The aim of XRTwinScape is to easily digitize a full environment by simply uploading a video. This project uses the AI based technology of Gaussian Splatting to recreate an existing environment in high fidelity and provides a fully web-based application to manage the creation and publication of the desired reconstructions.
The full Gaussian Splatting pipeline developed consists of Mast3r for the SfM part and Nerfstudio for the reconstruction training.
The application provides both an editor for modifying and annotating the reconstruction, and a VR and web-based viewer for reconstructions exploration.
This XRTwinScape project has been funded through the XR2Learn Open Call 2.
XR2Learn is a Horizon Europe-funded initiative designed to foster the cross-border creation of human-centric extended reality (XR) applications for education. At its core, it will deliver a one-stop-shop Digital Innovation Hub platform that brings together XR technology providers, application designers, educators, developers, end-users, and decision-makers. This platform will streamline communication, collaboration, and matchmaking across the XR-based educational supply chain, with a special focus on immersive training in manufacturing and distance-learning contexts.
The only requisite for running this project is an installation of Docker to run the containers.
To run this project just run the following command in the twin_scape folder
docker compose up -d
This work make use of the Mast3r project, Mast3r is distributed under the CC BY-NC-SA 4.0 License.