BLIP-Diffusion: Pre-trained Subject Representation for Controllable Text-to-Image Generation and Editing

TL;DR

BLIP-Diffusion is a novel subject-driven text-to-image generation model that leverages a pre-trained multimodal encoder for efficient, zero-shot, and customizable subject rendering and editing, overcoming limitations of existing models.

Contribution

It introduces a pre-trained multimodal encoder and a subject representation learning task, enabling efficient, zero-shot, and flexible subject-driven image generation and editing.

Findings

Supports zero-shot subject-driven generation

Achieves up to 20x faster fine-tuning

Enables flexible combination with existing techniques

Abstract

Subject-driven text-to-image generation models create novel renditions of an input subject based on text prompts. Existing models suffer from lengthy fine-tuning and difficulties preserving the subject fidelity. To overcome these limitations, we introduce BLIP-Diffusion, a new subject-driven image generation model that supports multimodal control which consumes inputs of subject images and text prompts. Unlike other subject-driven generation models, BLIP-Diffusion introduces a new multimodal encoder which is pre-trained to provide subject representation. We first pre-train the multimodal encoder following BLIP-2 to produce visual representation aligned with the text. Then we design a subject representation learning task which enables a diffusion model to leverage such visual representation and generates new subject renditions. Compared with previous methods such as DreamBooth, our model enables zero-shot subject-driven generation, and efficient fine-tuning for customized subject with up to 20x speedup. We also demonstrate that BLIP-Diffusion can be flexibly combined with existing techniques such as ControlNet and prompt-to-prompt to enable novel subject-driven generation and editing applications. Code and models will be released at https://github.com/salesforce/LAVIS/tree/main/projects/blip-diffusion. Project page at https://dxli94.github.io/BLIP-Diffusion-website/.