FFP-300K: Scaling First-Frame Propagation for Generalizable Video Editing
Xijie Huang, Chengming Xu, Donghao Luo, Xiaobin Hu, Peng Tang, Xu Peng, Jiangning Zhang, Chengjie Wang, Yanwei Fu
2026-01-07
Summary
This paper focuses on improving how we edit videos by changing their content while keeping things looking natural, specifically using a technique called First-Frame Propagation (FFP).
What's the problem?
Current video editing methods using FFP require a lot of manual input and guidance during the editing process, making them difficult to use. The main reason for this is that the datasets used to train these systems aren't good enough – they're often too short, the video quality is low, and they don't cover enough different types of edits to teach the system how to handle various situations effectively.
What's the solution?
The researchers created a new, much larger dataset called FFP-300K, containing 300,000 high-quality, longer videos designed for diverse editing tasks. They also developed a new system that doesn't need constant guidance. This system uses a clever technique called Adaptive Spatio-Temporal RoPE to separate how things *look* from how they *move* in the video, and a 'self-distillation' method to ensure the video stays consistent over time and doesn't change unexpectedly.
Why it matters?
This work is important because it makes video editing much easier and more automated. By creating a better dataset and a more sophisticated system, they've significantly improved the quality of edited videos, surpassing existing methods in terms of both visual appeal and how well the edits fit the original video content.
Abstract
First-Frame Propagation (FFP) offers a promising paradigm for controllable video editing, but existing methods are hampered by a reliance on cumbersome run-time guidance. We identify the root cause of this limitation as the inadequacy of current training datasets, which are often too short, low-resolution, and lack the task diversity required to teach robust temporal priors. To address this foundational data gap, we first introduce FFP-300K, a new large-scale dataset comprising 300K high-fidelity video pairs at 720p resolution and 81 frames in length, constructed via a principled two-track pipeline for diverse local and global edits. Building on this dataset, we propose a novel framework designed for true guidance-free FFP that resolves the critical tension between maintaining first-frame appearance and preserving source video motion. Architecturally, we introduce Adaptive Spatio-Temporal RoPE (AST-RoPE), which dynamically remaps positional encodings to disentangle appearance and motion references. At the objective level, we employ a self-distillation strategy where an identity propagation task acts as a powerful regularizer, ensuring long-term temporal stability and preventing semantic drift. Comprehensive experiments on the EditVerseBench benchmark demonstrate that our method significantly outperforming existing academic and commercial models by receiving about 0.2 PickScore and 0.3 VLM score improvement against these competitors.