Seen-to-Scene: Keep the Seen, Generate the Unseen for Video Outpainting

TL;DR

Seen-to-Scene introduces a unified framework combining propagation and generation for video outpainting, enhancing temporal coherence and visual realism over prior methods.

Contribution

It unifies propagation-based and generation-based approaches, using flow completion and reference-guided latent propagation for improved video outpainting.

Findings

Achieves superior temporal coherence and visual realism.

Outperforms prior state-of-the-art methods.

Provides efficient inference without input-specific adaptation.

Abstract

Video outpainting aims to expand the visible content of a video beyond the original frame boundaries while preserving spatial fidelity and temporal coherence across frames. Existing methods primarily rely on large-scale generative models, such as diffusion models. However, generationbased approaches suffer from implicit temporal modeling and limited spatial context. These limitations lead to intraframe and inter-frame inconsistencies, which become particularly pronounced in dynamic scenes and large outpainting scenarios. To overcome these challenges, we propose Seen-to-Scene, a novel framework that unifies propagationbased and generation-based paradigms for video outpainting. Specifically, Seen-to-Scene leverages flow-based propagation with a flow completion network pre-trained for video inpainting, which is fine-tuned in an end-to-end manner to bridge the domain gap and reconstruct coherent motion fields. To further improve the efficiency and reliability of propagation, we introduce a reference-guided latent propagation that effectively propagates source content across frames. Extensive experiments demonstrate that our method achieves superior temporal coherence and visual realism with efficient inference, surpassing even prior state-of-the-art methods that require input-specific adaptation.