In this episode, we dive into the creative mind of Oskar Stålberg, the developer behind many acclaimed indie successes like Bad North and Townscaper. He shares how procedural generation shaped the game, his experiences optimizing it for mobile and VR, and how he found the delicate balance between art and technical constraints. Stay until the end because if you are interested in procedural generation you will love this creative game masterpiece I will share.
The newsletter is also shared with our 10k audience on LinkedIn and if you would like to become a sponsor just answer this email or contact me on LinkedIn 👉 Gabriele Romagnoli.
Interview with Oskar Stålberg
Townscaper is a relaxing app that let’s you create cute towns on PC, mobile, VR and even on the browser 🤯.
What inspired you to create Townscaper, and how did the concept evolve?
Oskar Stålberg: It’s a long story. Townscaper started as a kind of tech and artistic demo. I’d been thinking a lot about the procedural generation of buildings, especially after working on The Division, where we had semi-procedural tools to create New York’s skyline. I wanted to explore the concept of procedural houses and started with a tool called "Brick Block" (you can play right here 🔗)—an early, modular version of what became Townscaper. Over time, I discovered new algorithms, like wave function collapse, that helped me achieve more complex architectural features.
How did Townscaper turn into a full game?
Oskar Stålberg: Honestly, Townscaper didn’t start as a game idea. It was a technical challenge at first—I wanted to figure out how to create colorful, procedural buildings. I couldn’t come up with gameplay to go along with it initially, so I set the project aside and worked on Bad North 🔗. After Bad North succeeded, I had the freedom to revisit Townscaper without worrying too much about commercial success. I focused on the beauty of the procedural generation itself, turning it into more of a creative tool than a traditional game.
Can you explain how procedural generation works in Townscaper?
Oskar Stålberg: In simple terms, the algorithm takes a set of puzzle pieces I’ve designed—small parts of buildings—and figures out how they fit together. It uses something called wave function collapse, which analyzes how these pieces connect, kind of like a jigsaw puzzle. Then, depending on where the player places blocks, it assembles a coherent building or structure.
How do you prevent procedural content from looking repetitive or too "programmer-made"?
Oskar Stålberg: A lot of it comes down to blending the algorithm’s constraints with handcrafted details. For example, in Townscaper, some pieces, like chimneys or roof types, are chosen randomly, which adds variety. I also allow the algorithm to prioritize certain features, like a garden appearing when an enclosed space is created. But I leave room for handcrafting, so details like the placement of windows or mailboxes don’t feel sterile or repetitive. It’s a mix of logic and artistry.
What are some interesting procedural generation techniques you used in Townscaper?
Oskar Stålberg: One fun example is the steel support beams in Townscaper. The algorithm looks at how the beams should connect and form a support network based on simple adjacency rules. It starts with the highest part of the structure and works its way down, coalescing into single pillars. Another example is the gardens—whenever you create an enclosed space with buildings around it, the cobblestones turn into grass, forming a garden. It’s all about playing with simple rules to create interesting results.
What were the biggest challenges in optimizing Townscaper for mobile and VR platforms?
Oskar Stålberg: One of the biggest challenges was developing an art style that looked good on lower-performance platforms. You can’t just take something that looks great on PC and port it directly to mobile or VR—it doesn’t work. I had to write my own lighting solutions and simplify things like ambient lighting. The art had to be visually appealing but also lightweight enough to run smoothly. Outlines and custom shaders became my go-to for keeping things crisp without overloading the system.
What unique design challenges did you face when porting Townscaper to VR?
Oskar Stålberg: The biggest challenge was rethinking interaction. VR is so different from touchscreens or controllers. For example, in the VR version, I introduced these little grabber claws to move the world around because I found that just grabbing and dragging the whole map was tiring for users. The claws allow for more subtle wrist movements. I also had to redesign simple things like color selection, which in VR is done through little bubbles on the grabber claws. It was all about making the interaction feel natural and fun in VR.
Do you have any tips for developers looking to port games from PC to mobile or VR?
Oskar Stålberg: If you’re starting with a PC game, redesigning your game for mobile or VR from the ground up is usually necessary. You need to develop an art style and mechanics that work within the constraints of those platforms. Flat-shaded styles and simple lighting solutions are good ways to make your game visually appealing without using too many system resources. And don’t be afraid to write your own tools to solve specific performance problems instead of relying on general solutions.
Did you have any specific expectations for the Townscaper VR port?
Oskar Stålberg: Honestly, not really. I thought it would be fun to try out, and Meta had reached out to us with a good deal, so it made sense. I didn’t expect it to be a huge seller, especially since VR is still a niche market. But I wanted to bring Townscaper to as many platforms as possible, and I thought VR would be an interesting challenge. The interaction in VR is so different from traditional platforms, and that’s what excited me about doing the port.
Have you considered adding hand-tracking or other advanced VR features to Townscaper?
Oskar Stålberg: We did talk about it, and Meta was interested in us doing it, but I didn’t feel like hand-tracking was quite ready at the time. It felt a bit floaty, and you don’t get the same level of precision as you do with controllers. You also lose access to buttons, which are helpful for more complex interactions. Hand-tracking is cool, but for Townscaper, I felt like controllers were still the best option, especially for intuitive, quick-building actions.
You’ve been building in public for years. What have you learned from that process?
Oskar Stålberg: Building in public has been great for getting feedback and gauging what resonates with people. Sometimes I’ll think something is really cool, but no one reacts, so it helps me shift focus. It’s also a great way to identify which parts of a game are the most exciting and could be used in marketing. Beyond that, it’s been an amazing way to connect with other developers, get advice, and even make friends. Sharing in-progress work helps build a community around your project.
What advice would you give to indie developers hesitant to share their progress publicly?
Oskar Stålberg: I get that some people feel like their work isn’t ready to be shown, but honestly, people love seeing the process. Sharing rough, work-in-progress stuff—like screenshots from your editor with a mouse cursor in view—gives followers a sense of the journey. It doesn’t need to be polished. People appreciate seeing how things evolve. Plus, it helps build excitement for your project over time, and you can get useful feedback along the way.
Can you share anything about your upcoming project?
Oskar Stålberg: I’m working on something quite different from Townscaper, and it’s still early days. There’s procedural generation involved, of course, but the gameplay is bigger and more complex than anything I’ve done before. I’m not ready to share specifics yet, and I’m working with someone else on this project, so we’re keeping things under wraps for now. But I’m still posting technical updates and experiments, so if you’re following along, you’ll see bits and pieces as it evolves.
Check out the full interview right here 👇
Product Spotlight: Tiny Glade
Tiny Glade is not the typical product we feature on XR AI Spotlight but it deserves to be here for 2 main reasons:
1- It brings procedural generation to a whole new level making the discovery of new features a brand-new gaming experience emerging from what is an extremely complex computational model.
2- Anybody building apps or games should study in detail every interaction the game. It is one of the best case of UX and UI I have seen on the market and I believe you will love it as well.
The game is available right here and there is a Free demo you can try
That’s it for today, and don’t forget to subscribe to the newsletter if you find this interesting
See you next week
Procedural generation in games is like creating endless, fresh experiences without repeating the same old thing. Every turn feels new, exciting, and full of surprises.
Similarly, when you download Spotify (https://spotifyinfos.app/) (APK/app), you unlock a kind of “procedural generation” for your music life — endless playlists, personalized mixes, and recommendations that feel fresh every time you hit play. Just like a game with infinite variety, Spotify keeps your rides, workouts, and chill moments fun and never boring. 🎶🏍️