Digital BW, The Print

> From: Clayton Jones
>
> No, it doesn't.  That's what everyone assumed up until this year's
> digicams hit the market.  The canon G6 has 7mp on a 1/8 chip, and has
> lower noise and greater dynamic range than the G5 with 5mp on the same
> size chip (it actually has lower noise and DR than my pro-1 with 8mp
> on a 2/3 chip-it's amazing).  That was the crux of Michael Johnston's
> column on the subject on the Lum. Lndscp web site awhile back, where
> he put forth convincing reasons for full frame sensors not becoming
> the standard.   Everything is trending smaller, not bigger.  They keep
> improving on it and nobody knows what the limits are yet.  All the old
> wisdom is up for grabs right now.

But that's been true all along. Semiconductor design advances, as well as
signal processing advances, have always decreased the noise levels for a
given pixel size. But it still remains true that for a given technology
making the pixels smaller increases the noise, so there's always a tradeoff
between pixel count and noise level.

Also, there is a theoretical limit, which is that ultimately light is a
"digital" phenomenon: it's made of photons. In daylight conditions, there is
such a torrent of them that they behave like an analog signal, but anyone
who's played with astrophotography knows that at very low light levels,
there is an intrinsic noisiness to the light itself, since it consists of a
stream of photons arriving at a certain average but irregular rate. Some
labratory sensors achieve something like 90% photon conversion efficiency,
which means that they've essentially reached the theoretical limit; the
remaining 10% would have no visible effect. They're still quite a ways away
from doing that in a CCD or CMOS sensor, though.

--

Ciao,               Paul D. DeRocco
Paul                mailto:pderocco@...