Digital BW, The Print

> From: Tyler Boley <tyler@...>

> 
> Yes, but "linearized", or "calibrated" to a selected standard, like
> LAB, a dot gain, a gamma. I believe that was where the thread started,
> which one, why, and whether the final output is in fact progressing by
> that standard.
> 

Good.  I think where differ is simply whether we achieve the goal of "the
final output is in fact progressing by that standard."  I initially started
the debate talking about whether we achieved a linearization/calibration to
Grey Gamma 2.2.  You highlighted the fact that perhaps LAB was better and
that this was the target chosen by QTR.  You note that the choice of target
is by no means without debate. I'll take your word on that.  I am simply
suggesting we don't get either with the present method.


> ...
>> Now we come to the question of whether or not we achieve the results we
>> intend....
> snip
>> I would suggest that when someone says I want to linearize to LAB, for
>> example, what they intend to achieve is a curvature of the pixel vs
> density
>> plot that matches the curvature of LAB.
> 
> ...
> and I thought we had concluded that that is indeed not what is
> happening by the conventional use of the term.

A curvature yes.  But what curvature and does the curvature meet the target?

> 
>> ...  What this would mean is that the
>> printer would reflect a change in density for a change in pixel
> value that
>> matches LAB, how the eye sees and all that good stuff.  If it
> doesn't then
>> we have failed to meet our goal.  If you say "well who cares if it
> doesn't"
>> then I would say "why reference your target to anything then, let alone
>> something which denotes a sense of rigour.  Pick any old number or
> target -
>> what the hell you can always soft proof the file and fix it with an
> edit."
> 
> I don't think that is the case at all, there is clearly a usefulness
> to selecting a target that is also the working space, and
> "linearizing" by conventional standards referencing that target.

Good.  Let's pick one and see if it works.

QTR picked LAB - I don't think it "works".

Not because it was the wrong target.  Simply because the output has not
achieved reference to target, in my opinion.  The density curve has
curvature and hence a gamma that isn't 1, but it isn't a LAB curve.
Furthermore, each paper curve has a different curve, different gamma.  This
is unnecessary.

> Clearly the resulting print bears a strong resemblance to the monitor
> image, even without softproof.
> That the entire curve has been proportionally compressed to match end
> points clearly breaks from your particular goal, and a strict
> technical definition of linearization to the selected standard, I'm
> not arguing that.
> I'd just like to use "clearly" in a sentence one more time.
> 
LOL!  Ok.

> I understand what you are after. All I am saying is that, despite it's
> semantic inaccuracy, the conventional process works extraordinarily
> well, is miles ahead of where we were before these tools were
> available, and allows myself and others to achieve goals previously
> barely attainable.

This I don't doubt.  I am simply suggesting that we can continue to move
forward.
> 
> You are defining a different workflow and different end result, the
> conventional approach to linearization may not be suitable for that as
> you have been saying.

In essence, yes.  I am suggesting that if we did achieve
linearization/calibration to target the workflow would simplify massively.
Soft proofing, for example, would become unnecessary.

> 
> I'm sure my inconsistent use of terminology and ADD when it comes to
> math is part of the problem here, I may not be the right guy to see
> this conversation through. I suspect, though, that we are more or less
> done and the approach I use, and the one so far used by many in the
> community, is of little use to you.

It is of great use to me.  I have printed some good prints using it.  But I
am suggesting that the workflow could be dramatically easier and the result
more precise.  I need either an Eye-One or access to another person's
soft-proof files and a lot of fiddling to get a sensible result that closely
matches my pre-print-phase image.

For example, as I said, soft proofing becomes unnecessary (except to proof
hue) and as a result the reintroduction of subjective matching of print
version to proofed version.

I very much like that this forum has endeavoured to make printing decent B&W
easier for people.  This I believe is a very worthy goal. I believe, quite
simply, that there is unnecessary complexity in the current workflow and
that if we achieved calibration to target this complexity would disappear in
one stroke.

> I hope I haven't sounded like I'm trying to talk you into it, not at all.
> I'd also add, in it's conceptual defense, that the craft (not the art)
> of photography has always involved dealing with and/or matching the
> range of the real world, through many steps and materials, down to the
> range of a print. Scene to film, film to print, and now the same
> concept in the digital realm with it's different steps along the way.

Yes.  I don't debate how that there are steps in the process.  I just don't
enjoy going 2 steps forward, 1 step back!

As I have said before, I don't have the necessary programming skills to
write my own RIP.  All I can do is suggest a simpler, more efficient way of
doing things and see if it gets adopted or ask Roy, as I have done in an
offline email, for some guidance as to how I might bend QTR to try to adopt
my revised workflow.  If I can't do that then I will stick with the status
quo and keep muddling on.

Have fun but don't keep striving forward...

:-)

Steve