Local Image Generation API (Ollama-like)

A FastAPI-based REST API for local image generation using AI models such as Stable Diffusion 1.5 and FLUX. The project is inspired by Ollama, but focused on generating images instead of text.

---

✨ Features

• 🚀 Fast & Local — Run models entirely on your machine using FastAPI + Docker.
• 🔄 Dynamic Model Management — Load and unload models on demand via API endpoints.
• 🎨 Text-to-Image (txt2img) — Generate images with customizable parameters:
  • Prompt & Negative Prompt
  • Width & Height
  • Inference Steps
  • Guidance Scale
  • Seed for reproducibility
• 🔑 API Key Authentication — Protect your endpoints with simple header-based auth.
• ⚡ Job Queue System — Support for async or queued image generation.
• 📦 Model Management — Use local models or automatically download from Hugging Face.

---

🛠 Requirements

• Docker & Docker Compose
• NVIDIA GPU with CUDA support (recommended)
• Python 3.10+ (if running without Docker)

---

🚀 Quickstart

Clone the repo and set up the environment:

git clone https://github.com/your-username/image-generator-api.git
cd image-generator-api

cp .env.example .env
mkdir -p models outputs
docker compose build
docker compose up -d

Check health:

curl http://localhost:7878/health

---

📥 Model Management

Load a Model

curl -X POST http://localhost:7878/models/load -H "Content-Type: application/json" \
  -d '{"name":"sd15","local_dir":"/models/sd15"}'

Unload a Model

curl -X POST http://localhost:7878/models/unload -H "Content-Type: application/json" \
  -d '{"name":"sd15"}'

List Loaded Models

curl http://localhost:7878/models -H "X-API-Key: your_api_key"

---

🎨 Image Generation

Generate an Image

curl -X POST http://localhost:7878/generate/txt2img -H "Content-Type: application/json" \
  -d '{"prompt":"desert sunset, cinematic","model":"sd15","width":512,"height":512,"steps":20,"guidance":7.5}'

The API will return an image_url you can open in your browser.

Async Mode (Jobs)

You can generate in async mode:

curl -X POST "http://localhost:7878/generate/txt2img?async_mode=true" -H "Content-Type: application/json" \
  -d '{"prompt":"futuristic city, neon lights","model":"sd15"}'

Check job status:

curl http://localhost:7878/jobs/<job_id>

---

⚙️ Configuration

All configuration is handled via the .env file:

# Ports
HOST_PORT=7878
APP_PORT=7878

# Auth
API_KEY=change-me

# Runtime
device=cuda
MAX_WORKERS=1

# Model Repos
SD15_REPO=runwayml/stable-diffusion-v1-5
SD15_DIR=/models/sd15
FLUX_REPO=black-forest-labs/FLUX.1-schnell
FLUX_DIR=/models/flux-schnell

---

📦 Dependencies

Main Python dependencies:

• fastapi
• uvicorn
• pydantic
• transformers
• diffusers
• accelerate
• huggingface_hub

See requirements.txt for the full list.

---

📸 Example Output

Prompt	Generated Image
desert sunset, cinematic

---

🛡 Security

• Use API_KEY in .env to require X-API-Key header for all API requests.
• Example:

curl http://localhost:7878/models -H "X-API-Key: change-me"

---

🤝 Contributing

Contributions are welcome! Please fork this repo and submit a PR.

---

📄 License

MIT License © 2025