Skip to content
official code of CVPR'18 paper "learning to generate time-lapse videos using multi-stage dynamic generative adversarial networks"
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Type Name Latest commit message Commit time
Failed to load latest commit information.

Learning to Generate Time-lapse Videos Using Multi-stage Dynamic Generative Adversarial Networks

This is the official code of the CVPR 2018 PAPER.

CVPR 2018 PAPER | Project Page | Dataset


  1. Requirements:
    • download our time-lapse dataset
    • python2.7
    • pytorch 0.3.0 or 0.3.1
    • ffmpeg
  2. Testing:
    • download our pretrained models
    • run python --cuda --testf your_test_dataset_folder
  3. Sample outputs:
    • in ./sample_outputsthere are mp4 files which are generated on my machine.


author = {Xiong, Wei and Luo, Wenhan and Ma, Lin and Liu, Wei and Luo, Jiebo},
title = {Learning to Generate Time-Lapse Videos Using Multi-Stage Dynamic
Generative Adversarial Networks},
booktitle = {The IEEE Conference on Computer Vision and Pattern Recognition
month = {June},
year = {2018}
You can’t perform that action at this time.