About quick battles

[Whiterider]

Something that needs to be improved in CM is the troops position in quick battles. It looks they are located randomly along the map, without sense, frequently the green areas and its surroundings are totally abandoned.

It´s really frustrating to spend a lot of time playing and discover at the end your enemy was sleeping far away :mad:

Could it be fixed?

[Whiterider]

As I remember it didn´t happen in CMX1 did it?

[A Canadian Cat]

Sounds frustrating. Here is what you need to be aware of: the AI does nothing but fight in place on its own. A quick battle map needs initial setup locations for the AI controlled troops and it needs an AI plan to move the AI controlled units from their setup areas to objectives or attack you.

The QB maps that came with the game are supposed to have those ready to go. Additional maps may or may not have the necessary plans - the author will hopefully say one way or another.

Since you are not specific about map, settings etc no one here can tell you more than that intro information.

[womble]

I believe there have been some issues tentatively identified with the QB AI possibly assigning its troops to groups that don't have any orders in the plan (or something like that), which exacerbates the problem.

[Doug Williams]

What ian said.

In order to have a decent game against the AI in a QB, the map must been made with QBs in mind, meaning that it has (more than one) AI setup locations and AI plans.

If you try to use a scenario map or a H2H only map for a QB against the AI, the AIs troops will just sit there in some random spot.

Scenario maps can, however, make good QB maps for H2H play.

[Whiterider]

I always play quick battle maps in quick battles. It´s more an issue of initial units position than something related to AI behaivour. Everybody can see it: just organize a few quick battle (meeting/attack/etc) and surrender in turn 2 then see the computer units location...

I supposed green areas were essential to win the game, but enemy troops don´t think so. Guys, this is an important issue even more than endless discussions about specific rules of weapons or graphics. Some people (ie most of my friends) love to play quick battles and can´t understand the reason why it happens. I only hope it can be fixed sooner or later.

Anyway thanks for your answers.

[womble]

Guys, this is an important issue even more than endless discussions about specific rules of weapons or graphics. Some people (ie most of my friends) love to play quick battles and can´t understand the reason why it happens. I only hope it can be fixed sooner or later.

Anyway thanks for your answers.

As has been said, it will depend on the plan. It's certainly possible for the AI to provide a QB opponent. Equally, there are times it fails. CMx2 games are pretty complicated things, and for "it" to be fixed, you'll need to be specific about what "it" is (down to the map, which side was attacking and which defending, possibly more details), since the reasons for any given failure can be varied. There have been various problems with QB deployments that have affected HvH as well as HvAI games, and BFC has tried to fix them. Often it's just a setting in the map data that needs to change. Sometimes a map's AI plan might have a FUBAR in it. If you actually want it fixed, give us a couple of examples, and those with the willing, knowledge and expertise can have a look and say if it's systematic or particular. Or if it's even reproducible.

[A Canadian Cat]

Guys, this is an important issue even more than endless discussions about specific rules of weapons or graphics.

Here is the thing: I almost never play against the AI - so close I could have written *never play against the AI* and I would have only been slightly wrong (although to be fair I am going through a phase where I am playing against the AI now but not Quick battles). There are lots of people on this forum that are like me an only play against humans (note I did not say *all* so please don't jump on me:-). So, lot of us have never seen what you are experiencing. Therefore we are not talking about it

We have been told that most of the customer base plays against the AI so what you are experiencing is entirely relevant just no something the majority of posters are experiencing.

Some people (ie most of my friends) love to play quick battles and can´t understand the reason why it happens. I only hope it can be fixed sooner or later.

Which is probably an indication that this is not an always happens problem.

Given that lots of official testers read this forum, heck the guy that is responsible for the QB maps reads and contributes here, Here is my suggestion: Fire up a QB take note of everything you set and what map is used. Save a turn when it first starts. Make sure you hit the problem you are talking about save another turn showing that. Grab an incriminating screen shot and post that that info to this thread. There is a good chance someone will play around and figure out what is up.

Probably will not be me though :cool:

[Whiterider]

I´ll try to explain the trouble in my limited english

In any quick battle (defense, attack, etc.) you play against the computer, enemy troops are usually located far from the green areas (sometimes all of them in one corner). You can check it by yourself: press "cease fire" in turn 2 and you´ll see.

I can´t believe I´m the first guy who noticed this strange issue :confused: (am I having hallucinations?:eek:). Please, let me know I don´t need to visit a psychiatrist. Can somebody who speaks english help me?

[A Canadian Cat]

I speak English - sorry you did not find my advice on how to get help to be well, helpful. Good luck.

[womble]

I´ll try to explain the trouble in my limited english

In any quick battle (defense, attack, etc.) you play against the computer, enemy troops are usually located far from the green areas (sometimes all of them in one corner). You can check it by yourself: press "cease fire" in turn 2 and you´ll see.

I can´t believe I´m the first guy who noticed this strange issue :confused: (am I having hallucinations?:eek:). Please, let me know I don´t need to visit a psychiatrist. Can somebody who speaks english help me?

Your problem is that you're looking at it in T2, and not considering what the AI plan has in store. I just started a game on QB-127, and you're right, almost none of the troops set up on the VL. A bunch set up just short of it, and the majority in the back corner. There were some outposts up front. But looking at the AI plan, I see that the defenders (it's an "Axis Assault" map, so the AI assigns the Allied plans to whichever side is actually defending) have orders to occupy the defined VL. So, by the time I've crossed the half dozen hedge lines, the Fusiliers I let the AI pick will be in position to contest my passage. The Flak guns I thought would be interesting to face with Airborne (d'Oh!) are another matter, of course, and the selection of which units to piquet with is a bit squiffy, but there's a reason they're set up off the VL, and that's because the plan has them moving onto it later. Your conclusion has been flawed by your early ceasefire, I'd suggest. It seems that it's probably a good idea for the defenders not to be on the VL from the start, since players do so love their preplanned bombardments, which would do almost no harm to this plan.

[Whiterider]

Ok, now I understand. The computer has its own plans to reach the victory getting the green area even if they´re initially located far away (a good explanation could be a surprise attack). Yep, some players like me love to bombard green areas and that gives a great advantage to attackers.

Thanks Womble, that makes me feel better

[Collingwood]

The mentioned QB behaviour is not news to me. I have often played QBs always using QB maps, and experienced absolutely minimal enemy forces throughout the battle, providing little or no defence of objectives. At the end of the battle I end up scouring the map, and there they are, on the edge of the map, or somewhere quite irrational.

No, I can't provide exact details either since I don't keep logs or journals of all my games, I just play and enjoy. But I don't think this is a really rare thing. IIRC, ND experienced something like this in one of his streams. Here we all are watching a QB and waiting for a decent stoush, which was not forthcoming. Sure that is a beta demonstration - but he did refer to the AI placement problem in QBs so I think BF are in some way aware there is an issue.

That said, I don't think QBs are meant to be the main way of playing, and I have no problems with how the AI follows plans in properly designed scenarios. Basically I play QBs rarely now unless I set it to "meeting engagement". These I find to be quite fun.

[Whiterider]

The mentioned QB behaviour is not news to me. I have often played QBs always using QB maps, and experienced absolutely minimal enemy forces throughout the battle, providing little or no defence of objectives. At the end of the battle I end up scouring the map, and there they are, on the edge of the map, or somewhere quite irrational.

Exactly! That happened to me several times. I don´t know what kind of problem is but it is real. I´ll follow your advice about meetings in QB.

[womble]

The mentioned QB behaviour is not news to me. I have often played QBs always using QB maps, and experienced absolutely minimal enemy forces throughout the battle, providing little or no defence of objectives. At the end of the battle I end up scouring the map, and there they are, on the edge of the map, or somewhere quite irrational.

No, I can't provide exact details either since I don't keep logs or journals of all my games, I just play and enjoy. But I don't think this is a really rare thing. IIRC, ND experienced something like this in one of his streams. Here we all are watching a QB and waiting for a decent stoush, which was not forthcoming. Sure that is a beta demonstration - but he did refer to the AI placement problem in QBs so I think BF are in some way aware there is an issue.

That said, I don't think QBs are meant to be the main way of playing, and I have no problems with how the AI follows plans in properly designed scenarios. Basically I play QBs rarely now unless I set it to "meeting engagement". These I find to be quite fun.

Part of the problem with AI plans in the QB was illustrated to me in the example I tried. I used a "Suggestion" for the AI side, and that included some Flak pieces (immobile). Of course, unless those are sited to be of use in keeping the attacker off the VL, especially on a Bocage map, they aren't going to be following any "move to the VL" orders later on. Similarly with all the arty pieces it picked (IGs and ATGs) leaving just the lone company of lightly-armed fusiliers to follow the plan to move onto the VL, and some mortars firing a "just outside my setup zone" preplanned barrage to give them time to do it.

It's a problem that could possibly be addressed if there was some way of coordinating force picks with plans and force types to groups of orders, but at the moment, it's very easy for the AI to totally bork its force pick at all kinds of level and choose units that either simply can't or can't effectively follow the orders of the plan, and it's not possible, AFAIK, to get the AI to assign mobile elements to groups with movement orders, and "offset" setup areas, while assigning static (or practically static) elements to setup areas associated with orders that have an intent of direct action in place. Even the most generic of order sequences in a plan can made useless by the AI force picker.

I have a suggestion for those of you with a circle of friends who like to play QBs: play each other HvH! But if you can't be doing with that, how about you set up QB opfors for each other?

[Collingwood]

Part of the problem with AI plans in the QB was illustrated to me in the example I tried. I used a "Suggestion" for the AI side, and that included some Flak pieces (immobile).

I confess, I often choose the AI forces for it.

I have a suggestion for those of you with a circle of friends who like to play QBs: play each other HvH! But if you can't be doing with that, how about you set up QB opfors for each other?

Great suggestion, which reminds me - There's something I've been working on for a while now: I set up a QB and choose all the forces. I then duplicate this, and vary the AI forces slightly. After I've got 3 or 4 QBs I put them into a folder, and have a script which randomly plucks one, and copies it with a new name into my scenarios folder/directory.

End result: I still get a sensible AI force to play against - but I'm never sure exactly what it will be. No matter which scenario was chosen by the script it copies it to the same generic name so I don't tip myself off, so to speak.

I have also been working on a similar idea to provide greater surprise value for scenarios.

[Freyberg]

Some of the time - reasonably often but not most of the time - I've had the AI set up forces in a completely bizarre way (including having many or most of the opposing forces in some weird place at the back of the map, which seems to happen more often when the map is too small for the number of troops), but more often I've found the AI plans lead to an average to decent battle. The more recent maps (eg. CMMG) have some quite interesting AI plans and sometimes it can be loads of fun.

When they aren't that challenging, at least QBs are quick.

QB maps with the new AI triggers should made quite a difference and should make for a pretty good scrap.

[MarkEzra]

I am the person responsible for making QB's for CMBN and it's modules (not CMFI).

Let me explain the current AI Plan/Orders method and provide some hope for CMRT QB's.

Current Defender AI Plans/Order Have a setup location on The Objective locations without any further movement orders. A second Group (or multiple groups) Have a placement order some place other than the Objectives location plus a Timed movement order to the Victory locations. The Time set for the Movement order is based on map size and the time I think the average Human Attacker will need to reach his Objectives. Since QB games can be set for any Weather, any terrain condition, and any hour of the day, AND the length of the game can be varied from 30 min to 2 hr, as well, the movement order may or may not actually happen. That's the current situation. Here's the future:

CMRT QB Maps will be Trigger sensitive. When the Human Attacker nears or reaches the Objectives that will trigger the AI Defender to respond.

[Collingwood]

Thanks Mark. Great news about triggers in CMRT QB maps, my love for them grows stronger

[Macisle]

I was actually running some tests on this the other day.

I do think there is an issue with the quality of the AI initial setup that is independent of the QB map designer's AI plan. The code should be looked at for ways to put ATGs in better positions and to make more use of buildings. The AI loves to mass its infantry around walls!

One really odd thing I found: when I repainted the setup area to try to force the units into more buildings, I had wacky things happen like nearly a full company, plus 2 ATGs, PLUS a tank on like...3 action spots. It happened more than once and the AI still didn't like to use buildings (there were a lot of painted buildings to choose from on the map, though).

If folks want to test, just grab a stock QB map, make sure it has AI plans, and choose the same units each time for the AI, hitting ceasefire or surrender (I used ceasefire) after running a few turns. Do NOT hit ceasefire during setup! If you do that, the AI will not be setup at all (ie, blobbed up in an odd corner of the map).

If you want to increase your chances of a good QB experience, choose a small/tiny map and give the AI a company at least, plus a number of ATGs and tanks. That can producing a challenging setup result. However, the AI may put all the ATGs in one spot!

The issue of the AI staying at the edge of the map until the player takes the VLs may be due to the AI only choosing one setup group out of multiple groups in the AI plan. So, for example, if the QB Map designer has assigned three groups in the AI plan with one group on the VL, one at the friendly map edge, and one forward of the VL, the AI may put all of its force in the group at the map edge, leaving the other two groups off.

So, that might be looked at as well, to see if code needs to be tweaked to make sure that the AI uses all the groups assigned to it, or at least doesn't only use a map-edge group.

[womble]

The issue of the AI staying at the edge of the map until the player takes the VLs may be due to the AI only choosing one setup group out of multiple groups in the AI plan. So, for example, if the QB Map designer has assigned three groups in the AI plan with one group on the VL, one at the friendly map edge, and one forward of the VL, the AI may put all of its force in the group at the map edge, leaving the other two groups off.

So, that might be looked at as well, to see if code needs to be tweaked to make sure that the AI uses all the groups assigned to it, or at least doesn't only use a map-edge group.

That's an important factor I think. I wonder if, when you're helping the AI pick its forces, whether having more formations will help all the groups get some troops assigned to them. Would picking three platoons, say, from three different Battalions make it more likely that a platoon would be assigned to a given group?

[womble]

MarkEzra, as the person in the world with the most experience setting up AI plans for QB maps, would it be useful to have a small selection of "key words" that could be associated with a given group in a plan, and which would be recognised by the QB "master" AI for setup purposes? For example, I'm thinking you could have "Holding force", "Counterattack", "Picket" available (to start with), and the AI would assign static assets like ATG or flak to a group that has "Holding force" as its keyword, while if there's a group with the "counterattack" keyword, it would be likely to put some or all of the armour in it. Such usage would be optional (default to "general" or somesuch) of course, and you'd not have to have a group for each keyword, or even have enough groups to accommodate all the keywords. It would need each unit assigning some sort of weighting towards the different keywords and a routine for the AI to use those keywords in its group assignments.

[Freyberg]

I assumed the AI chose the most appropriate location for units depending on the type of orders in the AI plan for that zone - like 700m ambush would be selected for AT guns, that sort of thing...

[Whiterider]

I am the person responsible for making QB's for CMBN and it's modules (not CMFI).

Here's the future:

CMRT QB Maps will be Trigger sensitive. When the Human Attacker nears or reaches the Objectives that will trigger the AI Defender to respond.

Very good news . Also, I´d love to see some day this improvement in other CMX2 games.

Thanks

[womble]

I assumed the AI chose the most appropriate location for units depending on the type of orders in the AI plan for that zone - like 700m ambush would be selected for AT guns, that sort of thing...

That may be true. But if the units are not assigned to Groups which have "relevant" orders and setup areas all the clever positioning in the world isn't going to help. It looks to me like formations are assigned alternately to available groups. The force for the defenders in my example referred to previously was composed of a Fusilier Company, some support elements from a PzGr Coy and a Flak Platoon (in, IIRC, that order on the "active units" screen) and, judging from their setup locations, the Fusiliers and the Flak are in Group 1, while the PzGr IG platoon are in Group 2. Also, the setup areas for the groups are quite restrictive, and I doubt there is a good place to set up a static Flak gun in the Group 1 setup area.

Distributing formations this way doesn't help the AI much, since it will often (IME) pick elements out of formations which can't fight effectively on their own, and then the Groups in the Plan force those elements to be hung out to dry.