AI Improvement

No I wish! Best moments always happen off camera :(

I think the Suppression script had him going prone and ASR to crouch so it worked out well for him at that wall :p

I hear you tho - it felt entirely authentic and was exciting to watch. Props also has to go to Vanilla AI -they seem much better at keeping cover in those forest battles. With friendly AI they are pretty much all taking trees when under fire and leaning in/out. I used to always have to worry in that if I moved too much -they'd break cover trying to regain formation. Now they seem to stay put (in firefight) unless I sprint forcing them to follow.

Best moments always happen off camera :(

Yes this is so true. I'll test out the TPWC script tonight, it sounds really good.

Had these vids scattered around the forum here and there, brought them together here as it seemed to fit this topic..

A very brief selection of the things we like our ai to do during our missions:

These were recorded using msi at only 50-70% quality, so they are not great. But they give you the idea of some of the things ai do during our missions, (720hd is available);

Inhabit buildings and position themselves correctly:

In combat use those positions to fire from in a reasonable manner, move around within the building or move buildings/go outside to gain a better position if they need to.

React to sound, i.e. gunfire sensibly

Follow up that sound reaction with a careful search for the source of the firing

React to suppressive fire in a reasonable manner, by using nearby buildings or structures to put themselves into cover

Do the same using whatever else happens to be nearby, in this case tree’s and the land contours

React to sighting smoke

Take up positions on balconies or rooftops for the best possible view of the surroundings

I hope to see this and much more in Arma 3, there are lots of really intelligent things that our groups ai do now, and have done for quite some time. Thanks to great mod makers who created these mods, SLX, GL4, Zeus, TPWC_suppress/los. We use pbo's from all these mods to get the ai to act the way we want them to act, as realistically as possible. They manage quite well, hope the same applies in Arma 3 as standard (not mods).

:)

The thing i most want of the AI to be improved, is my comrades AI. I want them to be attached to my ass, because now, every time i run for 150-200 meters, as a team leader, i have to check if my team members are behind me and most of the time they DON'T! Because they are always try to find the most long and elaborate way to pass through a bush or next to a rock, they are way too slow when you give them an order, sometime they just ignores it.

Look at the first Ghost Recon, released in 2001, your team members was actually usefull, they are always behind you, they always follow your orders quickly and in the way you tell them, they don't freak out when they see an enemy soldier, etc..

Please BIS, try to fix those things this time, for real i mean. Right now, having have an AI on your command is more like an handicap than a benefit.

They also didn't use cover or fight independently...

There is a significant disconnect between an effective AI soldier and a user friendly subordinate for the player.

In combat use those positions to fire from in a reasonable manner, move around within the building or move buildings/go outside to gain a better position if they need to.

React to sound, i.e. gunfire sensibly

really impressive, how you achieved such nice behavior? Because i can't make house-firing fight with ace,gl4 and slx. With gl4 enemy will garrison buildings, but when firefight starts, they just leave their house-positions and engage. I want some house to house fights :< Or at least soldiers shooting from windows, without using scripts.

There is a significant disconnect between an effective AI soldier and a user friendly subordinate for the player.

Yes indeed. It would be nice if leaders could set the "independence level" of their subordinates throughout a mission via command menu. Basically, humans leading ai should be able to set the weight of the ai priorities - following your orders vs. trying to stay alive/kill the enemy.

@ChrisB

It is nice to see ai using buildings so often. Would be awesome if BIS could make this default behaviour. But even in your vids, although it is better than nothing, the ai seem to not to be able to use buildings for their full potential. ie. I don't see units leaning out of doorways to fire or better yet peaking in and out of windows to return fire/avoid fire. Those are the types of things that for me, would make the ai really seem human - and that doesn't just go for buildings. cover outdoors is rarely used to its fullest by the ai. And with Arma 3's new animation there is even more "potential" for cover. Those are the type of changes I wish to see... but I have a feeling that it will not become a reality any time soon.

I am interested to know what the new and improved ai configs will do in arma 3, like what exact weird ai behaviour they will fix/improve.

Well I have to disagree, below are numerous pics of ai doing what they usually do in the missions we make. They use windows and doorways, o.k. they don’t lean out, but neither would a good soldier, only the ones in Hollywood lean out of windows or doorways during a firefight. They will use balconies and rooftops mostly in crouched or prone positions depending on cover, there is a pic of one firing from a balcony window, there’s all sorts.

It depends on what you have them set to also what other pbo’s from other mods are there. That’s the only way I can describe it, I cannot and do not script behaviour, I wouldn’t want to as you know whats going to happen. I and a couple of others from our group do lots of testing with pbo mixes, this is how these results were found.

They inhabit the building (GL4), they enter into combat from that position rather than leave the building, they will of course leave the building if they think its to their own advantage, but lots of the time they will move within the building from position to position, or indeed building to building, we think that is a combination of SLX and Zeus. Zeus because we run two ‘find cover’ pbo’s, slx & zeus, the reason for this is not so much find cover, as slx works really well with that, its because when we added the zeus ‘find cover’ the ai started to use outbuildings/small sheds etc that with GL4 they were avoiding, why, who knows, but that’s what testing showed us, so zeus find cover seems to effect where they want to be, so remained in the mix.

Stance wise TPWC has made a difference, they take up correct crouch positions in windows doorways etc and go into prone should they feel they need to, or stand, depends on the ai we leave it to them. Yes they make mistakes but that’s just part of the mix, if it happens and your there to take advantage of their mistake then great. They prove to be very hard ai to fight as you don’t know where they are when trying to take a town, they could literally be anywhere as set to defend they will move from building to building, set to static they take position in the nearest they feel is the best, not just the nearest to them as they may move past certain buildings to take up position in larger buildings with more positions available. It only takes 4-5 small groups of ai to defend a town really well. They can also call reinforcements if they wish, if they use air (heli) support it becomes very hard for our team (humans) to take control of the fight as they communicate with each other (GL4) and the heli will pass on our positions to ground troops etc.

This sounds a utopia to some who play the series, but really this behaviour is not new, GL3 did much the same as did other mods. Its just testing with parts from mods, not necessarily the whole mod.

Anyway, this is the way we as a group play against the ai, all adjustments to skill etc can be done using the mods configs, but having them set to the highest skill would mean you don’t win much, so default seems to be the best in most cases. I put the pbo mix we use in this thread down the page towards the bottom in a spoiler, so whoever wants to try the mix can. If your cpu is upto it there should not be a problem, don’t run other ai mods unless you check via testing as it may alter the behaviour discussed here and shown. ASR_ai works a little with the mix, but we have not checked recently as we do not use asr_ai, so checking yourself would be advisable. Other mods not concerned directly with ai will be fine, we run around 200 inc islands so there is no real problems. I44 doesn’t work with this mix as the ai seem to be disconnected, i.e. default BIS style which is pretty poor, don’t know why, but there you go. All the units/factions we have tested other than I44 work o.k. Even on arma islands such as United Sahrani etc, they work really well, plus the PMC/VTE islands works fine, there are a lot less enterable buildings however, but they use whats there if they need to.

Also its not just town/city fights, they act better out in the open, flanking more efficiently and they are in no hurry to win, which is great as firefights are more intense and may not end in any winner as they may withdraw to regroup and try a different direction. Just play and see is all the advice I can give, they are not our mods, however in our testing they have proved to us to be the best of the ai mods available, that’s why we use them. Must be noted these are for use in your own made missions not campaign missions or other user made missions unless they are made with this mix in mind or at least GL4 in there. I have never played any BIS made campaigns, its all about the Editor anyway with this series, but people have said on this and other forums that some mods mess around with campaign missions, so beware. But if your into making missions and use the mix here incorporated in the mission, plus have the mission laid out correctly (read the ‘readme’) then you should have no problems. The only advice I would give is, any units with waypoints set to ‘defend’ module (gl4) to make sure they reach each waypoint. Also never have an idle group, if it’s a combat area, simple patrol scripts are great and only effect where they go not what they do or act/react. Use reinforcements and synced units in the mission set-up, think when building your missions and it will work out great, well I hope, does depend on cpu available, our group use dedicated gaming pc’s, so there is little else for the cpu to think about, other than the game…

If your still reading this, thanks and sorry its long..

Pics;

all ai apart from the one who took the pic:

http://imageshack.us/photo/my-images/52/arma2oa2012070420272613.jpg/

http://imageshack.us/photo/my-images/210/arma2oa2012070420291219.jpg/

http://imageshack.us/photo/my-images/444/arma2oa2012062212593138.png/

http://imageshack.us/photo/my-images/685/arma2oa2012070115071564.jpg/

http://imageshack.us/photo/my-images/846/arma2oa2012070115061816.jpg/

http://imageshack.us/photo/my-images/208/arma2oa2012070115054411.jpg/

http://imageshack.us/photo/my-images/26/arma2oa2012071218161022.png/

http://imageshack.us/photo/my-images/845/arma2oa2012070403503412.jpg/

http://imageshack.us/photo/my-images/853/arma2oa2012071218222926.png/

http://imageshack.us/photo/my-images/171/arma2oa2012071619233003.png/

http://imageshack.us/photo/my-images/856/arma2oa2012071619222337.png/

http://imageshack.us/photo/my-images/834/arma2oa2012071619220055.png/

http://imageshack.us/photo/my-images/51/arma2oa2012071218140690.png/

http://imageshack.us/photo/my-images/96/arma2oa2012081718581362.png/

http://imageshack.us/photo/my-images/819/arma2oa2012083021495595.png/

http://imageshack.us/photo/my-images/571/arma2oa2012083012232053.jpg/

http://imageshack.us/photo/my-images/11/arma2oa2012083012182729.jpg/

http://imageshack.us/photo/my-images/221/arma2oa2012071816592698.png/

All this needs to happen in Arma 3 as default, then it can be even further built upon. Don't suppose it will, but there you go..

Well I have to disagree, below are numerous pics of ai doing what they usually do in the missions we make. They use windows and doorways, o.k. they don’t lean out, but neither would a good soldier, only the ones in Hollywood lean out of windows or doorways during a firefight. They will use balconies and rooftops mostly in crouched or prone positions depending on cover, there is a pic of one firing from a balcony window, there’s all sorts.

It depends on what you have them set to also what other pbo’s from other mods are there. That’s the only way I can describe it, I cannot and do not script behaviour, I wouldn’t want to as you know whats going to happen. I and a couple of others from our group do lots of testing with pbo mixes, this is how these results were found.

They inhabit the building (GL4), they enter into combat from that position rather than leave the building, they will of course leave the building if they think its to their own advantage, but lots of the time they will move within the building from position to position, or indeed building to building, we think that is a combination of SLX and Zeus. Zeus because we run two ‘find cover’ pbo’s, slx & zeus, the reason for this is not so much find cover, as slx works really well with that, its because when we added the zeus ‘find cover’ the ai started to use outbuildings/small sheds etc that with GL4 they were avoiding, why, who knows, but that’s what testing showed us, so zeus find cover seems to effect where they want to be, so remained in the mix.

Stance wise TPWC has made a difference, they take up correct crouch positions in windows doorways etc and go into prone should they feel they need to, or stand, depends on the ai we leave it to them. Yes they make mistakes but that’s just part of the mix, if it happens and your there to take advantage of their mistake then great. They prove to be very hard ai to fight as you don’t know where they are when trying to take a town, they could literally be anywhere as set to defend they will move from building to building, set to static they take position in the nearest they feel is the best, not just the nearest to them as they may move past certain buildings to take up position in larger buildings with more positions available. It only takes 4-5 small groups of ai to defend a town really well. They can also call reinforcements if they wish, if they use air (heli) support it becomes very hard for our team (humans) to take control of the fight as they communicate with each other (GL4) and the heli will pass on our positions to ground troops etc.

This sounds a utopia to some who play the series, but really this behaviour is not new, GL3 did much the same as did other mods. Its just testing with parts from mods, not necessarily the whole mod.

Well that sounds pretty good. I can't try it out for myself, because my computer will probably explode (It already had trouble handling zeus or slx alone let alone both combined) but I will take your word for it. One question. With this mod cocktail, are the ai less keen to leave cover. One of the problems with vanilla is that they are uber willing to leave cover. They might stay there for maybe 2-3 seconds and then they're bounding off to the next piece of cover which is 20 metres away. This needs to be fixed and I am wondering if one of those mods have done the trick when you play.

Concerning ai fighting from buildings, the impression I got from the vids is that the ai randomly take up position in nearby buildings, but that's about it. it didn't seem like they had a very good idea of how to actually use them. Ie. it seems like ai happen to go to the roof giving them a good view other the enemy rather than the ai deciding "hey that roof would be a good place to see the enemy from" and then going there. I don't know, as I haven't used the mods like you, but that was the impression I got - still tons better than vanilla though.

Well that sounds pretty good. I can't try it out for myself, because my computer will probably explode (It already had trouble handling zeus or slx alone let alone both combined) but I will take your word for it. One question. With this mod cocktail, are the ai less keen to leave cover. One of the problems with vanilla is that they are uber willing to leave cover. They might stay there for maybe 2-3 seconds and then they're bounding off to the next piece of cover which is 20 metres away. This needs to be fixed and I am wondering if one of those mods have done the trick when you play.

Concerning ai fighting from buildings, the impression I got from the vids is that the ai randomly take up position in nearby buildings, but that's about it. it didn't seem like they had a very good idea of how to actually use them. Ie. it seems like ai happen to go to the roof giving them a good view other the enemy rather than the ai deciding "hey that roof would be a good place to see the enemy from" and then going there. I don't know, as I haven't used the mods like you, but that was the impression I got - still tons better than vanilla though.

My own non-scientific impression is that the AI are more unwilling to leave cover than without these addons (which include the TPW suppression addon, probably the most influential). In particular it depends on the amount of return fire you provide.

All this needs to happen in Arma 3 as default, then it can be even further built upon. Don't suppose it will, but there you go..

As it is, last I heard from Gamescom (apparently Gaia by way of InstaGoat) the devs are more concerned with the AI marksmen not drawing handguns at range and stuff like that, than anything to do with cover (read: instead of "officially" adding stuff like this into vanilla ARMA 3, more concerned with the vanilla ARMA 3 AI not making "baby mistakes").

As it is, last I heard from Gamescom (apparently Gaia by way of InstaGoat) the devs are more concerned with the AI marksmen not drawing handguns at range and stuff like that, than anything to do with cover (read: instead of "officially" adding stuff like this into vanilla ARMA 3, more concerned with the vanilla ARMA 3 AI not making "baby mistakes").

Exploitation of cover should also be improved. They realized the problems when working on PMC. However, that said, I didn´t see a lot of AI action at gamescom, and from the videos, you can´t really discern any improvement, or lack of improvement.

This would require proper testing and observing once the alpha is out. Videos and 5 minute impressions don´t tell anything. That said, I have to say that I was a little disappointed with BI´s statements about the AI, mainly because a lot of the things I care about (exploitation of buildings, AI automatically gathering ammo/using backpacks, switching to appropriate ammo in tanks when facing tanks vs infantry, etc) will not be adressed if they don´t get to actually introduce those things.

I am not a friend of AI mods, as great as they are, they often interfere with missions that are not purposely designed around them. I would hope that Arma 3 introduces native AI improvements that allow people like me who don´t do PVP or MP in general to enjoy the missions without having to tack on mod after mod to bring on stuff that should ideally be present in the game as it is delivered. Some other things to mention would be proper handling of suppression/suppression and morale effects, self preservation of AI, and I guess many other things.

It goes to show that, no matter how many things the AI can already do, there´s still a ton of things that could (maybe should) be added to make it more complete. However, sometimes I am even beginning to think that what Arma 3 needs isn´t a revamp or expanded OFP-developed AI, but rather a completely new gen AI system, built from the ground up to deal with the highly complex battlefield they are faced with in A3.

In OFP, barely any buildings were enterable, there wasn´t a lot of clutter to take cover behind, weapons systems weren´t as complex and spotting systems weren´t either, there weren´t as many layers to combat as there are now, and of course the requirements and expectations of the customer have changed. Games like FEAR have redefined what people expect of a game AI, and Arma 2, despite the advances, still falls short rather often. Most AI mods do not really fix this either, but move the problems around to where they are more satisfying to the people that use those mods (such as me.).

However, building an AI from the ground up to the complexity we are currently having with Arma 2/Arma 3 AI would likely be a humongous, time and labour intensive task. It would also be very risky, because if it doesn´t turn out to be noticeably better/more capable, in the end it will be a waste to do.

As far as economy goes, fixing up the existing AI is a good move, because BI obviously doesn´t have the personell and cash reserves of companies like Blizzard, however ultimately it will leave the community with an AI that will still have the fundamental shortcomings the AI had ever since OFP. Dodgy use of cover, friendly fire issues, performance problems, pathfinding problems especially in vehicles, disabillity to use inventory/gather equipment on their own, and of course no intelligent use of buildings for tactical advantage. I don´t know, though, maybe they have an ace up their sleeve they´ve not told us about yet. One can dream.

Exploitation of cover should also be improved. They realized the problems when working on PMC. However, that said, I didn´t see a lot of AI action at gamescom, and from the videos, you can´t really discern any improvement, or lack of improvement.

Well, the answer they came up with in response to the PMC cover issue was, as described by RiE at E3, to simplify the AI's evaluating of cover.

things I care about (exploitation of buildings, AI automatically gathering ammo/using backpacks, switching to appropriate ammo in tanks when facing tanks vs infantry, etc) will not be adressed if they don´t get to actually introduce those things.

Unfortunately this seems to be one of the problems for BI with user input -- I get the sense that the player base is actually pretty fractured in the sense that everyone focuses on specific things and, while BI's development path may have been particularly influenced by what was most common and highest up on players' feedback chains, they may also suspect that they're only getting the most vocal who may not (I suspect probably not) reflect "the silent majority"...

I am not a friend of AI mods, as great as they are, they often interfere with missions that are not purposely designed around them. I would hope that Arma 3 introduces native AI improvements that allow people like me who don´t do PVP or MP in general to enjoy the missions without having to tack on mod after mod to bring on stuff that should ideally be present in the game as it is delivered.

I can respect this, although I would support this for a different reason: to cater to newcomers and give ARMA 3 a reputation that you don't HAVE to mod the game to have a solidly working experience, which then gets into the same "community split" issues that already exists (and that exists in more conventional games with DLC, though in the case of ARMA 2 that's thankfully been averted by BAF/PMC/ACR Lite), the emphasis must be on the unmodded vanilla experience, though of course someone else could then make the case that their preferred mod should be standardized by vanilla... even when that mod's concept and goal is directly contradictory to BI's own goals for ARMA 3? :rolleyes:

Games like FEAR have redefined what people expect of a game AI, and Arma 2, despite the advances, still falls short rather often. Most AI mods do not really fix this either, but move the problems around to where they are more satisfying to the people that use those mods (such as me.).

This goes back to what I was saying -- "more satisfying to the people that" do whatever, though I can't blame you considering what you've previously said to me (hence the need for a good SP experience), but it ties back into what I said about a split community, and not split by official or paid DLC, but by user-generated content.

However, building an AI from the ground up to the complexity we are currently having with Arma 2/Arma 3 AI would likely be a humongous, time and labour intensive task. It would also be very risky, because if it doesn´t turn out to be noticeably better/more capable, in the end it will be a waste to do.

Pretty much... and considering that the most likely to notice subtle changes are the long-timers who in turn were predisposed to buy ARMA 3 anyway... it makes more sense for BI to focus not on the AI being "noticeably better/more capable than in ARMA 2", but to focus on making it so that the newcomer's first impression is "it just works", and if the latter is likelier to result in increased sales for ARMA 3 and more positive brand name for the series and is less risky... then that's not selling out, that's growing up.

Mind you, I'm reminded of the time that when Call of Duty: Black Ops' Combat Training mode only had Team Deathmatch and Free-For-All modes because, as admitted by the game design director at Treyarch, they just couldn't get the AI to fight for the objective. I don't believe that we'll have hear anyone at BI say as much, nor do I actually believe that the coders, programmers and other AI-related devs at BI have this problem... but I can't help but wonder if just getting the AI to work is already pushing the bounds of BI's capabilities (including personnel and cash reserves).

As far as economy goes, fixing up the existing AI is a good move, because BI obviously doesn´t have the personell and cash reserves of companies like Blizzard, however ultimately it will leave the community with an AI that will still have the fundamental shortcomings the AI had ever since OFP.

Which is an odd claim considering your earlier quote (which my previous post was based on) that the AI development plan was in fact to fix fundamental shortcomings instead of piling on layers of "features" bloat over an unsustainably rickety foundation.

Yeah you guys are definitely right about BIS having reason not to invest in larger ai improvements. Not everyone wants it, those who do don't all agree on what "improvement" really is, It is very hard to develop ai, very likely to develop bugs in the process and you might not even create a product any better than the last. In fact there is a chance that you will actually make it worse. And even if you do make it better, people will still find new issues that need to be fixed. But instead you can do something else like improve the lighting which, I believe, is much easier, and is easily noticed and appreciated by all. I am surprised BIS is making any changes to the AI at all. I know I would love to see the ai really improved but logically it is not exactly the best use of resources for BIS. That being said, I hope BIs doesn't totally ignore it, and will try some new stuff in betas and patches to try and give the ai the last push it needs to be more human like.

I am not a friend of AI mods, as great as they are, they often interfere with missions that are not purposely designed around them. I would hope that Arma 3 introduces native AI improvements that allow people like me who don´t do PVP or MP in general to enjoy the missions without having to tack on mod after mod to bring on stuff that should ideally be present in the game as it is deliver

Although mods are great, for some reason I would prefer to go without them as well. I do use them, but they make the game feel less fluid for some reason. Not the actual performance, but the interface and interactions. For example ACE's weapon resting for some reason just doesn't feel like its part of the game. Or TPWC Suppression for the ai or playr. Its not that they don't work, but they seem kind of ... forced. It doesn't feel or look natrual compared to the rest of the game. I don't know how to describe it, but if possible, I always like to keep mods to a minimum. (And I mean know disrespect to any mods/modders, I still love the work you put out and couldn't play without your mods - its just I wish BIS had implemented all those features into the vanilla, so they would "click" with all the other aspects of the game nicer.)

Concerning cover, I kind of get the feeling that if anything, BIS is going to make the ai even less likely to "hold a defensive position". Jay Crowe was talking about the flow of the ai in some vid and he was saying that basically they wanted the ai to be more active and aggressive/responsive(to player). To me this indicates less cover usage/camping/getting "stuck" more banzai charge/fearless subordinates - could be misinterpreting there though.

I really do hope they do something about the ai cover use though. I really don't know much about the programming, but its seems the tools are all there - they seem very close to having it perfect. "sticky" cover positions are defined to the ai, and they know what the appropriate stances to shoot over/around those positions are. Now all that needs to be done is make it so ai "camps" more, and also knows what stance/position they need to use in order to fully protect/conceal themselves and use that stance when they are outgunned/numbered/underfire/taking casualties etc. If BI were to polish up TPWC suppression using some deeper programming/ and then embed it deeper in the ai routines, I think it would have really great effect for the ai cover usage.

Anyway, all that being said, I think that better configs will be great for the ai and I am thankful + looking forward to that. And at least the ai can now finally turn quickly to fight at reasonable pace in CQB:).

Yeah you guys are definitely right about BIS having reason not to invest in larger ai improvements. Not everyone wants it, those who do don't all agree on what "improvement" really is, It is very hard to develop ai, very likely to develop bugs in the process and you might not even create a product any better than the last. In fact there is a chance that you will actually make it worse. And even if you do make it better, people will still find new issues that need to be fixed. But instead you can do something else like improve the lighting which, I believe, is much easier, and is easily noticed and appreciated by all.

As for mods -- I can't say that I'm surprised at the mods "not clicking" and not feeling like they fit, especially (and I assume this to be moreso the case in ARMA 3) when they try to take the engine and gameplay in a distinctly different direction from what the engine was designed for, which sounds like part of the reason that the DayZ standalone won't use RV4: they're so divergent that both interface-wise and under-the-hood, ARMA 3 and DayZ's game design demand different and sometimes directly-opposing things, i.e. "openness" (see how often rocket's talked about why the mod version of DayZ can't be secured against hacking/scripting), and trying to develop both on the same engine would just pull the engine development in sometimes opposing directions.

Concerning cover, I kind of get the feeling that if anything, BIS is going to make the ai even less likely to "hold a defensive position". Jay Crowe was talking about the flow of the ai in some vid and he was saying that basically they wanted the ai to be more active and aggressive/responsive(to player). To me this indicates less cover usage/camping/getting "stuck" more banzai charge/fearless subordinates - could be misinterpreting there though.

I really do hope they do something about the ai cover use though. I really don't know much about the programming, but its seems the tools are all there - they seem very close to having it perfect. "sticky" cover positions are defined to the ai, and they know what the appropriate stances to shoot over/around those positions are. Now all that needs to be done is make it so ai "camps" more

... so, the direct opposite of what you believe will happen and what you say that Jay Crowe said that the ARMA 3 devs want? :D

If BI were to polish up TPWC suppression using some deeper programming/ and then embed it deeper in the ai routines, I think it would have really great effect for the ai cover usage.

This is assuming that BI even liked the idea of suppression -- and as the suppression thread showed, there were fundamental disagreements between posters, mods (?) and devs (though I believe only Celery weighed in?) on the very conception of what suppression is, complete with disagreements about immersion and physiology. How can BI take anything useful from that? :rolleyes:

I'm glad I only dipped my toe into that thread for only a few specific topics before bugging out.

This is assuming that BI even liked the idea of suppression -- and as the suppression thread showed, there were fundamental disagreements between posters, mods (?) and devs (though I believe only Celery weighed in?) on the very conception of what suppression is, complete with disagreements about immersion and physiology. How can BI take anything useful from that? :rolleyes:

Most of the ruckus in the suppression thread was about how it should affect the player (if at all). From what I could tell, most people seemed to agree that the AI should be suppressable.

I should think that the main concern of BIS AI routines is that they act consistently and reasonably, enough for reproducible campaign gameplay anyway. Also, there's got to be enough "room" for user-made solutions too, i.e. addons. I like AI addons, because with 4 or 5 of them I can mix & match & get different game experiences with them. Not everyone feels this way obviously :) but leaving room for addons is a requirement I guess.

---------- Post added at 08:31 ---------- Previous post was at 08:30 ----------

Most of the ruckus in the suppression thread was about how it should affect the player (if at all). From what I could tell, most people seemed to agree that the AI should be suppressable.

Plus, there already is player-suppression. That whole thread was rather useless :)

Plus, there already is player-suppression. That whole thread was rather useless :)

So many are these days. ;)

... so, the direct opposite of what you believe will happen and what you say that Jay Crowe said that the ARMA 3 devs want?

Yep, pretty much. I could always be wrong about what I believe the devs will do... I have my fingers crossed.

This is assuming that BI even liked the idea of suppression -- and as the suppression thread showed, there were fundamental disagreements between posters, mods (?) and devs (though I believe only Celery weighed in?) on the very conception of what suppression is, complete with disagreements about immersion and physiology. How can BI take anything useful from that?

Oh know I just meant ai suppression (the main purpose of that mod).TPWC suppression makes it so that ai change stance depending on incoming fire. So if an ai happened to be in some sort of cover, and bullets start coming his way, he will actually get down and hide himself with the cover, rather than just sit there and get shot up. Also makes the ai skills decrease while under fire to abstractly represent the use of suppression and makes the flow of the firefight more realistic. The thing is, many of these things would work much better if BIS implemented them "deeper down". For example many times if ai is in a sticky cover point, it will refuse to change stance like mentioned above because the stance attached to the cover point takes priority other the script. I didn't at all mean the player suppression stuff, which if implemented should be totally optional.

but leaving room for addons is a requirement I guess.

Yes I really hope they make ai even more accessible to modding.

Ah... I just wish the Ai could try to shoot me faster when they see me. A few new animations for stuff they might do when engaged would aslo be nice.

Edited September 14, 2012 by Ekko

tpw_ai_los is the mod for you my friend.

I would like to see my units engage enemy men who have left their vehicle. As it is now, they can accurately identify the vehicle as "enemy" from a great distance but the men are too far and are classified as "unknown". It makes sense that infantry is more difficult to identify but this is just illogical. In addition, perhaps my units should also engage unknown infantry who are just NEAR the enemy. This would add some depth to the game as friendly units and especially civilians could be in danger of friendly fire if they don't keep a responsible distance from the enemy.

Revealing units that disembark from a known enemy vehicle would probably be a simple AI tweak. Try suggesting it to Robalo in the ASR thread.

Ah... I just wish the Ai could try to shoot me faster when they see me. A few new animations for stuff they might do when engaged would aslo be nice.

Looks like that will be fixed in arma 3. Judgint by some of the vids, the ai actually look like they could be quite scary with there turn speeds in CQB.

It seems to me that AI reaction time is often a question of 'mental focus' and perceived player position rather than reflexes. ArmA 2 AI can shoot instantly, but is hamstrung by the aforementioned issues and a bizarre glitch that limits their arm traverse speed at times.