Most current research on automatically captioning and describing scenes with spatial content focuses on images. We outline that generating descriptive text for a synthesized 3D scene can be achieved via a suitable intermediate ...