Aligning Realities: A Registration Pipeline for Arbitrary Objects in Mixed Reality Using Controller-Based Point Selection
David Mertens, Steffen Sascha Stein, Kristoffer Waldow, Arnulph Fuhrmann
In: Proceedings of 33th IEEE Virtual Reality Conference (VR ’26)
Abstract
Photorealistic human representation is a key factor for embodiment and presence in Virtual and Social Virtual Reality. While recent approaches such as 3D Gaussian Splatting enable high-quality re-constructions, dynamic human capture remains challenging due to the need for synchronized multi-view data. Existing capture sys-tems are often costly and inaccessible, limiting reproducibility and dataset availability. We present a low-cost, reproducible multi-view capture rig consisting of 68 synchronized Raspberry Pi cameras arranged for full 360° coverage. The system enables reliable acquisition of dynamic human data suitable for 3D Gaussian Splatting and VR applications.
