New Zealand LEGO Map

The Planning

The first thing I realised was that QGIS was way clunkier than I would have hoped for; that paired with my close to non-existent knowledge about GIS tools caused some frustration. I knew what I wanted to do but I didn’t know the terminology to look for help.

It took me 10-15 hours over a weekend to figure out the basics and to get to a point where I had a first draft of an elevation raster for New Zealand. At that point I definitely jumped the gun (more on that later) when I started building the whole model in LEGO stud.io, a piece of software that lets you build virtual LEGO models.

The benefit of actually building something was that I ran into a bunch of questions I hadn’t thought of before. A 2 dimensional version is rather simple, with a third dimension I needed to think about how I want to layer the different colours. How do I make the whole thing stable? Do I start on the lowest layer and build up or do I have my baseline in the middle and build down and up? And how the hell would I hang it on a wall once it’s done?

I got in touch with Dirk who build the world map and he happily answered some of those and a bunch more questions. By the time I felt like I knew how I wanted to do it, I didn’t like my initial QGIS work any longer though..

• 

• 

• 

The Questioning

I’m a data person, which probably doesn’t come as a big surprise…This project is a data visualisation of sorts and to be happy with it, I had to be confident it represents the data in a way that makes sense; to me and any other audience.

So let’s see:

Scale: Each stud (the technical term for one LEGO square) represents 10km in the real world. This results in a scale of 1:1250000. I am happy with this, although I did create patterns for 12km and 15km squares because I questioned whether 10km would fit on our wall…In the end I did it anyway.

Colour palette: This was basically just verifying that Dirk’s work would work for me. The colours are a version of a standard colour scheme for elevation maps, which go from green to yellow to brown and white. Although there are discussion about how suitable that is (low areas do not always have green vegetation, etc.) I think that most people will be able to relate to it. Also, the colours Dirk chose are some of the more common LEGO colours which helped keeping the whole project within “a” budget. (more on that later as well).

Elevation value: Since each square needs one colour, I have to find one elevation value that represents that value. I started with the median of all values within a 10×10 square. The problem with this was that it ignores outliers (as medians do). I needed to have Mt. Cook (the highest peak in NZ) visible on the map and if I used the median, it wouldn’t show up. It would also “blur” everything so that there wasn’t a lot of difference in elevation. I tried the maximum value in each square and that worked better. It made the high peaks visible but neglected the deep valleys that exist on the South Island. I ended up going through it manually to adjust values. This helped to make for example Arthur’s Pass visible and made Marlborough Sound look more like a sound.

Elevation bins: This one was the most technical problem and one I didn’t find a clear answer for. I had my colours defined by how do I ramp them to 1. not break any mapping rules and 2. have the elevation represented in a relatable way. If I create equal bins based on percentiles, all high peaks disappear. If I divide the range of elevation by the number of bins, all the lower elevations blur into one. I ended up manually defining them as 0m-250m, 250m-500m, 500m-1000m and then 500m steps. This worked well to give a good impression of the NZ landscape.

This whole process actually took the longest time. I build 3 different base maps for 3 different scales, researched mapping best practise and still learned how to use QGIS but eventually I settled on something I was happy with

The (virtual) Build

• 

• 

• 

• 

• 

The Order

Now I said Stud.io integrates with bricklink, which means I can export the list of parts I used and order them on the website. Bricklink even offers to go through all available shops and pick the cheapest parts. However it does not take into account shipping fees and/or thresholds after which shipping is free. Also, in New Zealand there is limited amount of shops available. I can’t complain but especially larger orders of more obscure parts can get either very expensive or they might not have the necessary quantity available. I spent another 10-15 hours structuring my order to consolidate them in as few orders as possible to reduce shipping cost and to take advantage of cheaper prices where possible. Since my family is in Germany, I also had the option to place several orders in Europe, pay for local shipping and then have my family send them over in bulk, which made the shipping much cheaper. In total, all of this probably made a ~500NZD difference at the end. I am quite certain that I could have reduced the price further had I realised that the automatic conversion from Euro to NZD might not have been in my favour…

Anyway, I order around 3800 bricks in 12 orders. Following that, some of the stores didn’t have all the bricks in their inventory and I had to source them somewhere else AND the whole world struggled with COVID19, so even though my family sent the bricks before most of the flights where canned and countries went into lock-down, I wasn’t sure I would receive all the bricks any time soon.

The (physical) Build

Turns out I didn’t have to worry and my bricks were delivered in a reasonable timeframe. I spent some time sorting everything by colour to make it easier for myself, then I set myself up in the spare bedroom and started building. It turned out it was easier than I thought. I had to work around a few errors in my digital models which was fine and I even ended up with a bunch of spare bricks (at the same time I was missing a bunch of bricks that I thought would have been included in my orders).

Now I have a 7.8kg LEGO map in my spare bedroom that waits for the lock down to end so that I can buy material to hang it in my living room.

Final thoughts

It turns out I am definitely not the first one to approach a project like this. As I mentioned before, Dirk built a world map, as well as a globe. Dan Harris built a map of Scotland and just a few days back, somebody in NZ finished a map of Westeros.

It looks like I found a much simpler way to create a map like this than Dirk, while at the same time probably spending as much time as he did anyway. There is an R package (of course there is) which converts an image into a mosaic of LEGO bricks. With a bit of scripting the whole process could – in theory – be automated. However even if it was automated, there is still the problem of brick prices and availability, which means you might save time if you do several different maps but pay a premium for bricks.

I am very happy with how the map turned out. In one of my virtual builds I actually included a legend and lat/long markers in the frame, however I decided against it, since the main audience is myself (and I know the details) and although maps usually come with legends, it felt like it cluttered it too much.

Now I am keen to see what else I can build in LEGO and how other data points can be “physicalised”.

If you have questions or comments, if you want to build your own LEGO map or if you know other examples, get in touch and let me know! Thanks!

Update

The map got a bit of attention, so by now you can read about it in the NZ Herald and I talked to Adam on his “Project Geospatial” podcast, which is on Youtube and Spotify

• 

• 

• 

• 

• 

• 

•