txt2img

Text-to-Image generation service using Diffusers + FastAPI.

🇯🇵 日本語 (Japanese) | 🤖 for Agents

Documentation

• API Guide: Endpoints usage and SSE details.
• Models & Config: Supported models and parameters.
• Development: Project structure and dev commands.

Quick Start (Local)

1. Setup Environment:

   cp .env.example .env
   vim .env  # Set CONFIG (e.g., config/chroma/flash)
   

2. Run:

   task up
   # or: docker compose up
   

3. Access:

   • Web UI: http://localhost:5173
   • API Docs: http://localhost:8000/docs

Cloud / RunPod

For cloud deployments (e.g., RunPod), use the pre-built Docker image.

• Image: hdae/txt2img:latest
• Environment Variables:
  • CONFIG: Model preset path (e.g., sdxl/illustrious).
  • CIVITAI_API_KEY: Required for downloading models.
  • HF_TOKEN: Required for HuggingFace gated models.
  • VRAM_PROFILE: full (recommended for 24GB+), balanced, or lowvram.

RunPod Template Example

docker run --rm -p 8000:8000 --gpus all \
  -e CONFIG=sdxl/illustrious \
  -e CIVITAI_API_KEY=your_key \
  hdae/txt2img:latest