| Full name | Decoder-only View Synthesis Model Done Right |
|---|---|
| Publication title | Decoder-only View Synthesis Model Done Right |
| Publication authors | Cheng Sun; Jaesung Choe; Min-Hung Chen; Ryo Hachiuma; Yu-Chiang Frank Wang |
| Publication venue | |
| Publication url | https://arxiv.org/abs/2605.29891 |
| Project link | https://github.com/NVlabs/DVSM |
| Hardware | Training 64 A100 GPUs; Inference 1 A100 GPU. |
| Code | Code available with license: NVIDIA License |
| Use other data | Yes: DL3DV-10K |
| Last edited | 01 June, 2026 |
| PSNR | SSIM | LPIPS | |
|---|---|---|---|
| Original | 19.1502 | 0.8696 | 0.3211 |
| Color-corrected | 22.7158 | 0.8790 | 0.2748 |