abstract: a new technique for representing tabletop input within a game physics engine is explored. The technique allows users to manipulate digital objects in ways analogue to the manipulation of real objects. Vision-based input is represented directly within the games physics engine to utilize its capabilities of dynamic physical behaviour. The technique is able to sense arbitrary input information, allowing manipulation by fingers, hands, arms or physical objects. A user study of the different aspects involved in the technique showed that more work is required to further enhance and explore its capabilities.