[SIGGRAPH ASIA 2024] Frankenstein: Generating Semantic-Compositional 3D Scenes in One Tri-Plane - View it on GitHub