3D Visualisation and Presenting GIS

3D is seductive and often unnecessary, so start with a test: does the question involve height, view, skyline or shadow? If it does not, a well-designed 2D map is faster, clearer and more honest. If it does, flatness actively hides the answer.

There is also a ladder of 3D. The bottom rung, and the most useful, is 2.5D - flat footprints raised to a height stored in an attribute (extrusion). It is fast, needs only a footprint layer and a height field, and answers most massing and skyline questions. True 3D adds real geometry - roofs, facades, overhangs - and matters when detail carries the argument. City-scale work adds a notion of level of detail (LOD): blocks for a district study, detailed buildings only where they count. Reach for the lowest rung that answers the question; a district of lovingly detailed models when a block study would do is wasted effort and slow to share.

A GIS scene is assembled, not drawn. Three ingredients do most of the work. First, a terrain surface - a digital elevation model draped with imagery so the ground has real relief; in India CartoDEM via Bhuvan gives national coverage, and SRTM or ASTER cover the globe. Second, building footprints - polygons from OpenStreetMap, Microsoft or Google building layers, or a survey. Third, a height attribute on those footprints - real surveyed heights where you have them, or an estimate from storey count - which drives the extrusion.

Set the footprints on the terrain, extrude each by its height field, and you have a defensible 2.5D scene: relative heights are real, the ground is real, and every block still carries its attribute row. For genuine city models there are open data models to know: CityGML describes semantic 3D cities (buildings that know they are buildings, with parts and levels of detail), and 3D Tiles is the OGC standard for streaming massive 3D scenes to a browser. You rarely author these by hand, but you will receive and serve them.

A scene trapped in your software persuades no one, so the last technical step is export. On the free path, QGIS has a native 3D map view where you set the terrain and extrude a layer, and the qgis2threejs plugin exports that scene to an interactive three.js page the client opens in a browser - rotate, zoom, click, no software to install. On the Esri path, ArcGIS Pro authors local and global scenes, publishes them as web scenes to ArcGIS Online, and for design-led work ArcGIS Urban runs zoning and massing scenarios while CityEngine generates rule-based procedural cities.

Whatever the engine, you will produce a few output types for different rooms: an interactive 3D web scene for a client to explore alone; a still render or print board for a hearing wall; and sometimes a fixed flythrough video for a presentation where you control the pace. The scene is one; the outputs are chosen for the audience.

The hardest part of visualisation is not making the image - it is presenting it to people who will act on it. Two audiences, two disciplines. A client wants to understand and decide; give them one clear thing per view, an interactive scene to explore at their own pace, and plain labels. An authority - a plan-sanction reviewer, a hearing panel - wants to verify; give them the sources, the CRS, the data vintage, and an honest note on accuracy, because their job is to trust nothing they cannot check.

In Indian statutory settings this is concrete. When a 3D massing supports a plan under a master-plan or development-control process, the reviewer will ask which terrain and footprint data you used, whether heights are surveyed or estimated, and in which coordinate system it sits. A scene that answers those questions on its own title panel is an argument the authority can engage with; a beautiful scene that cannot is just a picture.

A rendering persuades before a viewer has read a single number - which is exactly why it demands restraint. The camera angle, the time of day, the greenery you scatter, the people you place: each can make a scheme look better or worse than the data warrants. That power is legitimate when it clarifies and dishonest when it oversells.

So hold two rules. Do not let beauty outrun the evidence - if the heights are estimated, say so; if the trees are illustrative, say so; do not present a rendered noon that hides the shadow a real winter afternoon would cast. And label uncertainty as a feature, not a weakness: a scene that distinguishes measured from assumed is more credible, not less. The most persuasive visualisation in a serious room is the one whose maker is visibly honest about what it does and does not know.