[UKCinC]

HI guys,
 This may have already been covered by another post but cant find it. I was jest woundering can you change the AI on CCV and CCTLD, I am new'ish to modding have played with teams and starting with maps.
As i said i am looking for tips or hints on modding AI ,if it can be dun.

Thanks all

[vonB]

No you can't, unless you have access to the Source Code and C++ Development Environment, or you are outrageously canny with dissassembly, but no-one in their right mind would try that route except for the most humble of tweaks, and certainly not to the AI!

All you can do is influence how the AI performs, and there are not many ways to do that.  The biggest influence that I know of is related to Maps.  The smaller they are and the less VL's they have, the better the AI is going to be for a very simple reason; the AI is VL focused.  The AI is constantly calculating and recalculating it's options, and the bigger the Map and the more VL's there are, the less processing it is able to do per unit time, and this is real time remember!

I am not familiar enough with the Data Files to be able to give you any sensible suggestions as to what you can do with those, so maybe someone skilled in that area can contribute for you.

[UKCinC]

Thanks,
I had a feeling it was going to be like that. I have another question but I will put up another post fo that one. Thanks again.

[CSO_Linebacker]

What we've actually been learning is that the more VLs there are on a map, the more agressive the AI tends to perform, at least in the most current of versions...scoring of the VLs is also another factor.  The exe allows for up to 16 VLs per map, and through testing we've seen a marked improvement in the performance of the AI when all 16...or something close to it...are used.  During testing of this for TLD, the movement and aggression of the AI opponent was noticeable almost immediately from the moment you pushed 'Begin'.

Smaller map size seemed to amplify the effect.

[vonB]

Interesting.  Not my experience, but then I don't play TLD a great deal.  The only difference I found with TLD was the Timer.  If it was 30 minutes, the Battles tended to become quite tedious (sometimes very!).  The AI sent out an assault group (usually a Command Team and 2 others), and when they were trashed, little aggression was subsequently shown.

Setting the Timer to 15 minutes changed that dramatically.

Did you do much in depth experimentation with few VL's?  1, 2, or 3 for example?  Most of my experientation was done with CCIII/CoI.  Taking the same Map and playing around with VL's (number and placement), there was a decided difference with few VL's, especially if they were not close to the AI's baseline.

On the whole, it is CCII that gives the best AI attacking.  III, IV, and V (and re-releases) seem fairly poor on the whole to me.

Would be good to get more feedback on other players experiences.  It can only come from mucking around with lots of Custom battles.  I don't know if many do that, and just play whatever Battles (whether single battle or part of an Operation or Campaign) come with the Release or Mod.

[CSO_Linebacker]

I think it was Schreck who did some pretty extensive testing playing around with VL values, and had some interesting conclusions.  VLs could have values other than 50,100 and 200...as long as exit VLs were always 50.  Didn't effect the scoring, but appeared to effect how the AI acted.  With fewer VLs, we were finding that the AI didn't do much.  Steve explained that with more VLs, the VLs almost act like a roadmap for the AI...continuously giving it something to strive for...therefore it will try and keep it's troops moving towards VLs. Can't recall if it was determined if map coding effected troops as much as it did vehicles.  Vehicle pathing is heavily reliant on map coding.

[squadleader_id]

Don't forget this simple "AI enhancing" registry editing trick for CC5:

[Pzt_Kanov]

[squadleader_id]

[UKCinC]

cool, thanks, i'am working on a mod at the mo for cc5. hoping that this combined with other factors i'am working on should make it a pretty tuff one, thanks for all you help.

[Spinlock]

with only 15 teams and no reserves to bring up, you can't really blame the A.I. for being conservative.
After the loss of a few teams the A.I. is at risk of losing the territory it already holds.
What's needed most of all is reserve teams if the A.I. tactics are going to improve.

[mooxe]

I have always thought making maps smaller, and having less VLs albeit more centralized will give you a better experience.

To actually improve how the AI makes decisions I have no idea. We can manipulate the AI by doing above suggestion. The trend seems to be the opposite, larger maps and more VLs apread out all over.

Crippling the human opponent and beefing up the AIs troops is another way, but the wrong direction for a permanent fix.

Morale should be on to help the AI survive longer. Smaller maps could also mean shorter timers, like 10min games. Imagine four 10min battles per day with morale on. That mean you have up to 40mins of battle to simulate one 24 hour period of fighting, which is reasonable in a grand campaign. Many people play CC5 at 15min intervals with only two turns per day.

Shorter time, with morale on, smaller maps with less VLs (but centralized) will help the AI survive longer and from experience it will make better decisions.

[Spinlock]

The small maps are always the most fun, when both sides are forced to bring their troops through choke point is when the action gets good.
Map size is the biggest reason IMO that RTB was crap.

[Pzt_Kanov]

Something happened after CC2, the next CC's where very easy compared to the first 2 and I don't think it was just because they had smaller maps, it was something else changed in the code because if you play cc2 the AI wants to get you, like really kill you and defends the maps very stubbornly till the last man (sometimes they agree on a truce if they can't continue) also they don't do the crawl of death that often when fired upon.

Lots of times I have been surprised by how persistent the AI was in CC2, be it by storming the rail bridge at Oosterbek as allies or delaying you at Son Bridge and Nijmegen as germans.

[platoon_michael]

[Pzt_Kanov]

It's not a case of being a new player back then, because I give the CC2 campaign a go every 2 or so months and is still hard with either side. On subsequent CC I can play it relaxed knowing I will win even though I've spend months not playing them. CC3 is special because the hard part is not winning which is easy, but to preserve your men while doing so so it's another type of difficulty.

[squadleader_id]

I agree with Kanov about the clever CC2 AI...tough and nasty!
10 min battles as Mooxe suggested sounds good too...using CCreq?

Also...try handing the CC5 AI 'all the cards'...no brilliant flanking or pincher manouvers...but the AI might surprise you from time to time.

[vonB]

Short and long of it is, nobody has done a serious in depth examination of the comparative performances, along with controls, to be able to give any definitive judgement.  Most of it is our own personal 'take', and no surprise that we are going to get different opnions.

But, we can see some common trends.  There seems to be some consensus on CCII being a more challanging AI game.  I agree with that.  As regards Map size and VL's, opnion seems mixed.  For myself, I did spend some time experimenting with CCIII, creating custom battles using the same Map(s), and changing VL's with the intention of observing what (if any) changes in behaviour there were.  It was a long time ago, but there is no question in my mind that VL placement and numbers can make a big difference.

There is no significant difference in the AI in any of the CC releases as far as I know, so the differences in performance must be due to other factors.  But then the AI is a dynamic process, so will naturally respond differently in different circumstances.  What we are examining are the influences that the 'environments' (for want of a better term) have.  The 'tactical' AI intelligence is not very sophisticated, and probably far more limited than most realise.  It is the psychological model that provides the gutsy play experience, but that does not make the AI a good 'tactician'.  No matter which version you play, if you use good tactics and play to the AI weaknesses, then you will normally win.

To stack the odds against you does not answer the desire for a more competent AI opponent.  VetBoB is just an abbatoir.  Good tactics and intelligent play is often impossible due to the shattered state of your Units.  You shouldn't knock this out of hand.  War can be just like that, but when it is followed relentlessly battle after battle, the novelty soon wears off, and where is the compelling tactical cut and thrust?

platoon_michael's comments are challanging, but neither he nor anyone else has any substantive evidence to demonstrate one thing or the other.  In fact, his comments invite such an approach, as it is the only way to get any definition into this.  At the moment it is essentially opnion based.  I possibly have more under the hood knowledge of CC than many, having worked inside the system, but I would by no means call myself an expert.  To get Code level analysis, you would have to get that from those who both have access to it, and understand enough about it to be able to give such commentary, and I have never seen anything significant forthcoming in that respect.  

However, when discussion these things with Steve 3 or 4 years ago, it was clear that the tactical AI is primitive.  It is the psych model that gives CC it's edge, but what most players seem to want (including myself) is a single player experience which is more challanging, not via the route of overwhelming odds, and stacking the table against you, but by an intelligent tactician who can make us work for our victories, or even hand us our arses on a plate.  That I cannot see happening, because the development costs would be prohibitive, so we look for whatever ways we can to find things that make a difference.  Long live the cause!

[platoon_michael]

If somebody's interested in putting the data from 1 CCII team up I would be willing to do the same from WAR.
We just need a manner in which everyone can see and understand the #'s
I.E. 4=Very High Moral
Are the chance tables the same?
etc etc
You can e-mail me if you want to,just be sure to put something in the subject matter so I know it's CC related as odds are your 1st e-mail will probably go to my junk folder.
or we can use PM.