We introduce kubric, an open-source python framework that interfaces with pybullet and blender to generate photo-realistic scenes, with rich annotations, and seamlessly scales to large jobs distributed overthousands of machines, and generating terabyte (tbs) of data.
We demonstrate the effectiveness of kubric by presenting a series of 13 different generated data sets for tasks ranging from studying 3d nerf models to optical flowestimation.
We release kubric, the used assets, all of the generation code, aswell as the rendered datasets for reuse and modification.
Authors
Klaus Greff, Francois Belletti, Lucas Beyer, Carl Doersch, Yilun Du, Daniel Duckworth, David J. Fleet, Dan Gnanapragasam, Florian Golemo, Charles Herrmann, Thomas Kipf, Abhijit Kundu, Dmitry Lagun, Issam Laradji, Hsueh-Ti, Henning Meyer, Yishu Miao, Derek Nowrouzezahrai, Cengiz Oztireli, Etienne Pot