SCAPE: Shape Completion and Animation of People
D. Anguelov, P. Srinivasan, D. Koller, S. Thrun, J. Rodgers, J. Davis

We introduce a data-driven method for building a human shape model that spans variation in both subject shape and pose.  The method is based on a representation that incorporates both articulated and nonrigid deformations.  We learn a pose deformation model that derives the nonrigid surface deformation as a function of the pose of the articulated skeleton.  We also learn a separate model of variation
based on body shape.  Our two models can be combined to produce 3D surface models with realistic muscle deformation for different people in different poses, when neither appear in the training set. We show how the model can be used for shape completion  ---
generating a complete surface mesh given a limited set of markers specifying the target shape.  We present applications of shape completion to partial view completion and motion capture animation. In particular, our method is capable of constructing a high-quality
animated surface model of a moving person, with realistic muscle deformation, using just a single static scan of that person. 

Paper:  [PDF - 1.8MB]
Video: [gzipped MP4 - 30MB]  You will need to have Quicktime installed to view this movie.

NEW! A different (and better) derivation of shape completion optimization in my thesis.
           Also, new results on non-linear pose regression.

Motion Capture Animations: [Boxing - animated GIF 4MB] [Walking - animated GIF 3.5MB] [Spinning - animated GIF 5MB]
These animations were obtained by shape-completing motion capture data using our model.

Pose Space Animations: [Sequence1 - Indeo AVI 1.8MB] [Sequence2 - Indeo AVI 2.1MB]
Pose and Body Shape Space Animations: [Sequence1 - Indeo AVI 2.6MB] [Sequence2 - Indeo AVI 2.6MB]
These animations were obtained by linearly interpolating between a few sample poses, and body shape parameters.

A subset of the SCAPE dataset has been made available for research purposes on this webpage . Please email prof. James Davis ( davis at ) to get the password. No commercial usage of the data is allowed.

The datasets contains 71 registered meshes of a particular person in different poses.
We provide:
Please email  if you have any questions.

[Back to the Projects page]