TerraLab Digital Twin
We built a 3D engine that turns geospatial data into editable 3D environments at any scale. We use it to help you visualize and present your projects.
Real-time 3D visualization from real Earth data
Scene Composition
DEM → PBR Lighting → 3D Vectors → Orthophotos Fusion → 3D Buildings & Vegetation
TerraLab Digital Twin
TerraLab Digital Twin is a 3D engine built from scratch in C++. It takes real geospatial data, satellite imagery, DEM, vectors, and turns it into a 3D environment you can explore and edit in real-time.
Satellite photos work well from above, but the closer you get to the ground, the more they break apart. Details disappear, textures become unreadable. Procedural generation, guided by actual geodata, reconstructs what the images can't provide: coherent terrain, surfaces, and structures at every scale.
The engine uses a quadtree architecture that subdivides the planet into tiles, streaming only what's needed for your current view. You can navigate from a global overview down to centimeter precision, reshape terrain, place infrastructure, and test different scenarios, all rendered in real-time on GPU.
Geospatial Data Mastery
We master Earth data before we talk about 3D
Multi-Source Data Harmonization
Radiometric normalization across sensors and seasons, cloud masking, gap filling, and cross-tile brightness balancing. Produces seamless mosaics ready for visualization and analysis.
Spatial Tiling & Real-Time Streaming
At the core of TerraLab: a quadtree spatial subdivision that structures raster and vector data into consistent tiles. Compressed by region and versioned over time, they feed 3D environments in real-time via our WorldMapServer, up to planetary scale.
Geospatial AI
Geospatial AI
AI-driven segmentation and land cover classification powered by geospatial foundation models. Automatically enriches Digital Twins with semantic layers: vegetation, hydrology, urban fabric, terrain features.
Real-Time 3D Technology
GPU-powered visualization with procedural enhancement
Planetary Scale to Centimeter Precision
Multi-resolution quadtree architecture with GPU-native rendering. Navigate from planetary overview to centimeter-level detail with no loading and no interruption.
Procedural Refinement
Linear → BiCubic → Dilatation → Erosion
Procedural Terrain Enhancement
DEM-driven terrain with runtime procedural refinement at every scale, from global landscape shaping to ground-level detail. Interpolation, erosion, and texture generation are guided by vector rules and polygon markers, producing PBR-ready albedo maps and textures. Terrain can be reshaped interactively in real-time.
Real-Time Data Fusion
Vectors → DEM → Normals → Slopes → Orthophotos → Final Fusion
Real-Time GIS Data Integration & Fusion
Orthophotos, DEM, normals, slopes, vector networks, and cartographic overlays combined into a unified 3D environment. Each data layer streams in real-time and updates as your project evolves.
Scene Composition
DEM → PBR Lighting → 3D Vectors → Orthophotos Fusion → 3D Buildings & Vegetation
Complete Scene Generation
From raw GIS data to fully populated 3D environments: terrain, PBR lighting, orthophoto fusion, 3D buildings, vegetation, and vector networks. All assembled and rendered in real-time.
From Project to Decision
Presentation and decision support for real projects
Aerial 3D Visualization
3D Project Visualization
Explore your own vector and raster data in real-time. Adjust the rendering style, from simplified to realistic, to match your audience and presentation context.
Fly Over, Build
Interactive 3D Editing
Edit roads, zones, and infrastructure directly in their 3D spatial context. Modify terrain, place objects, and iterate on designs with instant visual feedback.
Coming soon
Spatial Analysis for Decision-Making
Built on real-time data fusion, the engine supports interactive simulation with physics models: hydrology, viewshed computation, and environmental impact assessment directly within your 3D project environment.
FAQ
Common questions about TerraLab Digital Twin.
Contact us
Or email me directly: yvann.barbot@terra-lab.ai