Everything in Its Place: Benchmarking Spatial Intelligence of Text-to-Image Models - Explained Simply | ArXiv Explained