Originally posted by KjetilH If you are going to use SR, put in the lowest you are going to use.

SR is a counter movement of the sensor.

Let the amplitude of this counter movement be A. With a perfect SR system this means that the image without SR would shake with the same amplitude B=A.

The resulting shake has amplitude C=A-B (=0)

Now assume that real focal length f_b and dialed in focal length f_a differ, then

C = A-B = a f_a - b f_b = a (f_a - f_b).

This is minimal if you choose f_a in the

*middle *of the focal range you are going to use.

However, the efficiency of SR is easily reduced from 3 to only 1 step then. But still better than off for less than 4x zooms.