Digital BW, The Print

Is there an ideal, or optimal, greyscale ramp?

This is really a question about linearization.  The visual difference
between an ink density of .1 and .16 is obvious and easily seen.  The
visual difference between an ink density of 1.8 and 1.86 is not
obvious and may not even be discernable.  Is there an "optimal" ramp
rate for density to increase?

When linearizing curves with IJC/OPM there is a factor which can be
applied to change the rate at which density increases (slow increase
with small density and higher increase with increasing density). 
Something like gamma I suppose.  As I do more testing and curve making
a couple of questions have arisen.

1. Does this gamma or ramp rate change depending on Dmax?  The rate at
which denisyt changes near the Dmax point needs to be steeper with a
larger Dmax - I think?

2. Is it possible to create a gamma or ramp rate which matches a
hardware calibrated monitor at some specific gamma (1.8 or 2.2 or
whatever)?  Is that undesirable for some reason?

3. Is there a "formula" based on human perception that can be used to
determine what an optimal greyscale ramp would look like?

As always, thanks for your help while I try and nut these things out.

I suppose it's bad form to answer your own question but I found most
of what I was looking for in a long thread called "Linearization -
mathematical or perceptual?" from April of this year.


Thanks.

--- In DigitalBlackandWhiteThePrint@yahoogroups.com, "gulstenek"
<keving@d...> wrote:

> Is there an ideal, or optimal, greyscale ramp?
> 
> This is really a question about linearization.  The visual difference
> between an ink density of .1 and .16 is obvious and easily seen.  The
> visual difference between an ink density of 1.8 and 1.86 is not
> obvious and may not even be discernable.  Is there an "optimal" ramp
> rate for density to increase?
> 
> When linearizing curves with IJC/OPM there is a factor which can be
> applied to change the rate at which density increases (slow increase
> with small density and higher increase with increasing density). 
> Something like gamma I suppose.  As I do more testing and curve making
> a couple of questions have arisen.
> 
> 1. Does this gamma or ramp rate change depending on Dmax?  The rate at
> which denisyt changes near the Dmax point needs to be steeper with a
> larger Dmax - I think?
> 
> 2. Is it possible to create a gamma or ramp rate which matches a
> hardware calibrated monitor at some specific gamma (1.8 or 2.2 or
> whatever)?  Is that undesirable for some reason?
> 
> 3. Is there a "formula" based on human perception that can be used to
> determine what an optimal greyscale ramp would look like?
> 
> As always, thanks for your help while I try and nut these things out.

Kevin,

looks like you got your answers already. But here is some food for thought:


 
> 1. Does this gamma or ramp rate change depending on Dmax?  The rate at
> which denisyt changes near the Dmax point needs to be steeper with a
> larger Dmax - I think?


That's the reason IJC made the linearization gama variable. It's an easy way
to compensate for high dmax media and inks. With profiles that result in
dmax around 1.6 the slider is set to -6. However if you profile something
with a dmax of 2.4, you will inevitably have an overall darker scale if you
leave the slider at -6.  As you go down to -11 or -12, you are in essence
lightening the mid tones. This is much like lightening the mid tones in Photoshop: 
it decreases contrast in the highlights and increases it in the shadows (as you
note above). 



> 
> 2. Is it possible to create a gamma or ramp rate which matches a
> hardware calibrated monitor at some specific gamma (1.8 or 2.2 or
> whatever)?  Is that undesirable for some reason?

I'm sure it is possible, and rather desirable, but that's not how IJC is
set up.  It so happpens that with the Lin. slider set to -6 and dmaxes
in the 1.6-1.8 range it comes pretty close to gamma 1.8 (except the shadows
look a little too light). To really match to a monitor you'd have to
linearize by hand (trial and error) with the present version. If we are
lucky enough to have a future version where you can control the
aims, it would be a matter of writing a set of aims of your choice
to which IJC will linearize your profile. That would be far more
desirable than the simple pull of the midtone we have now.



> 
> 3. Is there a "formula" based on human perception that can be used to
> determine what an optimal greyscale ramp would look like?

There may be, but I don't know it. Kodak literature has densities for
optimal grayscales as they apply to their instruments and that became
the basis for the aims in IJC. But I am sure the committees who wrote
ICC specs and gammas etc for color management have arrived at
optimal grayscales in lots of other ways. There is also the old prepress
school of dot percent where 50% means half the dot is black and
half is paper-white. 


Hope this is helpful....



Antonis

Antonis-

Thanks for another thoughtful answer.
  
> > 1. Does this gamma or ramp rate change depending on Dmax?  The rate at
> > which denisyt changes near the Dmax point needs to be steeper with a
> > larger Dmax - I think?
> 
> 
> That's the reason IJC made the linearization gama variable. It's an
easy way
> to compensate for high dmax media and inks. With profiles that result in
> dmax around 1.6 the slider is set to -6. However if you profile
something
> with a dmax of 2.4, you will inevitably have an overall darker scale
if you
> leave the slider at -6.  As you go down to -11 or -12, you are in
essence
> lightening the mid tones. This is much like lightening the mid tones
in Photoshop: 
> it decreases contrast in the highlights and increases it in the
shadows (as you
> note above). 
> 

Thanks for the guideline numbers.  I had adopted -6 for the HPR and my
3000 with piezotones.  I got a 2200 the other day and looking at the
curves you and others had produced for ISP I wanted to understand
where the slider should be for larger DMax combinations and what the
slider actually meant.

I made some graphs yesterday and interestingly (to me anyway), when
the slider is at -6 for Dmax of 1.65(ish) this makes the luminance
values more or less a straight line. When the DMax is 2.3(ish) the
luminance is a linear straight line when the slider is at -10.

In the end, as long as you're close and consistenet I guess it's still
a 'salt to taste' issue.


> > 
> > 2. Is it possible to create a gamma or ramp rate which matches a
> > hardware calibrated monitor at some specific gamma (1.8 or 2.2 or
> > whatever)?  Is that undesirable for some reason?
> 
> I'm sure it is possible, and rather desirable, but that's not how IJC is
> set up.  It so happpens that with the Lin. slider set to -6 and dmaxes
> in the 1.6-1.8 range it comes pretty close to gamma 1.8 (except the
shadows
> look a little too light). To really match to a monitor you'd have to
> linearize by hand (trial and error) with the present version. If we are
> lucky enough to have a future version where you can control the
> aims, it would be a matter of writing a set of aims of your choice
> to which IJC will linearize your profile. That would be far more
> desirable than the simple pull of the midtone we have now.
>

Hmm.. I've been calibrating to a gamma of 2.2 and using a 2.2 working
space.  If it is closer to a gamma of 1.8 maybe it's worth the effort
to change for future images.  I know it's just fiddling with the fine
tuning dial but ...

> 
> > 
> > 3. Is there a "formula" based on human perception that can be used to
> > determine what an optimal greyscale ramp would look like?
> 
> There may be, but I don't know it. Kodak literature has densities for
> optimal grayscales as they apply to their instruments and that became
> the basis for the aims in IJC. But I am sure the committees who wrote
> ICC specs and gammas etc for color management have arrived at
> optimal grayscales in lots of other ways. There is also the old prepress
> school of dot percent where 50% means half the dot is black and
> half is paper-white. 
> 
> 
> Hope this is helpful....

Yes, thanks.

> I made some graphs yesterday and interestingly (to me anyway), when
> the slider is at -6 for Dmax of 1.65(ish) this makes the luminance
> values more or less a straight line. When the DMax is 2.3(ish) the
> luminance is a linear straight line when the slider is at -10.



That pretty much jives with my experience as well. Were these Lab graphs?



> 
> In the end, as long as you're close and consistenet I guess it's still
> a 'salt to taste' issue.


Yes, but if you follow ICC standards
of workflow and proofing, you would want these scales to be
consistent with those standards as well - and that's a bit trickier.




> Hmm.. I've been calibrating to a gamma of 2.2 and using a 2.2 working
> space.  If it is closer to a gamma of 1.8 maybe it's worth the effort
> to change for future images.


Do you mean you switch the gamma at OPM to 2.2 as well?
Try it and see what works best. Currently I don't think OPM reads
embedded profiles. Take a 2.2 image and print it both ways:
with OPM set to 1.8 and then again set to 2.2. Also try the
same with a 1.8 image. It should be that you get a match
when the image gamma matches the OPM setting.

You shouldn't need to change anything in your workflow. If you
are set on 2.2 in your Photoshop environment, all you have to
do is remember to set OPM to 2.2 (it always defaults to 1.8 on
launch).  
For myself, I find that any setting that doesn't stay
on relaunch is one more thing to forget when printing -
so I try to save a "Job" if at all possible.



Antonis

Antonis-

> That pretty much jives with my experience as well. Were these Lab
graphs?

Yes - sort of.  I derived the lab L from the X-rite 810 visual density
readings using: d=-log(Y) and Y=((L+16)/116)^3

 
 
> 
> Do you mean you switch the gamma at OPM to 2.2 as well?
> Try it and see what works best. Currently I don't think OPM reads
> embedded profiles. Take a 2.2 image and print it both ways:
> with OPM set to 1.8 and then again set to 2.2. Also try the
> same with a 1.8 image. It should be that you get a match
> when the image gamma matches the OPM setting.

Yes, I set the gamma in OPM to 2.2 - except when I forget.  Jobs
sounds like a good idea :)

My monitor is hardware calibrated to 2.2 and my working space is gamma
2.2.  I leave the files untagged and use a custom dot gain for soft
proofing.  I make all the changes to the file while it's in the
working space, save it as a master, duplicate it, apply the soft proof
and make one more layer adjustment until it matches the master.  I
print it with OPM and a gamma of 2.2 and then make any little local
adjustments based on what the print looks like not the screen.


It may be a little cumbersome but it seems to work well.

If there was a shortcut to go directly from the working space ( and
the master file )to the print that would be cool.  I haven't figured
out a way to do that. 

--
Kevin