Fast Personalized Text-to-Image Syntheses With Attention Injection

TL;DR

This work proposes an effective and fast approach that could balance the text-image consistency and identity consistency of the generated image and reference image while maintaining the inherent text-to-image generation ability of diffusion models.

Abstract

Currently, personalized image generation methods mostly require considerable time to finetune and often overfit the concept resulting in generated images that are similar to custom concepts but difficult to edit by prompts. We propose an effective and fast approach that could balance the text-image consistency and identity consistency of the generated image and reference image. Our method can generate personalized images without any fine-tuning while maintaining the inherent text-to-image generation ability of diffusion models. Given a prompt and a reference image, we merge the custom concept into generated images by manipulating cross-attention and self-attention layers of the original diffusion model to generate personalized images that match the text description. Comprehensive experiments highlight the superiority of our method.

Figures (6)

• Figure 1: The method we proposed is a fast personalized image generation approach, which does not require any fine-tuning or optimization and only needs one image for inference. It has shown better results in terms of text-image consistency and generation quality than other methods while maintaining identity consistency.
• Figure 2: Overall schematics of our method.
• Figure 3: The visualization of the cross-attention map after normalization.
• Figure 4: Our method results on different domains.
• Figure 5: The visual comparison with other methods.