10DFT wrote:
[...]
The last test I saw gave Sony's system a 1 stop advantage over no IS at all. And that wasn't on a very long lens where IS is actually useful.
Do you have any idea how far the sensor would have to move to correct a simple 2-3 degree movement on a 24 mm lens? How about 300 mm? 500 mm? You might want to study the concept a bit before slinging out insults.
For what it's worth, DPReview found the Alpha 100 to bring 2 stops better performance, tested on 50 mm focal length, 75 mm effective.
As for degree of movement, I just tested myself: At about 25 meter distance my aim drifts about 20 cm from the mark I was aiming for, say 40 cm total. That puts my camera shake at, give or take some, just under one degree.
300 mm focal length has a FOV of little under 5 degrees horizontal, corrected for APS-C crop. If my aim is steady within a degree, it means the sensor has to be able to move 1/5th of it's width to be able to compensate. An APS-C sensor is under 24 mm wide, so this proverbial on-the-back-of-a-napkin calculation suggests a correction shift of 2.4 mm away from centre.
Having said this, I know I have a pretty decent steady hand myself, I usually beat the 1/(effective mm) rule by a stop. A movement of up to 4 mm from centre seems possible, looking at the diagrams of the in-lens IS mechanism, so there is room left for more compensation. Even so, it's pretty clear to me that around or above 300 mm you hit the limit with in-body IS.
Oh, one additional note: All this talk about angles only says something on the range of the IS, it has no say on the speed of the IS. It's all fine to have a shift of -x to + x mm available, but as long as there is no information about response time, acceleration and speed, you can not tell which angle speeds can be compensated.