[filmscanners] Re: rebuild your scanner and get better results

["Arthur Entlich" <artistic-1@shaw.ca>]

Major A wrote:

> Austin,
>
>
>>Well, I'm not sure what you mean by "spatial profiles", but spatially, the
>>light sources AND the sensors are in "spatially" different places.  Also,
>>changing the color of the light source means differing the exposure time,
>>hence calibration for that light source.  In my experimentation having
>>designed a few scanners, I found that the optimum exposure times for each
>>color can vary widely, and thus, calibration for each color is very
>>important.
>>
>
> Exposure times are dealt with separately for each colour in the
> LS-8000, just as it has to be. There is no need to calibrate for each
> CCD element since the colour response of the CCD elements is
> identical. BTW, exposure times is the same throughout the CCD, unless
> you want to scan pixel-by-pixel and wait a week for a scan to
> complete.
>
>


This is contrary to everything I've understood about both calibration
routines and how CCD sensors operate.  Typically, color scanners (not
Nikon design) use a tri-line sensor, each with a color separation filter
on it; one red, one green, one blue.  Each line is composed on a line of
CCD sensors in a certain positional density.  The image is projected via
a lens and mirror optical path onto these lines of CCD sensors during
the scan. AS I understand it, during the calibration process, each CCD
sensor within each line is calibrated to a known black and white point.
  This is because not only does the response of each CCD vary, but it
can vary based upon use, temperature, etc, which is why a full
calibration is usually done before each final scan.

Now, with the Nikons either a single line CCD is used, or a tri-line
(with the LS-8000).  Since the Nikon doesn't use color separation
filters, but instead colored LEDs, the CCD line(s) is/are without
filter(s).  Since the same line is used for the R, G and B reading, only
one is required, but three (or four including the IR) readings are taken
at each film position, as the LEDs flash through their sequence.

The LS-8000 doesn't require a triline, but by incorporating it, 3 film
positions can be read at the same time.  This probably also explains why
the LS-8000 uses a diffuser because the 3 lines of sensors are spaced so
  the LED lighting needs to be evened out to lessen the variability.

Unlike other scanners, where each of the CCD lines is dedicated to one
color channel only, based upon the separation filter, with the Nikon
   each CCD line records the color separation image for all three color
channels.  Since I'd think CCD sensors' response differs with the color
frequency they are presented, I don't know what, if anything, Nikon does
to calibrate for the continually cycling colored LEDs.

If they aren't calibrating more than one of the three lines of the
LS-8000, that might explain banding fairly easily if all three are
incorporated (you can use just one with super fine mode).  But, I'd be
very surprised if Nikon is only calibrating the CCD "line" as a
totality, rather than individual CCD sensors.

There is another potential cause for banding with the Nikon design,
which I have posted in the past, so I'm not going to repeat it here in
detail.  It simply is that all three color separations are done at the
same time, at the same location, of the same line of film, with the same
CCD sensor line, unlike all other color scanners which scan each color
channel with a different CCD line, at a different locations at a
different film position.  Nikon's method would tend to amplify the
visibility of any slippage or stepper motor mispositioning or even any
film dimensionally variations.

Art



----------------------------------------------------------------------------------------
Unsubscribe by mail to listserver@halftone.co.uk, with 'unsubscribe 
filmscanners'
or 'unsubscribe filmscanners_digest' (as appropriate) in the message title or 
body