Well, theoretically in-lens image stabilization might degrade image quality like some sort of decentering defect: It moves lens-groups around in the construction that optimally should sit untilted and perfectly aligned in the center of the optical axis for optimum quality. BUT:
- I have tried to get one shot that made a visible difference between IS/OS/VC/VR=ON and IS=OFF. I couldn't manage to find proof.
- If this effect exists it will show itself at fast shutter-speeds (like say 1/500 sec) and
slow(er) shutter speeds. Because shooting through decentered glass is looking the same regardless of how fast you shoot.
What I'd do is look more on the up-side of IS/OS/VC/VR: under what conditions is it likely you can see a positive effect from stabilization?
Well, I can only offer my personal experience and thoughts on that, don't expect this to be conclusive:
- You profit more from IS at lower speeds: Every control-loop needs some time to react and adjust to do any good: So at 1/125 sec you should see more of an effect from IS than at 1/500 sec. At 1/30 of a sec even more so - but then you might be at such a low speed that the shot is ruined even with IS=ON.
- If you refrain from pixel-peeping (i.e. don't look at 100% magnification) the effect from IS may look better than at 100% viewing. My experience says that IS is much better at stabilizing to an accuracy of 2x2 pixels than at 1x1 pixel. This chimes well in with my warning on the type of fast-moving targets that you're shooting. On a pixel-level I'm pretty sure that you'd almost always might find some fault.
But as I said: that's my twopence only...