Today, we are excited to introduce Leffa, a cutting-edge project available on Hugging Face, which offers a unified framework for generating controllable character images. Leffa allows users to precisely manipulate the appearance (such as virtual try-on) and pose (such as pose transfer) of generated characters. The demo showcases impressive results, truly redefining the boundaries of what is possible with image generation.
The team behind Leffa has shared their research paper on [ArXiv](https://arxiv.org/pdf/2412.08486), while the project's codebase is available on [GitHub](https://github.com/franciszzj/Leffa). For a live demo, visit the [Hugging Face space](https://huggingface.co/spaces/franciszzj/Leffa).