PaintBoard: Prototyping Interactive Character Behaviours by Digitally Painting Storyboards

The creation of interactive worlds, such as those in video games, often include a set of computer controlled characters that must intelligently act and react in response to dynamic input from the user. These interactive behaviours usually require authors to programmatically define each behavior, reaction, and interaction the character should take in response to user input across a range of scenarios, a process that can take significant time. While this method can successfully create robust characters, the large development overhead is not conducive to the exploration, prototyping, and testing of new character ideas. We designed and developed PaintBoard, a system that enables users to rapidly prototype interactive character movement by digitally painting a storyboard. PaintBoard promotes prototyping by facilitating quick, visual authoring, and by enabling immediate testing by allowing the user to interact in real-time with a behaviour generated from a painted storyboard. To generate the interactive behavior, we developed a novel algorithm that analyzes a painted storyboard and uses machine-learning to generalize the painted examples to new situations. Our algorithm uses