WorldGenBench: A World-Knowledge-Integrated Benchmark for Reasoning-Driven Text-to-Image Generation
WorldGenBench evaluates Text-to-Image models' world knowledge and reasoning by testing them across humanities and nature domains using a structured checklist score. The results show proprietary models like GPT-4o excel in complex reasoning, while open-source diffusion models are strong but lack deep contextual understanding, highlighting the need for knowledge-aware generation.