2512.03964

Total: 1

#1 Training for Identity, Inference for Controllability: A Unified Approach to Tuning-Free Face Personalization [PDF2] [Copy] [Kimi] [REL]

Authors: Lianyu Pang, Ji Zhou, Qiping Wang, Baoquan Zhao, Zhenguo Yang, Qing Li, Xudong Mao

Tuning-free face personalization methods have developed along two distinct paradigms: text embedding approaches that map facial features into the text embedding space, and adapter-based methods that inject features through auxiliary cross-attention layers. While both paradigms have shown promise, existing methods struggle to simultaneously achieve high identity fidelity and flexible text controllability. We introduce UniID, a unified tuning-free framework that synergistically integrates both paradigms. Our key insight is that when merging these approaches, they should mutually reinforce only identity-relevant information while preserving the original diffusion prior for non-identity attributes. We realize this through a principled training-inference strategy: during training, we employ an identity-focused learning scheme that guides both branches to capture identity features exclusively; at inference, we introduce a normalized rescaling mechanism that recovers the text controllability of the base diffusion model while enabling complementary identity signals to enhance each other. This principled design enables UniID to achieve high-fidelity face personalization with flexible text controllability. Extensive experiments against six state-of-the-art methods demonstrate that UniID achieves superior performance in both identity preservation and text controllability. Code will be available at https://github.com/lyuPang/UniID

Subject: Computer Vision and Pattern Recognition

Publish: 2025-12-03 16:57:50 UTC