Apache-Talk @lexa.ru 

Inet-Admins @info.east.ru 

Filmscanners @halftone.co.uk 

Security-alerts @yandex-team.ru 

nginx-ru @sysoev.ru 




      :: Filmscanners
Filmscanners mailing list archive (filmscanners@halftone.co.uk)

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: filmscanners: ScanWit Yellow stain


I am a total newbie at this. I thought I would take a look at these to see
what I could learn. With that in mind, here are some observations I made
playing with the images. Maybe someone could correct my assumptions!

1. I took the raw data and rescanned it using Vuescan 7.1.3.
  a. I adjusted the gama down to 1.0(and then several steps up through 1.8.
  b. I left the black point on auto and adjusted the black point % from 0 to
10% in various steps.
  c. I set the white point to 1% and left auto-white point checked.
  d. I tried various compensation modes (auto-levels, white balance, etc.)


The raw scan is low resolution, so I ignored the "grainy" appearance.

The image appears to be very over exposed - which I believe you mentioned. I
found raising the black point and white point % recovered most of the
contrast and combined with the gamma correction the color intensity/range
was comparable to the scan of the print. While you mention that the scan has
less color information than the print, I found just the opposite. I found
lots of color info buried in there - hiding. And as I would expect the
dynamic range of the scanwit scan is much greater than the photo scan. In
the RAW data I saw much more detail in both highlight and shadows (look at
the road and the strips painted on it - they are there in the scan and
burned out in the photo.) It looks like the printer drove the contrast way
up to compensate for the over exposure and lost almost all highlights.

The "Yellow stain" is not just on one side, but is in fact on all four
sides, with the left and right being worse than the top and bottom. If you
darken the image enough you can see a definite "ring" about 10-15% of the
image width all the way around. With the yellow being worse on the sides, my
first impression was that the negative was not flat, but curved in the neg
holder and the "halo" of yellow was a result of the curvature - either
focus, refraction, or something... A second thought I had was that it
appeared as if the light source in the scanner was focused into a stripe on
the center of the negative instead of an even diffusion. Being yellow, it
would indicate the blue CCD sensors were less sensitive to low light levels
than the red and green. (Does anyone know if this is a characteristic of
blue CCD sensors in other scanners?)

On the banding what I saw was a series of vertical broken yellow stripes 1
pixel wide. Is this what you are refering to as banding? Again, I am a
newbie and dont always understand or see things that others know from
experience. The broken/dashed stripes of yellow appear only in the area with
the overall yellow cast. I have seen this affect before in scans from my
flatbed scanner on over exposed 400/800 ISO film and on my scanwit on
overexposed Kodax Gold 400/6. In researching this on the web I happened to
look at the example Q60 scans on Ed H's Vuescan page and noticed similar
stripes (banding) on what were considered "Good scans". I wonder if I am
seeing things, or if this is a trait of CCD scanners in very dense areas
(ie. is this CCD noise from low signal conditions?)

2. I took the raw data into PS 6.0.1 and played with it for a while. The
things I saw in Vuescan appeared the same in PS. I was able to achieve
reasonable color and image contrast, but color balance eluded me. I came
close using a horizontal gradient to modulate the blue layer - this came
close to canceling the effects, but was not a perfect match, so the results
were still not usable.

About that time my wife came home from work and made me go out to eat with
her. So, I couldnt do any more experiments.


1. I have spent about 6 to 8 hours a day for the past few weeks reading
everything I can find on the internet about scanning. It is almost
universally agreed that scanning film faster than 200 on a 2700 DPI scanner
is a real challenge. Several different theories exist as to why, I find the
idea of grain aliasing to be very convincing. This is the idea that
negatives are really a digital medium (the image is a series of random small
dots [grain?] - not a continous tone) and when you combine sets of digital
data (the scanning is digital dots) where the data content of the two sets
is close (or a harmonic) to the same frequency (number of dots per inch)
they will interfer with each other and create artificial lower frequency
dots (bigger and lower DPI.) This certainly would explain why in my case I
can see much "grainier" scans from negatives when compared to prints from
the same negative from the grocery store developer. I think part of what you
are seeing is this effect in the denser areas. I am changing to using 100
and 200 film almost exclusively now with good results.

2. The scan is NOT typical of what I am seeing from my Scanwit 2720 on
negatives (400ISO). There is a definite yellow/orange cast around the outer
edge of the image. You do indicate that you are seeing the banding/stripes
in other scans in light areas on negatives and dark on positives - this
would indicate that the banding/stripes is caused when there is low signal
levels from the CCDs (low light getting to the sensor.) Again this agrees
with what I have seen on several of my own scanners, and samples I have
looked at on the web of other "good" scanners.

I would like to suggest some possible causes in hopes that more experienced
individuals will take the time to shoot them down and teach me a bit more
about my new hobby:

1. The negative was not flat in the carrier which caused a change in focus
(fooled the auto-focus system) or exposure on the edges versus the center of
the negative. My negatives do not mount perfectly flat, but apparently they
are close enough that I don't see any significant difference in
focus/exposure between the edges and center of images. However, maybe your
scanners focusing system is faulty (or does the external software control
this?) Try adjusting the focus manually in Vuescan (Device/Focus) and see if
it affects the banding. This will not solve the problem but may point to the

2. The lamp is improperly focused in the scanner causing a "bright" middle
area - ie. an uneven exposure during scanning. Unfortunately with the
scanwit you cannot adjust the exposure to test this. But if this is the
problem, I would expect other overexposed images to show similar effects. I
would expect it to be present in any scanned image, but the "bright" spot
could be very marginally brighter and only show significant effects in very
dense regions of images.

3. There is a bright external light source (over head, desk, etc...) which
is leaking into the scanner and fooling the focus/exposure systems. It may
not be "bright" to the eye, but just the right color (frequency) to hit a
sensitive region on the CCD sensors. I tend to scan in the dark - not
because I should, I just like working with my lights off!

4. There is something obscuring the exposure/calibration slot on the
negative carrier resulting in a bad auto-exposure calibration. Check that
your neg strip doesn't have a "tail" sticking into this area - it is the
slot (1/4" x 1" just before the frame #1 opening in the carrier. It must be
clear and clean to work correctly.

5. There is some kind of foreign matter on the optics of the scanner - dust,
condensation, finger prints, cat hair? which causes the focus/exposure to
vary across the image - again, I would expect this to show up on other
images. However, if it is just a small effect, it could only show up in
cases of very low level signals - ie. dense areas of the negative.

To finish up (and quit wasting everyones bandwidth) in the short time I have
had my Scanwit I have jumped to the conclusion that something was wrong with
the scanner serval times, only to later find it was operator error. I have
also learned a great deal about operating Vuescan. With good negs Vuescan
does do a great job with the default settings, but with marginal negs you
need to get in and tweak things. I have found I can recover images which at
first appear to be completely useless, just by fiddling with gamma, black
and white points. I havent got the courage up to try to do color balance
using the manual white and black point settings yet - but I expect it can be

I expect this is a bad scanner, but to be sure you need to critcally examine
other scans for similar effects - ie. the sides being under/over exposed. If
this is eveident to any degree then check for possible causes - curved
negatives, external light sources, etc, or possible contamination - I have a
cat and it is veritually banned from my office now because cat hair gets
everywhere - could you have cat hair or dust in the scanner?

Good luck and if you exchange the scanner let us know how it goes.


"A newbie and proud of it!"

-----Original Message-----
From: owner-filmscanners@halftone.co.uk
[mailto:owner-filmscanners@halftone.co.uk]On Behalf Of Oostrom, Jerry
Sent: Thursday, June 28, 2001 2:27 AM
To: 'filmscanners@halftone.co.uk'
Subject: RE: filmscanners: ScanWit Yellow stain

Here are some of the scans I promised.

I made them by attaching the scanner to another current outlet group, but it
did not change things. I also swapped the negative to show you the error IS
in the scanner, not the negative (largest yellow band still appears at the
right side).

Here they are in the http://www.bigfoot.com/~jerfi/testscannerfix/ dirctory:

The film scans were made using Vuescan 7.1.3, the newest version yesterday.
I set:
*        the Vuescan settings to default,
*       used a scan resolution of 675dpi and resampled the image to a
quarter its size (corresponding to 337 dpi) using Picture Window Pro 3.0.
The crop file was saved from Picture Window Pro as a jpg at 100% quality
The raw scan file was also downsampled using PWP and may as a result have
obtained a color profile (perhaps sRGB, didn't take the time to check that).
My mistake.  However, you will still see the banding if you convert it to a
positive scan. BTW. the raw scan file is still in 16bit mode.

Now if you compare this to the flatbed scan from the print I received from
the printing service you know that the print shows much more color
information AND has no banding! Unfortunately, they just crop wrong!#$%

Film: Kodak Supra 400, shot at ISO320, me leaning almost over the car with a
The whole roll was shot at ISO320, but this particular negative seems more
Except for the yellow bands and strange color I obtained a scan with a very
nice soft! grain pattern.

The crop file:
http://home.wanadoo.nl/joostrom/testscannerfix/ToWebFilmScanquart.jpg 162k

The raw tiff 48bit file:
http://home.wanadoo.nl/joostrom/testscannerfix/scan0002quart.tif 858k!!!

The nice flatbed scan with the wrong crop but NO banding:
http://home.wanadoo.nl/joostrom/testscannerfix/toWebFlatbedquart.jpg 132k
(also PWP'd and saved at 100% jpg quality)

So is it user error, is it scanner error (bad lamp / bad CCD) or is it me
pushing the envelopes of what a decent scanner can do?

Thank you all for your interest, I've used you comments to make my point at
Acer NL,


BTW. All good intentions and manners aside: Mr. Honda Lo has become very
silent for weeks now. The same goes for the dutch contact person, whom I've
mailed several times (always received a return receipt, but unfortunately
that does not prove much). I am getting frustrated enough to write to the
bosses / managers / colleagues of this dutch contact person (whose email
addresses I got from Mr. honda Lo), just not there yet and still hoping for
some help from Acer NL. Are they just waiting for my 90 days warranty over
the previous repair to pass?
This mail is Bcc'd again to both Honda Lo and the dutch contact person.

> -----Original Message-----
> From: Oostrom, Jerry [SMTP:Jerry.Oostrom@Alcatel.nl]
> Sent: Wednesday, June 27, 2001 12:52 PM
> To:   'filmscanners@halftone.co.uk'
> Subject:      RE: filmscanners: ScanWit Yellow stain
> The scan that I made is indeed OFF the planet, even on my screen, but it
> has
> been done with the regular settings in which I scan normally exposed
> negatives (gamma 2.22). I overexposed the whole roll from ISO 400 to ISO
> 320, should not be too much I think, but this frame came out more
> overexposed than others. The print from the printing service is OK though,
> I
> could send a flatbed scan of it (the only problem is their cropping, they
> took off parts of the left door!@#$%).
> Anyway, before you all start thinking I am in error here, I think I'll
> send
> a raw scan downsampled to the list (I'll send only the link) so you can
> see
> for yourself if it's me or the Scanwit.
> The scanner does not work OK on normally-exposed materials in the sense
> that
> even though the errors are much less visible, they are still there, most
> noticeably on even coloured or light colored parts of a scan from negative
> and on dark colored parts of a positive. Its ruining the wedding shots we
> have taken from our family.
> Thank you for your interest,
> Jerry
> > info & comparisons


Copyright © Lexa Software, 1996-2009.