Training for Identity, Inference for Controllability: A Unified Approach to Tuning-Free Face Personalization

TL;DR

UniID is a unified, tuning-free framework that combines text embedding and adapter-based methods to achieve high-fidelity face personalization with flexible text controllability by mutually reinforcing identity features while preserving non-identity attributes.

Contribution

It introduces a novel unified approach that synergistically integrates two paradigms for face personalization, with a training-inference strategy ensuring identity fidelity and controllability.

Findings

Outperforms six state-of-the-art methods in identity preservation.

Achieves high-fidelity face personalization with flexible text control.

Demonstrates superior results in extensive experiments.

Abstract

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