MewtwoX23/VTE-Bench

SteerVTE: Precise Video Text Editing with Style and Glyph Control

💡 Abstract

Visual text editing aims to precisely modify text in images and videos while preserving stylistic consistency and visual realism. Despite significant advances in the image domain, video text editing remains largely unexplored: it is a localized task demanding stroke-level precision within small text regions, which compounds the challenges of cross-frame accuracy, temporal coherence, and stylistic fidelity. We introduce 🔥SteerVTE🔥, a unified framework that steers a frozen video diffusion model to perform precise Video Text Editing through style and glyph control. Built on a frozen diffusion transformer, SteerVTE attaches a lightweight text context adapter with two complementary modules: a style encoder capturing the original text's visual attributes, and dual-granularity glyph encoders encoding the target text at both the line and character levels. To overcome the inherently weak text rendering priors of video foundation models, we further propose a glyph-aware spatial-focal loss and a three-stage progressive training curriculum that scales from image to video data. To support large-scale training, we also develop an automatic synthesis pipeline and construct SteerVTE-1M, a dataset of one million triplets spanning diverse scenes, fonts, and stylistic effects. Extensive experiments demonstrate that SteerVTE substantially outperforms existing video editing baselines across text accuracy, style consistency, and temporal coherence.

📊 Dataset

VTE-Bench is a benchmark for video text editing, covering synthetic and real-world scenes with diverse text content, visual styles, and motion patterns. Each sample contains the source video, the target video or editing condition, a text mask, and reference images used to evaluate text editing quality.

The dataset is organized into three subsets:

• Synth: synthetic video text editing samples;
• Real: real-world video text editing samples;
• SceneText: scene text editing samples collected from diverse videos.

The corresponding CSV files (Synth.csv, Real.csv, and SceneText.csv) provide the sample metadata and relative paths to the videos, masks, and reference images.

📚 Citation

If you find SteerVTE is useful in your research or applications, please consider giving us a star 🌟 and citing it by the following BibTeX entry.

@article{zeng2026steervte,
  title={SteerVTE: Seamless Video Text Editing with Style and Glyph Control},
  author={Zeng, Kai and Li, Moran and Wang, Zhengwei and Yu, Yingchen and Lin, Yiheng and An, Ruichuan and Lu, Ming and She, Qi and Zhang, Wentao},
  journal={arXiv preprint arXiv:2606.23254},
  year={2026}
}