Vuescan - Minolta

[Ralf R. Radermacher]

From Ed's website:

What's new in version 8.1.39

* Improved Minolta Scan Elite 5400 calibration
* Improved auto-focus on some Minolta scanners
......

Ralf

 

[Fernando]

From Ed's website:

What's new in version 8.1.39

* Improved Minolta Scan Elite 5400 calibration
* Improved auto-focus on some Minolta scanners

Didn't have a chance tro try .39, but .40 is totally, totally broken
on my 5400.
Absurd colors and "interlaced" scan appearance (just as only half the
lines were scanned).
Of course I tried deleting the old .ini and I re-calibrated the
scanner... no luck.
I reported to Ed... but I wonder if he actually tries the new
releases. Luckily I always store all the releases I download, but it's
annoying anyway.

Fernando

 

[Fernando]

Didn't have a chance tro try .39, but .40 is totally, totally broken
on my 5400.
Absurd colors and "interlaced" scan appearance (just as only half the
lines were scanned).
Of course I tried deleting the old .ini and I re-calibrated the
scanner... no luck.

I now get proper scans after having initialized and calibrated the
scanner with Minolta Scan Utility!
Duh...

I'll see if the streaks are gone and report here.

Fernando

 

[Wilfred]

I now get proper scans after having initialized and calibrated the
scanner with Minolta Scan Utility!
Duh...

I'll see if the streaks are gone and report here.

That's strange - version 40 seems to work flawlessly with my DSE 5400
without calibration through Minolta SW. The previous version I used was
8.1.29.
Interesting that you mention an interlacing effect - I remember there
was a time VueScan produced these lines with my 5400, too. They were
hardly visible, except when USM was applied to the image.
By the way I noticed that I can now apply Shadow/highlight adjustments
in Photoshop CS to my VueScan-based scans without getting the lots of
noise in the shadows I used to get, both with older versions of VueScan
(before 8.1.12) *and* the Minolta software. I must say that I had
already given up on using this feature long time ago but I applied it to
my 'Bake-off' sample slide, which I scanned with VueScan 8.1.40, and it
worked fine.

 

[Fernando]

Still the same streaks...:

http://gundam.srd.it/PhotoPages/images/5400_vuescan_darkframesub_04.jpg

dark frame subtraction nicely gets rid of them, but I would have
preferred not having them in the first place.

Fernando

 

[Mendel Leisk]

From Ed's website:

What's new in version 8.1.39

* Improved Minolta Scan Elite 5400 calibration
* Improved auto-focus on some Minolta scanners
......

Ralf

--
Ralf R. Radermacher - DL9KCG - Köln/Cologne, Germany
private homepage: http://www.fotoralf.de
manual cameras and photo galleries - updated Jan. 10, 2005
Contarex - Kiev 60 - Horizon 202 - P6 mount lenses

Even if calibration can be sorted, or solved through work-around of
pre-calibrate through Minolta Scan Utility, I would like to see him do
SOMETHING about the current state of Vuescan cleaning. Not sure if it's
specific to my Elite 5400, or a problem with other scanners as well:

Working with 64 bit Vuescan raw file, setting the cleaning filter to
light, medium, or heavy, I see progressively cleaner image in preview.
However, when I save at ANY of these levels, the result is roughly the
same, cleaning at a level about half way between light and none.

The only change between settings, in the saved files, appears to be
progressively softer grain. Also, the actual cleaned areas are heavily
and obviously softened, compared to ICE.

 

[Kennedy McEwen]

Still the same streaks...:

http://gundam.srd.it/PhotoPages/images/5400_vuescan_darkframesub_04.jpg

dark frame subtraction nicely gets rid of them, but I would have
preferred not having them in the first place.

Perhaps you should offer Ed a deal for the use of your code, Fernando!
;-)

 

[Linus]

dark frame subtraction nicely gets rid of them,

Whatt's dark frame subtraction? Is it a custom function/tecnique you made on
your own? Or is it publicly available?

Thanks,
Linus (DSE 5400 owner)

 

[Fernando]

Whatt's dark frame subtraction? Is it a custom function/tecnique you made on
your own? Or is it publicly available?

Please cfr. the thread "D.F.S. v.0.1 (alpha-test)": it's a freely
available command-line utility (source included) that I wrote after
Kennedy's suggestions; it works by subtracting a "dark frame" scan
from the original image. This gets rid of "lines" that are visible on
some scanners, and are due to per-cell variation of residual dark
current noise (insufficient calibration); it may also improve shadow
reproduction a bit (by better defining the black point).

Please read the mentioned thread for details and download link.

Bye!

Fernando

 

[Don]

Perhaps you should offer Ed a deal for the use of your code, Fernando!
;-)

LOL! Rubbing salt in the wound... ;o)

Careful, though, some may accuse you of "bashing" - even if that puts
you in good company... ;o)

BTW, Fernando, has His Edness responded to your emails at all, or is
His Eminence still sulking because a mere mortal appears to have
solved his 2-year old problem in 5 minutes?

Don.

 

[Fernando]

Perhaps you should offer Ed a deal for the use of your code, Fernando!

)
I'd be glad to donate my code to Ed, but I don't think he would
incorporate a d.f.s. pass into Vuescan: it would probably slow down
the scanning too much. I told him about my tests, but he did not seem
interested in the technique.
But, this time he seems to have taken my issue more seriously (maybe
he have a bit more time to invest?): he is studying a variant of the
calibration routine. I'm testing a new version right now (it's Sunday
night here).

I sincerely hope it will work.

Fernando

 

[Bart van der Wolf]

SNIP

I'd be glad to donate my code to Ed, but I don't think
he would incorporate a d.f.s. pass into Vuescan: it
would probably slow down the scanning too much. I
told him about my tests, but he did not seem
interested in the technique.

He'd probably need to average some 40 rows during calibration for a
decent approximation.

But, this time he seems to have taken my issue more
seriously (maybe he have a bit more time to invest?):
he is studying a variant of the calibration routine. I'm
testing a new version right now (it's Sunday night here).

I sincerely hope it will work.

Yes, so we can go on and e.g. improve the IR cleaning effectiveness
for him ;-)
Then address the wrong D-max (gamma linearization) issue, and then
there is little left to wish for.

Bart

 

[Wilfred]

Then address the wrong D-max (gamma linearization) issue, and then there
is little left to wish for.

Which, indeed, seems to have resurfaced. Yesterday I scanned the
'bake-off' image with VueScan 8.1.40 and noticed that the left side of
the histogram was looking funny.

 

[Wilfred]

From Ed's website:

What's new in version 8.1.39

* Improved Minolta Scan Elite 5400 calibration
* Improved auto-focus on some Minolta scanners
......

What's new in version 8.1.41

• Fixed problem with Minolta Scan Elite 5400 with bad CCD pixels

 

[Bart van der Wolf]

SNIP

Even if calibration can be sorted, or solved through work-
around of pre-calibrate through Minolta Scan Utility, I would
like to see him do SOMETHING about the current state of
Vuescan cleaning. Not sure if it's specific to my Elite 5400,
or a problem with other scanners as well:

Part of it *is* specific to the DSE-5400 in general, or more specific
the much higher resolution. 'Filling in the blanks' is much easier if
the area to be filled is a few pixels. That could e.g. be done with a
median blur of surrounding pixels. With the DSE-5400 the defects are
82% larger than with a good 4000 ppi scanner, which calls for a better
(slower) algorithm than for almost all other scanners.

Of course it is Ed's choice to either do it right or not. There may be
a couple of group contributers that could help with suggestions (I'm
thinking about an enhanced type of Gaussian Pyramid interpolation),
but without him paticipating in this group it would take a lot of
effort to get the info (and associated sense of urgency, and award)
across.

Working with 64 bit Vuescan raw file, setting the cleaning
filter to light, medium, or heavy, I see progressively cleaner
image in preview.
However, when I save at ANY of these levels, the result is
roughly the same, cleaning at a level about half way
between light and none.

I never use anything stronger than light cleaning, but what you
describe seems easy to reproduce for Ed, so if he can he could change
some of the threshold settings.
Have you reported it to him? It won't change without him knowing it.

Bart

 

[Fernando]

Yes, so we can go on and e.g. improve the IR cleaning effectiveness
for him ;-)
Then address the wrong D-max (gamma linearization) issue, and then
there is little left to wish for.

Indeed, those three issues are the only things keeping Vuescan from
being perfect, at least from a technical standpoint (the GUI is not
the best in the world, IMHO; but I can live with it).
But on the other sides, those are 3 -important- issues.
I've seen that sometimes Ed is very active in pursuing bug reports
(see my present experience with the streaks), so maybe if we all
attach eloquent image samples and report the same problems...

Fernando

 

[Fernando]

calibration routine. I'm testing a new version right now (it's Sunday
night here).

I sincerely hope it will work.

It didn't; and the new 8.1.41 still shows streaks on my unit.
Maybe my 5400 is particularly unfortunate... anyway, I think that in
order to kill streaks, I'll have to deal with DFS, for a while at
least.

Fernando

 

[Mendel Leisk]

I emailed him March 06. I included a small crop of a 64 bit raw, plus
equiv cropped jpegs output at "no", "light", "medium" and "heavy"
cleaning. Also, a screen capture of display at "heavy" cleaning.

He responded March 10, to report he'd released 8.1.39 with improved
Elite 5400 calibration, and that that might fix the cleaning problem I
was experiencing.

I tried this version, saw no improvement, replied to his email (same
date), and have heard nothing since.

 

[Bart van der Wolf]

SNIP

I tried this version, saw no improvement, replied to his email (same
date), and have heard nothing since.

VueScan is up to version 8.1.41 last time I looked, so things may have
progressed.

Bart

 

[Mendel Leisk]

Well, Ed emailed me again lately, to say my cleaning problem is still
"on his list". You have to give the guy "A" for responsiveness. I've
emailed Minolta a couple of times, usually got no reply, or something
that looks like it was put out by software, without human intervention.