TryOn-Adapter: Efficient Fine-Grained Clothing Identity Adaptation for High-Fidelity Virtual Try-On

The core message of this work is to propose an effective and efficient framework, termed TryOn-Adapter, for virtual try-on that can maintain the identity of the given garment with low computational cost. The key innovations are: 1) decoupling clothing identity into fine-grained factors (style, texture, and structure) and tailoring lightweight modules to precisely control each factor; 2) introducing a training-free technique, T-RePaint, to further reinforce clothing identity preservation while maintaining realistic try-on effects during inference.