Manim Flow

Overview

Manim Flow is an AI-powered system that converts natural language prompts into 2D animated videos. Users generate animation code using AI, preview it in the browser, and render final videos through an isolated Python rendering service.

Problem

Generating educational or explanatory animations using Manim requires strong Python knowledge and a complex local setup. This creates a high barrier for users who want quick visualizations from simple ideas.

Solution

I built a full-stack pipeline that translates user prompts into Manim Python code using AI and renders videos inside a Dockerized environment. The system tracks rendering status asynchronously and delivers videos through cloud storage.

System Architecture

View Diagram

Core Features

• Multi-model AI code generation using Google Gemini and Groq AI for diverse animation styles.
• User authentication and profile management using Next.js API Routes.
• Community Gallery for sharing and discovering animations.
• Future support for Playground for testing and experimenting with animations.

Key Engineering Decisions

• Adopted Next.js full-stack for unified frontend/backend and type safety.
• Integrated Google Gemini and Groq AI for varied animation styles via API calls.
• Used PostgreSQL (Neon) for user data and animation metadata storage.
• Implemented isolated Python rendering service for secure video generation.

Tech Stack

Demo