Help, Minolta 5400 faulty scanner ???

[Patrick P.]

Hello everyone,

This is the second posting I've written about the problems I have with
my scanner.
Here are the problems: I'm getting rather flat looking scans. The
colors are dull. There is a magenta/red cast. There are sometimes
streaks (lines) in the shadows which are visible when the gamma is
turned way up. Many of the scans aren't focused correctly.


COLOR PROBLEM:
Typically I'm using the Minolta software with autofocus on,
autoexposure on, 16 bit and no color profile assigned (at scan).
Before I scan I crop and resize the exposure frame (by holding down
shift on the prescan image). I don't do any image adjustments in the
Minolta software. I use Velvia 100F Provia 100 and Kodak 100VS.

I've set up a page with example scans using different settings in
Minolta software and using vuescan demo version.

http://www20.brinkster.com/patrickp/minolta/minolta5400exp.html

Has anyone else had similar results with their scans? Is this normal?
If not should I return the scanner or is there another solution?

FOCUS PROBLEM:
I've heard others complain about focusing problems so this may be
normal. When the scans are out of focus (which happens a little too
often for my liking) I try using the auto focus pointer or manual
focus. I haven't been able to find any good help on the correct way to
use the manual focus. Matching the black and white bars seems totally
ridiculous to me. What do these bars represent? If the software is
able to produce these 2 bars why can't it match them automatically?

LINES IN THE SHADOWS PROBLEM:
They are only visible when I turn the gamma up. I've seen from other
threads that this is a common problem or a "feature" of the scanner.
Has anyone come up with a solution or reason for this?


From the reviews I read I thought this scanner was perfect but now I
really wish I'd bought a Nikon. So here is the big question, is it the
scanner or something else that causes all the problems? Any
suggestions would be appreciated.

Thanks for your help,
Patrick

 

[Wilfred van der Vegte]

Patrick P. wrote:

COLOR PROBLEM:
Typically I'm using the Minolta software with autofocus on,
autoexposure on, 16 bit and no color profile assigned (at scan).
Before I scan I crop and resize the exposure frame (by holding down
shift on the prescan image). I don't do any image adjustments in the
Minolta software. I use Velvia 100F Provia 100 and Kodak 100VS.

I've set up a page with example scans using different settings in
Minolta software and using vuescan demo version.

http://www20.brinkster.com/patrickp/minolta/minolta5400exp.html

Has anyone else had similar results with their scans? Is this normal?
If not should I return the scanner or is there another solution?

Patrick,

Without the original slide, it is difficult to judge how 'bad' these
scans are. There are clearly visible differences in color but no serious
color casts. The color shift caused by ICE has been reported before, see
http://www.xs4all.nl/~bvdwolf/main/foto/scan/se5400/se5400-5.htm. It
also occurs with Nikon scanners.
What you call 'perfect' can perhaps be attained by using a professional
profiling tool and creating custom profiles for your scanner and your
monitor. BTW, you mention that you are using a laptop and I think laptop
screens are still not very suitable for judging colors.

FOCUS PROBLEM:
I've heard others complain about focusing problems so this may be
normal. When the scans are out of focus (which happens a little too
often for my liking) I try using the auto focus pointer or manual
focus. I haven't been able to find any good help on the correct way to
use the manual focus. Matching the black and white bars seems totally
ridiculous to me. What do these bars represent? If the software is
able to produce these 2 bars why can't it match them automatically?

The resolution of the DSE 5400 may go beyond the limitations of your
camera, your camera handling and/or the film you used. At least that's
my experience. When scanning very grainy negatives with my DSE 5400, the
grain comes comes out very sharp, but Velvia or Provia slides shot with
high-quality equipment (a Contax G2 with Carl Zeiss lenses) only come
out completely sharp every now and then. Most slides that looked
perfectly sharp on my previous 2820dpi scanner now don't look sharp
anymore, probably because minimal amounts of motion blur become visible
at 5400 dpi. Still for most uses (e.g. printing on A3 paper) the slides
can still be considered to be perfectly sharp, and the scans do contain
more detail than they did at 2820 dpi.

LINES IN THE SHADOWS PROBLEM:
They are only visible when I turn the gamma up. I've seen from other
threads that this is a common problem or a "feature" of the scanner.
Has anyone come up with a solution or reason for this?

This is a common feature of all scanners, I think. If you turn up the
gamma to reveal shadow detail that simply isn't there, you will get
these lines. Insted of the detail that isn't there in the film, the only
'detail' that can be shown is an exagerration of the minimal differences
in sensitivity between the individual CCD sensor elements. The only
difference with a Nikon scanner may be that for a Nikon the lines
probably won't be colored, because Nikon scanners have a one-line CCD.
That only makes things worse, I think, because the same error will be in
all the three color channels together.
BTW, calibrating the scanner in VueScan is known to reduce the
visibility of these lines.

 

[Bart van der Wolf]

SNIP

When scanning very grainy negatives with my
DSE 5400, the grain comes comes out very sharp,

But given the amount of pixels the 5400 produces, you can either scale the
image size down or you do not need to enlarge so much. Also, use of the
Grain dissolver should help. And it will give better result when you scan at
a higher (the highest) resolution than stricly needed for the job at hand,
and bin the pixels together (VueScan File size reduction, or Photoshop
Filter|Pixellate|Mosaic followed by integer resizing).

but Velvia or Provia slides shot with high-quality
equipment (a Contax G2 with Carl Zeiss lenses) only
come out completely sharp every now and then.

Yes, it'll reveal lens and photographer shortcomings much better :-(
However, on the bright side; there's less need to enlarge the file.

Most slides that looked perfectly sharp on my previous
2820dpi scanner now don't look sharp anymore,
probably because minimal amounts of motion blur
become visible at 5400 dpi.

That's why I often carry a monopod with me (if the tripod is unpractical).

SNIP

BTW, calibrating the scanner in VueScan is known to reduce the
visibility of these lines.

Yes, but there is also still a VueScan issue with the Dmax of film in the
DSE5400.

Bart

 

[Wilfred van der Vegte]

SNIP


But given the amount of pixels the 5400 produces, you can either scale the
image size down or you do not need to enlarge so much. Also, use of the
Grain dissolver should help. And it will give better result when you scan at
a higher (the highest) resolution than stricly needed for the job at hand,
and bin the pixels together (VueScan File size reduction, or Photoshop
Filter|Pixellate|Mosaic followed by integer resizing).

Bart, this was not a complaint. I know the tricks you mention (except
the mosaic stuff - thanks, I will try it). I just brought this up to
point out that I don't experience a focusing problem like the OP did.
When the grain comes out sharp, at least there can't be a focusing
problem ;-)

Also, my point in making the following remarks was that it's very well
possible that the OP's scanner is not to blame for the unsharp scans he
gets. This can also be supported in a more quantitative way: in DoF
calculations, a 0.025 mm circle of confusion (CoC) is typically
accepted. Any unsharpness inside this circle is considered to be
unnoticeable (this can also be said for motion blur, of course). Now
consider that at 5400 ppi, one pixel is only 0.0047 mm. This implies
that the CoC area comprises five scanned pixels. Any soft transition
that takes up five or less pixels in an image scanned at 5400 ppi is
normally considered sharp! But someone who examines the scan at 1:1
pixel size on his screen will probably judge the scan as unsharp ...

Yes, it'll reveal lens and photographer shortcomings much better :-(
However, on the bright side; there's less need to enlarge the file.

Why would you want to enlarge the file? Are you referring to the need
for blowing up lower-resolution scans with software sch as Genuine Fractals?

 

[Bart van der Wolf]

SNIP

Bart, this was not a complaint.

Sorry if it came across like that, but it wasn't exclusively directed at
you. It was more a remark in general, such are the workings of a public
forum ;-)

SNIP

When the grain comes out sharp, at least there can't be a focusing
problem ;-)
Exactly.

SNIP
Any soft transition that takes up five or less pixels in an image
scanned at 5400 ppi is normally considered sharp! But someone
who examines the scan at 1:1 pixel size on his screen will
probably judge the scan as unsharp ...

Correct. It's a common mistake made by others, to not consider the final
output size, be it in print or on screen after downsizing.

SNIP

Why would you want to enlarge the file?

7800x5232 pixels (uncropped) will allow to print a non-interpolated
10.8x7.3 inch @ 720ppi (native Epson desktop printer resolution) with
awesome quality. It will also allow to control sharpening down to the single
pixel, without re-interpolation by the printer driver. Larger output
benefits from scaling the file size up to match the printer's parameters. I
do agree that the visible difference will be small, but why waste the
opportunity (especially if Qimage can handle the rescaling for you as you
print).
And in practice, most people are more likely to be confronted with scaling
down.

Are you referring to the need for blowing up lower-resolution
scans with software sch as Genuine Fractals?

IMHO, for most images GF doesn't produce very realistic images (too much of
a posterize look with surreal edges). I like Qimage, which just added a new
"pyramid" interpolation algorithm, better.
Some comparisons of this new interpolation method:
http://www.ddisoftware.com/testpics/pd-1.jpg
http://www.ddisoftware.com/testpics/pd-2.jpg
http://www.ddisoftware.com/testpics/pd-3.jpg
which was developed after an interesting discussion/poll:
http://forums.dpreview.com/forums/read.asp?forum=1003&message=7642362

Bart

 

[Patrick P.]

Sorry for taking so long to reply.

The out of focus is not due to the slide. When i use the manual focus
option I can usually get them into correct focus. It is the auto focus
that doesn't always work. I'd say 3 of 10 are not focused correctly. I
know they are focused correctly when I can see the grain. I hear that
vuescan has better odds than the minolta software.

The color shift caused by ICE has been reported before, see
http://www.xs4all.nl/~bvdwolf/main/foto/scan/se5400/se5400-5.htm.

Thanks for the interesting link. In my opinion my colors are really
bad. I've talked to some friends and decided to get an ICC made up for
my scanner. It seems like the most accurate and scientific way to fix
the color problem. It has got to be better than "eye balling" the
color. I'm hoping this will save me some time. What do you think?

BTW, you mention that you are using a laptop and I think laptop
screens are still not very suitable for judging colors.

Couldn't agree more, but don't have the choice. I've brought my images
into work and viewed it on my PC and a Mac. So at least that gives me
an idea of how my laptop is showing the colors.

I also scanned some images on my work PC and got the same results as
with my laptop.

Thanks,
Patrick

 

[Wilfred van der Vegte]

Sorry for taking so long to reply.

The out of focus is not due to the slide. When i use the manual focus
option I can usually get them into correct focus. It is the auto focus
that doesn't always work. I'd say 3 of 10 are not focused correctly.

Did you try to define of the location of the focusing point manually ?
Ideally, it should be somewhere off-center where the scanner is able to
recognize sufficient detail. When you focus on the center (I think
that's the default fucusing location), the limited DOF might bring the
corners out of focus.