Researchers at MIT have developed a new method that allows animators to have more control over their creations. Using mathematical functions called barycentric coordinates, they can now better control how 2D and 3D shapes stretch, move, and bend. Unlike traditional methods that only offered limited options for animation, this new method provides animators a level of control that allows them to tailor movements according to the desired “look” of their animations.
This breakthrough was made possible by a shift in traditional thinking. Instead of focusing on solving artistic problems, the researchers sought to provide the artists with the flexibility desired to create a final product that closely aligns with their vision. Ana Dodik, lead author of the paper introducing this technique, emphasized that artists aren’t interested in the intricacies of the algorithmic processes but value the “look” and flexibility of their creations.
This technology has the potential for wide-ranging applications. It can be used in areas such as medical imaging, architecture, virtual reality, and enhance computer vision in robots, helping them understand how objects move in the real world.
The new method developed by these researchers diverges from traditional methods by using a type of neural network to model the complex, but flexible, barycentric coordinate functions. Unlike other applications of neural networks that mimic human thought processes, this utilization is for mathematical reasons.
This neural network has the unique ability to output barycentric coordinate functions that always satisfy all mathematical constraints, streamlining the complex processes involved, ensuring that artists would not need to concern themselves with the mathematical aspects of their animations but focus on their creative objectives.
The researchers addressed the complexities of modern cage animation by covering shapes with overlapping virtual triangles that connect triplets of points on the outside of the cage. They used their neural network to predict how to combine these virtual triangles’ barycentric coordinates, resulting in a more complicated but smooth function that can be adjusted until the desired animation is achieved.
This method resulted in more naturally flowing animations. As an example, a cat’s tail created using this method would curve smoothly instead of folding rigidly near the vertices of the cage.
The team aims to continue refining this method, working on strategies to speed up the neural network. They also plan to incorporate it into an interactive interface that will allow artists to tweak their animations in real time. Funding for the research came from multiple sources, including the U.S. Army Research Office, the U.S. Air Force Office of Scientific Research, the U.S. National Science Foundation, and others.
