forge3d 1.21.0

forge3d: Rust/WebGPU core with PyO3 bindings

GPU-accelerated 3D terrain rendering for Python.
Built in Rust with WebGPU. Pre-built wheels — no Rust toolchain required.

---

This image was generated with 5 lines of Python. Load a DEM, launch the GPU viewer, position the camera, and capture a publication-quality snapshot. No shaders to write. No OpenGL boilerplate. Just pip install forge3d.

import forge3d

with forge3d.open_viewer_async(terrain_path="rainier.tif") as viewer:
    viewer.set_orbit_camera(phi_deg=225, theta_deg=35, radius=1.2)
    viewer.set_sun(azimuth_deg=315, elevation_deg=32)
    viewer.snapshot("render.png", width=3840, height=2160)

Install

pip install forge3d

Need Jupyter widgets or tutorial datasets?

pip install "forge3d[jupyter]"     # ViewerWidget for notebooks
pip install "forge3d[datasets]"    # on-demand sample DEMs, CityJSON, COPC
pip install "forge3d[all]"         # everything

Gallery

PBR terrain	Landcover overlay	LiDAR point cloud
Sun & shadow control	GeoPackage labels	CityJSON buildings
Vector overlays	Camera animation	Map plate compositor

See the full gallery →

What You Get

Open Source (Apache-2.0 / MIT)

Everything you need to go from raw elevation data to a rendered 3D scene:

• Interactive viewer — real-time orbit, pan, zoom via a Rust/WebGPU subprocess controlled from Python over IPC
• Terrain rendering — load GeoTIFFs or numpy arrays, PBR materials, 100+ colormaps
• Vector overlays — GeoJSON/GeoPackage polygons, lines, and labels projected onto terrain
• Point clouds — COPC and LAZ files with millions of points, colored by elevation or classification
• 3D Tiles — stream OGC 3D Tiles tilesets directly into the viewer
• CRS reprojection — automatic coordinate transforms via PROJ + pyproj
• Camera animation — keyframed flyover paths with frame-by-frame export
• Jupyter integration — ViewerWidget embeds the viewer inline in notebooks
• High-res snapshots — up to 8K PNG export from any camera angle

Pro

Professional cartography and production workflows:

• Map plate compositor — legends, scale bars, north arrows, multi-panel layouts
• SVG / PDF export — publication-ready vector output
• 3D buildings — GeoJSON, CityJSON, and 3D Tiles import with roof inference and PBR materials
• Mapbox Style Spec — load and apply Mapbox-compatible styles
• Scene bundles — save and share complete .forge3d scene packages
• Commercial licensing details → https://github.com/milos-agathon/forge3d#license

Tutorials

Two tracks depending on your background:

• GIS professionals — Visualize your first DEM →
• Python developers — Your first 3D terrain →
• Architecture overview — How forge3d works →

Jupyter

from forge3d.widgets import ViewerWidget

widget = ViewerWidget(terrain_path="dem.npy")
widget.set_camera(phi_deg=225, theta_deg=35, radius=1.2)
widget.set_sun(azimuth_deg=315, elevation_deg=32)
widget.snapshot()  # renders inline

Links

Documentation  ·  API Reference  ·  GitHub  ·  PyPI

License

Open-source core released under Apache-2.0 OR MIT. Pro features require a commercial license key set with forge3d.set_license_key(...). Licensing details live at https://github.com/milos-agathon/forge3d#license