In a message dated 9/21/2000 4:38:57 AM EST, [EMAIL PROTECTED] 
writes:

> > They have also asserted that Dmax isn't related to # of bits acquired.
>  
>  That's absolutely correct.

No, Dmax (or actually Dmax-Dmin) is at best log10(2^bits),
where "bits" is the number of bits above the noise level 
coming out of the A/D converter.  It's intended to mean
"how dark can the film get and still have image detail".

When the bits from the A/D converter are mapped to
an 8-bit, gamma-corrected space, the Dmax value is
more complex.  A good gamma mapping will include a
linear segment in the darkest part of the image, so
the Dmax formula still holds.  If the gamma mapping
is a simple power function, then the Dmax is worse.

For example, a scanner with a 14-bit A/D converter
will have (at best) a Dmax of 4.2.  If a scan from this
scanner is mapped to an 8-bit image with a gamma
mapping that has a linear segment for the first .018
out of 0..1 (i.e. the Rec. 709 or sRGB gamma mapping),
then you'll still be able to detect image detail in the
darkest part of this image.  If a simple power function
is used for the gamma mapping, then you won't be
able to detect as much image detail in the darkest
part of this image.

(VueScan uses this linear segment when doing the
gamma correction.)

>  If you get good scans that don't need any global tonal work, getting 24 bit
>  scans is just fine. I'd suggest they scan into a Working Space (which on
>  their scanner can be carried out in high bit). Having high bit scans IS 
nice
>  but not super critical if you get a good 24 bit file you don't intend to
>  really mess with global color and tonal corrections.

If you're picky about getting detail from the dark parts of scans, then
don't get 24-bit files from the drum scans.  If the scans are in a wide
gamut color space, then you also probably shouldn't get 24-bit files.

Regards,
Ed Hamrick

====================================================================
The filmscanners mailing list is hosted by http://www.halftone.co.uk
To resign, <mailto:[EMAIL PROTECTED]> with UNSUBSCRIBE FILMSCANNERS in the 
title, or UNSUBSCRIBE FILMSCANNERS_DIGEST if you are reading the Digest.

Reply via email to