Digital BW, The Print

You can get an idea of the differences with a few small PS experiments.
Take a nice full range B&W image.  Go into adjust curves.
Drag the 0,0 point to 10,10 and drag 100,100 to 80, 80
You ought to have a horizontal lines from 10,10 to the edge and
80,80 to the edge.  This is the change you are proposing.
Click Preview on and off to see the changes.  Shadows and Highlights
are clipped but the mid range stays the same.
Now move the points out to the edge along the horizontal lines.  The 
Input values will now be 0 and 100.  This shows the other way.  It
lightens the picture but there is no clipping.

Its a personal choice which you like, but my tendency is to go with the
readjustment to avoid clipping. 

My point about units was that 0 to 255 has no meaning for ink on paper.
The amount of light reflecting from paper depends on the light striking it.=

I was trying to say there are no absolute units that can relate to both
paper and a screen.

Roy

--- In DigitalBlackandWhiteThePrint@...m, Steve Kale <stevekale@=
b...> 
wrote:
> I think I would rather have all values between 80-100 be 80.  At least th=
e
> 80 value will be 80 and a 90 be at least as black as possible.  (And all
> units from 0 to 10 will be as white as possible.)  There will of course b=
e
> some loss of shadow detail but this is simply because it is beyond the
> rendition ability of the printer, ink and paper combination (in the same =
way
> that it can¹t produce a 5 because the paper isn¹t white enough).  Better
> this than a complete remap of the tone of the image and in effect lighten=
ing
> the whole image so that we retrieve the shadow (and highlight) detail at =
the
> margin (which we likely reverse with a curve at the time of proofing).  P=
ut
> another way, if we are happy with 80 as a deep enough black then we will =
not
> be too worried that a 90 only prints at 80.  (Of course the goal is to fi=
nd
> an ink and paper combination that is capable of expanding my 10 to 80 spa=
ce
> to a perfect 0 to 100.)  I suspect this is all much more of an issue in B=
&W
> with its simpler colour space, particularly when we typically drive for
> contrast with deep blacks offset with strong highlights.  A flat grey ima=
ge
> is exactly what we don¹t want.  In a colour image there is a lot more goi=
ng
> on and the various ways of dealing with out-of-gamut colours are more
> appropriate.   I agree that there should be good separation between 10 an=
d
> 80 so that we get a good ability to print in-gamut shades of grey.  The
> question is what happens at either end and the impact on the image of
> dealing with this.  As for units, I guess I am thinking of RGB values or =
any
> other means of measuring a grey scale space.  I used 0 to 100 as a simple=

> description.  I guess in RGB these would be equal R, G and B values from =
0
> to 255. 
> 
> 
> From: "Roy Harrington" <roy@h...>
> 
> Steve,
> 
> I think you are right that there is a "gamut compression" going on.  But
> I think there's a basic fitting of one gamut into another happening no
> matter what.  I'm not even sure how you'd compare a 75 gray in a file, a =
75=
> 
> gray on the screen and a 75 gray on paper in absolute terms.  What are
> the units?
> 
> Although I can see you point in your 10 to 80 example, I'm not to sure
> trying to do things "absolute" is great either.  Wouldn't you result in a=
ll=
> 
> the values from 80 to 100 being mapped to the same max 80 density?
> Maintaining the best separation you can throughout seems like a better
> alternative.   This is analogous to the color rendering -- using Perceptu=
al=
> 
> rather than Absolute Colormetric.
> 
> 
> 
> 
> 
> [Non-text portions of this message have been removed]