Datasets:
|
Download README.md from cstr/3dlf-scan: direct link, hf CLI and curl.
- Browser
- Download file 2.55 kB
-
https://huggingface.co/datasets/cstr/3dlf-scan/resolve/main/README.md
- Command line
-
hf download hf://datasets/cstr/3dlf-scan/README.md
-
curl -L -o README.md https://huggingface.co/datasets/cstr/3dlf-scan/resolve/main/README.md
2.55 kB
| license: cc-by-4.0 | |
| pretty_name: 3DLF-Scan (multi-sensor light-field and structured-light scans) | |
| task_categories: | |
| - image-to-3d | |
| - depth-estimation | |
| tags: | |
| - photogrammetry | |
| - light-field | |
| - structured-light | |
| - turntable | |
| - multi-view | |
| - 3d-reconstruction | |
| size_categories: | |
| - 10K<n<100K | |
| # 3DLF-Scan (full release, extracted) | |
| A re-host of **3DLF-Scan** version 1, extracted, so that individual files and objects can be | |
| downloaded without fetching the 9.8 GB archive. Seven 3D-printed figures (armadillo, asian_dragon, | |
| bunny, dragon, happy, lucy, thai_statue), each recorded in a full 360° turntable sweep by two | |
| calibrated light-field cameras (photonicSENS apiCAM PRO in `pro/`, apiCAM CUBE in `cube/`) and a | |
| Revopoint Miraco structured-light scanner (`revopoint/`): RGB images, raw and dense metric depth, | |
| per-view foreground masks, metric point clouds, poses, calibration and chessboard images | |
| (`calib_pro/`, `calib_cube/`), and fused structured-light meshes. | |
| For only the turntable photos of the PRO camera (about 100 MB per object), use | |
| [cstr/3dlf-scan-photos](https://huggingface.co/datasets/cstr/3dlf-scan-photos). | |
| `manifest.json` lists every file with its size and the CRC-32 recorded in the original archive. The | |
| archive itself was downloaded from Mendeley Data on 2026-10-07 and its SHA-256 equals the published | |
| value (`95ce4c76397c47bd8091fb85a529e38896284d163de5536fd27bdc2495d02f87`). | |
| **Not included:** the seven `<object>.stl` files at the root of the original archive. They are the | |
| canonical source models (Stanford 3D Scanning Repository; asian_dragon and thai_statue originally | |
| from XYZ RGB Inc.), which come with their own terms rather than CC BY; obtain them from | |
| <https://graphics.stanford.edu/data/3Dscanrep/>. Everything else is unchanged except that the | |
| archive's top folder `3DLF-Scan_dataset/` was dropped from the paths. | |
| ## Source and license | |
| **3DLF-Scan**, version 1, by Bohdan Vodianyk, Enrique Nava-Baro and Anton Popov. Mendeley Data, | |
| DOI [10.17632/ngvgpsvd8b.1](https://doi.org/10.17632/ngvgpsvd8b.1), | |
| <https://data.mendeley.com/datasets/ngvgpsvd8b/1>. Data article: Data in Brief, | |
| DOI [10.1016/j.dib.2026.112603](https://doi.org/10.1016/j.dib.2026.112603). | |
| Licensed [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/), as the original. Changes: the | |
| archive was extracted, the top folder removed from paths, and the canonical model STLs left out. | |
| The photographed and scanned objects are 3D prints of models from the Stanford 3D Scanning | |
| Repository, whose own terms apply to those models. | |