DTU -- scan1 (structured-light GT)

dtu_scan1 · DTU MVS 2014 · planned — not present locally yet

Ground-truth mesh

Not present locally. See the acquisition plan below.

Why this sample is in the benchmark

49-64 views with calibrated lighting and structured-light ground truth: high-precision geometry on a tabletop scene, and the standard MVS protocol that makes results comparable to the MVS literature.

Ground truth

GT type

laser

licence

Research

source

DTU MVS 2014

builder

- {'scene': 'scan1'}

views

49 × 1200px (orbit, fov 50°)

lighting

calibrated

Acquisition plan

kind script, automatic False, estimated size 900 MiB

1. Dataset home
   https://roboimagedata.compute.dtu.dk/?page_id=36
2. Fetch one scan (sample set)

   bash -c 'mkdir -p dtu/scan1 && wget -r -np -nH --cut-dirs=4 -R index.html* https://roboimagedata2.compute.dtu.dk/data/MVS/SampleSet/MVS%20Data/scan1/'

DTU MVS 2014: 124 robot-mounted scenes with STRUCTURED-LIGHT ground truth (high-precision point clouds), 49-64 views each with known poses and calibrated lighting. The standard MVS geometry benchmark; its 22-scan evaluation protocol is directly reusable.

Comparisons

No reconstruction compared yet. Run geombench compare dtu_scan1 --mock or point it at a real 3DGS output.