Fantastic start - and it is extremely unlikely he will use more than a few kilometers at a time. But part of the fun is making it procedurally generate till the end of time (or floating point errors).
My game is going to have semi-procedural terrain. It’ll be a small chain of islands, so it will generate in the beginning for those islands only. After the start of the game no new terrain is generated.
I have 2 questions for you though
1: How was the grass done?
2: How do you plan to have AI navigate the terrain? Generating a nav mesh takes too long for me in real time, even when I pre-calculate it at startup.
(thankfully the only enemy in the game will be swarms of braindead zombies, so the inability to give intelligent chase may help)
Very interesting, especially the fact that the game will be randomly generated, it’s ages that I have in plan to do the same thing… with today’s calculation power it’s finally surely possible, you only need to set up things very well to not find the map with impossible places and such, but I really think it can be done and Unity is a great tool to achieve this.
i think the demo space is large enough in my opinion, i was bored at the start but didn’t expect to see mountains and valleys which i spent a good 10 minutes exploring, nicely done look forward to updates. and this game has good performance.
It doesn’t require that size, it’s more of a “because I can” thing. And I love exploring, so I plan a game where I can explore to my heart’s content. It all depends on how interesting I can make the land…
1: There are several “Detail bands”, that specify DetailPrototype to use, amount, height at which the grass appears, preferred terrain type and “affinity” to slopes. Then for every point in detail map, a random amount of grass is generated, weighted by all these factors.
2: Ooh, good catch! I haven’t really thought of that, and that might become a problem… I guess I’d have to use some kind of steering instead of navmesh… need to think about it, hard (-8
Well, the trees are meant to not spawn on steep slopes: there’s a weight component in generation that makes a tree on a slope highly unlikely. But sometimes they still end up on steep parts, probably because of a bug in my slope calculation, or maybe low heightmap resolution allows for some weird spots…
Great start! You should render a skybox from your landscape, so it looks like it actually goes on forever (you can see the edges of the map from the top of high hills)