Digital BW, The Print

--- In DigitalBlackandWhiteThePrint@yahoogroups.com, "Tyler Boley" <tyler@t...> 
wrote:
... 
> That's what kind of bugs me, we do it to taste instead of an accepted
> standard. We can both be linear, but be printing considerably lighter
> or darker than each other depending on our eyes. Of course we are
> editing our files to suit, but one would think there could be some
> widely accepted standard that once we are all linearized to, the same
> file would print similarly across the board.

I think we're using "linear" to mean reasonably smooth, so there's many
reasonably smooth curves from white to black.  For QTR I took a
strict sense of the word meaning a straight line when using Lab values.

Here's a reference that talks about it -- see #5  Lightness.
http://www.poynton.com/notes/colour_and_gamma/GammaFAQ.html

"The perceptual response to luminance is called lightness, and is 
defined by the CIE Publication ...  "
The text doesn't call it "Lab" but the formula given is exactly what shows
up in color formulas.

This isn't to say that's the only "correct" answer.  But it would be nice
if all our files would be printable under different systems.  From
what I've seen some systems end up printing a little lighter than
this -- i.e. the midtones are lighter than linear Lab.  I wouldn't be
surprised if this wasn't influenced by the thought "I want more
shadow detail" or "I want the print more luminous i.e. lighter".

Tyler, since you've picked the dot gain 20% for your linearized system.
I'm wondering how that compares with linear Lab.  For instance, on
one step wedge I have:  0 step = 97 L, 50 step = 57 L, 100 step =17 L.
So the 50 step is right in the middle.   What would your system give?

The trouble with all this is that no matter what you have that's what
your files are edited to so.   It's painful to switch even if you think its
the right thing to do.   I don't know how you can handle other
peoples files.

Roy

> Really outstanding single channel icc profiles would put much of this
> to rest, if we both had them, we could print the same file
> theoretically identically.
> 
....
> 
> > On thin ice now and in my own words.... Lab is "more or less"
> > perceptually linear.
> 
> I'm showing my lack of formal color education here, but I think you
> are right. I think LAB was developed over time to come as close as
> possible to human visual perception.
> 
> >  On top of that the approach by colour
> > management applications isn't always in respect of the shape's
> > construction. Gamut mapping for instance asks for a more
> > complicated approach with (3d?) curved mapping (or tables) to get
> > good perceptual hue matches. Most of the time it isn't done like
> > that and shifts like the blue > purple hue occur. Similar problem
> > with density dot gain.  Some polishing of the original
> > mathematical setup of Lab is also needed apparently. All Bruce
> > Lindbloom's territory...
> 
> Lindbloom to the rescue again. I ran some tests, if we assume LAB is
> perceptually uniform from gray values of 0,0,0 to 100,0,0, then middle
> gray must be somewhere near 50,0,0. This means a dot gain of something
> like 31+%, much more than I would have guessed. Some other tests in
> photoshop confirm this. So I'm not sure if that helps at all.
> I've come at this problem from every conceivable direction, including
> density.
> At this point I can only conclude that "linear" simply means that as
> tones progress they do so in a mathematically consistent manner. Other
> factors thrown in to make it perceptually desirable, like gamma or dot
> gain, remain subjective decisions.
> But that makes comparisons of systems based on those factors (it
> prints too light, or too dark) useless, particularly if it can be
> varied easily by the user but remain "linear".
> Tyler