ArmA 3: Community wishes & ideas- DISCUSSION

[metalcraze]

BIS said there will be gear selection in the editor this time.

Better 11 years later than never

[maionaze]

I think it would be nice if there was a nice little gear editor built into the unit placement... device... thingy.

Don't get me wrong, I don't have a load of issues typing in code into the initialization box, but it is a bit time consuming.

What do you guys think? Am I totally in the wrong?

Nope You will be able to select your headgear, tactical vest, backpack/pouches, comouflage and weapons from a drop down menu within the editor.

[EDcase]

Yep, UI is one of the main things BiS has to work on in the editor.
More options in menus and less typing in fields.

Altering parameters of modules should NOT be done with text files that are made outside of the game
Make parameters with drop down options or numeric input

[Herbal Influence]

Please don't make the mistake of Battlefield 3. Please continue to allow offline usage and not-spying on your users. And allow non-DVD-usage after first upgrade. Thank you!

On the most frequented jounal on politics EA is since yesterday on the front page with the main headline: "EA spying on users with BF3" and there is - for good reasons - an enourmos reaction in the forums!

[Zoz]

After I posted a tought on ArmA's AI on offtopic I decided that maybe it would have suited this thread better.

I am talking about ArmA's AI on a higher than squad level and thinkin of a reason why we are gettin more special forces than combined arms action:

Maybe the reason we are gettin more OMG-ninja-delta-speshul-forces carp is that ArmA's AI cant really simulate a combat on a lets say even platoon level!

What I mean is:
If we have a platoon on the field (lets say 3 infantry squads,1 weapon squad,HQ) some things became really obvious, like:

-No chain of command
-HQs are just guys with guns and a high rank
-Each squad is on its own and doesnt really reads the situation(ie: enemy contact->go danger->attack without evaluating the situation and enemy forces)

Squads are supposed to be connected togheter and work towards the same objective, having them lying around and doing whatever they feel like doing makes it hard to simulate something bigger than squad level operation.

So maybe, since the squad is the biggest coordinated entity we can have on the field(without scripting or High command?) having this OMG-ninja-delta-speshul-forces carp esulating from any chain of command is what would make more sense?

Anyway, I have to say I never really used the High Command feature in my missions so, I ask you: would it be able to simulate something like this?

Also, on a slightly related note, I would like to hear some thoughts on the way the AI fires their weapons.
I think that when AI opens fire it is always aiming to kill. They dont really fire in your general direction just to gain fire superiority, they really aim at you!

Mind you, I'm not complaining about the so called deaths from "random" bullets. That quite never happened to me.


I just mean that the AI being so deadly every time they open fire, leads to shorter-than-real-life firefights.
Its even more weird when guerrilias squad are so deadly.

Just by looking at the bullets-per-kill ratio you can understand how winning a firefight is strictly related to gaining fire superiority.
This is the very role of machineguns after all. They havent got a 800rpm ratio because they need them to kill 800 persons in 1 minute!

Hope someone picks up the discussion so we can expand this a bit, as pretty much my every single post in this forum gets ignored! (I even got a new shiny avatar, so you can see me better!)

I personally would really like to hear some thoughts on that quote of myself above, as I think that the things I explained are what keeps ArmA away from the large combined operations we all love!

[CarlGustaffa]

AI being so deadly every time they open fire, leads to shorter-than-real-life firefights.

They have to be like this to counter our own incredible accuracy. As long as we keep fighting as if it is a video game and not like real soldiers (not fearing for our lives and no breath control to achieve aiming stability and no "forced fear"), that would only make the AI incapable.

[-Coulum-]

I think that you are right zoz. Ai squads must be able to at least try to work with one another. It doesn't have to be perfect but anything would be better than what bi has implemented so far... Which is nothing. I think that their should be another type of group which consists of several smaller groups.
The squad with the highest ranking soldier is the hq of the "platoon". When the platoon meets an enemy the hq will assess the friendly to known enemy ratio, assess the terrain and cover the platoon is in, assess it's overall objective and decide wether to attack, retreat, hold position or try to bypass the enemy. Depending on the chosen course of action the hq will assign squads to basic roles such as suppressive, aimed fire advancing and flanking roles. Depending on each squads role they will fight in different ways.

A really important part of this system is that the ai actually knows when to and how to retreat or seek advantageous terrain to make a stand in. Right now, too often ai will get attacked by greater numbers and instead of running away they go prone and make a stand in the middle of totally open terrain.

as far as ai weapon handling, the ai are far to accurate, far to quickly and rarely try to suppress each other. But as Carl Gustav pointed out, this is simply to balance the fact that the player is also uber accurate.

To solve this I think a more complex aiming system should be implemented that forces you to steady your weapon and breath in order to get accurate shots off. This would allow for the ai to be less accurate and thus would increase firefight times.

But in my opinion the thing that will improve firefights and stretch their length to more realistic times is making the ai not only know how to find cover but also know how to use it. Finding cover means that they can put an object in-between them and the enemy- they already do this relatively well. Using cover means when a shot wizzes by them they will actually hide behind the piece of cover making themselves totally out of sight from the shooter. Basically ai should actively be peaking in and out of cover depending on how much fire they are taking. Not only would this make it harder to shoot enemies quickly and make firefights longer and tougher, it would also introduce real suppressive fire.

[PuFu]

They have to be like this to counter our own incredible accuracy. As long as we keep fighting as if it is a video game and not like real soldiers (not fearing for our lives and no breath control to achieve aiming stability and no "forced fear"), that would only make the AI incapable.

They could always increase the sway some more. I mean i have a HUUUGE sway when moving around with sights up (even in the slowest mode available in A2), a lot more than anyone with 2 capable hands and somewhat acquainted with a rifle will ever be able to achieve without some sort of mechanical illness, but i get to keep the weapon straight on no matter of stance (which reminds me of the fact that one cannot affect the sway individually, crouch and prone are defined the same, while standing is different :|)...

[Zoz]

I think that you are right zoz. Ai squads must be able to at least try to work with one another. It doesn't have to be perfect but anything would be better than what bi has implemented so far... Which is nothing. I think that their should be another type of group which consists of several smaller groups.
The squad with the highest ranking soldier is the hq of the "platoon". When the platoon meets an enemy the hq will assess the friendly to known enemy ratio, assess the terrain and cover the platoon is in, assess it's overall objective and decide wether to attack, retreat, hold position or try to bypass the enemy. Depending on the chosen course of action the hq will assign squads to basic roles such as suppressive, aimed fire advancing and flanking roles. Depending on each squads role they will fight in different ways.

A really important part of this system is that the ai actually knows when to and how to retreat or seek advantageous terrain to make a stand in. Right now, too often ai will get attacked by greater numbers and instead of running away they go prone and make a stand in the middle of totally open terrain.

as far as ai weapon handling, the ai are far to accurate, far to quickly and rarely try to suppress each other. But as Carl Gustav pointed out, this is simply to balance the fact that the player is also uber accurate.

To solve this I think a more complex aiming system should be implemented that forces you to steady your weapon and breath in order to get accurate shots off. This would allow for the ai to be less accurate and thus would increase firefight times.

But in my opinion the thing that will improve firefights and stretch their length to more realistic times is making the ai not only know how to find cover but also know how to use it. Finding cover means that they can put an object in-between them and the enemy- they already do this relatively well. Using cover means when a shot wizzes by them they will actually hide behind the piece of cover making themselves totally out of sight from the shooter. Basically ai should actively be peaking in and out of cover depending on how much fire they are taking. Not only would this make it harder to shoot enemies quickly and make firefights longer and tougher, it would also introduce real suppressive fire.

Having some sort of basic coordination beetween groups would be very nice (read: essential) to have, but I know that would take a huge lot of work.
Just imagine what would be like to have the AI being to do all the things you posted! (Detection and classification of threats, understanding the environment they are acting into, being able to transmit infos on a realistic chain of command using: radio, visual contact, put on a coordinated and balanced effort, and so on..)
I would be watching AI meeting engagements all day!

Anyways, to have the "AI platoon" react in a realitic way, they should be told via editor the situation.
They must have clear objectives and knowledge of the surroundings and general situation. That would be hard to achieve. Still doenst mean we cant have something even very basic.

There would also be the need to simulate suppression effects a bit better!
Its all about using your assets to create fire superiority, thats how you not take casualties and can create room for manouvering. If you dont have fire superiority you fall back!

Sometimes I think AI doesnt understand how their lifes are at risk here!

Anyway, considered all I said before, I would even settle for a "small" change like:
AI squads (and I mean the squads we have now) being able to decide whether or not to retreat or attack. Just that. But this would probably lead to a infantry squad running around the island while being stalked by mech forces. Not that it would be a problem to me, but maybe to mission makers.. (module request!)

Right now every time a AI is spawned it doesnt care to know where it is or what it is doing. He just knows, that no matter what, he will need to attack!

That is sad.


On the matter of firefights, considering what you guys said, the best solution would probably be better cover system, as the easiest way to increase survivability is to cover most of body while fighting.
But I think we are getting that, already.

[CarlGustaffa]

Right now every time a AI is spawned it doesnt care to know where it is or what it is doing. He just knows, that no matter what, he will need to attack!

That is sad.

Not entirely true, because that is decision making left to the mission maker. We can manipulate waypoints, change their modes, cause them to flee or surrender, disable leader to not issue attack orders and so on. The reason few tries to do it is that it is painfully difficult and that it creates too serious overhead you really don't want to have.

I've done artillery decision making using fsm. Like evaluate situation and choose what to do based on that. It evaluates lighting, use of spotting rounds, mortar or artillery, enemy types (not density), choose weather or not to use illumination coordination (slightly random), re-evalution of the situation after spotting rounds, accounting for enemy movements, friendly activities in the area - basically all the stuff I would have tried to think of as a forward observer. Even at 84kB, this is a fairly easy decision making process compared to realistic "what to do and how to do it" behavior on a per unit/group basis. I fear that what can now run 100 AI units would only allow 10 AI units with this stuff running, but without causing the mission to get 10 times more difficult because of it.