I thought that the current canon cameras were actually hitting the ISO standard for exposure these days. They used to be under rated. (So ISO 100 would really be ISO 160 on an older canon camera. On the newer ones ISO 100 is ISO 100.)
[a href=\"index.php?act=findpost&pid=174045\"][{POST_SNAPBACK}][/a]
This is the metering. The real RAW sensitivity is still another issue. Cameras vary in how much highlight headroom they leave above middle grey in the metering (in the green channel, which is the reference channel), by anywhere from 2.2 to 4 stops in the samples I've seen, so in a sense, one could say that the actual saturation-based sensitivity varies about 1.8 stops, even given an accurate metering ISO.
The semantics get a bit stranger when you consider the fact that some cameras, like the Phase P45+ leave the full sensor DR in RAW files at all ISOs, so ISO 200 has a stop more highlights than ISO 100, 400 a stop more, etc.
What is still in a black box, really, even after many tests, is what the *absolute_signal* to noise ratio is at various ISOs. That's why it is sometimes difficult to make statements about cameras' absolute SNR sensitivities; what you can measure from RAWs from a camera is more related to DR and SNR throught the RAW's DR, but what it relates to in the real world may be off a bit from what is assumed. That's why I always suggest shooting the same scene, same FOV, same manual expsoure, in RAW, to compare cameras. Nobody seems to want to follow a hard line to the truth, so we are left to speculate.
All this stuff should be in the camera specs.
It would really be nice if someone could come up with a planar light source, like a light table, with calibrated and totally diffuse intensity, to provide a real-world reference that cameras could shoot in isolation, with and without transparency wedges.
