Minolta 5400 streaks

[Wilfred]

I strongly suspect this to be a communications problem. Here's why:

Communicating with peripherals directly is not something casual
programmers can - or should try to - do. Writing low level drivers is
not for the faint of heart - or for the petulant... ;o)

There is a (good!) reason why scans with Minolta software take
significantly longer. Commercial software is usually very conservative
by making sure the communication is reliable and that takes time (be
it due to handshaking or timing). By cutting corners one can speed
this up, but at the risk of dropping data.

Your hypothesis seems inconsistent with common observations that the
streaks always appear at the same locations. A communications problem
would lead to streaks at random locations and I don't see how it could
lead to streaks across the length of the scan.

You suggest the Minolta Software is better than VueScan. Numerous users
have been complaining here about clipping in the highlights when
scanning negatives with the Dimage Scan Utility. Despite a handful of
upgrades that came out since the introduction of the 5400, Minolta never
cared to solve this prolem. To me it seems that Ed Hamrick is at least
more responsive to his users than Minolta is.

 

[David O'Rourke]

A bit more information.

Using the Scan Elite, I scanned a black slide with Vuescan and Minolta's
utility. Pumping up the shadows produced streaks in both scans. The
Vuescan scan had considerably more horizontal lines of various colours
resembling noise. The Minolta scan was pretty clean with a couple of hard
defect lines. I was able with some large bars (multiple pixels) of noise to
match them up in the scans. I was unable to match single pixel lines.

David

 

[wim wiskerke]

It may be interesting to know that a on a German filmscanner forum, a
Nikon LS-5000 user has reported similar streaks when using VueScan. He
claims that the streaks started to appear after version 8.0.4.
For those who understand German:
http://f27.parsimony.net/forum67102/messages/5833.htm

I was about to ask about those streaks here.

They are vertical streaks. Not horizontal like the ones in the Minolta
scans. The Nikon did have a lot of horizontal streaking in dark areas
in the past.

(I have no experience with the Minolta, but I did operate the Canon
4000 and the Nikon 5000 next to each other for some time, both under
Vuescan.)

There is a hint of vertical streaks in the NikonScan scans.
If you move your navigator sliders in Photoshop up-down, you will see
a pattern (of 1 pixel wide streaks). If you move your sliders
left-right you will see no pattern.
In Vuescan it is very clear: There are alternating lines of higher and
lower density of 1 pixel wide in dark areas. I remember having read
somewhere that the nikon 5000 scanner reads with two rows of ccd's.
I suspect we are seeing these two ccd's here. And the <I> of TDI is
not working properly.
That said, the effect is not very apparent.
While under older versions of Vuescan I had a lot of horizontal
streaking. Those streaks were very visible in some B/W conversions I
used.
The vertical streaks are not visible while in actual pixels view.
As far as I can see, the horizontal streaking has vanished completely.

From the samples of the Minolta scans it seems to me that the scanners
have some hardware issues that the Minolta software is able to even
out during a calibration run. This is not uncommon in digital imagery
(remember Hubble?).
It allows for sloppy manufacturing. Provided you have good software,
nobody will notice.
I suggest Minolta owners go back to early scans and compare streaks
with recent scans. Maybe the Minolta calibration will simply
compensate for dust on the ccd. In digital camera's Olympus does that
in it's e-10 and e-20 with hot or stuck pixels.

I happened upon these vertical streaks while I was wondering if
Vuescan was exaggerating some form of chromatic aberration. (I think
it is - a whole new can of worms.)

For those who don't:
http://tinyurl.com/5uzcp

rotfl



regards, wim (who is somewhat ashamed for coming here with his
problems only)

 

[Hecate]

You suggest the Minolta Software is better than VueScan. Numerous users
have been complaining here about clipping in the highlights when
scanning negatives with the Dimage Scan Utility. Despite a handful of
upgrades that came out since the introduction of the 5400, Minolta never
cared to solve this prolem. To me it seems that Ed Hamrick is at least
more responsive to his users than Minolta is.

Minolta doesn't solve their problem, so they "never cared to".

Ed doesn't solve Vuescan's problem so "he's more responsive than
Minolta"

Ever heard the phrase "double standards?

 

[Wilfred]

Minolta doesn't solve their problem, so they "never cared to".

Ed doesn't solve Vuescan's problem so "he's more responsive than
Minolta"

Ever heard the phrase "double standards?

You are twisting my words. "So" was your addition, "at least" was
omitted by you. Yes, I have more confidence that at some time Ed will
solve the VueScan problems than that Minolta will solve the DSU problem.
This is from experience with two Minolta scanners. The first, a DSS
2800, had the same problem with clipping in the highlights when scanning
negs. I then switched over to VueScan (version 5.x) which handled the
problems nicely. It did have other problems by then but they were all
solved, even though it took some time. So was my problem with IR
cleaning on the 5400.
You and Don would have good reasons to kick against VueScan if there
would be any perfect scanning software to replace it. There isn't. Not
the Minolta software, not Silverfast (just have a look at the thread
"Nikon LS-30/2000 and SF6" started by Ralf Rademacher).

 

[wim wiskerke]

I happened upon these vertical streaks while I was wondering if
Vuescan was exaggerating some form of chromatic aberration.

It is not: it is a focusing problem exaggerating the CA.
I have no solution yet.

regards, wim

 

[celebrate]

You are twisting my words. "So" was your addition, "at least" was
omitted by you. Yes, I have more confidence that at some time Ed will
solve the VueScan problems than that Minolta will solve the DSU problem.
This is from experience with two Minolta scanners. The first, a DSS
2800, had the same problem with clipping in the highlights when scanning
negs. I then switched over to VueScan (version 5.x) which handled the
problems nicely. It did have other problems by then but they were all
solved, even though it took some time. So was my problem with IR
cleaning on the 5400.
You and Don would have good reasons to kick against VueScan if there
would be any perfect scanning software to replace it. There isn't. Not
the Minolta software, not Silverfast (just have a look at the thread
"Nikon LS-30/2000 and SF6" started by Ralf Rademacher).

Unlike Ed, other producers of less-than-perfect scanning software don't
come to this ng to bash those who point out the products' shortcomings.

 

[Wilfred]

Unlike Ed, other producers of less-than-perfect scanning software don't
come to this ng to bash those who point out the products' shortcomings.

In that respect the question is, of course, who started bashing. There
are some people on this ng whose highest goal in life is to wage flame
wars about VueScan. And no, I don't mean Ed Hamrick - currently he, like
Minolta, doesn't come to ths ng anyway. Only Silverfast does, sometimes.

 

[Don]

Your hypothesis seems inconsistent with common observations that the
streaks always appear at the same locations.

Is this *literally* at the same place (same byte numbers) regardless
of color depth or image size or, indeed, the scanner?

If so, then it looks to me like a 5-minute job to fix it. I always
assumed it was a spurious error (indicating a timing problem) as those
are the hardest to fix, but if the bug is easily reproducible then...

A communications problem
would lead to streaks at random locations and I don't see how it could
lead to streaks across the length of the scan.

I've read reports of both horizontal and vertical streaks. Anyway,
theoretically, a communications problem could produce all sorts of
patterns depending on timing and other conditions.

You suggest the Minolta Software is better than VueScan. Numerous users
have been complaining here about clipping in the highlights when
scanning negatives with the Dimage Scan Utility.

No, I only said that commercial software is much more conservative
when it comes to *communicating* with the device and err on the side
of caution. This is why it usually takes much longer.

Years ago I wrote a low level floppy driver which speeded up floppy
access by 300%. However, in certain circumstances it would drop data
as the system simply could not cope. To make it reliable I had to drop
the acceleration down by "only" doubling the speed. That shows how
careful commercial software is and how large their margin of error is.

As to clipping (if it helps any), once I turned off everything "Auto"
in NikonScan (in particular Auto Exposure!) there was no clipping.

Despite a handful of
upgrades that came out since the introduction of the 5400, Minolta never
cared to solve this prolem. To me it seems that Ed Hamrick is at least
more responsive to his users than Minolta is.

Even if we ignore his emotional outbursts, just repeating "it's been
fixed" is not being responsive when the bug is still there after a
year.

Like I wrote elsewhere, a reputable programmer would (at the very
least!) immediately remove the scanner from the supported list until
the problem has really been fixed. A real decent person would also
apologize and offer to give money back to those who were mislead into
a purchase by false claims.

By contrast, his "responsiveness" is limited to hurling obscenities
and being bad tempered.

Don.

 

[celebrate]

In that respect the question is, of course, who started bashing. There
are some people on this ng whose highest goal in life is to wage flame
wars about VueScan. And no, I don't mean Ed Hamrick - currently he, like
Minolta, doesn't come to ths ng anyway. Only Silverfast does, sometimes.

The critics of Vuescan started with bug reports on the product, and NOT
with bashing of Ed in person. In fact they NEVER bash Ed in person. When
Ed failed to fix the bugs, he responded with personal attacks of the
critics. That is BASHING.

 

[Fernando]

From the samples of the Minolta scans it seems to me that the scanners
have some hardware issues that the Minolta software is able to even
out during a calibration run. This is not uncommon in digital imagery
(remember Hubble?).
It allows for sloppy manufacturing. Provided you have good software,
nobody will notice.

I think you are absolutely spot on!
Still, the need to compensate for those streaks exists, and Minolta
Scan Utility and Silverfast both cope with it with good success.
Since I can't afford the excellent but very expensive Silverfast AI
and I hate the Minolta Scan Utility for its inability to focus
properly in a reliable way, I'd like to cope with this problem by
post-processing Vuescan streaked scans.
Any idea?

Thanks!

Fernando

 

[Dierk Haasis]

Still, the need to compensate for those streaks exists, and Minolta
Scan Utility and Silverfast both cope with it with good success.

I am not so sure about that. If the problem is with the hardware, the
software you mentioned cannot do an honest job of fixing. All it can
do is apply several filters (like blur) to cover up the streaks and
noise.

If, as you mentioned in an earlier post, the highest resolution and
sharpness is your goal, software that reduces effective res, applies
blur and other filters, and does perhaps shift the colours around a
bit, is not what you want.

No, I am not saying Silverfast or Minolta's sw actually do this. But
it is a possibility. Which would also clear up why VS doesn't do it,
as its design philosophy is another one.

Just a thought.

 

[Robert Feinman]

I think you are absolutely spot on!
Still, the need to compensate for those streaks exists, and Minolta
Scan Utility and Silverfast both cope with it with good success.
Since I can't afford the excellent but very expensive Silverfast AI
and I hate the Minolta Scan Utility for its inability to focus
properly in a reliable way, I'd like to cope with this problem by
post-processing Vuescan streaked scans.
Any idea?

Another observation to add to the mix. If I turn on the 5400 it wants
to be calibrated before use. I then start up the Minolta software and
it does this. I then quite the Minolta scan program and start up
Vuescan. At this point Vuescan recognizes that the scanner has been
"calibrated" and does not attempt to do this again.
So this would seem to indicate that "calibration" is a hardware or
firmware process residing in the scanner somehow.
If this is true then it seems odd that one program would get streaks
and another won't.
I think some investigation of noise removal, exposure settings,
clipping levels and perhaps even pixel averaging in the various
packages might yield some insights. At the moment I have no ideas
on how to test this, but I'll continue to think about.

On a separate issue, what do you all think of the idea of having
a separate Vuescan discussion group? Not on usenet, but somewhere
else, perhaps yahoo groups or the like. It could be moderated or
unmoderated and there would be an understanding that it is product
specific so questions about keeping usenet non-commercial would not
come up.
I asked Ed if he would like to start such a forum, but he said he
didn't have time to run it. Opinions?

 

[Fernando]

I am not so sure about that. If the problem is with the hardware, the
software you mentioned cannot do an honest job of fixing. All it can
do is apply several filters (like blur) to cover up the streaks and
noise.

If, as you mentioned in an earlier post, the highest resolution and
sharpness is your goal, software that reduces effective res, applies
blur and other filters, and does perhaps shift the colours around a
bit, is not what you want.

Dierk, I'm no newbie in the scanning world. I know when an imaging
software tries cheap tricks.
Not Minolta Scan Utility nor Silverfast use filters that lower the
actual resolving power of this scanner in a visible way. I can see it
pretty easily by running Imatest on Slanted Edge Test target slides.
The ultimate resolution (that is very good, by the way, as posted time
ago) is the same for Silverfast, Vuescan and Minolta Scan Utilty.
For me, it is a calibration issue with Vuescan. It fails to properly
recognize "weak" photodiodes and so it fails to properly compensate
the levels to cover the streaks.

Fernando

 

[Fernando]

Another observation to add to the mix. If I turn on the 5400 it wants
to be calibrated before use. I then start up the Minolta software and
it does this. I then quite the Minolta scan program and start up
Vuescan. At this point Vuescan recognizes that the scanner has been
"calibrated" and does not attempt to do this again.
So this would seem to indicate that "calibration" is a hardware or
firmware process residing in the scanner somehow.
If this is true then it seems odd that one program would get streaks
and another won't.

We have been through this (maybe not deeply enough), even with Ed
Hamrick's contribution, some time ago.
Yes, the firmware does perform some kind of initial calibration, and
stores some kind of information in a proper storage area inside the
scanner's memory (it seems to exist an appropriate RAM area into the
scanner's chipset). Or least, this is what Ed said.

Now the odd things:

1) Ed said something like "Yes, Minolta software writes the
calibration data inside the scanner's memory. Vuescan blanks that area
when it starts, but maybe it does not blank it properly. I'll have to
look into that".

Now: if Minolta software writes useful calibration data inside a
scanner's memory, why just not using them to compensate for the
streaks? I know, one should be forced to run M.S.U. before Vuescan;
but it's still better than nothing. I asked this to Ed, but received
no replies.

2) Vuescan performs its own calibration after the initialization. The
result of this operation is a binary file called "vuescan.se5", about
400KB.
I don't know if the contents of this file are then written into the
scanner's memory as Minolta Scan Utility seems to do.
I sent this file to Ed during a bug report, and he said "I see nothing
strange in the calibration file".

3) Regardless of if I run Minolta Scan Utility before Vuescan or not,
if I run the proprietary Vuescan calibration one time, two times or
never, if I switch on the scanner just before the calibration or not,
if I turn on GD/ICE or not, I still get streaks in Vuescan.
If I run Silverfast after having initialized the scanner with some
other software (MSU or Vuescan), I get lots of streaks. But if I
"re-initialize" the scanner under Silverfast, the streaks are gone.

On a separate issue, what do you all think of the idea of having
a separate Vuescan discussion group? Not on usenet, but somewhere
else, perhaps yahoo groups or the like. It could be moderated or
unmoderated and there would be an understanding that it is product
specific so questions about keeping usenet non-commercial would not
come up.
I asked Ed if he would like to start such a forum, but he said he
didn't have time to run it. Opinions?

Why not! I hate Yahoo groups for you get spammed by death, but hey, I
can setup a fake email address.
It seems a good idea to me.

Thanks Robert.

Fernando

 

[TimothyJP]

I am not so sure about that. If the problem is with the hardware, the
software you mentioned cannot do an honest job of fixing. All it can
do is apply several filters (like blur) to cover up the streaks and
noise.

If, as you mentioned in an earlier post, the highest resolution and
sharpness is your goal, software that reduces effective res, applies
blur and other filters, and does perhaps shift the colours around a
bit, is not what you want.

No, I am not saying Silverfast or Minolta's sw actually do this. But
it is a possibility. Which would also clear up why VS doesn't do it,
as its design philosophy is another one.

Just a thought.

The 5400 streak problem was widely reported soon after the product came
to market. After Minolta released rev 1.1.3 (?? check this), the reports
subsided considerably, and many users who experienced this problem found
the streaks had "disappeared". This release included firmware update,
which must be verified.

If the new release did make a difference, then the streaks were
"corrected" or "minimized" by Minolta's newer sw or firmware. But it
would still be inconclusive whether the problem's source was in the hw.
Third party sw developers like Vuescan or SilverFast might be aware of
this kind of information, but don't expect them to tell us.

 

[Wilfred]

On a separate issue, what do you all think of the idea of having
a separate Vuescan discussion group? Not on usenet, but somewhere
else, perhaps yahoo groups or the like. It could be moderated or
unmoderated and there would be an understanding that it is product
specific so questions about keeping usenet non-commercial would not
come up.
I asked Ed if he would like to start such a forum, but he said he
didn't have time to run it. Opinions?

It's OK if he doesn't have time to run it but it would be nice if he
would participate. Did he say anything about that?
The fact that yahoo groups require to be run by a moderator might be a
pro. OTOH, the moderator should be someone who is unbiased, which seems
difficult when it comes to VueScan.

Alternatives could be http://www.photo.net/community/forums or
http://www.photo-i.co.uk/cgi-bin/ikonboard/ikonboard.pl

 

[Kennedy McEwen]

The fact that yahoo groups require to be run by a moderator might be a
pro. OTOH, the moderator should be someone who is unbiased, which seems
difficult when it comes to VueScan.

Are you suggesting that Don might not be a candidate moderator?? ;-)

 

[Aguyathome]

The 5400 streak problem was widely reported soon after the product came

to market. After Minolta released rev 1.1.3 (?? check this), the reports
subsided considerably, and many users who experienced this problem found
the streaks had "disappeared". This release included firmware update,
which must be verified.

If the new release did make a difference, then the streaks were
"corrected" or "minimized" by Minolta's newer sw or firmware. But it
would still be inconclusive whether the problem's source was in the hw.
Third party sw developers like Vuescan or SilverFast might be aware of
this kind of information, but don't expect them to tell us.

I bought a 5400 late this past summer and I love it. I do all my scans with 16x
sampling, at 5400dpi, which makes a 233mb that looks incredible. Never had a
streak, but I have had a few scans come out dark and had to re-do them. Other
than that, and the fact that it takes an hour or two to do a single scan at the
highest setting, I wouldn't trade the thing for anything out there. It even
scan better than the $200,000-up drum scanners that we had when I worked at
Primedia Publishing in the late '90s.

Take care,
JD

 

[Anoni Moose]

It's OK if he doesn't have time to run it but it would be nice if he
would participate. Did he say anything about that?
The fact that yahoo groups require to be run by a moderator might be a
pro. OTOH, the moderator should be someone who is unbiased, which seems
difficult when it comes to VueScan.

Yahoo groups can be used for a moderated forum, but it doesn't have
to be fully (I'm the "leader" for one group). Whomever sets up
a group can require postings to be reviewed (moderated) or not. Of
course in the "not" case, it's still moderately moderated in that a
poster can be kicked off or can be individually set to having postings
reviewed (something I do) if they've a history of problem posting. So
when I "do" one, it's "tamed" somewhat but not squeeky clean. Mostly
a matter of keeping folk to the topic and not people-bashing.

If I get into the mood one of these nights I may set one up and see
what happens. More I think about it, the more likely I'll do it.

Mike