Will infantry be fixed in CMx2

[IICptMillerII]

It seems to me that you are both arguing the same thing by trying to explain it in different ways. You each have a different perspective of the same thing. 

Infantry are visibly represented 1:1. Each soldier has his own TacAI. One man in a squad can still see the enemy if the rest of his mates cant.

Hits are 1:1. If a bullet hits a pixel of a soldier then he is hit. Depending on where the bullet impacted the soldier determines if he is lightly wounded heavily wounded or killed. 

This is where the abstraction comes in. If you don't believe me here then I urge you to go in game and see for yourself. What I said above is true. If a bullet (or shrapnel or whatever) hits the pixel of a soldier then that soldier is hit. However this does not mean that the soldier must be killed or wounded. The game uses a save system to determine if soft factors have saved the soldier or not. Micro-terrain, furniture, stones, and any other number of small things that the game does not visually represent. I have seen my soldiers hit by direct fire many times and not be killed or wounded by it. You can even see the tracer "splash" on the pixeltruppen but he is not harmed in any way. This is the abstraction. 

The example of some soldiers clustered up behind a lamp post; if a mortar round lands on top of them then they will most likely be wounded or killed. If one or more is not it is because the abstracted saves occurred. If they are all taken out then the saves were not enough to actually save them. The locations of the soldiers and the rounds that impact them are all 1:1 but there are behind the scene saves that can occur. 

Lets take another example. A rifleman is shooting from behind a tree. You see return fire literally shoot him in the face, but the rifleman behind the tree is unaffected. (For simplicity sake lets just say hes a veteran or crack so he doesn't immediately dive for cover) Why is he not dead? Did the bullet actually hit him in the face? Yes. Is he unharmed? Yes. Then what gives? There was a cover save applied. The game determined that the soldier was leaning around the tree not just sitting out next to it all exposed like you see on your screen and the game decided that because he was leaning out from behind a tree he was not hit. 

I hope these examples and explanations are not confusing at all. For me when I see something such as the example above I take a moment to thin it through and then use a little imagination. That one German sniper in the window who just won't die? Maybe he stacked furniture or even some sandbags up there with him thats protecting him. The MG42 that has survived two or even three mortar barrages? Probably a bit more dug into the dirt berm of the hedgerow than the games visuals can show. Perhaps he even has some improvised overhead cover. The point is is that you can explain a lot of what you're seeing in game even if you can't directly see it. 

I also tend to treat situations like these as micro challenges. Just because the first MG you encountered was easy to wipe out does not mean the second one will be just as easy. These are the challenges that face the infantry every time they go into combat, and is one of the many reasons why the infantry are the most important on the battlefield. They are far and away the most adaptable to these challenges, whether they are making it harder for the enemy to kill them or trying to undo what measures the enemy has taken.

[Bulletpoint]

1 hour ago, Vanir Ausf B said:

If pixeltruppen were more spread out they would presumably un-nerf HE so the net effect might be a wash.

That would be nice, I think.

[A Canadian Cat]

56 minutes ago, IICptMillerII said:

It seems to me that you are both arguing the same thing by trying to explain it in different ways. You each have a different perspective of the same thing. 

Infantry are visibly represented 1:1. Each soldier has his own TacAI. One man in a squad can still see the enemy if the rest of his mates cant.

Hits are 1:1. If a bullet hits a pixel of a soldier then he is hit. Depending on where the bullet impacted the soldier determines if he is lightly wounded heavily wounded or killed. 

This is where the abstraction comes in. If you don't believe me here then I urge you to go in game and see for yourself. What I said above is true. If a bullet (or shrapnel or whatever) hits the pixel of a soldier then that soldier is hit. However this does not mean that the soldier must be killed or wounded. The game uses a save system to determine if soft factors have saved the soldier or not. Micro-terrain, furniture, stones, and any other number of small things that the game does not visually represent. I have seen my soldiers hit by direct fire many times and not be killed or wounded by it. You can even see the tracer "splash" on the pixeltruppen but he is not harmed in any way. This is the abstraction.

The way @womble and @llCptMillerll are explaining it is much more reflective of how the game works.  @shift8, as soon as you start talking about infantry as "just a "block" with xxxx fire, xxxx movement, xxxx cover, xxxx leadership xxxxx etc. " you are drifting away from how the game works.  You might think that you end up in the same place but that mental model is not correct and could lead to thinking that a squad with two MG42s and several rifles will always be more powerful than a squad with the same number of guys but only one MG42.  This is not the case.  If the two squads are positioned in such away that only two guys with rifles can see the enemy then that enemy experience the same incoming fire from each of those two squads.  It is *not* fire power xxxx / n it is just two guys with rifles.  This is because the game *does* represent infantry on a 1:1 basis. 

You are quite correct that there is math and abstraction in behind that but the 1:1 is real and there are not blocks of firepower etc. 

[shift8]

21 minutes ago, IanL said:

The way @womble and @llCptMillerll are explaining it is much more reflective of how the game works.  @shift8, as soon as you start talking about infantry as "just a "block" with xxxx fire, xxxx movement, xxxx cover, xxxx leadership xxxxx etc. " you are drifting away from how the game works.  You might think that you end up in the same place but that mental model is not correct and could lead to thinking that a squad with two MG42s and several rifles will always be more powerful than a squad with the same number of guys but only one MG42.  This is not the case.  If the two squads are positioned in such away that only two guys with rifles can see the enemy then that enemy experience the same incoming fire from each of those two squads.  It is *not* fire power xxxx / n it is just two guys with rifles.  This is because the game *does* represent infantry on a 1:1 basis. 

You are quite correct that there is math and abstraction in behind that but the 1:1 is real and there are not blocks of firepower etc. 

The thing here is that I completely agree with what Miller said. There is more or less no difference of opinion there. If you dont want to use the "block" terminology fine, but that is how I think of it on macro level. 

 

I am not sure though how you think my mental model wouldnt add up to what you said about the Mg. Given that Cpt Millers explanation is more or less in line with my own, I think you may be misunderstanding what I mean when I talk of blocks. My mental model of that situation works fine. For example, if the total abstracted fire power sum of the former squad is 10, and the later 8, in a open field all things being the same the former would most likely win. I think we would agree? However, if the Mg is obstructed, this would depreciate the firepower of the former squad accordingly, relative to the Fov of the user. Im not denying the specifics of the game mechanics, as I have already said. There is a certain degree of nuance to this that I think everyone might be missing. I am simply summarizing the units as blocks. You could also choose to further break each man into sub-blocks. However, in many situations the units would be affected more as a single block in certain geometries. 

 

So here is a more fine tuned definition of what I mean: Each infantry unit is a "block" made up of "sub-blocks" that are capable of having independent actions levied upon them. In the case of spacing for units, I consider it to be a non-issue because the units in general are re-affected by additional modifiers that reflect the terrain or position they are in. So long as the general modifier for a given space compensates for spacing, the end-result vulnerability is the same. Vanirs point on the HE is an example of this. So in summary, the spacing is a non-issue because the overall effectiveness of fires in respective environments is correct imo. 

Edited June 21, 2016 by shift8

[A Canadian Cat]

There is no fire power number. The game is tracking simulated bullets fired by simulated soldiers.

I think you should consider tweaking the way you are looking at it. Even though you might be working out the same result in this case the way you are thinking about it will will diverge from the way the game behaves at some point.

Said with the utmost respect and civility - you clearly think carefully about the way the game works.

[Ardem]

The problem with infantry is not the clumping, it is not the accuracy. What is the major problem is the lack of expect reaction to events, and even more so on blindly follow the final command when it is clear it should do something different.

Example under fire when moving, the slowness the AI react to coming under fire to either drop and return fire or move to the nearest cover at top speed (even faster then the current sprint) like a dash is the real issue. Or when entering a building and watching your lead soldiers in front get mowed down, curently the Tac AI either reacts by running back to only get shot in the back or continue to push forward and once again reaction is way to slow to what is happening. Instead at holding at the door and firing into the building from the outside.

The kneeling  looking for a target even though a soldier is being hammered by a platoon worth of firepower and only react which is way to long to hit the ground, not to curl up like a pinned guy but to just keep you head down until there is a lull.

Or the reaction of a pinned person staying curled up when at a short range there life is in danger completely and not react to either a dash away or at the very least return fire to a soldier that withing 15m and the pinned soldier knowing that curling up in a little ball will not protect them at that range.

It is the reaction of the TACT AI with infantry that needs the improvement, and better accuracy then can be increased, the reason accuracy may be dialed down is because of the slowness and the lack of reaction with infantry personnel to threats.

I would like to see blind firing included to a direction, firing at an enemy based on other sense like sound etc. I think huge penalties on accuracy and its not to hit but it is to help with suppression which is already in game. We have blind firing already included with players made choices, however it be nice to see the TAC AI reacting instead of waiting for a player to intervene.

More effort need to be made on the reaction to events to the TAC AI and the speed in which the troopers react to it. They seriously need to implement a dash, I hate seeing in a reaction sprint the Tact AI soldier almost moonwalking to where they need to go. I do not think Dash need to be a manual command but a TAC AI only feature, to get to closes cover and dive, not take 4 secs to kneel and lay down 

 

 

[John Kettler]

A most useful discussion. I get that the game follows the intersection of bullets with pixeltruppen to determine hits and their effect. What is unclear to me is how the game handles the situation of having the soldier, not in the pose we see, but in some state we aren't shown. By that I mean all those various body positions for which we currently have no/may never have animations. Does the game plot hit paths vs the men in postures we don't see?

Regards,

John Kettler

[Bulletpoint]

5 hours ago, John Kettler said:

What is unclear to me is how the game handles the situation of having the soldier, not in the pose we see, but in some state we aren't shown. By that I mean all those various body positions for which we currently have no/may never have animations. Does the game plot hit paths vs the men in postures we don't see?

It's very simple actually. Troops don't use cover well, but they are made tougher to compensate. So they get hit more often than in real life, but they "magically" survive many of the hits, based on a simple dice roll that takes place behind the scenes, with modifiers based on terrain etc.

I've even seen one soldier take a magazine of SMG fire straight through the chest and not only survive, but calmly return fire, killing several of the enemy (this was at 5 metre range)

[IICptMillerII]

19 hours ago, Ardem said:

The problem with infantry is not the clumping, it is not the accuracy. What is the major problem is the lack of expect reaction to events, and even more so on blindly follow the final command when it is clear it should do something different.

Example under fire when moving, the slowness the AI react to coming under fire to either drop and return fire or move to the nearest cover at top speed (even faster then the current sprint) like a dash is the real issue. Or when entering a building and watching your lead soldiers in front get mowed down, curently the Tac AI either reacts by running back to only get shot in the back or continue to push forward and once again reaction is way to slow to what is happening. Instead at holding at the door and firing into the building from the outside.

The kneeling  looking for a target even though a soldier is being hammered by a platoon worth of firepower and only react which is way to long to hit the ground, not to curl up like a pinned guy but to just keep you head down until there is a lull.

Or the reaction of a pinned person staying curled up when at a short range there life is in danger completely and not react to either a dash away or at the very least return fire to a soldier that withing 15m and the pinned soldier knowing that curling up in a little ball will not protect them at that range.

It is the reaction of the TACT AI with infantry that needs the improvement, and better accuracy then can be increased, the reason accuracy may be dialed down is because of the slowness and the lack of reaction with infantry personnel to threats.

I would like to see blind firing included to a direction, firing at an enemy based on other sense like sound etc. I think huge penalties on accuracy and its not to hit but it is to help with suppression which is already in game. We have blind firing already included with players made choices, however it be nice to see the TAC AI reacting instead of waiting for a player to intervene.

More effort need to be made on the reaction to events to the TAC AI and the speed in which the troopers react to it. They seriously need to implement a dash, I hate seeing in a reaction sprint the Tact AI soldier almost moonwalking to where they need to go. I do not think Dash need to be a manual command but a TAC AI only feature, to get to closes cover and dive, not take 4 secs to kneel and lay down 

 

 

At a certain point you have to accept that no AI in any game or sim is going to be perfect, or even a direct 1:1 representation of a human intelligence in the same situation. There are many sci-fi novels, movies and scientific theories about the development and impact a true artificial intelligence will have on the human race. Needless to say, whether you believe it will be a good thing for humanity or not, it will be a monumental technological breakthrough. Battlefront is a fantastic little company that is able to do things that companies with 10 times the size and budget could not. Creating a near human intelligent AI is not one of those things, and never will be. Well, unless of course the real reason @ChrisND is taking a break from streaming is to develop the first true AI Mk 1. 

The point is, the TacAI is not perfect and never will be. However, the TacAI is one of the most impressive things I've ever seen simulated in any game/sim, ever. Seriously, take a moment to think about it. Each soldier has its own little AI that is context specific to that soldier. This is why just because one soldier sees an enemy, it doesn't mean the rest of his squad magically does. I am continually amazed at the little things I see the TacAI do in game. When you see the AI do something you think is questionable, put yourself in the shoes of the soldier. See what he sees and is interacting around him, and it might just make sense after all. Also, truth IS stranger than fiction. for any silly thing you see the TacAI do, I am positive you can find a primary source account of a soldier doing something far more ridiculous in a real life battle. 

One of the main reasons why I love the TacAI so much is because it somehow manages to simulate these odd human-like moments on the battlefield. I remember reading the manual to CMSF a while back, and at the beginning they talk about the challenges of creating a war simulator. 

"Computers are essentially fancy calculators. They like
order, simplicity, and predictability. Chaos is not
something that a computer handles very easily or very
well. Depending on the circumstances, it might even
be impossible.
A tactical combat simulator, unfortunately, requires the
computer to simulate chaos - both natural and
manmade. Then, as the simulation is executed in
RealTime, the computer must calculate this chaotic
environment quickly and efficiently. As if this isn’t
demanding enough, the gamers using the simulator
require that the computer also devote a large amount
of its power to push around polygons to make the
simulation seem real."

There is only a bit more after that at the front of the CMSF manual, and I urge anyone interested to check it out. It helps to have a better understanding of the simulation as a whole. 

The next time you see the TacAI do something you think is silly, take a moment to think it through first from the perspective of the soldier, from the perspective of a real life account you may have heard, and then from the perspective of the simulation itself. Somewhere along the way it should make sense to you. And if for whatever reason you go through all of that and still can't figure it out, snap a quick video so the rest of us can enjoy the silliness as well, and just come to terms with the fact that its not now and never will be perfect. But what we have is pretty damn good, and far and away better than anything else that I've ever experienced. 

Edited June 22, 2016 by IICptMillerII

[A Canadian Cat]

17 minutes ago, IICptMillerII said:

The next time you see the TacAI do something you think is silly, take a moment to think it through first from the perspective of the soldier, from the perspective of a real life account you may have heard, and then from the perspective of the simulation itself. Somewhere along the way it should make sense to you.

+1 to that.  Just make sure you start with attitude that the pixel troops are not modelling super humans and not with the attitude "well *I'd* never do that". That makes a big difference.

[shift8]

1 hour ago, IICptMillerII said:

At a certain point you have to accept that no AI in any game or sim is going to be perfect, or even a direct 1:1 representation of a human intelligence in the same situation. There are many sci-fi novels, movies and scientific theories about the development and impact a true artificial intelligence will have on the human race. Needless to say, whether you believe it will be a good thing for humanity or not, it will be a monumental technological breakthrough. Battlefront is a fantastic little company that is able to do things that companies with 10 times the size and budget could not. Creating a near human intelligent AI is not one of those things, and never will be. Well, unless of course the real reason @ChrisND is taking a break from streaming is to develop the first true AI Mk 1. 

The point is, the TacAI is not perfect and never will be. However, the TacAI is one of the most impressive things I've ever seen simulated in any game/sim, ever. Seriously, take a moment to think about it. Each soldier has its own little AI that is context specific to that soldier. This is why just because one soldier sees an enemy, it doesn't mean the rest of his squad magically does. I am continually amazed at the little things I see the TacAI do in game. When you see the AI do something you think is questionable, put yourself in the shoes of the soldier. See what he sees and is interacting around him, and it might just make sense after all. Also, truth IS stranger than fiction. for any silly thing you see the TacAI do, I am positive you can find a primary source account of a soldier doing something far more ridiculous in a real life battle. 

One of the main reasons why I love the TacAI so much is because it somehow manages to simulate these odd human-like moments on the battlefield. I remember reading the manual to CMSF a while back, and at the beginning they talk about the challenges of creating a war simulator. 

"Computers are essentially fancy calculators. They like
order, simplicity, and predictability. Chaos is not
something that a computer handles very easily or very
well. Depending on the circumstances, it might even
be impossible.
A tactical combat simulator, unfortunately, requires the
computer to simulate chaos - both natural and
manmade. Then, as the simulation is executed in
RealTime, the computer must calculate this chaotic
environment quickly and efficiently. As if this isn’t
demanding enough, the gamers using the simulator
require that the computer also devote a large amount
of its power to push around polygons to make the
simulation seem real."

There is only a bit more after that at the front of the CMSF manual, and I urge anyone interested to check it out. It helps to have a better understanding of the simulation as a whole. 

The next time you see the TacAI do something you think is silly, take a moment to think it through first from the perspective of the soldier, from the perspective of a real life account you may have heard, and then from the perspective of the simulation itself. Somewhere along the way it should make sense to you. And if for whatever reason you go through all of that and still can't figure it out, snap a quick video so the rest of us can enjoy the silliness as well, and just come to terms with the fact that its not now and never will be perfect. But what we have is pretty damn good, and far and away better than anything else that I've ever experienced. 

Agree completely, all good points. 

 

Going to use this opportunity to make some points about game mechanic modeling:

For those of you who think there is something overall broken with the way infantry is modeled, you need to remember that when something is wrong in a 1:1 sense that attempting to go for a 1:1 simulation to fix it is not always better from a end result perspective. There are game companies that all the time attempt to model things too close to 1:1 and the end results as actually LESS realistic. The reason for this is that when you try to model something purely 1:1 you have to work from the bottom up you are banking on the idea that the sum of huge number of independent 1:1 mechanics will give your a final 1:1 result. However, this is predicated upon actually having a perfect model of all of these things. This is why in sim game development many mechanics are done from the TOP and move down to the cause mechanics. An Expert on something CAN know the general end result the game should have. 

 

Im going to give and example from  a totally different game in order to make my point here. It fits well with this debate because the infantry in RTS games are similarly complex. 

 

In the flight sim DCS world, there was an update back around 2012 IIRC that added what was called a "Advanced" Flight model to the missiles in game. No doubt, the new model was much more sophisticated than the older missile flight models, although it was still not perfect from a physics stand point. Missiles with the new AFM were suddenly subject to a litany of forces that had not been applied to the older model. HOWEVER, the new missiles were still using the OLD guidance system. The end result was missiles that were far less effective and far shorter ranged than they would have been in real life. Despite what might have seems like a overall upgrade, since the flight model was improved and the guidance static. However, it was a NET back step in the realism department. This was because once you start trying to to make something work 100% based on physics you have to get EVERYTHING right or you will get just as bad results as you might have gotten if you hadn't. The combination of what is still 1960's era guidance programming with a much closer to reality physics model meant that the guidance system was horribly inefficient at managing all the new variables it was confronted with, which resulted in BVR combat being so castrated it was essentially reduced to WVR. 

As a caveat to all that, there is a balance to be stuck between the real-world mechanics and the desired end result. I am not at all talking about "gameplay" here, but purely what realistic end result should occur. SOME mechanics can be modeled and SHOULD be modeled in highly linear 1:1 fashion because their nature is either simple enough or limited enough that there is not good reason not to. OTHER mechanics such a infantry are so incredibly complicated in terms of the number of geometry aspect that some aspects of them must be reduced to mathematics, and as Miller stated, the players ability to infer with his imagination what sorts of things the computer is actually simulating. 

-AGAIN, this doesn't apply to all mechanics. Nor does it apply equally to all aspects of the infantry, since some parts of them are in fact 1:1. This does not mean that there are not legitimate gripes to have with this game. If you look at some of my other posts you will see that I think there are some problems with the game (although few). 

 

So in summary, there is a nuanced balance when it comes to the exact procedures used to get a desired results with any game mechanic. 

 

[Michael Emrys]

13 hours ago, IICptMillerII said:

Also, truth IS stranger than fiction. for any silly thing you see the TacAI do, I am positive you can find a primary source account of a soldier doing something far more ridiculous in a real life battle.

Agree. In fact, our pixel soldiers behave more rationally most of the time than their real life counterparts. In case it has escaped anyones notice, please allow me to present to your attention that the world is largely populated by morons. Quite a few of them make it into armies. The bulk of them either got dumped into the rifle companies or the military academies. And then there were those who can't properly be described as morons, but who on their first exposure to combat were apt to behave like one long enough to get someone killed. It's just my opinion, but it seems to me that time and again, the AI does a fair job of representing all that. If anything, as I suggested earlier, a lot of the real-life stupidity gets edited out simply because a truer picture would not be tolerated by players. War is hell.

Michael

[Anthony P.]

12 hours ago, shift8 said:

In the flight sim DCS world, there was an update back around 2012 IIRC that added what was called a "Advanced" Flight model to the missiles in game. No doubt, the new model was much more sophisticated than the older missile flight models, although it was still not perfect from a physics stand point. Missiles with the new AFM were suddenly subject to a litany of forces that had not been applied to the older model. HOWEVER, the new missiles were still using the OLD guidance system. The end result was missiles that were far less effective and far shorter ranged than they would have been in real life. Despite what might have seems like a overall upgrade, since the flight model was improved and the guidance static. However, it was a NET back step in the realism department.

What, you're telling me that a small barrel loop or a gentle bank shouldn't be enough to spoof any sort of missile from either 6 or 12 o'clock (not even joking, that happens a lot)?

[A Canadian Cat]

16 hours ago, shift8 said:

<excellent explanation deleted>

So in summary, there is a nuanced balance when it comes to the exact procedures used to get a desired results with any game mechanic. 

Yes, too often some of us forget that.  Or probably never knew it since most people don't spend as much time thinking about this as you have.

 

4 hours ago, Michael Emrys said:

Agree. In fact, our pixel soldiers behave more rationally most of the time than their real life counterparts. In case it has escaped anyones notice, please allow me to present to your attention that the world is largely populated by morons. Quite a few of them make it into armies. The bulk of them either got dumped into the rifle companies or the military academies. And then there were those who can't properly be described as morons, but who on their first exposure to combat were apt to behave like one long enough to get someone killed. It's just my opinion, but it seems to me that time and again, the AI does a fair job of representing all that. If anything, as I suggested earlier, a lot of the real-life stupidity gets edited out simply because a truer picture would not be tolerated by players. War is hell.

Michael

+1

[cool breeze]

Hey, what going on here?!  Whats with all the reasonable discussion?  WHERE ARE ALL THE PEOPLE COMING OUT OF THE WOOD WORK TO JUMP ON THE CM SUX WAGON?!

[Michael Emrys]

1 hour ago, cool breeze said:

Hey, what going on here?!  Whats with all the reasonable discussion?  WHERE ARE ALL THE PEOPLE COMING OUT OF THE WOOD WORK TO JUMP ON THE CM SUX WAGON?!

Shhhhhhhh! In the words of good Ben Franklin (or if not him, whoever): "Let sleeping dogs lie."

Michael

[IICptMillerII]

2 hours ago, cool breeze said:

Hey, what going on here?!  Whats with all the reasonable discussion?  WHERE ARE ALL THE PEOPLE COMING OUT OF THE WOOD WORK TO JUMP ON THE CM SUX WAGON?!

I assure you it is only civil right now because the dissenting party has not had a chance to address the conversation. Brace for the counter attack!

 

9 hours ago, Anthony P. said:

What, you're telling me that a small barrel loop or a gentle bank shouldn't be enough to spoof any sort of missile from either 6 or 12 o'clock (not even joking, that happens a lot)?

Ralfidudes signature defensive maneuver! It can't be beaten! If only the Air Force had paid as much attention to Star Fox as Ralfidude did they would have discovered this ingenious move earlier! "Do a barrel roll!!!"

Back on topic @Michael Emrys brought up another aspect of the TacAI. It does indeed simulate less than clear thinking by soldiers under stress very well. Even if it is not intentional. I include these 'Sobel Moments' when I'm looking at things from the soldiers perspective. So if I go down to eye level and see what the soldier in question did and am still confused by it I usually am able to chalk it up to pure stupidity/incompetence. On the part of the soldier, not the TacAI.

Again the TacAI is not perfect but what has been described by myself and others is more than enough to explain/justify/make sense of 95% of what you see going on. 

[A Canadian Cat]

23 minutes ago, IICptMillerII said:

Again the TacAI is not perfect but what has been described by myself and others is more than enough to explain/justify/make sense of 95% of what you see going on. 

Agreed.  Lots of thoughts and ideas around to make the TacAI better but making each soldier a perfect uber soldier it not something I would like to see.

[Michael Emrys]

38 minutes ago, IanL said:

Agreed.  Lots of thoughts and ideas around to make the TacAI better but making each soldier a perfect uber soldier it not something I would like to see.

And I expect that BFC would not even contemplate such a move in the absence of a huge donation from some maniac.

Michael

[Ardem]

On 22/06/2016 at 7:18 AM, IICptMillerII said:

At a certain point you have to accept that no AI in any game or sim is going to be perfect, or even a direct 1:1 representation of a human intelligence in the same situation. There are many sci-fi novels, movies and scientific theories about the development and impact a true artificial intelligence will have on the human race. Needless to say, whether you believe it will be a good thing for humanity or not, it will be a monumental technological breakthrough. Battlefront is a fantastic little company that is able to do things that companies with 10 times the size and budget could not. Creating a near human intelligent AI is not one of those things, and never will be. Well, unless of course the real reason @ChrisND is taking a break from streaming is to develop the first true AI Mk 1. 

The point is, the TacAI is not perfect and never will be. However, the TacAI is one of the most impressive things I've ever seen simulated in any game/sim, ever. Seriously, take a moment to think about it. Each soldier has its own little AI that is context specific to that soldier. This is why just because one soldier sees an enemy, it doesn't mean the rest of his squad magically does. I am continually amazed at the little things I see the TacAI do in game. When you see the AI do something you think is questionable, put yourself in the shoes of the soldier. See what he sees and is interacting around him, and it might just make sense after all. Also, truth IS stranger than fiction. for any silly thing you see the TacAI do, I am positive you can find a primary source account of a soldier doing something far more ridiculous in a real life battle. 

One of the main reasons why I love the TacAI so much is because it somehow manages to simulate these odd human-like moments on the battlefield. I remember reading the manual to CMSF a while back, and at the beginning they talk about the challenges of creating a war simulator. 

"Computers are essentially fancy calculators. They like
order, simplicity, and predictability. Chaos is not
something that a computer handles very easily or very
well. Depending on the circumstances, it might even
be impossible.
A tactical combat simulator, unfortunately, requires the
computer to simulate chaos - both natural and
manmade. Then, as the simulation is executed in
RealTime, the computer must calculate this chaotic
environment quickly and efficiently. As if this isn’t
demanding enough, the gamers using the simulator
require that the computer also devote a large amount
of its power to push around polygons to make the
simulation seem real."

There is only a bit more after that at the front of the CMSF manual, and I urge anyone interested to check it out. It helps to have a better understanding of the simulation as a whole. 

The next time you see the TacAI do something you think is silly, take a moment to think it through first from the perspective of the soldier, from the perspective of a real life account you may have heard, and then from the perspective of the simulation itself. Somewhere along the way it should make sense to you. And if for whatever reason you go through all of that and still can't figure it out, snap a quick video so the rest of us can enjoy the silliness as well, and just come to terms with the fact that its not now and never will be perfect. But what we have is pretty damn good, and far and away better than anything else that I've ever experienced. 

I would disagree here the TAC AI is not that impressive. The tree branches used is pretty limited, but for the size company that Battlefront it is very good effort. I also disagree that it cannot be improved more so they it currently is.

More branches can be made to the AI decision making process, I am not having a go at what Battlefront has done, but i we were to take the Ai in say a big budget game an FPS, say for instance a new one like Rainbow Six Siege (yes you can compare FPS and RTS AI) imagine you looking at it from a 3D perspective above. Yes we can talking about CPU power etc, but there is obvious branches that the Battlefront team can work on in the reaction space.

Also do not mix up animations with TAC AI. Ignore the animation, that is only a representation of what has been called, it does not always correspond correctly to what it happening there is a only a limited number of animations that you can call on. I believe possibly the lack of repsonse is due to the amount of checks the computer runs on the unit, this may have been debraltily slowed to handle larger battles, else a check every microsecond would grind battle to a halt, however with faster CPUs they may want to reevaluate if they can increase the checks.

This is not slight at the great work they have done, however they know they can add branches and put more working into it if they wish. There is no need to white knight it the TAC AI could use some more work is all I am saying. I am sure the AI coder for Battlefront would agree, because knowing anyone that does AI coding they should never be satisfied with what is, but maybe he not given enough time or resource to do more then hat he/she has.

 

 

 

[IICptMillerII]

3 hours ago, Ardem said:

I would disagree here the TAC AI is not that impressive. The tree branches used is pretty limited, but for the size company that Battlefront it is very good effort. I also disagree that it cannot be improved more so they it currently is.

More branches can be made to the AI decision making process, I am not having a go at what Battlefront has done, but i we were to take the Ai in say a big budget game an FPS, say for instance a new one like Rainbow Six Siege (yes you can compare FPS and RTS AI) imagine you looking at it from a 3D perspective above. Yes we can talking about CPU power etc, but there is obvious branches that the Battlefront team can work on in the reaction space.

Also do not mix up animations with TAC AI. Ignore the animation, that is only a representation of what has been called, it does not always correspond correctly to what it happening there is a only a limited number of animations that you can call on. I believe possibly the lack of repsonse is due to the amount of checks the computer runs on the unit, this may have been debraltily slowed to handle larger battles, else a check every microsecond would grind battle to a halt, however with faster CPUs they may want to reevaluate if they can increase the checks.

This is not slight at the great work they have done, however they know they can add branches and put more working into it if they wish. There is no need to white knight it the TAC AI could use some more work is all I am saying. I am sure the AI coder for Battlefront would agree, because knowing anyone that does AI coding they should never be satisfied with what is, but maybe he not given enough time or resource to do more then hat he/she has.

 

 

 

As I've said many times now, yes the TacAI can be better. Yes the TacAI is not perfect. Yes I expect that it will continue to be developed and improved as we go on. 

The TacAI is not perfect and will be improved. 

There it is, literally spelled out for all to see. Hopefully we can move past this point now. 

The primary point here is that the TacAI is more than good enough for the game, and is not broken or flawed in a major way. It will still be improved. It has some issues. Everything does. My car has broken parts, and other parts that don't work correctly. It still gets me to where I need to go, reliably every time without a problem (knock on wood), and is in much better overall condition than a lot of other vehicles. Just because not everything is 100% does not mean its broken. There are literally hundreds of other examples just like this. The TacAI is the sum of its parts that work, and despite the flaws it still works very well. I honestly do not know how else to explain it. *Scrunches brow* 'Not broken. Work good.' *Thumps club*

As to the different types of AI and the difference between an FPS AI and a strategy AI, this is a rabbit hole I'm not going to bother going down. All I'll say is that the AI for each are extremely different. In Rainbow Six Siege its two teams of 5 if I'm not mistaken. So figure a max of 9 AI at most with only one human player. In CM, there can be battles that have HUNDREDS of soldiers  ON EACH SIDE. Each soldier has his own AI. That means there are as many AI's working as there are soldiers on the field. The difference should be clear, but if it is not I am sure someone else with a much better understanding of the workings of CPUs and super computers will chime in. Point is, comparing an apple to a rubix cube is silly in any situation, and thats what you've done by comparing strategy and FPS AI. 

[spawncaptain]

While I would certainly like to see improvements to the TacAI, way more than new content, IMO they should focus on adding some kind of mid-level AI that can do more than move a pre-designated group of units to a pre-designated place in a semi-coherent fashion.

[A Canadian Cat]

30 minutes ago, spawncaptain said:

While I would certainly like to see improvements to the TacAI, way more than new content,

I am sure you are not alone.  I am also pretty sure that there is a majority of people who want more content.  As @llCptMillerll has said it the Tac AI is actually pretty good.  Bottom line though BFC is constantly striving to improve everything about their games and keep older games up to date. So, I would imagine we will see improvements to the Tac AI as new releases come out.  It sure seems like the pace of Tac AI improvements will likely not match your expectations but it will very likely be non zero.

 

30 minutes ago, spawncaptain said:

IMO they should focus on adding some kind of mid-level AI that can do more than move a pre-designated group of units to a pre-designated place in a semi-coherent fashion.

First off that is outside the Tac AI.  I think the term BFC uses is the Strat AI for this level.  Us human players' units make use of the Tac AI just the same as the computer controlled units.  The part replaces / simulates us is an AI layer above the Tac AI (the Strat AI). I am not sure if Steve has answered this directly but the indications are that they see things differently.  The progress they have made has been to offer tools for the scenario designers to make those "move a pre-designated group of units to a pre-designated place" plans more nuanced.  Over time they have increased the number of groups, the number of orders and added conditional execution.  There have been asks for more unit groups more orders, plan branching plus a whole host of other small things.  It really seems that this is the way BFC have decided to go - create more flexible tools for scenario designers to create more complex and more realistic plans for the Strat AI.

[Duckman]

It will be a while until infantry is modelled on par with tanks in this and other games. Having said that, the entire game is a bit "compressed" meaning things happen at a higher than ultrarealistic speed for gaming purposes. The already cited "pushing too hard" effect also adds to this. 

What I can see perhaps helping infantry is if spotting becomes even more decentralized (final death of the Borg) and microterrain improved further. But I think infantry are already more survivable than in the bad old days of blatant Borg spotting when tanks would simply spot and mow them down from across the map (anyone else remember Close Combat?).

Other things that I sense may be issues, but which go beyond my knowledge of the game engine, are perhaps upping the protection of hard cover (stone buildings, log bunkers, even foxholes) a bit more and having a look at the effectiveness and speed of artillery. But those are hunches more than anything. Arty modelling in the game is already miles ahead of the competition and fosters good habits, unlike so many other games. But again foxholes, especially with overhead cover, is something that could perhaps be looked into in that context.

[Anthony P.]

Having seen countless 100mm AB hits exactly above foxholes in Black Sea having no effect on the occupants, I can't help but assume that overhead protection is already factored into the game, though not graphically.