Text-to-3D Human-Object Interaction Generation on T3Bench 100 prompts
31.4CLIP ScoreTHOM
Evaluation results
| Method | Links | ||
|---|---|---|---|
| THOMTime=2 hour 30 min.2026.04 | 31.4 | 2.6 | |
| ProlificDreamerTime=3 hours2026.04 | 29.1 | 1.2 | |
| GaussianDreamerProTime=1 hour 30 min.2026.04 | 28.5 | 1.4 | |
| DreamHOITime=3 hour 20 min., Adapted for hand-object interaction=true2026.04 | 27.8 | 1.6 | |
| Hash3DTime=25 min.2026.04 | 27.6 | 1.9 | |
| DreamFusionTime=35 min.2026.04 | 27.1 | 1.3 | |
| InterFusionTime=2 hour, Adapted for hand-object interaction=true2026.04 | 22.5 | 1.5 |