Diffusion Self-Distillation for Zero-Shot Customized Image Generation

TL;DR

This paper introduces Diffusion Self-Distillation, a method that uses a pre-trained diffusion model to generate a dataset for fine-tuning itself, enabling improved zero-shot image generation with fine-grained control.

Contribution

It presents a novel self-distillation approach that leverages a diffusion model's own capabilities to create training data, enhancing zero-shot image editing without test-time optimization.

Findings

Outperforms existing zero-shot methods in identity-preserving tasks

Competitive with per-instance tuning techniques

Does not require test-time optimization

Abstract

Text-to-image diffusion models produce impressive results but are frustrating tools for artists who desire fine-grained control. For example, a common use case is to create images of a specific instance in novel contexts, i.e., "identity-preserving generation". This setting, along with many other tasks (e.g., relighting), is a natural fit for image+text-conditional generative models. However, there is insufficient high-quality paired data to train such a model directly. We propose Diffusion Self-Distillation, a method for using a pre-trained text-to-image model to generate its own dataset for text-conditioned image-to-image tasks. We first leverage a text-to-image diffusion model's in-context generation ability to create grids of images and curate a large paired dataset with the help of a Visual-Language Model. We then fine-tune the text-to-image model into a text+image-to-image model using the curated paired dataset. We demonstrate that Diffusion Self-Distillation outperforms existing zero-shot methods and is competitive with per-instance tuning techniques on a wide range of identity-preservation generation tasks, without requiring test-time optimization.