Digital BW, The Print

> So just when are you going to define "noise" WRT to a silver print and an
> inkjet print so that we have a basis to judge whether it's existence is
> RELEVANT?

We have shown how to do it with inkjet prints with pretty reasonable
certainty, and shown some proposed methodologies for doing so with chemical
prints.

> Please stop skirting that issue.

Well, I haven't!  Both Johnny Brownlow and I HAVE spoken to this issue.

> Let me phrase it with a question: what is the analog
> equivalent of a
> bit?

It depends on what you are representing with that "bit" (voltage,
current...), what the range of values can be, and what the noise is in that
"system" (determines how many bits are needed for the A/D).

Since we're talking about scanners, sort of, I'll relate it to that.  For a
typical A/D, which has a voltage swing of +- 3V, that makes it have an
absolute range of 6V.  If the measurable noise in the system was, say, .01V,
then you would have a dynamic range of 6V/.01V or 600, and you would use an
A/D converter that would accommodate that "range"...which would be one that
has enough bits to cover 1024 (next binary step up from 600), and that would
take 10 bits.  So, 6V / 1024 means each bit represents a voltage change of
~0.005.  The bottom bit would be pretty much "in the noise", but that doesn'
t effect the overall voltage is represented by a particular binary
number...as it'll be accurate to the noise, or +-.005V (1/2 the noise will
be high, and 1/2 low).

In the case I gave, the analog equivalent of one bit is ~0.005 +- 0.005V ;-)

Austin