Reviews: Visual Dynamics: Probabilistic Future Frame Synthesis via Cross Convolutional Networks

Overall the paper addresses an interesting problem and the formulation is also intriguing. However, lack of more through ablation studies makes it unclear which parts of the model are truly important. The baselines are also rather weak as only a very simple kNN approach is tested as the baseline. A more through evaluation would make the paper considerably stronger. Generally I actually liked the paper for its ideas and addressed problem, but would like to read authors rebuttal before finalizing the decision.