Applying Guidance in a Limited Interval Improves Sample and Distribution Quality in Diffusion Models

TL;DR

This paper demonstrates that restricting guidance to an optimal noise level interval in diffusion models enhances image quality and inference efficiency, outperforming traditional constant guidance methods across various datasets and architectures.

Contribution

It introduces a novel approach of limiting guidance to specific noise levels, significantly improving FID scores and inference speed in diffusion models.

Findings

Guidance is harmful at high noise levels and unnecessary at low noise levels.

Limiting guidance to a specific noise interval improves FID from 1.81 to 1.40 on ImageNet-512.

The approach benefits multiple architectures, samplers, and datasets, including Stable Diffusion XL.

Abstract

Guidance is a crucial technique for extracting the best performance out of image-generating diffusion models. Traditionally, a constant guidance weight has been applied throughout the sampling chain of an image. We show that guidance is clearly harmful toward the beginning of the chain (high noise levels), largely unnecessary toward the end (low noise levels), and only beneficial in the middle. We thus restrict it to a specific range of noise levels, improving both the inference speed and result quality. This limited guidance interval improves the record FID in ImageNet-512 significantly, from 1.81 to 1.40. We show that it is quantitatively and qualitatively beneficial across different sampler parameters, network architectures, and datasets, including the large-scale setting of Stable Diffusion XL. We thus suggest exposing the guidance interval as a hyperparameter in all diffusion models that use guidance.