Amberspire News

[p]Note: these design diaries were originally published in spring/summer of 2025. The design, and definitely the screenshots, have changed a lot since then! I hope you'll find them an interesting look in the design and development process, and an insight into what does or does not change over the course of development. - nic[/p][p]I've been playing and thinking about city builders for a long time, and along the way I realized I wanted to do my own spin on houses and population. This was partly born out of thinking about cellular automata, but also realizing that typically in city builders I am not engaged by placing houses or residential districts. Unless I am giving myself a very specific external motivation for doing so (a visual appeal or extra challenge) I'll just do the bare minimum the game requires and nothing more. In a sense, I'd like the population to be either something I think about a lot less or a lot more.[/p][h3]Population[/h3][p]City builders use population as a kind of overall set of buckets for the systems the game is modeling: happiness, wealth, education etc, of which Amberspire does not have any. One reason for this is that the focus of the game is broader, the city in Amberspire can grow to be a massive metropolis, the interactions with the world around it are more political in nature, so we didn't want the design to be so complex as to track individual people. Another is that typically these systems heavy city builders are built on the assumptions of the 20th century, and typically in industrialized areas. Even with a displaced time period or location, I feel like the assumptions are still present. Which, in itself is neither here nor there, but I want Amberspire to feel like a science fantasy city builder, beyond just unusual names and architecture.[/p][p]Given other production constraints, I did not have the capability to design and implement the ideas I had for how a city several millennia in our future might operate – though those ideas are present in the writing and art.[/p][p]All that to say, I wanted the representation of labor, population, and people in Amberspire to be relatively simple, out of the player's control, and reinforce the science fantasy theme.[/p][p][/p][h3]Labor[/h3][p]Initially I was very confounded with how to represent labor in Amberspire. A typical city builder will track workers or some kind of value within a building, and modulate the building's output based on skill or morale or whatever else the game is concerned with. Since the focus of Amberspire is not that fine-grained, and building output is already variable with dice, how to should the game represent labor?[/p][p]Some ideas that didn't go anywhere, to start things off: dice themselves could change, with faces becoming blank or have some penalty modifier if rolled. Or when the resource is used in another building, its effectiveness is modified somehow.[/p][p]These probably could have worked, but didn't feel right and clashed with other plans I had. I didn't want the dice themselves to be overloaded with complexity or be so variable that the player couldn't guarantee what the die would be – before it's even rolled! This kind of 'double uncertainty' can very quickly go awry.[/p][p]Another promising idea: houses were buildings that rolled a ‘labor’ die, and to construct buildings you’d need to supply labor resources from house die. This somewhat works, but not at the range of scales I wanted to operate with. For a huge space fantasy metropolis, a building could be the neighborhood market or an entire spaceport. If a few ‘labor’ resources can build a market, then dozens would build a spaceport. Rolling dozens of dice was out of the question, a firm constraint for the game.[/p][p][/p][h3]Simple and Out of Control Houses[/h3][p]My solution is to both of these issues is first to remove the house building from the player’s toolkit altogether. As a player you will never just ‘build houses’. Residents, as houses, grow in the city as a function of you operating other buildings or various other events. Residents also don’t roll dice at all. When you construct a building, it must be able to reach a certain number of residents within a range of that building. While that building is under construction, those residents are reserved and unable to be used for constructing another building.[/p][p]Removing house building removes a key interaction for the player (which is the part I typically find unengaging) but adds a distinct texture to the game. Like cellular automata, residents grow semi-randomly, sometimes filling in gaps and sometimes creating unusual shapes. Buildings you construct as a player will affect an appeal value for spaces in the world, so the player does have some general direction, but not totally.[/p][p]The unusual shapes created by the residents will force you to build in odd and unconventional ways, working how you can with the residents. My goal here is to feel like the city has truly interesting districts and neighborhoods, arising both out of whim and chance and your desires as a player.[/p][p]The restriction for constructing a building addresses both the representation of labor and nice limitation on player construction together. It’s also an idea that is easily and immediately communicated to the player: you have strong but clear limitations about where buildings can go, and how many you can place. Buildings can also vary a lot in their needs, the aforementioned market only reserving a few adjacent houses while a spaceport requires dozens over a huge area of land.[/p][p]The population and neighborhoods will take on a life of their own exactly because the player didn’t place them, but watched them grow out of the city. My hope is that cities will have truly unique and unexpected designs because of this feature, and I’m very excited for it.[/p]

[p]From a design perspective, the factions and events form the third major element of the game, alongside the dice and the environment. The goal is to create a narrative quality to the game and city that is missing from the other two parts.[/p][p]The dice and environment provide a large and detailed systemic representation of the city. There's many little stories that come out of them, but they represent the city and environment as a series of processes with relatively fixed inputs and outputs.[/p][p]The factions and events represent the people, communities, and ideologies in the world, and they are weird! The events are meant to depict what the systems can't - strange occurrences, people being goofy or aggressive or helpful. The events also describe much of what happens beyond the city, in other settlements in the solar system and in systems beyond. This layer is something that narratively you are aware of as you control the city, but do not have direct control of.[/p][h3]Events[/h3][p]Designing the event system was tricky, and a lot of ideas came and went. I wanted to keep events and factions simple so that the focus remained on the city building, and so that it did not feel like the factions were playing a game without the player. A typical 4X or strategy game has the player and AI on (roughly) equal footing mechanically to simulate the idea that everyone is at the same table playing the game, which was not the goal here.[/p][p]Skipping over the dozen discarded ideas, the factions came down to being represented by two values: strength and disposition. If a faction is strong enough and has a disposition towards you, a mechanical effect is applied to the entire city at once. These can be positive or negative, but greatly alter the various rules of the game like weather, terrain, dice, and more. Events are now closely tied with factions and change strength or disposition of that faction, or others.[/p][p]The ultimate benefit of the simplicity of this system is communicating it to the player. To take a discarded idea of factions owning buildings and the you interacting with them: how is the state of the building's owner communicated to the player? Floating icons, art tinting, custom art, all have pros and cons that for the moment ruled out these ideas.[/p][h3]Influence & Dice[/h3][p]So factions have strength and disposition, events tell little stories to adjust those values and affect the game elsewhere. A few different ideas on how to incorporate dice came and went, with the resulting idea feeling very obvious in retrospect. If you can afford to you can choose to pay influence to select an event option, otherwise a die is rolled.[/p][p]A few other options here were interesting, like a special faction die to choose a type of event, or the player paying to change the faces of die to better guarantee the outcome they wanted. Ultimately simplicity won out, similar for reasons above.[/p]

[p]The first production decision made for Amberspire was choosing to render the game in 2D or 3D. By training and experience I am a 3D artist and comfortable with 3D games, so this seemed like a natural starting point. Once I actually got started on the prototype however, we decided to do the exact opposite: go full 2D and isometric.[/p][p]Bazaar[/p][h3]The Details[/h3][p]Cities are dense and detailed environments, and producing detailed 3D art is a lot of work. You need to conceptualize and model walls, roofs, windows props, domes, arches, tunnels, plants, – everything you want to see. Then unwrap and texture all those objects, before creating shaders for all the materials to render them. I have experience doing all of these things, but our goal was to keep the team size and production close to The Banished Vault - which had comparatively few small and simple models. Amberspire would need a few dozen unique buildings, all detailed and intricate.[/p][p]All this to say 2D art is not by any means easy, but it does have many fewer steps. It solves a lot of production problems all at once: once an artist draws the image, it just has to be placed on screen.[/p][p]More subtly, however, is an expected level of detail when comparing 2D and 3D. I did not want the city to feel like it was being represented abstractly, like buildings in The Banished Vault. With a space fantasy setting, specific details are important to communicate the setting and world to the player. My worry would be that with a setting so unusual that alienates the player a bit, art that is further removed from the 'accurate' depiction of the setting would be hard to understand and engage with.[/p][p]One small bonus here as well: 2D art is more iconic. Everything in a game is an interface the player, and which building is which is an important piece of information for the player. Making a 3D building recognizable is doable, but just that extra step harder than in 2D.[/p][p]Clockworks[/p][h3]The Style[/h3][p]The science fantasy mode that Amberspire is in is communicated through the art, and it looks best in 2D. Hand drawn, irregular, illogical, equally playful and sinister – I could go on. When our artist Yue started on the project one of the first things I said to her was “don’t worry about making the buildings perfectly logical”.[/p][p]There is a je ne sais quoi to 3D art that implies some kind of underlying logic. To make a closed shape, vertices have to be connected in a mathematical space and be rendered by the engine. Of course you can make illogical 3D art, but that comes with other tradeoffs.[/p][p]In Amberspire, the buildings are fully realized. They explain exactly as much of themselves as Yue and I want them to. The setting overall has many magical and fundamentally impossible aspects, and the art style and rendering are vital in communicating that.[/p][p]Temple[/p][p]
[/p]

[p]We've been quietly working away on Amberspire for a little while, and it's finally ready to be shared with everyone soon.[/p][p][/p][p]Amberspire will launch on Steam on May 6th 2026![/p][p][/p][p]We'll be posting a whole lot of sneak peeks and previews on our socials (@BithellGames on everything) between now and launch, so be sure to check those out in the meantime.[/p][p][/p][p]We're so excited to get Amberspire into your hands. Wishlist, tell a friend, and we'll see you soon![/p][p][/p][p]Watch the teaser now![/p]

Cellular automata have fascinated me from the moment I learned about them. How can simple rules and constraints can produce astounding complexity and unpredictability? Or even a simple recurring equation like the logistic map producing random noise, when tuned to the right value. It feels clipping through a wall in a game and seeing the underside of reality. A decade ago these ideas lodged in my brain and fused with the overwhelming desire to move from the suburbs and to a city, and I've been thinking about the two together ever since.

[h3]Cities of Life[/h3]
So over the years I’d try experimenting with cellular automata to create a city builder, somewhere between a game, toy, and visualizer. I tried to hold on to a powerful idea that would eventually prove impossible, that the player could adjust the rules of the automata system to change how the city would grow. If a player wants more farms, they can tweak the conditions under which farms grow, or denser/sparser street networks, and so on. It's easy enough to experiment with how this might work (1) using Conway's Game of Life (GoL), adjusting the thresholds for a cell's birth and death.

If you experiment with GoL you'll start to notice an issue: frequently a ruleset will set every cell in the system to alive or dead. Some rulesets won't do that, but after a few generations will stabilize in a random but completely static pattern, no longer exhibiting the intriguing fluctuations of Conway's specific rules.

From the point of view of making a game that someone can play, the variants on Life’s ruleset are not qualitatively equal. Conway’s ruleset is the optimal one that produces interesting results. What took me a long time was realizing how common this problem was across many types of cellular automata.

[h3]The Big Problem[/h3]
The appeal of a cellular automata system is its downfall when trying to put it into a game like how I imagined. That appeal - that a small set of deterministic rules can unexpectedly produce anywhere from structure, chaos, nothing or everything - is distinctly unappealing when applied to traditional game design and structure. This is the shared unsolvable problem (2) that I found, at least for me, my style of design, and my goals.

Generally speaking, players want the outcomes of their actions to be predictable, with some margin for uncertainty. Maybe the uncertainty is large but its effects are limited (a to-hit dice roll) or a small uncertainty not realized for a long time (several moves in Chess).

Broadly, types of cellular automata systems might be more or less predictable. I was exploring and designing various systems, trying to find one that allowed the player to adjust rulesets within them to be variously interesting and function as a game. While each ruleset had benefits and drawbacks, it was very hard to find one with a wide variety of qualitatively interesting rulesets. Many would respond like GoL does, just shutting down entirely.

If the player is fiddling with the rules, they might create something that has too much immediate uncertainty, or not enough. Or something that only starts producing random results after 1000 steps of iteration. Effectively, the result of adjusting rulesets in itself becomes too randomized. The outcomes don't fall within that narrow band of uncertainty that makes games interesting.

The appeal of decision spaces in games is that players can slowly grow to understand most if not all of the possibilities within it; the appeal of cellular automata is that a system can sometimes produce nothing, gibberish, or structure.

[h3]Moving On[/h3]
Now, you might be thinking, hey I can name a few games that use CA, so learn from them! And you’re right! Everything above is a historical realization I had maybe five years ago now, and only really applicable to me and a very specific goal of a certain kind of city builder. At some point I just had to set this idea down and move on.

What's funny is I do still think about games as non-linear systems and frequently try to analyze and design games with this method. Chess (or Go, etc) as a system is almost purely deterministic (the random elements are the players themselves) and produce highly random outputs. Solved games like tic-tac-toe do not interest people – we're interested in games (3) where we don't know the outcome.

This provides some key background to my thinking as I approach Amberspire. Cellular automata, chaos, and nonlinear systems have influenced me tremendously, and there are ideas in the game that are downstream of these. They are not quite true cellular automata, but there is chaos arising from order and player input. In future design diaries I'll be talking about the population and terrain tiles where these ideas are strongly represented.

- - - -

1 An online tool for experimenting with Game of Life (this can produce flashing images, increase the round delay value to mitigate this).

2 Or cursed.

3 Extend me some grace here about the definition of 'game', I am pretty aware of edge-case definitions and take them seriously!