FreeU: Free Lunch in Diffusion U-Net

TL;DR

This paper introduces FreeU, a simple method to improve diffusion U-Net image and video generation quality by re-weighting skip connections and backbone features during inference, without additional training.

Contribution

The paper reveals the distinct roles of U-Net components and proposes FreeU, a technique to enhance generation quality by adjusting feature contributions without retraining.

Findings

Improved image and video generation quality demonstrated.

Easy integration with existing diffusion models.

Requires only two scaling adjustments during inference.

Abstract

In this paper, we uncover the untapped potential of diffusion U-Net, which serves as a "free lunch" that substantially improves the generation quality on the fly. We initially investigate the key contributions of the U-Net architecture to the denoising process and identify that its main backbone primarily contributes to denoising, whereas its skip connections mainly introduce high-frequency features into the decoder module, causing the network to overlook the backbone semantics. Capitalizing on this discovery, we propose a simple yet effective method-termed "FreeU" - that enhances generation quality without additional training or finetuning. Our key insight is to strategically re-weight the contributions sourced from the U-Net's skip connections and backbone feature maps, to leverage the strengths of both components of the U-Net architecture. Promising results on image and video generation tasks demonstrate that our FreeU can be readily integrated to existing diffusion models, e.g., Stable Diffusion, DreamBooth, ModelScope, Rerender and ReVersion, to improve the generation quality with only a few lines of code. All you need is to adjust two scaling factors during inference. Project page: https://chenyangsi.top/FreeU/.