Depth-conditioned Image Generation on Conditional Generation Benchmark SD3-based
29.27G-CLIPSD3 + ControlNet
Evaluation results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| SD3 + ControlNetInput=Depth2026.05 | 29.27 | 24.25 | 10.28 | 31.55 | 33.28 | 31.44 | — | 5.502 | 40.33 | |
| COLLARInput=Depth2026.05 | 29.11 | 25.02 | 27.3 | 57.01 | 52.54 | 57.95 | 18.66 | 5.435 | 35.37 | |
| DreamRendererInput=Depth2026.05 | 28.82 | 24.63 | 17.44 | 45.72 | 43.78 | 46.34 | 21.53 | 3.849 | 24.29 |