Project report

»KOSMOS-3D in action: KulturGut Poggenhagen«

The primary goal of KOSMOS is to provide a cost-effective and easily accessible starting point for anyone wishing to capture Gaussian splats or engage in photogrammetry.

Project participants

Aram Badr Alexander Kutter Sam Wiemann
Prof. Vera Lossau
Kristina Herrmann

Project status

in progress

As part of a collaboration with Prof. Vera Lossau and Kristina Herrmann, our KOSMOS 3D scanner rig was deployed at KulturGut Poggenhagen last month to evaluate its performance in a complex spatial environment.

The main subject of the study was ‘Bovista’, a forest installation by Claas-H. Bläubaum. This site proved to be an ideal use case for capturing complex architectural forms that are seamlessly integrated into an organic environment.

Technical specifications

Hardware

Cameras

· 3 x GoPro Hero 13 Black Action-Cameras

Extension

· 2-metre-long painter’s pole/telescopic pole

Mounts

· Custom-designed and 3D-printed rig components

Devices

· Consumer-grade mobile devices

Software

Training Pipeline

· RealityScan by Epic Games for COLMAP data processing

Data processing

· LichtFeld Studio V0.5.2, developed by Janusch Patas

Control system

· A proprietary rig control application developed in-house, to be released shortly



Explore the scan

Further information on the scan route and data collection can be found in the description
Learn more

About Kosmos-3d

KOSMOS-3D is a research initiative led by Prof. Alexander Kutter, Aram Badr and Sam Wiemann from the KreativInstitut.OWL at the Ostwestfalen-Lippe University of Applied Sciences. The project builds on the pilot projects run by the Research & Transfer department at TH OWL.

About the pilot project



The primary goal of KOSMOS is to provide a cost-effective and easily accessible entry point for anyone wishing to capture Gaussian splats or engage in photogrammetry. The focus of our research is on Gaussian splatting.


We plan to release the scanner as an open-source project and to design it to be as modular and accessible as possible. By ‘accessible’, we mean not only the hardware and software, but also the data itself. Most commercial scanners currently restrict access to the data they generate. Whilst this is entirely sufficient for many applications (and we use such scanners ourselves), for research and educational purposes we require completely open access to the data.