This is the most interesting aspect of all of this. It works for music, too: what if your audio playback device could modify a song in real-time to add or remove instruments, or play in a different style. The same could be done with a movie. This also has great applications for musicians looking to learn a piece. YouTube's ability to slow a video without modifying pitch has been really great for this application, too.
You can already do vocal replacement and create new covers of old songs.
High quality pitch matching and vocal replacement is here [1] and is going to be big. You can generate Smash Mouth's "All Star" as every single president, celebrity, vocalist, or cartoon character. Or whatever you feed it.
Within the next year the field will move onto changing instrumentals and generating novel tracks.
[1] The tool I've written and will be deploying as the next version of FakeYou.com does this at high fidelity
Instead of selling the 4K version of an old movie, now they can re-imagine it in 10 different styles.