Prize committee’s comments:
Culture in the Crosshairs is a unique angle on a side of the Ukraine War that has not received as much attention as some other topics: the loss or damage to important cultural heritage sites. It’s a clear example of where data journalism can really add to our understanding of modern war and its impact, including documenting war crimes.
The team had no prior experience of using such large amounts of data — over a billion data points in some cases— and had to deepen their technical abilities in order to create this innovative piece of storytelling. The judging panel was very impressed that the team pushed themselves technically in order to provide an experience that is innovative and engaging for readers. The result is an extraordinarily compelling look inside the destruction of Ukrainian heritage sites, worthy of a Sigma award.
We followed the efforts of volunteers and human rights workers to document the destruction of treasured Ukrainian buildings.
We took point cloud data captured on the ground in Ukraine using a LiDAR laser scanner and used it to construct a fly-through of the buildings.
Because of the size of the point clouds, we had to use innovative techniques that allowed the data to load in blocks that allowed the visualisations to work for audience members without fast internet connections or older mobile devices.
The resulting story gave readers a close-up look at the damage inflicted on Ukraine’s historic buildings.
In our piece “Culture in the Crosshairs”, we followed the efforts of volunteers and human rights workers to document the destruction of treasured Ukrainian buildings.
The resulting piece provided our audience with a unique insight into the devastating toll of war and also the tactics employed by Russian forces accused of trying to eradicate Ukrainian cultural identity.
The piece also went beyond the data to look at the real world efforts of Ukrainians on the ground looking to hold Russian forces to account for alleged cultural war crimes. We spoke to lawyers and human rights workers compiling briefs of evidence on the alleged deliberate targeting of treasured Ukrainian cultural sites.
We also geolocated all of the sites listed on UNESCO’s compilation of damaged cultural monuments and buildings (there were about 170 in August, 2022), in order to map just how widespread this damage was across Ukraine. For several of the worst hit sites, we obtained satellite images and Google street view images to show them before and after the destruction.
The work inspired similar work on the same topic from other visual teams around the world.
For this project, we relied almost entirely on open-source software in the processing of the point cloud data, the 3D modelling and also for the front-end visualisations.
We used Cloud Compare(CC), both the GUI and command line interface (CLI), to drastically reduce the size of the point cloud data. The file size for each building capture was around 1.5 gigabytes or more than 1 billion points. Obviously, displaying data of that size would be nearly impossible in the browser, so we used Cloud Compare’s CLI to write a script that reduced the number of points in the files so that data points were no closer than a few centimetres to each other. In the original file, some points were as close as a few millimetres apart.
We then converted the files to a more common point cloud filetype (LAS/LAZ) and brought that into CC’s software that allowed us to tidy up the noise in the model, remove any unnecessary areas that had been captured and also run an ambient occlusion filter over the model to enhance its colours and shading.
This whole process whittled the file size, for each building, down to about 20mb/1 million points rather than 1.5gb/ 1 billion plus points.
We then used the LAStools software suite to split the data into tiles so that we could load the file in chunks rather than forcing our audience to wait for the whole thing to load at one time.
This project was then brought into ThreeJS, where we modelled what the users would see as they scrolled through the piece.
We also used QGIS to map the sites on UNESCOs list of damaged cultural sites, as well as processing satellite images.
Adobe Illustrator and Blender were also used.
Context about the project:
One of the biggest challenges we faced bringing this story to life was that none of us had done much work visualising point cloud data in the browser. We have little to no budget, so everything we created had to be in house and open source. We had experience creating and visualising 3D models, but in the past we had presented that data as image sequences or videos rendered from Blender or other 3D software. For this piece, we wanted to present the data in its purest format – the point cloud. This approach brought with it a whole series of challenges.
The first were the enormous file sizes, which we tackled using the approach I listed above. Then, we didn’t want our audience to have to wait for the whole file to load, so we came up with a way to split the data into tiles, so it could load in smaller sections at a time which meant people could see the models constructing themselves while they read through the story.
This meant the piece worked smoothly across all devices, mobile and desktop and through relatively poor internet connections as well.
The geolocation of damaged cultural sites was a significant undertaking as the location of sites wasn’t immediately obvious from the names supplied on the UNESCO list. For various reasons simply googling the name of the site to find the location was not always an option. We often had to trawl through social media, scour local news reports, check satellite imagery and cross check locations with sites listed on the Ukrainian Ministry of Culture’s own database.
We then used these sites and matched them up with satellite images and Google street view (a painstaking task in itself) to present before and after images for several of the worst hit sites across Ukraine.
What can other journalists learn from this project?
Two of the big takeaways for other journalists for the piece are:
1. We did this using entirely open-source software and code libraries. So everything we did could be entirely recreated without any budget constraints, other than the time it took us to pull it all together.
2. The other is that, at the beginning of this project none of us were 3D experts. But we were able to get our heads around the various different programs available and pull everything together. If we can do it, so can you!