Yahoo Groups archive

QTR-Quadtone RIP

Index last updated: 2026-04-28 23:12 UTC

Thread

Linearization - is the the most control available?

Linearization - is the the most control available?

2007-08-30 by spsguru

Hi everyone

Like many I am trying to create digital negatives using QTR on a 3800.
Well I am 22 test prints down the road and not able to get the control
I need. I get either print highlights good and blocked up shadows or
vise versa. Using the usual tools (gamma, shadow, highlight)only make
it worse.
Curves have also been unsuccessful using the method described in Ron
Reeder's - which is a GREAT resource I have just not been successful.

so I went to try and linearize. 

I read 21 steps with an xrite 810. and entered those readings into the
linearization line. Still no success. It ended up blocking up the
highlights and shadows with thinning out of the midtones.

Yet I feel like if I knew how to control the values in the linearize
line that would yield ultimate control of the ink densities.

So here are my questions - 

1. I have so many negatives now that I can read the value I want via
UV so I know exactly what the negative value should be for each
printed value.
---Is there a way to use linearize to get a specific ink output for
each value? In other words - I I know that I need UV of 1.5 for a 50%
grey - can I get use a value in linear to lay down ink for 1.5 at 50%?

2. I entered in the values as read by reflection densitometer into the
linearize curve - but the output was no where near linear - can
someone explain what linearize is really doing so maybe I can figure
out what direction to tweak the values to make the curve move correctly.

thank you for any help
Sean

Re: Linearization - is the the most control available?

2007-08-31 by clayharmon47

The answer is that it is very doable. And your first one is the hardest. Basically what you 
are outlining is the pretty much the right approach: You are telling the printer to use a 
specific ink percentage at a specific image percentage.

First, make sure your photoshop color settings are consistent and stay that way. I use 
gamma 2.2 for instance. Save the color settings file as 'digineg' or something. Also make 
sure you now the minimum printing time to get solid black through your inkjet substrate. 
You must be able to control the printing times and printing controls in genreral very 
accurately, or you will be attempting to linearize a moving target. I think this is where 
people can have a lot of problems with any inkjet negative method. Consistency is 
essential.

Make your basic ink profile that will deliver just about the maximum density you need for 
your proceess. I do palladium, so my basic profile produces a UV density of about 2.9.

In the linearization tab in QTRgui, or in the proper place in the .qidf file, set the basic 
linearization to the following:

"0;0 100:100" 

This last step is necessary to give you the ink mix that will be used for the rest of your 
linearization exercise. Don't know why, but you get different results if you leave it blank 
on your first pass. 


Invert the step tablet and use some text to identify it as Trial one or something similar. 
You will go crazy remembering which is which otherwise

Now print out your step tablet and then make a print in the process you have chosen.

With your reflection densitometer, measure the printed tablet. A 50% black should be a 
reflection density of about 0.48, assuming that your dmax is 1.35 and you are printing on 
a typical watercolor paper. Find the square that is closest to 0.48. It probably won't be the 
50% square. Say it is the 40% square. Now hold this up to your test strip on screen and see 
if you agree. The 0.48 number is a theoretical number from a function called the Yule 
Neilsen formula which equates dot percentage to reflection density. Like all true empirical 
formulas, there are some fudge factors in it, though, and you need to determine if it is 
giving you the right result.

This will anchor the middle of your curve. So the new linearization profile would  now 
read:

"0;0 50;40 100;100"

The middle of your curve is now anchored. Now go back and do it for your quarter tones 
at 25% and 75%. The target values should be about .21 and .83 respectively. Again, this is 
if you are printing a process with a 1.35 dmax on matte paper. And also once again, 
visually look to determine if this looks right when you compare the printed step tablet to 
the version on your screen. Say the percentage squares that do this are 35% and 65%. So 
now your linearization numbers are:

"0;0 25;35 50;40 75;65 0;100"

You can see where this is going. If you get close on this printing, it is a small mattter to 
print one more time and smooth your shadow and highlight areas by inserting more 
points. 

I have successfully linearized the 2200, 2400 and 7800 for palladium printing using this 
approach. It took about 16 pairs of input;output values to linearize the 2200 for instance. 
Patience and consistency will get you there very quickly. Good luck. I can send you a 
spreadsheet that has the Yule Neilsen formula imbedded in it if it will help.

Clay



--- In QuadtoneRIP@yahoogroups.com, "spsguru" <spsguru@...> wrote:
Show quoted textHide quoted text
>
> Hi everyone
> 
> Like many I am trying to create digital negatives using QTR on a 3800.
> Well I am 22 test prints down the road and not able to get the control
> I need. I get either print highlights good and blocked up shadows or
> vise versa. Using the usual tools (gamma, shadow, highlight)only make
> it worse.
> Curves have also been unsuccessful using the method described in Ron
> Reeder's - which is a GREAT resource I have just not been successful.
> 
> so I went to try and linearize. 
> 
> I read 21 steps with an xrite 810. and entered those readings into the
> linearization line. Still no success. It ended up blocking up the
> highlights and shadows with thinning out of the midtones.
> 
> Yet I feel like if I knew how to control the values in the linearize
> line that would yield ultimate control of the ink densities.
> 
> So here are my questions - 
> 
> 1. I have so many negatives now that I can read the value I want via
> UV so I know exactly what the negative value should be for each
> printed value.
> ---Is there a way to use linearize to get a specific ink output for
> each value? In other words - I I know that I need UV of 1.5 for a 50%
> grey - can I get use a value in linear to lay down ink for 1.5 at 50%?
> 
> 2. I entered in the values as read by reflection densitometer into the
> linearize curve - but the output was no where near linear - can
> someone explain what linearize is really doing so maybe I can figure
> out what direction to tweak the values to make the curve move correctly.
> 
> thank you for any help
> Sean
>

Re: Linearization - is the the most control available?

2007-08-31 by clayharmon47

One thing I forgot in my previous post - too little coffee, I guess - is that you will need to 
conceptually keep clear the difference between print % ink and negative % ink, since they 
are opposites. I will clarify my previous post this evening, but I think you can grasp that a 
high negative ink density will produce a low print density and vice versa. So it is necessary 
to correct  the ink % that the printer is laying down for the negative - not a print like the 
normal QTR process...

--- In QuadtoneRIP@yahoogroups.com, "spsguru" <spsguru@...> wrote:
Show quoted textHide quoted text
>
> Hi everyone
> 
> Like many I am trying to create digital negatives using QTR on a 3800.
> Well I am 22 test prints down the road and not able to get the control
> I need. I get either print highlights good and blocked up shadows or
> vise versa. Using the usual tools (gamma, shadow, highlight)only make
> it worse.
> Curves have also been unsuccessful using the method described in Ron
> Reeder's - which is a GREAT resource I have just not been successful.
> 
> so I went to try and linearize. 
> 
> I read 21 steps with an xrite 810. and entered those readings into the
> linearization line. Still no success. It ended up blocking up the
> highlights and shadows with thinning out of the midtones.
> 
> Yet I feel like if I knew how to control the values in the linearize
> line that would yield ultimate control of the ink densities.
> 
> So here are my questions - 
> 
> 1. I have so many negatives now that I can read the value I want via
> UV so I know exactly what the negative value should be for each
> printed value.
> ---Is there a way to use linearize to get a specific ink output for
> each value? In other words - I I know that I need UV of 1.5 for a 50%
> grey - can I get use a value in linear to lay down ink for 1.5 at 50%?
> 
> 2. I entered in the values as read by reflection densitometer into the
> linearize curve - but the output was no where near linear - can
> someone explain what linearize is really doing so maybe I can figure
> out what direction to tweak the values to make the curve move correctly.
> 
> thank you for any help
> Sean
>

Re:Linearization - is the the most control available?

2007-08-31 by Michael Mutmansky

Sean,

You cannot use the linearize tab when doing digital negatives, so if that's
what you are doing, it won't work, as the negative is the opposite of the
final result, and the formulas for linearization are built around the
positive, not the negative.

If you are attempting to linearize in the grey tab, you have to be very
careful about your math, otherwise it is possible to accidentally invert the
math, and double the nonlinearity, making the result even more nonlinear.

First, start with this in the adjustment/linearization data window in the
Grey Curve tab:

"0;0 100;100"

This linearizes the curve and starts you with what should be a linear
starting point for making adjustments.

Then, do a test stepwedge.  I would recommend a 51 step tablet, rather than
21, as the steps can be somewhat large for the curve in the highlight
otherwise.

From here, you should be able to follow the Reeder approach very precisely,
and you will get a reasonable starting point for more precise final
adjustments.

The trick is to take all the data readings but don't just apply them, think
about what the mean while doing it.  That way, when you are doing the
inversions, etc. you should be able tell if you are making the adjustment in
the right direction.

For example, lets suppose the midtones are too dark, and we have only three
steps to the test; 0, 50, 100.  Also, the endpoints are perfect, so no
adjustment will be done there.

If the midtones are too dark in the print, that means that not enough
density is being put down on the negative, so too much light is passing
through to the sensitized paper.  So, the mid-value needs to be set to a
higher value than it is now.

Using the Reeder approach, you would subtract the high values from the low
value to obtain the range, and then subtract the mid value and the low value
from the high to get the middle value within the range.  Lets say that the
50 value is producing a 70% print density and we want it to be 50%.

So when the printer is being sent a 50% image value, it is resulting in a
70% print value.  To lighten the print, we need to darken the negative, so
the value that we need to be in the linearization line at the 50 point will
need to be a lower number than 50 to get it to be darker.

If everything worked in a linear fashion, the proper value would be 30, so
the adjustment curve would look like this:

"0;0 50;30 100;100"

-----

Here's one of the real difficulties of working with QTR...

Image values are from 0-255 from black to light OR from 100 to 0 from black
to white depending on what color space you are working in.

QTR works from 100 to 0 for black to white in the curve creator (or 255 to
0)  with QTR translating 0 to mean absolutely no ink being laid down, and
100/255 being all the ink going down.

When working on a negative, however, these values flip, so 100/255 means
absolute paper white, and 0 means the darkest black you can produce
(assuming you have enough density to produce paper white with your process).

So there are a series if math inversions that have to be made to keep the
math working properly, hence the reason Reeder does those inversions to the
charts.

This is complicated by the Yuill-Nielson equations for visual linearity in
the print, and nonlinearity in the inks and receiving film, and lastly, the
response curve of the process you are using.  This is the reason the Reeder
approach will get you close to a linear result but not give you an
absolutely linear response even after a few iterations.

Regardless, once you have it working, you will find that you can produce a
negative that is better than can be produced in any other method that I have
seen or used, so it is worth the effort.


---Michael

Re:Linearization - is the the most control available?

2007-08-31 by Michael Mutmansky

Sean,

And in response to your initial questions, you cannot do anything linear.
Nothing in this process acts in a truly linear fashion. In my experience.
That's why you'll have to do some testing and a bit of iteration.

You cannot use the linearize function to produce a resultant UV density
except through a back-door approach of testing, and then taking the results
and applying them to the file as an adjustment curve.  That will get you
pretty close, though.

You can't use the reflected density values directly.  You need to go through
the inversions and the Y-N formula to produce an adjustment curve that will
properly linearize the negative.


---Michael

Re:Linearization - is the the most control available?

2007-08-31 by clayharmon47

Sean,


What Michael said. He knows this stuff better than anyone, I think.

I didn't realize that what you are attempting to do is use the 
linearization approach like you would for prints. As he said, you 
have to keep straight what you are trying to linearize - and that is 
you are trying to match a specific %gray on screen to the same %gray 
on the final print. The negative is the go-between, and the it is 
essentially 'backwards' from the print. More negative density equates 
to less print density and vice-versa. So if your print is too dark at 
a specific density on screen, you need to ADD density to the negative.

It is not that easy, but once done, gives beautiful results.

I'm sorry I forgot to put that essential piece of information in my 
original post, but except for that, it does describe the basic steps 
you will need to go through. 

Clay


--- In QuadtoneRIP@yahoogroups.com, Michael Mutmansky <michael@...> 
wrote:
>
> Sean,
> 
> And in response to your initial questions, you cannot do anything 
linear.
> Nothing in this process acts in a truly linear fashion. In my 
experience.
> That's why you'll have to do some testing and a bit of iteration.
> 
> You cannot use the linearize function to produce a resultant UV 
density
> except through a back-door approach of testing, and then taking the 
results
> and applying them to the file as an adjustment curve.  That will 
get you
> pretty close, though.
> 
> You can't use the reflected density values directly.  You need to 
go through
> the inversions and the Y-N formula to produce an adjustment curve 
that will
Show quoted textHide quoted text
> properly linearize the negative.
> 
> 
> ---Michael
>

Re:Linearization - is the the most control available?

2007-09-05 by spsguru

Michael - This is a wonderful response - I am still trying to digest
it all. But I am printing a new wedge and will be then working on the
curve values so I will be coming back here with a progress report.
Thank you for the help and the advice to avoid the linearization and
stay with the grey curve
Sean

--- In QuadtoneRIP@yahoogroups.com, Michael Mutmansky <michael@...> wrote:
>
> Sean,
> 
> You cannot use the linearize tab when doing digital negatives, so if
that's
> what you are doing, it won't work, as the negative is the opposite
of the
> final result, and the formulas for linearization are built around the
> positive, not the negative.
> 
> If you are attempting to linearize in the grey tab, you have to be very
> careful about your math, otherwise it is possible to accidentally
invert the
> math, and double the nonlinearity, making the result even more
nonlinear.
> 
> First, start with this in the adjustment/linearization data window
in the
> Grey Curve tab:
> 
> "0;0 100;100"
> 
> This linearizes the curve and starts you with what should be a linear
> starting point for making adjustments.
> 
> Then, do a test stepwedge.  I would recommend a 51 step tablet,
rather than
> 21, as the steps can be somewhat large for the curve in the highlight
> otherwise.
> 
> From here, you should be able to follow the Reeder approach very
precisely,
> and you will get a reasonable starting point for more precise final
> adjustments.
> 
> The trick is to take all the data readings but don't just apply
them, think
> about what the mean while doing it.  That way, when you are doing the
> inversions, etc. you should be able tell if you are making the
adjustment in
> the right direction.
> 
> For example, lets suppose the midtones are too dark, and we have
only three
> steps to the test; 0, 50, 100.  Also, the endpoints are perfect, so no
> adjustment will be done there.
> 
> If the midtones are too dark in the print, that means that not enough
> density is being put down on the negative, so too much light is passing
> through to the sensitized paper.  So, the mid-value needs to be set to a
> higher value than it is now.
> 
> Using the Reeder approach, you would subtract the high values from
the low
> value to obtain the range, and then subtract the mid value and the
low value
> from the high to get the middle value within the range.  Lets say
that the
> 50 value is producing a 70% print density and we want it to be 50%.
> 
> So when the printer is being sent a 50% image value, it is resulting
in a
> 70% print value.  To lighten the print, we need to darken the
negative, so
> the value that we need to be in the linearization line at the 50
point will
> need to be a lower number than 50 to get it to be darker.
> 
> If everything worked in a linear fashion, the proper value would be
30, so
> the adjustment curve would look like this:
> 
> "0;0 50;30 100;100"
> 
> -----
> 
> Here's one of the real difficulties of working with QTR...
> 
> Image values are from 0-255 from black to light OR from 100 to 0
from black
> to white depending on what color space you are working in.
> 
> QTR works from 100 to 0 for black to white in the curve creator (or
255 to
> 0)  with QTR translating 0 to mean absolutely no ink being laid
down, and
> 100/255 being all the ink going down.
> 
> When working on a negative, however, these values flip, so 100/255 means
> absolute paper white, and 0 means the darkest black you can produce
> (assuming you have enough density to produce paper white with your
process).
> 
> So there are a series if math inversions that have to be made to
keep the
> math working properly, hence the reason Reeder does those inversions
to the
> charts.
> 
> This is complicated by the Yuill-Nielson equations for visual
linearity in
> the print, and nonlinearity in the inks and receiving film, and
lastly, the
> response curve of the process you are using.  This is the reason the
Reeder
> approach will get you close to a linear result but not give you an
> absolutely linear response even after a few iterations.
> 
> Regardless, once you have it working, you will find that you can
produce a
> negative that is better than can be produced in any other method
that I have
Show quoted textHide quoted text
> seen or used, so it is worth the effort.
> 
> 
> ---Michael
>

Re:Linearization - is the the most control available?

2007-09-05 by spsguru

Clay - thank you too for the help. I will be updating this thread with
my progess
Sean

--- In QuadtoneRIP@yahoogroups.com, "clayharmon47" <clay@...> wrote:
Show quoted textHide quoted text
>
> Sean,
> 
> 
> What Michael said. He knows this stuff better than anyone, I think.
> 
> I didn't realize that what you are attempting to do is use the 
> linearization approach like you would for prints. As he said, you 
> have to keep straight what you are trying to linearize - and that is 
> you are trying to match a specific %gray on screen to the same %gray 
> on the final print. The negative is the go-between, and the it is 
> essentially 'backwards' from the print. More negative density equates 
> to less print density and vice-versa. So if your print is too dark at 
> a specific density on screen, you need to ADD density to the negative.
> 
> It is not that easy, but once done, gives beautiful results.
> 
> I'm sorry I forgot to put that essential piece of information in my 
> original post, but except for that, it does describe the basic steps 
> you will need to go through. 
> 
> Clay
> 
> 
> --- In QuadtoneRIP@yahoogroups.com, Michael Mutmansky <michael@> 
> wrote:
> >
> > Sean,
> > 
> > And in response to your initial questions, you cannot do anything 
> linear.
> > Nothing in this process acts in a truly linear fashion. In my 
> experience.
> > That's why you'll have to do some testing and a bit of iteration.
> > 
> > You cannot use the linearize function to produce a resultant UV 
> density
> > except through a back-door approach of testing, and then taking the 
> results
> > and applying them to the file as an adjustment curve.  That will 
> get you
> > pretty close, though.
> > 
> > You can't use the reflected density values directly.  You need to 
> go through
> > the inversions and the Y-N formula to produce an adjustment curve 
> that will
> > properly linearize the negative.
> > 
> > 
> > ---Michael
> >
>

Move to quarantaine

This moves the raw source file on disk only. The archive index is not changed automatically, so you still need to run a manual refresh afterward.