一個編輯器，多種編輯：用於多樣化影片編輯的統一免訓練框架

摘要

影片編輯涵蓋多種編輯典範，然而要在單一統一框架中同時實現高品質的指令引導與主體引導編輯仍具挑戰性。我們提出 EditVid，一個免訓練框架，結合了用於局部一致性的稀疏因果記憶、用於長程身分保持的基於對應關係之注意力後 token 注入，以及用於編輯局部性的軟潛在融合。此框架同時支援指令引導與參考引導的編輯，包括風格轉換、屬性修改、物體插入、部件級編輯與主體替換。在 FiVE 上，EditVid 達到 78.16 的 FiVE-Acc，相較於最強之已評估免訓練基線的 58.95，同時在 IVEBench 上亦取得具競爭力之成果。一項使用者研究進一步顯示，相較於 7 種競爭方法，EditVid 獲得 51.8% 的整體偏好。

English

Video editing spans diverse editing paradigms, yet achieving high-quality instruction-guided and subject-guided editing within a single unified framework remains challenging. We introduce EditVid, a training-free framework combining sparse causal memory for local coherence, correspondence-based post-attention token injection for long-range identity preservation, and soft latent blending for edit locality. The same framework supports instruction-guided and reference-guided edits, including style transfer, attribute modification, object insertion, part-level editing, and subject replacement. On FiVE, EditVid achieves 78.16 FiVE-Acc, compared with 58.95 for the strongest evaluated training-free baseline, while obtaining competitive results on IVEBench. A user study further shows a 51.8\% overall preference for EditVid over 7 competing methods.