The steady bottom line...
Our tests confirm what anecdotal evidence has suggested: You can generally expect to handhold lenses two shutter speeds slower than you ordinarily could, and sometimes stretch it to three.
But our tests also confirm that this technology is not a cure-all. Here are some caveats and tips regarding use of anti-shakers:
It doesn't prevent blur due to subject movement. Except specifically for panning, anti-shake is not well suited to action subjects. Sports photographers-except those specializing in wheeled sports-usually don't need it.
Anti-shake gives you consistency, not perfection. We got a huge variation in blur factor without stabilization as shutter speeds went down. In simple terms, sharp shots became a matter of luck. Anti-shake improves your odds, but won't cover for sloppy technique.
Practice makes better. Use the stabilization in the viewfinder as a form of biofeedback-wait until the image calms down, and fire the shutter gently during that lull. With Nikons, you can choose to have VR actuated only at the moment of exposure, but we don't recommend this.
You can use slower speeds the farther away the subject. We tested at moderately close range, where shake is magnified. At infinity, you might gain as much as two more shutter speeds over our "safe" speeds. Conversely, in tight close-ups, you may need faster speeds.
Think depth of field. Nature shooters love anti-shake-it lets them stop down an aperture or two for more focusing accuracy in close-ups.
Little cameras are steadiest. Good results handheld at 380mm or 420mm at 1/50 sec-and slower? The Lumix, PowerShot S1, and DiMAGE A2 blew everything else out of the water. These cameras are easy to hold steady, and have no SLR mirror slap or focal-plane shutter shudder.
You generally see a greater anti-shake effect at longer focal lengths than at wide focal lengths.
Steady shooters, beware. Paradoxically, human tripods can see greater quality dropoff as they go to slower speeds. Steady holders can get away shooting at lower speeds than shaky shooters, but at the point where they can no longer keep the camera steady enough, they may be using shutter speeds so low that even anti-shake won't help much.
Image stabilization really is rocket science. But it's easily understood rocket science.
Take a picture with shaky hands and/or a too-slow shutter speed and you get a blurry image. The image blurs because it's moving across the imaging plane while the shutter is open; the picture quite literally smears across the frame area.
Optical image stabilization works by shifting the image projected through the lens in an equal and opposite direction to the movement of the shake. This is done with a motorized 1.0X corrector lens or lens group located at the optical center of the lens. (The other system in use, Konica Minolta's Anti-shake in the DiMAGE A2, shifts the CCD to compensate, but the principle's the same.)
Fine, but how does the corrector know which way to shift, and by how much, and in real time? That's where the second major component of image stabilization comes in. Motion detectors (tech types know them as microaccelerometers, or gyro sensors) measure the amount and direction of shake, and transmit this information to a CPU built into the lens. The CPU in turn directs the corrector motors to shift accordingly.
And, voilà, your picture is sharp.