>
> What I meant is that it should be spec.ed at
> the number of independent |full color| output
> pixels....
I'm sorry, but you're several years too late with this argument.
The "pixel" count on all digital cameras (with the exception of
B&W cameras and one one other case mentioned below) is the count
of independent sensors. So the "6MP" *ist-D (and D100, D10, ...)
Have 6M sensors; 3M Green, 1.5M Red, 1.5M Blue. The missing values
at each sensor site are interpolated from the surrounding sensors.
While this isn't as good as 6M independent full RGB values (such
as you would get from a film scanner) it's actually quite a bit
better than a 1.5MP full-RGB scan; the human eye is far more
sensitive to changes in luminance than to changes in chroma (a
fact which is used to good effect both in TV transmission and in
image compression algorithms). As the sensor array does sample
the luminance at each receptor site detail resolution comes closer
to that from a 6MP image.
Note that this marketing hype extends even to the Foveon sensors,
which *do* measure R,G and B at each sensor site. There are 3.4M
sensor sites on the Foveon X3 sensor, yet it gets called a 10.2MP
device. This is an even worse exaggeration because there are not
10.2M spatially independant sites sampling the luminance.
The very worst marketing claims, though, come from (?some of?) the
Fuji FinePix cameras. In those the Bayer array is rotated by 45
degrees, thus:
* G *
R * B
* G *
Not only do these cameras interpolate the missing components at
the R/G/B sites; they also perform an extra interpolation step
to get values at the * sites, doubling the number of pixels that
are claimed for a given number of sensors.
