SteerVTE: Seamless Video Text Editing with Style and Glyph Control
SteerVTE is introduced, a unified framework that steers a frozen video diffusion model to perform precise Video Text Editing through style and glyph control and substantially outperforms existing video editing baselines across text accuracy, style consistency, and temporal coherence.