Math to Manim – Professional Edition

Transform natural language math & science descriptions into cinematic Manim animations — 100% local, offline, private.

✨ Highlights

• Completely local & private — no API keys, no cloud, no data leaves your machine
• Uses code-specialized LLM (qwen2.5-coder:7b recommended)
• Very aggressive code sanitization & repair (~40 common Manim pitfalls automatically fixed)
• Clean, modern Streamlit interface with quality presets (480p – 4K)
• Generation history, natural-language code refinement, custom resolution/FPS
• Automatic cleanup of old temporary files
• Fallback animation when generation fails

🎯 Typical Use Cases

• University lecturers creating step-by-step proof animations
• Math & physics YouTube / TikTok content creators
• Students preparing beautiful presentation videos
• Visual learners who want to see derivations, graphs, transformations
• Rapid prototyping of mathematical visualizations without writing Manim code manually

How It Works :

User enters natural language description

LLM creates structured, concise animation + math plan

Code-specialized LLM generates strict Manim CE v0.18.1 code

Heavy post-processing & error-fixing of the generated code

Manim renders MP4 video (selected quality preset)

Video preview + download appears in the browser

🚀 Quick Start (Windows – 2026)

Prerequisites (~25–45 min first time)

Software	Recommended version / Download	Purpose	Approx. size
Python	3.12.6	Runtime	~70 MB
FFmpeg	latest	Video encoding	~60 MB
MiKTeX	latest (basic + auto-install packages)	LaTeX / MathTex rendering	200+ MB
Ollama	latest	Local LLM runner	~100 MB
qwen2.5-coder	7b (or 14b if ≥16 GB VRAM)	Code generation model	~4.7 GB

Step-by-step Installation

1. Install Python 3.12.6
   https://www.python.org/downloads/release/python-3126/

2. Install FFmpeg
   https://ffmpeg.org/download.html
   → Extract → add the bin folder to your system PATH

3. Install MiKTeX
   https://miktex.org/download
   → It will automatically download missing LaTeX packages on first use

4. Install & start Ollama
   https://ollama.com/download
   Open a terminal / PowerShell:

ollama serve

In a second terminal window:

ollama pull qwen2.5-coder:7b

Install dependencies :

python -m pip install --upgrade pip
pip install -r requirements.txt

Launch the application :

streamlit run app.py

If you want to see a video of the examples that were created: go to example-result Directory

example Use MVP :

MPV-SHOW.mp4

Example animation Create :

animation_1771228341.mp4

For some mathematical examples, this project may give an error and not run correctly. We will try to optimize it soon so that it does not give an error.