Digital BW, The Print

A good test would start with an ISO target or the razor edge 
target like Imatest allows. With a flatbed scanner that has 
the film holder tweaked to the right focus. Using one of the 
1200>2400>4800 SPI (sampling per inch) resolution choices 
isn't a guarantee either that one hits the best optical 
resolution or the best effort/result choice in PPI (pixels 
per inch) resolution. With several flatbeds if one selects 
1200 or 2400 SPI the scanner actually could skip sampling 
steps if its hardware allows 3600, 4800, 6400 or up to 9600 
steps per inch, that choice of a lower sampling resolution 
then either degrades the detail and/or dynamic range etc 
available when using higher sampling resolutions. Selecting 
an odd resolution number that doesn't divide well can make 
things better. The scanner then usually takes a higher SPI 
resolution with all of the mentioned advantages and after 
the scan it downsamples to the odd resolution number before 
saving the data. If that odd number is set near to the best 
optical resolution possible as measured with a target and 
the scanner behaves as described and the downsampling 
routine in the driver is a good one (for example bicubic 
with some anti-aliasing and not a plain nearest neighbour) 
then you get the best effort/result scan possible with that 
scanner. If the scanner only gives its best quality (not 
just detail but also dynamic range, Dmax and S/N ratio) only 
at the highest SPI resolution setting (mainly because that 
brings the oversampling higher) and not by the trick 
mentioned earlier then you have little choice than using 
that one. Another method could be to use a lower output 
resolution choice in for example Vuescan and set the 
multisampling choice in that driver higher, in fact it will 
do the same then. I doubt it samples twice at the same step 
but just increases the sampling rate, downsamples and saves 
that lower resolution. This is not a multiscan choice where 
the scanner runs two times and combines the two readings, 
that choice usually gives worse detail on flatbeds as the 
registration of both scans suffers due to several causes. 
Silverfast has that multisample feature too for scanners 
that allow it.

Imatest allows something like 10 trials in its demo version. 
Scan at some settings and compare not just the optical 
resolution but also the dynamic range etc. Norman Koren has 
some pages on the Epson flatbeds too.



-- 
Met vriendelijke groeten, Ernst


|  Dinkla Grafische Techniek  |
|     www.pigment-print.com    |
|             ( unvollendet )            |