Dimage Scan Elite 5400 vs Dimage Scan Elite 5400 II

[Bart van der Wolf]

SNIP

I played with GEM and ROC a couple of times, but I found it a
complete waste of time.

My experience, also compared to NeatImage, is similar. I used the
separate demo utility from ASF/Kodak but I was disappointed by the
posterized results, and the lack of control (which it apparently
needed). NeatImage is much more flexible in addressing difficult
images, and it handles less challenging images fine (no softening of
image detail, but still removal of visible noise and highly
adjustable).

I don't know if Minolta's implementation is any better, but
getting a new preview for each change in GEM settings, and
the fact that it didn't do all that much compared to NeatImage,
made it a quickly forgotten feature.

(I don't have enough badly faded slides to compare ROC with
increasing the saturation in PhotoShop).

When I tested ROC (after being handed a demo CD at Photokina 2002 by
an ASF representative), I was unimpressed (due to risk of
posterization)by the 8-bit/channel file limit and by the level of
"automatic restoration". Maybe more recent versions are more capable,
so I can't judge that from personal experience, maybe my files haven't
faded enough (call me a control freak, but again lack of control
caused failed attempts, or was it the software?).

Bart

 

[Anoni Moose]

You seem to be under the illusion that GEM would be of no benefit with
GD, which is certainly not the case.

It also appears that you don't believe, having successfully licensed and
applied one Kodak product (ICE), that Minolta would have extended that
license to the full ICE3 package at the earliest opportunity (just as

Nikon did!) whether the grain dissolver was included or not.

Other part(s) of the ICE3 package seemed to already be in the
'version 1's software package, presumably for other devices
that Konica-Minolta sells. They were disabled when connected
up to the '5400 I'. I therefore assume they just didn't
want to pay the royaltees, it wasn't a matter of "getting to
it". I remember ROC in particular had nice big pictures
in the scanner's manual which one had to ignore.

Mike

P.S. - Unless you've got hard evidence as to Minolta's intent,
I might point out that you're just making assumptions
as well.

 

[Kennedy McEwen]

Fortunately (for me) I did this test already. See
http://misc.hq.phicoh.net/gem/details.txt

Between
http://misc.hq.phicoh.net/gem/ngem4.png
and
http://misc.hq.phicoh.net/gem/ngem0-NI.png
I know which one I prefer.

BOTH images are significantly overdone in terms of grain reduction and,
as I said, both algorithms fall over in different ways when used in
excess. In this particular image where the subject is a man made object
little looks particularly out of place on the subject when the grain
reduction is driven to the extreme - even though detail HAS been lost.
However on a natural object, such as the background, with surface
texture that level in both NeatImage and GEM is excessive. In your
example, much of the background detail has gone together with the grain,
as evidenced by how much background detail is visible in the
overprocessed GEM image that has been lost from the NI image!

Furthermore, it simply isn't possible to apply as much over-processing
in GEM as in NI, hence the top level of 4 does not plasticise the image
nearly as much as you have produced in NI. As I mentioned earlier, the
maximum level in GEM is far too much for general images, so why you
consider that to be a suitable demonstration level is beyond logic.

And there is not that much difference between level 4 GEM and the blur in
http://misc.hq.phicoh.net/gem/ngem0-g15.png

You clearly have a taste for overprocessed and artificial looking
images!

 

[Kennedy McEwen]

P.S. - Unless you've got hard evidence as to Minolta's intent,
I might point out that you're just making assumptions
as well.

Which is exactly the point I am making - without evidence all that is
left is assumption and just changing your point of view results in a
completely different assumption being made and conclusion being reached.

Some seem to think that because Minolta included GEM with the MkII it
*PROVES* that the diffuse illumination capability has been removed.

However, GEM may well have been introduced as a product update or, more
likely, that with a *SINGLE* scanner configuration (ie. *only* a diffuse
or collimated source) GEM could be tailored to a single known scanner
signature.

Consequently, the inclusion of a feature that was not included in the
original has absolutely no bearing whatsoever on whether the light
source of the MkII is diffuse (as the MkI was when the grain dissolver
was used) or collimated (as the MK1 and every other Minolta scanner was
without the grain dissolver or Scanhancer). To assume either way is
futile.

 

[miso]

I'd like to upgrade from my Artixscan 4000t and am looking at the
Dimage 5400I myself. I am a Vuescan user, so I am waiting for Ed to
support it.

Regarding the 5400 II, will the LED lighting cause it to have the depth
of field problems associated with Nikon scanners and film that isn't
perfectly flat?

 

[Philip Homburg]

You clearly have a taste for overprocessed and artificial looking
images!

Possibly. That's why I find GEM completely useless. Given that this a
4000 dpi scan, the sort of details that got lost are the ones that you
won't see in a print anyhow.

 

[Anoni Moose]

Some seem to think that because Minolta included GEM with the MkII it

*PROVES* that the diffuse illumination capability has been removed.

The LED source they now use also traditionally hasn't been
very diffuse. But I haven't take one of the new ones
apart to see if they did something non-traditional
rather than "look like Nikon" which it otherwise appears.
They MAY have decided that a big feature of their older
version is somthing that, although continued in the new
one, should no longer be highly featured in advertising,
what's more, not even be mentioned at all in the new model.
I MAY win the lottery next month too.

I only think it looks like a duck, quacks like a duck,
and smells like a duck. But you're right, it MAY
not be a duck! No DNA evidence....

However, GEM may well have been introduced as a product update or, more
likely, that with a *SINGLE* scanner configuration (ie. *only* a diffuse
or collimated source) GEM could be tailored to a single known scanner

signature.

Do you have proof of that? Else this is just a guess
of yours.



Mike

 

[Hecate]

I'd like to upgrade from my Artixscan 4000t and am looking at the
Dimage 5400I myself. I am a Vuescan user, so I am waiting for Ed to
support it.

Don't hold your breath.

--

Hecate - The Real One
(e-mail address removed)
Fashion: Buying things you don't need, with money
you don't have, to impress people you don't like...

 

[Loren Sauer]

I'd like to upgrade from my Artixscan 4000t and am looking at the
Dimage 5400I myself. I am a Vuescan user, so I am waiting for Ed to
support it.

Don't wait. The included DiMage Scan works quite well. I prefer it over
Vuescan with this scanner.

 

[miso]

I've used Vuescan since 2001. [Back then there was only one version.
The early users got automatic upgrades to the pro version.] It's a
great program if you know what you are doing.Think of it like a stick
shift in a car. Some people can drive a stick, and some can't.
Actually, it's more like a paddle shifter. You can use vuescan on
automatic most of the time, but can process the difficult images by
hand if you have to.

What I really lack is a scanner that can multisample without moving the
CCD. It really annoys me that the Polaroid version of the Artixscan
4000t can do this, but not Microtek's version. Had I known, I would
have bought the Polariod. Incidentally, there was a firmware upgrade on
the Artixscan. Mine broke, and when they repaired it I got a ROM swap.
I'm not sure what they changed.

 

[Kennedy McEwen]

Possibly. That's why I find GEM completely useless. Given that this a
4000 dpi scan, the sort of details that got lost are the ones that you
won't see in a print anyhow.

If you won't see the details that a 4000ppi scan reveals there is no
point in scanning at 4000ppi in the first place!

 

[Kennedy McEwen]

I'd like to upgrade from my Artixscan 4000t and am looking at the
Dimage 5400I myself. I am a Vuescan user, so I am waiting for Ed to
support it.

Regarding the 5400 II, will the LED lighting cause it to have the depth
of field problems associated with Nikon scanners and film that isn't
perfectly flat?

Why should it? It isn't the LED light source that produces the limited
depth of field on the Nikon scanners, it is the fast, low f/# optic that
does that!

 

[Kennedy McEwen]

The LED source they now use also traditionally hasn't been
very diffuse.

Minolta have not "traditionally" used LEDs. Nikon have, but in a unique
scanner configuration which is completely different from either the Mk1
or the MkII.

But I haven't take one of the new ones
apart to see if they did something non-traditional
rather than "look like Nikon" which it otherwise appears.

Another mistake that you are making - it does not "look like Nikon"!

Nikon use 4 optically co-located LEDs and a single line CCD which is
exposed in sequence to all 4 colours. The Minolta uses a tri-linear CCD
exposed simultaneously and hence has no need for an optical arrangement
to image each LED colour into the same location, so a white LED, or
several, can be used and they are perfectly capable of being made
diffuse.

I only think it looks like a duck, quacks like a duck,
and smells like a duck.

Your sense of smell, sight and hearing are defective!

 

[Philip Homburg]

If you won't see the details that a 4000ppi scan reveals there is no
point in scanning at 4000ppi in the first place!

There are plenty of reasons to over-sample in a digital system.

 

[Bart van der Wolf]

There are plenty of reasons to over-sample in a digital system.

Indeed, and one of them could be the removal of high frequency noise
without altering too much real detail ...

Bart

 

[Kennedy McEwen]

There are plenty of reasons to over-sample in a digital system.

Firstly, you are not oversampling - as is evidenced by the fact that
detail is present in one image and lost in the other. You have
subsequently overprocessed an *undersampled* scan to the point where
even the real detail that is resolved in the original is lost. Clearly
you don't understand what oversampling is or what it achieves! Stop
using terms you don't understand to justify a procedure you can't!

 

[Philip Homburg]

Firstly, you are not oversampling - as is evidenced by the fact that
detail is present in one image and lost in the other. You have
subsequently overprocessed an *undersampled* scan to the point where
even the real detail that is resolved in the original is lost. Clearly
you don't understand what oversampling is or what it achieves! Stop
using terms you don't understand to justify a procedure you can't!

If you want your own sandbox to play in, go ahead. I won't bother you anymore.

 

[miso]

I thought the wide aperture optics of the coolscan is to compensate for
the low light levels of the LEDs relative to that of a CCFT.

Incidentally, at the last CES Sony demonstrated a LCD TV that used
white led backlighting rather than CCFTs.

 

[Kennedy McEwen]

I thought the wide aperture optics of the coolscan is to compensate for
the low light levels of the LEDs relative to that of a CCFT.

Not exactly, but it is necessary to compensate for the LEDS and their
optical configuration. The MkII Minolta uses a white LED system to
replace the cold cathode tube of the Mk1, it doesn't use the same
multiple LED optical arrangement of the Nikon, so it is unlikely to
require any of the other features necessary to make the Nikon work.

Incidentally, at the last CES Sony demonstrated a LCD TV that used
white led backlighting rather than CCFTs.

Illustrating my case perfectly - white LEDs, as bright and diffuse as
the cold cathode lamps they replace!

Why assume that the MkII Minolta has the same optical restrictions as
the Nikon when the technology to simply replace the CC lamp and GD with
a white bright extended diffuse source (in other words - do the job
properly!) is widely available and almost certainly cheaper to implement
than Nikon's approach?

 

[miso]

I don't assume the 5400 II will have the same problems, but wonder if
anyone has evaluated the scanner to see if it has the same Coolscan
problems. If the optics haven't changed between versions of the 5400,
it is unlikely the new version will suffer from the Nikon depth of
field issues.