AttentionBender: Manipulating Cross-Attention in Video Diffusion Transformers as a Creative Probe

TL;DR

AttentionBender is a tool that manipulates cross-attention in Video Diffusion Transformers to help artists understand and creatively control the generative process, revealing complex entanglements and enabling novel visual effects.

Contribution

It introduces a novel method for visualizing and controlling cross-attention in video diffusion models, combining explainability with artistic exploration.

Findings

Cross-attention manipulation often causes distributed distortions and glitches.

Targeted edits in attention maps are highly entangled and resist localized control.

The tool enables both understanding of transformer mechanisms and creative aesthetic generation.

Abstract

We present AttentionBender, a tool that manipulates cross-attention in Video Diffusion Transformers to help artists probe the internal mechanics of black-box video generation. While generative outputs are increasingly realistic, prompt-only control limits artists' ability to build intuition for the model's material process or to work beyond its default tendencies. Using an autobiographical research-through-design approach, we built on Network Bending to design AttentionBender, which applies 2D transforms (rotation, scaling, translation, etc.) to cross-attention maps to modulate generation. We assess AttentionBender by visualizing 4,500+ video generations across prompts, operations, and layer targets. Our results suggest that cross-attention is highly entangled: targeted manipulations often resist clean, localized control, producing distributed distortions and glitch aesthetics over linear edits. AttentionBender contributes a tool that functions both as an Explainable AI style probe of transformer attention mechanisms, and as a creative technique for producing novel aesthetics beyond the model's learned representational space.