🪄 Text2Manim

Text2Manim is an open-source tool that lets you generate stunning math animations using Manim — just by describing them in plain English.

No need to write code.
Just describe the animation → Get Manim code → Render your video.

🎯 Aim

To make mathematical animations accessible to everyone — educators, students, and creators — by allowing them to describe what they want using natural language, and automatically generating the corresponding Manim Python code.

🧠 Motivation

Manim is a powerful animation engine but has a steep learning curve. Many educators and students don’t have the time to learn its syntax.
Text2Manim removes that barrier by converting natural descriptions into Manim animations, helping users bring abstract ideas to life with ease.

✅ Objectives

• Convert plain English descriptions into Manim code
• Render the code into animation previews or downloadable videos
• Provide a simple web interface to view animations

💡 Model Flexibility

Text2Manim supports any LLM of your choice — including OpenAI, Claude, and open models via OpenRouter.

To use other Models Change the model name and model provider

model = init_chat_model("gemini-2.5-flash", model_provider="google_genai")

To use OpenRouter:

from backend.utils.openrouter_llm import ChatOpenRouter
llm = ChatOpenRouter()

🔧 Setup (Run Locally)

Follow these steps to run Text2Manim locally on your machine:

1. Clone the repository

git clone https://github.com/your-username/text2manim.git
cd text2manim

2. Set up a virtual environment

python3 -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install dependencies

pip install -r backend/requirements.txt

4. Start Redis (via Docker)

Make sure Docker is installed and running, then start Redis:

docker run -d -p 6379:6379 --name text2manim-redis redis

5. Create .env file

Create a .env file in the project root:

AWS_ACCESS_KEY_ID=your_key
AWS_SECRET_ACCESS_KEY=your_secret
AWS_REGION=ap-south-1
GOOGLE_API_KEY=your-key

Replace with your actual AWS credentials and configuration.

6. Start the FastAPI server

uvicorn backend.api.app:app --reload

7. Run Manim rendering container

For sandbox Manim rendering, build the image first:

docker build -t manim-renderer -f backend/worker/Dockerfile .

Manim rendering will be handled in isolation when invoked from the backend.