Topic2Manim

Automatic educational video generator using AI and Manim. Converts any topic into a professional animated video with narration and mathematical visualizations.

User Interface

Examples

propmt: How do machines learn to recognize MNIST dataset numbers?

model: claude-sonnet-4-5-20250929

response:

propmt: What is a Markov chain and how are they related to LLMs?

model: claude-sonnet-4-5-20250929

response:

propmt: How does Cramer's rule work for system of linear equations?

model: claude-sonnet-4-5-20250929

response:

propmt: how chat gpt works?

model: gpt-5.2

response:

propmt: how tokenization works in chat gpt?

model: gpt-5.2

response:

Features

• Multi-LLM Support with automatic fallback (OpenAI GPT, Claude)
• Automatic script generation using advanced language models
• Educational animations with Manim Community Edition
• Multi-language support (automatically detects topic language)
• Optimized videos of ~60 seconds with multiple scenes
• Automatic concatenation of fragments into final video

Architecture

System Overview

Topic2Manim is a multi-agent system that orchestrates several specialized components to transform a topic into an educational video. The system follows a pipeline architecture where each agent has a specific responsibility.

graph TB
    subgraph Input
        A[User Topic]
    end
    
    subgraph "LLM Configuration"
        B[setup_llm_client]
        B1[Claude API]
        B2[OpenAI API]
        B -->|Priority 1| B1
        B -->|Fallback| B2
    end
    
    subgraph "Agent 1: Script Generation"
        C[animations.py]
        C1[generate_script_json]
        C --> C1
    end
    
    subgraph "Agent 2: TTS Generation"
        D[tts_generator.py]
        D1[generate_complete_audio]
        D2[generate_audio_fragment]
        D3[concatenate_audio_fragments]
        D1 --> D2
        D2 --> D3
    end
    
    subgraph "Agent 3: Manim Code Generation"
        E[manim_generator.py]
        E1[generate_manim_code]
        E --> E1
    end
    
    subgraph "Agent 4: Video Compilation"
        F[concat_video.py]
        F1[compile_video]
        F2[concatenate_videos]
        F3[merge_video_and_audio]
        F1 --> F2
        F2 --> F3
    end
    
    subgraph Output
        G[Final Video with Audio]
    end
    
    A --> B
    B --> C1
    C1 -->|video-output.json| D1
    C1 -->|video-output.json| E1
    D1 -->|audio durations| E1
    E1 -->|.py files| F1
    F1 -->|.mp4 fragments| F2
    D3 -->|audio.mp3| F3
    F2 -->|output_silent.mp4| F3
    F3 --> G
    
    style A fill:#e1f5ff
    style G fill:#c8e6c9
    style C fill:#fff9c4
    style D fill:#ffe0b2
    style E fill:#f8bbd0
    style F fill:#d1c4e9

Loading

Installation

First, install uv (if you haven't already):

# On macOS and Linux
curl -LsSf https://astral.sh/uv/install.sh | sh

# Or using pip
pip install uv

Setup

git clone https://github.com/mateolafalce/topic2manim.git
cd topic2manim

# Create virtual environment and install all dependencies
uv sync

# Activate the virtual environment
source .venv/bin/activate  # On Windows: .venv\Scripts\activate

cp .env.example .env

Usage

Start the Flask server:

python src/main.py

Then open your browser and navigate to:

http://localhost:5000

or

docker compose up