Personalized Text-to-Image Generation with Auto-Regressive Models

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Sun, Kaiyue, Liu, Xian, Teng, Yao, Liu, Xihui
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912333122502656
author Sun, Kaiyue
Liu, Xian
Teng, Yao
Liu, Xihui
author_facet Sun, Kaiyue
Liu, Xian
Teng, Yao
Liu, Xihui
contents Personalized image synthesis has emerged as a pivotal application in text-to-image generation, enabling the creation of images featuring specific subjects in diverse contexts. While diffusion models have dominated this domain, auto-regressive models, with their unified architecture for text and image modeling, remain underexplored for personalized image generation. This paper investigates the potential of optimizing auto-regressive models for personalized image synthesis, leveraging their inherent multimodal capabilities to perform this task. We propose a two-stage training strategy that combines optimization of text embeddings and fine-tuning of transformer layers. Our experiments on the auto-regressive model demonstrate that this method achieves comparable subject fidelity and prompt following to the leading diffusion-based personalization methods. The results highlight the effectiveness of auto-regressive models in personalized image generation, offering a new direction for future research in this area.
format Preprint
id arxiv_https___arxiv_org_abs_2504_13162
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Personalized Text-to-Image Generation with Auto-Regressive Models
Sun, Kaiyue
Liu, Xian
Teng, Yao
Liu, Xihui
Computer Vision and Pattern Recognition
Personalized image synthesis has emerged as a pivotal application in text-to-image generation, enabling the creation of images featuring specific subjects in diverse contexts. While diffusion models have dominated this domain, auto-regressive models, with their unified architecture for text and image modeling, remain underexplored for personalized image generation. This paper investigates the potential of optimizing auto-regressive models for personalized image synthesis, leveraging their inherent multimodal capabilities to perform this task. We propose a two-stage training strategy that combines optimization of text embeddings and fine-tuning of transformer layers. Our experiments on the auto-regressive model demonstrate that this method achieves comparable subject fidelity and prompt following to the leading diffusion-based personalization methods. The results highlight the effectiveness of auto-regressive models in personalized image generation, offering a new direction for future research in this area.
title Personalized Text-to-Image Generation with Auto-Regressive Models
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2504.13162