On-the-fly Repulsion in the Contextual Space for Rich Diversity in Diffusion Transformers

TL;DR

This paper introduces a novel on-the-fly repulsion method in the Contextual Space of diffusion transformers, significantly enhancing diversity in generated images without compromising quality or requiring costly optimization.

Contribution

It proposes a new framework for diversity by applying repulsion during the transformer's forward pass, acting on intermediate latents to improve variety efficiently.

Findings

Achieves richer diversity without losing visual fidelity or semantic accuracy.

Effective even in modern turbo and distilled models where traditional methods fail.

Imposes minimal computational overhead during inference.

Abstract

Modern Text-to-Image (T2I) diffusion models have achieved remarkable semantic alignment, yet they often suffer from a significant lack of variety, converging on a narrow set of visual solutions for any given prompt. This typicality bias presents a challenge for creative applications that require a wide range of generative outcomes. We identify a fundamental trade-off in current approaches to diversity: modifying model inputs requires costly optimization to incorporate feedback from the generative path. In contrast, acting on spatially-committed intermediate latents tends to disrupt the forming visual structure, leading to artifacts. In this work, we propose to apply repulsion in the Contextual Space as a novel framework for achieving rich diversity in Diffusion Transformers. By intervening in the multimodal attention channels, we apply on-the-fly repulsion during the transformer's forward pass, injecting the intervention between blocks where text conditioning is enriched with emergent image structure. This allows for redirecting the guidance trajectory after it is structurally informed but before the composition is fixed. Our results demonstrate that repulsion in the Contextual Space produces significantly richer diversity without sacrificing visual fidelity or semantic adherence. Furthermore, our method is uniquely efficient, imposing a small computational overhead while remaining effective even in modern "Turbo" and distilled models where traditional trajectory-based interventions typically fail.