Daniel Browning |
February 10th, 2010 11:04 PM |
Quote:
Originally Posted by Kalunga Lima
(Post 1484233)
I came upon this ISO vs Noise Level graph.
|
There are just so many things wrong with the "testing" for the graph to have any useful information. I don't even know where to start. For one, analyzing test frames with the lens cap on is fine if you know how to correlate that with actual performance. Clearly the tester who made that graph does not. Furthermore, the bend in the curve occurs at exactly the arbitrary tonal level selected by the tester. Other levels would have bent the curve in other places, which obviously removes all meaning from the graph. Even for a still photographer, it has no correlation to anything useful whatsoever, but that goes several times over for video use.
Quote:
Originally Posted by Kalunga Lima
(Post 1484233)
It would seem that the 5D respond best when set to it's "naive" ISO settings (100, 160, 320, 640, 1250) than to intermediary settings which it apparently must extrapolate.
|
Actually, it is the 100, 200, 400, 800, 1600, and 3200 settings which are not interpolated. 160, 320, 640, and 1250 are the result of digital manipulation.
Quote:
Originally Posted by Kalunga Lima
(Post 1484233)
Strangely enough, according to the graph, the 5D Mkii shows less noise at ISO 1250 than at ISO 125, after which things get exponentially nosier from ISO 1600 onwards.
|
For a fixed exposure and brightness, it is true that 1250 has less read noise than 125. And so does ISO 1600. But the test was not based on equal brightness, which was not the condition under which the test was made, so it only shows the correlation by accident.
In any case, fixed exposure/brightness only applies to raw photographers. For video use, exposure has to float with ISO to retain fixed brightness, and in that regime, ISO 1250 has far more noise than ISO 125. Anyone with a camera knows that already.
|