[filmscanners] Re: Using high res digital camera forscanning/duplicating

["Arthur Entlich" <artistik@shaw.ca>]

One aspect of most digital camera sensors (I think the only exception is
the Foveon chip) is they all use some form of color separating filters
on a black and white sensor, creating a color matrix.  In other words,
no digital camera sensor provides RGB separation per sensor.  The sensor
is only used to determine R G or B levels (or in the case of some Sony
lenses two different greens).  They still bias to the green accuracy.

To explain further, a standard scanner, flatbed or dedicated uses some
form of tri-color CCD strip (or in Nikon's case, a black and white CCD
with changing LCD back lighting).  Ever single location that is scanned
is done so by three unique sensors (other than Nikon again, that uses
another approach to get to the same place), R, G and B.  So every
location that is read for measures R, G and B content or color separation.

That is not, however, how digital cameras operate.

Digital cameras sensors are mainly some version of the Bayer pattern
matrix over a black and white sensor. The typical Bayer matrix is as
follows:

RGRGRGRGRGRG
GBGBGBGBGBGB
RGRGRGRGRGRG
GBGBGBGBGBGB

R = a red filter
B = a blue filter
G = a green filter

So any one location on the sensor grid is only color separated for one
primary color.  That sensor has no idea how much of the other primary
colors exists, only luminosity.  You will also note that in this basic
form, 25% of the sensor locations separate for Red and the same for
Blue, and 50% separate for Green.  Then sophisticated, but still
inaccurate, interpolation schemes are used to fill in the blanks.

Other electronic filters are used to try to screen out further errors,
and do so with relative effectiveness, but still color resolution is not
what it is on a flatbed or dedicated triline CCD.

There is no simple way to overcome this limitation, and it also explains
why a digital camera sensor can capture an image so rapidly, while a
scanner takes considerable time.  I suppose, just like with digital
camcorders, as prices continue to drop on CCDs, eventually a tri chip
still camera may come about at reasonable price, but for now, most if
not all (other than the Foveon) that are available to the general public
have this limitation.  I would guess that reproducing an analogue form
like film which has randomized dye clouds of colors might even further
confound this matrix grid approach.

So, while a "digital copy camera" may be a good quick answer for
non-mission critical work, the triline or similar CCD approach has some
advantages that aren't easily resolved with digital cameras today.

Art


mail@imagesbykehl.com wrote:
> Is the technology to the point yet where we could use a high quality
> duplicating lens and a high resolution digital camera (Canon/Nikon  12-16mp)
> to digitize slides and negatives.  What would the potential pitfalls be?
> Would Vuescan work to color correct the negatives?
>
> Forgive me is this subject has already been discussed to death.  I've been
> away from the list for a few years.
> If has already been addressed someone please direct me to the archives.
>
> Thanks
>
> Bob Kehl
> ImagesByKehl.com
> online for four years - thanks to this group
>
>

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