Thanks very much Dean. That is food for thought.
Hersch
At 09:26 AM 3/21/2001 -0800, you wrote:
I am surprised that some of the quotes I
included from the ROC patent did
not generate more response. I want to comment on some portions of
ROC (or
at least the patent that ROC seems to be based on).
> "The infrared scan is used to detect imperfections in the film
medium
> itself. As discussed in ... US Pat No, 5,266,805 ...
Unfortunately, in
> the
> infrared scan, there can be cross talk from the red sensitive,
cyan
> forming
> layer which would be identified as defects. The present
invention can be
> used to correct for the red crosstalk in the infrared scan.
>
This implies that the combination of ICE and ROC/GEM should be
greater than
the individual components. By using ROC, the influence of the red
channel
can be eliminated from the IR channel. Net result should be a
clearer
separation of defects from image, particularly the red dyes in slides
and
negatives.
The patent shows the absorption of the red dyes in negatives
extending
further into the IR than slides, so the improvement should be more
pronounced for negatives. Do any users of ICE see strong red colors
in
their negatives fooling ICE? This would probably show up in the
virtual
negative as a brighter red and in the reversed image as a darker
cyan
(compared to the scan with out ICE).
The coupling of ROC and ICE may be enough to allow ICE to work with
Kodachrome. It will be interesting to see if the new Nikons
actually use
this coupling and if the instructions remove their prohibition
against
Kodachrome. Unfortunately this will not help true B&W
film.
>
> "There is virtually nothing above about 40 line pairs per
millimeter
> spatial
> frequency recorded with today's lenses and film from real world
images.
> This cutoff corresponds to a 2000 by 3000 pixel scan of 35
millimeter
> film.
> Conversely, the grain noise begins with flat spectrum and is
attenuated
> only at very high frequencies by grain size and dye diffusion as
discussed
> above, which have an effect above 100 line pairs per
millimeter.
>
I really thought that this statement would get a few comments. If
true,
this would explain the Canon D30 reviews that indicate its image quality
is
greater than or equal to that of film. Probably true for the vast
majority
of images, even those shot using prime lenses with the camera on a
tripod.
The patent author points out that for real images the DOF severely
limits
the high spatial frequencies. For example at 100 lpm and a
FOV of 12 by 18
inches the depth of field is just +- 2 mm at F/2.8. You focus
on the
models eyes, but her eyelashes are out of focus.
>
> "A practical solution first isolates frequencies around 40 line
pairs to
> eliminate those parts of the image in which the energy seen at these
high
> frequencies is predominately from grain noise, and prunes out
or
> emphasizes
> those where the high frequencies also contains image detail.
For example,
> a
> sky, a blurred background, ... Because the noise is a constant
across the
> image, a region that contains more high frequencies than elsewhere
in the
> image is more active because o image detail...
>
This portion of the patent indicates that even if images are limited to
less
than 40 lpm, a scanner that can read at higher frequencies will
have
advantages when used with ROC and GEM. The higher frequencies will
clearly
separate the image from the grain noise, allowing better noise
removal
without affecting the image. It also indicates why reviewers think
that GEM
removes image detail - it does! However, I expect the authors would
argue
that the detail removed was not real, rather it was an artifact of the
grain
noise increasing the apparent resolution.
