QTR-Quadtone RIP

Ah, Roy, thank you.

First, what I am trying to do, actually what I WAS trying to do:

I just got a 2400, and find the ABW function rather useful for 
routine neutral(ish), warm, etc prints. It is not as versatile as 
QTR, mind you. In search for the ultimate in Dmax on Hahnemuhle 
papers, I swapped the K cart on the 2400 with one loaded with 
NanoChrome K ink. I get Dmax in the range of 1.95 with this ink on 
HWT and HPR (takes 3 days to dry-down to this value). I started 
making QTR curves with this ink, trying to fine-tune the color hue in 
the 60-100% range, where the K ink is most prominent. While I was 
waiting for the steps to dry-down, I just printed a step wedge using 
ABW, and lo-and-behlod, it held together very well. The densities 
were off in the 60-100% range because UC MK has a Dmax only of 1.65, 
while NanoChrome hits 1.95. So I wanted to linearize this.

Of course I could have continued on with QTR, but I thought: how 
about just creating a "linearizing ACV curve" instead? I could have 
done trial-and-error using the ideal density chart. So I was looking 
for a more "rationale" method instead.

But then, my south window fade testing of the Nanochrome K showed me 
that this ink does not work well with Hahnemuhle papers, and my 
experiment is moot. The 1.95 Dmax is SO MUCH of an improvement that 
against my best judgment, I am considering using it for prints on my 
wall, but not to give away. My suspicion is that the Epson Dye K 
might be a better option.

The latter part of your message is a little more involved than my 
puny mind can grapple right now. But, I understand how I could 
convert my image to the profile created using "create-icc" utility. I 
have not gone into that functionality of QTR yet, but it sounds like 
a fun thing to play around.

See you in Martin's Exchange.

Regards.

Shilesh

--- In QuadtoneRIP@yahoogroups.com, "Roy Harrington" <roy@h...> wrote:
>
> --- In QuadtoneRIP@yahoogroups.com, "Shilesh Jani" 
<shileshjani@b...> wrote:
> >
> > I want to rationally create a PhotoShop curve to linearize 
output. Is 
> > the linearization scheme used in QTR public domain? If so, can 
someone 
> > send it to me. My measurements will be in density D
> > 
> > Thank you.
> > 
> > Shilesh
> >
> 
> Hi Shilesh,
> 
> Conceptually, linearization is pretty simple.  Some of the details 
can get involved.
> 
> Basically you have a stepwedge such as K = 0,5,10,..95,100  and all 
the measured
> densities.  Think of plotting this with (x,y) points where  x = K 
value and y = 
> measured value converted into the K value of where that density 
should occur.
> E.G.  if K=50 measures d=.75  and .75 should occur at K=65 you have 
the
> point (x,y) = (50,65).   So when you plot the points out you see a 
graph of the
> actual output versus the input.  To make a correction curve just 
swap x and y
> for all the points.  Now for an input of 65 it gets corrected to 50 
which will output
> the density .75 which is what you want for an input of 65.
> 
> Pretty simple so far, but the big question is deciding where the 
d=.75 should
> occur.  "Linearization" implies that there is a straight line 
somewhere so you can
> interpolate and find any of the point in between.   However the 
units you use make
> a big difference.  QTR uses L values as the units for 
linearization. Density values
> would not produce a straight line.  In the QTR download there's a 
file called
> ideal_density.pdf that shows typical values for various dmax's.   
QTR justs converts
> densities to L-value and then draws a straight line from L-min to L-
max.
>    formulas:  
>          Y = 10 ^ (-d)
>          L = (Y^(1/3) * 116) - 16   for Y >= 8/903.3
>          L = 903.3 * Y                    for Y <   8/903.3
> 
> Linear L-values aren't the only choice but that's what QTR uses.
> 
> -----------------
> 
> I'm not sure what you are doing, but another thing you might 
consider is using
> the QTR-Create-ICC capability.   This does all the work of creating 
the correction
> curve and putting it into an ICC profile.  Then it can be used with 
the color
> management system to not just linearize to straight-line L-values 
but actually
> linearize to the builtin grayspace of the source file and do the 
gamut mapping
> of black point compensation in perceptual intent.   This is 
somewhat more than
> your original desire but maybe you are headed that direction anyway.
> 
> Roy
>