Nikon Scan Problems

[Chris Birkett]

I'm in the middle of a large slide scanning project, and I've been having
some serious issues with Nikon Scan. The version is 4.0.2, and I'm using it
with a Coolscan V.

Whenever I scan slides, they end up looking extremely soft, because the
autofocus totally misses. I did about 200 slides as tests before starting
on the actual scans, and while I noticed that many were very soft, I put it
down to poor technique, crappy cameras, etc. (these are all 20-50 years old,
and shot by my grandparents - not exactly Ansel Adams with a view camera).
I didn't think there was a problem until I scanned a slide of a sign showing
lots of small text. The scan was soft as usual, but there was absolutely no
fine detail whatsoever. I could see more detail with my homebrew loupe
(50mm lens reversed).

I fired up Vuescan, which I'd previously decided not to use because of
colour issues, and scanned the same frame. The Vuescan result was quite
significantly sharper. I went back into Nikon Scan and played around with
the settings, including manual focus, etc. I did eventually manage to get a
similar result after several tries, but I'm now finding it takes several
scans to produce a sharp result on almost any slide.

I'm doing about 1200 slides, so previewing and adjusting focus on each one
is not an option for me. I would be happy to use Vuescan, but the colour of
the final scan is rarely anywhere close to the original slide. Of course,
in Nikon Scan, the shadows often end up being far too dark, so either way
I'm screwed. I've tried various settings in Nikon Scan, but it makes no
difference, I always end up with a dark, soft result. I've put up a small
gallery with a couple examples to show what I mean, I would appreciate it if
you could check out the individual images:
http://www.pbase.com/bob_mcbob/scanner/

It's worth noting that I don't have any sharpness problems with negative
film. The difference between a typical slide scan and negative scan is like
night and day. I can see wisps of hair in the negatives, and I can barely
make out faces in the slides. Colour accuracy is also much better for
negatives, which is exactly the opposite of what I've heard in the past.

One reason I chose the Coolscan V over the Minolta 5400 was the better
autofocus with slides I often saw noted in reviews. Am I doing something
wrong here?

- Chris

 

[Kennedy McEwen]

I'm doing about 1200 slides, so previewing and adjusting focus on each one
is not an option for me. I would be happy to use Vuescan, but the colour of
the final scan is rarely anywhere close to the original slide. Of course,
in Nikon Scan, the shadows often end up being far too dark, so either way
I'm screwed. I've tried various settings in Nikon Scan, but it makes no
difference, I always end up with a dark, soft result. I've put up a small
gallery with a couple examples to show what I mean, I would appreciate it if
you could check out the individual images:
http://www.pbase.com/bob_mcbob/scanner/

There is certainly no question that your Nikonscan images are defocussed
compared to the others, however the clue *may* be that you are getting
sharp NikonScan results on negatives. As you may be aware, Nikonscan
can be set to operate with a different sequence of processes scanning
negatives and positive film, and also different for single scans and
multiple scans.

Under the Preferences section, look at *both* single scan and batch scan
settings and confirm that you have autofocus checked in both instances.
If you only have the autofocus action set on your Preview Settings then
it will only autofocus when a preview (*not* a thumbnail) is scanned.
After making these changes, close NikonScan, restart it and confirm that
it starts with autofocus checked, just in case you are also having
problems with the default settings.

Also, what does the scan look like if you perform a manual "auto-focus"
operation, rather than manually adjusting focus? This is the chequered
button on the Preview window - click that to autofocus on the centre of
the frame.

Finally, check that the autofocus cursor is positioned to an area in the
image frame - if you have set this to include some of the slide mount
then results can be wrong. Best results are usually achieved with the
autofocus cursor positioned about 1/3 of the way between edge and centre
of the frame. With Windows, use Ctrl-Click on the Autofocus button to
set the actual autofocus position. Once you have set this, you may need
to press the "Settings/Set User Settings" option for this to become the
default position for your scans, otherwise it will return to its
original default on your next slide.

 

[wim wiskerke]

I'm in the middle of a large slide scanning project, and I've been having
some serious issues with Nikon Scan. The version is 4.0.2, and I'm using it
with a Coolscan V.

Whenever I scan slides, they end up looking extremely soft, because the
autofocus totally misses. I did about 200 slides as tests before starting
on the actual scans, and while I noticed that many were very soft, I put it
down to poor technique, crappy cameras, etc. (these are all 20-50 years old,
and shot by my grandparents - not exactly Ansel Adams with a view camera).
I didn't think there was a problem until I scanned a slide of a sign showing
lots of small text. The scan was soft as usual, but there was absolutely no
fine detail whatsoever. I could see more detail with my homebrew loupe
(50mm lens reversed).

I fired up Vuescan, which I'd previously decided not to use because of
colour issues, and scanned the same frame. The Vuescan result was quite
significantly sharper. I went back into Nikon Scan and played around with
the settings, including manual focus, etc. I did eventually manage to get a
similar result after several tries, but I'm now finding it takes several
scans to produce a sharp result on almost any slide.

Where is your focus point in NikonScan?
Is there absolutely no part of the picture in focus?
Do you use GEM? (under Ice)

I'm doing about 1200 slides, so previewing and adjusting focus on each one
is not an option for me. I would be happy to use Vuescan, but the colour of
the final scan is rarely anywhere close to the original slide. Of course,
in Nikon Scan, the shadows often end up being far too dark, so either way
I'm screwed. I've tried various settings in Nikon Scan, but it makes no
difference, I always end up with a dark, soft result. I've put up a small
gallery with a couple examples to show what I mean, I would appreciate it if
you could check out the individual images:
http://www.pbase.com/bob_mcbob/scanner/

In the Vuescan Input tab do have your Media set to <Image> ?
Try in the Color tab the settings <neutral> <white balance> or <auto
levels>.

regards, wim

 

[Chris Birkett]

"wim wiskerke" <wimATwiskerke.com> wrote in message

Where is your focus point in NikonScan?
Is there absolutely no part of the picture in focus?
Do you use GEM? (under Ice)

The focus point is whatever Nikon Scan selects with autofocus, I don't
change it manually when doing bulk scanning. The entire image is soft,
exactly like the examples I posted. Of course, on some boxes of slides,
autofocus works perfectly. I think Nikon Scan just hates me.

In the Vuescan Input tab do have your Media set to <Image> ?
Try in the Color tab the settings <neutral> <white balance> or <auto
levels>.

I usually have it set to white balance, both black and white points 0, and
either generic color slide film or ektachrome.

- Chris

 

[Chris Birkett]

I'm in the middle of a large slide scanning project, and I've been having
some serious issues with Nikon Scan. The version is 4.0.2, and I'm using it
with a Coolscan V.

I added another photo to show a problem I'm having with Vuescan (another
reason why I'm loath to use it). The IR cleaning sometimes works extremely
poorly on regular colour slides (no Kodachrome here). On large open areas
like skies I can often see little specks of dust that ICE in Nikon Scan
totally removes. It also leaves strange artifacts from time to time where
it performs the cleaning. They range from blurry patches to large 'holes'
like the example shows. I've been using it on the 'medium' setting.

- Chris

 

[(Pete Cresswell)]

RE/

I fired up Vuescan, which I'd previously decided not to use because of
colour issues, and scanned the same frame.

This is from somebody who knows next to nothing...so take it as such.

What about using VueScan to scan to raw image files and then using NikonScan to
convert same into the desired output? Seems like that would work around the
focus issue, but let NikonScan make the decisions about color.

 

[Wilfred]

"wim wiskerke" <wimATwiskerke.com> wrote in message



The focus point is whatever Nikon Scan selects with autofocus, I don't
change it manually when doing bulk scanning. The entire image is soft,
exactly like the examples I posted. Of course, on some boxes of slides,
autofocus works perfectly. I think Nikon Scan just hates me.




I usually have it set to white balance, both black and white points 0, and
either generic color slide film or ektachrome.

As Wim says (implicitly), you shouldn't use the 'slide' setting in
VueScan, and not-at-all you should you use any of the 'specific' film
profiles ever. They simply don't work, or at least I've never heard from
anyone who ever had success with them.
As Wim says (explicitly), use 'image' instead of 'slide'. It doesn't
sound logical but it works. It's what Ed Hamrick recommends, too.

There *is* a way that 'slide' can be used successfully with certain
Nikon scanners by using the 'advanced workflow' in the VueScan manual.
It doesn't work with my Minolta, but Erik Krause has reported success
with it on an LS-40 in this newsgroup, IIRC.

 

[wim wiskerke]

I added another photo to show a problem I'm having with Vuescan (another
reason why I'm loath to use it). The IR cleaning sometimes works extremely
poorly on regular colour slides (no Kodachrome here). On large open areas
like skies I can often see little specks of dust that ICE in Nikon Scan
totally removes. It also leaves strange artifacts from time to time where
it performs the cleaning. They range from blurry patches to large 'holes'
like the example shows. I've been using it on the 'medium' setting.

Yes this is a big problem. I found out this is a focus problem too.
With the slide in focus, there are no holes. When out of focus, the
holes appear. Also the chromatic aberration increases, further
enlarging the holes. Velvia is really bad in this respect.
I have been doing a lot of focus testing on my cs 5000 these days. I
found wildly varying readouts even with a really very flat piece of
microfilm. Both with Nikon Scan and Vuescan. So it is a hardware
issue. Or at least machine-based. (Could be the firmware.)

Others have mentioned the inaccuracy of the slide holder already.
And the curvature of slides are a big problem.
Also slide mounts vary in thickness from side to side. Paper mounts
tend to be very unstable. Sofar I get best results using Gepe mounts
with metal masks inside. I am waiting for a filmholder FH-3 to
arrive. I hope it will bring some stability to this scanner. (It is a
bloody shame it is not included with the scanner. It is out of stock
everywhere in the US and Europe. This must tell Nikon something.)

I had no luck looking for an explanation of the focus problem or even
the focus mechanism or proces on the Coolscans.
Does anybody here know how focusing works on the Coolscans?
Has anybody done some focus testing?

In a previous thread there has been some discussion about the units
used to measure focusing in both Vuescan and Nikon Scan. How long can
the total of the focus track be? And what could the resolution be? I
found different results for every unit in Vuescan. It did not map to
the closest step available, but seems to actually correspond with a
step. (I used jpeg size to judge this, just like the *best picture*
features in some digital camera's do.)

Having more focus points would help: think AF camera's. Are there any
scanners using this?

regards, wim

 

[wim wiskerke]

"wim wiskerke" <wimATwiskerke.com> wrote in message


The focus point is whatever Nikon Scan selects with autofocus, I don't
change it manually when doing bulk scanning.

Yes, but where is it?
In the Layout Tools, choose the crosshair.

regards, wim

 

[Ken]

I usually have it set to white balance, both black and white points 0, and
either generic color slide film or ektachrome.

I always use Neutral, White Balance does some odd things. If I'm doing a
large number of slides I can batch process them in Photoshop to do Color
Correction. Any images that don't look correct I then go back to the
original scan and manually work on it.

 

[Erik Krause]

There *is* a way that 'slide' can be used successfully with certain
Nikon scanners by using the 'advanced workflow' in the VueScan manual.
It doesn't work with my Minolta, but Erik Krause has reported success
with it on an LS-40 in this newsgroup, IIRC.

Thats true. Interesting that it shouldn't work on Minolta or doesn't
negative work either? The methode the film base color is determined is
the same for slide and for color negative. You can try this if you scan
a raw negative scan as slide or vice versa.

The only point is that for a negative there is usually the film border
as clear area from which the film base color can be determined but
there isn't such an area for slide. Hence media type Slide only works
with advanced workflow (but this should be the case for Minolta, too,
since it is pure post processing that can be done even from raw scan
files with no scanner present).

 

[Chris Birkett]

I've been doing a lot of testing of Vuescan and Nikon Scan over the last
couple days, and I've been experiencing another problem I can't explain.
I'm not sure whether it's worth posting a new thread about it, so I'll leave
it as a reply here.

I've put up a couple of examples under the original post's url. When I'm
scanning in Vuescan, the preview image/histogram usually isn't the same as
the final scan. Sometimes it's COMPLTELY different, to the point that the
preview is totally useless for judging anything whatsoever. In the example
I posted, the original is a fairly low contrast image of a slightly
overexposed/faded slide. Very easy to add a little contrast and clean it
up. The final scan is a horrible overexposed mess! Check out the
difference between the two histograms. I've included the colour settings in
the screenshots so you can get an idea of the settings I used, but it's the
same no matter what I change.

Any ideas? I'm stumped. I'm finding colour in Vuescan is a total
crapshoot, and the ROC equivalent is nowhere near as effective as ROC.

- Chris

 

[Wilfred]

Thats true. Interesting that it shouldn't work on Minolta or doesn't
negative work either? The methode the film base color is determined is
the same for slide and for color negative. You can try this if you scan
a raw negative scan as slide or vice versa.

The only point is that for a negative there is usually the film border
as clear area from which the film base color can be determined but
there isn't such an area for slide. Hence media type Slide only works
with advanced workflow (but this should be the case for Minolta, too,
since it is pure post processing that can be done even from raw scan
files with no scanner present).

Strange enough, the advanced workflow does work with negs on my Minolta
but when using it with slides, the resulting image has a terrible color
cast. I don't remember exactly - I tried only once - but I think the
cast was reddish. Perhaps the base color sample wasn't sufficiently
overexposed? Anyway I like the results my scanner gives with the 'image'
setting and the 'primitive' workflow.

 

[degrub]

Low contrast slides are hell for autofocus as well.
Frank

 

[ThomasH]

I've been doing a lot of testing of Vuescan and Nikon Scan over the last
couple days, and I've been experiencing another problem I can't explain.
I'm not sure whether it's worth posting a new thread about it, so I'll leave
it as a reply here.

I've put up a couple of examples under the original post's url. When I'm
scanning in Vuescan, the preview image/histogram usually isn't the same as
the final scan. Sometimes it's COMPLTELY different, to the point that the
preview is totally useless for judging anything whatsoever. In the example
I posted, the original is a fairly low contrast image of a slightly
overexposed/faded slide. Very easy to add a little contrast and clean it
up. The final scan is a horrible overexposed mess! Check out the
difference between the two histograms. I've included the colour settings in
the screenshots so you can get an idea of the settings I used, but it's the
same no matter what I change.

Any ideas? I'm stumped. I'm finding colour in Vuescan is a total
crapshoot, and the ROC equivalent is nowhere near as effective as ROC.

I am astonished about your judgment. Vuescan's stronghold is its
ability to provide a perfect color balance, while NikonScan will
always demand from you an image per image manual tweak, very tedious.

My experience with both NikonScan and Vuescan is approx. 4-5 years.
NikonScan's color management is very simple minded and barely usable.
I use LS4000 + Sa30 roll film adapter + SF200 slide feeder.
I scan using Vuescan and ICC profiles made using Wolf Faust's
IT8 targets. I use only scanner targets, what is an approximation,
a compromise. Technically speaking I should have also made film
targets and use them as well. As it is, the optical difference
between my scanner profiles and the buildin profile is often not
recognizable, nut in some cases it is. Foremost, I am getting
much better gradation in shadows.

While looking at your examples from
http://www.pbase.com/bob_mcbob/scanner,
I cannot follow your opinion. Let's take a look at the White House
(images without numbers, I mean 3rd row from the top.) NikonScan
image renders the house as violet blue, while both Silverfast and
Vuescan do a much better job in restoring its white color.

All scans suggest that they are drastically underexposed. You
should not use Abode RGB in Vuescan, when you monitor images
on sRGB device, such as a common computer screen.

The histograms which you have posted are very strange indeed.
The preview histogram suggests that the slide was severely
underexposed during preview.

Maybe you could try out the workflow which I practice for a long
time: Use NikonScan with Nikon Color Management off, max bit depth,
apply ICE (and GEM if desired.) Never use ROC, unless you have really
faded historic slides. This will provide you with smooth 48bit *.tiff
files for archiving.

http://www.pbase.com/phototalk_thh/use_ns312_and_vue

Use Vuescan with these files, what will allow you to make
numerous experiments, quite fast. You could also post one of
such raw *.tiffs to pBase and let us experiment with it.


Scanning is not a straight forward process, you will have to
develop some craftsmanship and experience to achieve good
results. I just played with your Vuescan scan and by raising
gamma *drastically* and doing some contrast/color saturation
I got quite a nice looking image.

Probably most of us were *stunned* by the bad results at the
beginning. I needed 2-3 months before I begun to have results
which I could consider "extreme close to projected slide" or
to an negative image printed in a good lab! Every of the scanning
programs has its "dark sides," it is simply so that film scanning
compared to digital image processing is a marginal business.
Not enough research flows into this discipline and every
scanning program and exposure algorithm fails in some situation.

Sometimes I think that best 'slide scan' and a high-quality
macro image made using any of the top DSLR's!! I making now
some experiments using EOS-20D and I will post results.

Thomas

 

[Wilfred]

You
should not use Abode RGB in Vuescan, when you monitor images
on sRGB device, such as a common computer screen.

This is not true. If you specify a profile for your monitor (e.g., sRGB,
or a custom profile generated with a measuring device) VueScan will
adapt its output to the screen, whatever the color profile of the output
file is.
On the other hand, before you post images to the web, you should always
convert to sRGB.

 

[John K.]

The version is 4.0.2, and I'm using it with a Coolscan V.

With NikonScan 4.0.2, I get similar dissapointing results as yours when
I have "Image Enhancer ON" and "Digial Ice on Fine".

For best results, I turn OFF Image enhancer and Digital Ice on "Normal".

 

[ThomasH]

This is not true. If you specify a profile for your monitor (e.g., sRGB,
or a custom profile generated with a measuring device) VueScan will
adapt its output to the screen, whatever the color profile of the output
file is.
On the other hand, before you post images to the web, you should always
convert to sRGB.

And you claim that this profile would convert Adobe color space
to a profiled sRGB??? I will not exclude that such profile can
be generated, I think however that in this case you are mistaken.

 

[Norbert Preining]

Hi Thomas!

http://www.pbase.com/phototalk_thh/use_ns312_and_vue

Nice description. I have only one question: The images created by
vuescan directly and the one created via NikonScan->vuescan look rather
different, too. You state that the image directly from vuescan looks
like the one you see projecting the slide. But the one NS+VueScan really
looks -- well -- not poor, but quite different. Is there a reason for this?

Best wishes

Norbert

 

[ThomasH]

Hi Thomas!


Nice description. I have only one question: The images created by
vuescan directly and the one created via NikonScan->vuescan look rather
different, too. You state that the image directly from vuescan looks
like the one you see projecting the slide. But the one NS+VueScan really
looks -- well -- not poor, but quite different. Is there a reason for this?

I am not sure what causes the difference, except that by elimination
of possible reasons I can only suspect that ICE+GEM processing, which
I use for my raw files, in not neutral to the color balance!

Anyway, please take this method and its description with the
caution becasue it relate to older NikonScan and older Vuescan.

We do have a new NikonScan 4 now, which has changed quite drastically
in its color reproduction (without Nikon's description what and why.)
The nasty magenta cast in green reproduction is gone. Also the "raw
files" look completely different! NikonScan 3 raw files were very
dark, white was at approx. 60% level(!!), whereas NikonScan 4 raw
files have gamma elevated significantly and this provides much better
latitude for a processing in Vuescan or in Photoshop.

Furthermore, Vuescan made also significant advances. It supports
now color profiles and its color reproduction has improved because
Ed used the profiles to calibrate the build-in profiles as well.
Also many of the posterization problems in conjunction with black
point setting have been resolved.

I still use this workflow, albeit I am by myself not conclusive
regarding Media type: "To Image or not to Image" its the question!

Image seems to be working perfect with the scanner profile, but
"Media type=Slide" allows to setup the film base color.... In some
situations however (low contrast images or low temperature such as
late evening) "Media type=Slide" does not provide satisfactory
results or even fails completely, whereas type=Image seems to
be very linear in the processing of the raw data.

Now its the time for Christmas... We have to hit the road to
be ahead of the traffic!

Merry Christmas to everybody!

Thomas