Amy Hennig Talks More About Struggling With EA's Frostbite Engine

Since this thread has evolved into a game AI thread, I thought I might post a link to the website of the Game AI Pro books

Game AI Pro

There are three books as of now and a fourth if forthcoming. The chapters are written by people from the industry. You can read the first two books for free. There is a nice chapter about FEAR and some chapters about Last of Us. Maybe some of you will find this a worthwhile read.

Also, by pure accident, i stumbled upon this book called Artificial Intelligence and Games just now. Apparently the PDF version of that book can be downloaded here. It seems to be less hand-on and more a scientific overview. I don't know if it is any good but thought some might want to check it out anyway.
 
lefantome said:
Thanks for the reply but I still don't get it and the MGSV/Halo comparison is not clear to me.

Is it a form of director AI like in Left 4 Dead?
Click to expand...
Meta is really all about memory, and every genre of game has Meta. Fighting games have meta, RTS games have meta, chess games have meta. Meta is taking a look at the current landscape of what the players are doing, and drafting a new strategy to specifically combat that one entirely.

Game developers make balance changes to games to create new possibilities of new Metas so that there isn't just a single dominant strategy that can't be countered.

As Shifty said, Meta AI might not be the best term to use. Meta often being used as 'current winning strategy', might be the wrong way to look at it. It might be better to describe Meta AI, as AI that has memory of past events.

Normally you have a set of inputs for any AI, but if you add memory to the list of inputs, how it responds to the same situation will be different depending on the memory parameter.

There are different ways to describe what memory AI is, but in RPG, it's like an action you took earlier on in the game comes back to affect your later game. So in Bioware games, this is done a lot in game, and people have memorable moments when the game comes back to haunt you for a small decision you didn't think mattered. But even further, they take your save game from the previous mass effect and move it to the next title, and once again the actions that you took in the previous mass effect affected many scenarios in the next title. And I think they have grossly undervalued what charm their titles have in terms of Meta. In others like MGSV, the enemy adapts to how you like to infiltrate enemy stations. So your strategies are less effective or harder to pull off. In a MMORPG, say like EvE Online, there was a massive battle at Jita that I got to be apart of with all the players involved. And to commemorate that event, they put a statue there in that system.

It can be a small thing, but this is what players remember.

And all of these things that we do, as people, I know we can do in games, and they don't need to be crazy calculated AI. We've had these types of elements in games for a long time. Just recently though, a lot of games have moved away from it, moved away from more decision making to less decision making, moved away from more systems to simpler systems, all in the name of getting the game out of the way so that the story and narrative can have a better spotlight.
 
Frostbite ain't going nowhere :smile2: Yesterday was its 10 years anniversary & EA is seems fairly happy to communicate about the fact that all of their IPs are using it (minus NHL & NBA IIRC this year).

 
I feel Frostbite gets a bad rap, it's easy to lay all the blame on one thing but it's usually a combination of things. I mean look at pubg it uses unreal 4 but no one really blamed unreal 4 for the issues they had.
 
iroboto said:
Meta is really all about memory, and every genre of game has Meta. Fighting games have meta, RTS games have meta, chess games have meta. Meta is taking a look at the current landscape of what the players are doing, and drafting a new strategy to specifically combat that one entirely.

Game developers make balance changes to games to create new possibilities of new Metas so that there isn't just a single dominant strategy that can't be countered.

As Shifty said, Meta AI might not be the best term to use. Meta often being used as 'current winning strategy', might be the wrong way to look at it. It might be better to describe Meta AI, as AI that has memory of past events.

Normally you have a set of inputs for any AI, but if you add memory to the list of inputs, how it responds to the same situation will be different depending on the memory parameter.

There are different ways to describe what memory AI is, but in RPG, it's like an action you took earlier on in the game comes back to affect your later game. So in Bioware games, this is done a lot in game, and people have memorable moments when the game comes back to haunt you for a small decision you didn't think mattered. But even further, they take your save game from the previous mass effect and move it to the next title, and once again the actions that you took in the previous mass effect affected many scenarios in the next title. And I think they have grossly undervalued what charm their titles have in terms of Meta. In others like MGSV, the enemy adapts to how you like to infiltrate enemy stations. So your strategies are less effective or harder to pull off. In a MMORPG, say like EvE Online, there was a massive battle at Jita that I got to be apart of with all the players involved. And to commemorate that event, they put a statue there in that system.

It can be a small thing, but this is what players remember.

And all of these things that we do, as people, I know we can do in games, and they don't need to be crazy calculated AI. We've had these types of elements in games for a long time. Just recently though, a lot of games have moved away from it, moved away from more decision making to less decision making, moved away from more systems to simpler systems, all in the name of getting the game out of the way so that the story and narrative can have a better spotlight.
Click to expand...

Thanks, I just didn't know it was named meta and using that word sounds weird to me.
 
DavidGraham said:
Oh, innovation has stagnated alright.

When was the last time we saw a game as large as Daggerfall? 15000 cities scattered across a 160000 km2 massive map, populated with 3/4 of a million people that you can talk to and interact with? This was released 20 years ago, no other game has attempted to do such thing. And no, No Man Sky doesn't count.

Last game with honestly good AI? F.E.A.R. 1!

Last game with good total destruction? Red Faction Guerrilla, and maybe Mercenaries 2.

Last game with good massive environment interaction? Crysis Warhead, or Max Payne 3!
Click to expand...
darn, I got TES: Daggerfall at GoG, in a Bethesda sale, but never checked! How did they achieve that?

AI wise, I'd also add Halo.
 
Cyan said:
darn, I got TES: Daggerfall at GoG, in a Bethesda sale, but never checked! How did they achieve that?
Click to expand...

The same way Elite did 8 galaxies with 256 planets on 8-bit CPUs with 32kb RAM: procedural generation. That wasn't a limit of 8-bit systems, Elite was originally planned to have 2,814,74,976,710,656 galaxies but they rolled it way back. You can go too far, like No Man's Sky, which is stupidly big and diverse but where much of the diversity is really inconsequential and uninteresting.

Size isn't everything though, compare many Bethesda RPG open world to Rockstar worlds, the latter are much larger but are they more interesting?
 
And in Daggerfall, there was basically zero diversity.
 
https://kotaku.com/how-biowares-anthem-went-wrong-1833731964

a story of indecision and mismanagement. It’s a story of technical failings, as EA’s Frostbite engine continued to make life miserable for many of BioWare’s developers, and understaffed departments struggled to serve their team’s needs. It’s a story of two studios, one in Edmonton, Alberta, Canada and another in Austin, Texas, that grew resentful toward one another thanks to a tense, lopsided relationship. It’s a story of a video game that was in development for nearly seven years but didn’t enter production until the final 18 months, thanks to big narrative reboots, major design overhauls, and a leadership team said to be unable to provide a consistent vision and unwilling to listen to feedback.
Click to expand...
 
Forstbite ugliness shows it's head one more time in Anthem. This time it's confirmed by at least 3 different developers working on Anthem.
“Frostbite is like an in-house engine with all the problems that entails—it’s poorly documented, hacked together, and so on—with all the problems of an externally sourced engine. Nobody you actually work with designed it, so you don’t know why this thing works the way it does, why this is named the way it is.”
Click to expand...
“Part of the trouble was you could do enough in the engine to hack it to show what was possible, but then to get the investment behind it to get it actually done took a lot longer, and in some cases you’d run into a brick wall. Then you’d realize, ‘Oh my god, we can do this only if we reinvent the wheel, which is going to take too long.’ It was sometimes difficult to know when to cut and run.”
Click to expand...
“If you can hack around it, you hack around it, as opposed to fixing it properly.” said a third developer who worked on Anthem and continued: “I would say the biggest problem I had with Frostbite was how many steps you needed to do something basic. With another engine I could do something myself, maybe with a designer. Here it’s a complicated thing.”
Click to expand...
“It’s hard enough to make a game. It’s really hard to make a game where you have to fight your own tool set all the time.”
Click to expand...

https://kotaku.com/how-biowares-anthem-went-wrong-1833731964
 
At the same time, it seems like Bioware themselves have a really bad track record with project management.

Along with the supposed re-allocation of engineering resources (for FIFA), what a recipe for disaster.
 
I get the impression they need good management with no underestimation of engineering resources, the latter of which may be problematic if the talent just isn’t available.

Put another way, I’m not sure what choice they would have had otherwise. UE4 has different challenges. Bungle and Blizzard built their own engines from scratch anyway.
 
You must log in or register to reply here.
Back
Top