Be Yourself.

Bounded attention for multi-subject text-to-image generation.

Text-to-image diffusion models have an unprecedented ability to generate diverse and high-quality images.

---

Join the discussion on this paper page.