A preprint tests an object-centric model on KITTI-MOT, reporting top reconstruction scores but mixed results when the future camera is predicted.