Try a 30 second exposure and post your results
2 seconds doesn't give enough time for the IS to screw things up
Damn! I posted the wrong image. When I did the test I saw there was zero difference at 2s, so I repeated it with a longer exposure. I didn't realise I'd processed and saved the 2s data.
Here's a comparison of the 24-105 and the 100-400 with a 22s exposure....


I will be taking care of that little problem soon, as the I have a sale pending on my 100-400L, and am sure the MKII version will be better on a tripod with the IS on.

