RE: 8 verses 16 bit

[rafeb@xxxxxxxxxxxx]

On 5 Mar 2002 at 8:30, gary wrote:

> Note that the ability to digitize the noise is what lets averaging reduce
> noise in the first place.

But by the same token, simple filtering can be done in the 
analog realm with little or no loss and very low cost (eg., 
an RC filter into a high impedance input.)  I see no point 
of extra data (or bandwidth, for that matter) unless there's a 
benefit somewhere.


> I don't have a problem with the data conversion hardware being a bit better
> than the analog source, i.e. a little margin is good.


I agree, but let's not pretend that it adds information, 
or that it's necessary for good results.

FWIW, the standard 12-bit AFE (analog front end) used in 
most desktop CCD scanners is a $1.25 part made by 
Wolfson Microelectronics.  So let's not kid ourselves that 
the A/D is a major factor in the cost of a scanner.

It is surely a good thing that most scanners do their 
internal white point/black point compensation using 
16-bit arithmetic, because quite often there's a lot 
of compensating to do.  Most of that takes place 
under the hood and is invisible to the user.  If we 
started with an 8-bit A/D, we'd probably end up with 
5 or 6 useable bits at the output.  If we start with a 
12-bit A/D we end up with 8 or 9 useable bits 
(rough guess.)  CCDs are very crude devices...


rafe b.

-
Turn off HTML mail features. Keep quoted material short. Use accurate
subject lines. http://www.leben.com/lists for list instructions.