On 2009-10-01, John Mills <johnmi...@speakeasy.net> wrote:
>> With cameras which use more advanced versions of the Apical Iridex
>> hardware or firmware (starting with Sony, but Nikon is reported to be
>> in process of catching up), the situation is not as clear.  I did not
>> see any report of RAW processor which can match Apical-style "Dynamic
>> Range Optimizations".
>> So: there might be one respect (tonal mapping, sometimes called
>> "dynamic range") in which RAW-processed-JPEG might be not as good as
>> in-camera one...

> I'm not sure I follow that, unless the sensor's bit-depth and that of the 
> camera's RAW format are different.

Sensor bit-depth is an absolutely bogus metric (unless one uses it as
"an indicator of amount of R&D", which may correlate with other,
important issues; such as read noise and correlation of noise of
nearby pixels).

If RAW files were compressed to 8-bit gamma=2, they won't loose
"practically any" information; 9-bit would be a significant overkill
(assuming full-well about 70K electrons, as typical large-sensor dSLRs
have).  gamma=2.2 is very similar.  (The special significance of
quantization after gamma=2 is that Poisson noise becomes
"constant-width", thus dithers in dark parts as well as in

If you do not know what DRO is, look on dpreview, and/or look for
examples on Apical site.


Gimp-user mailing list

Reply via email to