Elitechrome 100 Slide Scanning with Coolscan V ED

[Gordon Moat]

I get what you're saying, and it makes sense.

OTOH, when I scan at 5400 on E100S I get strong grain appearance
that does not show projected. The same film scanned at 1/4 the
res of the scanner does not exhibit the grain as strongly (and of
course nor does it show the detail). So pehaps grain aliasing is
the wrong term, but something is happening (or not happening)
resulting in the chunky looking grain. Other films (incl. GX,
Sensia, Velvia) do not show these artifacts...

I have noticed similar problems scanning some Kodak and Fuji films on Nikon
film scanners. I have rarely had the same problems using Canon or Polaroid film
scanners. Perhaps that may be due to the light source, or the focusing, or the
software, or some combination of items. The few times I have experienced the
same trouble (like with Ektachrome 400X), I found that manually focusing the
scanner produced better results, and did not affect the captured detail amount.
Another thing to try is adjusting the scanner exposure, rather than the gain,
or gamut level.

Ciao!

Gordon Moat
A G Studio
<http://www.allgstudio.com>
<http://www.agstudiopro.com> Coming Soon!

 

[Kennedy McEwen]

I get what you're saying, and it makes sense.

OTOH, when I scan at 5400 on E100S I get strong grain appearance that
does not show projected. The same film scanned at 1/4 the res of the
scanner does not exhibit the grain as strongly (and of course nor does
it show the detail). So pehaps grain aliasing is the wrong term, but
something is happening (or not happening) resulting in the chunky
looking grain. Other films (incl. GX, Sensia, Velvia) do not show
these artifacts...

A lot depends on the downsampling process used in the scanner driver.
Most scanners simply drop samples to produce a low resolution scan
because this makes for faster scans. The correct way, which avoids
exaggerated aliasing, is to scan at full resolution and downsample using
an averaging process. Clearly this makes the entire process slightly
longer for low resolution scans than for high resolution ones. This is
essentially what Nikon does on their scanners and there used to be a FAQ
to that effect on the Nikon web-site soon after the LS-4000 was released
because it was so different from the normal expectation for scanners. I
don't know what algorithm your Minolta is using since it is not
something I have investigated with it, but it should be a relatively
simple thing to work out from some test scan images and MTF
measurements.

 

[Alan Browne]

A lot depends on the downsampling process used in the scanner driver.
Most scanners simply drop samples to produce a low resolution scan
because this makes for faster scans. The correct way, which avoids
exaggerated aliasing, is to scan at full resolution and downsample using
an averaging process. Clearly this makes the entire process slightly
longer for low resolution scans than for high resolution ones. This is
essentially what Nikon does on their scanners and there used to be a FAQ
to that effect on the Nikon web-site soon after the LS-4000 was released
because it was so different from the normal expectation for scanners. I
don't know what algorithm your Minolta is using since it is not
something I have investigated with it, but it should be a relatively
simple thing to work out from some test scan images and MTF measurements.

My typical practice is to scan archive at full res, crop and
store as TIF. From that I downsample (Photoshop) to the size
needed for printing or screen viewing. However, with some scans
that I know are just going to be web presented, I scan at 1800
and then downsample from there (Photoshop). At 1800, the
'digital artifact' grain for some films is not as bad as it is at
full res. I have no precise idea how Minolta do lower res
scanning, however it is most certainly fewer samples as the scan
time is less. If they were downsampling from full res, the scan
would take as long as a full scan, I expect. (Note the standard
scan densities in the Dimage Scan s/w are all all evenly divisble
into the max. (5400, 2700, 1800, 1350, 900, 675, 540) suggesting
simple undersampling).

 

[Kennedy McEwen]

I have no precise idea how Minolta do lower res scanning, however it is
most certainly fewer samples as the scan time is less. If they were
downsampling from full res, the scan would take as long as a full scan,
I expect. (Note the standard scan densities in the Dimage Scan s/w are
all all evenly divisble into the max. (5400, 2700, 1800, 1350, 900,
675, 540) suggesting simple undersampling).

They may well be downsampling using a binning process on the CCD,
whereby 2, 3, 4 etc. elements are summed together to produce an
effective large pixel in the CCD axis. This reduces the readout time of
the CCD considerably. Alternatively they may be reading the entire CCD
and downsampling correctly in that axis but then multi-stepping the
head, resulting in incorrect downsampling in that axis. Either approach
would reduce the scan time but still offer some advantage in terms of
aliasing - but not as much as proper downsampling from the full
resolution.