0h7c8bggs3o0hh72h4fi4_source.mp4 -

: Synchronizes a virtual cursor with the narration to highlight specific areas of the slides.

: Uses Vision-Language Models (VLMs) to create narration subtitles and visual-focus prompts.

If you are looking for a of a specific section or want to know how to run the code , let me know! 0h7c8bggs3o0hh72h4fi4_source.mp4

: Automatically generates and refines LaTeX-based slides from the paper's text.

You can find more details, the full paper, and video demos on the official Paper2Video Project Page . : Synchronizes a virtual cursor with the narration

: Generates a realistic, personalized talking-head video using a portrait and voice sample.

Paper2Video: Automatic Video Generation from Scientific Papers the full paper

The paper introduces , a multi-agent framework that automatically converts academic papers into professional presentation videos. It breaks the process down into four distinct "builders":