Minolta 5400, Vuescan, 1px lines

[Maxim Kammerer]

Hi,

I own a Minolta Dimage Scan Elite 5400 and I get 1px high lines in
scanning direction.

I'm scanning only negatives. The lines are mostly red, there are
around 5 major lines and lotsa minor, and when looking close there are
also a few green/yellow ones. I put up a picture at
http://www.directupload.net/images/050206/3gk642v7.jpg
(first part example for yellow/green, other 2 parts red lines)

These lines occur only with Vuescan and only when I enable the grain
dissolver. Scanning with the Minolta-Software in any configuration or
with Vuescan without the grain dissolver give no lines.

Things like reinitialising or initialising with the Minolta-Software
don't help.

The lines are always at the same position.

First I thought the lines were there only at pictures with high
contrast, but meanwhile I think, they are always there, but sometimes
less visible.

Any ideas?

Yours,
Maxim

 

[Kennedy McEwen]

Hi,

I own a Minolta Dimage Scan Elite 5400 and I get 1px high lines in
scanning direction.

I'm scanning only negatives. The lines are mostly red, there are
around 5 major lines and lotsa minor, and when looking close there are
also a few green/yellow ones. I put up a picture at
http://www.directupload.net/images/050206/3gk642v7.jpg
(first part example for yellow/green, other 2 parts red lines)

These lines occur only with Vuescan and only when I enable the grain
dissolver. Scanning with the Minolta-Software in any configuration or
with Vuescan without the grain dissolver give no lines.

Things like reinitialising or initialising with the Minolta-Software
don't help.

The lines are always at the same position.

First I thought the lines were there only at pictures with high
contrast, but meanwhile I think, they are always there, but sometimes
less visible.

Any ideas?

Yes, Vuescan isn't implementing the dark current calibration correctly.
This has been a problem for Ed to resolve, because Minolta aren't
providing him with any support, so he is working blind. He appears to
be gradually making some progress, according to the public posts and
questions concerning it, but it has been a long slow slog and, based on
your account, isn't completely over yet.

 

[Hecate]

Hi,

I own a Minolta Dimage Scan Elite 5400 and I get 1px high lines in
scanning direction.

I'm scanning only negatives. The lines are mostly red, there are
around 5 major lines and lotsa minor, and when looking close there are
also a few green/yellow ones. I put up a picture at
http://www.directupload.net/images/050206/3gk642v7.jpg
(first part example for yellow/green, other 2 parts red lines)

These lines occur only with Vuescan and only when I enable the grain
dissolver. Scanning with the Minolta-Software in any configuration or
with Vuescan without the grain dissolver give no lines.

Things like reinitialising or initialising with the Minolta-Software
don't help.

The lines are always at the same position.

First I thought the lines were there only at pictures with high
contrast, but meanwhile I think, they are always there, but sometimes
less visible.

Any ideas?

Yes. Don't use Vuescan with a Minolta scanner. The software can't
cope.

 

[Wilfred]

Hi,

I own a Minolta Dimage Scan Elite 5400 and I get 1px high lines in
scanning direction.

I'm scanning only negatives. The lines are mostly red, there are
around 5 major lines and lotsa minor, and when looking close there are
also a few green/yellow ones. I put up a picture at
http://www.directupload.net/images/050206/3gk642v7.jpg
(first part example for yellow/green, other 2 parts red lines)

These lines occur only with Vuescan and only when I enable the grain
dissolver. Scanning with the Minolta-Software in any configuration or
with Vuescan without the grain dissolver give no lines.

Things like reinitialising or initialising with the Minolta-Software
don't help.

Did you try to calibrate the scanner from within VueScan? Are you using
the latest version of VueScan?

 

[Bart van der Wolf]

SNIP

Did you try to calibrate the scanner from within VueScan? Are you
using the latest version of VueScan?

And do try the VueScan (re-)calibration just before scanning your
negative. As the scanner warms up, the electronics may need
recalibration. Also try to avoid external electro-magnetic sources
like speakers/phones/monitor by creating some distance. That will not
guarantee success, but it will help the odds.

There are apparently things happening in the scanner that are hard to
reverse engineer for the author of VueScan. Apparently Konica/Minolta
isn't being of much help to him either.

Bart

 

[Fernando]

Hi,

I own a Minolta Dimage Scan Elite 5400 and I get 1px high lines in
scanning direction.

Unfortunately, it's not uncommon.
I have the same problem even with the latest version of Vuescan
(albeit it's much better now than some months ago. Vuescan improved
the handling of this issue over time).
It has to do with the calibration of the sensor response, and Vuescan
seems to miss some trick that Minolta Scan Utility exploits.
Try the latest version of Vuescan (and remove any old vuescan.ini file
that could be sitting around somewhere), launch the Minolta Scan
Utility first, so that it can initialize the scanner before Vuescan,
and Calibrate the scanner (Scanner->Calibrate) just before the actual
scan.
If this does not help, you'll have to leave Vuescan and stick to
Minolta Scan Utility (or consider Silverfast), until Ed Hamrick will
finally (hopefully) sort this calibration issue once and for all.

Bye!

Fernando

 

[Robert Feinman]

Hi,

I own a Minolta Dimage Scan Elite 5400 and I get 1px high lines in
scanning direction.

I'm scanning only negatives. The lines are mostly red, there are
around 5 major lines and lotsa minor, and when looking close there are
also a few green/yellow ones. I put up a picture at
http://www.directupload.net/images/050206/3gk642v7.jpg
(first part example for yellow/green, other 2 parts red lines)

These lines occur only with Vuescan and only when I enable the grain
dissolver. Scanning with the Minolta-Software in any configuration or
with Vuescan without the grain dissolver give no lines.

Things like reinitialising or initialising with the Minolta-Software
don't help.

The lines are always at the same position.

First I thought the lines were there only at pictures with high
contrast, but meanwhile I think, they are always there, but sometimes
less visible.

Any ideas?

Yours,
Maxim

I wonder about these types of concerns. Under what conditions would a
1px line be visible in a final image? Under any normal usage the final
image is not going to be kept at the 5400 dpi setting. That would imply
a printed image of about 18x27 inches or so at 300dpi. Even at that
size the defect would be 1/300 inch in size. You would need to look
pretty closely to actually see it.
Under any other scenario the defect will be down-sampled and averaged
to create the final product, either print or online.

Sometimes, in real life, we have to settle for less than perfection.

On a more positive note, the latest version of Vuescan now produces a
"scan" view at the same time as the "preview" view. The scan view shows
the final image with the cropping applied and thus can save a step if
you like to do a scan without saving first.

 

[rgbcmyk]

Minolta's native sw works fine for me on the 5400. What are you trying
to achieve with Vuescan that cannot be done with the native sw?

 

[Wilfred]

Minolta's native sw works fine for me on the 5400. What are you trying
to achieve with Vuescan that cannot be done with the native sw?

Although I'm not Maxim, I can mention a few things that VueScan can do
on the Minolta 5400, and that the Minolta software can't:
* cleaning a scan based on IR data without having the grain dissolver
switched on
* scanning negatives without clipped highlights
* profiling the scanner using IT8 targets
* profiling printers using IT8 targets (although I haven't foun out how
to make this work)
* adding copyright information to the file info

I have certainly left out some other features but these alone might be a
reason to use VueScan.

 

[Fernando]

I have certainly left out some other features but these alone might be a
reason to use VueScan.

More reliable AF, ability to automatically correct the scanner
response with an ICC profile (and re-convert the pic to whatever
profile you like), ability to store the IR channel with the image (to
do ad-hoc manual corrections with Photoshop), ability to tweak the
level of the dust-cleaning action (3 levels of strength), sensible MF
interface (I love numbers), larger histograms (also in Log mode), Film
profiling.

And I'm sure I leaved something out as well.

Fernando

 

[Don]

Yes, Vuescan isn't implementing the dark current calibration correctly.
This has been a problem for Ed to resolve, because Minolta aren't
providing him with any support

It's not Minolta's job to do that!

They provide the SDK. That's all the support you can expect from a
company! If they didn't provide the SDK then you could blame them.

He *chooses* to bypass this SDK - for whatever reason - and talk to
the scanner directly, at which point it becomes his problem, not
theirs.

You can't expect Minolta to help him bypass the officially sanctioned
method and *get mired* in his own programming incompetence! They
provide the SDK exactly in order to avoid this!

Fact 1: Minolta uses their own SDK drivers and has no problems.

Fact 2: SilverFast uses Minolta's SDK drivers and has no problem.

Fact 3: VueScan *bypasses* the SDK and *has problems*!

Sure, it's VueScan's prerogative to bypass the SDK - for whatever
reason! - but with that choice comes *responsibility*! Any problems
caused by such willful action are *self-inflicted* which nobody else
can be blamed for.

Don.

 

[Don]

I wonder about these types of concerns. Under what conditions would a
1px line be visible in a final image?

While that may be true literally, it misses the point completely.

There are many aspects of scanning which, due to *objective* physics
of scanning, may only be visible under magnification but may not
visible in the final product (e.g. grain). Perfectionists (counting
yours truly) do worry about such minutiae and - every now it then -
it's very helpful to, indeed, step back and ask oneself the above
question.

However, and that's a big "however", the decision to ignore such an
artifact only applies if it's the result of the *objective* physics of
scanning (e.g. grain) and not of programming incompetence.

When such problems are self-inflicted and a result of *subjective*
programming inadequacies, then it's a totally different story. These
VueScan Minolta stripes are a clear *indication* of how sloppy VueScan
programming is!!

Therefore, the question is not "Can I see this 1px line?" but "What
other such programming bugs are there?". In case of VueScan, as
numerous complaints by frustrated users illustrate almost daily, there
are indeed very many.

Don.

 

[Kennedy McEwen]

It's not Minolta's job to do that!

I didn't suggest it was, nevertheless, it is a statement of fact.

 

[Bart van der Wolf]

SNIP

And I'm sure I leaved something out as well.

Like being able to save Raw files (never having to rescan the delicate
original) including IR channel (take benefit of improvements in the IR
cleaning algorithm), and it can drive different scanner brands with
the same interface, and it has a built-in "densitometer", and it's
available on multiple OS platforms (e.g. Linux), and it provides
visual clipping warnings, and Logarithmic histogram display, and I
probably also left out some things I use less frequently... ;-)

Bart

 

[Fernando]

and Logarithmic histogram display

Hey, I said that!!!

....so I'll add the faster focusing, faster start-up, built-in negative
film profiles for many films, no writing into Windows registers,
useable email support (try writing Minolta...), consistent CCD
exposure setting interface (never really understood Minolta Scan
Utility one), ability to launch external viewers/programs and the end
of the scan, ability not to save a scan if you don't like it, film
base color correction (lockable)...

;-)

Just now, Hamrick put out 8.1.30.
It adds single-pass multi-scanning for my Polaroid SS120, a feature
that the Polaroid Scan software never supported. Hamrick sorted this
out from the twin-brother Microtek Artixscan 120TF.

Now if only Ed could put an end to those calibration / dynamic range
issues with Minolta scanners...

Fernando

 

[Bart van der Wolf]

Hey, I said that!!!

Sorry, I lost track of all the benefits ;-)

[...]

consistent CCD exposure setting interface (never really
understood Minolta Scan Utility one),

[...]

To me it looks like plus or minus 2 exposure control results in plus
or minus 1 "stop" or double/half the base exposure. That would allow a
maximum correction factor of +/- 4 (combining both master and RGB
controls).

Bart

 

[Hecate]

I wonder about these types of concerns. Under what conditions would a
1px line be visible in a final image?

Depends on the direction of the main light source, even a 1px line can
look like a canyon <g> OK, it's not hat bad, but if you're producing
a largish image for display and the line/light source are aligned the
right (or wrong depending on your point of view) way even small
defects can become visible.

 

[Wilfred]

sensible MF
interface (I love numbers)

Fernando,

Could you explain how you can put these numbers to use? I would love to
have more control over focusing but I haven't the faintest idea about
how to manipulate these numbers in order to get the sharpest focus. The
only approach I could think of is trial-and-error, trying dozens of
settings and comparing the results at 1:1-pixel magnifcation. That would
be too time-consuming for me because it has to be repeated for every scan.

You and Bart, thanks for the additions.
One more:
Ability to preview the areas in your scan where clipping will occur (R,
G and B separately), and ability to preview where the dust and the
scratches are.

 

[Fernando]

Fernando,

Could you explain how you can put these numbers to use?

When I insert a set of slides, they are typically curved/bulged all in
the same direction (it's inherent of the way film dries). So if I
manage to acquire a good focus on one of them (for example because it
presents a visible, sharp edge where I can put the cross on), I have a
reference focus number, with sign, that comes handy if then the
scanner has troubles with the other slides of the same batch (for
example, because they lack proper sharp spots where I can put the
cross on).
Say I get "-0.145" as the focus value for the "good" slide; if the
second one says "0.20", I know it's very out of focus. At that point I
can either move the focus point to another spot, focus manually by
trial & error analyzing partial scans, or simply put it "-0.145"
knowing I'll be in the ballpark at least.
I'd love the Minolta software to give a similar indication... for it's
even worse at auto-focusing.

Ability to preview the areas in your scan where clipping will occur (R,
G and B separately), and ability to preview where the dust and the
scratches are.

Good one!

Fernando

 

[Bart van der Wolf]

Fernando,

Could you explain how you can put these numbers to use?

SNIP

There are several ways I use it:
1. Focus on center (by dragging the focus marker and select
Scanner|Focus from the menubar) and notice the relative focus number
it generates. As the scanner heats up the film (doing a few previews
can help to speed it up), it is likely to change shape. So after some
acclimatization re-focus from the menu. Repeat untill the number
stabilizes around a certain value. The film has now reached a certain
equilibrium.
2. Since the film is often somewhat warped/curled (especially the
first/last frame from a strip), sample at several points across the
crop area and note the minimum/maximum values. By typing in the
average number you'll optimize depth of field for that frame.
Additional benefit is that the film is even less likely to change
shape during the final scan pass because it had more time to
acclimatize, with the lamp turned on.
3. Sometimes you may need the absolutely best possible focus of a
small area (e.g. a slanted edge test target itself, or an image of one
on film). By making a sequence of multiple scans with small increments
of the manual focus settings, you can be sure that one image will be
the best possible.

One more:
Ability to preview the areas in your scan where clipping will occur
(R, G and B separately), and ability to preview where the dust and
the scratches are.

Yes, and also important if you have a colormanaged workflow, is the
visual "out of gamut" clipping warning. That may prompt to use a wider
gamut output colorspace like "ProPhoto RGB", and optionally convert in
Photoshop to a different space using the desired rendering intent.

Another benefit of VueScan is that it allows to preview at
maximum/scan resolution, and then "scan from preview" (no need to scan
the film again after preview, so no loss of time).

And another benefit is that VueScan can simultaneously save to
multiple fileformats (with different sizes) from a single scan. You
can e.g. produce an original TIFF and/or Raw for archiving and a small
version for for email while building a separate index.

Bart