FonTS: Text Rendering with Typography and Style Controls [ICCV 2025]

Wenda Shi¹, Yiren Song², Dengming Zhang³, Jiaming Liu⁴, Xingxing Zou^1,*

¹The Hong Kong Polytechnic University, ²National University of Singapore, ³Zhejiang University, ⁴Tiamat AI ^*Corresponding author

This repository is the official implementation of FonTS, a two-stage DiT-based pipeline to achieve word-level typographic control, font consistency, and artistic style consistency in text rendering tasks.

Environment

# Create conda environment
conda create -n fonts python=3.12

# Activate environment
conda activate fonts

# Install dependencies
pip install -r requirements.txt

or：

# Create environment from yml file
conda env create -f environment.yml

# Activate environment
conda activate fonts

Inference

python /path/to/FonTS/flux+SCA-only/infer_flux+SCA-only.py

Evaluation

Benchmark download: 🤗 ATR-Bench

Dataset

Trainset download: 🤗 SC-artext

Serial Work

WordCon: Word-level Typography Control in Scene Text Rendering

Citation

If you find this work helpful, please consider citing our paper or give a star🌟:

@InProceedings{Shi_2025_ICCV,
    author    = {Shi, Wenda and Song, Yiren and Zhang, Dengming and Liu, Jiaming and Zou, Xingxing},
    title     = {FonTS: Text Rendering With Typography and Style Controls},
    booktitle = {Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)},
    month     = {October},
    year      = {2025},
    pages     = {18463-18474}
}

Acknowledgments

This implementation is built based on xflux, IP-Adapter, Flux, diffusers.

This work was substantially supported by a grant from the Research Grants Council of the Hong Kong Special Administrative Region, China (Project No. PolyU/RGC Project PolyU 25211424) and partially supported by a grant from PolyU university start-up fund (Project No. P0047675).