Visualizing openstreetmap data


I found blender-geo, a project on github to create meshes in blender from openstreetmap data. You simply export a region of osm data and then use the plugin to import that. I made some contributions to the project, but this weekend wanted to use it for a little hack/side experiment using NL-specific data.

The dutch government offers a huge dataset of lidar data (basically, data from a plane that flies with a laser over a territory for the purpose of deriving the height of terrain+constructions) at 5cm resolution covering the entire country. It’s basically an enormous point cloud of x,y,z data of the land coverage. Some smart people figured out what’s ‘street level’ and what are objects like houses and cars and separated this into different datasets.

So from that data, I derived the ‘street level’ for a polygon of osm data, then searched the bunch of point data that should be the roof and calculated an average for both. Then I applied that as a measure for building height. This run took 3 hours to complete, because there’s quite a bit of data involved.

Here’s the blend file if you want to experiment and look further:

https://github.com/gtoonstra/blender-geo/blob/experiments_with_ahn/blender/delft.blend

Look into the “custom properties” for objects to get more info.

All materials applied are pretty basic with an intent. This blend file is how the scene comes out the importer. The idea is that artists work on the materials to improve the visualization. So just by editing the materials, the look of the scene should improve very rapidly.

If you look on bing and search for Delft, around “Markt” and “Nieuwe Kerk”, you’ll see this little area mapped out. You can then see how it approximates the height data quite nicely.

Larger image: https://raw.githubusercontent.com/gtoonstra/blender-geo/experiments_with_ahn/images/screenshot.jpg