Omegance: A Single Parameter for Various Granularities in Diffusion-Based Synthesis

TL;DR

This paper introduces Omegance, a simple yet effective method that uses a single parameter during the denoising process of diffusion models to control the level of detail in generated images and videos, without retraining or architectural changes.

Contribution

It proposes a novel single-parameter approach for granularity control in diffusion-based synthesis, applicable during inference and compatible with existing models.

Findings

Effective control of detail levels demonstrated in image and video synthesis

No need for retraining or model modifications

Applicable to advanced diffusion models

Abstract

In this work, we show that we only need a single parameter $\omega$ to effectively control granularity in diffusion-based synthesis. This parameter is incorporated during the denoising steps of the diffusion model's reverse process. This simple approach does not require model retraining or architectural modifications and incurs negligible computational overhead, yet enables precise control over the level of details in the generated outputs. Moreover, spatial masks or denoising schedules with varying $\omega$ values can be applied to achieve region-specific or timestep-specific granularity control. External control signals or reference images can guide the creation of precise $\omega$ masks, allowing targeted granularity adjustments. Despite its simplicity, the method demonstrates impressive performance across various image and video synthesis tasks and is adaptable to advanced diffusion models. The code is available at https://github.com/itsmag11/Omegance.