Digital BW, The Print

> From: Paul Roark <paul.roark@...>


> Of course, a single 50% patch would be rather meaningless.  But it's a place
> to start.  

I agree.  What I meant though is that it may be fairer to simply show them
your mid grey only.  As this is the only number you target this is the only
one that is fair to test.  With a dMax of 1.56 (from your 21-step-K156.tif)
your printer can't even print anything to the left of your 80 step!  So
asking someone to compare a step wedge on the screen either Adobe RGB or
Gray Gamma 2.2 (both gamma 2.2 workspaces) with the printed version is
guaranteed the outcome you are getting especially if they are influenced by
comparing steps in the darker end (we have better means to get closer to
pure white than pure black).

(Take your 100% step for example - pure black, K=100%, LAB = 0.  Even your
monitor can't reproduce this but it gets a darn sight closer than your
printer.  Your Xrite measures your monitor's luminosity at pixel value = 0
(LAB=0 etc) in order to profile the monitor.  Unfortunately I do not know to
access the figure it measures (nor for the value measured by Eye-One) - I am
sure someone can tell us how.  With the number in hand you can calculate its
density equivalent and compare it with 1.56.)

I you are showing someone a full step wedge - the only fair test is for them
to look at it on screen with a curve that compresses the image tonal range
to the 0-255 equivalents of 1.56 and 0.04, with the curve in between such
that the on screen densities are those that you intend.  For example, in a
2.2 workspace if it goes through 128/128 then you intend a 0.66 density for
your 50% step.   Just for a laugh do this:  pull up your step wedge image;
do a curve with the end input/output points of 0/49 and 255/243; passing
through 128/135; run the eyedropper over the 100, 50 and 0 steps and record
the LAB figures; and then calc the densities.  You will I believe get
roughly 1.56, 0.62 and 0.04.  Those three steps (on a well profiled monitor)
should now match the printed version of K156.tif (ok so the 50 is a fraction
off and I am sure you can fiddle for the right input/output combo that hits
the nail on the head!).  Fit all the other points and you get a look at the
complete transfer curve that is taking place in the print end of your
workflow.

> The major "problem" is actually the way the monitors compress the
> darkest few steps to the point that, in my view, it's just wasting our
> limited grayscale.

I think this is a different kettle of fish entirely - much like Tyler's
reference to the question of what OUGHT to be mid grey or what OUGHT to be
the right workspace.  This is WAY beyond my knowledge and not something I
intend to tackle. I have no clue. I am currently very happy with Adobe
RGB/Gray Gamma 2.2.

I don't care which is chosen - what I do find ironic though is that people
set a workspace of gray gamma but linearize to a different workspace such as
LAB.  At a minimum they OUGHT to be the same to remove one layer of transfer
complexity when we are not working in a colour synchronised workflow (that
would manage this complexity for us).  I would love someone to give me a
credible explanation as to why they are different?


> Download one of them and take a look.  See
> http://home1.gte.net/res09aij/index.htm for my general index.  Of course,
> you'll need to be familiar with the inkset densities for the curves to make
> sense.

I'll definitely take a look.

> All RIPs have a transfer function buried in them at some level.  I don't see
> any one telling the  printer in machine code which dots to put where.  The
> Epson driver is a RIP -- but with some "hard-wired" cross-overs (which are
> excellent and useful if consistent with the desired results).

Yep.  But a programme like QTR does tell the printer what density (in
percentage ink terms) to lay down where and this is quite "open".  This is
defined in the code associated with the linearization.  QTR does not go so
far as to allow a user to define the type of linearization though - that
really would be quite a masterpiece.

 
> The Photoshop controls are more limited than a good RIP, but the differences
> are a matter of degree.  There isn't that much difference.  The only place
> I've been able to see a qualitative difference is where one needs to get
> control of individual jets and avoid the built-in driver cross-overs.
> 
> The ink differences are largely taken care of in the inkset design, where I
> work backwards from the printer and drive to the ink densities.  I prefer to
> work with printers designed for pigments, because then the differences
> between the OEM design assumptions and what I have available are minor.
> It's amazing the dye printers work as well as they do.   But, again, since I
> do the inkset design work also, I fit the 2 together.
> 
> The problems I have here are, for example, with what I'm doing now.  I'm
> trying to work with an existing inkset -- the UT-FSN -- and make an inkset
> that will print top notch matte and gloss prints on my 7500 and other large
> format printers of that generation without the need to change the black ink.
> The FS ink densities were never intended for this.  However, I've found that
> only one new ink will be needed -- a medium gray that is denser than the
> existing medium gray FS ink.  However, even here, where I'm not sure there
> will be enough volume for MIS to be interested in making the inkset, I find
> a way to easily mix the needed ink from exiting stock (in this case UT-FSN-M
> and EZ-N).  But, so far, I'm finding the printer controls I have with the
> Epson drive and Photoshop appear to give me what I need to do the job.
> 
> I'm not intending to restrict myself to the Epson driver.  When reasonably
> priced RIPs can improve the situation, I'll use them.
> 
I can't even begin to think about fussing with ink design!  Hopefully you
won't stop doing what you do.

>> ...
>> So the workflow becomes: 1. Get the image looking nice on screen ...
>> save as file copy.  2. Apply a print curve which specifically takes
>> into account the tonal range of the ink/paper combo ...
> 
> That's it.
> 
>> ... 3.  Make any last minute, image specific adjustments - ...
> 
> I'd do this in #2. However, in reality, I always assume that I need to print
> some draft prints and look at hard copy.  Not infrequently I make a slightly
> modified curve for the specific print, saving it with the information I need
> to identify it as a printing file for a specific paper, size, etc.  I just
> don't think any monitor can take the place of viewing the print on paper.


Completely agree.  A tweak would be to the same curve but would affect
points within the tonal range of the printer only. My proposed workflow
simply dials in the end points of the paper/ink combo, initially without
moving "mid grey" (which I am happy to have be 0.5 in gray gamma 2.2) and a
smoothing of the image compression with a nice S curve. Tweaks can be
infinite but tweaks at this stage relate to the effect of tonal compression
- the fact that I can't print the full image as I would otherwise like (I
can't print the tonal range of my monitor) - and largely likely in relation
to the knee points (highlights and deep shadows).

 
> 
>> ...But it would be easier if we got what we were looking for when we
>> push a button or shifted a curve.  That is, full logical transparency.
> 
> But that is where the skill and artistry comes in.

What I meant was the technical end.  You targeted 50% to a density of 0.61
yet it sits at 0.66 on screen and have explored the reason as to why people
have said it seems to print light.  What I meant was that it would be nice
to look at 0.66 and have it print 0.66.  This was not meant to be a
substitute for the artistry associated with which part of the image is 0.66!

Cheers

Steve