OpenDCAI's Paper2Any Turns Research PDFs Into Editable Slides and Diagrams
Paper2Any is an open-source Python toolkit that turns research PDFs, screenshots, or plain text into editable PPTX slides, scientific figures, draw.io diagrams, and academic posters in one click.
Takeaways
- What it is: Paper2Any is an open-source toolkit that converts research PDFs, text, or screenshots into editable PPTX, SVG, and draw.io files.
- 12 sub-tools: Covers Paper2Figure, Paper2PPT, PDF2PPT, Paper2Poster, Paper2Citation, Paper2Rebuttal, PPTPolish, and more.
- Architecture: Built on LangGraph agents, FastAPI backend, React frontend, with MinerU and SAM3 for PDF parsing and image segmentation.
- Model-agnostic: Supports GPT-4o, Qwen-VL, Gemini, and any OpenAI-compatible API endpoint via config.
- Self-hostable via Docker: One-command Docker deploy; Apache 2.0 license allows commercial use.
- Online demo available at dcai-paper2any.nas.cpolar.cn; a commercial Studio version is being rolled out separately.
Paper2Any is an open-source toolkit from OpenDCAI that automates one of the most tedious parts of academic work: turning a research paper into presentation-ready visuals. Drop in a PDF, a screenshot, or even a plain-text description, and the system generates editable PowerPoint slides, model architecture diagrams, technical roadmaps, draw.io flowcharts, and academic posters. Everything comes out as a native, editable file, not a static image you have to redraw from scratch.

The problem it actually solves
Anyone who has tried to recreate a model architecture diagram from a paper knows the pain: you read the methods section, open PowerPoint or Inkscape, and spend an hour manually placing boxes and arrows. Paper2Any addresses the challenge of transforming research papers, text, or topics into editable visual assets like figures, technical diagrams, and presentation slides. The key word is editable. The outputs are .pptx and .svg files you can open and modify, not screenshots.
The project employs AI-powered multimodal workflows to process inputs such as PDF documents, screenshots, or plain text, leveraging specialized agents and toolkits to generate editable outputs including scientific figures, presentation slides, and video scripts.
What's inside the toolkit
From paper PDFs, images, or text, the system generates editable scientific figures, slide decks, video scripts, academic posters, and other multimodal content in one click. The toolkit is split into focused sub-tools:
- Paper2Figure: generates model architecture diagrams, technical roadmaps (PPT + SVG), and experimental plots with editable PPTX output.
- Paper2PPT