Ken N. wrote:
>>
Just a comment regarding scanning resolution vs grain: I did a maximum
resolution scan of a Tri-X neg and compared the "grain" with a chemical print
I made of the same frame. Even though the film grain is larger than the
pixels, the pixels are not round, nor do they vary in size and position. It
really takes about 9 pixels (3x3) just to get a halfway realistic
representation of a single grain. The scan appears NOISY as compared to the
photographic print as during the scan a pixel may pick up the average of
several grains or it might pick up just the center of a single grain. This
is a subject of interest to me, because I KNOW that Provia-F is as close to
grain-free as possible, but digital scans seem to pick of "grain" that isn't
there in a normal photographic printing process.
<<
Ken,
This is related to "Aliasing" (in photo speak "moire"). We need to
sample the picture with more than 2 pixels per line of resolution in any
direction to avoid this. (this is the so called theoretical Nyquist sampling
criterion). In practise you need a bit more than the theoretical 2 pixels ,
especially since the pixel spacing is not even the same on the diagonal as it
is in vertical or horizontal scan directions. Aliasing, from pixel sampling
adds low frequency noise that does no exsist in the original. The fact that
the pixels have spaces between them also increases noise slightly because the
sample does then not average over the theoretical whole pixel area.Your 9
pixel estimate for a dot translates to about 2 pixels per "line"
(line=1dark,1light transition) in any direction. Right on the limit. Even
cheap scanners can improve aliasing a little by scanning and averaging
overlapping pixel positions in the direction of the scan.
Regards,
Tim Hughes
>>Hi100@xxxxxxx<<
< This message was delivered via the Olympus Mailing List >
< For questions, mailto:owner-olympus@xxxxxxxxxxxxxxx >
< Web Page: http://Zuiko.sls.bc.ca/swright/olympuslist.html >
|