Citystate Metropolis - Development update and first screenshots

Finally, here they are- Citystate Metropolis first screenshots.

I apologize for the radio silence. Things have moved more slowly than I expected, the development itself has consumed all my time and attention. With this post I'll try to address concerns from the community regarding the changes that can be expected with this new approach. I hope these updates and first official screenshots can make up for the wait. 

A quick update on my progress so far. The development of Citystate Metropolis started in December 2022, right after Citystate II final patch was released. This means I've just passed the 2 years mark, exactly the amount of time CSII took to develop... Not surprising really, Metropolis is taking more time than I thought. For one, it requires significantly more development time to build interactive systems such as 3D construction and design tools than it takes to build stats-based systems (such as sliders and checkboxes). But this is what it's all about, moving away from a spreadsheet approach and towards an interactive and player-centric approach. Moreover, I didn't design my two previous games with a long-term plan in mind but that's not the case with Metropolis. I want to make sure that this game can be worked on for years after launch. This means that the core features have to be designed for long term maintenance and expandability. And for that to happen, I need to make sure the most basic systems such as road and zoning tools are as good as they can (realistically) be.

So far here are the three fundamental and most complete parts of the game.

TERRAIN 

The terrain is procedurally generated, or rather semi-procedurally generated. I’ve had a love-hate relationship with procgen terrains for years. In fact, CSII initially used procedural maps but I eventually got fed up with their artificial-looking mountains (the fractal patterns, the lack of map-wide river beds, etc.) and decided to create all the maps by hand. However, this time I think I found a good balance between real world data, handmade stuff and procedural generation. Here’s how it works, roughly: The map generator first picks a high-resolution real world heightmap with lots of details. It then 'multiplies' it by a hand-made lower resolution heightmap with larger scale geographical features (think island, peninsula, coasts, etc.), and then it adds on top several layers of noise and a layer of computer generated erosion. The vegetation is then added similarly as in CSII, from a mixture of noise and rule-based generation. And finally, the sources of rivers are added from pre-recorded and manually set positions. While not truly infinite, this allows for thousands of unique maps, real world topology, and tremendous options for player customization. As a technical note, the map is 67sqkm (about 6 times bigger than the biggest map in CSII). 

ROADS 

Roads are probably the aspect of CSII I was the least satisfied with. The code was anything but elegant and did not allow for any room for expansion. Moving away from a tile-based system, I’ve switched to a more contemporary spline-based approach, allowing curved roads. Even though this system was initially faster and easier to implement than the tiled approach, it became exponentially more complex as soon as traffic and interactions with buildings had to be implemented. Commutes and pathfinding are one of the most complicated aspects of the development, especially given that all the computations happen each frame and had to be multi-threaded. However, at the moment it can already support hundreds of thousands of agents simultaneously, supported by a GPU instancing technique and other boring optimizations. The agents (the citizens) are no longer just short-lived visual representations like in the previous game but carry with them real data from the simulation systems. The citizens will not be individually simulated like in Skylines; the simulation will happen at the level of households and buildings. The citizens do not interact directly with the city by going from place to place. Rather, the simulation (and more precisely, the building they live in) dictates their movements, wealth, education and progression. The main advantage of that approach is to be able to simulate at least one million citizens without having to cap the FPS at 30... But most importantly for me, it prevents a micromanagement hell and the long list of cheats and tricks a pure agent-based simulation inherently comes with. In my opinion, having the simulation work at the level of the buildings is the proper level to simulate a city's social dynamics with more pragmatic and real-world results. I could write a thesis about it, but I guess I have a game to finish first. 

BUILDINGS 

Residential, commercial and industrial buildings are procedurally generated in Citystate Metropolis. Land lots no longer need to be square either; the zoning tool can fill any type of polygonal or curved shape and allows for more ‘organic’ looking neighborhoods. The buildings’ general shape is still somewhat rectangular since their actual land footprint is based on a grid inside the lot. However, the assets used to feed the procedural generation can still have curves and more exotic shapes, giving us a nice balance between customization and realism. 

What I really like about this system is that it allows us to directly draw where the building should be, just like a city planner would. Setting the size, shape and position of the building, and how the rest of the land lot should be used, tile by tile, how much parking space, how many trees, etc. More importantly, these choices are not just aesthetic, they will affect the affordability and attractiveness of the building. Here, the player needs to find the balance between esthetics, environmental and economic needs. There isn’t a single winning strategy, it all depends on what the player wants to achieve: cheap unregulated housing for fast growth and poverty reduction, or eco-friendly super modern design for comfort, beauty and international attractiveness. Here, this is exactly how Metropolis differs from the previous games. Instead of setting policies with sliders and checkboxes, we design the city with powerful urban planning and architecture tools; we set the rules from a creative approach rather than a bureaucratic one.  

In the end, the level of customization we want to work at is entirely up to us. It's possible to design each building in detail one at a time or just choose a preset and zone multiple lots at once. 

On a side note, I believe that one can’t create 'the most realistic city builder ever' without addressing meaningfully socioeconomic issues: the relationship between education level and income, income disparities, spatial segregation, etc. After all, these games do shape how we think about cities!

2025 

The game will be released between September and November 2025 in early access. Before that, there are still major parts of the game that need to be completed. Public transit, service buildings, modding support and the nitty gritty of the actual economic systems. But I remain optimistic, the foundations are looking good and it should allow for things to move rapidly. 

My next goal is to have the steam page set up and ready with a short trailer within the next 4 weeks. The soundtrack is already complete (3-hour long!) and I’d like to show it off soon too. In the meantime, you can subscribe to the newsletter at the bottom of the page to stay tuned with the upcoming news. Wish me luck!