3D Map Explorer turns any Windows workstation into a complete, offline geospatial hub - 3D visualization, true 3D / Z editing over GeoPackage, and a full GEOINT workstation: stereo photogrammetry, SAR/InSAR, hyperspectral & target detection, IMINT exploitation, TAK/C2, multi-INT sensor fusion, local AI, RF sensing, offline routing - and a complete UAS / drone loop from preflight Go/No-Go to oriented-imagery exploitation. No cloud. No ArcGIS license. No infrastructure.
Valuable geospatial data is trapped on individual workstations. Enterprise GIS servers cost a fortune and take weeks to stand up. Cloud tools raise data-sovereignty concerns. Field teams go dark the moment the network drops.
Enterprise GIS stacks demand dedicated hardware, licenses, and a half-FTE just to keep the lights on.
Cloud sync is a non-starter for classified, regulated, or commercially sensitive geospatial data.
Field, tactical, and remote teams lose every collaborative capability the moment the uplink fails.
An embedded high-performance geospatial server inside the desktop: 30+ formats and the databases you already run go in, open standards come out - OGC, Esri REST, I3S, STAC, vector tiles, streams. Nothing leaves your network.
Full technical architecture with request flows and data pathways in ARCHITECTURE.md
With its embedded local server, the desktop hosts, serves, edits, analyzes, and shares geospatial data - fully offline if required.
True 3D (Z + M) editing over GeoPackage through an ESRI-compatible FeatureServer and the native Editor widget - round-trips elevation exactly.
Deep-dive →One click turns the workstation into a LAN-wide OGC / ESRI / 3D-Tiles endpoint. Zero config, zero infrastructure.
How it works →I3S SceneLayers from .slpk, Gaussian splats, OGC 3D Tiles, LiDAR point clouds, and streaming Cloud-Optimized GeoTIFF.
See the stack →Shadow & solar potential, viewshed, slope/aspect, routing, spatial statistics - with live 3D visualization.
Explore tools →USGS quakes, NOAA weather, AIS vessels, FAA aircraft, Meshtastic mesh, TAK/CoT, APRS, and IoT sensors.
Live data →GeoPackage, GeoJSON, ESRI Geodatabase, Shapefile, KML, PMTiles, FlatGeobuf, COG - read, convert, serve.
Full format list →Beyond the six pillars - search public imagery catalogs, map your existing databases, fly drone video geo-locked to the terrain, import CAD, stream live feeds straight into ArcGIS, and serve terrain & imagery to Cesium, Unreal, Unity, and Godot with no Cesium ion or ArcGIS Online dependency.
Search Element 84 & Planetary Computer from one box, and publish your own rasters and point clouds as a spec-conformant STAC API 1.0.0.
Search & publish →Postgres, MySQL, SQLite, SQL Server, BigQuery, Google Sheets - spatial tables auto-discovered and served as MVT, OGC Features, and FeatureServer.
Map your data →STANAG/MISB KLV playback with a live sensor footprint, DJI→STANAG conversion, and optional YOLO-E detection - RTSP/UDP/SRT supported.
Geo-locked video →DWG / DXF / DGN onto the map via control-point or source-EPSG georeferencing, served as GeoJSON, MBTiles-vector, or GeoPackage.
Import drawings →Subscribe to SensorThings, SOS, GTFS-rt, and Moving Features into DuckDB - then expose moving data as a native Esri WebSocket StreamLayer.
Live streaming →Air-gapped terrain & imagery for Cesium, ArcGIS Maps SDK for Unreal/Unity, and Godot - quantized-mesh, WMTS, TMS XML, and LERC. For mission-planning & simulation enclaves without Cesium ion or ArcGIS Online.
Engine integrations →The same geoprocessing and serving engine as a back-end Linux service - DEB/RPM packages, a Helm chart, and Docker images. No desktop and no bundled viewer: host your own web map from a folder, or pair it with MapLibre GL Native on QNX or hardened Linux.
Deploy headless →The full feature hub - the six pillars plus everything above, with deep-dive pages and the promo deck.
All features →The newest releases turn the desktop into a full intelligence platform: tactical comms, radar and SAR exploitation, imagery intelligence, local AI, RF sensing, offline routing, GIS-server emulation - plus SharePoint lists as editable layers, commercial satellite tasking, LoD2.2 city reconstruction, and a plugin SDK.
TAK Server + air-gapped multicast, blue-force overlay, GeoChat, Mission Board kanban with AAR replay, geofencing, MIL-STD-2525.
Tactical picture →ASTERIX CAT-048 tracks with a live Track Table, CPA calculator, velocity vectors, and 7500/7600/7700 squawk alerting.
Track the picture →SAR amplitude viewing with speckle filtering, spectral indices, YOLO-E detection, GEOINT annotation, and RPC mensuration.
Exploit imagery →Gemma 3 / Phi-4 chatbot + NL→SQL, SAM2 segmentation, YOLO-E detection - all on your hardware, no cloud AI.
On-prem intelligence →Resolved-canonical-entity graph fuses every feed into one typed object model, exposed as MCP tools an AI agent calls directly - resolve_entity, find_paths, get_neighbors - over Streamable HTTP.
Agent-callable COP →Predict where a resolved entity will be in N minutes (constant-velocity + least-squares), then task a DJI drone to orbit the predicted position - the full detect → resolve → predict → task → collect loop, all air-gapped.
Close the loop →Plan UAS surveys on the map - FlyPath turns a drawn polygon into an optimized flight grid with GSD, photo-count, and battery estimates, exports DJI WPML .kmz, and checks FAA airspace (Class B/C/D, SUA, TFRs, UASFM grids).
Plan the flight →Connect a PX4 or ArduPilot autopilot over MAVLink (UDP/TCP) and watch the drone fly as a live green track - position, battery, flight mode, GPS health, armed state - shared to the TAK team as a friendly air track.
Watch it fly →Supervised & unsupervised imagery classification - SAM, Min-Distance, Maximum Likelihood, K-means, plus RF/SVM - trained from ROIs or GPS points, with a full confusion-matrix accuracy assessment.
Classify the scene →ADS-B, RemoteID drones, AIS, APRS, scanner audio, and rtl_433 IoT - the RF spectrum fused onto one map.
Listen in →Air-gapped A* routing over your own road graph, multi-waypoint turn-by-turn, GPX/KML export, graph analysis.
Plan the route →Any SharePoint list as an editable map layer - Entra ID sign-in, five geometry shapes auto-detected, edits flow back via Graph.
Map the list →Search, task, and order from Planet, Sentinel Hub, Skywatch, and SkyFi - your accounts, your imagery, straight to the map.
Task the bird →LoD2.2 building reconstruction, splat training, forestry LiDAR - and a drop-in plugin platform to extend it all.
Rebuild & extend →Beyond GEOINT, the platform now runs the entire drone loop on one desktop - preflight Go/No-Go, live gimbal payload control, RTK/PPK corrections, BVLOS awareness, counter-UAS detection, flight-log replay, fleet management, orthomosaic/SfM processing, and oriented-imagery exploitation. Plan it, fly it, process it, exploit it - without leaving the workstation.
Controlled-airspace overlay, TFRs, a METAR wind gate, mission-footprint intersection, and a checklist → Go/No-Go verdict. LAANC-aware.
Get the verdict →Live gimbal attitude, RGB/thermal switching, and zoom - the platform's first outbound MAVLink command-send path.
Drive the sensor →NTRIP → RTCM → MAVLink relay for live RTK, plus PPK trajectory post-processing against base-station RINEX.
Centimeter accuracy →Fused traffic picture (Remote ID + MAVLink + ADS-B), terrain-sampled comm-link line-of-sight, and deconfliction alerts.
See the picture →Radar-webhook ingest, conservative threat assessment, protected-asset zones, and C-UAS ↔ Remote ID de-escalation.
Defend the asset →Ingest ArduPilot .tlog/.bin and DJI .csv logs; animate the trajectory and run post-flight analysis. Feeds PPK + fleet.
Replay the flight →Drone registry, flight-hour aggregation, maintenance-due tracking, and per-battery cycle logging - the airworthiness ledger.
Manage the fleet →Local OpenDroneMap CLI sidecar for air-gapped photogrammetry, plus WebODM/NodeODM cloud import.
Process the sortie →Exploit drone photos by capture pose - auto-derived from FMV frames or EXIF, served as an ArcGIS Feature Service.
Exploit the photos →The headline capability of this release: a real, end-to-end 3D (Z + M) editing pipeline over GeoPackage, wired into the running app and the ArcGIS Editor widget.
hasZ / hasM read from gpkg_geometry_columns.edit3d.gpkg (POINT Z, POLYGON Z) on first run.// Z-enabled feature, absolute height mode { "attributes": { "name": "Tower-A" }, "geometry": { "x": -74.006, "y": 40.712, "z": 1500.5 } } → stored as POINT Z · hasZ: true · z preserved exactly
Drop a file - or point at a database - and it's on the map and on the wire. One process, zero installs, ETag-cached, LAN-wide.
* Capability counts reflect the documented platform scope. ROI and TCO figures appear on the ROI page and are illustrative planning targets.
The capabilities that close the gaps to the dedicated exploitation workstations and fusion platforms - Textron RemoteView, ArcGIS AllSource, L3Harris Excalibur, and Hexagon Luciad. Stereo → DSM, multi-INT fusion with NATO-standard export, virtual mosaics, full image classification, annotated GEOINT reporting, tip-and-cue, dynamic projection, business data, and interactive charts.
Air-gapped COP/SA/C3 for maritime: S-57/S-100 ENC charting, satellite & terrestrial AIS fusion, and a benchmarked dark-vessel detection chain - detect → associate → fuse → adjudicate.
Explore maritime COP →RPC triangulation + dense matching turns a stereo pair into a geocoded DSM, dual-registered as a COG and a terrain source. The RemoteView "3D Pro" parity.
Extract 3D terrain →One COP from ADS-B, AIS, ASTERIX, SDR, RemoteID & TAK. Export track history as STANAG 4676 TrackImageFile XML - the NATO JISR exchange format.
Fuse tactical tracks →N COGs composed into one seamless layer via a lightweight VRT - no pixel copy, no preprocessing. TB-scale imagery on demand.
Stream virtual VRTs →SAM, MaxLik, MinDist, K-means (air-gapped) + RF/SVM/XGBoost. Accuracy assessment with confusion matrix + Kappa, change detection, OBIA, radiometric calibration.
Classify rasters →Annotated map → editable PowerPoint, KMZ/KML, PDF, or GeoTIFF. The RemoteView annotated-reports deliverable in one route.
Generate map decks →FMV AI detection → SAR/EO re-image tasking with a human approval gate. The wide search finds it; the hi-res confirms it.
Automate tip-and-cue →Polar Stereographic, Gnomonic, Lambert Conformal Conic - switch the displayed projection at runtime. The Luciad-parity capability.
Switch projections →Customers, orders & sales density on the map - read-only, geocoded from the ERP middleware. The spreadsheet becomes a map.
Map business data →Plotly charts with bidirectional chart↔map selection linkage. Lasso in the chart, highlight on the map - and back. The DataPlotty parity.
Link charts & map →INTREP/IPB/AGM doctrinal map-sheet templates, geo-accurate neatline graticules, and an atlas-series orchestrator. The RemoteView finished-map parity.
Design map sheets →The newest release broadens the platform in three directions at once: an immersive 360° / StreetView viewer that geolocates tags from panoramas; a data-and-automation tier - ServiceNow, DLT data loading, and a pure-SQL web crawler - that pulls enterprise and web sources onto the map; and a project foundation plus event-driven hooks (DuckDB trigger→webhook, hot-folder auto-ingest) that make the whole thing respond when something happens.
Immersive equirect / cubemap / 360-video viewer with click-to-place geolocated tags, AI auto-tagging (YOLO-E), multi-source telemetry, and live cross-window map FOV sync.
Step inside the scene →Incident tickets + CMDB assets on the map via DLT, write-back create/update/close direct to the Table API, and an opt-in poller. No Python middleware.
Map the ticket queue →REST APIs, 60+ verified sources (Salesforce/HubSpot/Notion/Stripe…), and any SQLAlchemy database → staging DuckDB → served layers, geometry-aware.
Load any source →Crawl pages, sitemaps, or discovered link graphs; extract tables, CSS, JS vars, or articles (the IMPORTHTML analog); preview live - pure SQL, no sidecar.
Crawl the web →Trigger→webhook for DuckDB - forward query & transaction events to webhooks, an in-app SSE feed, JSONL audit log, or a local command. Off by default.
Make DuckDB react →A project is a portable folder + one .tmgproj GPKG, backed by an in-memory DuckDB hub with VIEW-default no-copy storage. Copy the folder; the whole project travels.
Drop a .geojson / .parquet / .gpkg into a watch folder and it appears on the map within ~2 s - the data landing-zone workflow, air-gapped.
From tactical operations to city planning - wherever secure, real-time, 3D geospatial intelligence matters.
Secure sharing within classified networks, TAK/CoT fusion, air-gapped situational awareness.
Live damage assessment, resource routing, real-time field-to-command data flow.
3D city models, shadow & solar studies, multi-department infrastructure coordination.
Underground utility modeling, 3D asset editing, network planning.
Mining, forestry, geology - sub-surface modeling and LiDAR point clouds.
Data sovereignty, inter-agency collaboration, citizen service delivery.
Civil & environmental design, building information modeling, change-order reduction.
Universities and research centers analyzing spatial data at scale.
“Network Data Sharing reduced our emergency response time by 45 minutes, potentially saving lives during critical incidents.”
- Emergency Management Director, Major Metropolitan Area
Run the installer, and in seconds your workstation is hosting, editing, and sharing 3D geospatial data - no cloud, no license server, no waiting.