“Interface techniques for 3D control of spatial keyframing” by Mordatch, Coleman, Singh and Balakrishnan

Spatial Keyframing offers a compelling approach for interactive control of 3D characters, as user motion directly controls the timing of an animated interpolation among a set of target poses [Igarashi et al. 2005]. In the original system, the user sets up a collection of target icons with associated character poses in a two dimensional control space that is overlaid with an interactive rendering of the 3D character. Mouse pointer location in this control space defines a multi–target pose interpolation, allowing navigation to become a form of interactive character animation. We are investigating techniques for using 3D control spaces (Figure 1), as the extra dimension provides more freedom in terms of specifying motions among the target poses, similar to how 2D control provides more potential motions than a one–dimensional slider that transitions through a fixed sequence of poses. This presents control difficulties, however, as mouse motion and traditional displays are inherently two- dimensional, which makes setup, control, and visualization more difficult. We are developing techniques that assist users with positioning new pose targets and choosing control space views that maximize the number of reachable targets. In addition, the 3D control space allows us to use novel interfaces to capture expressive 3D hand motion without the constraints associated with mouse navigation. We are taking this approach to create a virtual puppeteering system, in which users navigate a true 3D control space to control a 3D volumetric rendering of an interactive character.

1. Grossman, T., Wigdor, D., and Balakrishnan, R. 2004. Multi-Finger Gestural Interaction with 3D Volumetric Displays. In Proceedings of the UIST 2004, 61–70.
2. Igarashi, T., Moscovich, T., and Hughes, J. F. 2005. Spatial Keyframing for Performance-Driven Animation. In Proceedings of SCA 2005, 107–115.