High-quality object-space dynamic ambient occlusion
AssigneeElectronic Arts Inc.
InventorsBinh Huy Le, John Peter Lewis
Life of this patent
Estimating the term…
Abstract
Systems and methods are disclosed for calculating dynamic ambient occlusion (AO) values for character models to yield high-quality approximations of global illumination effects. The approach utilizes a dual component machine-learning model that factorizes dynamic AO computation into a non-linear component, in which visibility is determined by approximating spheres and their casted shadows, and a linear component that leverages a skinning-like algorithm for efficiency. The parameters of both components are trained in a regression against ground-truth AO values. The resulting model accommodates lighting interactions with external objects and can be generalized without requiring carefully constructed training data.
Claims20
A computer-implemented method comprising:
- during runtime execution of a video game application on a user computing device,
- applying a non-linear layer of a machine-learning model to a three-dimensional (3D) virtual character model in order to:
- generate an arrangement of elements representative of a pose of the 3D virtual character model;
- compute shadow values for elements in the arrangement, wherein individual shadow values are based at least in part on a shadow cast by at least one of the elements in the arrangement;
- applying a linear layer of the machine-learning model in order to:
- calculate ambient occlusion values for the 3D virtual character model based at least in part on the shadow values for the elements in the arrangement;
- and
- rendering a frame including at least a portion of the 3D virtual character model based at least in part on the calculated ambient occlusion values.
A computing system, comprising:
- one or more processing devices configured to execute instructions in order to:
- apply a non-linear layer of a machine-learning model to a three-dimensional (3D) virtual character model in order to:
- generate an arrangement of elements representative of a pose of the 3D virtual character model;
- compute shadow values for elements in the arrangement, wherein individual shadow values are based at least in part on a shadow cast by at least one of the elements in the arrangement;
- apply a linear layer of the machine-learning model in order to:
- calculate ambient occlusion values for the 3D virtual character model based at least in part on the shadow values for the elements in the arrangement;
- and
- render a frame including at least a portion of the 3D virtual character model based at least in part on the calculated ambient occlusion values.
Non-transitory computer readable storage media comprising computer-executable instructions that, when executed by a computing system, cause the computing system to:
- apply a non-linear layer of a machine-learning model to a three-dimensional (3D) virtual character model in order to:
- generate an arrangement of elements representative of a pose of the 3D virtual character model;
- compute shadow values for elements in the arrangement, wherein individual shadow values are based at least in part on a shadow cast by at least one of the elements in the arrangement;
- apply a linear layer of the machine-learning model in order to:
- calculate ambient occlusion values for the 3D virtual character model based at least in part on the shadow values for the elements in the arrangement;
- and
- render a frame including at least a portion of the 3D virtual character model based at least in part on the calculated ambient occlusion values.
Dependencies are inferred from the claim text.