Digital BW, The Print

Tyler

Yes I agree that there are many ways to replicate WYSIWYG.  But this
proposed workflow is not intended to revolve solely around that.  It is
intended to more clearly and explicitly relate the tonal range of the
printer into the final steps of the workflow process.

The big problem with today's setup is that there is a hidden transfer
function (a remapping of pixel and their density/luminance values) from
workspace to print space which is HIDDEN from the view of the user - it is
imbedded in the RIP code and curve generation process.  It does not conform
with the user's workspace.  What's more, the nature of this transfer
function - it's end points, representation of mid grey, and all other points
between - is different for every paper and ink combination!  Move from a
curve for matt paper such as EEM to a glossy "curve" on glossy paper and the
way all your mid tones are reproduced is different.  A PS adjustment curve
for one will not give the same result for another.  Over the last 2 years I
have witnessed hundreds if not thousands of posts on this forum which are
the symptoms of this problem.  People get their prints looking nice on
screen using the full tonal range available to them and then send it to the
printer without any understanding of the tonal range limitations of the
paper/ink combination they are using nor the way the printer will remap
their original (file's) tonal range to fit the narrower one.

Both the current methodology and the one I am proposing may get you to the
same point in the end, but I am reminded of Ansel Adams' remarks in Chapter
3 Exposure in his "The Ansel Adams Photography Series Book 2: The Negative":

"I can recall seeing Edward Weston....using his meter in rather unorthodox
ways. He would point it in several directions, take a reading from each, and
fiddle with the dial with a thoughtful expression.  "It says one-quarter
second at f32, I'll give one second."  His approach was empirical, based on
long experience combined with very deep sensitivity and intuition, and his
extraordinary results speak for themselves.  My own approach relies on
experience and intuition for the visualization of the image, but I prefer a
more methodical system for executing the visualized photograph."

Now I am not suggesting that the current setup isn't methodical but I am
suggesting that the current remapping of tonal range from the very broad
workspace to the narrow print space is not as clear nor intuitive as it
could very easily be.

(In all this discussion I will try to stick with a normalised pixel value
range of 0 - black - to 1 - white.  Things can get confusing because PS
levels uses a range of 0 (black) to 255 (white); PS curves use a range 0
(white) to 100 (black) or 0 (black) to 255 (white) depending on which way
they are set; and we print step wedges from 0 (white) to x (black) depending
on the number of steps!)

A good exercise is to plot the tonal response curve of gamma 2.2 (our
workspace) and overlay on top of that the tonal response curve of, say, a
QTR EEM ink curve.  By tonal response curve I mean log10(pixel value) on the
x axis and log10(luminance) [or -Density] on the y axis.  It is readily
clear that never the two shall meet.

Yet a printer/paper/ink combination has a natural tonal response curve which
is exactly analogous to photographic film or photographic paper.  At a
certain point, dmax of around 1.65 for EEM, we can't get it any darker by
throwing any more ink at the page - much the same way that we can't get a
photographic paper any darker by throwing more light through the NEGATIVE at
it - and at a certain point, around dMin of 0.04, we can't get it any whiter
because we have hit paper white.  The file pixel value of these two
inflexion points are readily calculated.

If we follow my proposed methodology, we align the two inflexion points with
their corresponding gamma 2.2 pixel values, anything outside this range
would "flat line" at the respective limit, and the points in between would
be "linearized" to a gamma of 2.2 (the same as our workspace).  Now plot
this tonal response curve.  Looking at the chart we would see a line
beginning flat at -dMax, inflecting at log10(45/255) where the printer's
tonal range begins and it can reproduce changes in density, travelling at a
slope of 2.2 to the next inflection point log10(243/255) and then
horizontal.  This curve should be readily identifiable with a photographic
paper curve.  It is also very intuitive.  The upward sloping portion of the
curve would overlay that of gamma 2.2 and would represent the "dynamic
range" (to use an Ansel Adams' Zone term) of the print space.

(Take a look at http://www.normankoren.com/makingfineprints1A.html )

Now turn to PS.  You've got the image looking great on the screen in the
very broad workspace Gray Gamma 2.2. (Actually not quite because our screen
also flat lines at some point in the deep black but this is so far outside
of the tonal range of the print space that we need not worry about it.)

Guys like Ansel were very conscious of the tonal range of their print space
all the way through there workflow.  They knew that they had to work to a
narrower range than the range of the scene.  After all, in most cases, the
tonal range of the scene > tonal range of the capture medium (film or
digital) > tonal range of the print space.  Tonal range compression
(principally through the use of filters) was necessary else clipping was a
reality.  The natural result was some mix of the two.  We are prepared to
sacrifice some loss of detail in the shadows (clipping) so long as the
important parts of the image are well in view.  The really cool thing about
digital is that we have much greater flexibility for managing this issue
(and for changing the setting of "Zone V" and hence we can with a digital
camera "expose to the right").

I would suggest that many digital photographers today are much less
conscious of the reduced tonal range of their print space at the time of
printing.  How many know how many f-stops is inherent in a range of density
from 0.04 to 1.65 or what pixel values can be faithfully reproduced with
this range?  This is manifested in all the complaints of my images print
lighter than on screen, they look flat, how do I get a good black etc etc.
Most just send their nice looking image to the printer and then wonder what
happened.  Gradually they learn to fiddle with S curves in PS and perhaps
even to use soft proof techniques.  But even a soft proof technique is a
very fiddly exercise - trying to realign the proofed image's relative tonal
situation to the un-proofed original. The curves are not conducted within
the readily identifiable pixel value range of the printer (those pixel
values that can be reproduced by the printer) and it involves second
guessing the other hidden transfer function imbedded in the RIP.

If on the other hand, the printer RIP is set up as I am proposing, I know
which pixels in my image will be clipped by the printer to either paper
white or dMax IF I SIMPLY SEND THE FILE AS IS.  Furthermore, and
importantly, I know that all OTHER pixels WILL be accurately reproduced.
(For Zone people, my Zone V will be faithfully where I set it.)  Take a look
at the PS curve that would adjust the file so that it sent only those pixels
that were within the dynamic range of the printer.  First of all it would
look like the tonal response curve of the print space (with just one
difference: the slope of the diagonal portion is 1:1 rather than 2.2:1).  It
also says that there is a 1:1 mapping of points, ie NO CHANGE, that sit
within the tonal range of the print space. Nice and intuitive.  Most likely,
we will then make a decision to not clip abruptly but rather smooth at the
extremes the adjustment in image tonal range required to fit the printer
tonal range - and so the PS curve is modified to some form of "S" curve but
the anchor points representing the two ends of the printer tonal range
remain in view.  (Furthermore, we can choose some clipping if we so desire.)
The PS curve in this case is very intuitive and we know that what we have in
this curve represents the ENTIRE tonal transformation that will take place -
nothing is hidden.

(Run through what would happen if we decided to print the same image on a
printer or paper and ink with greater tonal range.  We would extend the PS
curve's end points showing a wider dynamic range. Again very intuitive.)

Hopefully you can now see why I think this methodology is much more
transparent, intuitive, visual and, in the end, easier to comprehend.  To be
honest I am surprised that printer RIPs didn't evolve this was in the first
place because it is so analogous to the darkroom world.  If one were to
print a 101 step wedge (from 0 black to 1 white!), measure the densities and
plot the tonal response curve, a darkroom guy would immediately recognize
the graph!

As I commented to Roy last night, I do not have the programming skills to
write a RIP or modify one like QTR to put this to a practical demonstration.
But I do believe that this is a better basis for the work.

Food for thought over the weekend!!

:-)

Steve