Model production for animation libraries

A system comprising:

• one or more computing devices, including at least one processor, for:
• producing a first plurality of polygon meshes from a first plurality of images depicting positional changes of a surface feature of a first object in the plurality of images as captured during a first motion capture session, the first plurality of polygon meshes including a polygon mesh representing a location of the surface feature for each image in the first plurality of images;
• decomposing the first plurality of polygon meshes to produce a first motion model that may be used to substantially reconstruct representations present in the first plurality of images;
• producing a second plurality of polygon meshes from a second plurality of images depicting positional changes of the surface feature as captured during a second motion capture session different than the first capture session, wherein the second plurality of images are optical images captured from one or more cameras, the second plurality of polygon meshes including a polygon mesh representing a location of the surface feature for each image in the second plurality of images;
• decomposing the second plurality of polygon meshes;
• producing an enhanced motion model by combining the decomposed second plurality of polygon meshes with the first motion model;
• producing an image of the first object from the enhanced motion model.

A computer program product tangibly embodied in a storage device and comprising instructions that when executed by a processor perform a method comprising:

• producing a first plurality of polygon meshes from a first plurality of images depicting positional changes of a surface feature of a first object in the plurality of images as captured during a first motion capture session, the first plurality of polygon meshes including a polygon mesh representing a location of the surface feature for each image in the first plurality of images;
• decomposing the first plurality of polygon meshes to produce a first motion model that may be used to substantially reconstruct representations present in the first plurality of images;
• producing a second plurality of polygon meshes from a second plurality of images depicting positional changes of the surface feature as captured during a second motion capture session different than the first motion capture session, wherein the second plurality of images are optical images captured from one or more cameras, the second plurality of polygon meshes including a polygon mesh representing a location of the surface feature for each image in the second plurality of images;
• decomposing the second plurality of polygon meshes;
• producing an enhanced motion model by combining the decomposed second plurality of polygon meshes with the first motion model;
• producing an image of the first object from the enhanced motion model.

A computer-implemented method for generating a model for an animation library, the method comprising:

• receiving a first motion model produced from a first plurality of images depicting facial expressions associated with a performance of an actor captured during a first capture session, wherein the first motion model may be used to substantially reconstruct the facial expressions present in the first plurality of images;
• receiving second content depicting facial expressions associated with a performance of an actor captured during a second capture session different than the first capture session;
• comparing common features represented in content from the first and second capture sessions to determine one or more calibration values;
• aligning content from the first capture session with the second content from the second capture session using the one or more calibration values and combining the aligned content to produce combined session content;
• decomposing the combined session content;
• and
• producing an enhanced motion model by combining the decomposed second plurality of polygon meshes with the first motion model, wherein the enhanced motion model may be used to substantially reconstruct both the facial expressions present in the first plurality of images and the facial expressions captured during the second capture session.