A group of MIT researchers have developed a technique that will allow artists better control over their 3D animations. The method uses mathematical functions known as barycentric coordinates, allowing 3D shapes to be manipulated. This offers more flexibility than traditional animation methods, which require starting from scratch for every change in animation. The developed method also has the potential for use in fields like medical imaging, virtual reality, and even helping robots understand how objects move in reality.
The method involves modifying the ‘cage’ around the shape, giving more control on how the shape changes when the cage is moved. However, the traditional calculations inherent in this approach often limit artists’ control over the smoothness of the figure. To expand on this, the researchers have developed a technique that incorporates artists’ ideal vision of the ‘smoothness’ and the final look of the figure.
Applying the concept of barycentric coordinates, initially introduced in 1827, the researchers used the model of how the corners of a shape interact and impact the shape’s interior. By applying a type of neural network to mimic these barycentric coordinate functions, the team simplified the typically complex, non-triangular cage calculations. The network offers solutions that are always appropriate, reducing the constraints and making it easier for artists to manipulate the figures without having to consider the mathematical aspects.
Building on the simple barycentric coordinates of triangles, the researchers developed overlapping ‘virtual triangles’ that joined the points of a complex figure, effectively making the neural network predict the combination of virtual triangles to create smooth transitions in animations. The fluidity offered by the neural network in calculating these transitions have potential to make animations appear much more natural.
The researchers will continue to develop this technique and aim to build it into a real-time, interactive interface that would allow artists to tweak animations with less effort. This research was backed by a number of organisations, including the U.S. Army Research Office, U.S. Air Force Office of Scientific Research, and the U.S. National Science Foundation.
