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] Re: Channel alignment & Revelation on LS-8000 banding



I forgot to mention that this misalignment would also be happening with
color films, but it wouldn't be as obvious what the cause was, since
each channel is recording different info.  In fact, it is more of an
issue with color, because with B&W you could just use one channel as you
image channel and toss the rest if it didn't all align.

BTW, you can't use the LS-8000 for this experiment. Since it uses one or
three monochromic sensor lines. All three color channels are created
from either one sensor line (in the fine scan mode) or the three
channels are a mixture of all three lines mixed at the same film position.

In fact, this puzzle has just lead me to the answer as to why the
LS-8000 introduces banding when used in the 3 line mode!

What you are experiencing as "channel misalignment" on a standard
tri-line CCD sensor, where each channel is made up of the result of one
sensor line, producing on color channel, the Nikon LS-8000 patches
together a channel from all three sensor lines in an overlay like an
interleaf pattern.  What is even worse about this design is that any
misalignment would be identical for all three channels in the position,
since three separate scans are take before the film is moved again, only
by changing light source color, so the misalignment would be in
identical places on each channel, in effect multiplying the visibility
of the defect! The slightest misalignment of the film due to dimensional
change, stepper motor slippage, film carrier position, or even changes
due to temperature changes would show up on all three color channels in
the same place.

This explains why Nikon has been unable to fix this problem other than
by limiting the whole scan, all three channels to one sensor line (super
fine mode).

The method used by most other scanner designs might introduce other
problems, but any banding or spacing differences would tend to be
mitigated by the fact that it would not be reproduced in the same
location of each color channel.  It might slightly soften or misregister
the color, but it would hid banding, the opposite of what the LS-8000
would do.

If Nikon also did not calibrate more than the one sensor line, this
would be even more likely to cause banding.

This is actually a Eureka moment I'm having about the LS-8000.

That design with using the 3 lines would have to have very exacting
tolerances to avoid any banding.

Very interesting

Art

Simon Lamb wrote:

> There is a debate on-going on the Imacon users list regarding channel
> misalignment when scanning black and white stock as RGB.  If anyone has the
> time to undertake a smal test could you do the following:
>
> 1. Scan a black and white negative as Positive and save at max resolution as
> 16-bit RGB.
> 2.  Open the file in Photoshop and view at 100%.  Using the channels
> palette, skip between the R, G & B channels (Control 1, 2 & 3) and look to
> see if the image moves (sometimes by between 2 and 4 pixels) between each
> channel.
> 3.  If it moves then it shows other scanners have the same problem aligning
> channels when scanning monochrome stock using RGB sensors, which in some
> cases causes the image to show stretching across the grain.
>
> Useful scanners to try this on would be those competing in the Imacon Photo
> arena (Nikon 8000, Sprintscan 120, Multi Pro etc.).
>
> Thanks to anyone who can do this test and post their findings.
>
> Simon
>
>
>


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



 




Copyright © Lexa Software, 1996-2009.