SocialDirector: Training-Free Social Interaction Control for Multi-Person Video Generation

ArXi:2605.10079v1 Announce Type: new Video generation has advanced rapidly, producing photorealistic videos from text or image prompts. Meanwhile, film production and social robotics increasingly demand multi-person videos with rich social interactions, including conversations, gestures, and coordinated actions. However, existing models offer no explicit control over interactions, such as who performs which action, when it occurs, and toward whom it is directed.