On Fri, 12 Jan 2001 13:07:25 -0500 Austin Franklin ([EMAIL PROTECTED]) wrote:
> Scanners use DMax as part of their specs, which is what this whole
> discussion has been about. What's your point?
No they don't. They use OD (Optical Density) range - which is the number you get when
you subtract DMin from DMax, for a film target. This is a logarithmic expression of a
density range which could equally be expressed as F-stops, a *ratio*.
Any consideration of how bit depth relates to ODR will be mistaken unless this is
borne in mind. It is not simply a matter of chopping up any ODR into 256 or more
equal values - what appears at the CCD is *not* logarithmic, like ODR, but an
exponential ratio of voltages with the CCD noise floor as the minimum value. This may
or may not be the same as DMax for any given film sample - DMax is a film property,
not a scanner parameter.
At some maximum illumination - lightsource unimpeded by film - voltage from the CCD
will reach a maximum value. In practice, any film target will reduce this somewhat,
because of its DMin (base opacity + base fog).
The ratio of the minimum:maximum voltages between these two end points is the CCD's
analogue proportional response to film transmissive densities DMin:Dmax in F stops.
And like F stops, it is a ratio, it is not logarithmic.
A ratio of (say) 1:1024 *cannot* be mapped linearly to any smaller ratio, eg 1:256 of
8bits. So if a film exhibits 1:1024 (ODR=3.0, or 10 stops), but you try to digitise it
using an 8bit ADC, you will lose data from either or both ends of the range. In this
case, 2 stops.
Yes, you /could/ do it if you frigged around with the analogue voltages from the CCD,
upstream of the ADC by using some non-linear analogue amplification scheme which
changed the ratio, but generally the sort of scanners we are talking about don't do
this. Bit depth thus imposes a ceiling to ODR in these units - which is why Julian was
correct to say that bit depth imposes a theoretical maximum value.
This theoretical maximum is what some mfrs quote as ODR, and this appears to be
Nikon's stance with their new scanners. In practice, the ODR will always be lower,
thanks to noise etc, which is why measurement is better. However, as Polaroid's Dave
H. says, there is no standard measurement scheme, hence no comparability between
different mfr's quoted ODR's even when they are the result of measurement instead of
theory uncontaminated by reality.
Regards
Tony Sleep
http://www.halftone.co.uk - Online portfolio & exhibit; + film scanner info &
comparisons
