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]

filmscanners: Skin tones and acer noise

(Jerry, I copied you on this because it appears Alan has a similar issue to
yours. I am still trying to get my website updated so I can post the
material about correcting images with these problems. But, now that I have
seen Alan's scan's I am beginning to think that you and he may have bad or
marginal units.)

(filmscanners list: I copied the list on this reply to Alan Womack - part of
an off line exchange we have had concerning noise in the blue channel of
Scanwit scanners. I hope other Scanwit owners will send me
(frank@theNichols.net) clips from images in high density/neutral tone areas
from thier scanners. It appears some Scanwits have very high noise in the
blue channels as shown in the attached examples and if I can collect enough
samples maybe we can get ACER's attention. - sorry about being so long
winded, it is a personality flaw of mine!)


I agree your 200 (aw fugi 200.tif) looks more like my 800 (fn kodak
800.tif), but if you look at the scans (blue channel) at 600% magnification
I see two significant differences:

1. There is a very strong horizontal component in your scan, and I bet these
horizontal lines are along the axis of the CCD - ie. perpendicular to the
motion of scanning? This is different from Jerry's macro/visible banding
which we have called "yellow stain", although his scan also shows these same
horizontal lines at the micro level (jo clip.tif).

2: Your histogram vs mine (blue channel)! Yours is the second sample I have
seen with a sparse/spikey/noisey blue channel histogram like this. The first
one came from Jerry Oostrom - his is even worse than yours. I realize that
facial tones don't have much blue component, but I would still expect what
is there to have a smoother variation.

Is your scan of an overexposed negative (or under exposed slide) which would
make the emulsion very dense?

The issue seems to be the blue (and to some lesser extent green) channels
are less sensitve and so end up operating down in the noisey region of the
operating curve of the CCD.

I want to try an experiment on my scanner, if you could do the same it would
be very enlightening. I am going to find a piece of very light blue filter
and lay it over the negative and rescan my "baddies". The goal is to fool
the autoexposure system into increasing the exposure level on the blue
channel. (I may have this wrong with negatives, it may need to be red and
green filters added - or maybe a neutral?) I don't expect this to work - I
expect the light level is not affected, but instead the duration of exposure
will be changed - which won't help if the blue channel is down in the noise
region as I expect. But if it does work, then this could be a "hack" to help
with very dense film. The idea would be to scan twice, once with and once
without the filter, then combine the blue channel from the scan with the
filter with the red/green channels from the scan without the filter - argh!

I took my scanner apart a while back in hopes that there would be a manual
adjustment on the lamp, but alas no luck. Without reverse engineering the
circuits around the lamp, I don't think fiddling is going to help. I tried
placing a mirror behind the lamp to increase the light though the film, but
it made no difference I could detect.

My guess is that the only hope is to collect samples from a number of owners
of scanwits - if we could collect enough samples to show a distinct set of
two groups - bad/marginal (yours), better/okay(mine) - maybe we could get
ACER's attention and get some form of fix for the worse set. However, if the
samples show a smooth distribution in the amount of noise between owners,
then it is probably just "normal" varition in less expensive components and
ACER probably won't do anything except for the worst cases.

Also, I don't have access to a densiometer(sp?). Do you? If not maybe we
could get someone on the list to test the negatives (yours and mine) and see
if you are just outside the specified range of Dmin/Dmax for the machine. In
which case, you got a "normal" scanner, and I (and others) got an
exceptional one.

As a partial cure - have you tried using a very slight blur on the blue
channel? I use about .2 to .3 radius of Gaussian blur on the blue channel
when it gets this bad in mine and it seems to help at the expense of a
little sharpness.

Attached are smaller snips from our scans - more suitable for posting to the
mailing list - these show almost no image but at 600% magnifcation they
demonstrate the noise characteristics of each. They are labeled: aw fugi
200.tif - Alan's, jo clip.tif (Jerry's) and fn fugi 200.tif and fn kodak
800.tif (Franks/mine).

Suggestions - comments?


> -----Original Message-----
> From: Alan Womack [mailto:arwomack01@worldnet.att.net]
> Sent: Sunday, July 22, 2001 1:31 AM
> To: frank@theNichols.net
> Subject: Skin tones and acer noise
> Frank.
> Your 800 scan looks more like what I get with 200 speed film.
> Attached is a private label fuji 200, take a look at the blue channel.
> alan
> Epson Inkjet Printer FAQ: http://welcome.to/epson-inkjet

Attachment: fn fugi 200.tif
Description: TIFF image

Attachment: fn kodak 800.tif
Description: TIFF image

Attachment: jo clip.tif
Description: TIFF image

Attachment: aw fugi200.tif
Description: TIFF image


Copyright © Lexa Software, 1996-2009.