Advertisement

Need some input for new project: Orbis Infinitas

Started by August 05, 2013 04:38 PM
6 comments, last by wodinoneeye 11 years, 5 months ago

Orbis Infinitas will be a procedural generated roleplaying game, where everything will be generated based on a given seed at the game start. This includes the world itself, cities and towns, dungeons and forests, empires and politics, characters and monsters, but also the whole story, background and quest system. Everything will be generated on the fly and a player can be part of a world that seems to have no borders. You can walk and walk into one direction and maybe you will never find the end of the world.

The graphics will be in a style you may know from older rpgs some years ago. If you can remember games like “Secret of Mana” or “Terranigma” from the good old SNES, you know what you can expect from this game project. I do not want to do the best graphics out in the world, it’s more important – in my opinion – to have the best gameplay and a well done procedural generated game structure. Additionally I just like the good old graphics of the golden SNES age and I’m sure that you do not need modern 3D grahics to create a good and successful game.

I’m not a native speaker. I’m from Germany and so German is my first language. So excuse me, if I do not write in perfect English. I hope you just understand what I want to say… ;)

I put up a webpage to show the progress of the project:

http://www.orbisinfinitas.com/

What sort of input is it that you are looking for exactly?

Advertisement
I know that I am working at a massive project. I already have some experience from creating one of the biggest Elder Scrolls Oblivion mods called "Adash - City of Magic" - but this project is far bigger.

The input I'm looking for: I'm sure some experienced people here can tell me if my goals are realistic and where there are problems I can not see yet. Further I just want to have some opinions about my project and my ideas, that might help me.
You describe "standard" graphics and client-server communication which have been done many times before, without any special requirement. Therefore you should assume such aspects will be easy (at the very least, easy to find help for) and concentrate your experiments and prototypes on the difficult side of the project: automatically generating and storing a large and complex world. Apart from automatic generation of something fun, the great technical challenge of a large-world procedurally generated RPG is the tradeoff between regenerating things procedurally, losing the game's memory and limiting what can be done, and storing data and running simulations, at unbearable memory and processing costs; you'll need to design your game to forget information as early as possible, but not so early that the player notices artifacts.

Omae Wa Mou Shindeiru

It's doable but not easy by any means. There are a few examples of these type of games, the most famous being Dwarf Fortress, SimEarth, Spore, etc.. Look into how the sim games use cellular automata for their world creation, i believe Dwarf Fortress uses something similar. The newer Sim games use agent based systems to simulate their worlds, which is just a few steps up from the cellular automata model.

Also u might do well to look into genetic simulations for "evolving" the AI and wildlife esp their behavior, there is a host of research done on that topic.

Good Luck!

Your game idea needs a lot of technology that is not a game. The procedural stuff will be very hard to make, and might not actually improve the game. Maybe the seed you choose will start you in a vast wilderness, far from any cities or dungeons. Maybe your procedurally generated story will be boring. A lot of times, my Dwarf Fortress worlds will be terrible, with all the empires destroyed except the elves, and all the dragons and monsters killed. Nothing good. Just elves.

You will have to work very hard to be sure that each of the infinite possible worlds is fun and engaging. Maybe you should make it more like Diablo. The towns and NPCs are set by writers and map designers, but the wilderness areas and dungeons are more random. You could have a starting town that's designed by you, and have a couple places that are important to the basic story built like that, too. Then have the algorithm design all the other towns, dungeons and wild places on the fly. That way, you design the skeleton of the game, to make sure it has the right shape and does the right things, and then let the computer put the meat on, filling the world with new content every time you play.

Advertisement

If anything, I'm interested.

It sounds like programming the heuristics to such a procedural generation would be fun. I can see several ways to go about that, but you need to have a strong base.

Have you thought about the inherent logics that govern the generation of continents, etc?

It would be a mistake to start by just randomly generating bits of land.

I've seen on your website that you've started to generate coastlines. I believe you're not being hollistic about your understanding of the challenge, and by doing "fine-tuning eye-candy" first, you might seriously impede your ability to deliver on the core underlying principles of your terrain generation logic.

I'm currently working on something loosely similar, where relative positionning of elements in a procedurally generated environment is critical, so I do have a grasp of the basics and could avoid you the trouble of making similar mistakes if you'd like to discuss this a bit more in-depth.

Good luck!

You will find that unless the whole game is to be a huge homogeneous mess of uncohesivec/conflicting elements/placements you will have to base it on using a complex/large hierarchy of preexisting templates/patterns/generation-rules to make what is 'on-the-fly' look like it fits together aethetically as well as functionally.

When you start dealing with ecosystems and worse societies of people things start getting very complex in the way individual objects relate to each other. You cannot just throw things together without it looking wrong or for it to make any sense to be playable.

So to do what you are asking, there will only be limited 'random' elements and most of the development will be groupings of objects for your world which belong together, which then at some level can be mixed/substituted properely together for variation or local diffences/themes

--------------------------------------------[size="1"]Ratings are Opinion, not Fact

This topic is closed to new replies.

Advertisement