The paper studies GAN training from a short video clip. Reviewers appreciated the challenges of the task, the fact that it hasn't been attempted in this form and also the combination of VAE and GAN. Reviewers were concerned about the use cases, realism, simplicity of the motion and length of the videos. These concerns remained after reading rebuttal. Three reviewers valued novelty of the task more strongly and AC concurs. This will be a baseline for future improvements.