When we think about encroaching Goliaths like Everquest Next, we tend to think in the macro: high-powered executives, valedictorians from the best colleges, Skynets-in-training. Yet, at the heart of the MMO everyone seems to be talking about lately, there sits a pair of indie outfits: Storybricks and Voxel Farm. If either of those names seem familiar, it might well be because you've seen Storybricks' ill-fated Kickstarter campaign (and the numerous mentions of them in the media after that) or know of Voxel Farm's function as the vertebrate of the highly ambitious indie game Starforge. And while Sony has definitely done phenomenal things with the technologies provided by these two engines(Did you know that Voxel Farm is being powered by a one-man maestro of procedural generation?), it's still ridiculously cool to be able to say: Indies did this.
At least, in part.
"The start of it was about building emotional A.I." Brian Green, a veteran of the MMO scene, began. "We wanted to give the spark of life to characters so that when you went and interacted with them, they wouldn't just have pre-scripted actions or just stand there looking static. We wanted them to react. You know, if somebody got poisoned, we wanted them to be afraid and want the antidote. If a guard saw a mugging, we want his sense of justice to kick in and go and stop that mugging."
It sounds like an MMO alchemist's dream, an ideal as unattainable as the Holy Grail. However, late last week, Sony publicly announced that the presence of emergent A.I in Everquest Next. Though the proof is in the pudding, the idea was, at the very least, real enough to place into the hands of the media.
According to Green, Storybricks is currently working on assisting Sony in the development of combat A.I and ambient A.I for the upcoming MMO. "What we were working on as a company before that is also what we call 'emotional A.I' so that you know, you can set up a character to have an emotional reaction. What this is generally portrayed through is body language or facial expression and this is one of the exciting things about the style of art for Everquest Next. It has very expressive faces so that way you can see if the character's happy or sad or worried. And so, we're looking at ways to kind of exploit that for other aspects to make that with the goal of making of basically making a living world that people can experience."
Director of Development David Georgeson explained during the reveal that it was possible to install 'tags' on the various critters populating Norrath. Is there an upper limit as to how many of these characteristics can be pinned to a resident orc? Green says that there can be as many or as few as the designers desire.
"I think the really emergent theme is that you don't need very many. Like Dave Georgeson said in the reveal: you need basically one attribute. Desire for wealth. And that influences how Orcs - they will look for where the wealth travels - move along paths. We want players to say, 'Oh, Orcs like wealth. Therefore, if I want to hunt Orcs, I look for places where there's wealth. A caravan carrying money between these two cities - I'm gonna go along with the caravan because I want to hunt orcs and I figure Orcs will probably be attacking this caravan.' If you get too many of these, things can get a little confused. Our mission is to be of help to the designers in deciding how many of these attributes they want to make. But, you know, it's potentially limitless."
In spite of the exciting future, Storybricks had relatively humble beginnings. May last year, the company launched a Kickstarter that did not quite make it to the finishing line. Green says that one of the problems they encountered then was marketing. "We had something that is very exciting but I think we weren't able to communicate that terribly well. What happened was that when we got to talk to people individually, they would get very excited about it as soon as they understood but that somehow didn't quite come through with the campaign itself."
The original vision of Storybricks differs slightly from their current endeavor. Today, the company is working hand-in-hand with Sony to come up with code specific for the game. Green was quick to note that they were not helming the direction of the game but simply providing suggestions as to its course.
Having said that, it's certainly intriguing to see how their influence might help mold Everquest Next. Green went into further detail about how combat may possibly work. " What we want to do is we want to have combat to be, you know, an encounter. Something that's interesting to the player. One of the things we're working on is basically different types of monsters. Tying into something that we were talking about, if we have a monster type that is going to be timid, they might not try to engage in direct combat. Maybe, they're going to do something like lead you through traps or maybe try flanking maneuvers or something like that. You're going to have to adjust your tactics to what kind of monster. And as you adjust your tactics, they're going to adjust their tactics. What we want is to have it so that combat doesn't just become 'I see something, I hit the right buttons, and then it dies and I take the treasure.' We want it to be an interesting part of gameplay."
There's more. "It might also be something like maybe if you're fighting something and a second monster comes and the second monster hates the monster you're fighting more than it hates you, it might actually join in on your side. This is a possibility; something that could happen. It depends on the situation."
What is most fascinating, perhaps, is that the local fauna may well be driven by more than simple desire. Green explained, "What we are thinking is that basically all orcs will have certain personalities and then maybe have a special orc that is a part of the story. Maybe, an Orc that's sorrowful or regretful who feels differently about things."
"That is something we could do and it's up to the designers if they want to do this." Green added, quick to impose the necessary disclaimer.
As a result, sieges, should those arise in the final product, may well feel like actual sieges as opposed to endless waves of single-minded hostility. "It's not necessarily going to be a bunch of enemies pounding on the wall but it could be if you have a bunch of unintelligent enemies. They might just go and beat on the doors and get picked off easily. But, if you have more structured, warlike Orcs, they might bring siege machinery or, like Dave said during his example, the Valhallas. They might bring siege machinery to actually try to crush the walls and then as soon as there's a breach in the walls, they'll get through. What the enemies can do is going to depend on the situation."
We pushed deeper into the viscera of Storybricks. Originally built on Unity and Python, Storybricks now incorporates languages that Sony is using. As complicated as it all might seem on the outside, Green's explanation of how the technology works seems almost disarmingly simple: "We're using a lot of influence maps. What an influence map does is it looks at the land and says 'Okay, where are the points of wealth?' And that's the highpoint and it kind of fades off from there. So, we've got two points of wealth and if somebody walks into the influence, they're going to be attracted to that. They're going to try to seek out the middle of that. A lot of what we're doing for the other stuff, like the emotional stuff is basically very much unique to the stuff we're working on in Story Bricks. There hasn't been a whole lot of game AI that has taken that direction."
Miguel Cepero speaks on Voxel Farm
"My real background is in electric engineering, automation and things like that." Cepero remarked over an impromptu Skype interview.
To recap for those who have yet to hear about it, Voxel Farm is a jaw-dropping, voxel-based procedural engine capable of, among other things, procedural terrain, a voxelization component, a networking component, a sandbox edition and a voxel instancing system. Though Everquest Next possesses a more stylistic art direction, Cupero's Voxel Farm, as evidenced by the sample videos on his website, seems amply capable of photo-realism.
How did someone with background in electric engineering progress into the creation of an engine like Voxel Farm?
"Early in my career I saw you could do a lot more with software and then I became obsessed with creating virtual worlds. Games is a big part of that because they always need some sort of virtual reality."
While many would think that Minecraft served as a large inspiration for Cepero, Voxel Farm's creator was already at it long before; the engine has been in development since 2008. Nonetheless, Cepero claims that Minecraft's success changed everything. "It showed people really liked to create things. I was building something more for the game studios but, as it turns out, this can be really democratic."
StarForge and Everquest are but two of the companies that have licensed Voxel Farm. But it appears unlikely that they will be the last (Cepero has already hinted that other organizations have purchased a license). Like Storybricks, Voxel Farm feels like a dream made into flesh and code, a realization of an industry's long-running desires. And, in some ways, Voxel Farm is exactly that. When asked as to how Voxel Farm came about, Cepero began by explaining the problems that he is looking to solve.
"If you look at the game industry today, it turns out its growth is being limited by human creativity. We are past the point where the hardware and software was the bottleneck. AAA games are very expensive to develop to the point the industry is at danger of stagnating. There may be no reason to push the hardware any further since it is too expensive to exploit it anyway. And this is a hard limit, because now it is the human factor slowing us."
He continued, "Solving that problem was key for Voxel Farm.The engine has two main pillars to address that: Procedural Generation and Crowdsourcing. I believe that this is how we can get things moving again. Crowdsourcing is what Minecraft has shown to be huge."
"Think of what Youtube did to regular TV. Think of what Facebook did to your entertainment. You can have people producing their own content and entertaining themselves. Not everything has to be created in a studio and games are no different."
The trick is, Cepero explained, compensating for a lack of skill. This is where the second tenet comes in. I do not like to call it procedural generation, because it kinds of misses the point. It gets confused with old-school procedural stuff. What procedural really means is automation. Computers will be able to automate many of the tasks artists do. They could learn by watching artists, and reproduce what they achieve so, when I look at procedural generation, what we need to be doing is automating the artist."
"Now, these artistic automatons, it is something anyone can use. They could create content for others to enjoy, and it would be looking good as if a professional artist had done it. That was the plan when I started Voxel Farm."
Disappointing as this may be for those envisioning a world where armor and cloth may be shredded and ruined, Voxel Farm deals exclusively with the architecture and landscape. Characters, the clothing they wear and the physics governing the swish of a cape all lie outside of its command. That said, such focus may mean even more incredible things for the future of the engine.
"In general, I want it to create things at the level of what you see in Pixar movies, for instance. They have amazing artists and I would like to automate what they do at least for buildings and natural environments"
"Then, I want to get into creatures."
The concept of voxel-based creatures might seem ludicrous but Cepero already has an idea in regards to what he might do. Cepero wants to build them from ground up. "I am considering starting from the behavior so you pick a bunch of things the animal can do -how it runs, how it will attack, etc - and you can think of these as basic animations that will determine which bones are required for the animal. Once you have the bone structure, you can look at the animation again so you can discover where muscles are needed. If it is flapping wings, you would need to connect these bones to some powerful muscles. Then you add other layers, like fat and veins. All these are things you can control as a creator of the animal. Then, skin."
He added, nonchalantly. "I think it is doable, not very difficult to implement, but it can produce a large number of creatures. It is just an approach, I am sure I may consider something else along the way."
Due to the haphazard manner in which the interview ("Are you free for an interview?" "Sure?" "Right now?" "Sure!") was conducted, Cepero didn't have time to stay too long. However, he managed to answer two other questions. The tools demonstrated during the Everquest Next reveal are, it appears, standard Voxel Farm features. "You can have voxel brushes made of anything. You can also go back and smooth and fiddle with the voxels you have placed."
As for how latency would work in the face of so much destructible terrain, Cepero seemed optimistic about the servers withstanding the load. "There is some advantage in using procedural generation and voxels. For instance, when broadcasting network data, you can send only the bits of the world that have changed since all involved parties can generate the same data on their own. In general, I am not too concerned about these issues."
"BUT." He stressed. "Now we are talking about games that are not static anymore. When you have static games there is a lot of tricks you can use to bring memory costs down with fully dynamic worlds. We will need higher requirements almost for sure because we are getting a new feature that we did not have before: ability to change the world on the fly. This necessarily has a cost."