Diffusion Models Beat GANs on Image Synthesis

TL;DR

Diffusion models surpass GANs in image synthesis quality, achieving state-of-the-art results on ImageNet with improved architecture and classifier guidance, while maintaining efficiency and better distribution coverage.

Contribution

The paper introduces improved diffusion architectures and a classifier guidance method that significantly enhance image synthesis quality over GANs and previous diffusion models.

Findings

Achieved FID of 2.97 on ImageNet 128x128

Matched BigGAN-deep performance with fewer passes

Enhanced image quality with classifier guidance and upsampling diffusion

Abstract

We show that diffusion models can achieve image sample quality superior to the current state-of-the-art generative models. We achieve this on unconditional image synthesis by finding a better architecture through a series of ablations. For conditional image synthesis, we further improve sample quality with classifier guidance: a simple, compute-efficient method for trading off diversity for fidelity using gradients from a classifier. We achieve an FID of 2.97 on ImageNet 128$\times$128, 4.59 on ImageNet 256$\times$256, and 7.72 on ImageNet 512$\times$512, and we match BigGAN-deep even with as few as 25 forward passes per sample, all while maintaining better coverage of the distribution. Finally, we find that classifier guidance combines well with upsampling diffusion models, further improving FID to 3.94 on ImageNet 256$\times$256 and 3.85 on ImageNet 512$\times$512. We release our code at https://github.com/openai/guided-diffusion