Development Blog & Reveals

Now you have to play through 500 more similar situation where the ai simply walks into the building one at a time each one being repeatedly slaughered by your friend camping in the hardcorner of the room, while no ai thinks maybe they should go through the back door, toss a grenade or even enter the building facing that hardcorner instead of going into it obliviously.

What I'm saying is that no doubt the ai sometimes does awesome stuff but it only happens once in "insanely large number" times and this ratio suggests that those flashes of brilliance are not infact purposeful but pure fluke. Could be wrong but thats simply the impresion I get. It is very hard to get killed in a building against the ai.

And I still have never seen them take up good firing positions within a building or even decide, on their own, to occupy a building.

That at least is my experience and is why I think they need to be improved to at least know the basics of fighting in/between/from buildings.

Pretty much this. I've also seen AI navigate to top floor of building garage to get me but again its pretty rare.

In contrast to that story I was playing some Co-OP with a friend in which for whatever reason the AI montage of mods we had running enabled the enemy AI a too high of a Knowsabout reading on us. We were seperated and I was holed up in a Taki house with literally dozens of enemy AI attempting to come and kill me. I finally walked away unscathed after a 30+ killstreak all due to the fact that these guys would pretty much just run to or past the front door where I had my bead and die one at a time. Imagine if AI had lobbed a grenade, peaked the doorframe or that window that would have easily given them the kill they deserved.

These types of things can only be implemented at the core level by BI no matter how talented our AI modders are (and they are quite good).

These types of things can only be implemented at the core level by BI no matter how talented our AI modders are (and they are quite good).

Precisely. This is why its really painful to hear that BIS is doing stuff like improving the ai by only giving it "config tweaking". Modders can already do that well enough. Maybe not quite as "under the skin" as BIS will but ASR or similar mods already fix the ai accuracy/spotting etc. to be more realistic. What we need is those base level, individual decision making and action improvements. Being able to move to and shoot out windows. knowing how to scan and engage a room quickly and effeiciently. determine what is and isn't dangerous based on their buddy who just got blown apart walking into a room. improve/fix the fundamental behaviours so than modders aren't trying to fix them and can instead expand on the ai rather than fix its obvious flaws.

I understand its all easier said than done though.

Have the showcase videos stopped becuase there not done with the helicopter model or something or are they canceling the line of e3 showcase and starting a games con video showcase?

Have the showcase videos stopped becuase there not done with the helicopter model or something or are they canceling the line of e3 showcase and starting a games con video showcase?

The showcases we are waiting for a helicopters and infantry. My guess is that infantry is causing a hold up, the main advancements we are waiting for are large improvements to the AI, BIS is probably fixing problems with the AI and making sure that the walkthrough demonstrates smart behaviour and not the weird movements and behaviours exhibited by A2 AI. Comparatively the other videos feature very little or no AI.

The showcases we are waiting for a helicopters and infantry. My guess is that infantry is causing a hold up, the main advancements we are waiting for are large improvements to the AI, BIS is probably fixing problems with the AI and making sure that the walkthrough demonstrates smart behaviour and not the weird movements and behaviours exhibited by A2 AI. Comparatively the other videos feature very little or no AI.

We've already seen helicopters demonstration. The next one is support where they show off some artillery and a helicopter attack.

As the number of render targets increases, so the FPS/source reduces. In the vehicle demonstrated, there were something like 7 RTs defined (mirrors, on-board TV for driver, TV for commander, TV for Gunner).

Consequently, each screen shared ~1/7th of the overall resources dedicated to RTT. Of course, I like our optimism, and willingness to push barriers, but, indeed, perhaps that's not going to work, at least in the current state. :o

In truth, we've already identified the issue, and have considered a few solutions: simply a more intelligent deployment of render targets, maybe a more 'optimised' implementation, perhaps some more intricate/half-way-house solutions, and, of course, empowering players with user-defined selection of quality.

Overall, though, the focus will need to be on usability. So, if I'm going to use an on-board TV screen for spotting/ targeting enemies, for example, it better be rendered at a useful rate! If I'm not able to do so because some tiny-ass mirror I can barely see is sucking my performance, I'm going to nuke that from orbit.

Partly, we need to make better design decisons about that (function/ form), and partly we need to put some tools in the players hands to have it as he prefers (a la Take On). These are just some of things we need to balance and iron out, and it forms part of our enormous tasklog as we march valiantly towards a public release!

Best,

RiE

Couldn't you just turn off the RTT if your not in the position with the screen. The driver probably couldn't see the gunners screen and the gunner probably could see the rear view screen or mirrors. So why waste the resources showing them when those positions cant see it. Unless its just programmed to the vehicle the player is in. Get in a vehicle and all the RTT turns on.

I don't know much about it though.

Precisely. This is why its really painful to hear that BIS is doing stuff like improving the ai by only giving it "config tweaking".

[...]

I understand its all easier said than done though.

I think the point is not that more 'deep rooted' improvements to the AI aren't warranted/ desired; rather, it's that to be confident about which behaviours need to be changed/ corrected, one must be sure that the configuration is balanced appropriately. It's also worth bearing in mind that more high level observations about AI decisions are not simple to translate into technical requirements.

Naturally, I wouldn't dare to suggest that specific repros aren't available (The CIT and Beta Patch forums have been and continue to prove to be invaluable resources for our most skilled programmers: the ones with the experience to make core changes happen), but, again, we need clean repros and it can be a challenge to achieve that at this stage of production, where a big number of factors interplay (i.e. significant environment, model, config, animation changes on a frequent basis) on top of the significant resource demands of implementing new features.

Looking forward, we know that AI refinement will be necessary. The above isn't intended to excuse not achieving it; rather, it frames the problem and doing that helps us plan an appropriate schedule for getting it done. :)

Have the showcase videos stopped becuase there not done with the helicopter model or something or are they canceling the line of e3 showcase and starting a games con video showcase?

We enjoyed producing the showcase videos after E3, but, with GC looming so quickly, we had to divert resources towards preparations for that. I think it's a happy (or ingeniously planned :p) coincidence that the recent GameStar videos have been able to supplement our own coverage.

We're aware that helicopter and infantry walky-talk-throughs are high priority for our future assets. :)

Couldn't you just [...]

That sort of thing falls under the:

some more intricate/half-way-house solutions

Category :)

trolling trolls today

Quite, indeed!

Best,

RiE

Edited August 3, 2012 by RoyaltyinExile

For those of you concerned about AI indoors, while this is not confirmed to be either an AI or a player, make what you will of this screenshot.

Nope, it's rather the other way around, wait for more info :icon_twisted:

Hopefully you don't mean anything more complex than the stance modifier key that Evan Lahti described in PC Gamer and as Ivan Buchta demonstrated at E3... I'm very pleased that Gaia has streamlined the controls!

We enjoyed producing the showcase videos after E3, but, with GC looming so quickly, we had to divert resources towards preparations for that. I think it's a happy (or ingeniously planned :p) coincidence that the recent GameStar videos have been able to supplement our own coverage.

We're aware that helicopter and infantry walky-talk-throughs are high priority for our future assets. :)

I believe you guys already DID a helicopter showcase though? EDIT: Apparently there was, the Support Showcase was after that, and the Infantry Showcase was to be the sixth and last post-E3 showcase.

Is 'Firing Drills' the renamed Infantry Showcase? After all, we already saw the time-based shooting course and the points-based shooting course, is Firing Drills separate from those?

In addition to tactical pace and the stance modifiers, we've seen both standing-in-place weapon transition (I'm still interested in whether weapon switch can be on without forcing the character to stay in place, hopefully the infantry video shows this!) and the tightness/responsiveness/smoothness of the infantry combat... so besides whether or not we can weapon-switch while remaining moving, what more wonders do you have to show us? :eek:

Quite, indeed!

Best,

RiE

The best troll-of-the-trolls (yes I think of the old fanatics as trolls) I've seen from you was the GameSpot presentation at E3: "Let's not be afraid of that word, streamlined!"

My personal favorite though:

Bohemia has tightened up the shooting mechanics to make it feel much more like a traditional shooter when you’re on the ground; there’s a sense of responsive twitchiness to the infantry combat that ARMA has lacked before.

Edited August 3, 2012 by Chortles

For those of you concerned about AI indoors, while this is not confirmed to be either an AI or a player, make what you will of this screenshot.

This is anything but a proof.

Hence my own disclaimer. Would be nice if one of BI were to answer the implicit question ("was that a player or an AI") of course...

EDIT: InstaGoat, I threw that screenshot in because for all of its dubious utility re: AI, it's also the ONLY screenshot I am aware of for ARMA 3 that is set indoors.

Edited August 3, 2012 by Chortles

I think the point is not that more 'deep rooted' improvements to the AI aren't warranted/ desired; rather, it's that to be confident about which behaviours need to be changed/ corrected, one must be sure that the configuration is balanced appropriately. It's also worth bearing in mind that more high level observations about AI decisions are not simple to translate into technical requirements.

Naturally, I wouldn't dare to suggest that specific repros aren't available (The CIT and Beta Patch forums have been and continue to prove to be invaluable resources for our most skilled programmers: the ones with the experience to make core changes happen), but, again, we need clean repros and it can be a challenge to achieve that at this stage of production, where a big number of factors interplay (i.e. significant environment, model, config, animation changes on a frequent basis) on top of the significant resource demands of implementing new features.

Looking forward, we know that AI refinement will be necessary. The above isn't intended to excuse not achieving it; rather, it frames the problem and doing that helps us plan an appropriate schedule for getting it done. :)

We enjoyed producing the showcase videos after E3, but, with GC looming so quickly, we had to divert resources towards preparations for that. I think it's a happy (or ingeniously planned :p) coincidence that the recent GameStar videos have been able to supplement our own coverage.

We're aware that helicopter and infantry walky-talk-throughs are high priority for our future assets. :)

That sort of thing falls under the:

Category :)

Quite, indeed!

Best,

RiE

I think with the AI, best thing to do to test their behaviour is set up "clean" arenas like I did way back when with my Arma 2 AI videos:

Right now it looks like I´ll be at gamescom. Are visitors allowed to peek into the Editor, or is the showcase scenario the only thing you´re allowing us to play with? Because if I could get into the editor, I´d set up something like this just to see what your tweaks so far have done. :P

The AI is super-important to me, because I am on terrible net and more often than not cannot play online, especially large-scale and pvp scenarios. So I kind of need to use what´s in the game, and I appreciate every AI mod that I -don´t- need to use (because they can screw with vanilla content, like campaigns or missions).

As for the screenshots, I can set up really tight CQB screenshots in OA (or even Arma 1) in twenty minutes, using the editor. Screenshots prove nothing, only video does.

I think the point is not that more 'deep rooted' improvements to the AI aren't warranted/ desired; rather, it's that to be confident about which behaviours need to be changed/ corrected, one must be sure that the configuration is balanced appropriately. It's also worth bearing in mind that more high level observations about AI decisions are not simple to translate into technical requirements.

Naturally, I wouldn't dare to suggest that specific repros aren't available (The CIT and Beta Patch forums have been and continue to prove to be invaluable resources for our most skilled programmers: the ones with the experience to make core changes happen), but, again, we need clean repros and it can be a challenge to achieve that at this stage of production, where a big number of factors interplay (i.e. significant environment, model, config, animation changes on a frequent basis) on top of the significant resource demands of implementing new features.

Looking forward, we know that AI refinement will be necessary. The above isn't intended to excuse not achieving it; rather, it frames the problem and doing that helps us plan an appropriate schedule for getting it done.

Thanks for the feedback RiE. If I am understanding you correctly, you're basically saying that ai can't be fixed if the environment + features they are using isn't fully stable first. Makes perfect sense and I glad to hear that AI improvement is on your radar, even if it takes longer to get to the point where you can make significant changes to it. And I love it when the devs communicate with us lowly players, especially when its concerning these kind of things, as it helps us better understand just how complex the process of building this game is. Also makes us feel special.:o

Have a good one.

Edited August 3, 2012 by -Coulum-

@RoyaltyinExile: I think you'll find many participents to run any AI repros you need on almost a full time basis here :)

Also, thanks for the reassurances!

This is anything but a proof.

Agreed. Theres a picture of Game 2 from way back in the OFP days that had 3 soldiers stacked up behind a building and I've never seen that replicated (I haven't been able to find a working picture of it). Additionally, I've never seen the AI do anything like this.

@RoyaltyinExile: Are you just saying that outsourcing QA to the community for free is more fruitful for A3 than let BIS devs test & find issues/bugs?? You should at least think of giving these people some funny badges/medals for use in this forum and ingame!! Something they can/must show everyone.... ;)

I'd like to see jets next :D (heres hoping gamestar covered them)

Agreed. Theres a picture of Game 2 from way back in the OFP days that had 3 soldiers stacked up behind a building and I've never seen that replicated (I haven't been able to find a working picture of it). Additionally, I've never seen the AI do anything like this.

This one?

Hard to believe it was 7 years ago now, was interesting to visit E3 :)

Rest of the pics are here in case it was another one: http://www.ign.com/images/games/arma-combat-operations-pc-748793/4fa6c9f8cdc388ed13eb252b

This one?

http://img225.imageshack.us/img225/8224/46569679.jpg

Hard to believe it was 7 years ago now, was interesting to visit E3 :)

Rest of the pics are here in case it was another one: http://www.ign.com/images/games/arma-combat-operations-pc-748793/4fa6c9f8cdc388ed13eb252b

No it was actually this one:

The link to the bigger version is dead though and I can't find it anywhere else online. Apparently it was 4 guys stacked up though. And yeah it is hard to believe it was so long ago. I'm getting fairly close to being a member of these forums for half of my life now.

Edited August 3, 2012 by Jakerod

Aha of course :)

on some game 2 screens, there were woods and roads crossing in exactly as in Chernarus, i wonder if Chernarus or it's prototype was developed before arma 1 release?

The link to the bigger version is dead though and I can't find it anywhere else online.

Here you go: http://4gamers.ru/albums/screens/action/armedassault/armedassault_1160433949.jpg

These old screens looks insane! It's more prototype for ArmA than real game, although I would like to play it once, even if it had a lot of bugs ;)

Perfect damage system for cars WIP?:)

Here you go: http://4gamers.ru/albums/screens/action/armedassault/armedassault_1160433949.jpg

Thank you! I rather liked that picture.

And now that we've gone 7 years off-topic we should probably get back to blog reveals.

BI should probably get back to releasing them...

Apparently it was 4 guys stacked up though

Stacked up properly as well....