Shadow = height ÷ tan(altitude)

Data: Open-Meteo (last year)

− = sink into ground  ·  + = lift up  ·  type a value + Enter to set exactly

⊙ Reset Position ✕ Remove Model

.dae — SketchUp: File → Export → 3D Model
.obj — Blender: File → Export → Wavefront (.obj)

DXF (AC1015/AutoCAD 2000): AutoCAD / GstarCAD / SketchUp Pro (3D, can embed the street tile as a real image — see Base Map Texture below)
Collada .dae: SketchUp native import

Road centerlines into the same DXF/Collada file (not a separate download), split by class onto ROAD_MAJOR / _SECONDARY / _MINOR / _SERVICE / _PATH, plus street names on ROAD_NAME (romanised where OSM has it). Read straight from OpenStreetMap, so each road comes through whole — the map no longer jumps around during export.

Separate layers: TREE (crown circles), WATER, WATERWAY, GREEN_PARK / _GRASS / _FOREST. Tick to check what OpenStreetMap has for your area — coverage is hand-mapped, so it is rich in central Bangkok and often empty upcountry.

The same 16-sector rose as the on-screen one, drawn beside the site as real geometry with reference rings and an arrow in the direction the wind flows toward. Needs wind data loaded (Wind panel) first.

Independent of each other and of whatever the map is currently showing on screen — tick either, both, or neither. Each becomes its own image on a separate layer/group (BASEMAP_IMAGE / BASEMAP_IMAGE_SATELLITE in DXF; two hideable groups in SketchUp), so with both ticked you can freeze one and thaw the other to switch views without re-exporting. The .dxf/.dae references its texture(s) as external .png files, so all of it comes down together in one .zip — extract it to a folder and import from there. GeoJSON/PNG/PDF ignore this.

Adds sun.csv (altitude/azimuth every 15 min across the analysed date) and wind.csv (hourly readings for the selected month + the same 16-sector summary as the Wind Rose) to whichever export you run — for charts and tables in a presentation sheet. Bundled into a .zip alongside the main file whenever there's more than one file to download.

Area in sq.m and rai–ngan–sq.wah, perimeter, site coordinates, analysis date/time, sun altitude & azimuth, building height, units. Written in ASCII (DXF text is code-page based, so Thai script would arrive garbled in AutoCAD).

Labels every site boundary segment along the edge, e.g. 24.50 m  N45-30'E — survey-plan style, for checking against the title deed.

Dashed line offset inward from the boundary; the remaining buildable area is reported in the Site Info block.

One shadow per clock hour on the selected date, each on its own layer (SHADOW_0900…), plus a combined envelope of everything shaded during that window. Hours when the sun is down are skipped.

A true 3D line from the site centre toward the sun at the selected time — align SketchUp's shadow settings or a directional light straight to it.

Metres (1 unit = 1 m) Centimetres (1 unit = 1 cm) Millimetres (1 unit = 1 mm)

Scales the whole .dxf and sets $INSUNITS to match, so it drops into a drawing already set up in that unit without rescaling. Collada carries its own unit tag and GeoJSON is in degrees, so both ignore this.

Surrounding buildings are closed solids by default, so SketchUp and AutoCAD can report their volume. Tick this to drop the floor faces — they sit at ground level under each footprint and are never visible — which cut a real 2,626-building Bangkok export by 25.6% (5.47 MB → 4.07 MB). The masses then stop being solids. The site building always keeps its floor.

Saves site boundary, camera, Sun/Wind settings, and the imported 3D model (if any) to a single .siteanalysis.json file. Wind/sun data itself is re-fetched fresh on load, not stored.