Smart-Insertion-V: Photorealistic Video Insertion via a Closed-Loop Feedback Dual-Stream Framework

Mask-free video object insertion has emerged as a challenging task, requiring harmonious integration of reference objects into source videos. However, existing methods struggle when references exhibit severe stylistic domain gaps with the source scene. To overcome this, we propose $\mathit{\text{Smart-Insertion-V}}$, an end-to-end $\mathbf{\text{Dual-Stream}}$ framework that concurrently conducts video insertion and image style transfer. Within this framework, the image stream synchronously guides the video generation process, while a $\mathbf{\text{Closed-loop}}\mathbf{\text{Feedback}}$ mechanism is further incorporated to ensure robust insertion. Inevitably, integrating these diverse conditioning signals results in feature entanglement and style leakage. To tackle this issue, we design $\mathbf{\text{Dual-World-View}}\mathbf{\text{RoPE}}$ to distinguish different signals via spatial-temporal offsets without incurring heavy training overhead. Furthermore, to facilitate spatial grounding and stylistic adaptation, we introduce a $\mathbf{\text{Decoupled}}\mathbf{\text{Guidance}}\mathbf{\text{Module}}$ that leverages a Vision-Language Model for semantic reasoning while preserving original temporal guidance with native text encoder. To bridge data gap for harmonious reference insertion task, we propose a data curation pipeline and will release an $\mathbf{\text{open-source}}\mathbf{\text{dataset}}$. Experiments demonstrate that our method can insert objects into plausible positions while achieving the most harmonious results.