TextBoost: Towards One-Shot Personalization of Text-to-Image Models via Fine-tuning Text Encoder
This addresses the challenge of creating controllable personalized images from a single reference for users of text-to-image models, though it is incremental as it builds on existing fine-tuning methods.
The paper tackles the problem of one-shot personalization in text-to-image models, which often overfit to a single reference image, by proposing a selective fine-tuning strategy for the text encoder with techniques like augmentation tokens and a knowledge-preservation loss, resulting in efficient generation of high-quality, diverse images with reduced memory and storage requirements.
Recent breakthroughs in text-to-image models have opened up promising research avenues in personalized image generation, enabling users to create diverse images of a specific subject using natural language prompts. However, existing methods often suffer from performance degradation when given only a single reference image. They tend to overfit the input, producing highly similar outputs regardless of the text prompt. This paper addresses the challenge of one-shot personalization by mitigating overfitting, enabling the creation of controllable images through text prompts. Specifically, we propose a selective fine-tuning strategy that focuses on the text encoder. Furthermore, we introduce three key techniques to enhance personalization performance: (1) augmentation tokens to encourage feature disentanglement and alleviate overfitting, (2) a knowledge-preservation loss to reduce language drift and promote generalizability across diverse prompts, and (3) SNR-weighted sampling for efficient training. Extensive experiments demonstrate that our approach efficiently generates high-quality, diverse images using only a single reference image while significantly reducing memory and storage requirements.