How lothnic/zographika works
Project type and purpose
• This is a full-stack web application (“Zographika”) whose goal is to let end users enter text prompts in a React/Vite front-end, forward those prompts to a Flask back-end, use Google’s Gemini API to generate Manim scripts, render animations with Manim, and return the resulting media to the front end for display or download.
File structure
• Root files:
• app.py – entry point for the Flask service
• Dockerfile – builds the back-end container
• render.yaml – Render.com deployment config
• vercel.json – Vercel config for the front end
• requirements.txt – Python dependencies (Flask, Manim, Gemini client, etc.)
• Front-end (frontend/):
• Vite-powered React app (App.jsx, components/, assets/, etc.)
• .env.production holds the back-end URL
• package.json/npm scripts, public/index.html
• Generated artifacts:
• generated_scripts/ – auto-created Manim .py scripts
• templates/ – Jinja2 HTML templates for Flask (e.g., index.html)
• Miscellaneous:
• my_animation.py – example/utility script
• deploy-frontend.sh – helper to build & push front end
README for additional insights
• Workflow: user → React UI → Flask API → Gemini API → Manim CLI → video/gif → React UI
• No persistent database; scripts and outputs live on the file system or in memory.
• Deployment: front end on Vercel; back end on Render.com (Docker).
Detail without over-complication
• Logical component (even if it’s just a small script directory).
• Model data flow from user input → API → rendering → output.
• Deployment boundaries (cloud providers, container boundaries).