MIT researchers have unveiled a groundbreaking technique that could revolutionize the way animated characters move in movies and video games. The method, based on mathematical functions known as barycentric coordinates, provides artists with unprecedented control over the motions of 2D and 3D shapes.
Traditionally, animators faced limitations with inflexible techniques that offered only a single option for barycentric coordinate functions for a given character. Ana Dodik, the lead author of the research paper, emphasized the importance of flexibility for artists, stating, "What artists care about is flexibility and the 'look' of their final product."
The newly introduced technique allows artists to choose or design smoothness energies for any shape, giving them the freedom to experiment with different looks. Dodik highlighted the significance of this approach, noting that it goes beyond artistic applications and could find utility in diverse fields such as medical imaging, architecture, virtual reality, and computer vision.
When animating characters, artists commonly use a cage – a set of points connected by lines or triangles – to manipulate and deform the character. The challenge lies in determining how the character moves when the cage is modified, a task typically governed by barycentric coordinate functions.
MIT researchers departed from traditional methods by employing a special type of neural network to model these coordinate functions. Unlike past approaches, their neural network knows how to output functions that satisfy constraints exactly, eliminating the need for artists to grapple with intricate mathematical details.
Barycentric coordinates, introduced by German mathematician August Möbius in 1827, have been a fundamental concept in geometry. However, adapting them to complex cages posed significant challenges. The MIT team addressed this by using triangular barycentric coordinates and virtual triangles, employing the neural network to combine these virtual triangles into smooth functions.
The researchers demonstrated the practical impact of their method by generating more natural-looking animations, such as a cat's tail that moves smoothly without rigid folds near the cage's vertices. Dodik emphasized the flexibility provided by neural networks, stating that it enables artists to iterate on animations in real time.
Looking ahead, the team aims to explore strategies to accelerate the neural network and integrate the method into an interactive interface for real-time animation iteration. The development marks a significant stride in empowering artists and enhancing the creative process in animation.
