RoadCraft Devblog: building the engine!

​Hey everyone, and welcome to this RoadCraft devblog!

Today, we’re diving into the origins of the custom Swarm engine (World War Z, Space Marine 2) built specifically for RoadCraft.

But what’s the Swarm Engine? It’s a tool that makes crowds and traffic behave more naturally and react just like in real life.

You’ll discover how it came to life, what features made the cut, which ones didn’t (and why), and even get a glimpse at some future-facing tech that might shape the games to come. 😉

☀️ RoadCraft’s Birth

The idea for RoadCraft started to take shape about a year after SnowRunner launched. At first, it was just a simple pitch: 

What if we built a Swarm engine demo that could pull off terrain deformation like the one we have in SnowRunner?

Back then, the big question on the tech side was how to bring SnowRunner features into the latest version of the Swarm engine, mainly because keeping two engines alive at the same time just wasn’t realistic. 

And yeah, trying to move stuff over from one engine to another came with a lot of headaches.🤯

Once we realized we couldn’t just keep supporting the older engine while building new features, the real challenge became figuring out: 

• What features do we bring over first? 

• In what order? 

• And what will they actually look like in the new Swarm tech?

💪 Rolling up our sleeves

Building RoadCraft meant getting our hands dirty, for real. 😅

We had to rethink everything from terrain technology to lighting. While we brought over some tech from our previous projects, adapting it to a new environment turned out to be way more work than expected.

One big challenge was letting several team members work on the same terrain without stepping on each other’s toes. 

In past Swarm projects, we mostly used geometry instead of full-on terrain tech, so this was new ground for us. 

Then came the RoadCraft gameplay features: pouring sand, laying asphalt, tearing down trees… At first, they were basic, just proof-of-concept. But as they evolved, they started affecting terrain deformation in unexpected (and often chaotic) ways. We spent months just making sure the terrain wouldn’t implode every time a player messed with sand or trees.

At one point, we even thought about cutting asphalt laying altogether to keep the project scope realistic. But in the end, we pushed through and we’re glad we did.😊

Speaking of which, vehicle damage is something we couldn’t include in RoadCraft for now but we’ll come back to that topic in a future devblog! 

Here’s how the terrain looked during the project’s first prototypes.

The terrain from our very first prototype level, featuring the early version of the factory. This is where we first tested the tech that allowed multiple people to work on different parts of the same level simultaneously. Back then, these sections called “terrain blocks” in the engine were much smaller than they are now.

🚧 Sand, forest and asphalt: a love-hate story

Adding sand mechanics was a big leap for us. Our terrain system wasn’t originally built to handle such intense modifications but we made it work. 

What really caught us off-guard were trees. Yep, trees. 🌳

• When trees fight back

Turns out, when you let players dig the terrain, they can also dig out the ground under trees. And if you do that, those trees should logically fall. Easy in theory but tricky in practice. 

And all that forest is only physicalized to some extent. We can not just make all the trees around you “truly” physical

So we had to get smart: trees only become physical when it matters, when you're nearby, when an AI’s interacting with them or when there’s an actual collision.

This led to a lot of bugs (some of which we’re still fixing) but it also made for some great moments. 

Most trees can be taken down, cut, bulldozed or flattened with a heavy enough vehicle. A few trees are indestructible by design, thanks to the level design team. 

There has to be some challenge left, right?

One of our first attempts at creating a “dense” forest in the game level, once we had enough nature assets to make it happen. And yes, that’s the Snowrunner Tayga with the dozer blade. No, it’s not in the game but who knows, maybe it will be in the future. (Though if it does, we’ll need to rebuild it from scratch since this version was just for prototyping).

• Sand challenge

Sand came with its own set of issues. In early tests, players could stack up massive piles of it, which quickly caused memory problems. So we added limits on how much sand you can pile up in one spot and how deep you can dig.

There’s a reason we don’t have excavators in the game: supporting deep, fully physical terrain would come at too high a cost. So the system focuses only on the area directly around the player, which is where all the cool interactions happen anyway.

• Asphalt meltdown

Asphalt was by far the biggest technical challenge. 

Our engine didn’t like having more than two terrain materials in the same area. Three caused issues. Four looked ugly and moreover, was causing rendering chaos. 

We were stuck between cutting features or rewriting core systems. And since road building is central to RoadCraft, asphalt had to stay. That meant saying goodbye to the gravel layer, even though it would’ve made things more realistic. Maybe we’ll find a way to bring it back in a future update or even a sequel.

When the terrain really hates you all the way 😅. Early prototypes of terrain-vehicle interactions were brutal.

Somewhere along the way, terrain material problems were an everyday pain in all the places. And what’s that in the bottom left corner…

The “physics world” looks different… (a peak at how Havok debugger looks like)

👀 Besides the core! 

One of the biggest challenges outside of gameplay was save file size. Every time you cut a tree, pour sand, lay asphalt or simply drive around and deform the terrain, the save file gets bigger. And at one point, we were seriously wondering if we could even fit it in console memory.

Optimizing this beast was a process half a year and half our collective nervous system out the window. It’s still a heavy file, but it’s manageable now.

That said, if you’re planning to create an asphalted parking lot of all the maps in the game, your hard drive will be grateful to you. 😂 

This devblog has come to an end! 

As with anything in life, the development of a completely new title on new technology is a massive undertaking: one that demands careful planning, difficult choices, and constant efforts to avoid burning out the team or pushing the project over budget.

There’s still a lot more we’d like to share, and we’ll dive deeper into other aspects of RoadCraft in upcoming devblogs.

Thanks for reading and be prepared for May 20! 😉

— Vitaliy Y. - Game Director