SAIN

The repo is the code implementation for paper "Bridging the Gap: Sketch-Aware Interpolation Network for High-Quality Animation Sketch Inbetweening".

Paper

Training

Download the dataset and use the following scripts.

python main.py --data_root /path/to/std12k_points --batch_size 4 --test_batch_size 4 --loss 0.7*L1+0.3*LPIPS

Dataset

Our dataset can be downloaded in the following link. google drive

The pre-trained model can be downloaded in the following link google drive

The training & test dataset containing region & stroke level correspondence data can be downloaded in the following link. google drive

Reference

Some great video interpolation resources that we benefit from:

• AnimeInterp Code
• VFI-Transformer Code
• EISAI Code

Citation

@inproceedings{shen2023sain,
    title={Bridging the Gap: Sketch-Aware Interpolation Network for High-Quality Animation Sketch Inbetweening},
    author={Jiaming Shen, Kun Hu, Wei Bao, Chang Wen Chen, and Zhiyong Wang},
    Booktitle = {Proc. of ACM International Conference on Multimedia (MM’24)},
    year={2024}
}