Place-it-R1: Unlocking Environment-aware Reasoning Potential of MLLM for Video Object Insertion

Researchers introduce Place-it-R1, an AI framework that uses Multimodal Large Language Models to insert objects into videos while maintaining physical realism. The system employs Chain-of-Thought reasoning to ensure inserted objects interact naturally with their environment, addressing the gap between visual quality and physical plausibility in video editing.

• →Place-it-R1 combines MLLMs with video diffusion to create physically realistic video object insertion.
• →The framework uses Chain-of-Thought reasoning to understand scene physics before placing objects.
• →MLLM-guided Spatial Direct Preference Optimization enables the system to score and improve its own outputs.
• →Users can choose between plausibility-focused mode (allowing environment modifications) and fidelity-focused mode (preserving original scene).
• →The system outperforms existing commercial solutions in creating physically coherent video edits.