Archived from groups: rec.photo.digital (
More info?)
In message <0slv01hnjbk30rv8p8vsu8lq4npcifedgt@4ax.com>,
drs@canby.com wrote:
>On Sun, 13 Feb 2005 20:32:02 GMT, JPS@no.komm wrote:
>
>>20D seems to be straight-up at ISO 100, unless Canon covered their
>>tracks very well. They leave the data in 20D ISO 3200 shots all even,
>>showing that it is 1600 pushed to 3200, so I don't know if they'd try to
>>hide anything (perhaps they were ridiculed for the fake odd numbers in
>>the 10D's 1600 and 3200 modes, which are both pushed a stop).
>
>I quickly get more information than I understand. And that's fine.
>
>Does it accurate to think of the changes in ISO settings to be linear
>as far as potential noise generation is concerned?
If the quantization of the data were not an issue, this would be true.
Imagine, however, that if by some magic, all of a sudden the camera had
ISO 50, 25, 12.5, 6.25, 3.125, 1.5625, etc. All of the super-low ISOs
have the same limitation; the bit depth of the image, which breaks up
the shadows, by posterizing them. It's not noise, per se, but it might
as well be. Once you visualize this, then it is easy to see how the
"noise" in the shadows at ISO 200 is not twice as bad as the noise at
ISO 100, even though it is about twice as strong, in the sensor itself.
It's like, do two drops of rum in a glass of coke taste twice as strong
as one drop? No, but 4 ounces will certainly taste stronger than 2
ounces of rum.
--
<>>< ><<> ><<> <>>< ><<> <>>< <>>< ><<>
John P Sheehy <JPS@no.komm>
><<> <>>< <>>< ><<> <>>< ><<> ><<> <>><