US 2019/0156548 A1 · REPURPOSING EXISTING ANIMATED CONTENT

Abstract

Systems and methods for automatically animating a character based on an existing corpus of animation are described. The character may be from a previously produced feature animated film, and the data used for training may be the data used to animate the character in the film. A low-dimensional embedding for subsets of the existing animation corresponding to different semantic labels may be learned by mapping high-dimensional rig control parameters to a latent space. A particle model may be used to move within the latent space, thereby generating novel animations corresponding to the space's semantic label, such as a pose. Bridges may link a first pose of a first model within the latent space that is similar to a second pose of a second model of the space. Animations corresponding to transitions between semantic labels may be generated by creating animation paths that traverse a bridge from one model into another.

Claims20

Dependent claims

A computer-implemented method for rendering new animated content of a character based on existing animated content of the character, the method comprising:

accessing a latent space that comprises a first semantic model, a second semantic model, and a bridge, wherein:
the first semantic model corresponds to a first plurality of frames of the existing animated content of the character and is generated based on a first set of model parameters,
the second semantic model corresponds to a second plurality of frames of the existing animated content of the character and is generated based on a second set of model parameters, and
the bridge links the first semantic model and the second semantic model;
and
rendering the character with a pose that is based on a location of a particle within the latent space, wherein the location of the particle within the latent space is controlled via user input, and wherein rendering the character with the pose that is based on the location of the particle within the latent space comprises:
determining a new model parameter that corresponds to the location of the particle within the latent space;
determining a new rig parameter that corresponds to the new model parameter;
and
rendering a new frame of animation associated with the determined new rig parameter.

7 more elements

A non-transitory computer-readable storage medium storing one or more programs for rendering new animated content of a character based on existing animated content of the character, the one or more programs configured to be executed by one or more processors and including instructions for:

accessing a latent space that comprises a first semantic model, a second semantic model, and a bridge, wherein:
the first semantic model corresponds to a first plurality of frames of the existing animated content of the character and is generated based on a first set of model parameters,
the second semantic model corresponds to a second plurality of frames of the existing animated content of the character and is generated based on a second set of model parameters, and
the bridge links the first semantic model and the second semantic model;
and
rendering the character with a pose that is based on a location of a particle within the latent space, wherein the location of the particle within the latent space is controlled via user input, and wherein rendering the character with the pose that is based on the location of the particle within the latent space comprises:
determining a new model parameter that corresponds to the location of the particle within the latent space;
determining a new rig parameter that corresponds to the new model parameter;
and
rendering a new frame of animation associated with the determined new rig parameter.

7 more elements

An apparatus for rendering new animated content of a character based on existing animated content of the character, the apparatus comprising:

one or more processors;
and
memory storing one or more programs configured to be executed by the one or more processors, the one or more programs including instructions for:
accessing a latent space that comprises a first semantic model, a second semantic model, and a bridge, wherein:
the first semantic model corresponds to a first plurality of frames of the existing animated content of the character and is generated based on a first set of model parameters,
the second semantic model corresponds to a second plurality of frames of the existing animated content of the character and is generated based on a second set of model parameters, and
the bridge links the first semantic model and the second semantic model;
and
rendering the character with a pose that is based on a location of a particle within the latent space, wherein the location of the particle within the latent space is controlled via user input, and wherein rendering the character with the pose that is based on the location of the particle within the latent space comprises:
determining a new model parameter that corresponds to the location of the particle within the latent space;
determining a new rig parameter that corresponds to the new model parameter;
and
rendering a new frame of animation associated with the determined new rig parameter.

10 more elements

Dependencies are inferred from the claim text.

Repurposing existing animated content

Life of this patent

Repurposing existing animated content

Life of this patent

Abstract

Claims20