Thin-Plate Spline-based Interpolation for Animation Line Inbetweening

August 25, 2024 · View on GitHub

Tianyi Zhu, Wei Shang, Dongwei Ren*, 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

Input starting frame Input ending frame Inbetweening results

📖 Overview

model architecture
An overview of the pipeline.

⚙️ Run inference demo

  1. Download the Gluestick weights and put them in './model/resources'.
  2. Download the TPS-Inbetween pretrained weights here and then put it in the './ckpt' directory.
  3. 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'.