Minolta 5400 - Vuescan problem

  • Thread starter Thread starter Robert Feinman
  • Start date Start date
R

Robert Feinman

I was using version 8.3.24 of Vuescan with the Minolta 5400 and
found an issue with IR cleaning. I've been scanning panoramas
(55mm wide) in two chunks. After setting everything up for the
first part I lock all the settings and do the scan.
I then shift the film to the same frame and do the second part.
In the second part I see streaks in the IR image. These run
along the scan direction and sort of look like processing defects.
They are not visible to the eye or the RGB image. These false
defects translate into visible streaks when cleaning is applied.
They are not the perfectly straight and even lines that occur when
the sensor calibration is off.

Since part of the film is scanned twice (the overlap), and the
IR defects only appear the second time, the problem lies with
the software.
I reverted to version 8.3.22 and the problem stopped. Since
some of the fixes listed since .22 involve IR cleaning and
Minolta scanners anyone having this setup my want to check and
see if they are affected.

I've written to Ed Hamrick, but not gotten a response as of yet.
 
Robert said:
I was using version 8.3.24 of Vuescan with the Minolta 5400 and
found an issue with IR cleaning. I've been scanning panoramas
(55mm wide) in two chunks. After setting everything up for the
first part I lock all the settings and do the scan.
I then shift the film to the same frame and do the second part.
In the second part I see streaks in the IR image. These run
along the scan direction and sort of look like processing defects.
They are not visible to the eye or the RGB image. These false
defects translate into visible streaks when cleaning is applied.
They are not the perfectly straight and even lines that occur when
the sensor calibration is off.

Since part of the film is scanned twice (the overlap), and the
IR defects only appear the second time, the problem lies with
the software.
I reverted to version 8.3.22 and the problem stopped. Since
some of the fixes listed since .22 involve IR cleaning and
Minolta scanners anyone having this setup my want to check and
see if they are affected.

I've written to Ed Hamrick, but not gotten a response as of yet.

I'm using the Linux version of 8.3.26 with the Minolta 5400. I can't
reproduce your streaks, but I find the IR cleaning does not work at all
now.

With the IR clean set to None, the IR channel shows some dust on a grey
background. With the IR Clean set to Medium the dust remains on the RGB
view, but the IR channel shows pure white. The pixel colours display
doesn't show any IR.

Reverting to 8.3.10 with the same slide it all works correctly.

So there is clearly something wrong.

Another problem I found was when the 'number of samples' is set to 4 or
more the scanner skips, and strips of the picture are repeated. Has
anyone else seen that?

John
 
Same here. Streaks in shades, visible only with IR switched on.

I saw the effect only in full resolution (5400dpi) scans.
Curiously the streaking was gone if I choose to scan, workign from raw
files, onyly the part of the fram containing the artefacts.

[...]
On Sun, 5 Feb 2006, it was written:
I'm using the Linux version of 8.3.26 with the Minolta 5400. I can't reproduce
your streaks, but I find the IR cleaning does not work at all now[...]

Yes, 3.25 and 3.26 do not have streaks but IR cleaning seems to be
(mostly?) gone.
With the IR clean set to None, the IR channel shows some dust on a grey
background. With the IR Clean set to Medium the dust remains on the RGB view,
but the IR channel shows pure white. The pixel colours display doesn't show
any IR.

Reverting to 8.3.10 with the same slide it all works correctly.

3.16 seems to be OK as well.

Another way of getting good IR cleaning is to scan in 16bits linear
using Minolta software and then "rescan" it using Vuescan.
Weird I know, but I like the colors I get from IT8 calibrated vuescan
better then than those from Minolta's soft.
So there is clearly something wrong.

Another problem I found was when the 'number of samples' is set to 4 or more
the scanner skips, and strips of the picture are repeated. Has anyone else
seen that?

No.

I all honesty I am getting tired of this hit and miss game.

As I said I still like the colors from Vuescan better then what I get from
manufacturers software and the IT8 calibration seems to be the key.
For this reason I'd be rather unhappy to go back to using Minolta and am
considering Silverfast. The colors from demo version of SE seem to be
closer to the original then Minolta's even without the calibraition.
I understand the Ai version includes IT8 calibration module, but
the demo version does not allow me to use it.

Robert, I know from your site that you have used this combination (5400
wih Silverfast). Did you use the calibration with IT8 target? If so, how
much difference does it make? It is not terribly "well defined" question
perhaps, but I'd grateful for any comments.

My main interest is slides.

Regards
MZ
 
On Sun, 5 Feb 2006 12:44:33 -0500, Robert Feinman

Curiously (and ironically) wasn't it you who - not so long ago -
posted thinly veiled spam gushing with praise of Vuescan?
I was using version 8.3.24 of Vuescan with the Minolta 5400 and
found an issue with IR cleaning. I've been scanning panoramas
(55mm wide) in two chunks. After setting everything up for the
first part I lock all the settings and do the scan.
I then shift the film to the same frame and do the second part.
In the second part I see streaks in the IR image.

Could this be "The return of the notorious Vuescan stripes"?

On Minolta 5400 specifically, the author was unable to fix that bug
for over two years and - as has been documented often enough - Vuescan
bugs do return at regular intervals.
I've written to Ed Hamrick, but not gotten a response as of yet.

Be very grateful for that! ;o)

Every time Minolta is mentioned the author tends to explode with
insults and obscenities! Indeed, it was his inability to fix this very
bug that caused him to theatrically resign from this newsgroup as he
ran away from hordes of duped and angry Vuescan users.

BTW, it's notable that once again (!) an "avid defender" of Vuescan
ends up braking down, admitting their frustration and coming clean
about Vuescan.

Does that make you a "Vuescan basher" too? Would you now follow your
own advice and apologize for such name calling of others in the past?

Don.
 
Robert, I know from your site that you have used this combination (5400
wih Silverfast). Did you use the calibration with IT8 target? If so, how
much difference does it make? It is not terribly "well defined" question
perhaps, but I'd grateful for any comments.
About calibration:
There seems to be unlimited confusion about this topic so here
is my addition to the muddle (wink).

As far as I understand calibration is a technique which allows one
to define the color rendering deficiencies of a physical device,
that is a scanner, monitor or printer (I guess a digital camera
would fall into this as well).

Each physical device creates some distortion when it tries to
represent the full range of visible light. So what calibration does
is to attempt to compare a known color target with the results of the
physical device and create a "map" which allows the colors to be
corrected so as to be closer to reality. There are problems with
this because no matter how good the calibration software is there are
just some colors that can not be handled by the device. It is then
up to the software and/or the user to decide what sort of compromise to
adopt to deal with these "out of gamut" colors.

For a scanner and specific scanning software there needs to be only
one such calibration map made. All the scanning setting are set in a
fixed manner and then the setup is calibrated. These exact settings
are then used for every subsequent scan.

I find the instructions from both Vuescan and Silverfast unclear.
They don't explain what they are calibrating so I don't use this
functionality, I use Monaco EZcolor to calibrate instead. The scanner
profile is then assigned to the scan and the file is then converted
into Adobe RGB space as part of the opening sequence in Photoshop.
Perhaps the scanner software can do this as part of its output function.

There are those who feel the need to calibrate for different types of
film. If they have a calibrated slide with the appropriate data file
this would imply that the calibration of the scanner/film combination
would produce an output which looked more like the original scene and
tended to remove any of the color characteristics of the individual
brand of film. Whether this is successful or not I don't know, I've
never had a calibrated slide of this type to try.

In one of my tips I suggest creating a color adjustment curve in
Photoshop which is used to correct for any residual distortions after
the image is opened. To do this I photograph a standard color chart with
sample colors and a gray scale in the type of lighting I usually use.
This means one for sunlight and one for open shade, for me.
This image is then scanned, converted to Adobe RGB and the curves are
created to remove any last residual color shifts in the gray scale.
Some of the individual patches can be tweaked as well if necessary, but
this is done by eye.

These curves are then saved and reapplied to subsequent files shot with
the same film and lighting conditions.
It seems that Silverfast may be able to do this with its color sampler
function and I have a tip on that as well, it just seems less intuitive
to me since I don't really know what the software is doing.
If you get this to work then you can save several steps in each scan
which is good for those doing lots of scanning.

Calibrating monitors and printers is a separate issue which is dealt
with by me and many others. If you are only trying to create your
own inkjet prints the problem is much easier than if you are trying
to create files which will be output on a variety of devices for
different uses.

I hope this discussion doesn't add to the general confusion...
 
About calibration:
There seems to be unlimited confusion about this topic so here
is my addition to the muddle (wink).

Thanks for food for thoughts.
It is a bit over my head, serves me well for asking questions without
doing solid homework well. :)

In the meantime 8.3.29 appears to be working correctly with my 5400 again.

Regards
MZ
 
Back
Top