MIT scientists have introduced a revolutionary system, “Home3D,” designed to enable users to enjoy 3D movies at home without the need for cumbersome special glasses. Although 3D movies have remained popular in theaters, the transition to home entertainment has been hindered by the inconvenience of requiring viewers to wear special glasses. Theatres typically utilize special polarized light or project paired images to create a sense of depth, but glasses have proven inconvenient for home 3D TVs.
MIT researchers aim to overcome this limitation with “Home3D,” a system that converts traditional stereo 3D movies into a format compatible with “automultiscopic displays.” These displays, showing promise for home theater systems, are continually improving in resolution. Petr Kellnhofer from MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) explained, “Automultiscopic displays are not as popular as they could be because they cannot play the stereo formats that traditional 3D movies use in theaters. By converting existing 3D movies to this format, our system helps open the door to bringing 3-D TVs into people’s homes.”
Home3D operates in real-time on a graphics-processing unit (GPU), making it compatible with systems like Xbox or PlayStation. The researchers envision Home3D potentially evolving into a chip that could be integrated into TVs or media players such as Google’s Chromecast.
The system’s algorithms also provide users with the flexibility to customize their viewing experience, adjusting the desired level of 3D for each movie. In a user study featuring clips from movies like “The Avengers” and “Big Buck Bunny,” participants rated Home3D videos as higher quality 60% of the time compared to 3D videos converted using alternative methods. Home3D converts 3D movies from “stereoscopic” to “multi-view” video, displaying three or more images that simulate the scene from different perspectives. This approach allows each eye to perceive the scene from a specific location, enabling the brain to naturally compute depth in the image.