Action! (1998)
This system represents about five years of R&D effort on my part. Action! is an integrated system for creating and delivering multi-participant dramatic/cinematic experiences.
The system provides a set of tools for building virtual dramatic worlds, which are presented to human players by
artificial director agents. The director agents use a novel artificial intelligence (AI) architecture to create the
graphic and narrative representations seen by human players. Players control virtual actors through
a high-level interface that can be used to make the actors perform complex physical actions, speak, or alter their
facial expressions. The director agents encode dramatic and cinematic rules, which can be modified
by a human author to convey certain artistic qualities specific to a given experience.
Director agents carry out two novel processes: cinematic and literary rendering.
In the case of cinematic rendering, agents create a series of comic book-like
panels by controlling parameters such as the position/angle of a virtual 3D camera,
the size/shape of panels, and the placement of panels on the page. With literary
rendering, agents generate English language descriptions of dramatic events and
the inner states of virtual actors (i.e., character inner-voice.) In both processes,
simulated emotional models of the virtual actors influence the choices of the director
agent. Thus, if an actor is frightened, the director agent will attempt to create images
and sentences that dramatically convey a sense of fear. Below are some screen captures
from a dramatic scenario, Dry Bones, created with the system.
