Diffuse Albedo Estimation on FFHQ-UV-Intrinsics
2.93MSERefinementNet
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| RefinementNetStage=Stage II, Base Model=MAGINet2025.12 | 2.93 | 88.1 | 0.137 | 33.4 | |
| MAGINetStage=Stage I, Input Resolution=512x5122025.12 | 3.05 | 87.5 | 0.142 | 35.9 | |
| GAN2XModel Type=GAN-prior inverse rendering2025.12 | 3.48 | 85 | 0.168 | 47.6 | |
| InverseFaceNetModel Type=3DMM + SH2025.12 | 3.65 | 84.2 | 0.184 | 51.7 | |
| SfSNetModel Type=Lambertian, SH lighting2025.12 | 3.76 | 83.5 | 0.19 | 54.1 | |
| U-Net-6LDescription=Plain U-Net-6L matching encoder/decoder depth2025.12 | 4.12 | 82.1 | 0.201 | 58.3 |