English static mirror for SEO/GEO · AI-assisted translation · Read Chinese original

Aurora: An Agentic Framework for Unified Video Editing with Tool-Using VLMs

Forum topic · 小凯 · 2026-05-20

Summary

Aurora is an agentic video editing framework that pairs a tool-augmented vision-language model (VLM) agent with a unified video diffusion transformer. Recent unified video editing models rely on a single diffusion transformer that jointly consumes text, source video, and reference images, but they assume users already provide model-ready prompts, references, and spatial grounding, which real requests often lack. Aurora addresses this gap by letting the VLM agent translate raw user requests into a structured edit plan aligned with the transformer's conditioning channels, resolving textual and visual underspecification before generation. The authors train the VLM agent with supervised data for full edit planning and reference image selection, and use preference pairs to improve tool-use robustness and instruction refinement. They also introduce AgentEdit-Bench, a benchmark for evaluating agent-augmented video editing under text and visual underspecification. Experiments on AgentEdit-Bench and two existing benchmarks show that Aurora outperforms instruction-only baselines and that the trained VLM agent transfers to compatible frozen video editing models.

Paper Overview

  • Field: Computer Vision (CV)
  • Authors: Yongsheng Yu, Ziyun Zeng, Zhiyuan Xiao
  • Published: 2026-05-19
  • arXiv: 2505.14307
  • Summary

    Recent video editing models have converged on a unified conditioning design: a single diffusion transformer jointly consumes text, source video, and reference images, and one set of weights covers replacement, removal, style transfer, and reference-driven insertion. The design is flexible, but it assumes that the user already provides model-ready text, reference images, and spatial grounding for local edits, which real requests often omit.

    Aurora is an agentic video editing framework that pairs a tool-augmented vision-language model (VLM) agent with a unified video diffusion transformer. The VLM agent maps a raw user request to a structured edit plan aligned with the transformer's conditioning channels, thereby resolving textual and visual underspecification before generation.

    The VLM agent is trained with supervised data to produce complete edit plans and select reference images, while preference pairs are used to enhance the robustness of tool use and instruction refinement.

    The authors introduce AgentEdit-Bench, a benchmark designed to evaluate agent-enhanced video editing under text and visual underspecification.

    Key Contributions

  • Agentic editing framework: Decouples *intent resolution* (handled by the VLM agent) from *video generation* (handled by the diffusion transformer), allowing the editing model to receive fully specified conditioning inputs.
  • Training recipe: Supervised training for full edit planning and reference image selection, combined with preference-pair optimization for tool-use robustness and instruction refinement.
  • AgentEdit-Bench: A new benchmark specifically targeting underspecified editing requests that lack ready-made prompts, references, or spatial localizations.
  • Transferability: The trained VLM agent can be transferred to compatible frozen video editing models without retraining the underlying diffusion transformer.
  • Experimental Results

    Experiments on AgentEdit-Bench and two existing video editing benchmarks show that Aurora outperforms instruction-only baselines. The transferability of the VLM agent to compatible frozen video editing models is also demonstrated, suggesting that the agentic layer is a reusable, model-agnostic front-end for unified video editing systems.

    Resources

  • Paper: https://arxiv.org/abs/2505.14307
  • Project page: https://yeates.github.io/Aurora-Page

Tags

#video-editing#diffusion-transformer#vlm-agent#agentic-ai#reference-driven-editing#benchmark#arxiv-2505-14307

This page is an English static mirror generated for search and AI citation. It may be a full translation or structured summary of the Chinese original. Canonical interactive discussion lives on the Chinese page: https://zhichai.net/topic/177620483