Do ArmA Firefights Exsist?

[-Coulum- 35]

Cover...

Life..

Aim..

Agree with what you listed, but I think there is also something to be said for the ai as well. There are some problems with them that limit firefights as well.

Spotting - The ai are able to see you far to easily in some situations, Ie. when you are ghilied up, prone and just visible between a bunch of bushes and trees. Of course the ai are also horrible at spotting in other situations, but their ability to spot camouflaged and partially concealed units leads to more point fire, less area fire, more of a need to kill rather than to suppress and shorter firefights. It would be hard to truly fix this, but even primitive measures could be taken to help. Ie. if a unit is within 2 metres of a bush he is 100% harder for the ai to see, if he is within a metre of the interior of a window he is 50% harder to see, if he is in rubble he is 75% harder to see etc etc.

Communication - To amplify the above spotting problem, the ai is far to quick and far to good at sharing info with their squad mates. This means when one guy sees you ghillied up and prone hiding in the bushes, the entire squad also knows your exact location. They can all achieve pinpoint accuracy now. you might as well be out in the open in an orange jump suit.

My take on "reality" is that if one squad member was lucky enough to pinpoint your location (which among the chaos of a firefight isn't likely) he would try and communicate your general position to the rest of the squad who will try and lay down area fire on your general location... and this communication wouldn't happen nearly as fast as it does in arma. Fixing this so that ai squad members could help other's see the enemy but not see for them would make firefights last longer already - instead of 10 pinpoint accurate shots coming at the target you would have one accurate shot and 9 suppressing shots. It would also help to eliminate the illusion that ai are aimbots or have X-ray vision.

Lack of emotion - The ai don't feel fear or panic or stress. Thus there ability to spot and shoot is not hampered, even when explosion are going off, sonic cracks are snapping all around them, bullets are kicking up dirt infront of them, and "buddies" are getting mowed down beside them. They still aim in their slow, methodical and precise way and spot with the same speed and accuracy as if they were not under fire. This leads to the need for kill shots (because simply firing near him won't effect him as it might in reality) and also leads to accurate return fire from ai even when all hell breaks loose around them. Both of which will obviously shorten the fire fight and make the need to kill that much more important than the need to simply keep the enemies heads down. Ideally aim and spotting ability of the ai should be dependant on the situation they are in. Ai should also be willing to sacrifice accuracy for speed if they are underfire.

Use of cover - well this is one that I know is hard to fix but I will still harp on about. The ai doesn't often take available cover or even concealment, and when it does, it doesn't use it to its full potential. The great thing about cover is that when someone is firing at you, you can duck down and they will have no chance of hitting you. The ai fails to understand that, and even though they are making themselves harder to hit by covering a portion of their body it merely means it will take a couple extra shots to kill them. As is, ai's use of cover doesn't help them much more than going prone in the open - it just makes them a harder target. Ideally, ai should be switching between a firing stance and a hiding stance when behind cover, depending on the situation they're in, to maximize their survivability, and to make it so that they are not only making themselves a smaller target, but also making themselves not a possible target at all.

These ai issues, combined with the issues JetlinerX listed cause firefights to play out more like skirmishes. Infact I would actually say that they play more like timed target practices, where if you don't hit the targets before time is up you die. IMO pretty boring, although still possibly entertaining when you have a good sound mod, some friends to play with, and shut down the part of your mind saying "this isn't even close to how things would really go down".

[orcinus 121]

@-Coulum-

Good points (not the only ones in this thread, ofc). I was considerably impressed by the way AI units avoided crossing open spaces using "findcover" from Gunter Severloh's reworked SLX (COSLX). Findcover has more to do with movement than firefights AFAICT; in combat the Dodge feature kicked in. However since at least 1.61 neither seems to work any more :(

I guess there's some conflict with either the additional cover-finding in the core game, or with asr_ai &/or tpwcas. Way beyond my puny skills to try to update them.

[chrisb 196]

@-Coulum-

Good points (not the only ones in this thread, ofc). I was considerably impressed by the way AI units avoided crossing open spaces using "findcover" from Gunter Severloh's reworked SLX (COSLX). Findcover has more to do with movement than firefights AFAICT; in combat the Dodge feature kicked in. However since at least 1.61 neither seems to work any more :(

I guess there's some conflict with either the additional cover-finding in the core game, or with asr_ai &/or tpwcas. Way beyond my puny skills to try to update them.

Both SLX and Zeus find cover pbo’s work fine, right upto and including the last beta.

Nothing changes much in the core game concerning ai, I still have the same reaction from ai that I have always had since using the mods or parts of mods I use and that’s great reactions.

Finding cover, as seen in many of my ai vid test’s (in sig), is an important part of A2 and realism, without it you just don’t really have a realistic game. It’s the same for using buildings, rooftops, balconies etc, what’s the terrain without ai using them, again plenty of examples in the test videos..

I don’t use ASR_ai. TPWC_suppress does not interfere with find cover, in-fact it helps enhance the movement toward cover, better use of stances, (need I say, lots of examples in the video's:rolleyes:)..

Short example:

Using Zeus find cover..

Edited January 21, 2013 by ChrisB

[-Coulum- 35]

in combat the Dodge feature kicked in. However since at least 1.61 neither seems to work any more

What exactly did the dodge feauture do? As far as I know the SLX findcover should still work although I haven't tried that mod since the summer (not sure if that was pre 1.61).

The problem with the modded solutions to finding cover is that they often confuse the ai. They'll be moving toward the cover that BIS has assigned them then be told by SLX or Zeus or even certain versions of TPWC to go to another location of cover. They'll be told to advance by BIS and then told to retreat by mod x. They'll be told to stop and shoot and then told to run into some building obscuring their shot. My personal experience with mods that try to replace what bis has already done is the ai is confused and appears to have some attention disorder, never sitting down and focusing on completing a single task. They often become very easy to dispatch at this point, even if they are harder targets due to their unpredictable behaviour, because they are not often shooting at you. BIS default ai with a bit of skill modification may not go into buildings, and debatably may not use exterior cover as much, but at least for the most part it is able to focus on completing a task. IMO an ai lying down in the grass shooting at you is more challenging then an ai running back and forth like a chicken without its head and happening upon cover, occasionally finding the time to shoot a couple rounds your way (if you let it live that long that is). Neither are realistic however, and both lead to a real lack of good quality fire fights.

I believe the best thing bis can do for the cover problem is to improve upon their "sticky cover system", making it so there are more cover position related to each object, including windows and doors in the list of possible cover points an ai can use, and attaching a safe and firing stance/lean to each cover position. To help modders/scripters, they need to allow us some way of getting access to those cover points and telling ai to move to them or change their stance behind them. Something similar to the "findcover" command that is now outdated, as well as a "isincover" type command. This way modders/scripters can set up more mission specific behaviours for ai. This kind of access is necessary because it would be quite impossible for bi to program the ai to react to every situation in the way the mission designer intended. Ie. if the ai is made to be super campers, always staying in cover, a D-day mission would be impossible to make. Basically, fix the low level individual abilities of the ai to interpret and navigate the map, and let scripters use those abilities to create the behaviours they desire.

[katipo66 94]

Both SLX and Zeus find cover pbo’s work fine, right upto and including the last beta.

SLX find cover has amazed me at times, ive seen ai move quickly to cover in a crouched position without hesitation, when ai use cover well it's a different game, its just getting them there, of course there will always be the times when they are caught out in the open And easy pickings with a scoped rifle, more fun with iron sights.

[Tonci87 163]

@ChrisB

You made me curious and now I have two tasks for you

Task 1

Download Troopmon 2 (A great tool to monitor AI behaviour, you can get it from Armaholic)

Create this small scenario.

Takistan

Chardarakht (South-East on the Map)

Blufor Squad (2xRifleman, 1xLMG, 1xDMR, 1xTeamleader) enters the town from the west with a move waypoint at the opposite side of the town

OPFOR Squad (2xRifleman, 1xLMG, 1xDMR, 1xTeamleader) enters the town from the east with a move waypoint at the opposite side of the town

All have max skill level.

Monitor the firefight with Troopmon, or camera.sqs

Task 2

Download the campaign Chesty Puller by Bardosy

http://forums.bistudio.com/showthread.php?87392-CAMP-Chesty-Puller

and play the beginning of the last mission (use campaign cheat).

Use high command to send Teams 3,4,5 into the town.

Observe their advance.

Upload them to YT

Task 1 will get some nice results in urban CQC while task two will show Ai behaviour in large scale engagements.

Please also list your settings (and if you used PC1 or PC2 [bTW, I see you are using the Roccat Kone+ , is it good?]) + average FPS in both scenarios if possible.

The results should be very interesting!

Edited January 21, 2013 by Tonci87

[ProfTournesol 956]

The only time i'm afraid to die is when the mission is long and there's no savegame available.

[Tonci87 163]

The only time i'm afraid to die is when the mission is long and there's no savegame available.

I´ve played large scale TvT Tournaments where a games goes for two hours and there is no respawn.

That´s as close as you can come to being afraid to die in Arma. Never did I run away that often in firefights.

On the other hand, it´s great if you manage to win a fight, because every life counts.

Best situation ever: 10 Minutes after game start 3 of us moved down the road from Stary to Novy Sobor. Suddenly we hear a MTVR approaching from behind. We tokk cover at the side of the road and opened fire as the truck was only a dozen meters away from us. The driver was killed immediately and the vehicle crashed into a bush. Our AKs and my RPK quickly dispatched the disembarking US soldiers (the truck was full). A whole squad had to find something else to do for that evening. That feeling=Priceless.

[NoRailgunner 0]

Default AI needs to be finetuned to observe something like good firefight and for longer firefights you need to supply them with ammunitions. Cross fingers that in A3 the default AI can use and rearm themselves - at least through their own backpack/pouch/belt.

[chrisb 196]

@ChrisB

You made me curious and now I have two tasks for you

Snip::

The results should be very interesting!

I have replied in the spoiler.

The pc I use for A2 is pc1. My other pc is waiting for A3, I am sorry now I bought it so early, may end up changing the card when A3 comes out, not sure yet.. Both are gaming pc’s.

The Kone + is great, simple button functions, two buttons on the side I have set to ‘walk’ and ‘freelook’, right click on the mouse wheel I have set to ‘right mirror’ left to ‘left mirror’, (blakes mirror’s), then when I’m out of a vehicle they’re set to ‘Gear’ & ‘Map’. Two buttons behind the wheel are set to ‘Reload’ & ‘change sights’. Button to the front of the wheel is for dpi changing on the fly. Mouse wheel click, I don’t use..

The software is good and the set-up is easy, can be a little temperamental regarding gaming surface, if you use the TCU tool to scan your surface it helps.. There are plenty of set-up tools to use/change.

As to your tasks, I don’t have or download any campaigns, just play our own ongoing campaign.

.

Regards Troopmon, it’s a while since I tried it, I have it on one of my external hdd’s, it's good as a tool if you like that and are used to using it (I’m not).

However, I would be interested to see your small test mission, if your used to using TM, do the task and upload it.

Regards fps, I lose 30-35fps recording (msi), so it does hit performance a lot, that’s why many vids are short and in low quality settings, quicker to upload to YT.

Settings I use are mainly high-v/high, the view distance in urban would be 1500 max, possibly as low as 1000, no lower than that, open areas 3000 max (on the ground).

With your task 1, my guess is; as you describe the layout and using the amount of units you say, my fps would be reasonably high probably between 55-70 in that location, any desert terrain I do quite well. I have some vids on my channel showing fps performance on different terrains, although as said, there is that fps loss recording. GPU usage would be top end of the 90’s and the gpu mem usage would be 1100-1200 ish or a little higher, I have a 2gb card.

Regards the campaign fps, I have no idea, where is it set and how many ai?

I hope that ai in A3 are better equipped for CQC, I know with a mix of pbo's it can be very good in A2, but its never going to be perfect, even A3 will not achieve that, its just not that type of game, but it can be really good, just have to test really..;)

[Tonci87 163]

I have replied in the spoiler.

The pc I use for A2 is pc1. My other pc is waiting for A3, I am sorry now I bought it so early, may end up changing the card when A3 comes out, not sure yet.. Both are gaming pc’s.

The Kone + is great, simple button functions, two buttons on the side I have set to ‘walk’ and ‘freelook’, right click on the mouse wheel I have set to ‘right mirror’ left to ‘left mirror’, (blakes mirror’s), then when I’m out of a vehicle they’re set to ‘Gear’ & ‘Map’. Two buttons behind the wheel are set to ‘Reload’ & ‘change sights’. Button to the front of the wheel is for dpi changing on the fly. Mouse wheel click, I don’t use..

The software is good and the set-up is easy, can be a little temperamental regarding gaming surface, if you use the TCU tool to scan your surface it helps.. There are plenty of set-up tools to use/change.

As to your tasks, I don’t have or download any campaigns, just play our own ongoing campaign.

.

Regards Troopmon, it’s a while since I tried it, I have it on one of my external hdd’s, it's good as a tool if you like that and are used to using it (I’m not).

However, I would be interested to see your small test mission, if your used to using TM, do the task and upload it.

Regards fps, I lose 30-35fps recording (msi), so it does hit performance a lot, that’s why many vids are short and in low quality settings, quicker to upload to YT.

Settings I use are mainly high-v/high, the view distance in urban would be 1500 max, possibly as low as 1000, no lower than that, open areas 3000 max (on the ground).

With your task 1, my guess is; as you describe the layout and using the amount of units you say, my fps would be reasonably high probably between 55-70 in that location, any desert terrain I do quite well. I have some vids on my channel showing fps performance on different terrains, although as said, there is that fps loss recording. GPU usage would be top end of the 90’s and the gpu mem usage would be 1100-1200 ish or a little higher, I have a 2gb card.

Regards the campaign fps, I have no idea, where is it set and how many ai?

I hope that ai in A3 are better equipped for CQC, I know with a mix of pbo's it can be very good in A2, but its never going to be perfect, even A3 will not achieve that, its just not that type of game, but it can be really good, just have to test really..;)

You should try Bardosys campaign.

The last Mission is your Teams entering Zelenogorsk from the south while there is already fighting in the town.

I suspect your FPS to go down considerably compared to the first test scenario, and I think that your AI groups will find a cover position and stay there until all enemys are dead instead of advancing on the town as ordered. Problem is that they will most likely run out of ammo until this happens.

That is one of the many problems of the PBOs you use. They work fine for small scale Team vs. Team engagements but fail in big battles.

I´m thinking about getting the new Kone XD...

[chrisb 196]

@tonci, in spoiler

I understand what your saying, I have mentioned on many posts regards the pbo list’s I use, that missions should be made using a GL4 format.

If ‘Bardosys’ campaign is not made with that, then it would probably break.

The mix works for any size battle large or small, depends on cpu of course. They will move up as a group but only when they feel it safe 'tactically wise' to do so, no WW1 tactics here, 'over the top and keep moving' belongs to the past.;).

Kone +, good mouse.. Of the three I have at the moment, the Kone+ is doing well.. I'm looking at the Rat9 however.:).

On topic:

I re-comfirm to anyone wanting to use a pbo mix that includes GL4, use it for missions you are making, it really is completely reliable in GL4 format missions and will give you by far the best results for A2, imo.

Other missions/campaigns, however, may break, that means your fire-fights may never happen:butbut:

I have never played any other campaigns, either BIS or community. No point, I started playing 'War Games' in the late 70's and have progressed with a certain campaign format, which luckily enough this series brought to life just a decade ago with the, utterly great, 'Editor', prior to that the only reliable way to set out our missions was on a board/pc..:)

[iceman77 18]

Default AI needs to be finetuned to observe something like good firefight and for longer firefights you need to supply them with ammunitions. Cross fingers that in A3 the default AI can use and rearm themselves - at least through their own backpack/pouch/belt.

Doesn't the AI go to the 1st available source to rearm if they are out of ammo?

[NoRailgunner 0]

@Iceman77 - default AI are turning into sitting, walking or running ducks and moaning that they run out of ammo even if their own backpack is full of ammo. ASR AI helps to get AI ammo on their own or even another weapon with ammo. Another issue is how often the AI uses binoculars, rangefinders or laser designators during combat or if the group should move/advance. Sadly the best workaround to enjoy a mission is still to remove binoculars, rangefinders, laser designators and pistols/handguns from AI.

[katipo66 94]

Why not use the function in stealth mode that allows AI to move along cover and include it in the standard move option, so that when they are on the move they might at least stay close to cover as much as possible instead of strolling in big open fields where they can easily be picked off.. It might not help in creating longer firefights but it might reduce the amount of times they get caught in the open and provide a greater challenge?

[armazingerz 1]

I guess it's much harder to aim someone when you have to hold on a 4 or 5 kg rifle

[-Coulum- 35]

I guess it's much harder to aim someone when you have to hold on a 4 or 5 kg rifle

Yes, but more importantly, its much harder to aim when all you want to do is hide, and the enemy is lying in the high grass, bushes and shadows 300 metres away, and you have no clue where to start looking for him, while you think he has his sights aimed right for where your head will pop up when you try and look.

I do agree though, heavy weapons should have more sway, or at least should take longer to "settle down" after movement or sudden moves of the mouse. (The latter being a feature I would love to see for all weapons depending on weight)

[sidekilla09 11]

Can you PM me that pbo mix? For years since i started playing Arma 2 i always wanted ai to use cover the right way and use buildings most addons dont do it correctlly. Can you send that list to me i would like to run some test of my own if you dont mine..

---------- Post added at 07:02 PM ---------- Previous post was at 07:01 PM ----------

@tonci, in spoiler

I understand what your saying, I have mentioned on many posts regards the pbo list’s I use, that missions should be made using a GL4 format.

If ‘Bardosys’ campaign is not made with that, then it would probably break.

The mix works for any size battle large or small, depends on cpu of course. They will move up as a group but only when they feel it safe 'tactically wise' to do so, no WW1 tactics here, 'over the top and keep moving' belongs to the past.;).

Kone +, good mouse.. Of the three I have at the moment, the Kone+ is doing well.. I'm looking at the Rat9 however.:).

On topic:

I re-comfirm to anyone wanting to use a pbo mix that includes GL4, use it for missions you are making, it really is completely reliable in GL4 format missions and will give you by far the best results for A2, imo.

Other missions/campaigns, however, may break, that means your fire-fights may never happen:butbut:

I have never played any other campaigns, either BIS or community. No point, I started playing 'War Games' in the late 70's and have progressed with a certain campaign format, which luckily enough this series brought to life just a decade ago with the, utterly great, 'Editor', prior to that the only reliable way to set out our missions was on a board/pc..:)

Can you PM me that pbo mix? For years since i started playing Arma 2 i always wanted ai to use cover the right way and use buildings most addons dont do it correctlly. Can you send that list to me i would like to run some test of my own if you dont mine..

[chrisb 196]

Can you PM me that pbo mix? For years since i started playing Arma 2 i always wanted ai to use cover the right way and use buildings most addons dont do it correctlly. Can you send that list to me i would like to run some test of my own if you dont mine..

---------- Post added at 07:02 PM ---------- Previous post was at 07:01 PM ----------

Can you PM me that pbo mix? For years since i started playing Arma 2 i always wanted ai to use cover the right way and use buildings most addons dont do it correctlly. Can you send that list to me i would like to run some test of my own if you dont mine..

There is a good ai pbo list in my sig, just have a look and collect the pbo’s required. Once you have them all, pm me and I will run through it with you. There are some config alterations etc that you will have to do, but best just start by collecting the mods you need, then go from there..:)

This post also applies to anyone else that has asked me and I've forgotten, just collect the pbo's (from the mods) required and pm me, thanks, and sorry for forgetting..:o

_______

EDIT:

Take a look at this two part ai firefight test (too long for youtube 15mins limit, so second part is only a couple of mins long).

It’s a similar format to the one I often use for this type of test, but this time I try and stay out of the action, just observing instead, although temptation gets to me in the end.

The set-up: ‘two opposing sides running straight at each other with no cover in-between, the only cover provided is some way off to each side of each team, in the form of two small settlements. One waypoint given to each side ‘Move’ to the opposing sides start point, everything else set to ‘No Change’, that’s it. I have raised the ai skill level a touch, so they are now around the mid skill range’.

Its an interesting fire-fight, they carry it off really well overall, staying at longer ranges and in cover most of the time. The enemy are launching over grenades, not quite hitting target but certainly getting closer each time, the enemy also try moving down a little to get a better angle on the grenade launching.

As said, overall really good ai behaviour, some repeating actions are going on, but there again that’s to be expected, especially when the area between the two forces has no cover, whatsoever. They have to think harder to complete the task, rather than just moving forward towards each other all the time, out in the open without some viable cover, in a painfully unrealistic way.

Instead they are made to think, they choose to fire-fight at longer range, using high positions, moving around from position to position where possible, also taking good cover in buildings etc.

We join the video here, just after our team leader has ordered us to move to cover.

Pt 1

https://www.youtube.com/watch?v=N46bQ9_bPRo

Pt 2

http://www.youtube.com/watch?v=QI7RrTPS15Q

:)

Edited February 6, 2013 by ChrisB
Additional content.

[ollem 4]

The problem with the modded solutions to finding cover is that they often confuse the ai. They'll be moving toward the cover that BIS has assigned them then be told by SLX or Zeus or even certain versions of TPWC to go to another location of cover. They'll be told to advance by BIS and then told to retreat by mod x. They'll be told to stop and shoot and then told to run into some building obscuring their shot. My personal experience with mods that try to replace what bis has already done is the ai is confused and appears to have some attention disorder, never sitting down and focusing on completing a single task.

I agree with you and that is why I've implemented a basic 'Find Cover' solution in TPWCAS (which can be switched off in config as well)

The main objective has been to keep it as simple as possible (less cpu strain) and also to interfere as little as possible with default AI behavior.

- if unit is already in cover (there is cover between shooter and target) no action let default AI behavior continue

- if there is cover nearby (7 m default configurable value) the AI is ordered to move to cover side of the cover object (only objects with minimal size are considered)

- if unit is moving in wrong way move order is skipped and default AI behavior takes over again

- if unit takes too long to move to cover move order is skipped and let default AI behavior take over again

- If unit moved into cover position, after a short random time default AI behavior takes over again

- if unit is within 25m (default configurable value) tpwcas is ignored for the situation and find cover is therefor also ignored

Code is minimal and doesn;t check if a unit can

I've added debug option to monitor behavior and though of course it's not perfect, it certainly often moves AI into a safe position.