freeu

We propose FreeU, a method that substantially improves diffusion model sample quality at no costs: no training, no additional parameter introduced, and no ...

2023年12月11日 — FreeU is a method that modifies the U-Net noise predictor in Stable Diffusion to enhance the image generation process. It is described in FreeU: ...

FreeU Framework. We propose FreeU, a method that substantially improves diffusion model sample quality at no costs: no training, no additional parameter ...

此工作使用了一个超级简单的方式做了一个即插即用的方法来提高扩散模型的生成质量，无需训练，可以和现有的使用了Unet架构的所有扩散模型相结合。

FreeU is applied during inference and it does not require any additional training. The technique works for different tasks such as text-to-image, image-to-image ...

implementation of FreeU as an a1111 sd webui extension. At each of the 3 stages of the UNet decoder: Apply a scalar on a window over the features of the ...

由 C Si 著作 · 2023 · 被引用 17 次 — Abstract:In this paper, we uncover the untapped potential of diffusion U-Net, which serves as a free lunch that substantially improves the ...

2023年12月19日 — FreeU框架是通用的，与现有的扩散模型无缝集成。通过各种基于扩散的方法证明了样品质量的显著改善，显示了FreeU在不增加额外成本的情况下的有效性。