# rec.photo.misc

## Subject: Re: Superior Technology: Foveon X3

"Xiaoding" wrote:

>That is a classic mathematical answer to a non-mathematical problem.

I have spent a lot of time on this subject. Any point I am not likely
to get probably doesn't have one.

>And yes, the Bayer
>does, indeed, fill in the missing color information with it's "guess"
>of what should be there.

Guessing is not a good choice of words. Interpolating is more accurate.
It's nothing like, what's that dark pixel against the sky; is it a bird?
Let's draw a bird there". Most demosaicing works on a deterministic
alorithm; no guessing.

>I call it crayon, a more colorful metaphor,
>which is ironic, since the Bayer has a problem with color. :)

Bayer doesn't have any special problem with color. It simply doesn't
sample color at the full resolution of the image. The color
discrimination is very good; much better than the X3. The X3's
frequency response graphs in the Foveon whitepaper show an absence of
three similarly-shaped bell-curves staggered for the three colors, like
most Bayer cameras do. Either the green or blue is flat over a range of
frequencies, and neither is particularly steep where they cross.
Foveon, before teaming with Sigma, stated that a special set of filters
used in conjuction with the sensor are needed to get good color
discrimination. Sigma did not use these filters, probably because that
would add to the expense of manufacture, and would reduce the
sensitivity of the camera.

>Color
>is not resolution.

Nobody said it was. Color *HAS* resolution. A Bayer with 4x as many 2D
pixels captures just as much color resolution as an X3 for red and blue;
one with 2x as many has as much resolution of green.

--

<>>< ><<> ><<> <>>< ><<> <>>< <>>< ><<>
John P Sheehy
><<> <>>< <>>< ><<> <>>< ><<> ><<> <>><