Devil is in the Details: Density Guidance for Detail-Aware Generation with Flow Models

TL;DR

This paper introduces Density Guidance, a novel method for controlling image detail and density in flow-based generative models, improving the balance between realism and detail during image synthesis.

Contribution

It presents a principled approach to exact log-density control in flow models, explaining prior guidance techniques and extending control to stochastic sampling.

Findings

Density Guidance enables precise control over image detail.

The method maintains high sample quality while adjusting density.

Score alignment explains the effectiveness of prior guidance techniques.

Abstract

Diffusion models have emerged as a powerful class of generative models, capable of producing high-quality images by mapping noise to a data distribution. However, recent findings suggest that image likelihood does not align with perceptual quality: high-likelihood samples tend to be smooth, while lower-likelihood ones are more detailed. Controlling sample density is thus crucial for balancing realism and detail. In this paper, we analyze an existing technique, Prior Guidance, which scales the latent code to influence image detail. We introduce score alignment, a condition that explains why this method works and show that it can be tractably checked for any continuous normalizing flow model. We then propose Density Guidance, a principled modification of the generative ODE that enables exact log-density control during sampling. Finally, we extend Density Guidance to stochastic sampling, ensuring precise log-density control while allowing controlled variation in structure or fine details. Our experiments demonstrate that these techniques provide fine-grained control over image detail without compromising sample quality. Code is available at https://github.com/Aalto-QuML/density-guidance.