The animation pipeline typically used in video games and feature films requires creation of detailed 3D characters. Each character must be built from hand-designed skeletons and meticulously tailored skin and fabrics, all of which then gets animated frame by frame. Each stage in this sequence is time-consuming and requires expertise in complicated animation tools, making the process inaccessible to novices. Furthermore, this sequence is linear. A model cannot be edited once it's been rigged and animated, impeding iterative design. Motivated by this observation, we created AniMesh, a system that unifies modeling and animation.
Using AniMesh, anyone, experts and novices alike, can build 3D characters in minutes (not hours), by simply drawing sketches. Then, the user can bring her creation to life by demonstrating motions using her body in front of a commodity motion tracking camera. Unlike current industry standard tools, however, should the user decide to change the shape, structure or position of her character, AniMesh will automatically adjust the animation to the new shape of the character.
Our design goal, unlike most existing animation suites, is to support tasks traditionally poorly served by modern tools, like rapid prototyping, being accessible to non-expert users, facilitating exploration. In these scenarios a non-linear and non-destructive workflow can offer advantages over more traditional techniques. With this in mind, every component and algorithm in AniMesh was designed to allow a single user to interleave these tasks and significantly decrease their iteration time, and, ideally, increase their prototyping ability.