arXiv (Cornell University) · 2016 · 38 citations · 0 references
ComputingAutonomous SystemsRobotics EngineeringCinematographyComputer VisionMotion PlanningCinematography AnalysisMotion CaptureCinematographic Motion PlanningComposition PrinciplesDrone CinematographyDrone Tracking ShotsReal-time Trajectory ControlReal-time Trajectory PlanningResearch With Human SubjectsSubjects
We present a system to capture video footage of human subjects in the real\nworld. Our system leverages a quadrotor camera to automatically capture\nwell-composed video of two subjects. Subjects are tracked in a large-scale\noutdoor environment using RTK GPS and IMU sensors. Then, given the tracked\nstate of our subjects, our system automatically computes static shots based on\nwell-established visual composition principles and canonical shots from\ncinematography literature. To transition between these static shots, we\ncalculate feasible, safe, and visually pleasing transitions using a novel\nreal-time trajectory planning algorithm. We evaluate the performance of our\ntracking system, and experimentally show that RTK GPS significantly outperforms\nconventional GPS in capturing a variety of canonical shots. Lastly, we\ndemonstrate our system guiding a consumer quadrotor camera autonomously\ncapturing footage of two subjects in a variety of use cases. This is the first\nend-to-end system that enables people to leverage the mobility of quadrotors,\nas well as the knowledge of expert filmmakers, to autonomously capture\nhigh-quality footage of people in the real world.\n