I’ve noticed that a lot of architectural visualisations don’t always include the context, partly due to the fact that it’s very time consuming to model and texture every building surrounding the proposed project. I want to change that as it is very important to see how the design adapts and responds to its environment. It also makes for interesting renders!
I’ve outlined the first part of this process below for getting a 3D model of the context for anywhere in the world:
- Site location
- OSM vs Overture
- Downloading the data
- Extracting tiles
- Exporting shape files
- Converting to OSM
SITE LOCATION
First thing to do, is to search for the area in question in Google Maps to get an idea of the size and scale. If you’re lucky, Street View will be available and you can explore the area in first person (even better if you can visit in person!) For this example and for a project I’m working on, I’m looking at the whole of Dubai.
I found it’s about 40km2 which would be a nightmare to model if the internet and data produced by mapping companies did not exist. Although most projects probably don’t require this huge of an area for context, I felt it would be a good exercise to see the limits of what could be visualised.
OPENSTREETMAP VS OVERTURE
Google Maps unfortunately doesn’t provide vector data that could be used in 3D modelling apps so I did some searching and OpenStreetMap seems to be the most popular and comprehensive when it comes to data around the world, partly due to it being open source and having many contributors.
View Larger Map
Exporting from their site is as easy as selecting a box around the area of interest and then clicking export. The only issue I found is that for larger areas it might not always be possible through the web interface. The real benefit of these .OSM files is that they have a lot of useful metadata embedded into them through the use of “Tags” which include information such as height of the building and building type. The only issue I found is that not all buildings currently have all their tags filled out.
This is when I found out about the Overture Maps Foundation, which founded in 2022, improves upon the data found in OSM. It too is also open source but is also supplemented with data from private companies, which makes it the best high quality open map source right now. Their site though is not as user friendly as OSM and you cannot just search for a location on the browser but after watching some YouTube tutorials, I managed to get the data for Dubai on my screen.
DOWNLOADING THE DATA
I use QGIS, an amazing open source public project for viewing and editing geospatial data. To view the Overture maps, I had to download this Python script from GitHub and then run it in QGIS by editing the release field to the most recent data file ("2025-03-19" as of now) from Overture which looks like it gets updated every month. I added in the “buildings” and “transportation” themes for viewing the building profiles and road networks.
I did have some issues trying to select the elements in version 3.42.1 so I downloaded 3.38.3 and the issue was resolved.
EXTRACTING TILES
As you can probably imagine, there’s a lot of data to visualise on screen for an area this large but QGIS was able to render all of this out in a few minutes. However, selecting these individual elements became an issue: I tried selecting the whole of Dubai in one go, and it estimated it would take around nine hours! I was not prepared to wait that long (and the fact the file would be too big to make use of) so I decided there must be a better way to export the vector shapes into a file.
Tiling the area in question with a polygonal grid was the route I took. To create it, I used the Create Grid tool and edited the spacing to be around 6km horizontal and 6km vertical leading to a 60km2 10×10 grid. I then placed this grid layer on top and lowered the opacity to around 40% and gave it an orange colour. I then separated each tile into its own layer and this resulted in 100 tiles/layers ranging from A1 for the bottom left to J10 for the top right. I then was able to batch process each one using the Clip tool which resulted in 100 cropped layers with a more manageable number of buildings in each. I was now ready to save out these selections and import them into my 3D DCC.
EXPORTING SHAPE FILES
The only thing QGIS doesn’t seem to do, is to be able to export as .osm files which is what I wanted (due to computability with OSM data). This led me into researching the best output format for conversion and I found that the ESRI Shapefile (.shp) would allow me to do this and still have the all important metadata passed down.
CONVERTING TO OSM
The final step is to convert the .shp file into an .osm file. This is done by utilising JOSM (Java OpenStreetMap Editor) which doesn’t open the .shp natively but does once the opendata plugin is enabled. Once the file is opened, it’s just a matter of saving it out as an .osm file.
Which should result in a clean .osm file with the buildings profiles exported as polylines and all the metadata included. This will come in very handy when computing the generation of these contextual building models in 3D.
Leave a Reply