At 06:56 PM 10/27/2002, Winsor wrote:
><snip>
>That is interesting and I am sure you are right, but I am not entirely
>convinced. If you put a recessed sensor in a chip and then create one 3 times
>as wide and 3 times as deep I am not sure that you have gained anything.
>Since the bigger sensor elements are probably thicker it is probably more than
>3 times as deep. <snip>
I think there are several reason for better potential performance of
the Foveon chip. The Foveon chip uses the fact that different
wavelengths of light are absorbed in doped silicon at different depths.
Shorter wavelengths are absorbed near the surface (blue light) while
longer wavelengths absorbed deeper (red light). The ion implantation is
adjusted to the different color depths to create junction isolation in
the silicon to electrically capture the different wavelengths. There
are no filters stacked on top of the silicon. The doped silicon itself
is the color separator. The deepest junction is only 2 microns deep for
red light. The junction depth of a CCD charge bucket is probably deeper
(10um? more) plus you have the RGB filter coating thickness. Hence the
steeper the angle of the light before it is absorbed and converted to
photos is going to be worse for CCD type images. It is filtered before
it is absorbed in the silicon. Combine with the RGB pattern and you get
the effect of chromatic aberration.
The Foveon chip does not use filters to separate the light, but rather
the natural separation caused by the doped silicon itself. The
filtering and absorption is at the same depth, so the angle of light
has little effect. I think 2um thin is comparable to the best film
emulsions. Check out:
<http://www.dpreview.com/news/0202/02021102foveonx3tech.asp> for a
diagram of the silicon junction depths for the color separation. I can
see no reason why a full frame Foveon chip would not work fine with
conventional wide angle lenses.
What would be really cool is an image sensor like I saw in the Solid
State Circuits conference a few years ago that did exponential
conversion of the light (mantissa plus exponent) and easily had 10
stops of dynamic range. If they could combine that with the Foveon
technology - now we're talking serious digital photography.
Wayne
< This message was delivered via the Olympus Mailing List >
< For questions, mailto:owner-olympus@xxxxxxxxxxxxxxx >
< Web Page: http://Zuiko.sls.bc.ca/swright/olympuslist.html >
|