*A TensorFlow implementation of the models described in [Unsupervised Learning for Physical Interaction through Video Prediction (Finn et al., 2016)](https://arxiv.org/abs/1605.07157).*
*A TensorFlow implementation of the models described in [Unsupervised Learning for Physical Interaction through Video Prediction (Finn et al., 2016)](https://arxiv.org/abs/1605.07157).*
This video prediction model, which is optionally conditioned on actions,
This video prediction model, which is optionally conditioned on actions,
predictions future video by internally predicting how to transform the last
predicts future video by internally predicting how to transform the last
image (which may have been predicted) into the next image. As a result, it can
image (which may have been predicted) into the next image. As a result, it can
reuse apperance information from previous frames and can better generalize to
reuse apperance information from previous frames and can better generalize to
objects not seen in the training set. Some example predictions on novel objects
objects not seen in the training set. Some example predictions on novel objects
