ns66 wrote:
[...]
IMHO in-body IS can and will work with lens IS and ADDs to stablization, they don't need any kind of coupling. here is why: when image pass through IS lens, to the sensor, it just appears to be a more stable image, just like someone with steadier hands, now why would in-body IS freak out? in-body IS will try to compensate the already-quite-steady image and make it MORE stablized, period. they don't need to be coupled or even aware of each other's existance, and they will make stablization only better.
[...]
Ok, so the lens-IS detects a movement of, say, 5 degree per second and shifts the floating element at the appropriate speed. Now, the body-IS uses the very same type gyro sensors, and will also detect the very same 5 degree per second camera shake...
How will the body-IS electronics know how effective the lens-IS was at taking care of the shake? Remember, the current systems have no live preview, the shutter is closed, the mirror is down, the sensor CAN NOT do the detecting. (Even so, framerate of CCDs aren't that extreme to do high-speed shake detection.)
The point is: the Body-IS system of a hypothetical dual-IS system can not detect/measure how effective the lens-IS was. It uses the same kind gyro sensor input. There is no detecting how much shake there is left to compensate.
The only half-harted way to make these work together is to know EXACTLY how effective the lens-IS system is, and know EXACTLY how much left-over shake will remain... For any possible lens-IS, with every possible lens-IS revision, at any conceivable focal length, at all possible available battery power levels, at any possible combination of pitch acceleration and pitch speed and at any translation acceleration and translation speed... Only then the body-IS electronics will be able to GUESS what sensor shift needs to be applied.
If you don't see what kind engineering, controlling and signal processing nightmare this is, not to mention a debugging nightmare and a fine-tuning nightmare, then I don't know what will.