Digital BW, The Print

One reason to have more bits in the A/D is to give yourself a margin for 
error in an image for under or over exposure or in RF systems for bad 
gain settings. So there are two issues, the S/N required and the dynamic 
range that insures that S/N is still captured even in the case of a 
screw up.

Truman

Paul D. DeRocco wrote:

> >
> I actually ran some rigorous tests on my Minolta DiMage 7. Since its raw
> output is uncompressed, I was able to write a program that rounded it 
> off to
> fewer than the 12 bits that the camera produces. On a test shot of 
> blue sky,
> I found that I had to reduce to 8 bits in order to see any difference 
> after
> applying a drastic curve, and even then I couldn't see any actual banding.
> My conclusion was that with that noisy CCD, the extra four bits of raw 
> mode
> data were complete rubbish.
>
> My Canon 10D has nearly two bits better S/N, based on comparing results at
> various ISO settings. This means that a 10-bit converter would be enough,
> although 12 bits isn't a significant extra expense. I've never used a
> scanning back, or seen any tested, but I see no reason why they couldn't
> perform as well as a good CCD film scanner, which is probably around 12
> bits.
>
> Of course, the only reasons for needing more S/N is so that you can 
> shoot in
> very low light, or shoot very contrasty subjects and pull detail out 
> of the
> muck. I suspect scanning backs are intended to be used in the studio, 
> where
> one has control over the lighting, so don't even really need an enormous
> dynamic range. Their primary design goal is just to get huge numbers of
> pixels. The _really_ quiet sensors are used in astrophotography.
>
> --
>
> Ciao,               Paul D. DeRocco
> Paul                mailto:pderocco@...
>
>


[Non-text portions of this message have been removed]