That’s an interesting solution that I’ll have to look at in the future.
The method I settled on was to not attach the camera to the skeletal mesh at all. If you pay attention to games like Mirror’s Edge you can see they’ve actually just added a body to a typical First Person setup, rather than adding a First Person Camera to a third person setup. That way the arms move with the camera and any hand animations are always going to be in the correct place, and then the body is only visble when you look down. In my experience since this original post, it’s a much easier system to work with since the body only needs movement animations.
I wrote an updated tutorial with my results a few years back:
