The striking differences between geo-engines and game engines

A look at how it works in practice

In your opinion, what are the biggest technical challenges involved in processing large geospatial datasets in 3D?

“Everything has its value and its place,” Joris begins. “One significant trend we’ve been seeing lately is the widespread adoption of hybrid solutions. We’re currently collaborating with Nvidia to visualize reality capture, digital twins real-world models at various levels. What we’ve noticed is that the standard formats used in the geospatial industry are difficult to process at scale within game engines.”

Joris continues: “For smaller scenes—such as when a client asks us to visualize a townhouse with a high level of detail and visual fidelity—we consider creating those 3D scenes using photogrammetry or point-cloud meshes. We can then apply real-time ray tracing to them. But if you want to do more with it—such as using sensor data and adding all kinds of vector data—it becomes a lot more difficult. Technically, it is possible to achieve this in a game engine, but it’s not easy.”

“Visuals aren’t unimportant, but they aren’t the most important thing. From a geospatial perspective, we naturally try to visualize PBR (Physically Based Rendering) shading and processing aspects. Shadow analysis is something that does occur in our field, but ray tracing isn’t particularly relevant.”

From the perspective of municipal practice

Corné Helmons, Advisor Digital Twin, GeoBIM, and Standardization at HCI Infra BV, also sees the differences. He has led digital twin for clients including the municipalities of Rotterdam and The Hague. He shared his vision for applying his work using a game engine:

“The biggest technical challenges we face mainly have to do with the accuracy of our data—specifically, ensuring that the georeferences are correct and kept up to date. We took a training course on how to use a game engine, and it works like a charm,” he says with a smile.

Corné pauses to think for a moment and continues: “We do notice that there are limitations to the scalability of our data processing.” That’s why I challenged the management team to form partnerships with other organizations that can help us with this, such as IMAGEM, whom we met during our trip to the Smart City EXPO event in Barcelona.”

Corné: “Well, that’s exactly the core issue we’re currently facing at the City of Rotterdam. The way I see it, the Open Urban Platform (OUP) will serve as a hub connecting a vast amount of data, allowing us all to align the same references, BIM models, and datasets. That way, it won’t matter whether we Luciad with Unreal or Luciad —we can be confident that the data is accurate, or at least acceptable for the project we’re working on. The goal should be the result, not the path to it.

For now, our platform is designed for visualization without a "Z-component," so we use a flat surface. This doesn’t matter for BIM models because they don’t incorporate any curvature either. This makes things easier for all of us, given the size of our work area within Rotterdam. Moreover, our geolocation generally works quite well, as long as we stick to the NAP level. However, as soon as we start working on larger-scale projects, such as the Groene Boog or a tram line from the beach to the city center, we do run into challenges.”

Joris adds: “Datasets are always so large that, as soon as you want to stream visualizations and work with a high level of detail, you face different challenges. Game developers have been dealing with this for years, often relying on manual work to ensure that everything looks and functions properly. For example, within a room or space, they know that the outside world is omitted because it doesn’t need to be displayed. As a result, the engine doesn’t have to render that data, which ensures optimal use of the hardware.”

“For example, a church tower in the distance might be rendered less sharply. You can adjust that manually, but it remains a challenge to tailor the right ‘level of detail’ to the situation. We’re also seeing hybrid environments being created that we load into Unreal using our datasets. We have clients who render 3D data processing and 3D tiles exclusively in the Unreal Engine, but then load them via Luciad. This demonstrates the flexibility of this engine in combination with our platform, allowing us to adapt relatively quickly and effectively based on the client’s needs.”

Joris really loves the versatility and variety in all the data that’s collected: “What I enjoy is the diversity and sheer volume of data that’s generated. Take data collection (measuring, scanning, etc.), for example, where data is generated in all sorts of different ways and formats. But also the analyses we can apply to aerial photos, data, and real-time video as they’re collected, which can yield surprising results. All this data and the challenge of bringing it all together is what motivates me to get out of bed every day.”

Corné is primarily interested in the availability and speed with which data can be applied: “I think that IMAGEM Luciad , is Luciad very good addition to the OUP, offering the perspective and scale needed to approach the entire world. Especially for public datasets we access from other municipalities, this can make it easier to seamlessly align digital twin beyond our own scope. If we can then get all the georeferences to align and use the same standards, I think the world would already look a lot better.”