which was released earlier this year — can generate an interactive, real-time scene from a single image and text description (e.g. “A cute humanoid robot in the woods”). In this way ...