DreamBooth: Fine Tuning Text-to-Image Diffusion Models for Subject-Driven Generation

TLDR

Large text‑to‑image diffusion models can generate high‑quality, diverse images from text prompts, yet they cannot faithfully reproduce specific subjects from a limited reference set. This work introduces a method to personalize diffusion models so they can generate novel images of a specific subject. By fine‑tuning a pretrained model on a handful of subject images and attaching a unique identifier, the authors employ a class‑specific prior preservation loss to embed the subject into the model’s latent space, allowing the identifier to generate diverse, photorealistic renditions across scenes, poses, views, and lighting. The method successfully performs subject recontextualization, text‑guided view synthesis, and artistic rendering while preserving the subject’s key features. Project page: https://dreambooth.github.io/.

Abstract

Large text-to-image models achieved a remarkable leap in the evolution of AI, enabling high-quality and diverse synthesis of images from a given text prompt. However, these models lack the ability to mimic the appearance of subjects in a given reference set and synthesize novel renditions of them in different contexts. In this work, we present a new approach for "personalization" of text-to-image diffusion models. Given as input just a few images of a subject, we fine-tune a pretrained text-to-image model such that it learns to bind a unique identifier with that specific subject. Once the subject is embedded in the output domain of the model, the unique identifier can be used to synthesize novel photorealistic images of the subject contextualized in different scenes. By leveraging the semantic prior embedded in the model with a new autogenous class-specific prior preservation loss, our technique enables synthesizing the subject in diverse scenes, poses, views and lighting conditions that do not appear in the reference images. We apply our technique to several previously-unassailable tasks, including subject recontextualization, text-guided view synthesis, and artistic rendering, all while preserving the subject's key features. We also provide a new dataset and evaluation protocol for this new task of subject-driven generation. Project page: https://dreambooth.github.io/