AI Improvement

[DeclaredEvol 10]

As said in another post I made, Bohemia Interactive does not have Open Source Code for ArmA. Therefore, enhancing the AI to reasonable use would be beyond our capabilities. You'll be playing the same game, with some extra features... and less capability for ACE to turn it into a Simulator again. It almost seems like BI didn't want ArmA III to be realistic?? Putting us into the future with fake materials is going to make it harder for us to work with, requiring people to literally remodel and start from scratch. Nothing is ever easy anymore...

Its even MORE harder to work with their new features (PhysX, DX10, More Structures) since its more advanced... don't expect super-awesome mods any time soon :P

[Nicholas 5]

Putting us into the future with fake materials is going to make it harder for us to work with, requiring people to literally remodel and start from scratch. Nothing is ever easy anymore...

Its even MORE harder to work with their new features (PhysX, DX10, More Structures) since its more advanced... don't expect super-awesome mods any time soon :P

It seems like nearly everyone expects everything handed to them on a silver platter. When OFP was released, everyone had to model from scratch. No one was worried if OFP was "realistic" or not, they just tried to make really cool stuff. Everyone now is only worried about making "realistic" addons it seems. I personally think new content is better than retextures and revamped content from a previous game.

[DeclaredEvol 10]

Wrong and right answer, what I said stands firm. You pay 50-60 Dollars for a game, and you've agreed it is worth the label of a 50-60 Dollar Game. It all depends on the customer and how easy it is for him to want the product, now of course I've been playing ArmA since 2006-2007... so I understand where you're coming from. My major concern is what this will change ArmA into, and originally ArmA was more close to reality than it was to some fantasy op shooter. If ArmA III is successful, then it might just keep going in that direction which people are afraid of. One day you had a game where you could play some neat special operations on ArmA I. And the next you're playing zombie mods, so I think this may end up being more of a sandbox engine than a war engine in the end. The creativity is beyond the roof, which I like... I am not just military player. I've played games since the early 90's before military was truly an option at all.

The only thing else I could disagree with you is that you said everyone wants things on a silver platter, and I think you're implying to what I said about the AI. And no, we all want different things and we all except different results. But ArmA being a simulator with educative material means it needs to stand strong when presenting it self, I can't tell you how many times I've had a friend or my dad walk into my room and say what the hell is that guy doing. And that sir is a bot, its not perfect... but it is given basic guidelines. What I don't understand is why its easier for Bohemia Interactive to create integration for PhysX and not enhance some weak areas of their AI such as their survival. Its not long from the start of a mission I see a bot do something stupid like run out while bullets are flying and get hit... or stand up straight in the middle of a field while a technical is near him. The fact that they've not enhanced AI is what I am concerned about, its not about these incredible digital effects anymore... its the game play. Bottle-necking AI for Close Quarters and Survival is what a lot of noobs are going to notice right off from the start. Plus the annoying Microsoft Sam voices lol... (To Be Honest, I was willing to let Bohemia Interactive work till 2013-2014 so we could get a superior engine) And gladly they've taken some more leaps which its wise to be cautious about your release. I HONESTLY hope Bohemia pays attention to all of the AI Modders out there and take them seriously because we as people of great respect cannot fix some of the loop holes they leave us. When the game is out, it is too late and that shit happened to people with BF3 and other games as well. I like a game that takes 5 years, because when it comes out people will never quit playing it and the new players just keep coming. Plus time is money, and a release date being pushed back means less money at first but more in the end. So if they are going to mess around for another few months before it is released, they have to make fast choices of what they truly want to do.

[^Th0mas^ 10]

So we didnt get to see any AI at E3? :(

Will there be working AI in the community alpha?

[onlyrazor 11]

Wrong and right answer, what I said stands firm. You pay 50-60 Dollars for a game, and you've agreed it is worth the label of a 50-60 Dollar Game. It all depends on the customer and how easy it is for him to want the product, now of course I've been playing ArmA since 2006-2007... so I understand where you're coming from. My major concern is what this will change ArmA into, and originally ArmA was more close to reality than it was to some fantasy op shooter. If ArmA III is successful, then it might just keep going in that direction which people are afraid of. One day you had a game where you could play some neat special operations on ArmA I. And the next you're playing zombie mods, so I think this may end up being more of a sandbox engine than a war engine in the end. The creativity is beyond the roof, which I like... I am not just military player. I've played games since the early 90's before military was truly an option at all.

The only thing else I could disagree with you is that you said everyone wants things on a silver platter, and I think you're implying to what I said about the AI. And no, we all want different things and we all except different results. But ArmA being a simulator with educative material means it needs to stand strong when presenting it self, I can't tell you how many times I've had a friend or my dad walk into my room and say what the hell is that guy doing. And that sir is a bot, its not perfect... but it is given basic guidelines. What I don't understand is why its easier for Bohemia Interactive to create integration for PhysX and not enhance some weak areas of their AI such as their survival. Its not long from the start of a mission I see a bot do something stupid like run out while bullets are flying and get hit... or stand up straight in the middle of a field while a technical is near him. The fact that they've not enhanced AI is what I am concerned about, its not about these incredible digital effects anymore... its the game play. Bottle-necking AI for Close Quarters and Survival is what a lot of noobs are going to notice right off from the start. Plus the annoying Microsoft Sam voices lol... (To Be Honest, I was willing to let Bohemia Interactive work till 2013-2014 so we could get a superior engine) And gladly they've taken some more leaps which its wise to be cautious about your release. I HONESTLY hope Bohemia pays attention to all of the AI Modders out there and take them seriously because we as people of great respect cannot fix some of the loop holes they leave us. When the game is out, it is too late and that shit happened to people with BF3 and other games as well. I like a game that takes 5 years, because when it comes out people will never quit playing it and the new players just keep coming. Plus time is money, and a release date being pushed back means less money at first but more in the end. So if they are going to mess around for another few months before it is released, they have to make fast choices of what they truly want to do.

Slow down there. First off, somehow you went from talking about AI improvements to Arma 3's realism and the future, which is better suited for a thread such as this if it doesn't directly tie into AI. Also, Real Virtuality is a sandbox engine. If BIS wanted a war engine, they'd just lock out any modding capabilities, removed the editor and said "There's war. Enjoy."

Did anybody at BIS say they didn't enhance AI? For all the community knows, the AI could have had hundreds of minor enhancements since work started on Arma 3. I'm fairly certain that of all people, BIS take modders very seriously. Concerning AI development, modders might be particularly important, though, so I agree with you there. AI mods make Arma a lot more interesting to play.

I'd say more, but I'm tired right now, so maybe later. Nobody take this post the wrong way, even though it does appear harsh.

[froggyluv 2130]

Interesting read on state of AI in gaming and why AAA title developers aren't necessarily advancing it:

Videogames have journeyed from pixel sprites to near-photorealism, but our expectations for how characters should look have been satisfied more easily than our desires for how they should behave. The next generation of technology may be able to render the way light glistens on the meniscus of an eye, or scatters beneath soft skin, but will the owner of said skin be able to find her way out of a room without hitting herself with the door?

-Edge

http://www.edge-online.com/features/future-videogame-ai

[^Th0mas^ 10]

Maybe more decent driving AI in ArmA 3?

http://forums.bistudio.com/showthread.php?137266-How-is-AI-driving-in-the-latest-patch&p=2193885&viewfull=1#post2193885

[onlyrazor 11]

Maybe more decent driving AI in ArmA 3?

http://forums.bistudio.com/showthread.php?137266-How-is-AI-driving-in-the-latest-patch&p=2193885&viewfull=1#post2193885

Dwarden said something about this:

yes I do agree driving AI should be improved more ... but that's task for next title ...

[metalcraze 290]

The main problem with driving is AI driver thinking his vehicle is much wider than it really is. That's the reason for all the problems you have with incoming traffic and why AI can't drive in between 2 bushes even if they are 5+ meters apart. Which is more than enough for a car.

xAitment pathfinding which comes to VBS2 is designed with a heavy city traffic in mind (which can be seen in one of its demo videos) - but alas BIS decided we won't have it in ArmA3

[froggyluv 2130]

xAitment pathfinding which comes to VBS2 is designed with a heavy city traffic in mind (which can be seen in one of its demo videos) - but alas BIS decided we won't have it in ArmA3

Hehe, I've been pushing for Xaitment for a long time but after trying 7554 - color me unimpressed. In theory and tech demos its decent but I have yet to see a game, especially in our genre, live up to its claims.

[mazza 1]

Q: Will the AI no longer be able to see through grass, and how will the grass be displayed in multiplayer?

Dan: Unfortunately, there's no progress on this. Staying the same so far.

Source http://www.armaholic.com/page.php?id=17263

LOL How long have we been asking for this..?

[Rak 0]

Source http://www.armaholic.com/page.php?id=17263

LOL How long have we been asking for this..?

Oh wow, he dodged pretty much every question there. "Maybe", "Not so much" or straight "No"s.

TBH I don't think AI will be improved at all, even Ivan Buchta admitted in one video that they dumbed down AI's cover searching to allow them to keep up with the player. It seems to be BIS is only focusing on graphics and animations for ArmA 3, not "gameplay features".

[killerwhale 1]

there are bigger problems than driving, that is the last thing to talk about. Ai being freak in taking cover behind objects or leaning from walls. all urban combat is a mess for Arma2 AI. do the guys that release the game actually play it?

[Ekko 1]

there are bigger problems than driving, that is the last thing to talk about. Ai being freak in taking cover behind objects or leaning from walls. all urban combat is a mess for Arma2 AI. do the guys that release the game actually play it?

They do, but they are blind.

[x3kj 1247]

TBH I don't think AI will be improved at all, even Ivan Buchta admitted in one video that they dumbed down AI's cover searching to allow them to keep up with the player.

They "dumbed" it down so the AI can make effective decisions. What's the point in considering a wooden plank that lies flat on the ground as cover? That's what Arma 2 AI does. So how exactly is that "dumbing" it down? And that's exactly what they tried to improve according to the interview. If you can't interpret the answers the devs give you correctly you should stop posting :j:

Edited July 30, 2012 by Fennek

[chortles 263]

TBH I don't think AI will be improved at all, even Ivan Buchta admitted in one video that they dumbed down AI's cover searching to allow them to keep up with the player. It seems to be BIS is only focusing on graphics and animations for ArmA 3, not "gameplay features".

I'd rather the AI keep up with me, so dumbing down the AI's cover searching is ah-oh-kay by me! Honestly though, I'd rather see the animations improved more than "gameplay features".

What Jay Crowe (the actual speaker) said -- this was in GameSpot's version of the E3 2012 presentation -- was that in PMC/Zargabad the AI would keep trying to evaluate EVERYTHING as cover, so BI had to simplify the cover-deciding process.

[froggyluv 2130]

What Jay Crowe (the actual speaker) said -- this was in GameSpot's version of the E3 2012 presentation -- was that in PMC/Zargabad the AI would keep trying to evaluate EVERYTHING as cover, so BI had to simplify the cover-deciding process.

That ain't dumbing down -thats streamlining :)

[metalcraze 290]

That ain't dumbing down -thats streamlining :)

Is there a difference?

Previously when in combat AI was immediately jumping under the closest rock, now an armored car sprays them and they run around like headless chicken. And in infantry E3 demos they simply run at your gun.

[-Coulum- 35]

Okay I could be wrong but in that vid where Jay talked about cover finding wasn't he simply giving an example of changes already made in OA betas that will be carried on into A3. I didn't actually think he was talking about new features in A3... or was he?

Dumbing down/ streamlining whatever you want to call it, sometimes these kind of changes are needed imo. For example if an ai sees an enemy 10 metres away, there should be no complex decision making. Just shoot the guy. Similarly, if a burst of rounds snaps by an ai he should just get the fuck down. Simple but effective changes like these are needed.

[metalcraze 290]

No he said the problem was (for whom?) is that AI navigated too slow because it took a lot of cover around it into account. And then the video showed that AI doesn't try to use any cover or concealment ever, like we are back in OFP again where AI sometimes knew that there are bushes and trees in the world.

[chortles 263]

That ain't dumbing down -thats streamlining :)

The real best part though was Jay having to say "let's not be afraid of that word, streamlining"...

[x3kj 1247]

And then the video showed that AI doesn't try to use any cover or concealment ever, like we are back in OFP again where AI sometimes knew that there are bushes and trees in the world.

E3 didn't show a plane in action, does that mean they can't be manually controlled? ... seriously dude

the showoff was supposed to show off (surprise!) action and not the full game. What's the point in showing how good the AI can hide behind bushes and having the visitor crouch through the hole village for 30min in order to find the damn AI, with 10 other people waiting behind you to play. They were certainly set to aggressive and attack to enforce action.

[metalcraze 290]

E3 didn't show a plane in action, does that mean they can't be manually controlled?

But E3 showed AI in action. Problems with reading?

the showoff was supposed to show off (surprise!) action and not the full game. What's the point in showing how good the AI can hide behind bushes and having the visitor crouch through the hole village for 30min in order to find the damn AI, with 10 other people waiting behind you to play. They were certainly set to aggressive and attack to enforce action.

Yes because all AI does in ArmA is hiding in bushes, pissing its pants in fear when it sees the human player. And AI in ArmA is also known for being so good at hiding you will spend half of an hour to find it in between 3 buildings. And of course just for E3 BIS spent time and money making an "action mode" AI that will just run at player's gun barrel. Sounds plausible man.

Multiple videos showed AI acting dumb in every single occasion (that included being killed by a vehicle as they simply ran left and right in the open even though there were a ditch and bushes like 10m away) and obviously worse than in OA. Jay Crowe said they ditched old cover-seeking algorithms. Quit fanboying.

Edited July 30, 2012 by metalcraze

[NeMeSiS 11]

No he said the problem was (for whom?) is that AI navigated too slow because it took a lot of cover around it into account. And then the video showed that AI doesn't try to use any cover or concealment ever, like we are back in OFP again where AI sometimes knew that there are bushes and trees in the world.

To be honest, if you try the same thing in A2 you will get pretty much the same results. The A2 AI will get to cover eventually, but any player with at least 1 working hand will have killed them long before that.

[-Coulum- 35]

To be honest, if you try the same thing in A2 you will get pretty much the same results. The A2 AI will get to cover eventually, but any player with at least 1 working hand will have killed them long before that.

Yes indeed. A3 ai didn't look any better than a2 but it certainly didn't look any worse.

I wonder how/if BIS will improve ai spotting. I am tired of shooting one shot two shots off in the dark and immediately being dismantled by the DSHKM 300 metres away. I don't think that the ai should be able to target my approximate location and even shoot me after a while but I really dislike the contrast between instant knowledge of my exact location and total inaction. Conversely, I also hate how the ai can't spot me at 600 metres in broad-daylight, in the open, after I empty my mag, run around in circles and scream at the top of my lungs. BIS said they were tweaking ai configs so could these problem be fixed by this?