Posted Mon, Jan 06, 2014 by Veluux
As both a Minecraft and EverQuest enthusiast, it would be a terrible mistake if I attempted to put my excitement for EQNext and Landmark into words. It just simply wouldn’t do my emotions justice. Landmark in particular is shaping up to be an excellent combination of those two games, with no small amount of unique features that will give it a very distinct feel of its own. I’ll be covering many of those various aspects, mechanics and game features over the coming weeks; but one in particular has been nagging at me lately.
I’ve seen so many different feelings expressed about SOE’s choice to use a procedurally generated world for EverQuest Next, across a broad spectrum ranging anywhere from adoration to disgust. I felt that an accurate description of what procedural generation truly is (as well as what it is not) was in order; along with some of the positives and negatives to implementing it. I’ll also take a quick look at the specific use being employed for EQNext and Landmark within its version of the Voxel Farm engine.
Simply put, procedural generation is the systematic production of digital content that can be used in a multitude of applications. In EQNext and Landmark it will primarily be implemented for terrain generation (which is where people usually see proc-gen visually occur, despite its various other rending uses).
Terrain-based proc-gen is a mathematical system used to take data variables and manipulate the visual layout of a given world. It can be subdivided down into regions, areas, zones, or what-have you - and even those can have procedurally generated traits that describe where and when they occur.
This system of adjustable values makes the process of world creation extremely simple and straight-forward. In fact, it extends the visual decision making of the world-creating process into not just the hands of the artists, but the programmers and coders as well; because everybody knows what looks believable and what doesn’t. Instead of requiring artists and programmers to work (painstakingly hand-in-hand), now coders can look at concept art and continually tweak the variables to get the computer-generated terrain to match the concept and style the artists and designers are looking for.
Totally random. Worry not, those of you that fear floating islands, upside down trees, and purple clouds. Procedural generation is a controlled process. How much random creation that is produced is directly under the control of the development team. All the variables, and the relationships between them and be adjusted to control and prevent nearly all oddities and extreme renders. Making the terrain generation more detailed, specific, and accurate - is mostly a matter of time-commitment - and in direct relation: budget. For a company as big as SOE, and a franchise as storied as EverQuest; I don’t think the budget (or even time) will be an issue.
However, procedural generation is also not totally realistic. No matter how large the budget or how much time you have to make adjustments, replicating a real-life forest or desert perfectly is going to be impossible; but all we really need is to maintain a suspension of disbelief. So partially unrealistic shouldn’t be an issue, and for a fantasy game - that should actually be a goal.
One of the greatest benefits by far of procedural generation is how time and cost-effective it can be for creating large-scale worlds. Instead of modelers having to hand render and hand-place every single terrain object, the process becomes automated. That leaves more time and money to spend on other things. I also want to note that this doesn’t just mean non-terrain related things. Freeing up the artists and 3D modelers means they have more time to spend on making the automated terrain more believable. More grasses, flowers, types and styles of trees, as well as other terrain-related props. The less time they have to spend doing ridiculously menial tasks like placing trees along the edge of a massive forest (that very few will even venture over to see), the more time they have to spend on everything else.
One of the other major bonuses of procedurally generated terrain is how much it opens itself to dynamic environments. You want to see the world change with the seasons, meaning snow, autumn leaves, dormant trees, or spring flowers? Then procedural generation is your new best friend. Now when a new season rolls around, new models or additional models can be “subbed” into the same algorithm (that will produce the same terrain, rock & tree placement, etc) to produce a completely changed environment. How cool would that be? That brings me around to one of the final benefits of having automated terrain generation: It “feels” real.
It doesn’t matter if you believe in Evolution, Creation, or some other form of universal origin. The world around us has a pattern and an order to it that is truly mathematical in nature (even if we don’t completely understand the equations behind some of the more complex aspects). Procedural generation is the digital method of replicating nature - which is, itself, a system of rules and laws not unlike a terrain-generation algorithm. Water flows downhill. Trees of different types can only grow in differing proximities of one another (due to root systems, soil quality, rainfall, and other factors). When you see a pro-gen’d environment, your mind naturally gravitates to its believability. Minecraft is a perfect example; even in a world of cubes and blocks, our brains want to believe it.
I’m a huge proponent of porecural generation - for all of the above reasons, and many more. Perhaps it comes from my artistic experience and background; I know how bad it kills the motivation to be working on the “un-fun” parts of an artistic endeavor. It can be a huge burnout factor. That being said, it doesn’t mean there aren’t some negatives to procedural generation of terrain.
If not handled properly, or given enough detailed variables, pro-gen’d terrain can feel repetitive and predictable even when it is truly unique and different. That makes things bland, and bland is bad for gamers. We want to escape into a new and interesting world - not a boring one.
Procedurally generated terrain is also difficult to perfect. Sure there are lots of variables you can adjust, and you can constantly add new parameters to rein things into the realm of realism, but you’re never going to get it exactly right. There will always be some parts of an environment that just don’t quite fit. It’s inevitable.
So are anomalies. They happen. They are the black cat of The Matrix. Fortunately, like most other glitches and bugs, they can be edited out. Procedural generation does not deny permanent hand-made changes. The problem is that they typically have to be observed first. Fortunately for EverQuest Next, Landmark should drastically help iron out any procedural glitches, bugs, or anomalies that might pop up - since we will be seeing the exact same terrain that will be featured in EQN (per comments made by the dev team).
We’ve covered what proc-gen is, and what it isn’t. We’ve also covered some of the pros and cons of implementing it in a game. We know we’re going to see it heavily used in Landmark; and we’ve got confirmations that it will also exist even in EQN. So you should be glad to know that Landmark and EQN will likely be some of the largest-scale MMO worlds to date. When the developers keep hinting about how massive the world will be, I lean more towards examples like this:
That video is an early, early sampling of the Voxel Farm engine (which EQN and Landmark will be featuring) rendering an huge, mountainous environment. The viewer is taken on a nearly 20 minute journey descending just one massive mountain. Even if the mountains in EQN are half this big, imagine the potential for a humungous world that doesn’t just feel, but actually is truly enormous?
I hope you enjoyed the read and got to learn a bit more about procedural generation and where it could potentially go in the hands of SOE. I’m looking forward to exploring these massive proc-gen’d environments within the world of Landmark and the wilds of Norrath in EverQuest Next; and I hope you are too. I’ll see you out there - if we can find each other!