Research map / Image generation / Text-to-image diffusion
ImagenWorld: Stress-Testing Image Generation Models with Explainable Human Evaluation on Open-ended Real-World Tasks
Samin Mahdizadeh Sani, Max Ku, Nima Jamali, Matina Mahdizadeh Sani, Paria Khoshtab, Wei-Chieh Sun, Parnian Fazel, Zhi Rui Tam, Thomas Chong, Edisy Kin Wai Chan, Donald Wai Tong Tsang, Chiao-Wei Hsu and 14 more
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| editing tasks vs generation tasks | unified models · ImagenWorld six tasks | overall human rating: about 0.1 lower on editing | Hurt |
| closed-source GPT-Image-1 vs Gemini 2.0 Flash | unified models · ImagenWorld | overall human rating: about 0.1-0.2 higher; TIE 0.82 vs 0.69 | Improved |
| targeted synthetic text-rich data curation (Qwen-Image) | Qwen-Image · ImagenWorld TIG textual graphics | human rating | Improved |
| symbolic and text-heavy topics vs artworks/photorealistic | all models · ImagenWorld topics | average human rating: about 0.78 (A,P) vs about 0.55 (S,I) | Hurt |
| autoregressive-diffusion hybrid architecture for editing | OmniGen2, Step1X-Edit · ImagenWorld editing | rate of regenerating new image: 17% new-image rate vs 0.6% IC-Edit | Hurt |
| VLM judge in place of human raters | Gemini-2.5-Flash · ImagenWorld | Kendall accuracy: VLM-human 0.78 vs human-human 0.76 overall; artifact 0.75 vs 0.76 | Mixed |
Closest papers by meaning
- ImagenHub: Standardizing the evaluation of conditional image generation models (ICLR 2024 · similarity 0.90)
- MMIG-Bench: Towards Comprehensive and Explainable Evaluation of Multi-Modal Image Generation Models (NeurIPS 2025 · similarity 0.89)
- Factuality Matters: When Image Generation and Editing Meet Structured Visuals (ICLR 2026 · similarity 0.88)
- OneIG-Bench: Omni-dimensional Nuanced Evaluation for Image Generation (NeurIPS 2025 · similarity 0.88)
- ImgEdit: A Unified Image Editing Dataset and Benchmark (NeurIPS 2025 · similarity 0.88)
- IDEA-Bench: How Far are Generative Models from Professional Designing? (CVPR 2025 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
