Text-to-Image Generation on InterActing Functional Relation scenario, sampled
4.28Human Likert ScaleDetailScribe
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| DetailScribeBackbone=Stable Diffusion 3.52025.04 | 4.28 | 4.88 | 1.761 | 99.8 | 44.9 | |
| DALL·E 32025.04 | 3.94 | 4.68 | 1.535 | 88 | 22.6 | |
| SD + GPT RewriteBackbone=Stable Diffusion 3.5, Prompt Strategy=GPT Rewrite2025.04 | 3.77 | 4.44 | 1.546 | 92.1 | 24.5 | |
| SD + GPT RefineBackbone=Stable Diffusion 3.5, Prompt Strategy=GPT Refine2025.04 | 3.45 | 4.6 | 1.524 | 95.1 | 29 | |
| SDBackbone=Stable Diffusion 3.52025.04 | 3.36 | 4.48 | 1.657 | 97.1 | 36.6 | |
| SD + Inf ScaleBackbone=Stable Diffusion 3.5, Strategy=Inference Scaling2025.04 | 3.27 | 4.56 | 1.718 | 98.5 | 43.4 |