We present a photon mapping technique capable of computing high quality global illumination at interactive frame rates. By extending the concept of photon differentials to efficiently handle diffuse reflections, we generate footprints at all photon hit points. These enable illumination reconstruction by density estimation with variable kernel bandwidths without having to locate the k nearest photon hits first. Adapting an efficient BVH construction process for ray tracing acceleration, we build photon maps that enable the fast retrieval of all hits relevant to a shading point. We present a heuristic that automatically tunes the BVH build's termination criterion to the scene and illumination conditions. As all stages of the algorithm are highly parallelizable, we demonstrate an implementation using NVidia's CUDA manycore architecture running at interactive rates on a single GPU. Both light source and camera may be freely moved with global illumination fully recalculated in each frame.

Mesh deformations are generally specified by manipulating an underlying control structure. This may have tediously been built in a preprocessing step or computed ad-hoc from user sketches. We propose an extension to the latter approach, deriving three-dimensional representations of a skeleton and the surrounding region of interest from one or two simple strokes. This provides a substantial improvement over existing techniques, which build only two-dimensional structures on the fly. As a single planar sketch is not sufficient to fully describe a threedimensional skeleton, its shape must be deduced by interpreting the input strokes in the context of the model. We explain the challenges involved, analyze several algorithms for interpreting the input and conclude by sketching out the direction of our ongoing work on the topic.