★ New · feat/cad-import

Decades of as-builts, on the map

Every utility has file cabinets of substation one-lines, yard layouts, and line drawings locked in CAD. Bring them onto the map in their native DWG, DXF, and DGN formats, rubber-sheet them into place with control points or a source CRS, and serve the result as GeoJSON, MBTiles vector tiles, or GeoPackage — rendered alongside your basemaps, corridor imagery, and 3D scene layers.

DWGDXFDGN EPSG:4326EPSG:3857MBTiles-vectoraffine · polynomial-2

What it reads

Three CAD formats, three outputs

DXF is read in-process via the managed OGR driver (always available). DWG and DGN — the formats most substation and transmission as-builts live in — fall back to external ogr2ogr with the OpenDesign / ODA driver; the response carries a driver field so the UI can tell you when a portable GDAL is needed.

Outputs served as layers: GeoJSON (GeoJSONLayer), MBTiles vector tiles (VectorTileLayer), and GPKG (FeatureLayer).
All output geometry normalized to EPSG:4326 (WGS84); the MBTiles path additionally reprojects 4326 → 3857 during tiling.
Files written in-process for DXF/GPKG — no external GDAL required for the common case.

POST /api/cad/convert (multipart)

// DWG/DXF/DGN → georeferenced layer
fields:
  file        = site-plan.dwg
  format      = mbtiles-vector
  georefMode  = controlPoints
  controlPoints = [
    {src:[0,0],   wgs84:[-74.00,40.71]},
    {src:[500,0], wgs84:[-73.99,40.71]},
    {src:[0,500], wgs84:[-74.00,40.72]}
  ]
  transformKind = affine   // or polynomial-2

→ driver:"ogr2OgrFallback"
→ served at /api/tiles/{id}/…

Two ways to georeference

From drawing coordinates to real-world positions

As-builts live in arbitrary drawing units. VoltGrid gives you two robust paths onto the globe, both with a live RMSE preview before you commit — so a 1970s yard drawing lands exactly on today's substation fence line.

🎯

Control points

Click ≥3 matching points between the drawing and the map — fence corners, structure bases, road intersections. A least-squares affine (6-param) or polynomial-2 (12-param) transform is solved with RMSE reported — so you see the fit before importing.

🌐

Source EPSG

If the plan already carries a real projected CRS (e.g. the State Plane or UTM zone your engineering standards use), supply it directly. OSR reprojects to WGS84 with no manual clicking.

In-App Screenshots

CAD Georeferencing, Vector Layers & Layout Designer

VoltGrid desktop — CAD drawing import dialog with source CRS selection — CAD drawing import modal configuring source Coordinate Reference System (CRS / EPSG) and georeferencing mode

VoltGrid desktop — CAD point feature layer loaded over Africa — Georeferenced CAD point feature layer imported and rendered directly over the map canvas

VoltGrid desktop — CAD arc and vector linework layer rendered on map — CAD arc and polyline vector layers converted and rendered with hardware-accelerated symbology

VoltGrid desktop — Map Layout Designer and sheet composer — Map Layout Designer child window with cartographic canvas tools, map frames, and elements for engineering print deliverables

The conversion API

Convert, preview, serve

Capability	Endpoint	Notes
Preview	POST /api/cad/preview	Bounds, layers & GeoJSON in source coords
Convert	POST /api/cad/convert	Multipart → GeoJSON / MBTiles-vector / GPKG
Features	GET /api/cad/{id}/features.json	Transformed GeoJSON, ETag / 304 cached
Tiles	GET /api/tiles/{id}/{z}/{x}/{y}	MBTiles-vector rendering
Style	GET /api/tiles/{id}/style.json	Vector tile style
Spatial facades	/api/spatial/cad_{id}/…	MVT / OGC Features / FeatureServer
Remove	DELETE /api/cad/{id}	Removes from registry + deletes files

Georeferencing supports affine and polynomial transformations via least-squares adjustment; reprojection is handled via GDAL/OSR.

Engineering drawings meet the grid

Overlay DWG/DXF/DGN as-builts on imagery and 3D terrain — georeferenced, tiled, and served from the desktop. VoltGrid is built for the same design-to-GIS bridge utilities run with PLS-CADD structure and conductor data, with FileGDB, PostGIS, GPKG, and GeoJSON on the way out.

All data formats → Read the docs