Jump to content

AI never in attack.


semmes
 Share

Recommended Posts

12 hours ago, slysniper said:

As they made the AI programming more difficult to allow the designer more ability to make better AI plans, Less players have spent the time to learn to use the tools.

Now adding more tools and increasing the complexity is good for those who want to design better AI plans, but keep in mind, as you create more complexity. Those who master it will be fewer and fewer.

This is a valid concern.  However, no one is forcing players who want to make their one scenarios to use all the complex features.

The fact is that these days the best scenarios and campaigns (and mods) seem to be made by a very small group of very highly talented folks who are willing to commit hundreds of man-hours to produce superb work.  So let them have the tools to improve their amazing efforts.

Link to comment
Share on other sites

The two big things I've wanted to be able to do as a designer with the AI is:

- Triggers for reserve forces. Would allow the 'rescue/activation' of additional units once specific location is reached on the map by friendly or enemy forces.

- The ability for designers to run both side's AI plans in parallel as part of the testing process. Essentially so they push play and watch a battle unfold. This allows them to see where forces start spotting each other and if the AI plans are broadly working as intended before they send it out to another person(s) to test blindly. Playing one side at a time when you know everything about the AI opponent is a bit of a wasteful time sink IMO.

Link to comment
Share on other sites

This topic is why I'm often on the board talking down the notion of maxed out map sizes and maxed out scenario times. The AI functions admirably (1) if the scenario designer is sufficiently adept to make it perform logically, and (2) if the the scenario is within the game's 'sweet spot'. Event the best scenario designer is at sea if he can't anticipate where the opponent is likely to be on a  4km map after 3 hrs 30 minutes of gameplay. You may have noticed the occasional huge scenario show up with the tag 'Play (X) side only'. That means the designer's given up on formulating offensive AI plans. It may be difficult to max out map and force size in the game but its pretty easy to max out the scenario designer's ability to orchestrate events.

Link to comment
Share on other sites

55 minutes ago, MikeyD said:

This topic is why I'm often on the board talking down the notion of maxed out map sizes and maxed out scenario times. The AI functions admirably (1) if the scenario designer is sufficiently adept to make it perform logically, and (2) if the the scenario is within the game's 'sweet spot'. Event the best scenario designer is at sea if he can't anticipate where the opponent is likely to be on a  4km map after 3 hrs 30 minutes of gameplay. You may have noticed the occasional huge scenario show up with the tag 'Play (X) side only'. That means the designer's given up on formulating offensive AI plans. It may be difficult to max out map and force size in the game but its pretty easy to max out the scenario designer's ability to orchestrate events.

Being able to play equally good scenarios at these larger sizes would be great fun though...😎

Hopefully the designers will eventually get the tools  needed to make this happen...

Link to comment
Share on other sites

Some very simple things could make life so much easier for scenario designers, which would mean more people would create more and better scenarios:

  • An undo function.
     
  • Not having the "draw line function" appear when cycling through the various orientations of walls and hedges.
     
  • A better way to choose building windows and doors than to hold down Ctrl and manually click through every single possible configuration on every single side of every single floor of every single building.
     
  • Being able to raise and lower elevation in the 3D-view.
     

But don't take my word for it. I've only done two scenarios. They took me literally hundreds of hours though. Ask the beta testers and those who make the big campaigns that come with the games. I know many of them have suggested many of these things before. Listen to them.

Link to comment
Share on other sites

Lets make it more simple, Any map design feature would be nice to do in the 3d view, I hate the time wasted in imagining what it looks like from imput in 2d- then verifying in 3d, then having to go back to 2d to try and correct it. All main features should be functional to change and add right in 3d . That would make a huge difference in map making time.

Link to comment
Share on other sites

2 hours ago, slysniper said:

Lets make it more simple, Any map design feature would be nice to do in the 3d view, I hate the time wasted in imagining what it looks like from imput in 2d- then verifying in 3d, then having to go back to 2d to try and correct it. All main features should be functional to change and add right in 3d . That would make a huge difference in map making time.

Being able to make changes to the map in 3d would be good...but i think that i would also want to keep the 2d option...i think it might actually be easier to...paint the big picture...in 2d and then do some tweeking in 3d...

 

Link to comment
Share on other sites

10 hours ago, MikeyD said:

Event the best scenario designer is at sea if he can't anticipate where the opponent is likely to be on a  4km map after 3 hrs 30 minutes of gameplay. 

PERSISTANT MAP-DAMAGE might be a good cure for this decease...😁

If we could use large maps and play several shorter battles on the same map might make it somewhat easier on the AI...

Link to comment
Share on other sites

Make that two!  ;)

With persistent map damage, we could make some remarkable things.....Consider the possibilities of @LongLeftFlank's Ramadi, @Macisle's Kharkov, or @benpark's Berlin, used as a Master Map in the manner of a CM-1 'Static Campaign'.  Just imagine what those maps might look like after ten or more heavy battles!  :o

It would be (IMHO), fekkin awesome!  B)

Edited by Sgt.Squarehead
Link to comment
Share on other sites

1 hour ago, RepsolCBR said:

Event the best scenario designer is at sea if he can't anticipate where the opponent is likely to be on a  4km map after 3 hrs 30 minutes of gameplay. 

Isn't that where activation areas are useful?   If a unit enters an area it activates the AI response.

Link to comment
Share on other sites

On 9/30/2020 at 4:14 PM, Erwin said:

the AI plan (US) not only responded in a clever way but also initiated well-constructed attacks vs the human (German).  So, a very good AI attack can be created with current tools.  Since we don't experience this very often, presumably the effort takes work and a lot of playtesting.

I haven' seen i , first of all, but "responded", sure? If you attack on the right it is going  to counterattack on he right? Two forces?, a patrol activates the left counterattack and your attack the counterattack on the right? Yes, you can plan a slow methodical attack with enough troops  to cover both flanks all the time, or just add a 2nd, 3rd and 9h wave  to break through... if there is anything  to break through.

 If you need all that effort and playtesting, aren't you saying that the design is flawed? 

 



			
		
Link to comment
Share on other sites

On 10/1/2020 at 2:55 AM, MOS:96B2P said:

can generally be handled by the placement of the trigger. 

 

So you have to tell the HP where to move his tanks, he cannot fire and keep the distance? Maybe some lines  to help him  park too?

You forget the basic one, a counter but you have to tell he HP  exactly where to cross  so he can be counted.

So, you provide help?, really?, by avoiding the elephant in the room? All those new games were "planning for he past" then?

Yes, if you read the Editor you read the word "Stance". For example: 9 tanks are burning behind but number 10 is still moving  to the next order, all by himself, whatever "Stance" they had.

Link to comment
Share on other sites

1 hour ago, Sgt.Squarehead said:

Make that two!  ;)

With persistent map damage, we could make some remarkable things.....Consider the possibilities of @LongLeftFlank's Ramadi, @Macisle's Kharkov, or @benpark's Berlin, used as a Master Map in the manner of a CM-1 'Static Campaign'.  Just imagine what those maps might look like after ten or more heavy battles!  :o

It would be (IMHO), fekkin awesome!  B)

It most certainly would !! 😛😛😜🤩🥰

Link to comment
Share on other sites

On 10/1/2020 at 6:55 AM, Zveroboy1 said:

the possibility of a counter-attack.

I agree a 100%.

I am always going  to have a reserve and  to organize a secondary attack even if only as a"fair play" touch -and because it is amusing.

Pity most of the time counterattack only means units crisscrossing the battlefield. Trigger and timer are a too blunt instrument.

 



			
		
Link to comment
Share on other sites

58 minutes ago, Erwin said:

Isn't that where activation areas are useful?   If a unit enters an area it activates the AI response.

A trigger only tells the AI when to move...not where...

On large. huge maps having the AI to be in the right place...at the right time...with an addaquet force becomes rather tricky...

Link to comment
Share on other sites

TBH all of this just emphasises the need for a bit of psychology and a s**tload of testing on the scenario designer's part.....It's easy enough to criticise the inadequacies of the game, but figuring out ways around them can be rather more challenging and IMHO, it's much more fun (when it works).  ;)

 

Edited by Sgt.Squarehead
Link to comment
Share on other sites

19 hours ago, semmes said:

So you have to tell the HP where to move his tanks, he cannot fire and keep the distance? Maybe some lines  to help him  park too?

You forget the basic one, a counter but you have to tell he HP  exactly where to cross  so he can be counted.

I prefer to have the human player make his own decisions.  Decisions that are interesting because they have both benefits and consequences.  Also, in the editor, it's not possible or desirable to tell the human player where to move his tanks.  You could tell the player in the briefing but I think it is more fun to let the player make the command decisions.  If you read the 4.0 engine manual, ask questions and gain design experience you'll develop a better understanding of what's possible / desirable in the editor. 

The human player will have as many choices as the designer provides, through terrain, TOE etc.  In the above example the human player could attack the AI position with a deliberate assault (and probably trigger the counter attack), suppress the AI position from a distance while his maneuver element takes a different avenue of approach, smoke & by-pass, flank, etc.       

In one of the posts up above the example of where to place an enemy armor only trigger was given in response to the following general situation;  The desire for the AI to counter attack a main attack and not be triggered by a recon probe.  In my experience the thoughtful placement of the counter attack trigger will most of the time have the desired result.  However, the editor is versatile enough to usually provide more than one solution to a situation.  It just takes experimentation and testing. 

  

19 hours ago, semmes said:

So, you provide help?, really?, by avoiding the elephant in the room? All those new games were "planning for he past" then?

Yes, if you read the Editor you read the word "Stance". For example: 9 tanks are burning behind but number 10 is still moving  to the next order, all by himself, whatever "Stance" they had.

We provide help to each other on the forum by discussing (in a friendly helpful manner) how to use the tools provided.  We also often wish for new and improved tools. 

All the new games were a further, continued improvement on the current game engine (CM2) with additional TOE and environments.  CM3 (which BFC has mentioned) I guess would be the "planning for the future".

"Stance" orders are just one of the many factors that go into an AI plan.  There are also the movement type orders, building orders, vehicle orders, open hatches/closed hatches (3D only & one time only), equipment, experience, morale, leadership, fitness, AI area fire, AI facing, AI withdrawal etc.  These combine with order triggers, terrain triggers, timers, terrain, weather, ground conditions, etc. which are tested together then adjusted to obtain the desired results most of the time.  Again, the AI is not as good as the average human but can provide an interesting, fun battle.  If not careful the AI can provide a very difficult battle, especially on defense (IMO not usually fun).  More and better design tools would be welcome.       

In the above example it seems an AI plan was created that resulted in 9 out of 10 AI tanks burning.   This should have been noticed during scenario author test.  If it was not the desired result, but occurred most of the time, then some of the above factors maybe need to be adjusted?  Also the AI attack is only as good as the designer's plan.  Should this AI group attack after an artillery strike, does the AI need infantry, does the AI need a support / overwatch AI group, should the AI group have used area fire orders while approaching / passing a likely human player position? etc.   

Part of the fun of scenario design is playing / testing the plans.  Part of the danger of scenario design is adjusting the AI vs the Beta testers until the AI is very difficult for the player who plays the scenario for the first time (compared to beta testers who played it dozens of times during testing).  @Erwin has reminded me of this :D.             

 

On 9/30/2020 at 4:01 AM, semmes said:

I did mention the trigger, the problem is that it can be activated by a scout or a T-70, instead of by an IS-2 platoon.

An enemy armor trigger cannot be activated by an enemy infantry scout team.  Even an enemy jeep, truck (any soft skin vehicle) will not activate that trigger.  It will only activate when enemy light armor or armor touches the trigger.  It is fair & accurate to say an enemy armor only trigger will activate for both a single enemy T-70 or a platoon of IS-2s. 

After you learned your misunderstanding of how AI triggers worked (your above quote) did you ever go back and try again?  Hopefully you haven't gotten so frustrated that you've given up.  There is also a scenario design AAR book in PDF format that is located in the Combat Mission file location.  This PDF can be helpful when learning scenario design.  

And to demonstrate the AI in action.  An AI controlled tank broke through the human forward defenses and located the human Battalion Command Post.  

U43jMUNh.jpg

K1qCbqzh.jpg

Cool stuff.  :) 

Edited by MOS:96B2P
Link to comment
Share on other sites

8 minutes ago, Sgt.Squarehead said:

I still owe you some feedback on the flipside version of that one.....But things got a bit too real on that front for a while, as you know!  Is the version I have still current?

Current enough, I think.  I improved a few small things but nothing major if I recall.  

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

 Share

×
×
  • Create New...