Re: istD bit depth

[Paul Stenquist]

Thanks John. We can always count on you to keep us up to date. It will 
be interesting to see where Pentax goes with their next step up the 
ladder. I can't imagine they would call a halt to things now that 
they've plunged headfirst into the digital SLR market.
Paul
On Oct 23, 2004, at 5:36 PM, John Francis wrote:

The scanner has a single row of pixels, so you can use the space
next to the pixel for processing elements.  A camera has to put
another row of pixels there (and another row beyond that, etc.,
etc.) for a 2D array of sensor elements.
The 'blad is probably using larger 'pixels' in its sensor.  That
is partly to get the signal-to-noise ratio down to the level where
having 16 bits makes sense (you want signal in those extra bits,
not just more bits of noise) - larger pixels have a much better
signal-to-noise ratio - and partly because you need a larger area
of silicon for the other circuitry (A-to-D, logic processing, etc.).
All the current 35mm DSLRs seem to use 12-bit sensors; I'm sure there
are 14-bit DSLRs in development, but I'd be really surprised if Nikon
went directly to 16-bit.
At a risk of boring you with numbers:  the current technology seems to
support little more than 8MP from an APS-sized sensor & 12-bit pixels.
Signal-to-noise ratio goes roughly linearly with area, so to get those
extra 4 bits of signal we need 16x the per-pixel area.  A sensor twice
the size of a 35mm frame would only really require a 14-bit sensor for
any resolution greater than that same 8MP.  Perhaps Hasselblad (and/or
Imaco) have some technological edge that really requires those extra
two bits, or have come up with some clever multi-pass exposure tricks
(I doubt if they've managed to get a sensor larger than 2 35mm frames).
Or perhaps it's just marketing ...
Shel Belinkoff mused:
Hi John ...
Couldn't forget that linear stuff since I never knew it <vbg>
Don't really understand the 2D thing.  Are there two rows of pixels, 
one
below the other?  Nah, that can't be it?  So how come the 'blad can 
have a
16-bit sensor, and some DSLR cameras 14-bit?  Is it a matter of space
(which is what I'm inferring from your remarks)?  I heard talk of a 
Nikon
D3 with a 16-bit sensor, BTW ...

Shel

[Original Message]
From: John Francis <[EMAIL PROTECTED]>
To: <[EMAIL PROTECTED]>
Date: 10/23/2004 12:42:22 PM
Subject: Re: istD bit depth
Shel Belinkoff mused:
The istD has a bit depth of 12.  I seem to recall some DSLR with a 
bit
depth of 14 ... maybe.  The specs on the new Hasselblad claim a bit
depth
of 16.  Why is it that so many DSLR cameras are using a bit depth of
12?
Is there a physical or design reason?  Cost?  My little Nikon 
scanner
has a
bit depth of 16 ... why not a DSLR?
Shel
Don't forget that your scanner only has a single row of sensors, not 
a
two-dimensional array, and that it only has to work at a single 
speed.

Of the two, the fact that it's only a linear sensor is more 
important.
You can put the extra processing elements, etc., alongside the sensor
without having to worry too much how much room they take up.  In a 2D
sensor you're trying to put another row of pixels there.