Video images live in a 3D world. This project is about unlocking that 3D information to allow complex special effects with simple user interaction.
Video looks at a 3D world, so users working with video should be able to interact with that 3D world. Editing and interaction with the video, however, remains based on 2D interface paradigms which have evolved little from the era of celluloid, scissors and tape.
In our view of video manipulation, massive automatic precomputation on the video adds metadata which then allows users to make complex special effects with very simple real-time user interfaces. For example, we show how 3D objects can be inserted into the video, and how these virtual objects can be correctly occluded by real objects in the scene.
Examples (click images for video)
 |
 |
 |
| Animated logo | Without Occlusion | With Occlusion |