Digital BW, The Print

Hi Mark,

> > The only operation that benefits from 16 bits is tonal
> adjustment.  All the
> > things you mention above don't benefit from higher bit depth,
> unless your
> > output takes 16 bit data.
>
>
> Can you explain exactly what you mean by tonal adjustments?
> Do you mean Levels and/or Curves adjustments only?

Pretty much, yes.

> Not to be
> a dork, but is it possible to describe this advantage in simple
> English; not like N gets converted to blah blah?

Well, um, I may not be the best person to explain this in simple English.
It does require SOME technical background to understand what the actual
"issue" is...

> Could you describe a simple scan workflow example? And how
> doing levels/curves would be better in 16bit?

What happens when you make tonal adjustments with 8 bits is you loose tonal
values.  For 8 bit data, you have 256 possible tonal values, 0 to 255.  When
you make a tonal adjustment, you are changing/shifting the current values to
new values.  Any time you do this, you are going to get values that were
once different, now to be the same, therefore decreasing the number of
different values.

Say you have four values for tones, 0, 1, 2 and 3.  You want to "make a
move" and you move 0 to 1, 1 to 2, 2 to 3 and then what do you do with
3...it stays 3...so now you decreased the number of tones from 4 to 3.

The solution is, you need to make tonal adjustments in a larger bit space
than the actual image data you want, so you don't "overlap" values.  The
ACTUAL image data in a 16 bit file typically doesn't occupy all of the 16
bits, there are MANY of the 65k values that are unused.  It is because of
the unused values that you don't lose any values.  The data has "room" to
move without stepping on other datas toes.

Yes, this gets pretty technical, and when I get the time I have wanted to do
a write-up that explains this very clearly with pictures, circles and
arrows...but I haven't had the time to do it yet...

Regards,

Austin