless you are you 
> prepared to undertake a lot of testing.

I was afraid you'd say that.
 
> Roy's suggestion of taking a preexisting curve for a similar paper (in 
> particular find one with similar ink limit characteristics) and 
> applying an icc profile to it is a very good one which will likely
> save you a lot of time.

I agree.  My problem is that I'm trying to create a curve for Epson
VFA on a 2100, and there isn't a pre-existing curve.  As a first step
I'm trying to do it for Epson Enhanced Matte, which most of us use as
a proofing paper.  I'm doing this because, like a number of people, I
find that the supplied Cone 2880 curve is putting down too much ink,
and putting waves into the paper.  Also EEnhMatte is a lot cheaper to
experiment with.  

I could probably solve the problem with EEnhMatte by setting certain
parameters in QTRGui, but the VFA problem remains.

Regards

Ferdinand

Before I launch into a small amount of experimentation, a question. 
Has anyone actually successfully built and used curves using Howards
.ACV files, or modified versions thereof?  I gather Howard hasn't, and
my searching couldn't find anyone else who has either?

And yes, I read Roy's warning about the limitations of linearisation,
and the need to use a known good curve as a starting point.  I guess
it won't cost much to try it on EEnhMatte.

What would be a good starting point for VFA?

Ferdinand

Yes.  Here is an example.
http://tech.groups.yahoo.com/group/QuadtoneRIP/message/4175

On May 4, 2008, at 6:01 PM, ferdinand_paris wrote:

> Before I launch into a small amount of experimentation, a question.
> Has anyone actually successfully built and used curves using Howards
> .ACV files, or modified versions thereof?  I gather Howard hasn't, and
> my searching couldn't find anyone else who has either?
>
> And yes, I read Roy's warning about the limitations of linearisation,
> and the need to use a known good curve as a starting point.  I guess
> it won't cost much to try it on EEnhMatte.
>
> What would be a good starting point for VFA?
>
> Ferdinand

Thanks.  I learning things in leaps and bounds.  Unfortunately since
I'm trying to create curves for a 2100 I can't use yours.  However I
can use the file created by Scott Schroeder in the files section.

A couple of questions:

(i)  Did Roy ever create a way to extract the curves from a .quad
file, as it was suggested that he was going to do?

(ii) Carl - in your definition file I see that you've specified some
parameters like 

GRAY_HIGHLIGHT=4
GRAY_SHADOW=4
GRAY_OVERLAP=5
GRAY_GAMMA=1

and

TONER_HIGHLIGHT=2
TONER_SHADOW=2

Do these actually do anything when you've specified a curve?  If so,
how did you choose them?  Howard in an earlier response said that
"None of the gray curve or toner curves facilities are needed as you
can make all those adjustments with the shape of the loaded curve",
implying that they either not needed or not used.  

Sorry for all these questions, but I'm trying to limit the amount of
blind speculation and testing that I am about to do.  It pays to do a
little research and testing before starting to burn ink and paper, and
in order to minimise the loss of sanity.

Ferdinand


--- In QuadtoneRIP@yahoogroups.com, Carl Schofield <list@...> wrote:

>
> Yes.  Here is an example.
> http://tech.groups.yahoo.com/group/QuadtoneRIP/message/4175
> 
> 
> On May 4, 2008, at 6:01 PM, ferdinand_paris wrote:
> 
> > Before I launch into a small amount of experimentation, a question.
> > Has anyone actually successfully built and used curves using Howards
> > .ACV files, or modified versions thereof?  I gather Howard hasn't, and
> > my searching couldn't find anyone else who has either?
> >
> > And yes, I read Roy's warning about the limitations of linearisation,
> > and the need to use a known good curve as a starting point.  I guess
> > it won't cost much to try it on EEnhMatte.
> >
> > What would be a good starting point for VFA?
> >
> > Ferdinand
>

No, they are not used and can be deleted.  The only items needed are  
ink limits and linearization.

On May 5, 2008, at 9:45 AM, ferdinand_paris wrote:

> Thanks.  I learning things in leaps and bounds.  Unfortunately since
> I'm trying to create curves for a 2100 I can't use yours.  However I
> can use the file created by Scott Schroeder in the files section.
>
> A couple of questions:
>
> (i)  Did Roy ever create a way to extract the curves from a .quad
> file, as it was suggested that he was going to do?
>
> (ii) Carl - in your definition file I see that you've specified some
> parameters like
>
> GRAY_HIGHLIGHT=4
> GRAY_SHADOW=4
> GRAY_OVERLAP=5
> GRAY_GAMMA=1
>
> and
>
> TONER_HIGHLIGHT=2
> TONER_SHADOW=2
>
> Do these actually do anything when you've specified a curve?  If so,
> how did you choose them?  Howard in an earlier response said that
> "None of the gray curve or toner curves facilities are needed as you
> can make all those adjustments with the shape of the loaded curve",
> implying that they either not needed or not used.
>
> Sorry for all these questions, but I'm trying to limit the amount of
> blind speculation and testing that I am about to do.  It pays to do a
> little research and testing before starting to burn ink and paper, and
> in order to minimise the loss of sanity.
>
> Ferdinand
>
>
> --- In QuadtoneRIP@yahoogroups.com, Carl Schofield <list@...> wrote:
>>
>> Yes.  Here is an example.
>> http://tech.groups.yahoo.com/group/QuadtoneRIP/message/4175
>>
>>
>> On May 4, 2008, at 6:01 PM, ferdinand_paris wrote:
>>
>>> Before I launch into a small amount of experimentation, a question.
>>> Has anyone actually successfully built and used curves using Howards
>>> .ACV files, or modified versions thereof?  I gather Howard hasn't,  
>>> and
>>> my searching couldn't find anyone else who has either?
>>>
>>> And yes, I read Roy's warning about the limitations of  
>>> linearisation,
>>> and the need to use a known good curve as a starting point.  I guess
>>> it won't cost much to try it on EEnhMatte.
>>>
>>> What would be a good starting point for VFA?
>>>
>>> Ferdinand
>>

Thanks again for advice.  A few more questions if I may.

(i)  I'm hoping to produce a 1440 curve for EpEnhMatte, since the 2880
curve seems to use too much ink for this paper.  Is there any point in
trying to base a 1440 curve on an attempt to simulate a Cone-produced
2880 curve, either by using ACV files, or by entering points?  That
is, will such 2880 curves inherently make the ink limits too low for
1440?  Would I be better simply to try and create a 2880 curve and
lower the limits, based on the ink separation page?  

Perhaps another way of asking this question is whether the individual
curves that you enter for each ink are absolute or relative curves? 
Hope this makes sense.

(i)  Are there any software or Excel tricks for taking two columns of
data and producing the "xx.xxx;yy.yyy" pair that are required for
entering points in QTR?  And conversely, i.e. for taking the pairs and
producing columns in a spreadsheet?

Ferdinand

ferdinand_paris wrote:
> Thanks again for advice.  A few more questions if I may.
> 
> (i)  I'm hoping to produce a 1440 curve for EpEnhMatte, since the 2880
> curve seems to use too much ink for this paper.  Is there any point in
> trying to base a 1440 curve on an attempt to simulate a Cone-produced
> 2880 curve, either by using ACV files, or by entering points?  That
> is, will such 2880 curves inherently make the ink limits too low for
> 1440?  Would I be better simply to try and create a 2880 curve and
> lower the limits, based on the ink separation page?  
> 
> Perhaps another way of asking this question is whether the individual
> curves that you enter for each ink are absolute or relative curves? 
> Hope this makes sense.

You just need to use lower ink limits with 2880dpi, probably in the 
region of 35-60% lower.

I'm still not clear why you have rejected the "take existing curve for 
similar paper/adjust ink limit at print time (if necessary)/make ICC for 
new paper" method that Roy suggested. It sounds a lot simpler.

> (i)  Are there any software or Excel tricks for taking two columns of
> data and producing the "xx.xxx;yy.yyy" pair that are required for
> entering points in QTR?  And conversely, i.e. for taking the pairs and
> producing columns in a spreadsheet?
> 

I just copied & pasted straight from Carl's email and all the points 
appeared correctly. The only problems was that their appear to be more 
points than boxes available so I guess you'll have to stick to editing 
the text file if you want that number of points.

Howard

--- In QuadtoneRIP@yahoogroups.com, Howard Shaw <glassman@...> wrote:
> I'm still not clear why you have rejected the "take existing curve for 
> similar paper/adjust ink limit at print time (if necessary)/make
> ICC for new paper" method that Roy suggested. It sounds a lot simpler.
 
Perhaps because I misunderstood something.  I didn't quite understand
Roy's comment at the time.  I thought that you had to extract the
underlying curve shape and input it into the curve editor in order to
be able to set limits.  Reading again I understand.  Thanks.

Question - how do I judge the ink limit setting in the print dialog? 
Is this measured in the same units as ink limits in the curve creator?
 I.e. so that -10 is like 90 in the curve creator, or is it entirely
suck it and see?

It's a neat solution, assuming that I can select a sufficiently
similar paper.  I'll probably use it for VFA.  The basic shape of
Cone's curves don't seem to vary all *that* much, so I am hoping that
the linearisation works well enough.

However I am not so optimistic about using this technique for
EpEnhMatte.  The shape of the black in Cone's curve is quite different
(lower) from his other curves and I'm already having trouble with ink
overloading.  I suspect that I'll need to be able to specifically
tweak the black.  For this paper I may attempt to simulate his curves
in the curve editor.

> > Perhaps another way of asking this question is whether the individual
> > curves that you enter for each ink are absolute or relative curves? 
> 
> You just need to use lower ink limits with 2880dpi, probably in the 
> region of 35-60% lower.

This might prove to be irrelevant, but I there's something I still
don't understand, and I can't seem to describe it properly.  If I
enter a set of curves for an ink in the curve creator, either through
an AVC or by points, how do I know how to scale those curves?  I could
extract the shape of the curves from the .quad, reduce them to 50
points, but I've got to scale them somehow.  If I scale them too low,
adjusting the limits isn't going to help me, as I can't use a limit
more than 100, can I?
 
> > (i)  Are there any software or Excel tricks for taking two columns of
> > data and producing the "xx.xxx;yy.yyy" pair that are required for
> > entering points in QTR?  And conversely, i.e. for taking the pairs and
> > producing columns in a spreadsheet?

Again, I had difficulty stating my question clearly.  I wanted to take
the points from the curve creator and paste them into Excel and
manipulate the curves.  Then turn them back into points and past them
into the curve creator.  I've worked out how to split the two numbers
separated by a ";" into two separate cells and how to recombine them
using a simple formula.

Thanks for your patience.

Ferdinand

I realise that my last post was a bit long, which does not encourage
responses.  This one is shorter.

I've created a curve definition for EpEnhMatte on my K7 2100 by
mimicking the Cone supplied .quad file.  (I've uploaded the
spreadsheet that I created to do this to the Files section, in case
it's of general interest.)  I've printed out the separation page in
calibration mode with the limit at 100.

Question is, how do I, or indeed can I, set the ink limits using the
calibration printout?  It seems to that the 7 sets of 50 points that I
entered into the curve definition already have limits implicit in
them, non?   Should I solve the too-much-ink problem by setting a
black only limit, or setting a global limit lower than 100?  And can I
determine this from the calibration printout?


F_P

At last I think I've realised my problem.  EEM can't handle the volume
of ink that the Cone curves put down at 2880.  Even in the midtones
there's a fair bit of overlap between the cartridges, and the volume
of ink that the curves put down to achieve the tone is more than the
paper can handle.

I wonder if it's possible to tweak these curves for 1440, or whether
the same problem will occur there.  Perhaps the traditional QTR
approach with limited curve overlap might be the only way for EEM.  

(I know that K7 might seem overkill for EEM, but it's useful for some
purposes.)

Ferdinand

Why don't you just decrease the default ink limit in the profile to  
75, print at 2880 and re-linearize? This setting used to work fine for  
me in a 2400.  If it is still too high for your printer try a limit  
near 50.

On May 11, 2008, at 6:28 PM, ferdinand_paris wrote:

> At last I think I've realised my problem.  EEM can't handle the volume
> of ink that the Cone curves put down at 2880.  Even in the midtones
> there's a fair bit of overlap between the cartridges, and the volume
> of ink that the curves put down to achieve the tone is more than the
> paper can handle.
>
> I wonder if it's possible to tweak these curves for 1440, or whether
> the same problem will occur there.  Perhaps the traditional QTR
> approach with limited curve overlap might be the only way for EEM.
>
> (I know that K7 might seem overkill for EEM, but it's useful for some
> purposes.)
>
> Ferdinand

I would have thought that this would have reduced density across the
tonal range and made the print overall lighter if I was using a set of
curves specified by inputting points.  Is this not the case?  I guess
I'll have to try it.  I don't get the arithmetic of the relationship
between the curve definitions in the .qidf and the .quad

F_P


--- In QuadtoneRIP@yahoogroups.com, Carl Schofield <list@...> wrote:

>
> Why don't you just decrease the default ink limit in the profile to  
> 75, print at 2880 and re-linearize? This setting used to work fine for  
> me in a 2400.  If it is still too high for your printer try a limit  
> near 50.
> 
> 
> On May 11, 2008, at 6:28 PM, ferdinand_paris wrote:
> 
> > At last I think I've realised my problem.  EEM can't handle the volume
> > of ink that the Cone curves put down at 2880.