Steering Guidance for Personalized Text-to-Image Diffusion Models

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Park, Sunghyun, Choi, Seokeon, Park, Hyoungwoo, Yun, Sungrack
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866915421525901312
author Park, Sunghyun
Choi, Seokeon
Park, Hyoungwoo
Yun, Sungrack
author_facet Park, Sunghyun
Choi, Seokeon
Park, Hyoungwoo
Yun, Sungrack
contents Personalizing text-to-image diffusion models is crucial for adapting the pre-trained models to specific target concepts, enabling diverse image generation. However, fine-tuning with few images introduces an inherent trade-off between aligning with the target distribution (e.g., subject fidelity) and preserving the broad knowledge of the original model (e.g., text editability). Existing sampling guidance methods, such as classifier-free guidance (CFG) and autoguidance (AG), fail to effectively guide the output toward well-balanced space: CFG restricts the adaptation to the target distribution, while AG compromises text alignment. To address these limitations, we propose personalization guidance, a simple yet effective method leveraging an unlearned weak model conditioned on a null text prompt. Moreover, our method dynamically controls the extent of unlearning in a weak model through weight interpolation between pre-trained and fine-tuned models during inference. Unlike existing guidance methods, which depend solely on guidance scales, our method explicitly steers the outputs toward a balanced latent space without additional computational overhead. Experimental results demonstrate that our proposed guidance can improve text alignment and target distribution fidelity, integrating seamlessly with various fine-tuning strategies.
format Preprint
id arxiv_https___arxiv_org_abs_2508_00319
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Steering Guidance for Personalized Text-to-Image Diffusion Models
Park, Sunghyun
Choi, Seokeon
Park, Hyoungwoo
Yun, Sungrack
Computer Vision and Pattern Recognition
Machine Learning
Personalizing text-to-image diffusion models is crucial for adapting the pre-trained models to specific target concepts, enabling diverse image generation. However, fine-tuning with few images introduces an inherent trade-off between aligning with the target distribution (e.g., subject fidelity) and preserving the broad knowledge of the original model (e.g., text editability). Existing sampling guidance methods, such as classifier-free guidance (CFG) and autoguidance (AG), fail to effectively guide the output toward well-balanced space: CFG restricts the adaptation to the target distribution, while AG compromises text alignment. To address these limitations, we propose personalization guidance, a simple yet effective method leveraging an unlearned weak model conditioned on a null text prompt. Moreover, our method dynamically controls the extent of unlearning in a weak model through weight interpolation between pre-trained and fine-tuned models during inference. Unlike existing guidance methods, which depend solely on guidance scales, our method explicitly steers the outputs toward a balanced latent space without additional computational overhead. Experimental results demonstrate that our proposed guidance can improve text alignment and target distribution fidelity, integrating seamlessly with various fine-tuning strategies.
title Steering Guidance for Personalized Text-to-Image Diffusion Models
topic Computer Vision and Pattern Recognition
Machine Learning
url https://arxiv.org/abs/2508.00319