Re: VARIABLE REZ TEST: Minolta Scan Multi Pro!

[Kennedy McEwen <rkm@xxxxxxxxxxxxxxxxxxxx>]

In article <101836035802@wi.net>, david/lisa soderman <scapes@wi.net> 
writes
>>>but it seems to fly in the face
>>>of the 4800ppi's "extra sharpness" as being the culprit re: grain/noise.
>>>
>> Well, for the moment, I disagree with that conclusion David, and I will
>> tell you why in due course, but I am interested to hear how you arrived
>> at it.
>> --
>> Kennedy
>
>Nothing profound.  All I meant was to point out that none of the previous
>noisy scans were done at interpolated 4800ppi...and that the lower rez scans
>of 3.2k (supposedly opticel)/1.6k sport the same noise /grain.
>
Fair enough, but I think you need to know how the Minolta generates its 
3200 and 1600ppi scans before you can come to that conclusion.

Correct me if I am wrong here, but I don't think this scanner has an 
optical zoom arrangement and, if not, then these lower resolution scans 
are not optical resolutions, but downsampled derivatives of the full 
resolution of the scanner.  Now, depending on how that downsampling is 
implemented, the noise could show a similar level or, if implemented 
correctly (as it is in the Nikon range but resulting in identical scan 
times for all resolutions) significantly reduce.

For example, simplistically reducing from 4800ppi to 1600ppi should 
result in a noise reduction by a factor of three - since each new sample 
is taken from the average of 9 original pixels.  (Noise reduces 
according to the square root of the number of samples.)  That would 
closely simulate an optical resolution of 1600ppi, where the pixels size 
and pitch increase proportionally.  This step also filters alias 
components by the larger pixel area before resampling, rather than 
simply increasing aliasing at the much lower Nyquist point.

However, you can compare your lower resolution scans to Photoshop 
downsampled versions of the 4800ppi scan - and they are very similar. 
Knowing that the PS algorithm takes no steps to eliminate aliasing from 
the downsampling process, it is clear that the Minolta algorithm doesn't 
either.

Hence my conclusion is that the extra sharpness of the Minolta is still 
a candidate for culprit - all the aliasing and other crud that is 
present at 4800ppi is just redistributed at 1600ppi.  You won't find 
that with the Nikon (unless the CS8000 downsampling algorithm is 
substantially different from the CS4000 one) which Rafe can confirm for 
you.
-- 
Kennedy
Yes, Socrates himself is particularly missed;
A lovely little thinker, but a bugger when he's pissed.
Python Philosophers
-
Turn off HTML mail features. Keep quoted material short. Use accurate
subject lines. http://www.leben.com/lists for list instructions.