chenxwh / fastcomposer

Tuning-Free Multi-Subject Image Generation with Localized Attention

  • Public
  • 33.9K runs
  • GitHub
  • Paper
  • License