Noise Consistency Regularization for Improved Subject-Driven Image Synthesis

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Ni, Yao, Wen, Song, Koniusz, Piotr, Cherian, Anoop
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910993146183680
author Ni, Yao
Wen, Song
Koniusz, Piotr
Cherian, Anoop
author_facet Ni, Yao
Wen, Song
Koniusz, Piotr
Cherian, Anoop
contents Fine-tuning Stable Diffusion enables subject-driven image synthesis by adapting the model to generate images containing specific subjects. However, existing fine-tuning methods suffer from two key issues: underfitting, where the model fails to reliably capture subject identity, and overfitting, where it memorizes the subject image and reduces background diversity. To address these challenges, we propose two auxiliary consistency losses for diffusion fine-tuning. First, a prior consistency regularization loss ensures that the predicted diffusion noise for prior (non-subject) images remains consistent with that of the pretrained model, improving fidelity. Second, a subject consistency regularization loss enhances the fine-tuned model's robustness to multiplicative noise modulated latent code, helping to preserve subject identity while improving diversity. Our experimental results demonstrate that incorporating these losses into fine-tuning not only preserves subject identity but also enhances image diversity, outperforming DreamBooth in terms of CLIP scores, background variation, and overall visual quality.
format Preprint
id arxiv_https___arxiv_org_abs_2506_06483
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Noise Consistency Regularization for Improved Subject-Driven Image Synthesis
Ni, Yao
Wen, Song
Koniusz, Piotr
Cherian, Anoop
Graphics
Artificial Intelligence
Computer Vision and Pattern Recognition
Machine Learning
Image and Video Processing
Fine-tuning Stable Diffusion enables subject-driven image synthesis by adapting the model to generate images containing specific subjects. However, existing fine-tuning methods suffer from two key issues: underfitting, where the model fails to reliably capture subject identity, and overfitting, where it memorizes the subject image and reduces background diversity. To address these challenges, we propose two auxiliary consistency losses for diffusion fine-tuning. First, a prior consistency regularization loss ensures that the predicted diffusion noise for prior (non-subject) images remains consistent with that of the pretrained model, improving fidelity. Second, a subject consistency regularization loss enhances the fine-tuned model's robustness to multiplicative noise modulated latent code, helping to preserve subject identity while improving diversity. Our experimental results demonstrate that incorporating these losses into fine-tuning not only preserves subject identity but also enhances image diversity, outperforming DreamBooth in terms of CLIP scores, background variation, and overall visual quality.
title Noise Consistency Regularization for Improved Subject-Driven Image Synthesis
topic Graphics
Artificial Intelligence
Computer Vision and Pattern Recognition
Machine Learning
Image and Video Processing
url https://arxiv.org/abs/2506.06483