Create fictional worlds with AI. Describe a world — and the neural net will enrich it with lore, generate 1–4 images of different scenes, and optionally animate each scene into a short video.

🎨 Made with pollinations.ai — unified API for text, image, and video generation.

You:                     "a dark cyberpunk city where it rains LEDs..."
    ↓
1. GPT-5.4 enriches the description → invents a name + 3–4 paragraphs of lore
    ↓
2. GPT-5.4 generates N image prompts (different scenes of the same world)
    ↓
3. gpt-image-2 renders N images
    ↓
4. (Optional) ltx-2 generates a video for any image

git clone https://github.com/MrMegnis/world-generator.git
cd world-generator

Get your key at enter.pollinations.ai and add it to .env:

cp .env.example .env
# Edit .env:
POLLINATIONS_API_KEY=pk_xxx_xxxxx

💡 pk_ keys are suitable for client-side apps (rate limit: 1 req/IP/hour).
🔑 sk_ keys are for server-side apps (no rate limits).

pip install -r requirements.txt

uvicorn main:app --host 0.0.0.0 --port 8000 --reload

Open http://localhost:8000 in your browser.

docker build -t world-generator .
docker run -p 8000:8000 -e POLLINATIONS_API_KEY=pk_xxx_xxxxx world-generator

1. Describe your world — vibe, aesthetics, inhabitants, rules
2. Choose the number of variants (1–4)
3. Click ⚡ Generate Worlds
4. See the results — world name, lore, and images
5. Click 🎬 Generate Video under any variant

Description:

A desert world where the sand is made of tiny mirrors. Two suns — blue and purple. Civilization lives underground, in crystal caves.

Result: world name, 3–4 paragraphs of lore, 4 different images of the world's scenes, and a video for each scene.

world-generator/
├── main.py              # FastAPI backend
├── static/
│   ├── index.html       # Frontend
│   ├── style.css        # Dark cosmic styles
│   └── app.js           # Frontend logic
├── requirements.txt
├── .env.example
├── .gitignore
├── Dockerfile
└── README.md

MIT