Originally posted by JohanBene if 70% of the light reaches the sensor that equals (ln 0,7)/(ln 2) = -0,515.
A loss of half an EV-step
Correct.
And the effect on ISO is even simpler.
E.g., one has to compare e.g.,
SLR ISO 1600 vs. SLT ISO 1120 (1120 = 0.7 * 1600)
when comparing sensors and camera electronics.
Or when judging cameras,
SLT ISO 1600 should look like SLR ISO 2285 (2285 = 1600 / 0.7).
The effect of the semitransparent mirror is like a 1/SQRT(0.7) or 1.20 crop factor which is close to APS-H (if it were full frame) and increases the crop factor (as far as ISO is concerned) from 1.53 to 1.83.
So, the ISO performance of the Sony SLT cameras should be expected to be in the middle between APS-C and FourThird systems and if one ignores APS-C for a second, is almost (within 10% distance) like FourThirds.
Originally posted by Groucho ...and apart from the D700, most don't have dramatically better high ISO performance than, say, the K-x.
The almost 2 years old D3X sensor (made by Sony?) is the point of reference. It matches the D700 in high ISO performance but has
far better dynamic range. The newer D3s is a bit better in high ISO performance though.
K-x is more than one stop below D3X in high ISO performance and one stop below in dynamic range despite it being DR king in APSC land.
That's why I made a distinction between FF cameras and crippled FF cameras in my blog article about Photokina and FF.