Nikon CoolScan 5000 Color Shift Problems

[Ken]

My trouble deals with color shifts when scanning some older (1970's -
1990's) Ektachrome slides. I know some people have trouble with
Kodachrome slides but those work real well for us and we do not have
any of these problems. When we have post processing turned on and use
ROC, we can get severe color shifting. Below is a link to an example
(compressed) which shows an extreme magenta color shift:

http://www.usadiscountgenerators.com/ScannerProblem.html

If I keep rescanning the same slides over and over, I get different
results for the same slide - some times the slide is perfect, other
times the slide is color shifted (varying colors for the same slide).
Only some slides in a batch are color shifted.

We have been working with Nikon on this for a year and I am afraid to
say I am not impressed. Their recommendation was not to use ROC or GEM
as these might cause difficulties with their scanner. By the way, I am
using NikonScan but the same thing happened with VueScan. If I do not
use ROC, about 20% of the slides are too dark to be usable. I have
thousands of slides to do so I can not keep rescanning many times until
I get a good slide.

To eliminate variables, I have tried the scanners on multiple PC's and
Macs, all running different versions of system software and the problem
always shows up. I have also tried with and without the slide feeder

Nikon Service says it must be the slides but that would not explain why
scanning the same slide sometimes gives good results and sometimes bad.
It seems to me that it can not be the slides or the software since the
same input should give the same output. The only thing that makes
sense to me is an intermittent problem in the hardware (fluctuating
intensity of the diodes, power fluctuations on the sensor, etc.).
Nikon tested the machine and said it is working fine on their test bed.
I sent them some of my slides and they were able to reproduce the
problem if they use ROC but they say the scanner is fine and that I
will just need to turn ROC off and individually manipulate each slide
in Photoshop - that's not an option for 7,000 slides.

Is anyone else seeing this problem? Any help would be appreciated!

Thanks,

Ken

 

[CSM1]

My trouble deals with color shifts when scanning some older (1970's -
1990's) Ektachrome slides. I know some people have trouble with
Kodachrome slides but those work real well for us and we do not have
any of these problems. When we have post processing turned on and use
ROC, we can get severe color shifting. Below is a link to an example
(compressed) which shows an extreme magenta color shift:

http://www.usadiscountgenerators.com/ScannerProblem.html

If I keep rescanning the same slides over and over, I get different
results for the same slide - some times the slide is perfect, other
times the slide is color shifted (varying colors for the same slide).
Only some slides in a batch are color shifted.

We have been working with Nikon on this for a year and I am afraid to
say I am not impressed. Their recommendation was not to use ROC or GEM
as these might cause difficulties with their scanner. By the way, I am
using NikonScan but the same thing happened with VueScan. If I do not
use ROC, about 20% of the slides are too dark to be usable. I have
thousands of slides to do so I can not keep rescanning many times until
I get a good slide.

To eliminate variables, I have tried the scanners on multiple PC's and
Macs, all running different versions of system software and the problem
always shows up. I have also tried with and without the slide feeder

Nikon Service says it must be the slides but that would not explain why
scanning the same slide sometimes gives good results and sometimes bad.
It seems to me that it can not be the slides or the software since the
same input should give the same output. The only thing that makes
sense to me is an intermittent problem in the hardware (fluctuating
intensity of the diodes, power fluctuations on the sensor, etc.).
Nikon tested the machine and said it is working fine on their test bed.
I sent them some of my slides and they were able to reproduce the
problem if they use ROC but they say the scanner is fine and that I
will just need to turn ROC off and individually manipulate each slide
in Photoshop - that's not an option for 7,000 slides.

Is anyone else seeing this problem? Any help would be appreciated!

Thanks,

Ken

It is a well known fact that film dye layers fade and shift over time.
Ektrachrome often fades long before Kodachrome.

Some dye layers are effected more than other layers. Thus the common color
shift is to the reddish or magenta ranges.

Kodak's information on processing Ektrachrome and its effect on dye fading.
http://www.kodak.com/global/en/service/tib/tib5200.shtml

Some more information:
http://photo.net/bboard/q-and-a-fetch-msg?msg_id=00DGgG

 

[Ken]

Thanks for responding.

I can understand the slide scan being faded or color shifted because
the original is faded or shifted - I expect that. However, every time I
scan the exact same slide with the exact same settings, I should get
the exact same result (shifted or not). However, the exact same slide
will some times give perfect results and other times give results
similar to the picture url posted. So unless I am misunderstanding what
you are saying, I do not believe that fading of the slide is causing
the problem I am seeing. Am I wrong?

 

[rafe b]

My trouble deals with color shifts when scanning some older (1970's -
1990's) Ektachrome slides. I know some people have trouble with
Kodachrome slides but those work real well for us and we do not have
any of these problems. When we have post processing turned on and use
ROC, we can get severe color shifting. Below is a link to an example
(compressed) which shows an extreme magenta color shift:

http://www.usadiscountgenerators.com/ScannerProblem.html

I have never seen anything quite so pathological from
my Nikon scanner, though mine is the LS-8000.

I would start by turning off ROC and GEM, and
leaving all controls in their "neutral" position.

Before doing auto-exposure, make sure the selection
marquee in the image preview includes *only* the real
image (and nothing else.) Best to be conservative
about this and leave off a bit of image on each edge.

Also: try scanning *without* auto-exposure, all
controls centered/neutral, etc.

If all else fails, clean out (ie., remove the contents of)
the appropriate Nikon directory in the
\Documements and Settings\... directory. I can't give
you a more specific path, as I'm away from my "home"
computer at the moment.

The Nikon scan driver un-installer doesn't remove
or clean this directory, and it really should. Nikon
does supply a utility to do a registry cleanup (after
uninstall) but it's not quite thorough enough.


rafe b
www.terrapinphoto.com

 

[Ken]

Hi Rafe,

Thanks for your help. I have done multiple clean installs onto
multiple computers that have never had the software/scanner installed
on them and that did not make the problem go away.

I did try limiting the size of the selection of the image as you
suggested. I even tried selecting a very small area in the center of
the image and that did not help.

If I do not enable ROC, the image does not color shift but I need that
feature. If I turn ROC on, regardless of settings, I still get the
same slide color shifting some of the time.

I would have thought if the problem were inherent to the software,
hardware, or slides, then many people would be having this problem. I
have downloaded different copies of the software and installed those as
well as the software off of the CD so I do not think the problem lies
with a bad copy of the software. I do not think it can be a bad slide
since I should consistently get bad results each time I scan the same
bad slide. It still seems to me that the only possibility that makes
any sense is that there is an intermittent problem/flakey component in
the hardware but Nikon says no.

I get different results from the same slide and Nikon drives me nuts
because they say that is because the slide is bad but they can not
explain why a bad slide does not give consistantly bad results. One of
the reasons I bought this scanner is because Nikon has a good
reputation and because I have liked their other equipment that I own.
But this experience with their service department really has me
rethinking this reputation. I will owe them a big apology if this ends
up being something i am doing but I do not see how that is possible.

Maybe I can find someone local who has one of these scanners and have
them run my slides and see what they get.

Thanks again for your help.

Ken

 

[rafe b]

If I do not enable ROC, the image does not color shift but I need that
feature. If I turn ROC on, regardless of settings, I still get the
same slide color shifting some of the time.

I would have thought if the problem were inherent to the software,
hardware, or slides, then many people would be having this problem. I
have downloaded different copies of the software and installed those as
well as the software off of the CD so I do not think the problem lies
with a bad copy of the software. I do not think it can be a bad slide
since I should consistently get bad results each time I scan the same
bad slide. It still seems to me that the only possibility that makes
any sense is that there is an intermittent problem/flakey component in
the hardware but Nikon says no.

If the problem only occurs with ROC turned on, then it
seems to me it's not a hardware problem.

I can't comment much on ROC because I don't use it.


rafe b
www.terrapinphoto.com

 

[Ken]

Hi Rafe,

ROC uses information from the prescan, etc. to make its corrections.
If I leave the slide in there and use the cached prescan information,
then I will always get the same result (good or bad) until I do another
prescan. I am not in front of my system now and I can not remember if
the auto exposure is part of the prescan or not so i may be getting
prescan and autoexposure mixed up (this problem got me so frustrated
that I had to take a break from it for a few months). It seems to me
at that prescan stage, the software gets bad data from the scanner
which it then uses to correct the image. Another possibility is that
the software sends information about how to scan to a buffer in the
scanner and for some reason this buffer is not functioning properly. I
find these types of problems much easier to troubleshoot when I have
designed the system and software. Not knowing exactly how they are
doing things makes it difficult to trouble shoot the hardware and
software.

Thanks again,

Ken

 

[rafe b]

Hi Rafe,

ROC uses information from the prescan, etc. to make its corrections.
If I leave the slide in there and use the cached prescan information,
then I will always get the same result (good or bad) until I do another
prescan. I am not in front of my system now and I can not remember if
the auto exposure is part of the prescan or not so i may be getting
prescan and autoexposure mixed up (this problem got me so frustrated
that I had to take a break from it for a few months). It seems to me
at that prescan stage, the software gets bad data from the scanner
which it then uses to correct the image.

You're describing dICE, not ROC or GEM.

The prescan takes advantage of the CCD's inherent
ability to "see" infrared. That's used by dICE to create
a map of defects on the film surface that need to be
patched.

As far as I know, neither ROC nor GEM use the
prescan info. They are done entirely in post-processsing.

IIRC, "ROC" is an acronym for "Restoration of Color.
"GEM" has something to do with grain reduction.

As I said earlier -- I don't use either of these features.
But I do use dICE, almost always.

If you're getting a nasty color cast over the whole image,
and it only happens when ROC is turned on, then I'd
say that points to ROC as the culprit.


rafe b
www.terrapinphoto.com

 

[degrub]

"If I do not
use ROC, about 20% of the slides are too dark to be usable. "

This would point to the slides being under exposed or too much of the
area around the image being used by NS to calculate exposure. If the
slides were truely under exposed, then given the dye fading, you may
have lost most if not all of the information from that dye layer and ROC
will not have anything to work with.

However, since you are getting different results on the same slide with
each prescan, that points to the scanner software calculating the
exposure incorrectly. ROC assumes the image information is from a
reasonably exposed slide and that the dye response curves have changed
in a predictable way.

In the image you give as an example there is a large black area around
the actual image. If that is consistent on most of the slides you are
having issues with , then i suspect the problem is in what part of the
total area is used to calculate exposure.

The only other thing i can think of that would cause this issue is an
unstable supply of power to the scanner that is changing the LED
intensity (again - exposure issue). Do you have a UPS in front of the
scanner, preferably one with good voltage regulation ?

i have used my CS4000 for years on E and K slides with ROC and never
seen this issue. If Nikon can replicate this on their bench, then it
likely is something in the slide or the power to the scanner.

Another way to test this is to take an IT8 calibration slide and see if
you can replicate the prescan to prescan variation. Another way to test
would be to prescan and scan the same slide several times without ROC
and look at the histogram very closely to see if there is any change in
the R G and B channel individual histograms. IF you can, check the power
to the scanner for voltage issues. An analog voltmeter should be good
enough. Watch for large needle movement. Anything below 100 volts is
definitely a problem. Scanners like stable power.

Just for the sake of trying something different, download a demo copy of
silverfast ( www.silverfast.com )for your scanner and see if you get
the same results. They do not use ROC , but have a different calculation
to do the same thing. You could also try a demo of the photoshop plugin
ROC from www.appliedsciencefiction.com to see if you get the same
results without using the NS version of ROC during the scan.

regards,

 

[Ken]

Hi Degrub,

Thanks for your detailed response.

I have the scanner on a Tripp-Lite UPS and I have tried the scanner at
2 other locations which are 35 miles away from me and each other.
However, I will check the voltages like you suggest. I have been
thinking that the scanner has a faulty internal power supply which is
sending fluctuating voltage to the scanner.

The slide posted is just one compressed example that I happened to
have. One of the first things I thought might be a problem was the
black border so I did try significant cropping of the scan area but
that did not help. As a side note, the black border did not adversely
affect the Kodachrome slides although it does cut down on the number of
bits of color available just to resolve the actual slide.

"If I do not
use ROC, about 20% of the slides are too dark to be usable. "

This would point to the slides being under exposed or too much of the
area around the image being used by NS to calculate exposure. If the
slides were truely under exposed, then given the dye fading, you may
have lost most if not all of the information from that dye layer and ROC
will not have anything to work with.

The actual slides do not appear to be nearly as underexposed as the
scans of them.

Another way to test this is to take an IT8 calibration slide and see if
you can replicate the prescan to prescan variation.

I have a IT8 calibration slide and will give this a try.

Another way to test
would be to prescan and scan the same slide several times without ROC
and look at the histogram very closely to see if there is any change in
the R G and B channel individual histograms.

Another good idea!

Just for the sake of trying something different, download a demo copy of
silverfast ( www.silverfast.com )for your scanner and see if you get
the same results. They do not use ROC , but have a different calculation
to do the same thing. You could also try a demo of the photoshop plugin
ROC from www.appliedsciencefiction.com to see if you get the same
results without using the NS version of ROC during the scan.

And two more good ideas. Thanks very much!!!

I will try these and also see if our local camera specialty store has a
scanner I can test the slides on.

Ken

 

[degrub]

if it is the power supply or some output element, then it should be
repeatable on any slide and you should see it in the histogram.

regarding the actual slides not appearing underexposed - i assume on a
light table or projection - the human eye adapts very quickly where the
scanner does not. THe autoexposure attempts to find a compromise for
what it sees as approximate black and white and fit the available inputs
to the scanner 0 and 255 values ( actually Nikon may bias the darker
areas to reduce noise).

Hope you find something - hardware or software.