Blendshapes From Commodity RGB-D Sensor

August 10, 2015 | Los Angeles, CA

Speaker: Dan Casas, Graham Fyffe, Ruizhe Wang, Evan Suma, Ari Shapiro
Host: The 42nd International Conference and Exhibition on Computer Graphics and Interactive Techniques

This talk demonstrates a near-automatic technique for generating a set of photorealistic blendshapes from facial scans using commodity RGB-D sensors, such as the Microsoft Kinect or Intel RealSense. The method has the potential to expand the number of photoreal, individualized faces that are available for simulation.