Marching.....Spacing

[Garnier]

It should be easy to create a few generic offset values since GCM games have divisions have only a couple infantry brigades and 1 or 2 batteries.

GCM divisions have no hard limit on size, and can be up to 100 brigades with random/historical OOBs. Won't be too hard though, I have plenty of geometry code already there for generating maps.

Meanwhile I can swap the names of DRIL_Lvl4_Inf_Line and DRIL_Lvl4_Inf_Reserves to get them to deploy right with sandbox=. Do you know if the AI uses formation names or just the id numbers in column A from drills.csv?

[Marching Thru Georgia]

The numbers listed in column A are just the order of the formations in the drills file. It's strictly a visual aid. As far as I know, it is only used to light up the appropriate button on the toolbar via the '#formnum=' statement. So if you put a new formation in between two existing ones, the wrong button will light up since it's been hardcoded with that statement. That's a pain, because adding a new formation means having to change all the numbers in gui.csv. Hopefully, NSD won't do that in SOWW. However, putting that new formation in between two others will not screw up the game since the numbering is done at the start of each game.

[Davinci]

What problems are you seeing in your open play games? The AI decides whether to maintain a reserve force and if/when to commit it.

Thanks, hopefully this is something that will change in the next edition of the game. The player should not have the knowledge of knowing that the AI will place a part of it's army in reserve.

There should be a hidden level of mystery here.

That is not something that can be modded. Although maybe it is influenced by the Style value of the commander. I've never looked at that.

There was a time when I changed the numbers from 0 - 6 to a 9 for the purpose of trying to get around the games-code. At the time it seemed to work, but now the AI seems to hold back until I continue to probe forward before they will commit to battle.

Use of the defines.ini setting for initializing the troop's starting formation is a straightforward solution, but as Garnier points out, that is not usable for MP.

Understood!

We (well MTG) found that lowering the style rating of the generals stopped them sending some divisions off into useless corners of the map. Strangely enough the worse they are rated the less they split their forces up. Personally I'd expect the opposite to be correct so maybe its an unintended feature, or possibly I'm misunderstanding what NSD mean by the "style" of the generals.

I tried lowering their values at one time to stop them from wandering forward, but decided that slowing their speed down reduced or basically solved that problem. It is extremely rare for one of the enemy Generals to die by moving too far forward.

But, You are saying that giving them a lower rating of 0 or 1 makes them commit more troops to a battle. I agree with you that this would appear to me that they should be more cautious with committing troops.

The biggest-problem with experimenting with the OOB is that this is one of the files that can't be changed with a saved game. So it makes testing any changes extremely slow.

@ MTG - Hey, did You find a way of spreading out the units in Column-Formation at the start of a game?

For one reason or another, I never could get them to spread out.

Thanks,

davinci

[Saddletank]

Davinci - we have used style=0 or style=1 many times now in battles co-op vs the AI and it seems to make the AI better in that it all comes at you at once. Our previous problem in co-op vs AI games was that it sometimes took us most of the 90 minute or 2 hour game time to just find the enemy! There was one famous 2 hr game on one of teh Pipe Creek maps where we never did find him, and that was a corps vs corps game!

If you just set the C-in-C to a low style number but leave his division commanders as normal you can get some frightful attacks handed at you.

I admit that we made this change purely for better MP co-op vs AI games because we had a time limit of 2 to 3 hours and wanted a battle in that time. If you play solo and can save your game and spend as long as you like playing, this might not help as it does determine that the AI won't have a reserve although you still won't know what he's sending where.

[Davinci]

Davinci - we have used style=0 or style=1 many times now in battles co-op vs the AI and it seems to make the AI better in that it all comes at you at once.

Yes, I would like one of those battles, but I might be causing the very problem that I'm complaining about.

There was one famous 2 hr game on one of teh Pipe Creek maps where we never did find him, and that was a corps vs corps game!

That reminds me of something that I meant to ask on the forums a long time ago.

One of my Corps Commander was placed at the exact spot to view a enemy Division marching cross-country in an Open-Play-Game. I would guess that this is why I never could spot him using any of the roads. Without warning the enemy suddenly reversed-course, marched back to the crossroad, and then starting marching on the road.

I'm under the impression that the AI will march a certain route, and if they don't come into contact within a certain amount of time, they will leave the road-system and march cross-country.

If you just set the C-in-C to a low style number but leave his division commanders as normal you can get some frightful attacks handed at you.

Thanks, I'm going to try this out this week-end.

If you play solo and can save your game and spend as long as you like playing, this might not help as it does determine that the AI won't have a reserve although you still won't know what he's sending where.

Hmm, this is the area that I think that I might be causing the problem that I'm complaining about.

By, slowing down the game and playing it in intervals, it takes eight-hours of game time to locate the enemy units. The units march at a very-slow speed, and somehow the running is disabled from something that I modded.

I'm under the impression that nothing happened real-fast in the Civil-War, so I slowed down my game to reflect that. So it basically takes hours to move the units around on the map. This might also affect the AI units as far as the calculations go, I'm not sure.

I probably need to experiment with different settings in the OOB to see which one works best for me.

davinci

[Marching Thru Georgia]

Since the KS group plays a number of co-op games, we have spent a lot of time trying different things in order to force the AI into bringing most of his forces to the battle, sometimes on multiple axis. Here is what we have found. The top level of the AI, the part that decides on an initial movement plan, is not very well developed. So you have to help it if you want a battle where the AI concentrates his forces.

If you want the AI to concentrate his force, an objective needs to be placed on the map. Without it the AI will always default to Hunt Them Down mode, where the AI just randomly assigns sectors to each division with no thought on mutual support. A sandbox Defend game will have the AI marching most of his forces to the objective. A corps with 5 or 6 divisions will usually hold one back in reserve. That's perfectly rational. If you want the AI to use those reserves before the battle is lost, place the AI commander near the place of battle. He will get his division commander's request for help much sooner, and sometimes actually anticipates their need and bring the reserves up in a support role.

If you would like to attack the AI, then select the Attack option. Note that after the game starts, you have to exit it and go into scenario.ini and change sandbox=2 to sandbox=1. That is a game bug, attack is choice 1. As a result of this bug both of your forces will be some distance away form the objective. But the AI will concentrate on the objective.

The AI does a better job of managing a more coordinated battle if a high percentage of the commanders have a style rating of 1-3. I don't recommend 0. Forget about historic values if you want a decent battle.

There are a few caveats. The AI is impatient. Once an AI division reaches its destination sector, it will wait a few minutes, (5-10), and then send back a notice that there is no enemy present. There is a high probability that the AI corps commander will send new orders. The division will then march cross country to that and any subsequent sector. I think that was to give the impression of a unit marching to the sound of the guns, even if it is not. So if the AI forces reach the objective and there are no enemy forces to be seen, the AI will slowly give new orders to the divisions to hunt you down. The new destinations will be randomly distributed on the map. Cohesion dissipates. If the AI division does not find the enemy at its new location, the odds of it moving to a third location are drastically reduced. So you will need to arrive at the objective quickly or else the enemy will disperse, usually keeping only one division at the objective.

With that behavior in mind, it becomes apparent that the map size is an important factor. The map, regardless of size, is divided into 25 sectors. The area of the sectors change significantly with the size of the map. The original maps had sectors of 0.5 mi. on a side, (area = 0.25 sq.mi.). For the newer 5 mile maps, sectors are 1 mi. on a side, (1 sq. mi.). Your 10 mile map has sectors nearly the size of a small map, 2mi. on a side, (4 sq.mi.). So the odds of the AI finding the enemy by randomly distributing its forces goes down significantly as the map size increases. Also the ability of the AI to send reinforcements in a timely manner drops too.

Hope that helps.

[Davinci]

Marching Thru Georgia - Thank You!

Not only does that help me, but I'm sure that that information will help a lot more players that read it.

As I have stated above, I’m probably causing some of the problems that I’m complaining about, but the game is still great.

I’ll try changing the enemy-officers rating first to see what happens.

I’m trying to avoid adding any objectives since I don’t want to influence the AI into moving a certain direction.

I like the idea of dividing my forces at the start of a game and trying to bring them back together when I finally locate the enemy-forces.

Thanks Again!

davinci

[Davinci]

When I mentioned that the AI shouldn't always hold back a reserve, I didn't mean such as several brigades, but I'm under the impression that it holds back a whole division.

But, as noted, I might be causing this problem, so I'm not sure.

This is a typical Open-Play-Game that I have had for the past several years. A different part of the battle-field, but usually the same results.

Both armies have three-corps divided into twenty-three brigades.

General Hood has the strongest corps so I will always send him marching first as the search and main-attack force.

General Ewell has the smallest of my corps, so I'll send him out after the roads are clear of General Hood's men, but have him marching using a different road-system.

General Pickett - is only used as the reserve force for when one of the other two finds the position of the enemy.

In the picture below, General Hood has located two of the three enemy divisions, while General Ewell is sending him three of his six brigades.

General Ewell's remaining three brigades will act as a sort of blocking force while General Pickett sends General Hood four of his eight brigades.

Problem - I know that I can do this without encountering the third-enemy-division due to the fact that 99% of my games the enemy has never attacted with all three of his divisions. I shouldn't have that knowledge of knowing that.

Question - Do any of you have similiar battles, or do you encounter all of the enemy divisions?

The attachment 1_2014-01-25.jpg is no longer available

davinci

[Saddletank]

From what you are saying it sounds like this is a scenario you have played several times. No AI system is going to give you a fresh experience time after time. Most of my solo games are sandbox hunt them down games so every encounter is fresh and I never know what I am facing and where the enemy is or is likely to be. In these cases it doesn't matter if the AI high command performs well or badly because I simply have no idea.

I don't think I have ever replayed a single SoW game I've done. Not one.

Another self-imposed rule is I never click on distant enemy flags and generals to find out who they are (because you wouldn't have that knowledge). I sometimes check them out once they get inside musket range because then you'd begin to have wounded and deserters and POWs straggling about and you could find these things out, but sometimes I'm too busy commanding to worry about even that.

So not knowing what enemy formations you are facing helps cut out a sense of knowing too much.

Do you pay with a limited camera view at all? I find that helps to increase the sense of immersion as well. With a camera view that doesn't see over trees or move far from your general you cannot gain unfair knowledge about what your opponent is doing.

These things all combine to make the battles seem far more real and less like games against an artificial intelligence.

[Marching Thru Georgia]

I agree you shouldn't have that information, but since it looks like you are playing a hunt them down game it is unavoidable. The AI doesn't know about mutual support between divisions and corps in that situation. As I don't see the Union CinC at either fight, he will not even think about releasing the reserves until he receives a request from one of his subordinates. That will occur when a command takes 1/3 casualties and again when it reaches 2/3. However, given that this is the 10 mile map, the show will be over long before they could be of any help.

To have a fair fight, especially on a map that size you'll need to help the AI as I mentioned before or send your corps/divisions off to random sectors as the AI does. That's the only way to even the odds.