Awesome

Thin-Plate Spline-based Interpolation for Animation Line Inbetweening

Tianyi Zhu, Wei Shang, Dongwei Ren<sup>*</sup>, Wangmeng Zuo

This repository is the official PyTorch implementation of "Thin-Plate Spline-based Interpolation for Animation Line Inbetweening".

✨ News/TODO

Examples on MixiamoLine240 of TPS-Inbetween.
Source code of inference demo and pretrained weight.
Source code of mannual matching demo.

🖼️ Resluts

<table class="center"> <tr style="font-weight: bolder;text-align:center;"> <td>Input starting frame</td> <td>Input ending frame</td> <td>Inbetweening results</td> </tr> <tr> <td> <img src=assets/input1_0.png width="250"> </td> <td> <img src=assets/input1_1.png width="250"> </td> <td> <img src=assets/ours1.gif width="250"> </td> </tr> <tr> <td> <img src=assets/input2_0.png width="250"> </td> <td> <img src=assets/input2_1.png width="250"> </td> <td> <img src=assets/ours2.gif width="250"> </td> </tr> <tr> <td> <img src=assets/input3_0.png width="250"> </td> <td> <img src=assets/input3_1.png width="250"> </td> <td> <img src=assets/ours3.gif width="250"> </td> </tr> </table>

📖 Overview

<p align="center"> <img src="assets/model.jpg" alt="model architecture" width="800"/> </br> An overview of the pipeline. </p>

⚙️ Run inference demo

Download the Gluestick weights and put them in './model/resources'.
Download the TPS-Inbetween pretrained weights here and then put it in the './ckpt' directory.
Run the following command to get inbetweening results.

python demo.py --image1 'assets/input1_0.png' --image2 'assets/input1_1.png' --xN 30 --save_path './output'

You can change 'xN' to get arbitrary frame rates results. The reuslts are saved in the folder './output'.