MultiSet AI Launches World’s First 3DGS-to-VPS Pipeline, Advancing True Scan-Agnostic Spatial Computing

SAN FRANCISCO, CA – MultiSet AI has launched the world's first 3D Gaussian Splatting (3DGS) to Visual Positioning System (VPS) integration. By enabling the upload of metric-scaled Gaussian Splats, MultiSet AI extends its foundational spatial computing infrastructure to support the rapid market adoption of explicit radiance fields, delivering centimeter-accurate, cross-platform AR maps with zero hardware lock-in.

Market Context and the Acceleration of 3DGS Adoption

‍3D Gaussian Splatting has officially shifted from an experimental research novelty to an enterprise rendering standard. With the 3DGS tools market projected to reach $7.19 billion by 2033, and authoritative consortiums like the Khronos Group advancing the KHR_gaussian_splatting glTF standard into its final release candidate stages in early 2026, industry adoption is accelerating exponentially. Enterprises are aggressively deploying 3DGS to bypass the massive computational overhead of Neural Radiance Fields (NeRFs) and the manual topology constraints of traditional mesh workflows. MultiSet’s new integration capitalizes on this momentum, providing the critical missing link: functional, real-world spatial localization from 3DGS assets.

Additive Capabilities: Expanding a Robust Geometric Foundation

‍This release represents a major expansion of MultiSet AI’s scan-agnostic capabilities. Building upon our robust, highly accurate localization pipeline for traditional geometric point clouds (E57, LAS) and triangulated meshes, the platform now seamlessly ingests explicit 3DGS primitives (.PLY).

Rather than replacing existing workflows, the integration of 3DGS augments MultiSet's proven spatial mapping infrastructure by natively combining extreme geometric accuracy with photorealistic fidelity. This unlocks new operational advantages for enterprises:

• Unified Digital Twins: Deploy a single, photorealistic spatial asset that simultaneously powers human-wearable AR navigation and autonomous robotics localization.
• High-Fidelity Experiential AR: Anchor persistent, centimeter-accurate digital overlays in retail or entertainment spaces using maps that perfectly mirror real-world visual conditions, including reflective and translucent surfaces that historically disrupt LiDAR.
• Large-Scale Campus Mapping: Utilize MultiSet’s MapSet feature to stitch multiple 3DGS captures across sprawling environments. The system natively mixes new .PLY files with legacy LiDAR point clouds, ensuring comprehensive coverage without creating isolated map islands.

Hardware Synergy: XGRIDS and Metric Scale

‍While universally hardware-agnostic, the pipeline officially launches with optimized support for the XGRIDS hardware ecosystem, including the PortalCam and Lixel K1/L2 Pro. By utilizing "Portability" mode during XGRIDS reconstruction, operators generate a strictly metric-scaled .ply output where one digital unit equals one physical meter. This deterministic scaling is critical for absolute real-world alignment and drift prevention.

The Technical Workflow: Reconstruct, Upload, Localize

‍The developer experience remains tightly integrated and friction-free:

1. Reconstruct: Process environmental captures into metric-scaled representations.
2. Upload: Bundle the metric-scaled .ply file and corresponding poses.json (containing timestamp, translation, and rotation camera trajectory data) into a single .zip archive via the MultiSet Developer Portal.
3. Localize: MultiSet’s Vision Fusion pipeline processes the payload, generating a highly compressed VPS map capable of resolving camera pose in under 1500 milliseconds.

Future-Proofing with Open Standards

‍MultiSet aligns natively with the trajectory of the spatial web. The platform fully supports impending industry standards, including the aforementioned Khronos glTF extension and the open-source SPZ compression format. Maps are immediately deployable across MultiSet's cross-platform SDKs, supporting Unity, WebXR, iOS, Android, Meta Quest, and ROS 2.

Get Started

• Review the Documentation: MultiSet 3DGS Upload Specifications
• Hardware Integration: XGRIDS Gaussian Splat Guide
• Start Building: Access the free tier at multiset.ai.

About MultiSet AI

‍MultiSet AI provides the foundational spatial intelligence infrastructure for enterprise augmented reality. By delivering the world's most robust scan-agnostic Visual Positioning System, MultiSet enables organizations to build, self-host, and scale persistent AR experiences with uncompromising accuracy and full data sovereignty.