aimcompute writes:
> As for Kodak's assertion that their films are specially formulated for
> scanning... I'm not sure I believe it, nor do I think that if true, it
would
> necessarily be more desirable than having a film that is specially
> formulated for faithfully and accurately recording the image. I think
all
> films are formulated for scanning by virtue of the fact they can all be
> scanned.
Let me explain how I understand scanning works, and try to offer a theory as
to why some films can be optimized for scanning.
Basically, all film scanners shoot a whitish light through the film, and
record the results in some fixed amount of resolution (8, 10, 12, 16-bits,
etc.). The scanner CCD is calibrated to this white light in hardware using
the scanners' particular calibration method, but not for each scan. In
addition, the scanner itself has a characteristic response curve to the
different colors at different levels. (Using an ICM color profile file for
you scanner is important). I don't think it is possible for the scanner to
change the hardware gain values on each color based upon readings from the
film base edges, as mounted slides don't have this film based edge to read,
and doing so would also change the color profile of the CCD output for every
scan, so it relies upon software to correct for film based differences, of
which some programs are better than others.
When a film is scanned, particularly negative film that has an strong orange
base, the range of available colors is shifted, and compressed. Because of
this compression of the range as captured by the scanners limited pixel
depth, you further reduce the range of pixel depth by a few bits or more.
When correcting for the color shifts, it is not simply a matter of raising
the high & low ends of each color a particular amount, but also a matter of
raising each color level a particular amount based on it's existing (raw)
value (a full correction curve). This is what Vuescan does well (and also
what ICM files provide).
As an aside, I think there is a market for ICM files for particular films.
Does anyone know if these can be bought on the web anywhere?
Now, I will continue with my theory. When a film such as Superia 400 is
scanned, the high amount of variability of correction leads to a loss of
information. Once a scanned image is edited and saved, you can never go
backwards completely and recreate the original file from the edited file.
Essentially, all of the major transforms done during image editing are
one-way streets, and result in the loss of a slight amount of information.
The more corrections that need done, the more loss in information.
A film could be designed to be optimized for scanning by normalizing and
flattening the inherent characteristic color response curves. In this way,
the most information is captured in the raw scan. Users of sRGB monitor
profiles might understand this as a film calibrated closely to sRGB. (This
is also where an film ICC profile will come into play.) So, a film
optimized for scanning is also probably a film optimized to "faithfully and
accurately" capture the image.
Does this seem plausible?
Cheers,
Gerald
-
This message is from the Pentax-Discuss Mail List. To unsubscribe,
go to http://www.pdml.net and follow the directions. Don't forget to
visit the Pentax Users' Gallery at http://pug.komkon.org .
