Well…
Same way newer crysis shots look worse than the old ones, I guess reality will ruin some of those promises. But it looks like far cry 2 won’t be a re-hash to get some money. Maybe it’ll turn out to be like just cause, just without its problems (Mainly, awful gameplay)
If it’s anywhere near as awesome as Just Cause, then… awesome.
Ditto. I have a friend working on Far Cry 2 and the info he’s been sneaking to me is absolutely awesome. I’m really looking forward to seeing what they can do.
edit: And he assures me that the screenshots of things on fire are not bullshots.
Snort.
Did it fail at going under the bridge, or did that turn out to be scripted or something?
Everytime you post an opinion about a game I realize how insane you are.
I heard Just Cause was actually pretty good.
Bob’s not insane, he just has excessively narrow tastes.
Well, while we have no way of knowing what section they were referring to, I saw nothing to suggest this was emergent, it looked like a coded element of the strider routine, if height > X then go low.
Wouldn’t emergent behavior be more like give AI entities specific changing desires and motivations, then give them generic tools to navigate and interact with everything in the world to accomplish those goals? In other words, Radiant AI (in Oblivion) is one of the game implementations that comes closest to this particular school of thought.
Not if you script the strider to walk at the bridge, no. What did they expect? It to explode?
Not really. Emergent behavior is unexpected and unforeseen interactions that create gameplay or interesting moments. In your situation, the AI needs to already be able to understand the possible interactions.
The AI when doing what they know could theoretically result in emergent gameplay, but not in a way specific to any goals the AI may have.
edit: But just to agree with you; AI navigating the world has jack to do with emergent gameplay. It’s simply the AI navigating the world.
At the very least, the animations would have had to have been provided in order for the strider to do that, so it was anticipated by the designers.
Forgive my ignorance, as I’m not a game designer…but emergent or adaptive behavior? Are we talking about a character in a game doing something that surprises the people who coded the thing? An entity that creatively augments its own behavior to fit a situation (not based on a huge decision tree), and thus rearranges the terms of its own abilities and existence within the game world is probably a long way off, eh? Purely computer-based ‘enitites’ that modify their own computer code by ‘learning’ things not originally available to (or within the design of) that code is currently not possible, right? I mean, Radiant AI is neat, sure, but there are still rules that govern the ability to augment abilities…even if those rules are iterated dynamically and self-referential, they are still possibilities that exist within the original code/logic, and not things that the computer ‘thought’ of itself.
My fear is that I may be taking the term too far outside the realm of game development, here…do people in the industry have some other definition/concept of this term?
You never played Galapagos?
You are reading way too much in to it.
It’s like a developer making grenades, and allowing grenades to destroy wood, and then without realizing it, level designers make a giant wood platform, and then an AI lobs a grenade at you, but instead of exploding and killing you, it explodes and destroys the platform.
It’s a weak example, but I’m tired and it’s not the kind of thing you can often sit down and think out. It just happens. It has nothing to do with AI learning, or crazy shit like that.
From what I understand, in some instances, the strider destroyed the bridge to pass through it, sometimes it went under it. The fact that the strider AI made a choice to sometimes blow up the bridge rather than pass under it was a choice the designers hadn’t programmed into the AI specifically.
Now, calling this “emergent” is perhaps tooting your own horn a bit, but it is pretty fucking cool when an AI behaves in a way that you didn’t intend and does so effectively.
Getting back on topic, I’m fucking sick of perfectly good FPSs being ruined by the “mutant/alien/monster” cliche. I was REALLY loving Far Cry prior to the emergence of the Trifucks. A Far Cry sequel with less monsters, more commandos, and the same type of gameplay? Sign me the fuck up.
That’s not emergent. The bridge had to be built in a fashion that would let it be destroyed – objects in Half Life 2 don’t have that level of freedom except where the designer decides it. And since the player couldn’t destroy it (or have reason to), then it means it was designed that way so the strider could do it.
I think what he’s saying is that that specific choice (to go under the bridge rather than through it) wasn’t programmed in. Some similar choice the strider would have to make was obviously programmed in, but to have it applied to the bridge was unanticipated.
I’m not sure if the exact mechanism of how the AI accomplishes its feats is necessarily tied to emergent behavior. What we are looking at is how the AI character behaves, and not how it thinks. If you want a learning AI, you can do that as well and still give it the tools that any type of agent will have. Agents are defined by a series of 0 to N number of inputs that let them determine things about the “world” around them (game state for discrete simulations, or some other type of “view” into the world for a continuous simulation), and some number 0 to M of actuators that let them effect the world around them (could be as simple as moving some part of the Agent around, or as complex as launching/shooting something or changing the mode of one of their input “views”).
I think if you took a big game animal like a lion, and gave its game AI agent inputs for hunger and thirst, internal meters of fear, anger, and reproductive drive (for males) or maternal drive (for females), give them movement in three dimensions including stalking, walking, running, leaping in various degrees, plus claw and bite attacks of various types, you may be able to tweak it around such that each one runs a genetic algorithm and can simulate a pride of lions in interesting ways. You would have new lions being born and dying, and you could simulate micro-ecosystems such as the pride’s relationship with prey animals and human encroachment.