First Order Motion Model for Image Animation
Aliaksandr Siarohin, Stéphane Lathuilière, Sergey Tulyakov, Elisa Ricci and Nicu Sebe. In NeurIPS, 2019.

The original page: [Project Page]

We implement the Vox model for facial videos and images. The official weight are converted to jittor. Download the weight and put it in ./weights/

The Google Drive link: https://drive.google.com/drive/folders/1pMODD4H_pb_9-YAFDXkv8W59mRzqxW2_?usp=sharing

Reconstruct a video with single image :

python demo.py --source_image exp/img.jpg --driving_video exp/ori.mp4 --result_video exp/recon.mp4 --find_best_frame

Drive a single image with relative mode:

python demo.py --source_image exp/frame00043.jpg --driving_video exp/ori.mp4 --result_video exp/relative.mp4 --relative --find_best_frame

Drive a single image with absolute mode:

python demo.py --source_image exp/frame00043.jpg --driving_video exp/ori.mp4 --result_video exp/relative.mp4 --find_best_frame

This repository borrows partially from the original codes.