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.

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://data.mendeley.com/datasets/ngvgpsvd8b/1. Data article: Data in Brief, DOI 10.1016/j.dib.2026.112603.

Licensed CC 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.