I guess the solution is to render fake video on the fly doing pose and expression emulation which is possible with a few static shots, open source tools, and a halfway decent machine. It might not fool a human, but you dontnhave too and I suspect it won't be long before fooling the average human will be pretty easy too.