Research map / Video / Video generation
Motion Forcing: Decoupling Ego and Object Motion via Sparse Inputs for Structured Video Generation
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's abstract only. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Proposes Motion Forcing, a method for structured video generation (e.g., autonomous driving) that decouples ego and object motion using sparse user inputs instead of dense 3D annotations.
Problem
Structured video generation such as autonomous driving needs precise spatial guidance, but text-to-video models lack it and dedicated driving world models require expensive dense 3D annotations that are hard for users to provide.
Approach
Uses sparse, user-accessible control inputs and decouples ego-vehicle motion from object motion to bridge the gap between simple human control and dense physical reality in structured video generation.
Experimental setup
Baselines: text-to-video models autonomous driving world models
Closest papers by meaning
- Generating Multimodal Driving Scenes via Next-Scene Prediction (CVPR 2025 · similarity 0.88)
- Wan-Move: Motion-controllable Video Generation via Latent Trajectory Guidance (NeurIPS 2025 · similarity 0.87)
- Learn the Force We Can: Enabling Sparse Motion Control in Multi-Object Video Generation (AAAI 2024 · similarity 0.87)
- Grounding Video Models to Actions through Goal Conditioned Exploration (ICLR 2025 · similarity 0.87)
- Motion Prompting: Controlling Video Generation with Motion Trajectories (CVPR 2025 · similarity 0.87)
- OptiWorld: Optimal Control for Video World Generation under Physical Constraints (NeurIPS 2026 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.