Epson F-3200 vs. Minolta Dimage Multi II

  • Thread starter Thread starter Ralf R. Radermacher
  • Start date Start date
R

Ralf R. Radermacher

If you had to chose between the new Epson F-3200 and a Minolta Dimage
Multi II (not the later Pro version!), what would you do?

I already have a scanner for 35 mm, so the new scanner would only be
used for MF negative scanning (645 to 6x12).

Has enyone ever compared the Minolta to the Epson Perfection flatbeds
for MF scanning, bearing in mind the Minolta's optical resolution of
just 1000 dpi?

Ralf
 
Ralf R. Radermacher said:
If you had to chose between the new Epson F-3200 and a Minolta Dimage
Multi II (not the later Pro version!), what would you do?

Neither. I'd buy the 4870. Even if it's worse than I think, NeatImaged,
sharpened, and downsampled to 2000 dpi, it should produce nice sharp images.
And that's 13MP quality pixels from 645. Enough to be useful.

The F3200 has no IR ICE, uses the same sensor as the 3200 dpi flatbeds, and
may cost more than the 4870. Three strikes.

I fail to see any reason for the excitement over the F3200.
I already have a scanner for 35 mm, so the new scanner would only be
used for MF negative scanning (645 to 6x12).

Has enyone ever compared the Minolta to the Epson Perfection flatbeds
for MF scanning, bearing in mind the Minolta's optical resolution of
just 1000 dpi?

At 1000 dpi, 645 is 1600 x 2200 = 3.52 MP. You'd be better off shooting a
5MP digicam.

David J. Littleboy
Tokyo, Japan
 
David J. Littleboy said:
Neither. I'd buy the 4870. Even if it's worse than I think, NeatImaged,
sharpened, and downsampled to 2000 dpi, it should produce nice sharp images.
And that's 13MP quality pixels from 645. Enough to be useful.

I had a Prefection 2450 for about two years, then ordered a 4870, tested
it extensively during a good week before I sent it back and bought a
second-hand 3200* as a stop-gap measure.

The 2450 is prone to problems with uneven lighting that won't bother
most people but can be quite annoying if you're trying to assemble
scanned frames into panos.

The 4870 just wasn't enough of a progress to justify the expense.

The 3200 is O.K. considering what I paid for it.

BUT... and this is the point... none of them produces more detail from
645 than what I get from 35 mm with a Nikon LS-2000.

That's why I'm trying to determine if there's something better but still
affordable or if I'll simply give up MF.

Ralf

*) Perfection 3200, not the new F-3200
 
Ralf R. Radermacher said:
BUT... and this is the point... none of them produces more detail from
645 than what I get from 35 mm with a Nikon LS-2000.

Hmm. I'd think the 4870 scans, Neatimaged, sharpened lightly, downsampled to
2400 dpi and sharpened lightly again would be better than what one could get
from 35mm, whatever the scanner. Wrong again???
That's why I'm trying to determine if there's something better but still
affordable or if I'll simply give up MF.

After seeing what 1800 dpi (that's all, folks) Kodak ProPhotoCD scans from
645 looked like, I bought a Nikon 8000...

David J. Littleboy
Tokyo, Japan
 
David J. Littleboy said:
Hmm. I'd think the 4870 scans, Neatimaged, sharpened lightly, downsampled to
2400 dpi and sharpened lightly again would be better than what one could get
from 35mm, whatever the scanner. Wrong again???

If the information isn't there, no amount of neating, sharpening, up- or
downsampling, or whatever will magically reveal it.

Such trickery may work with things like tropical sunsets but it's of no
use if you're doing stuff like industrial archeology where you want to
see every little bit of pipework, provded it's on the neg, of course.
After seeing what 1800 dpi (that's all, folks) Kodak ProPhotoCD scans from
645 looked like, I bought a Nikon 8000...

:-/

Ralf
 
Ralf R. Radermacher said:
If the information isn't there, no amount of neating, sharpening, up- or
downsampling, or whatever will magically reveal it.

Of course. The object of downsampling is to get the information actually
captured into a form that's more easily used. Looking at Epson scans at 100%
on the screen is seriously depressing (since they're so grossly soft), and
makes it hard to evaluate what you really have.

More importantly, most PCs/Macs will choke on a 4800 dpi scan of 6x7 and
take forever for 645. It's easier to sharpen for final printing from a file
that will print at 300 dpi than one that will print at 600 dpi. If the
downsampling loses information, you downsampled too far.

But my question is (since I don't own either a 4870 or a 35mm camera): is
645 at the 2400 dpi produced by the 4870 really no better than 35mm scanned
at 4000 dpi? 4000 dpi scans don't look all that great (at 100% on the screen
or 300 ppi on a print), and the number of pixels is quite comparable.

Have you taken the same shot with 645 and 35mm, scanned with the 4870 and
LS-2000, printed to, say, 11x14, and asked people which looks better???

My point of mentioning the Kodak ProPhotoCDs, was that it seemed to me that
a mere 1800 dpi from 645 gets more than is reasonable to even dream of from
35mm, so the 4870 should be in the ballpark for making 645 be better than
35mm.

If you throw away the information at the point of capture by using a
subminiature format, you're dead in the water. Of course, it's no fun to
have to pay lab scanning/printing charges to get what's on your film.
Such trickery may work with things like tropical sunsets but it's of no
use if you're doing stuff like industrial archeology where you want to
see every little bit of pipework, provded it's on the neg, of course.

Again, since the 4870 isn't capturing 4800 dpi of detail, it makes sense to
downsample it to a point where one is using one's pixels sensibly*. Even the
downsampling trick has the problem that neatimaging and sharpening at 4800
dpi is seriously painful on most PCs/Macs.

*: Someone here claimed that the 4870's limiting resolutions are 30 lp/mm in
one direction and 40 lp/mm in the other. Since 3 pixels per line pair is
plenty at the limiting resolution, that's 120 * 25.4 = a tad over 3000
pixels, Kennedy McEwen's estimate of the 4870's resolution. My 2400 dpi may
be a tad on the low side.

David J. Littleboy
Tokyo, Japan
 
David said:
one could
get

Of course. The object of downsampling is to get the information actually
captured into a form that's more easily used. Looking at Epson scans at 100%
on the screen is seriously depressing (since they're so grossly soft), and
makes it hard to evaluate what you really have.

More importantly, most PCs/Macs will choke on a 4800 dpi scan of 6x7 and
take forever for 645. It's easier to sharpen for final printing from a file
that will print at 300 dpi than one that will print at 600 dpi. If the
downsampling loses information, you downsampled too far.

But my question is (since I don't own either a 4870 or a 35mm camera): is
645 at the 2400 dpi produced by the 4870 really no better than 35mm scanned
at 4000 dpi? 4000 dpi scans don't look all that great (at 100% on the screen
or 300 ppi on a print), and the number of pixels is quite comparable.

Have you taken the same shot with 645 and 35mm, scanned with the 4870 and
LS-2000, printed to, say, 11x14, and asked people which looks better???

My point of mentioning the Kodak ProPhotoCDs, was that it seemed to me that
a mere 1800 dpi from 645 gets more than is reasonable to even dream of from
35mm, so the 4870 should be in the ballpark for making 645 be better than
35mm.

If you throw away the information at the point of capture by using a
subminiature format, you're dead in the water. Of course, it's no fun to
have to pay lab scanning/printing charges to get what's on your film.
course.

Again, since the 4870 isn't capturing 4800 dpi of detail, it makes sense to
downsample it to a point where one is using one's pixels sensibly*. Even the
downsampling trick has the problem that neatimaging and sharpening at 4800
dpi is seriously painful on most PCs/Macs.

*: Someone here claimed that the 4870's limiting resolutions are 30 lp/mm in
one direction and 40 lp/mm in the other. Since 3 pixels per line pair is
plenty at the limiting resolution, that's 120 * 25.4 = a tad over 3000
pixels, Kennedy McEwen's estimate of the 4870's resolution. My 2400 dpi may
be a tad on the low side.

David J. Littleboy
Tokyo, Japan
 
Well, actually nothing but a full-quote of other peoples' contributions
to the topic at hand.

So....???

David:

The trouble with all those Epson flatbeds, and I have owned or tested
three of them, so far, must at least in part be due to internal
reflections or stray light. The scans always look like those from my
Nikon LS-2000 if it needs a good cleaning. Reasonably sharp but with a
sort of of aura around everything, if you know what I mean.

This is always present, no matter how new or old the scanner is and
regardless of how often you clean it. Must be an inherent problem.

Ralf
 
The trouble with all those Epson flatbeds, and I have owned or tested
three of them, so far, must at least in part be due to internal
reflections or stray light. The scans always look like those from my

See Flextight. The problem is with the way that they hold and scan
the negatives.

http://www.imacon.dk/sw323.asp

"The Flextight holder


What do we want to achieve with an original holder? We want to be able
to keep the original in perfect focus across the scan line. We want to
avoid Newton rings. We want to have as few surfaces in the light path as
possible to clean and to break the light waves. Finally we made it as
easy as possible to mount the original in the holder.
All this we have achieved in the Flextight original holder, which is
glass free, with a magnetic rubber overlay, which makes mounting a very
easy task. The perfect focus across the scan line is achieved by bending
the original around a virtual drum. When you bend a media in one
direction it becomes perfectly straight and stable in the other direction.

The holder concept makes it possible to take the holder from the
scanner, after preview, and place it on a light table for comparing and
adjusting the scanner preview. When finished the holder is easily put
into the scanner for final scanning. This feature is not available on
most of the competitive scanners.


Focal Length


Why is the scanner so high? It is because we want to have a long focal
length. By having a long focal length we only utilize the lens where it
is best, which means less distortion and therefore fewer corrections.





» Larger image



The glass free optical path


By building the scanner like we did we have actually achieved a glass
free optical path. This means that we do not have to correct for
unwanted effects like chromatic aberration or diffusion caused by glass
plates, prisms or mirrors. It also means that less mechanics are to be
adjusted and finally that less surfaces are to be kept clean.


CCD facing downwards


A simple but obvious thing is to place the CCD face down, as it is
unlikely that dust gathers on the light sensitive cells in this position.


The direct analogue to digital conversion


When the analogue signal produced in the CCD is to be converted to a
digital signal it has to move from the CCD to the AD converter. As the
analogue signal is a low tension electric current it is very sensitive
to electrical noise, which very likely is produced by other electrical
units in and around the scanner.

Imacon has therefore decided to place the AD converters directly on the
CCD board to keep the travelling distance as short as possible. And to
lower the risk of noise we have even decided to have one AD converter
for each color channel instead of sending the signals through a
multi-plexer, which has the potential risk of adding more noise.


Keeping heat out/down


Every time the temperature raises with 10° Celsius the electrical noise
in the sensor doubles. This means that it is very important to prevent
the sensor from heating up.
The most obvious source of heat is the power supply. By removing that
from the cabinet and placing it outside the scanner one big problem is
eliminated.
A second thing is the light tube. This light tube is a cold catode tube,
which produces very little infrared waves and therefore produces very
little heat.

The sensor itself is an electronic device that heats up when the clock
frequency is speeded up. This is usually done to make fast previews. The
sensor samples full information even at preview scans, so every cell has
to be emptied to produce even a low-resolution preview.

Imacon has found a way to add the information in the sensor together
directly, which means that we do not have to speed up the readout of the
sensor, therefore not producing the heat, which is common in this process.
At the top of the line scanner Imacon has even chosen to cool the sensor
actively, which means that an electronic device is placed directly on
the CCD to cool it down. This will increase signal to noise ration with
1 to 2 bits.


Stable metal construction


To keep the scanner stiff and stable all mechanic parts as well as the
cabinet are made out of metal. And due to the fact that the scanners are
high speed and have attachable batch feeder devices we expect that they
will make thousands of scans and are therefore build to last.
"
 
Back
Top