Digital BW, The Print

Hi Clayton

First of all let me say that I don't "miss the human side".  I simply try to
understand the tools that I use.  Trial and error without understanding is
of little interest to me.  Learning is.  An artist, like anyone else, needs
to learn to understand and use the tools of their trade to the best of their
ability.  A painter must understand how to turn a brush and mix paint.  A
photographer must know how to use their camera.  A digital photographer must
learn and understand the camera, their editing programme and, in our case,
their printer.  Ansel Adams was a strong proponent of this technical
understanding.  Why?  Not because it made him better at previsualisation but
because it allowed him to better effect his previsualisation.  So it pays to
learn what you can and understand the "why" as much as possible.

This thread started by addressing one of the big issues in digital
photography:  matching monitor and print.  I am simply making a few basic
points:

1. start with a well-profiled monitor (as well profiled as you can get) and
don't be tempted to "dumb down the monitor to match the print"

2.  Understand why a print will not match the image on screen without the
aid of a soft proof

3.  Understand that if the print space (the colour generated for each and
every possible pixel value) is closer to the image's colour space (the ICC
profile with which the file is tagged) then the degree of mismatch will be
less.  The corollary to this is if I can do something to make the two spaces
closer then I am achieving something with regard to the issue.

4.  But you WILL need a soft proof if you want to come close to matching
monitor and print so start thinking about how you might get/make one.  In
other words, pursuing 3 doesn't get you very far.






> From: Clayton Jones <cj@...>

> 
> I think you are both right.  BO only comes into play when its output
> time, but an initial color to BW conversion does map the image (if
> that is the correct term) to the default gray workspace.  So the
> initial screen images for DG20 and GG2.2 conversions of the same color
> image, for example, will look identical, but the initial prints will
> be very different (assuming the printer profile is SAS).  However,
> once converted and changed to grayscale, any profile may then be
> assigned to an image, and only the screen image will change, not the
> print (again, assuming the printer profile remains SAS).

Clayton, I am sorry but this is simply not true.  A simple test proves this.
Print a step wedge where you have assigned Dot Gain 20% and then print a
step wedge where you have assigned Gray Gamma 2.2 (choose whatever printer
driver gamma setting you like so long as it is consistent).  Measure the
results and (unsurprisingly to me at least) you will find they are very
different.  I just did this.


> Beyond that
> I'm not sure what happens when you check the Black Only box, but your
> description may explain what I've been experiencing. Changing the
> driver gamma setting does change the print.

And of course changing the gamma setting in the printer driver changes the
results.



But now for some good news for you....

I just ran a quick test.  I printed a step wedge tagged/assigned as Gray
Gamma 2.2 with a printer driver setting of 1.8 and then 2.2.  I did the same
with a step wedge assigned Dot Gain 20%.  I then measured the results and
compared them with the file values and added up the sum of the errors
(absolute errors).  This only cost one piece of paper and about 20mins.  <g>
<g>

Previously I had been talking in terms of Gray Gamma 2.2 as the starting
workspace - recall this conversation began with Paul's workflow and his
comments about the pervasiveness of Adobe RGB as a colour workspace and Gray
Gamma 2.2 as a greyscale workspace.  I made the comment that if working in
Gray Gamma 2.2 and sending a file to the printer Same as Source (ie Gray
Gamma 2.2) then one would be better off, if using printer colour management
which Black Only does, setting the gamma setting at 2.2 rather than 1.8
because then the print space would be closer to the workspace.  This is born
out by the test.  The total error for a Gray Gamma 2.2 step wedge printed
with the printer driver gamma set at 1.8 was 221.2 L units (total Lab L
error, measured output vs input).  With the printer gamma set at 2.2 the
error was 142.3.

Following Clayton's comments and as a test of Clayton's "hours of trial and
error" I decided to run the test with a workspace of Dot Gain 20%.  At a
printer driver gamma setting of 1.8 the total error was 89.7 and at 2.2
gamma the error was 127.6.  So yes, on the 2100 printer, the combination of
Dot Gain 20% and a 1.8 setting in the Epson driver will yield a closer match
than the other combinations tested.

(If any wants the results of the test I can post the numbers.  It is likely
informative to know which parts of an image are varied to what extent.)

As to why this is the case I don't know but perhaps the difference between
you and I, Clayton, is that I would like to understand why.  It is the case
for the 2100 but I do not know if it is the case for other printers.
Without understanding the why I can't make a projection to the application
of the results with other printers.

Be that as it may, though, it is still not WYSIWYG.  In fact, far from it.
To get a decent WYSIWYG scenario you will need, as noted above, a soft
proof.  Once you have a soft proof, quite frankly it doesn't matter what
combination you use except that you will need a greater degree of image
adjustment for some combinations than others.  In other words, if you want
to address the topic of this thread you need to think about getting your
hands on a soft proof.  Carl, has made available PS soft proof files for the
21/2200, Gray Gamma 2.2 and Epson driver gamma 2.2 combination.  If people
want I am happy to make one for the 21/2200, Dot Gain 20% and Epson driver
gamma 1.8 combination.

The very last point I made on BO printing and Matching Monitor and Print was
that it would be great if we could use the same technique that Roy used for
constructing an ICC profile of the BO print space (1.8 and/or 2.2).  Then
rather than messing with adjustment curves we could use PS' built in colour
management with perceptual intent to map images from whatever workspace we
choose to work in to the BO print space.  This point remains and is still a
valid goal.  Why?  Because if this part of the process can be automated, a
novice need not bother with image adjustment curves designed to bend the
image file to the print space - they can sit back and let perceptual
rendering do the job for them (and believe me it does it quite well).  What
could be simpler than that?  How to do this is beyond my understanding for
the moment...but I would love to understand how!  <g>

Now, I have 3 big fat trout to dress

Steve