high-iso scanning: minolta 5400 or nikon LS-50 ??

[Simon]

Hello,

I've been using a Minolta Scan Dual II to scan my films for the past
year. However, I had the unit on loan and recently had to return it.
Therefore I'm in the market for a new scanner. Based on my experience
with the DS2, it should

1) have an IR channel. With the DS2 I have noticed that dust spotting
is not my idea of fun...

2) deal better with grain in images. The DS2 scans often came out
horribly noisy (much more than prints or other scans). Probably due to
grain aliasing with my ISO 800 films?

The demand of an IR channel essentially narrows down my choice to two
models: the Minolta 5400 and the Nikon LS-50 (the canon fs-4000 being
older, scsi and with less than wonderful software). Based on some
research on the net and in magazines, I came up with the following
shortlist of strong and weak points of each of these units:

Minolta 5400
+ resolution (maybe less grain aliasing ?)
+ grain diffuser
- sloooow
- lamp drift over time

Nikon LS-50
+ recommended over the Minolta by Ed Hamrick, author of Vuescan, for
having better colours and channel separation
- shallow DOF
- worse Dmax for B&W (according to test in ColorFoto 3/2004)
+ light stability because of LED lightsource


Of course which scanner is best will greatly depend on my usage of the
machine. I shoot negative film almost exclusively, the majority of
rolls being high ISO colour film (400-1600). In addition I have a
decent amount of older B&W material, most of it also in the ISO
400-3200 region. Slides are less important. I regularly shoot one or
more rolls in bars and on parties and would like to be able to
batch-scan most of the images with reasonable results. Of course I
don't mind spending extra time on the _good_ shots.

Does anyone have experience with one or both scanners? Suggestions
concerning aspects I overlooked, overstressed or otherwise
misinterpreted? Any help is greatly appreciated!

Thanks,
Simon


PS - For those of you who read both the filmscanners mailing list and
comp.periphs.scanners , I apologize for double-posting this message.

 

[Kennedy McEwen]

I came up with the following
shortlist of strong and weak points of each of these units:

Minolta 5400
+ resolution (maybe less grain aliasing ?)
+ grain diffuser
- sloooow
- lamp drift over time

Nikon LS-50
+ recommended over the Minolta by Ed Hamrick, author of Vuescan, for
having better colours and channel separation
- shallow DOF
- worse Dmax for B&W (according to test in ColorFoto 3/2004)
+ light stability because of LED lightsource

Well, it makes a change when someone has actually done their homework -
its more difficult to make a recommendation. You have basically dug out
all of the facts on these scanners and their competitors and your key
points are pretty much the key issues.

One other thing that may make a difference is the way the film is
handled. On the Minolta the film strip is placed in a holder first and
this is inserted into the scanner. A similar option is also available
with the Nikon, and that eliminates the limitations of the shallow DoF
you refer to above. However there is a much more convenient mode as
well, using the SA-21 motorised adapter included with the scanner, where
the film strip is just inserted into the machine. Since nothing holds
the film flat, you can have DoF issues with curled film as mentioned in
the reviews, but the vast majority of the time it is perfectly OK.
However, since there is less handling of the negative, there is less
potential for damage - and it is much more convenient and faster. The
only down side of the direct feed is the frame size is 0.7mm shorter
than the 24mm width of a standard 35mm frame, which can be a problem if
you have some very tightly cropped originals, but there is always the
other adapter as a fall back.

For your main applications, colour negatives, the lower alleged Dmax of
the Nikon won't matter, although in theory it could with your secondary
applications, traditional monochrome film. Having said that, I haven't
encountered any problems with that type of film on the LS-4000, which is
basically the same scanner with support for bulk options and Firewire
instead of USB2. Being an issue with the denser regions of the
negative, and thus the image highlights, lack of Dmax is less critical
with negatives than it is with dense slides.

The grain dissolver on the Minolta is a neat addition, but it adds to
the scanning time - which may be related to your conclusion of "slow",
although you have to remember that the Minolta is producing twice as
much data, so it is inevitable that it will be slower. However I doubt
if the grain dissolver will make the slightest difference with your fast
films - or the higher resolution.

With 800ASA film, either scanner should deliver the goods and I doubt
that you will be able to tell any difference between them.

There have been a few reports of problems with the Minolta out of the
box - someone posting here recently had 3 on the trot that were faulty
although the 4th worked fine. That could have been teething problems
and the reports seems to have quietened down recently.

I am always reluctant to make recommendations based on what you do today
- all that will do is guarantee you will be unsatisfied tomorrow. So I
think you should also consider what stretch potential each of these
scanners offer you beyond your current work. That might help to make
the decision for you.

I don't own a Minolta, and nothing that has come onto the market since
buying my Nikon LS-4000 has persuaded me to upgrade... yet. ;-)

 

[Simon]

Thank you for your deailed answer! However, a few additional questions
popped up in my mind:

However I doubt
if the grain dissolver will make the slightest difference with your fast
films - or the higher resolution.

Wouldn't it do just what it says...dissolve the grain? Can you comment
on the appearance of high ISO scans from your ls-4000, if you ever use
that type of film?

I am always reluctant to make recommendations based on what you do today
- all that will do is guarantee you will be unsatisfied tomorrow. So I
think you should also consider what stretch potential each of these
scanners offer you beyond your current work. That might help to make
the decision for you.

That sounds like good advice. I have been toying with the idea of
shooting slide film occasionally. I presume that both of the scanners
do a good job on this. Maybe the Nikon will deliver a little more
consistent colour because of the LED light source?

Finally, can you comment on the quality of the colour inversion for
negatives in the NikonScan software? Can I expect reasonable results
from that, or should I switch to Vuescan (no problem, I already have a
license)?

If so, is there a possibility to make `default' scans of negatives in
NikonScan and apply ICE and GEM, doing the colour inversion later in
Vuescan? This would be my preferred workflow, as I have read that ICE
and GEM are better than their vuescan equivalents.

Thanks in advance!
Simon


PS - I would have checked the workflow questions myself, if only I had
access to the documentation. However, Nikon likes me to register one
of their products before I can download anything from their support
site Why they would want to do this is beyond me..

 

[Kennedy McEwen]

Thank you for your deailed answer! However, a few additional questions
popped up in my mind:


Wouldn't it do just what it says...dissolve the grain?

The grain dissolver was originally developed by Erik de Goederen and
sold through his organisation called Scanhancer, which have several
comparison images on their web site at http://scanhancer.iddo.nl/
showing you what it is capable of. It was copied by Minolta for the
Elite 5400 scanner. I say "copied" quite deliberately here because I
initially thought that they had licensed the technology from Erik, but
see no reference to this in any Minolta documentation and only a cryptic
comment at the bottom of the Scanhancer site to the effect that Minolta
have "implemented it". I assume that this was with Erik's consent since
he posts a response from Minolta on his site as:
"Thank you very much for your kind support with your diffusor." -
Minolta Europe

The Scanhancer is a cleverly simple device that overcomes grain
aliasing, which is primarily what causes the grain on scanned images to
appear more severe than it does in analogue prints, or indeed on
professional drum scanned images. Aliasing occurs because the scanner
CCD can respond to finer detail than it can resolve, which is not the
case with analogue systems. The unresolved information gets mixed in
with the resolved information and they can never be subsequently
separated. Scanhancer significantly reduces this unresolvable
information before it reaches the CCD using controlled diffusion. So it
virtually eliminates aliased grain, but does not eliminate the grain
itself - consequently the results are very similar to traditional
analogue prints.

Can you comment
on the appearance of high ISO scans from your ls-4000, if you ever use
that type of film?

Generally speaking I am very pleased with the grain levels in my scans
from all emulsion speeds. One of the main reasons I had for upgrading
to the LS-4000 was the reduction of grain aliasing which is a side
effect of the higher resolution, rather than the raw resolution itself.
The largest prints I make are around 13" wide, which corresponds to
200ppi on the page from my previous 2700ppi scanners - which is only
marginally below the resolution limits of a human eye. So going up to
4000ppi didn't really gain very much in the raw resolution stakes -
perhaps the ability to crop a little more but, since I try to crop at
the taking stage (35mm film doesn't have enough space on the film plane
to be magnanimous!) that was of little true value. However the
reduction in grain aliasing and the appearance of that grain on the
final print was very significant - the results just looked much more
comparable to the grain in analogue prints.

Some emulsions work better with a very slight defocus of the scanner
lens - only two or three steps from the optimum focus - but these are
generally medium speed, sharp grain negative emulsions, such as Superia.
The grain on high speed film is sufficiently coarse that the scanner has
adequate resolution to capture it without significant aliasing even at
the optimum focus. I have no doubt however that Scanhancer would reduce
the excessive visibility of grain even further though, and I note that
Erik is still looking for volunteers to evaluate his device in Nikon
(and Canon) scanners.

I tend to use GEM (Grain Equalisation Management - one of the ICE3/4
components) at its mid level. Like all of these software only
solutions, GEM has no means of discriminating between grain, aliased
grain or real image content other than guesswork, so it inevitably
introduces some image artefacts, mainly softening. The same is true of
NeatImage, which a lot of people recommend. With NeatImage you can
match the filter to the spatial characteristics of the grain structure
by analysing a relatively uniform area of the image which should only
contain grain, which should offer an advantage over GEM, but I have not
found that to be the case in practice. The default grain filters in GEM
work just as well as NeatImage and are faster.

That sounds like good advice. I have been toying with the idea of
shooting slide film occasionally. I presume that both of the scanners
do a good job on this. Maybe the Nikon will deliver a little more
consistent colour because of the LED light source?

It certainly delivers purer colours than the broad spectrum of the
alternatives, however it is not without its downside. The most common
of these is the "Kodachrome Problem", caused by the different (and
slightly variable) spectral characteristics of Kodachrome dyes from
those of conventional slide films. That can result in some colour
variations between batches of Kodachrome which are not visible with the
broad spectral response of the eye - or other scanner technologies. I
should emphasise that this is a very minor variability with modern
Nikon's like the LS-4000 or LS-50.

Generally speaking, slide film will produce images with less visible
grain than equivalent negative film. This is because the negative
emulsion records the image in a compressed density range and the scanner
software (or the chemical printing process) has to expand that out to
produce the density range in the image. This is a direct trade with the
increased exposure latitude of negative film, however expanding the
density range of the image recorded on the film also enhances the
visibility of the emulsion's grain structure.

Finally, can you comment on the quality of the colour inversion for
negatives in the NikonScan software? Can I expect reasonable results
from that, or should I switch to Vuescan (no problem, I already have a
license)?

I have no problems at all with the colours from NikonScan for either
slides or negatives. One tip is to set the black point on the emulsion
base, whether slide or negative, but particularly so with negatives - it
is completely unexposed and so, by definition, is the true black. This
only need be done once for each film type and stored for future use.
Black point errors compound once gamma adjustments are introduced and
this may be one of the main reasons that, in my opinion, it produces a
more dramatic improvement to colour balance than white or even mid-tone
balance. Even so, NikonScan provide eye-dropper tools so that adjust
the balance of white and mid grey based on specific points in each image
- however the default colour inversion, especially after black point
correction, is normally very close to ideal.

NikonScan defaults to using the ROC (Restoration of Colour) filter on.
Whilst this is intended for faded originals, it contains automatic white
balance which is the main advantage of Vuescan. However it also
enhances the colour saturation to a defined level, which I personally
find too severe, even at the lowest level. Consequently I normally
leave this turned off unless the original material really is faded.

If so, is there a possibility to make `default' scans of negatives in
NikonScan and apply ICE and GEM, doing the colour inversion later in
Vuescan? This would be my preferred workflow, as I have read that ICE
and GEM are better than their vuescan equivalents.

I haven't tried this combination myself, but it seems to be possible and
I understand some people are using it. Hopefully they will respond in
the thread.

PS - I would have checked the workflow questions myself, if only I had
access to the documentation. However, Nikon likes me to register one
of their products before I can download anything from their support
site Why they would want to do this is beyond me..

A recent 'innovation' - ostensibly so that they can concentrate on
responding to technical queries from real clients, but my own experience
is that this has not improved in the slightest. Similar to software
activation really - general inconvenience all round without delivering
any of the practical objectives it promised, ultimately losing clients.
:-(