Instruction-based video editing has witnessed rapid progress, yet current methods often struggle with precise visual control, as natural language is inherently limited in describing complex visual nuances. Although reference-guided editing offers a robust solution, its potential is currently bottlenecked by the scarcity of high-quality paired training data. To bridge this gap, we introduce a scalable data generation pipeline that transforms existing video editing pairs into high-fidelity training quadruplets, leveraging image generative models to create synthesized reference scaffolds. Using this pipeline, we construct RefVIE, a large-scale dataset tailored for instruction-reference-following tasks, and establish RefVIE-Bench for comprehensive evaluation. Furthermore, we propose a unified editing architecture, Kiwi-Edit, that synergizes learnable queries and latent visual features for reference semantic guidance. Our model achieves significant gains in instruction following and reference fidelity via a progressive multi-stage training curriculum. Extensive experiments demonstrate that our data and architecture establish a new state-of-the-art in controllable video editing. All datasets, models, and code is released at https://github.com/showlab/Kiwi-Edit.

翻译：基于指令的视频编辑技术已取得快速进展，但现有方法常因自然语言在描述复杂视觉细节上的固有局限而难以实现精确的视觉控制。尽管参考引导编辑提供了稳健的解决方案，但其潜力目前受限于高质量配对训练数据的稀缺。为弥合这一差距，我们引入了一种可扩展的数据生成流程，该流程利用图像生成模型创建合成的参考支架，将现有视频编辑对转化为高保真度的训练四元组。基于此流程，我们构建了专为指令-参考跟随任务定制的大规模数据集 RefVIE，并建立了用于全面评估的 RefVIE-Bench。此外，我们提出了一种统一的编辑架构 Kiwi-Edit，它通过可学习的查询与潜在视觉特征的协同作用来实现参考语义引导。通过渐进式多阶段训练策略，我们的模型在指令跟随与参考保真度方面取得了显著提升。大量实验表明，我们的数据与架构在可控视频编辑领域确立了新的技术标杆。所有数据集、模型与代码均已发布于 https://github.com/showlab/Kiwi-Edit。