SimInsert: Seamless Video Object Insertion via Regional Sparse Attention Fusion

2026-05-22

AI Summary

Existing video object insertion methods often rely on explicit motion modeling or extensive retraining, struggling to simultaneously achieve spatiotemporal consistency, interaction realism, and generalization. This work proposes a training-free, text-driven framework that decouples the task into single-frame editing and semantic motion description, leveraging the generative priors of image-to-video diffusion models to seamlessly integrate inserted objects into dynamic scenes while preserving the background. The key innovation lies in a non-intrusive guidance mechanism combined with region-sparse attention fusion, which effectively maintains structural consistency, ensures seamless boundaries, and mitigates fidelity drift during denoising. Experiments demonstrate that the proposed method significantly outperforms state-of-the-art approaches, achieving substantial gains in PSNR (+18.8%), SSIM (+20.1%), and LPIPS (−44.1%), thereby enabling high-fidelity video editing.

Abstract

Video object insertion requires ensuring spatio-temporal coherence and interactive realism, extending far beyond simple content placement. However, current approaches are often hindered by a reliance on explicit motion engineering or resource-intensive retraining, restricting their flexibility and generalization. To bridge this gap, we present \textit{SimInsert}, a training-free paradigm that efficiently decouples the task into intuitive single-frame editing and semantic motion description. By harnessing the robust generative priors of image-to-video diffusion models, SimInsert propagates edits temporally, strictly preserving background invariance while enabling plausible, text-driven interactions between the inserted object and the dynamic environment. Our approach hinges on non-invasive guidance mechanisms that enforce structural consistency, facilitate seamless boundary fusion, and counteract the fidelity drift that typically accumulates during the denoising trajectory. Extensive quantitative experiments validate our efficacy: SimInsert surpasses state-of-the-art methods with an 18.8\% gain in PSNR, 20.1\% in SSIM, and a 44.1\% decrease in LPIPS, offering a streamlined solution for high-fidelity video editing.

Problem

• video object insertion
• spatio-temporal coherence
• interactive realism
• background invariance
• fidelity drift

Innovation

• training-free
• regional sparse attention
• image-to-video diffusion
• spatio-temporal coherence
• non-invasive guidance