TeleStyle: Content-Preserving Style Transfer in Images and Videos

Shiwen Zhang, Xiaoyan Yang, Bojia Zi, Haibin Huang, Chi Zhang, Xuelong Li
Institute of Artificial Intelligence, China Telecom (TeleAI)

[Project Page] [arXiv] [Hugging Face] [GitHub]

Abstract

Content-preserving style transfer—generating stylized outputs based on content and style references—remains a significant challenge for Diffusion Transformers (DiTs) due to the inherent entanglement of content and style features in their internal representations. In this technical report, we present TeleStyle, a lightweight yet effective model for both image and video stylization. Built upon Qwen-Image-Edit, TeleStyle leverages the base model’s robust capabilities in content preservation and style customization. To facilitate effective training, we curated a high-quality dataset of distinct specific styles and further synthesized triplets using thousands of diverse, in-the-wild style categories. We introduce a Curriculum Continual Learning framework to train TeleStyle on this hybrid dataset of clean (curated) and noisy (synthetic) triplets. This approach enables the model to generalize to unseen styles without compromising precise content fidelity. Additionally, we introduce a video-to-video stylization module to enhance temporal consistency and visual quality. TeleStyle achieves state-of-the-art performance across three core evaluation metrics: style similarity, content consistency, and aesthetic quality.

Latest News

• Jan 28, 2026: We release the technical report , code and model of TeleStyle.

Todo List

• Release inference code
• Release models
• Release technical report

How to use

1. Installation

pip install -r requirements.txt

This environment is tested with:

• Python 3.11
• PyTorch 2.4.1 + CUDA 12.1
• diffusers 0.36.0
• transformers 4.49.0

2. Download Checkpoint

Download the TeleStyle checkpoint to a local path for example weights/:

We provide Image and Video checkpoint:

• Image (reference style image + content image -> stylized image)
  diffsynth_Qwen-Image-Edit-2509-Lightning-4steps-V1.0-bf16.safetensors; diffsynth_Qwen-Image-Edit-2509-telestyle.safetensors 37

• Video (stylized first frame + content video -> stylized video)
  dit.ckpt; prompt_embeds.pth

3. Inference

We provide inference scripts for running TeleStyle on demo inputs for each task:

Image Stylization

python telestyleimage_inference.py --image_path assets/example/0.png --style_path videos/1.png --output_path results/image.png

Video Stylization

python telestylevideo_inference.py --video_path assets/example/1.mp4 --style_path assets/example/1-0.png --output_path results/video.mp4

Citation

If you find TeleStyle useful in your research, please kindly cite our paper:

@article{teleai2026telestyle,
    title={TeleStyle: Content-Preserving Style Transfer in Images and Videos}, 
    author={Shiwen Zhang and Xiaoyan Yang and Bojia Zi and Haibin Huang and Chi Zhang and Xuelong Li},
    journal={arXiv preprint arXiv:2601.20175},
    year={2026}
}