Artist-directed volumetric dynamic virtual cameras

A method for authoring a virtual camera used by an application that exposes a virtual three-dimensional (3D) application space, comprising:

• receiving an input for selecting a portion of the virtual 3D space for which training data is created, the training data comprising player positions in the virtual 3D space and corresponding camera parameters;
• enabling creation of a training dataset for the selected portion of the virtual 3D space, the creation including
• (i) receiving an instruction for selecting a training player location within the selected portion of the virtual 3D space,
• (ii) enabling the virtual camera to be configured to capture a scene of the selected portion of the virtual 3D space that includes the training player location,
• (iii) capturing camera parameters corresponding to the training player position for the configured camera,
• (iv) adding the captured parameters to a training dataset, and
• (v) iterating steps (i) to (iv) until the training data for the selected portion of the virtual 3D space is completed;
• and
• providing the training dataset to a component that implements a radial basis function (RBF) algorithm that interpolates the training dataset to produce camera parameters for player positions in the virtual 3D space other than the training player position.

One or more computer-readable memories storing instructions which, when executed by one or more processors disposed in an electronic device, implement a camera system used in a virtual three-dimensional (3D) game space and performing a method comprising:

• receiving a training dataset that maps respective virtual player positions in a convex hull in the virtual 3D space to corresponding sets of camera parameters, the convex hull being fitted to sample points in a camera sample dataset;
• receiving a current position of a virtual player in the convex hull during game runtime;
• interpolating the data in the training dataset to generate interpolated camera parameters for the current virtual player position;
• receiving a current position of the virtual player in the virtual 3D space that is external to the convex hull;
• extrapolating the data in the training dataset to generate extrapolated camera parameters for the current virtual player position;
• sending either the interpolated camera parameters or the extrapolated camera parameters to a camera system so that the camera system can capture scenes in the virtual 3D game space in accordance with either the interpolated camera parameters or the extrapolated camera parameters.