Anvh wrote:
It makes sense Ian but that doesn't guarantee it's on focus.
There are some margins build in the lens and the AF sensor of the camera so the tolerances aren't 100% meaning the lens might front or back focus.
I've checked my lenses and they all needed some correction to make them focus correctly with the AF sensor of the camera.
Live view works differently, it use the actually image to verify focus so it bypass all those tolerances by looking at the final image.

So to show the differences you're doing here, focus first with live view since that one is the most accurate and set the lens to MF so that the camera doesn't focus any more.
Then take a photo with live view and then one without live view while leaving the lens on MF, that should deal with the variation of the focus in your photos.

johnriley wrote:
Did you use the 2 second self timer to ensure lack of vibration?

There's no mirror slap anyway with Live View, so the difference just maybe could be down to that.