TiPGAN: High-Quality Tileable Textures Synthesis with Intrinsic Priors for Cloth Digitization Applications

This is the official implementation of TiPGAN (CAD 2025). You can watch our demo video by clicking the below teaser image: If you find our repository helpful, please cite using this bib.

Environment

We tested our environment on CUDA 11.8. This repository should not have specific requirements on CUDA or PyTorch version. You may should make sure the CUDA version and the PyTorch version are compatible.

conda create -n tipgan python=3.10
conda activate tipgan
pip install torch==2.4.1 torchvision==0.18.0 torchaudio==2.3.0 --index-url https://download.pytorch.org/whl/cu118

Install the requirements.txt:

pip install -r requirements.txt

Train

Before training your texture, you should download the VGG checkpoints from our Hugging Face repository. You can do this by running the following command:

huggingface-cli download taited/TiPGAN --local-dir ckpt

Next, specify the image path and start training by running:

python train_tipgan.py --img_path media/nature_0005.jpg

Inference

You can generate the seamless texture and its tiled result by running the following command:

python inference.py --img_path media/nature_0005.jpg --ckpt_path ckpt/nature_0005.pth

We have uploaded some of the checkpoints for the synthesized seamless textures to Hugging Face. The table below shows the generated seamless textures.

	flora_0003	honeycombed_0123	nature_0005
Input
Seamless Texture
Tilled Texture

	chequered_0199	dotted_0190	zigzagged_0036
Input
Seamless Texture
Tilled Texture

	banded_0109	banded_0141	banded_0145
Input
Seamless Texture
Tilled Texture

Citation

If this repository is helpful to your research, please cite it as below.

@article{HE2025103866,
title = {TiPGAN: High-quality tileable textures synthesis with intrinsic priors for cloth digitization applications},
journal = {Computer-Aided Design},
volume = {183},
pages = {103866},
year = {2025},
issn = {0010-4485},
doi = {https://doi.org/10.1016/j.cad.2025.103866},
url = {https://www.sciencedirect.com/science/article/pii/S0010448525000284},
author = {Honghong He and Zhengwentai Sun and Jintu Fan and P.Y. Mok},
}