Re: Re[2]: filmscanners: Film Scanners and what they see.

[Rob Geraghty]

Byron wrote:
> > Surely the bit range used by negs should be in the midrange of possible
> > values precisely because it is compressed?
> Yes but these values are quantized to 10 bits on a LS30 rather than 12
bits as
> in other scanners.  So there is four times the quantization noise.  When
> decompressed (ie expanded) this extra quantization noise becomes
objectionable.

Does anyone have evidence of this, or is it supposition?

> > Wouldn't the least significant bits be the dark areas in slides and
beyond
> > the brightest parts of negs?
> No, the least significant bits apply at all brightness levels not just the
> darkest/lightest.

Huh?  Why?  I realise it takes every bit to make a number, and I can see
that things like thermal errors from the CCD will always apply, but surely
the noise is most important when the signal drops to the noise floor (ie
when the signal is defined only by the least significant bits)?

Rob