🎨 TeleStyle V2 Release — Open-Source Style Transfer Model
The TeleAI team has released the second version, TeleStyle V2, for style transfer in images and videos. The model utilizes self-distillation and Distributional Matching (DMD), allowing it to work with combinations of realistic and stylized content. QIE2509 LoRA models are already available for the diffusers and diffsynth libraries.
🌍 The emergence of a high-quality open-source solution reduces developers' dependence on closed proprietary APIs and expands the possibilities for creating content while preserving the original structure.
👤 You can use the new LoRA models in your pipelines for professional media stylization, achieving results comparable to top commercial neural networks.
Source 1: https://github.com/Tele-AI/TeleStyleV2 Source 2: https://huggingface.co/Tele-AI/TeleStyleV2/tree/main