ACD: Direct Conditional Control for Video Diffusion Models via Attention Supervision
This addresses the need for more precise control in video generation for applications like content creation, though it is an incremental improvement over existing methods.
The paper tackled the problem of limited controllability in video synthesis by proposing ACD, a framework that uses attention supervision to align video diffusion models with external control signals, achieving superior alignment with conditioning inputs while preserving temporal coherence and visual fidelity.
Controllability is a fundamental requirement in video synthesis, where accurate alignment with conditioning signals is essential. Existing classifier-free guidance methods typically achieve conditioning indirectly by modeling the joint distribution of data and conditions, which often results in limited controllability over the specified conditions. Classifier-based guidance enforces conditions through an external classifier, but the model may exploit this mechanism to raise the classifier score without genuinely satisfying the intended condition, resulting in adversarial artifacts and limited effective controllability. In this paper, we propose Attention-Conditional Diffusion (ACD), a novel framework for direct conditional control in video diffusion models via attention supervision. By aligning the model's attention maps with external control signals, ACD achieves better controllability. To support this, we introduce a sparse 3D-aware object layout as an efficient conditioning signal, along with a dedicated Layout ControlNet and an automated annotation pipeline for scalable layout integration. Extensive experiments on benchmark video generation datasets demonstrate that ACD delivers superior alignment with conditioning inputs while preserving temporal coherence and visual fidelity, establishing an effective paradigm for conditional video synthesis.