Video input (camera-based monitoring of the user and his or her actions) offers an extremely promising medium for a rich vocabulary of user interaction. We see a clear role for video as an input channel to advance the use of highly interactive tools and visualizations. But by itself, video input cannot solve all of the problems of the user interface. Video input needs to be carefully coupled with other means of providing enhanced manipulation as well as with salient feedback channels so that the user can act and perceive in a tightly coupled loop.