AI

[jomni]

Is the AI still the same as the original? No improvements?

I found the original quite easy to win (either side, at high difficulty) because of the predictable AI. Logistics planning (reinforcement pool) is not as punishing as Cross of Iron or Bridge Too Far's points purchase method.

[Peter Fisla]

Based on my play so far (and I have played CC since the CC1 demo came out) I would say developers did a good job improving the AI. AI will now flank you more often (on Attack I always leave some reserve squad in the back )...Tanks movement has improved as well I think after CC1 for me WAR is the second best CC. AI will never be able to match human player however if you set your side to Veteran and AI to novice you will have a lot of fun, at least I am.

[jomni]

Ok, i might give it a shot.  AI issue aside, I think that the force pool system makes the game easier to win...(though making the campaign more realistic).

[Iron Stringbean]

Playing as Axis forces in the 25 day grand campaign w/ limited supply for Axis, my infantry BG's are attacking with about 15-18 rifle squads and a few MG squads, and are facing off against Allied infantry BG's with over 70 rifle/bar with all sorts of support teams.

Add in that all Allied BG's can reinforce, this is going to be a LONG campaign. [:)]

[mooxe]

I started a new GC. Morale was off and timer at 15mins. Morale off is a huge disadvantage to the AI, I just diudnt want to prolong death. Settings are German Elite vs American Recruit. After five battles, I disbanded the Americans four times and was just too short of time to disband him the 5th time. There are three more battles left this first turn.
 
The AI deployment is its 1st huge area for improvement. I can literally run fast my men from the start because I know they have not setup any good defensive line. The mortars may get me but it just slows you down. ATG are deployed in horrible spots, tanks the same. Attacking Hosington, 50% of the AI forces got caught floundering on a bridge. If they deployed in all the two and three story houses I would of had a game. Hosington on recruit only gives the Germans a single tank. Only one single team met me within 200m of the jump off point. One team alone will always die.
 
My total losses are higher then they should be since I was rushing around. A 30min game would of resulting in five disbands and much less casualties. So is the AI better than the previous versions? Not that I can tell, but it has been awhile since I played Battle of the Bulge. Destroying the AI is just as easy as it ever was.
 
I wish more emphasis was put on multiplayer here at Matrix.

[squadleader_id]

The AI is still a bumbling idiot when attacking:
- Massing tanks to take VLs then getting stuck in the traffic jam.
- Sending tanks to take VLs near buildings without infantry support.
- Sending infantry teams forward piece-meal.
- A lot of teams/units camp in their start-up positions not participating in the attack

On defense the AI is a bit better.  The AI initial deployment is poor of course...and he likes to advance and leave good defensive positions.
Most maps force you to engage tanks at close range (due to blocked LOS), so that can be a challenge for the (German) attacker.

Better than the original CC4...but we still need a VetMod for experienced players.

[Platoon_Michael]

Might have been nice to have done away with the square box deployment.
 
I too see tanks moving grouped together when moving towards a VL or even when attacking.
Mooxe had already mentioned the ambush problem and I see that as well,a recon team was within 20 or so  meters and wouldnt fire but a rifle team from a greater distance did.This was even after the AI saw my recon team and went for cover.
I dont know what to think about the mortors.
I lose my AT guns every time within a few minutes of the game starting.
But having a slot wasted on a team that cant kill anything sucks too,so whats one to do?
 
I dont like my bazooka teams as they can't hit squat unless its real close.
The AI is using support against me,and very effective to.
 
I would have liked to have seen the trees added to the zoomed out view of the map
(what is that the .ovm?I forget)
It can be a real pain with the new maps and all the woods featured when trying to deploy of fight.
especially now that vehicles dont get through them without trowing a track or breaking a wheel.
 
I havent seen the AI use much smoke yet.
 
Overall I like it,
plays much better than the original and not quite as hard as VETBoB is.

[Stwa]

mooxe, squadleader_id, platoon_Michael ...

So, would you guys say, the game is using basically the CC5 AI.

If so, I would add ...

Map size will have a lot to do with AI success (large would be better for the AI). You cant run all of your teams to all of the objectives, and be able to concentrate your force at each objective in 15 minutes.

squadleader_id's comments sound like the same AI from CC5 to me. Its better (perhaps) from CC4. But having the AI not commit his entire force to an attack and dedicating some of his troops to ambush, I view as a feature.

On many occasions, and usually on larger maps, the AI can make credible deployments, except for vehicles and guns. But then depending upon the map is hard to deploy guns.

The tank grouping thing that Platoon_Michael mentions, is a real drag, and has been prevelent since CC4. The only effective way to deal with this is to just not have that many vehicles on the map. I try to limit the game to 2 vehicles per side, and generally put them in the support platoon. On extremely large maps like in GJS, you might get away with having more, but sometimes I think if the tanks are in one of the other platoons (not support), they will try to drive together toward the objective.

I like the AI as an attacker, usually on larger maps, with no time limit, force moral turned on or off. Can you still kick his butt, of course, but on a big map to hold the objectives you have to thin out your teams, and the AI can catch one of your weaker platoons, etc...

[squadleader_id]

With no time limit, how long will a typical battle last (with the AI attacking)?
The AI is sluggish even on 15 min battles...and he doesn't commit all his forces in attacking.

BTW, in WAR I've noticed that sometimes when the AI is attacking...after the first wave of assault...he will camp...then about 5 min before the timer runs out...he will commit another big assault (VL rush?).  A pattern I've seen in some CC5 VetMods.

[mooxe]

I can't see a difference in the AI. If there is a difference its not enough to stop me stomping its head into the dirt repeatedly.

For example I give the AI 1/10 in CC5.

WaR gets a 2/10.

So you can see its improved, but still horrible and easily trounced.

Anyways we've bashed the AI since 1997 or 1998. Untill the game is rewritten then theres not much chance of having a fulfilling game vs the pc. I really wonder who has ever finished a complete GC in CC4/5/WaR vs the AI. The % of people who have finished would be a tiny tiny fraction of all those who started. If Matrix put more emphasis into online play around here we'd probably have a better time with it.

ORIGINAL: Stwa

mooxe, squadleader_id, platoon_Michael ...

So, would you guys say, the game is using basically the CC5 AI.

If so, I would add ...

Map size will have a lot to do with AI success (large would be better for the AI). You cant run all of your teams to all of the objectives, and be able to concentrate your force at each objective in 15 minutes.

squadleader_id's comments sound like the same AI from CC5 to me. Its better (perhaps) from CC4. But having the AI not commit his entire force to an attack and dedicating some of his troops to ambush, I view as a feature.

On many occasions, and usually on larger maps, the AI can make credible deployments, except for vehicles and guns. But then depending upon the map is hard to deploy guns.

The tank grouping thing that Platoon_Michael mentions, is a real drag, and has been prevelent since CC4. The only effective way to deal with this is to just not have that many vehicles on the map. I try to limit the game to 2 vehicles per side, and generally put them in the support platoon. On extremely large maps like in GJS, you might get away with having more, but sometimes I think if the tanks are in one of the other platoons (not support), they will try to drive together toward the objective.

I like the AI as an attacker, usually on larger maps, with no time limit, force moral turned on or off. Can you still kick his butt, of course, but on a big map to hold the objectives you have to thin out your teams, and the AI can catch one of your weaker platoons, etc...

[Tejszd]

Don't know how many have finished the GC but you know from posts on the other CC forums that a large number of people still play the AI. So while the AI in WAR could be better it is at least a definite improvement over CC5.

As for online play improvements, I do not know if that is the answer to make a hit CC game. How many play CCMT online? It was geared to online play but you do not see much activity/posts here or in the other CC forums about it....

[Stwa]

ORIGINAL: Tejszd

Don't know how many have finished the GC but you know from posts on the other CC forums that a large number of people still play the AI. So while the AI in WAR could be better it is at least a definite improvement over CC5.

As for online play improvements, I do not know if that is the answer to make a hit CC game. How many play CCMT online? It was geared to online play but you do not see much activity/posts here or in the other CC forums about it....

Interesting, but what specific areas of AI behavior have been improved by the developers. Just listening to the guys makes me think is still basically CC5.


squadleader_id,

Trying to respond to your question, everything is so map dependant, its hard to give a good answer that is short. But on small maps, having things set to no time limit doesn't really change things much. Its a visibility thing I think. I like the fact that the AI won't Rambo you, so on a large map he will attempt to be more stealthy. This does make the game take longer. I do play more relaxed myself in these circumstances, and this lets me enjoy some of the battles more, and believe it or not, sometimes the AI puts up a memorable performance.

mooxe,

I agree in one sense, if you could take all the work effort over the years for CC map making, and plow the same effort into either the AI or MP, then this series would be unstoppable. Too bad no producer has the kahones to finally put an end to puny AI behavior.

Its as if ongoing AI development does not make economic sense? Precisely the argument that is given for H2H play. So we are screwed I guess.

[mooxe]

The ONLY thing that "geared" CCMT to play online was the hacked in 4 by 4 multiplayer. What it had working against it was.... Old CC game not many interested in, modern battlefield, BHQ, all friendlies were the same colour on the map as you, oversized maps, weapons too leathal, no auto connect, only single battles, no GC, no Ops.... I mean come on, people just dont play CCMT at all for a variety of reasons. Matrix games could do some promotional online stuff at the very least instead of just putting a game out the door and focusing on whats next. Do we see ANYONE in BHQ at all except for the CC2 players? All that was ever done for CCMT was to mention it can be played by more than two people.
 
The trend the past few years for the popular games is to play them online. Is it the answer for CC? If it can be done right then yes. Otherwise the only answer it to have some breakthroughs in how the AI fights.
 
Anyways my point was that you cant see and appreciate some of the games bugs or mistakes when playing vs the AI. More emphasis in testing on multiplayer should be done to fine tune.
 
So what could Matrix Games or S3T do? Host some tournaments, include some "signs" in game on where to hook up for a game,

[Perturabo]

ORIGINAL: Tejszd

As for online play improvements, I do not know if that is the answer to make a hit CC game. How many play CCMT online? It was geared to online play but you do not see much activity/posts here or in the other CC forums about it....

Most of people, including me got put off by connection problems.

[jomni]

I think that, even with poor AI (1/10) CC will thrive if the campaign system is challenging. 
I really find the purchase point system more challenging than the force pool system.
With the point system, purchasing powerful units have it's (long term) repercussions (from not being able to muster cheap fodder to losing the whole sector if things don't go well). 
In the force pool system, you can just use up all your most powerful units immediately and pound away the enemy AI which cannot optimize unit choice and use powerful units properly.

My grognard heart is still with Close Combat 2 (Bridge too Far).  

[Stwa]

Im sorry guys,

But everytime a CC release comes out we all go into this AI discussion. And I just have to say, I am not that dissapointed with the AI. Some of its limitiations have been discussed here, but I see exceptions to those remarks during gameplay all the time. For instance, I have seen the AI attack with vigor, and perform flanking movements.

So, I would like to offer another screenshot, again from CC5 (my apologies, I will buy WaR soon). This screenshot shows a vigorous AI assault down the road in the center of the map. The AI has captured the VL on the wings (see minimap), and is trying to flank the German positions in the center. I have cut the Germans down to 10 squads to help with the odds. The AI just can take out that MG42 just north of the road, and it causes a lot of problems.

So it got me to thinking that maybe cover is just valued to heavily in favor of the defender, and this can be adjusted in the data. This would make for a better AI attacker, the whole dataset being revamped. Despite the heavy AI losses in this encounter the results (as will most CC games), still seem realistic to me.

[Sgt.Fury25]

The AI is pretty good imo.It seems agressive and cautious at the same time.On the attack it will throw a good punch,but also keep back some squads as backup.Anyways its a hell of alot better than the ai in ww2rtv!

[squadleader_id]

Stwa, if you must play as defender vs the AI...give CC5: Stalingrad Der Kessel (play as Germans) a try.
Have you played CC4 VetBoB?
There's also a Regular CC5 Vetmod based on TT's work.

The Vetmod boosted AI won't be brilliant...but as the defender...you'll be reeling and scrambling more often than when facing the stock AI  [:)]

[Stwa]

Nah, I tried the Vetmods many years ago. They are very gamey and appeal mainly to gamers who mostly want the 15 minute fix. After all its a game and if it takes more than 15 minutes that would be bad, and if everyone stops shooting, I'll bet plenty of players hit escape after about 30 seconds of silence.
 
I like the AI, its not perfect, and I have had many great games against the AI. I just recently rolled everything in from backup so I am just getting re-aqainted with the game. Big maps are fun for a lot of reasons, but those old CC4 maps, are still pretty cool. Remember back then, the old timers were complaining about the size of the CC5 maps being to big and all, and it would take forever to engage.
 
I play very differently against the AI, and I do like it when the AI attacks because there is less for me to do, and I can watch the action. In some of this conversation the AI is getting a bad rap, when maybe some factors can be tweeked. (After all thats what Vetmods are, just tweeks, oui?)
 
You gotta figure, when you play a 15 minute game against the AI, a lot of decisions you make are with the full knowledge that the game will end in 15 minutes. Like taking all 15 squads to a single objective, and then running all of them to the next objective, etc.. And I'll bet after all the shooting has died down and their is 30 seconds left, people scramble to get a few more objectives, while the AI is down, oui?) And ammo, who cares just get to that last objective before time runs out.
 
Thats partly why I like the no time limit game. I do have house rules though. If the AI is attacking and wants a truce then I give it. Otherwise I will try to take one more objective.
 
And just looking at the replies above, remember meeting engagements, that's where you would try to run as quickly as possible (damn the mortars), to the objectives, so you could ..... what .... camp, or take cover. After all, modern warfare gives a decided advantage to a concealed defender. Hard to critisize the AI who wants to do the same thing. Its that concealed defender (in cover), that perhaps can be tweaked to provide a better chance of success during attack. Perhaps a few tweaks here might help.
 
[:)]

[Perturabo]

ORIGINAL: Stwa

So it got me to thinking that maybe cover is just valued to heavily in favor of the defender, and this can be adjusted in the data. This would make for a better AI attacker, the whole dataset being revamped. Despite the heavy AI losses in this encounter the results (as will most CC games), still seem realistic to me.

AFAIK attacking without a big advantage in numbers isn't something that is a good idea. The problem is that despite CC4-5 pretence to be a strategic game, you won't experience things like attacking one map with multiple units (on strategic map) and battles on tactical map will almost always be against units equal in size.

Also, attacking AI will always be at disadvantage when compared to player because defending units in cover usually fire long before they are spotted by enemy soldiers.
The player can cheat by looking for tracers and muzzle flashes and reacting accordingly.
AI can't do it because it doesn't see tracers and muzzle flashes.
Which is why they should be invisible until they are spotted by soldiers and then marked on map so that the AI and player in would know their position and could target them.