paper-2-web

Paper2All: Automated Tool for Converting Academic Papers and Generating Promotional Materials

Overview

Paper2All is an academic paper conversion pipeline that can automatically convert research papers in LaTeX or PDF format into interactive websites, conference posters, and presentation videos, for conference preparation, research promotion, and enhanced preprint presentation.

Use Cases

1. Academic Conference Materials Preparation

Prepare a complete promotional package for conference submissions or attendance, including print-ready posters for poster sessions, presentation videos for oral talks, and a paper companion website accessible via QR code for attendees. Supports custom poster sizes (e.g., 4'×3', A0, etc.) and video durations.

2. Preprint Enhancement

Add an interactive homepage for papers submitted to preprint servers such as arXiv and bioRxiv to improve accessibility and dissemination. The generated site supports responsive design, interactive charts, and citation links, making it suitable as a permanent showcase for research outputs.

3. Research Promotion and Outreach

Convert published papers into formats better suited for social media and lab websites, including video abstracts (1–3 minute social-media versions or 5–10 minute journal versions), printable conference posters, and easy-to-share web versions.

Core Features

Paper2Web: Interactive Website Generation

Automatically generate a responsive academic homepage from the paper source, including multi-section layouts, interactive charts and tables, and mobile-friendly navigation. The system can automatically discover institutional logos (requires configuring the Google Search API) and performs aesthetic optimization and quality assessment of the design. The generated site can be deployed directly to GitHub Pages, a university server, or a cloud platform.

Paper2Video: Presentation Video Production

Automatically generate slides from the paper structure, synthesize voiceover narration, and synchronize cursor movements and highlighted annotations. Supports multilingual speech synthesis and optionally adding a virtual presenter video based on Hallo2 (requires a 48GB GPU). Suitable for creating video summaries, conference presentation videos, or online course materials.

Paper2Poster: Conference Poster Generation

Create print-ready academic posters with support for arbitrary custom sizes, professional design templates, and institutional branding elements. Automatically generate QR codes linking to the paper or website and output high-resolution PDFs (300+ DPI). The generated posters can be submitted directly to professional printing services or university print shops.

Frequently Asked Questions

What input formats does Paper2All support?

The system recommends using LaTeX source files for the best content extraction results; the supported directory structure should include a main file (main.tex), a figures folder, and references. High-quality PDF input is also supported, but PDFs must contain selectable text (not scanned images) and high-resolution figures (recommended 300+ DPI).

How long does it take to generate each component?

Using GPT-4 models, website generation takes about 15–30 minutes, poster generation about 10–20 minutes, and video generation without a virtual presenter about 20–60 minutes. Adding a virtual presenter increases processing time to 60–120 minutes. In urgent cases, it is recommended to generate in the priority order: website → poster → video.

Is a GPU required for video generation?

Standard video generation (slides + narration) does not require a GPU—only a CPU and sufficient memory (16GB minimum, 32GB recommended). A 48GB VRAM NVIDIA GPU (e.g., A6000) is only required when enabling the virtual presenter feature. GPU requirements are high, so this feature is recommended only when a professional presentation video is needed.