Virtual Cinematographer

As the virtual reality (VR) industry continues to push the envelope on creating unique immersive content, the question of how to effectively capture VR experiences for an audience remains a challenge. The dominant method of showcasing VR applications through first-person point of view video is often lacking in cinematic interest: failing to capture other viewpoints and the essence of the moment. Meanwhile, manually setting up cameras and splicing video together requires planning and expertise on behalf of the user. 

In this paper, we propose the use of artificial intelligence (AI) to automatically identify and select the most compelling predefined viewpoint in a VR environment at a given moment. Our system switches between multiple cameras in real time, capturing crucial moments in a scene through analysis of user interactions and spatial cues. Utilizing the new dimension of information we receive in the virtual space, we create a technique that enhances the communication of immersive experiences in a user-friendly way.