[Leica] Re: a thought on leica digital

[Jim Brick <jim@brick.org>]

Unfortunately, it's the other way around. We can make sensors the size of
MF film (6x6) and larger, but we cannot get pixels any smaller than, at the
very very best, 3x3 microns at .13 micron semiconductor geometries. Even at
this size, the signal to noise ratio is in the dumper and the dynamic range
of such tiny pixels is in the mud. I have had a 24x35mm 6mp sensor for over
two years. Alpa (yes Alpa) has had a 6x6 24mp sensor for two years. These
are expensive and made with 5 and 7 micron pixels. Dynamic range, color
fidelity, etc, is outstanding. Pixel resolution is simply OK. Remember, it
takes four pixels to produce one color pixel (Bayer pattern.) 5+5=10,
10x10=100 square microns for a color pixel.

With each pixel, in a good, high quality sensor, averaging 5x5 microns,
this is 25 square microns for one 8, 10, or 12 bit pixel. 1024 shades from
white to black. A grain in a fine grain film emulsion averages 1x1 micron
(one micron square) and contains 20 billion silver halide molecules.
Somewhat more density possibilities than 1024, in 1/25th the space.

There will have to be a technical breakthrough in a different technology,
or a catalytic technology, before electronic storage of electrons,
representing the light value of a point, can be much better than what we
have today. Today's technology is against the wall. We can make bigger
sensors. But not higher density.

Jim Brick
Senior Scientist
Imaging Electronics Division
Agilent Technologies
The New HP


At 10:13 PM 9/29/00 -0400, Johnny Deadman wrote:
>A point to mull. The critical problem in chip design is size, not density.
>So once the technology exists to make 35x24mm chips, you can expect the
>packing density of sensors to be very high. This means that it may be
>possible to make cameras that take your M lenses that have the resolving
>power of 10x8 field cameras. Cool, huh?
>
>-- 
>Johnny Deadman
>
>http://www.pinkheadedbug.com
>