Digital BW, The Print

Hi Clayton


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

> 
>> Since most people work in Gray Gamma 2.2 for B&W
> 
> What do you mean by this?  There are three different places to assign
> things: the image profile, the printer profile, and the driver gamma
> setting.  What does "work in Gray Gamma 2.2" mean in terms of these
> three settings?  And how do you know that "most people" do whatever it
> is you mean?

I was referring to people generally setting their workspace and tagging
their B&W images as Gray Gamma 2.2 (not the printer driver for now).  I note
that your website refers to your use of Dot Gain 20% but it has been
mentioned broadly here that Gray Gamma 2.2 is very very widely used and it
is also nice for those converting from colour (where the de facto standard
is Adobe RGB) because Gray Gamma 2.2 is a subset of Adobe RGB.

> 
> 
>> it would make sense to have a print space gamma as close to this as
>> possible.  
> 
> What do you mean by "print space gamma"?  Do you mean the printer
> profile setting or the driver gamma setting?

Any print space has a gamma = rate of change of density as you move from a
pixel value of 0 (pure black) to 256/65536 (white).  When using Black Only
the Epson driver allows you a choice of settings which influence the output.
> 
> 
>> Choosing 1.8 rather than 2.2 in the Epson driver simply makes the
>> mismatch worse (and, by corollary, the adjustment of the image
>> needs to be greater to get satisfactory prints vs if you had
>> selected 2.2).
> 
> Again, a nice theory, but it doesn't hold up in the real world.  In
> actual use I find just the opposite.

OK but the greater mismatch is clearly demonstrated by printing and
measuring a step wedge printed with each.  I have posted the results of
this.



> Selecting 2.2 ends up requiring
> more work.  It has been over two years since I worked this stuff out,
> so I decided to go back and test your theory to see if I was missing
> something.  This morning at church I made a portrait of a woman, using
> a digital camera, so I decided to use it as a test case. I opened it
> in PS and did a straight grayscale mode conversion (I wanted to do as
> little manipulation as possible for this test).
> 
> The pic was well exposed and the BW image is very pleasing and will
> not require much work.  The grayscale tones have values that closely
> match the color image (IOW, the image doesn't look under or over
> exposed in comparison to the color version, and the contrast is
> likewise very close).  Keep in mind that this image was automatically
> tagged with my default DG18% profile, and the printer profile is
> always set to Same As Source.
> 
> I then made a straight print (no curves, no manipulations) with my
> standard driver gamma setting of 1.8.  The resulting print looks just
> like the screen image (given the differences between a print and a
> screen).  These are the settings that I have found give the best
> WYSIWYG.
> 
> Then I tried your way.  First I set the image profile to GG 2.2 (which
> immediately made the image extremely dark on screen, not at all
> resembling the original).

Converting the image profile should not have changed the way it looked at
all so I don't know what you did here.  This would be a fundamental
breakdown in the "test".  Because the representation on screen is colour
managed the image should have been displayed exactly as before.

Again, if your monitor is halfway decent (and I am sure it is) and is
halfway profiles (even its factory default will do) you wont get immediate
WYSIWYG for no other reason than your monitor can display a much darker
black can you can print.  This is a simple fact.  If you get WYSIWYG between
a display and a print on matte paper without a form of soft proof (either
ICC or curve based) then you have at least dulled down the monitor's black
point (ie reduced the darkest black it can produce) to one that matches dMax
on matte paper (ie only a density of about 1.6-1.65).


>I left the printer profile at SAS (because
> it is the same as changing it to GG2.2, so in effect the printer
> profile is now also GG2.2), and printed it with the driver gamma set
> to 2.2.  The resulting print was similar to the first print, but was
> slightly darker and had significantly lower contrast, resulting in a
> duller and less pleasing print).  And of course there was terrible
> WYSIWYG - the screen image was very dark and bore little resemblance
> to the print (or the original).
> 
> So by doing it your way I got a duller starting print and terrible
> WYSIWYG, which means the image would require more work to get a nice
> print, and it would be done under more difficult conditions (poor
> WYSIWYG), which would probably require more time, paper and ink.
> 
> Now, what actually happenend here?  First, setting the image profile
> to GG2.2 made absolutely no difference in the print.  This is because
> whenever the printer profile is the same as the image profile (SAS) it
> has the opposite affect and cancels out whatever the image profile
> does (as far as what goes to the printer - the image profile does
> affect what we see on screen).   So the only thing that affected the
> print was the driver gamma setting.  Just to reassure myself on this
> point, I made a 3rd print with image profile back at DG18 and driver
> gamma at 2.2, and got an identical print to #2.  So the only thing
> that made any difference in the print was the driver gamma setting.
> 
> So why do I prefer 1.8 to 2.2?  Because I tried all three available on
> the 2200 (1.5, 1.8, 2.2) and found that 1.8 gives a print that most
> closely resembles a screen image with profile at DG18 or 20, and I
> prefer DG18 or 20 for the image profile because it produces a screen
> image that most closely resembles the original image, which in most
> cases is well exposed and closely resembles the original scene.  In my
> experiments I found that driver gamma 1.5 prints were lighter and more
> contrasty than 1.8 prints, and 2.2 prints were darker and less
> contrasty.  So the main points are these:
> 
> 1) With printer profile set to SAS, the image profile does not affect
> the print, it only changes the screen image.

With Same as Source, the image profile is passed through to the printer
unchanged.  Alter the image profile and that different image profile gets
passed to the print driver.  With Same as Source all you are doing is
telling Photoshop (and Photoshop only) to not do any profile conversion
prior to sending the print to the driver.  Same as Source doesn't mean "send
no profile" - it means send the same profile with which the file is tagged.
As you likely know, there are two places where one can do colour management
in a fully colour managed workflow. Either in PS on the way out by selecting
the print profile there and having no printer colour management.  Or by
using Same as Source in PS and using the driver to do the colour management.
In both cases, the profile accompanying the document as it leaves PS is
important.


>Therefore we can choose
> whatever image profile that gives us the best WYSIWYG.  This may vary
> depending on monitor and graphics card.  I recommend DG20 in the
> article because it's in the middle of the brightness range of all the
> canned profiles and is a good starting place.  For my own setup, I
> found that a custom DG18 curve was an improvement over DG20.
> 
> 2) The only thing that affects the print (other than changes to the
> image itself) is the driver gamma setting.

Sorry but I really do not think this is correct.  The profile with which the
image is tagged is a fundamental starting point.

> I recommend 1.8 because
> it's in the middle of the brightness/contrast range of the three
> choices, and, from experience, produces the most well balanced print,
> requiring less work.  Also, because because 1.5 and 2.2 significantly
> alter the contrast as well as brightness, 1.8 is also the easiest to
> find a matching image profile for good WYSIWYG.

I don't know what you mean by these terms here.  The dynamic range of the
print can't change in all circumstances: paper white is paper white and ink
black is ink black.  The contrast or gamma of the print will alter depending
on the gamma setting.  What do you mean by "brightness"?  (I think you are
using it to describe gamma/contrast)


> 
> 3) Putting it all together, I have found that the combination of image
> profile at DG18 or 20, printer profile at SAS, and driver gamma at
> 1.8, produces 
> - a screen image that most closely resembles the original
> - the most well balanced starting print
> - the best WYSIWYG
> 
> So it's a "straight through" workflow, with all three stages looking
> alike, requiring the least amount of work under the best possible
> conditions. 
> 
> None of this is theory.  It is the result of countless hours of trial
> and error experiments (and an ungodly amount of ink and paper - MIS
> loves me <g>).  It's easy enough, anyone can try it for themselves in
> a few minutes.
> 


I am glad it works for you.  But as I have simply suggested WYSIWYG is not
possible on a well profiled decent monitor without a softproof mechanism -
if only because of the black point.  That is the case regardless of how the
file is tagged and what gamma is selected.  So simply tagging a file DG20%
and selecting 1.8 in the driver won't get you there for a very large
proportion of today's monitors and particularly LCDs.  You arrived at your
workflow through countless hours of trial and error.  I'm glad it works for
you.  The question is whether it can now be refined so that less people have
the "matching monitor and print issue" - or more simply whether your
starting point is best recommended for new users.  One step is to make the
print space closer to the workspace.  The next step is to provide a soft
proof which people can use to preview their images on screen and amend them
accordingly.  Both these steps are very easy.  I will post a soft proof ICC
file for Eboni on EEM later today (I am out for most of the afternoon and
evening, UK time). 

I don't even use Black Only but, given the countless hours that people spend
on this issue, it is worth spending some time on the subject.  The issue was
posted by Paul for his workflow.  I have simply noted that the issue is
largely dead for QTR (and by default IJC/OPM) users and could very easily be
dead for many more users.  There will still be the vagaries of different
instrumentation, ambient lighting etc etc (the disappointment which Bruce
Fraser refers to), but at least the bulk of the issue can be dealt with.

[Take a look again at the results I posted for printing a step wedge with
the four different methodologies (in answer to a post quite some time ago) -
QTR, Epson driver colour managed, Epson BO 1.8 and Epson BO 2.2.  There you
can see the measured input vs output results.  Just one man's test but
objectively measured output.  Let me know if I need to post the link to the
pdf of the results again.]

Cheers

Steve