Stonex has officially stepped into the world of radiance field representations with the release of GOpost v2.4.3, introducing support for 3D Gaussian Splatting (3DGS) across its mapping and reality capture pipeline.
Stonex is now leveraging radiance fields to generate lifelike 3D content by combining panoramic video, SLAM, and LiDAR to create visually rich and spatially accurate environments. Much like XGRIDS, they ground their reconstructions with SLAM and LiDAR, but the actual imagery is captured using an external 360° camera.
Currently, the company now supports a dedicated 3DGS workflow using the Insta360 X4 and X5, with direct import into GOpost, without needing InstaStudio. The recommended capture settings involve recording in 5.7K, 30fps, using the “Timeshift” mode at 10x speed, with auto exposure or manually tuned ISO. Stonex encourages fixed ISO settings (e.g. ISO 100 for outdoor daylight, ISO 800 for dim interiors) and a focus on consistent motion to support downstream SLAM stability.
Once captured, the imagery and trajectory are processed inside GOpost, where a new “3DGS PreTraining” option initiates the splat pipeline. After specifying the data directory, users can launch pre-processing directly in the interface. Currently StoneX recommends using Postshot for the actual reconstruction portion, using the camera poses created by Stonex.
The Stonex team has made a couple of demo captures available here. There has been no pricing information made publicly available about Stonex devices yet.
This marks the latest entrant into a rapidly evolving space where camera design, capture strategy, and neural rendering are converging to make lifelike 3D reconstruction more accessible. Learn more at stonex.it, or watch their full tutorial on capturing with Insta360 and PostShot to see how this comes together in practice.
New to Gaussian Splatting? Start here
