Digital BW, The Print

> From: Roy Harrington <roy@...>

 
> Hi Steve,
> 
> I've actually been following the thread all along and it sure has sparked a
> lot of interesting questions.
> 
> I like comparing things and linearizing in Lab because it's supposed to the
> ideal
> match for human vision.  I.e. linear Lab ought to be linear visually -- in
> other
> words a difference from Lab 15 to 20 (5%) is comparable to a difference from
> Lab 50 to 55 (also 5%) and Lab 85 to 90.  This makes sense as the best use of
> the 256 values you get for an 8bit data file.

Yep Tyler sold me on this.  But I don't believe QTR "linearizes in LAB".
This is a crucial point and one that I would like to explore further with
you - offline if necessary (everyone else breaths a sigh of relief!).  LAB
has a defined density curve - as pixel values (ie exposure) increase(s) from
0 (black) to 1 (white) LAB defines a particular shade of grey (if a=b=0), ie
density output.  If this curve is "the way the eye sees" then we need to
have our printed output from 0 to 1 match this curve.  We know we can't get
perfect 0 and perfect 1 but we can get a portion of the range in between.  I
believe that to be linearized to LAB this portion should take on the same
characteristics as the LAB curve.

Setting 0 to dMax and 1 to dMin and straight lining the LAB values in
between is analogous to creating an entirely new colour space. Plot the
density curve and compare it with LAB (and even gamma 2.2).
> 
> Unfortunately Photoshop doesn't have a Lab oriented grayspace.  We could use
> Lab spaces but first they have 3 channels so your files will be triple the
> size with
> no more info.  

Yes agreed.

> Second I think when you print, PS will get you "out of Lab" in
> the
> conversion to grayscale doing a profile mapping for you (or is it to you :) ).
> 
Sorry I am not sure what you mean by this.

> The main reason I've always used gray gamma 2.2 along with adobe RGB is that
> with the eye dropper you can see the 8 bit values in the file.  K gives only
> 0-100
> but underneath its really 0-255. You can see 0-255 in the RGB values but they
> are only correct if the gray and color spaces have the same gamma. (Profile
> conversions are always going on).

Yes that is correct.  But as we know from Bruce Lindbloom's site the gamma
best able to approximate LAB greyscale is a gamma of around the "standard"
2.2.  As I said earlier, I might tolerate working in 2.2 and then printing
in LAB only if the printer were actually calibrated to LAB ('cos only then
would 2.2 be roughly close to LAB).


Cheers

Steve