Text-to-Image Generation on InterActing (entire)
4.57Qwen2.5 ScoreDetailScribe
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| DetailScribeBase Model=Stable Diffusion 3.52025.04 | 4.57 | 1.46 | 92.3 | 38.1 | |
| DALL·E 32025.04 | 4.519 | 1.222 | 86 | 31.2 | |
| SDBase Model=Stable Diffusion 3.52025.04 | 4.228 | 1.323 | 90.2 | 33.6 | |
| SD + GPT RefineBase Model=Stable Diffusion 3.5, Modification=GPT Refine2025.04 | 4.141 | 1.255 | 88 | 30 | |
| SD + Inf ScaleBase Model=Stable Diffusion 3.5, Modification=Inference Scaling2025.04 | 4.126 | 1.354 | 92.2 | 36.5 | |
| SD + GPT RewriteBase Model=Stable Diffusion 3.5, Modification=GPT Rewrite2025.04 | 4.116 | 1.193 | 88 | 26.8 |