Advancing Diffusion Models: Alias-Free Resampling and Enhanced Rotational Equivariance

TL;DR

This paper introduces alias-free resampling layers into diffusion models to reduce artifacts and improve image quality, while also enabling user-controlled rotation of generated images without extra training.

Contribution

It proposes a novel alias-free resampling method integrated into diffusion models and a modified diffusion process for controllable image rotation, both without increasing model complexity.

Findings

Consistent improvements in FID and KID scores on benchmark datasets.

Enhanced rotational equivariance in generated images.

Maintained computational efficiency with no extra trainable parameters.

Abstract

Recent advances in image generation, particularly via diffusion models, have led to impressive improvements in image synthesis quality. Despite this, diffusion models are still challenged by model-induced artifacts and limited stability in image fidelity. In this work, we hypothesize that the primary cause of this issue is the improper resampling operation that introduces aliasing in the diffusion model and a careful alias-free resampling dictated by image processing theory can improve the model's performance in image synthesis. We propose the integration of alias-free resampling layers into the UNet architecture of diffusion models without adding extra trainable parameters, thereby maintaining computational efficiency. We then assess whether these theory-driven modifications enhance image quality and rotational equivariance. Our experimental results on benchmark datasets, including CIFAR-10, MNIST, and MNIST-M, reveal consistent gains in image quality, particularly in terms of FID and KID scores. Furthermore, we propose a modified diffusion process that enables user-controlled rotation of generated images without requiring additional training. Our findings highlight the potential of theory-driven enhancements such as alias-free resampling in generative models to improve image quality while maintaining model efficiency and pioneer future research directions to incorporate them into video-generating diffusion models, enabling deeper exploration of the applications of alias-free resampling in generative modeling.