Skip to main content
LivestreamHear more about how Niantic Spatial is working with NVIDIA and Flexion to close the Sim-to-Real gapWatch now
Scaniverse appCapture and reconstruct real places.PricingPlans for teams and enterprises.
Contact sales
Industries

Where people and machines work in the real world.

Field operator at a refineryOil and gasIntelligent field operations. Aerial view of a coastal site, measured from drone imageryDefense and intelligenceMission rehearsal and GPS-denied operations. Robots in a real-to-sim workshopEmbodied AIReal-to-sim training for robots.
Company

The team, partners and resources behind Niantic Spatial.

About
About usEventsCareers
Partners
Developer partner programAcademic and community partnersPartner with us
Guides and support
Scaniverse FAQNiantic Spatial SDK docsDeveloper forums
NewsResearch
A family of foundation models, trained on real-world data to establish ground truth for physical AI.
ReconstructionMetric-scale splats and aligned meshes of real places, from a single room to city-scale.LocalizationPrecise vision-based positioning, even where GPS doesn’t work.UnderstandingObjects, change over time and answers at every 3D point of a real place.
Inputs
CaptureFrom any sensor or archive: phone, 360°, LiDAR, drone, satellite and your own data.
Where people and machines work in the real world.
Oil and gasIntelligent field operations.Defense and intelligenceMission rehearsal and GPS-denied operations.Embodied AIReal-to-sim training for robots.
The team, partners and resources behind Niantic Spatial.
About
About usEventsCareers
Partners
Developer partner programAcademic and community partnersPartner with us
Guides and support
Scaniverse FAQNiantic Spatial SDK docsDeveloper forums
Scaniverse appCapture and reconstruct real places.PricingPlans for teams and enterprises.
Contact sales

Reconstruct any environment at metric scale.

Fuse data from phones, drones, LiDAR, and other ground-, air-, and space-based sensors into scalable 3D models that are geometrically accurate and georeferenced, not just photorealistic.

Contact salesTry Scaniverse

From input to output

GroundPhones, 360° cameras and LiDAR scanners
AirDrone and aerial imagery
OrbitSatellite imagery
Field platformsRobots and ROVs
ArchivesExisting scans, point clouds and imagery
Reconstruction
AccuracyMetric scale, georeferenced
CloudMulti-scan alignment, highest fidelity
On deviceSingle-scan splats and meshes in Scaniverse
CoverageFrom one room to an entire city
Gaussian splatsSPZ · PLY
MeshesFBX · GLB
Sim-ready scenesUSDZ · splat + collision mesh

Explore a real world environment in 3D

A high-fidelity reconstruction of Lands End as an interactive canvas where digital content is anchored to real-world locations within the splat.

Splat viewer demo

From multi-modal capture to one georeferenced 3D model

  1. Scan alignment
    Process images, video, LiDAR, and other sensor data from phones, drones, robots, and satellites, aligning scans across sources and refining their poses in real-world coordinates.
  2. Depth, semantics, and georeferencing
    Infer depth and semantic information, and anchor the reconstruction to real-world coordinates.
  3. High-fidelity 3D, ready for your workflow
    Create a unified, georeferenced digital twin with high-quality Gaussian splats and textured meshes, built for real-time rendering. Export in SPZ, PLY, FBX, GLB, or USDZ.

Reconstruct on device, in the cloud, or in your environment.

On deviceTurn a single scan into splats and meshes with Scaniverse on iOS and Android, with fast, private processing that can run fully offline, built for the field.
In the cloudBring multiple scans together into our highest-fidelity, georeferenced models of entire sites, built for multi-team projects.

Industries we support

Oil and GasTurn existing footage and new scans into measurable 3D models of facilities and subsea assets for inspection, maintenance, and change assessment.Explore oil and gas Defense and IntelligenceReconstruct operating environments at metric scale for mission planning, rehearsal, and simulation.Explore defense Embodied AIBuild simulation-ready scenes from captured sites, pairing Gaussian splats with aligned meshes for policy training and evaluation.Explore embodied AI

Start in Scaniverse

Scaniverse is the front door to our reconstruction pipeline. Bring in phone, 360°, drone or archived captures and get back georeferenced Gaussian splats and meshes.

Try Scaniverse

Talk to us

Piloting a site or scaling across many? We’ll scope reconstruction around your sensors and data, in the cloud or on-premises, delivered as splats, meshes and VPS maps.

Scope a pilot

Related

LocalizationPosition cameras, robots, and people within the reconstructed environment using visual localization even where GPS is unavailable or unreliable. A shared coordinate system connects navigation and field workflows to the same real-world site.UnderstandingQuery the reconstructed environment, identify assets, and compare captures to detect changes over time. Combine geometry with semantic information to generate site-specific answers and inspection reports.

Frequently Asked Questions

What data can I use to create a reconstruction?

Our Reconstruction pipeline works with images, video, and sensor data from commercial off-the-shelf devices including phones, drones, robots, 360° cameras, and LiDAR scanners as well as satellite imagery. It also supports challenging inputs such as archival footage and subsea video. Supported inputs depend on the workflow. For an entry-level, self-service option, Scaniverse supports mobile capture and uploads from supported 360° cameras.

Can I use a reconstruction for measurement and mapping?

Yes. Our Reconstruction pipeline produces geometrically accurate 3D models anchored to real-world coordinates not just photorealistic scenes. These models support measurement, mapping, infrastructure inspection, and comparison of the same location over time. Georeferencing also enables integration with GIS, simulation, and autonomy platforms.

What is the difference between a Gaussian splat and a textured mesh?

A Gaussian splat represents a scene using many small 3D elements that reproduce its visual appearance, making it useful for photorealistic exploration and visualization. A textured mesh represents the scene through connected geometric surfaces, supporting geometry-based workflows such as simulation and training. Available exports include SPZ, PLY, FBX, GLB, and USDZ, depending on the capture workflow and asset type.

Can I combine multiple scans into one reconstruction?

Yes. Our Reconstruction pipeline aligns and merges multiple capture passes into a unified, georeferenced 3D model. This includes stitching multiple drone scans into large-area reconstructions, with overlapping imagery helping establish alignment. The capture approach depends on the source data and project requirements. For an entry-level mobile workflow, Scaniverse lets multiple team members contribute scans from different devices to a shared site model.

Can reconstruction run offline, or does it require cloud processing?

We support both on-device and cloud processing. Cloud workflows support large-scale mapping, multi-user projects, and enterprise digital twins.