SVBRDF Multi-view Consistency on 5 synthetic scenes (test)
0.056Basecolor ConsistencyUNet-HF
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| UNet-HFInput modalities=RGB, normals, and depth maps, Model Category=Regression2025.12 | 0.056 | 0.05 | 0.041 | 33,534 | |
| UNet-HFInput modalities=Training without normals and depth maps, Model Category=Regression2025.12 | 0.06 | 0.064 | 0.048 | 40,912 | |
| UNet-RGBInput modalities=RGB, normals, and depth maps, Model Category=Regression2025.12 | 0.064 | 0.078 | 0.06 | 361 | |
| GenPerceptInput modalities=RGB, normals, and depth maps, Model Category=Regression2025.12 | 0.072 | 0.104 | 0.069 | 694 | |
| Zeng et al.Model Category=Regression, Citation=[ZLH*22]2025.12 | 0.073 | 0.1 | 0.105 | 415 | |
| GenPerceptInput modalities=Training without normals and depth maps, Model Category=Regression2025.12 | 0.075 | 0.108 | 0.069 | 742 | |
| UNet-RGBInput modalities=Training without normals and depth maps, Model Category=Regression2025.12 | 0.089 | 0.117 | 0.077 | 362 | |
| Kocsis et al.Number of predictions averaged (xN)=10, Model Category=Generative2025.12 | 0.124 | 0.1 | 0.104 | — | |
| Zeng et al.Model Category=Generative, Citation=[ZDG*24]2025.12 | 0.155 | 0.247 | 0.542 | — | |
| Kocsis et al.Number of predictions averaged (xN)=1, Model Category=Generative2025.12 | 0.281 | 0.227 | 0.219 | 6 |