This question already has an answer here:
I have an image stabilizer in my 18-55mm kit lens that came with the Canon EOS 2000D DSLR. I recently tested it with the largest possible focal length of this lens. I saw that the viewfinder image vibrated far less with the stabilizer, so apparently it's doing something.
My 10 test shots with 1/5 second shutter and IS were occasionally shaken, even though I attempted to hold the camera steadily.
However, the 10 test shots with 1/10 second shutter speed were okay if I attempted to hold the camera steadily.
Then I tested with image stabilizer off, attempting to hold the camera steadily.
The 10 test shots with 1/20 second shutter and no IS were occasionally shaken.
The 10 test shots with 1/40 second shutter and no IS were basically okay.
Also, I have tested fast-moving subjects like cars to capture motion blur. As it's wet outside, I took the images through my small window so I have to be quick whenever I see a car driving past. My experience is that with 50mm fixed lens without IS, 1/80 second shutter will be fine even if I have to take the image really quickly and have no time to ensure my hands are steady. Also, my experience with 18-55mm kit lens with IS is that 1/20 second shutter is required with IS on, if I have to capture the image really quickly.
So, my experience is that trying to keep my hands steady, the IS helps only for 2 stops. Similarly, when taking the picture quickly, the IS also helps only for 2 stops.
However, the manufacturer claims the IS is a 4-stop IS. As the stops are exponential, the manufacturer is claiming 16-fold improvement whereas I'm observing only 4-fold improvement.
Can I trust the manufacturer's claims for IS improvement? Or is it just false marketing data?
Is the IS as effective with any focal length, or might it be the case that for this particular lens, the 4 stops advertised only works for a certain optimal focal length, and at the longest focal length, it's not 4 stops?
Note my results with steady hands somewhat contradict the 1/(focal length * crop factor) rule. 1/40 seconds was fine with steady hands even though the rule gives 1/88 seconds, with crop factor 1.6 and focal length 55 mm. However, the 1/(focal length * crop factor) rule seems to be valid for me if I have to capture the image really quickly. But perhaps that's the subject of another question...