Tri-Prompting: Video Diffusion with Unified Control over Scene, Subject, and Motion
Tri-Prompting is a unified video diffusion framework that controls scene composition, subject consistency across multiple views, and motion separately. It achieves precise control by using dual conditioning: 3D tracking points for the background and downsampled RGB cues for the foreground subject. This integrated approach significantly improves motion accuracy and identity preservation compared to existing specialized models.