Generating video sequences from a single input image, extending it with motion and temporal dynamics.