Color shift along CCD in SE5400 and VueScan

J

Juha Koivisto

I noticed a strange problem with Minolta Scan Elite 5400 and VueScan
(8.0.7, Linux): the colors in the dark parts of a scanned slide have a
reddish cast in one end (top?) of the ccd, while the other end tends
toward green.

The effect is very subtle and would not be seen in most real scans, but
I did notice it in a photo with a lot of dark sky, which prompted me to
do some more testing... I put two test scans here temporarily:

<http://www.hut.fi/u/jkoivist/se5400/color.html>

Granted, the test is a bit artificial, as no-one would want to scan pure
black slides with colors boosted to the level of the first example, but
nevertheless, the problem is there and (like said) can sometimes be seen
in real photos as well.

The colors are not in the film - this is quite clear, but just to be
methodical about it, I flipped the film around and scanned it again,
with identical results.

The problem seems to affect only dark colours, and (as far as I could
tell with some quick testing) the Minolta software does not suffer from
it. It also seems that the colors may shift slightly more toward green
as the scanner warms up, but I'm not sure about this yet.

I have two theories about the cause: Either it's because of temperature-
dependent dark current in the ccd (after all, it's mounted vertically,
and heat rises up - but then why are red and green channels affected
differently?), or it could also be due to uneven lamp color which would
be properly compensated for in the "clear end" but the algorithm would
somehow overcompensate in the dark end.

The problem may be worse with negatives, as it would then affect bright
colors which are more visible. I don't have any photos with evenly
illuminated white background so I couldn't test it much (though I did
some scans of fully exposed film and the colors were there) - I'll try
to include some test shots on the next roll.

In any case the effect seems predictable enough that it should be easy
to fix. I'll be happy to do further testing if needed.

Could some other VueScan users also try scanning dark slide film? Do
you get similar results?
 
X

xphoto

Yes, I do have the similar problem. Not only the dark slides, but also
other color like pure white looks a little bit of pink ...
 
J

Juha Koivisto

Yes, I do have the similar problem. Not only the dark slides, but also
other color like pure white looks a little bit of pink ...

I'm not sure if we're talking about the same thing here... My problem is
not an overall color cast (which would be relatively easy to correct) but
that the colors are different in the right and left edge of the film.
Does your scanner also do this?

Could you (or someone else) try a simple test, scanning a piece of
exposed negative or unexposed slide film with Vuescan, and see if there
is a color shift across the width of the scan?

It seemed to me that the Minolta software did not have this problem,
though I'm not absolutely sure (I'm running linux so I can't test that
as easily).
 
J

Juha Koivisto

Juha Koivisto said:
Could you (or someone else) try a simple test, scanning a piece of
exposed negative or unexposed slide film with Vuescan, and see if there
is a color shift across the width of the scan?

Anyone... please? Or is really no-one interested to see if their scanner
suffers from this problem?
 
J

Juha Koivisto

Juha Koivisto said:
I noticed a strange problem with Minolta Scan Elite 5400 and VueScan
(8.0.7, Linux): the colors in the dark parts of a scanned slide have a
reddish cast in one end (top?) of the ccd, while the other end tends
toward green.

I did some more testing and discovered that if I initialize the scanner
with the Minolta software and then reconnect to the linux machine running
Vuescan, the problem is gone! Two more test scans at

What's more, the effect seems to persist after briefly powering off the
scanner (but not powering off overnight or unplugging the power cord,
and apparently not powering off with the USB cable disconnected either).

Also, after running the Minolta software, scans in Vuescan are more alike
with grain dissolver off as with grain dissolver on, while before GD on
was darker and the color balance a bit different as well. I'm not sure
why this is or how important it is, but I'm starting to wonder if there's
something wrong with SE5400 initialization/calibration in Vuescan.
 
F

Fernando

On Fri, 16 Jul 2004 15:29:45 +0300, Juha Koivisto
why this is or how important it is, but I'm starting to wonder if there's
something wrong with SE5400 initialization/calibration in Vuescan.

I agree. I have similar problems, with odd, uneven casts that are very
difficult to correct even with a custom profile (because of the
unevenness). I too think that there's some kind of calibration issue
with the 5400 and Vuescan, for Minolta Scan does not shows those
problems (but I don't like it: too little control over some
parameters, and I hate GD while loving ICE).

Fernando
 

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments. After that, you can post your question and our members will help you out.

Ask a Question

Top