19 Nov
2010
19 Nov
'10
9:43 a.m.
On Thu, Nov 18, 2010 at 07:52:06PM -1000, david wrote:
I work with 48-bit color TIF images, with color
dynamic ranges way
beyond any printer or monitor's ability to differentiate them. Sort
of the equivalent (to me) of 96 or 192KHz sample rates.
The equivalent of more bits per sample really. For images that
certainly makes sense - nobody ever claimed that 8 bits per
component were sufficient, plus there is the disadvantage of
non-linear reproduction (gamma).
The equivalent of higher sample rates for images is higher
resolution, and again this makes sense. When working on images,
operations that amount to resampling are much more frequent than
they are in typical audio processing (any change of size means
resampling).
In audio there is nothing *wrong* with working at higher sample
rates, but generally nothing is gained by doing it. Except that
some algorithms may become easier to implement. For example
zita-at1 (auto-tuner) will do parts of its internal processing
using the double sample rate when working in 44.1 or 48 kHz.
Ciao,
--
FA
There are three of them, and Alleline.