Features
3D Gaussian Splatting
Accurate and efficient 2D and 3D reconstruction
LiDAR Data Processing
Processes data collected by DJI LiDAR
High-Fidelity Models
High-accuracy 1:500 mapping without GCPs
Stunning, Photo-Quality 3D Models
Quickly generates detailed and realistic 3D models from photos, accurately restoring cultural relics, architectural projects, urban blocks, and scenic gardens. Significantly enhances representation of fine structures, transparent and reflective objects, and vegetation compared to traditional photogrammetry.
Twice as Efficient for Large-Scale Scenes
Processes ~500 photos per hour, completing single-block modeling in 30-40 minutes. For 10,000+ photo scenes, reconstruction speed is twice that of Terra Mesh. City-scale reconstruction supports up to 30,000 photos per task with only 4 GB GPU memory and 32 GB RAM required.
City-Scale Reconstruction
Up to 30,000 photos per single task
Meets urban, industrial park, and infrastructure needs
Entry-Level Hardware
32 GB RAM for smooth operation
No high-performance computer needed
Universal Data Formats
Seamless integration with third-party platforms
Surveying, gaming, and film production ready
Distortion-Free Output
Distortion-free orthomosaics
Further improved result quality
High-Fidelity 2D/3D Models
Rapidly generates high-precision 3D models, point clouds, TDOM, and DSM from photos. Adjustable mesh count and resolution. Export to various industry-standard formats.
High-Accuracy 1:500 Mapping
Centimeter-level accuracy without ground control points via RTK-integrated DJI drones. Dedicated accuracy enhancements for Matrice 4E. Consistently meets 1:500 standards in complex environments.
Fast, On-Site Results
Standalone: 30,000 photos in 3 days, 5,000 photos in under 2 hours (2D). Cluster: 30,000 photos in 21 hours (3D), 5,000 photos in 80 minutes (2D). Real-time results for smaller projects.
LiDAR Reconstruction
Deep LiDAR-visual data fusion. Single acquisition generates point clouds, 2D maps, and 3D models. Centimeter-level accuracy via RTK/PPK without GCPs. Handles ~1 GB point cloud per 1 GB RAM.
Multispectral Reconstruction for Agriculture
Generate NDVI and NDRE vegetation index maps from Mavic 3 Multispectral and Phantom 4 Multispectral data. Create prescription maps for DJI Agras drones with variable-rate spraying. Supports radiometric correction and reflectance maps for remote sensing research.
Cluster Reconstruction — Superior Efficiency
Ultra-large-scale mapping with multi-GPU and multi-node collaboration. Processes 6,000 photos with just 1 GB RAM. Five worker devices complete 30,000 photos in 21 hours. Flexibly scale as projects grow.
Hardware Synergy
Matrice 4E, Matrice 400, Zenmuse P1, L2, D-RTK 3
Cost-effective visible light & LiDAR solutions
Terra & Modify Workflow
Point cloud smoothing, noise reduction, classification
Comprehensive solution: survey → model → edit → share
PPK Calculation
Local and cloud-based PPK support
Centimeter-level accuracy without RTK
Control Points & Check Points
Semi-automatic picking with sub-pixel accuracy
Verify reconstruction accuracy at any stage
Comprehensive Coordinate Systems
8,500+ built-in coordinate systems. Supports local systems, seven-parameter transformations, PRJ file imports. Intelligently recommends optimal system based on user location.
Annotation & Measurement
Accurate measurement of points, lines, areas, and volumes. Click 3D model to view camera position and original images. Smooth model-to-site switching for efficient inspection.
Rich Usage Scenarios
Auto-generates waypoints from 3D model or point cloud. Simulated camera view for precise waypoint selection. Supports Matrice 4T/4TD for detailed inspections across industries.





