Lifelike 3D Human Generation On Thuman2 0
Metrics
CLIP Similarity
LPIPS
PSNR
SSIM
Results
Performance results of various models on this benchmark
Model Name | CLIP Similarity | LPIPS | PSNR | SSIM | Paper Title | Repository |
|---|---|---|---|---|---|---|
| SiTH | 0.8978 | 0.1396 | 17.0533 | 0.8963 | SiTH: Single-view Textured Human Reconstruction with Image-Conditioned Diffusion | |
| Human-VDM | 0.9235 | 0.0957 | 20.068 | 0.9228 | Human-VDM: Learning Single-Image 3D Human Gaussian Splatting from Video Diffusion Models | |
| SIFU | 0.8663 | 0.1500 | 16.4600 | 0.7931 | SIFU: Side-view Conditioned Implicit Function for Real-world Usable Clothed Human Reconstruction | |
| PaMIR | 0.8861 | 0.1461 | 16.6267 | 0.8924 | PaMIR: Parametric Model-Conditioned Implicit Representation for Image-based Human Reconstruction | |
| Ultraman | 0.9131 | 0.1338 | 17.4877 | 0.8958 | Ultraman: Single Image 3D Human Reconstruction with Ultra Speed and Detail | |
| PIFu | 0.8501 | 0.1615 | 15.0248 | 0.8884 | PIFu: Pixel-Aligned Implicit Function for High-Resolution Clothed Human Digitization |
0 of 6 row(s) selected.