ID-Aligner: Enhancing Identity-Preserving Text-to-Image Generation with Reward Feedback Learning

TL;DR

ID-Aligner is a novel framework that improves identity preservation and aesthetic quality in text-to-image generation by using reward feedback learning, compatible with various model adaptation techniques.

Contribution

It introduces a universal feedback fine-tuning framework that enhances identity retention and aesthetic appeal in ID-T2I models, compatible with LoRA and Adapter methods.

Findings

Significant improvement in identity preservation on SD1.5 and SDXL models.

Enhanced aesthetic quality with human-annotated preference feedback.

Effective integration of face recognition feedback for identity consistency.

Abstract

The rapid development of diffusion models has triggered diverse applications. Identity-preserving text-to-image generation (ID-T2I) particularly has received significant attention due to its wide range of application scenarios like AI portrait and advertising. While existing ID-T2I methods have demonstrated impressive results, several key challenges remain: (1) It is hard to maintain the identity characteristics of reference portraits accurately, (2) The generated images lack aesthetic appeal especially while enforcing identity retention, and (3) There is a limitation that cannot be compatible with LoRA-based and Adapter-based methods simultaneously. To address these issues, we present \textbf{ID-Aligner}, a general feedback learning framework to enhance ID-T2I performance. To resolve identity features lost, we introduce identity consistency reward fine-tuning to utilize the feedback from face detection and recognition models to improve generated identity preservation. Furthermore, we propose identity aesthetic reward fine-tuning leveraging rewards from human-annotated preference data and automatically constructed feedback on character structure generation to provide aesthetic tuning signals. Thanks to its universal feedback fine-tuning framework, our method can be readily applied to both LoRA and Adapter models, achieving consistent performance gains. Extensive experiments on SD1.5 and SDXL diffusion models validate the effectiveness of our approach. \textbf{Project Page: \url{https://idaligner.github.io/}}