Software animations create both realistic and fanciful characters, but their makeup and capabilities need not match those that are possible in the real world. Harvard's software, however, translates the primary characteristics of the on-screen characters into articulated components that together realize a figurine that can be created by a 3-D printer.
By observing the on-screen appearance and actions performed by the character, the Harvard algorithm determines the ideal locations for the character's joints—either ball-in-socket or hinged—then optimizes their size and location using the physics of the real 3-D world. Once the reverse rendering operation is complete, a detailed file is sent to a 3-D printer, which creates a completely assembled version of the action figure.
The Harvard researchers expect their invention to be useful not only for personalized action figures for consumers, but also for professional animators who today create mannequins by hand. With the reverse rendering algorithm, animators will be able to quickly create action figures which they can use to experiment with different stances and motions in real world recreations of virtual worlds.
Harvard’s Office of Technology Development has filed a patent which it aims to license to a cloud-based service that will used 3-D printers to create customized, user-generated figurines from existing animation software.
Funding for the product was supplied by the National Science Foundation, Pixar and the John Simon Guggenheim Memorial Foundation.