Notes on OpenStreetMap

Although OpenStreetMap (OSM) has a very different goal and purpose (cartography) than the VTP (full 3D reality), it is potentially a useful starting point in some parts of the world, providing a base map, and in some cases even some primitive 3D data.

OSM Data

• OSM is stored in a large database, but also has an XML format (.osm) for export/interchange.
• It is simply points ("nodes") and polylines ("ways"), with topology, using a very large set of standardized attributes (key/value pairs), which are applied very rigorously in some areas and inconsistently in others, due to the volunteered nature of the data.
• Although OSM began as an effort to produce a free streetmap of the world, by 2010, OSM was increasingly being used to represent things besides roads - water, building footprints, and even small features like fences and hedges.
• Using the default web interface on the home page, you can export a region of interest to an .osm file but this is only supported for fairly small regions.  For larger regions, there are services like the Geofabrik free download server, and there are tools like Osmosis (with a GUI, OSMembrane) to filter for features of interest.
• Aside from the main wiki page on Map Features, a very useful reference on which tag key/value pairs are actually used is at taginfo.openstreetmap.org.

Data collection

• There are three main ways that people add data to OSM:
  1. Importing an existing dataset, which usually involves a lot of cleaning up the data.
  2. Consumer GPS tracks, for example as described in Grassroots GIS (2010)
  3. Tracing off of permitted aerial imagery.
• All three of these have significant pitfalls (GPS and aerial accuracy/agreement, etc.)
• GPS was the most common approach during OSM's early years, but now tracing aerials is more common.

Roads

• Some small drawbacks to OSM's road model are:
  1. Roads are simple centerlines, which of course fails badly for visual representation of large intersections, freeway exits, etc.
  2. They store every single point in every polyline as its own node with attributes, which is more overhead than one might hope for an efficient visualization system.
• Some good things:
  1. It supports areas for features like parking lots, which could capture a lot of visually significant paved areas of cities.
  2. Since roads have a direction, you can flag the presence of features like sidewalks using left/right/both.

Buildings

• OSM supports building footprints: Using the same linear features used for roads, the feature is closed and attributed with "building=yes", or a more specific type such as "building=hotel" (see OSM Key:building).
• Some physical attributes potentially useful for visualization are supported, such as height, building:levels, roof:shape, but these seem to be very sparsely supplied in the OSM map.
• OSM: Simple 3D Buildings suggest some modest tags for adding some 3D to the map.
• OpenBuildingModels - Towards a Platform for Crowdsourcing Virtual 3D Cities (2012) goes further, suggesting that links could be made in OSM to 3D structure models in other places, for example the 3D Warehouse.

Power

• OSM's Power networks include power:pole and many supported kinds of power:tower, although in practice it is usually just "power=tower" with no "design" or "tower:type".

3D View

• It is relatively simple to take the OSM export data and extrude the primitive features (road centerlines, building footprintf) in 3D.  There are a number of software tools that do this:
• Kendzi3D is a plugin for the popular JOSM editor, which gives a live 3D view (on a flat plane) of your data as you edit:
  
• The VTP software (VTBuilder, Enviro) support importing from OSM:
  
• Several other tools and projects (OSM-3D, OSM2World ... see OSM Wiki: 3D Development)
• It can even be done in a webpage using only 2D graphics to draw the 3D buildings, as on osmbuildings.org
• Maperitive is a standalone tool for rendering OSM, optionally draped on hill-shaded elevation (SRTM is the default source), with output to bitmaps, SVG or Collada.

Metrics

• Interesting ways to track OSM usage: Your OSM Heat Map and How did you contribute to OpenStreetMap?