Olvand is a little multiplayer sandbox RPG, where the players live in self-built towns and can go on all kinds of adventures together. Imagine living with your friends in a small town in the mountains, or creating a new group of friends in a pub in the metropole you all live in. There will be several minigames the inhabitants of a server can play together, among which will be combat based games like King of the Hill or Capture the Flag. You will be able to play against other people in your city, or as a city against another city, or as a whole server against another server. The combat works with self-built guns, in which all kinds of powers can be combined to create unique effects. You can sign up to be a tester on olvand.com.
Some pictures of what I want to add in the future:
Hi all, all right then: the cave generator. The nice thing about generator algorithms is that they are almost standalone pieces of code; they don't depend on anything already in the game, and thus I can work on them anywhere, without access to my latest additions to the Olvand codebase, or even access to the internet. The cave generator is no different: I started it almost a year ago in the plane to Bratislava, where I had to be for a business trip, and then worked on it on and off for months in the train between my home and my job. I say months, yes, because it took my quite some time to get it right. Let me tell you why.
So what we have is a large piece of rock, and what we want is to 'carve' a nice and interesting level out of that. How to start? My first try was to randomly take out blocks, in the hopes that caves would form naturally. Well, they did, but most of the beautiful things that emerged turned out to be unreachable by the player. This picture should give you an idea of what went wrong:
Once I saw this, I passionately started to add code that kept track of which blocks where removed, and then made sure all of these blocks where connected. After several iterations this still didn't work as intended, and with each iteration my code had almost doubled in complexity. While planning for a next iteration that made things even more complicated, I realized this all should and could work in a much simple and elegant way, so I deleted everything and started over.
The better solution
The code is now based on something which I would like to call 'the mole': an object that starts at the entrance, and digs tunnels in random directions. After every intersection, the mole makes a random decision whether to duplicate or not. If so, the cave splits in two. It works more or less like this:
This approach hopefully looks super-simple and straightforward to you, but it took me quite some time to realize this was the way to go. And on top of this, while the idea might be simple, for the implementation there all kinds of details I have to take care of. For example, the mole should never leave the piece of rock I'm working with (otherwise you get a corridor that leads to nothing), there should be a boss room that is always reachable from the entrance, etc.
And besides this all, the algorithm also keeps track of what the various blocks looks like. For example, if the mole digs north into new territory, the block should behave as a vertical corridor, but if it discovers there already is a horizontal corridor there, it should turn this into a crossing. I don't want to waste client resources for this, so this is all done server side and then sent to the client as well.
To quickly see if a new addition works as intended, the generator can also show an ascii art version of what it created. This way, I don't have to load every creation into the game before I can inspect it. [Here] [are] [some] [examples] .
Unfortunately, I've discovered that building a cave generator algorithm is one thing, but going from the the blueprint to an actual cave you walk around in is another; there are several practical hurdles to overcome... but let's save that for next week :). See you then!
Latest tweets from
It can take up to a few hours for tweets to begin appearing.