What is this project?

Built something cool that turns prompts into 2D animated videos using the Manim animation engine!

Perfect for teaching, Explaining dev concepts or just geeking out. Just type a prompt → Get a slick animation

What problem does it solve?

Writing code directly with the Manim library to generate 2D videos is quite challenging. Thankfully, LLMs are excellent at generating text (including code). So, we use an LLM to generate Manim code from a prompt, and then use the Manim animation library to render that code into a video.

image.png

Tech Stack & Why This Choice?

Frontend – Next.js + shadcn/ui

Backend – Node.js, WebSocket, redis

Storage – Azure Blob – similar to S3, used for storing video assets.

Deployment – Docker + Azure – containerized setup deployed on Azure for better scalability and integration.

Architecture

High-Level Flow

  1. User logs in (Google OAuth via NextAuth)
  2. Sends prompt → hits backend via API/WebSocket
  3. Worker picks up job via Redis queue
  4. Uses Manim to generate video