| license: mit | |
| pipeline_tag: image-to-3d | |
| # LVSPM | |
| [Paper](https://huggingface.co/papers/2610.10960) | [Project Page](https://burningdust21.github.io/Projects/LVSPM) | [Code](https://github.com/Burningdust21/LVSPM) | |
| Model checkpoints for novel view synthesis and camera pose estimation. | |
| | Checkpoint | Resolution | | |
| | --- | --- | | |
| | [LVSPM_448x256.ckpt](https://huggingface.co/LittleFrog/lvspm/resolve/main/LVSPM_448x256.ckpt) | 448 × 256 | | |
| | [LVSPM_512x288.ckpt](https://huggingface.co/LittleFrog/lvspm/resolve/main/LVSPM_512x288.ckpt) | 512 × 288 | | |
| For comparison with baselines, pose is evaluated at 512 × 288 and novel view synthesis at 448 × 256. | |
| Place the downloaded files in the LVSPM code repository's `checkpoints/` directory. Follow its README for installation, inference, data preparation and evaluation. |