Downloads · 30 days
0
k4fka/autophotogrammetry-nordic
autophotogrammetry-nordic is a machine learning model from k4fka. Use it for the machine learning task on the model card, and read the license before you ship it in a product.
This repository contains a verified 3D Gaussian Splat PLY reconstructed from the Wikimedia Commons video "The Destroyer" in Gothenburg.
Downloads · 30 days
0
Access
Public
Updated Aug 21, 2026
Repo size
165 MB
Likes
0
Public
Click a slice to open those files.
.ply163 MB · 99%
From the Hugging Face model README
This repository contains a verified 3D Gaussian Splat PLY reconstructed from the Wikimedia Commons video "The Destroyer" in Gothenburg.
splat.ply — selected default Splatfacto export, 30,000 iterationspreview.png — one hold-out evaluation renderartifact-manifest.json — exact size, SHA-256, provenance, and metricsorientation-evidence.json — accepted orientation contract for the exact PLYevaluation.json — hold-out PSNR/SSIM/LPIPS evidencepreflight.json — source screening and download verificationThe selected default variant scored PSNR 15.6369, SSIM 0.4120, and LPIPS 0.6714 on the deterministic one-frame hold-out split. The render was also visually reviewed against the scale-regularized comparison; the default export preserved the harbor, ship, and waterfront structure with substantially less scattering.
The source download was independently verified at 677,541,013 bytes with
SHA-1 b05afd8dc11977bd42d704f24b25c0296e357ae0 and SHA-256
ceaf2d7852a4d1edf3c3fa09ac1fec775253f3c909855d118c705fe9dde3b54e.
Source: Wikimedia Commons — "The Destroyer" in Gothenburg
Author: Swedrone. The source is licensed under CC BY 3.0. Please retain this attribution when redistributing the derived artifact. The original video is not mirrored in this repository.
The reconstruction used the pinned Nerfstudio revision
50e0e3c70c775e89333256213363badbf074f29d, gsplat 1.4.0, and the local
autophotogrammetry:quality-ns50e0e3-sm86-local CUDA environment on an RTX 3060.
All exact hashes and paths needed to audit the result are in the accompanying
JSON evidence files.