Improved Anti-tank gun behavior in CM2

[Warren Peace]

(This has probably been discussed before, but searching just does not work well.)

Anti-tank guns in CM are not modeled correctly. The basic problem is that they are too easy to spot after firing. I made a scenario in which two Pak 88/43 need to hold an objective against a british Armored platoon (1xfirefly, 3xshermans) The Map is 1600x800, gentle hills, farmland, light trees. Clearly in the real world the Shermans would be toasted. But not in CM. What generally, happens is that the tanks get fired on, one gets destroyed, and then the tanks target the guns and take them out after a few shots.

What is not modeled correctly, IMHO, is that it should be much more difficult for the tanks to locate the ATGs at 700 meters (about when they open up). Also, these ATG's are in heavy woods in Fox-holes and should be harder to take out with the 75s.

As near as I can tell from reading the tanks would get slaughtered by the 88's because it would take the troops a while to figure out where the shells were coming from.

Any thoughts?

Warren

[CMplayer]

I would like to repeat the situation.

What were the experience levels of the

involved units?

thx,

--Rett

[offtaskagain]

Theres a good story about this in "Panzer Commander", the autobiography of Hans Von Lucke. While serving as a regimental commander in the 21st Panzer near Caen, he located a battery of Luftwaffe 88s sitting around doing nothing in the rear lines. He ordered them to move to the front after nearly shooting the commander. Happily for him, they got into place in an orchard behind a farm field right before a full division worth of British tanks showed up. I believe this was during Operation "Goodwood". The 88s fired for like 20 minutes without being spotted while managing to torch something like 50 tanks. Von Lucke said you could see the ripples of the shells as they traveled through and over the grain.

[YECoyote]

it's very easy to reproduce this, because it happens ALL the time. If the tank is unbutton, they will spot the ATG right after the first shot. In All or Nothing scenario, my 88 would kill 1 tank, and the rest of his friends would blow me to kingdom come in 20 seconds after that.

The map size and the terrain (in CM) do not lend itself to the 88's advantage. In desert warfare, Rommel usually used the strategy where he first attacked with tanks. The tanks then immediately retreat. The Brits gave chase, and the 88's, at long range, would decimate the armour formation.

[ 06-27-2001: Message edited by: YECoyote ]

[MikeyD]

In CM2 at least the early scenarios are going to feature 37mm AT guns firing at point-blank range against almost blind T36/76s. How the heck BTS is going to model that correctly is going to be anyone's guess.

[Stacheldraht]

Another problem with AT guns is how the Tac AI gets befuddled when faced with multiple targets, manically switching aim between a few tanks without ever firing during a turn. Just take a shot, please.

[CMplayer]

<BLOCKQUOTE>quote:</font><HR>Originally posted by Warren Peace:

Any thoughts?

<HR></BLOCKQUOTE>

I just tested this some. I describe

the test below, but the general conclusion

is that for 88's to have the advantage

over a large number of 'vanilla' shermans

you want to engage at 1200+ meters.

(Also, I have had 57mm AT guns take out

3-4 enemy vehicles without being spotted

due to their getting first shot hits,

and firing through keyholed slots so

that not many pairs of eyes are looking

in their direction.)

Test results

2 regular pak88 at edge of woods in foxholes.

10 regular vanilla sherman tanks in open

flat ground.

one of the 88s in C&C of a +2? commander.

five battles at each of 3 initial ranges,

800, 1100, 1400 meters

engaging at 800 meters tanks killed/game

4,2,1,5,2 (both 88's KO) average 3 kills/game

or 1.5 kills/gun

1100 meters

5,8,1,7,4 (both 88s KO) average 5 kills/game

or 2.5 kills/gun

1400 meters

10!(one 88 KO),4,10!(one 88 KO),5,8

average 7.4 kills/game or 3.7 kills/gun and

gun survival in two cases.

Spotting: at 800 and 1100 meters the guns

were spotted within 10 seconds of opening

fire. At 1400 meters it could take up to

a turn or two, with the tanks closing range

to about 1300 meters before the guns were

spotted.

Also, the farther into the woods you put

the guns, the lower the chance of their

scoring a hit, (according to the LOS

targetting line in setup) but this may

positively affect their ability to stay concealed.

--Rett

[ 06-27-2001: Message edited by: CMplayer ]

[WWB]

I think this points to the broader issue which is vehicles and guns are way too easy to spot when in covering terrain.

Believe me, in Central Virgina, near my dads house, those cops, without using any car-sized ghillie suits, can hide in among the trees with ease, and you never would see them if they do not turn your lights on.

Also, does anyone know if sun direction will be simulated in CM? It was a major factor on the Eastern Front.

WWB

[Battlefront.com]

There is no easy solution here. The main problem is that the spotting system does not lend itself to having wide open fields of vision with many eyes and no distractions. It is basically a problem with Absolute Spotting vs. Relative Spotting (do a Search for more info).

For example, if 10 targets each have a 10% chance of spotting something then it won't take more than a shot or two before the gun is spotted. The problem is that once ONE target locates the gun, the other nine are made aware of it with absolute precision. In reality the one tank that spotted the gun would have to communicate its location which is a very difficult thing to do in most battlefield conditions. At least to the degree necessary for counter fire.

Now... you guys might say "well, just make the guns harder to spot". They already ARE very hard to spot. If we make them harder to spot then that means a single target will have next to no chance of seeing what is shooting at it. Then you get the opposite problem where guns are now unrealistically difficult to spot.

Until we move to a Relative Spotting system (which I don't believe any wargame currently has?) there isn't much we can do.

Most early war tanks will be particullarly blind and due to inexperienced crews very difficult to move around. Too complex a topic to get into here.

When the TacAI can't "make up its mind" it is most likely because of poor quality LOS. In other words, it spots the best target to shoot at but it moves out of LOS before a shot can be taken, so after a couple of seconds it will switch targets. This is to avoid a huge problem known as "target fixation", which is FAR worse than the ocasional "make up its mind" problem. There is code in there to make the TacAI less likely to switch human designated targets as well as to not switch around too much. But in a target rich environment with lots of small LOS obstructions, the system gets stressed to the maximum. There is little we can do about extreme situations like this. After all, we can't program in the human brain so there is only so much we can do.

Steve

[Vanir Ausf B]

<BLOCKQUOTE>quote:</font><HR>Originally posted by CMplayer:

Spotting: at 800 and 1100 meters the guns

were spotted within 10 seconds of opening

fire. At 1400 meters it could take up to

a turn or two, with the tanks closing range

to about 1300 meters before the guns were

spotted.

Also, the farther into the woods you put

the guns, the lower the chance of their

scoring a hit, (according to the LOS

targetting line in setup) but this may

positively affect their ability to stay concealed.<HR></BLOCKQUOTE>

CMplayer hits it on the head. If you want your guns to be harder to spot, don't put them at the edge of the tree. Sit them back 10-15 meters. Also, range makes a huge difference. Most CM AT engagements are at fairly short ranges.

[Battlefront.com]

More thoughts...

Rett's "test" shows that this isn't as much of a problem as some think it is (we have been over this issue many times in the last year ). Some people pick bad ambush locations, either because there are too many LOS ops for the enemy or because the gun is not in the most ideal location. It is best to put it back in the woods as far as possible while still getting a strong LOS out to your intended kill zone.

wwb_99, I hear what you are saying about concealment in the woods. I live in deep woods and spend a lot of time there. But there is a difference between a cruiser parked in cover and a AT gun that creates a lot of noise and smoke when it fires. Now... if there was a cronic problem of a non-firing AT gun, in woods, in a foxhole, being spotted at even a couple of hundred meters, then there would certainly be a problem. But to the best of my knowledge, this is not the case.

As for the power of the Sherman 75... it packs a really big HE punch, therefore firing at AT guns should be a rather dangerous event for the gun crews. So I don't see this being at all a problem for 75s whacking guns. AT guns are very fragile when taken under direct fire.

Oh... and another point to keep in mind. And that is vehicles currently do not have Morale. In real life engagements 10 Shermans would have taken 3-6 casualties and then the remainder would have withdrawn, leaving the AT guns intact (higher chance at least). Right now players, and the TacAI, keep things going as a slugfest. CM2 will change this when optional vehicle Morale is introduced.

Finally, if the tankers have the advantage of numbers, they *might* come out on top. Well, if you can call 6-8 dead Shermans for 2 AT guns a "victory". However, CM's scoring system does NOT think this way. Losing that many tanks is a good way to lose a battle.

[Warren Peace]

Steve, Thanks for the quick response.

I thought making them harder to spot might be the solution, but I guess not. What about making the guns a little harder to knock out once they have been spotted? Does the size of the gun have an effect on how easy it is to hit? Perhaps dug in guns could be smaller targets.

Warren

[CMplayer]

<BLOCKQUOTE>quote:</font><HR>Originally posted by Big Time Software:

There is no easy solution here. <HR></BLOCKQUOTE>

Maybe there is, but you are overlooking

it. See below.

<BLOCKQUOTE>quote:</font><HR>

For example, if 10 targets each have a 10% chance of spotting something then it won't take more than a shot or two before the gun is spotted.

<HR></BLOCKQUOTE>

The easy solution is to calculate spotting

from the spotted's end, not the spotter's

end. This makes sense and fits the

absolute spotting system.

What I mean is this, if 10 units have a

ten percent chance of spotting a gun,

each time you do a spotting check you

put all the units which COULD spot it

together, and do a single check based

on that. BUT don't just add up the percentages linearly. Take the best

one, and then for each additional spotter,

in descending order of likelihood add

say, 50% of its chance, and for the next

on 25% of its chane, and for the next one

12.5% of its chance etc. Then do one

single spotting resolution.

In the example you gave, 10 shermans

with a 10% chance to see the gun, are

summed up approx like this 10 + 5 + 2.25 +

1 + .5 + .25 etc giving about a 21% chance

of being spotted at that resolution. (say

one resolution per turn) If that is still

too much, then make the drop-off even

steeper.

This way additional eyes count, but they

don't have the negative side effect you

describe.

Since you have 'borg' spotting, putting

the spotting together like this seems

like an elegant and appropriate fix to me.

regards,

--Rett

[ 06-27-2001: Message edited by: CMplayer ]

[Battlefront.com]

Warren Peace:

<BLOCKQUOTE>quote:</font><HR>What about making the guns a little harder to knock out once they have been spotted? Does the size of the gun have an effect on how easy it is to hit? Perhaps dug in guns could be smaller targets.<HR></BLOCKQUOTE>

Yes, target size and degree of cover are huge factors already. So a dug in Pak38 in Tall Pines is going to be a tough nut to crack compared to a non dug in Flak 36 parked in the middle of Open Ground.

The bigger the target, the greater the chance that a near miss is going to cause problems. Many times what I see happening is that the AT gun doesn't get knocked out, directly, but instead gets hammered so much that the crew buggers off or remains unable to return fire (Pinned) until something does actually finish it off.

Another tip about AT guns... Plan on having dead spots for protection. Putting a gun in the best LOS spot is not necessarily the wisest choice. Remember... LOS works both ways. If you can see the whole map, then everything on the whole map can see you. If on the defensive that means a lot of enemy attention can possibly come your way all at once. A good example of this is Valley of Trouble which came with the Demo. Each side has some great hills, but they are quite dangerous to be on if the other side knows what it is doing.

Relative Spotting (which will NOT be in CMBB) will make this a little less dangerous, but not hugely so. Since the human/AI player will always be able to unrealistically coordinate units, a high profile target in a spot that can be hit from a lot of different places will always have a slight unrealistically higher degree of danger for the defending unit positioned there.

Steve

[PVLLVS MAXIMVS]

I have two things to say.

1> I am continually astounded by the fact that BTS puts so much effort into not just supporting the CM community, not just listening to the CM community, but actually being a PART of the CM community. Props to BTS, cuz they kick butt.

2> What experience I have at this game has made me think that there are 2 solutions to this problem, short of rewriting the game itself...

A> Use AT guns to cover approaches which are tight enough that onle one or maybe two tanks can come at you at once. If you have 5 tanks firing at one gun, sure, its toast. But five tanks which have to, say, come around a corner one at a time will get slaughtered.

B> If your map is too open to allow for a setup like above, then use multiple guns to cover the approach. Obviously 5 tanks facing 3 guns have less chance than 5 vs 1. Of course, that makes you vulnerable to artillery, but you cant have your cake and eat it too.

[Renaud]

A tactic I use in AT ambushes is to use some light artillery (typically mortars) to cause all the intended targets to button up a minute or so before you spring the AT ambush. This makes a huge difference in how long your AT guns last. It's practically a requirement for me now. Of course infantry eyes are a problem so I try never to open up when I know a horde of infantry can see the AT gun.

Ren

[ParaBellum]

Another problem with AT guns:

I have placed a 75mm gun in a nice ambush position in tall pines in a foxhole. Good chances to get of some shots before beeing noticed by the enemy. A Sherman moves towards the gun, distance 350m. I order the gun to unhide and target the Sherman. First shot-kill. Great!

The tank never knew what hit them. But what's happening now??? The gun crew decides to switch target to some infantry in a patch of light woods about 400m away!!

Result: Enemy sees gun and knocks it out quickly with mortar fire.

I've seen this a lot...

I really would like to see a command for CM2 to only engage tanks or vehicles.

I want my AT guns to engage tanks, not single squads!

Or tweak the gun's AI so that they are more reluctant to fire on infantry.

It's really hard when you see an AT gun after hitting a tank, instead of going back to "hide" to ambush another tank, to start shelling minor threats, getting spotted and killed fast.

Ah, and since I didn't say it for quite some time: CM is the best game I've bought in a long, long time. I'm now playing it for a year and in this time the CD left the drive only for some very brief moments.

[ 06-27-2001: Message edited by: ParaBellum ]

[Battlefront.com]

Rett,

First of all, my example of a "10% chance" thing is a gross over simplification of what happens in the game. The 10 Sermans would each have a variable chance of spotting the AT gun based on LOS quality, buttoned status, experience, facing, size of the gun, terrain the gun is in, if it is dug in or not, range, if it fired, etc. etc. In other words, there is a HUGE number of factors involved to come up with the % chance of being spotted for a given slice of time.

The 10 Shermans on second #52 might actually have this degree of a chance of spotting the AT Gun ranging from ZERO to a few percentage points. But the longer the engagement goes on the greater the chance that the vast quantity of Shermans has of spotting the target. And once a SINGLE Sherman has spotted it, every single unit in the game (including the other Shermans) knows where the target is. Your suggestion does not fix this problem and it is the single biggest reason for situations where the attacker spots the defender too quickly.

<BLOCKQUOTE>quote:</font><HR>What I mean is this, if 10 units have a ten percent chance of spotting a gun, each time you do a spotting check you put all the units which COULD spot it together, and do a single check based on that.<HR></BLOCKQUOTE>

This really can't work in CM. At least not in a way that would likely result in better behavior. The reason why is that units do not all spot at the same second, so it is impossible to "add up" all possible spotting units and come up with one result. And if we did move to that sort of system I think the chances of things being spotted would go up, not down. Currently each unit is allowed to spot on so many times during a turn. But WHEN those chances come up are situationally dependent. For example, a tank firing at Target A might only get a chance to spot other targets once every 6 seconds, and then the precentage chance decreases depending on the location of the target to that tank's arc (i.e. directly behind is most likely going to be 0%). So the inherent "chaos" of spotting actually works in favor of the defender. Your system would require removing this chaos and that would, at least as it appears, be a step backwards.

Steve

[Jeff Duquette]

<BLOCKQUOTE>quote:</font><HR>YECoyote Said: The map size and the terrain (in CM) do not lend itself to the 88's advantage. In desert warfare, Rommel usually used the strategy where he first attacked with tanks. The tanks then immediately retreat. The Brits gave chase, and the 88's, at long range, would decimate the armour formation.<HR></BLOCKQUOTE>

I disagree. CM map sizes can easily be cranked up to 3 or 4 klicks. The games map scale is, in theory, very conducive to modeling long-range anti-tank fire and tank gunnery.

Regarding easy spotting of ATG's…I agree that this is a condition of the game. Part of ATG spotting is a function (believe it or not) of ground conditions. Warm dry days tend to result in a fair amount of dust being generated as a result of gun discharge. ATG's were typically spotted from dust clouds generated when the ATG fires. Muzzle flash or "marginally" smokeless powders were also tell-tale signatures. In the desert it is very difficult to conceal an anti-tank guns position once it opens fire. A lot of dust. British training manuals generated from experience in N. Africa for the 2-pdr and 6-pdr stress wetting the ground down in front of an ATG or spreading camouflage netting below the barrel and muzzle to reduce the potential for kicking up dust when firing. Problem is water is a commodity in the desert and I reckon soldiers were at times reluctant to poor their meager water ration onto the ground.

Moist or wet surficial soil conditions reduce dust generation when firing. Soil cover ala grass and other vegetation also reduce dust. Wind also acts to dissipate dust quicker.

With respect to simulating random spotting in a game…this seems to be an age old groan on the part of wargamers, and a insurmountable game side effect for wargame designers. One unit sees for all units. However, in my humble opinion, conditional spotting is more suited to CM turn sequencing than any other TAC sim out there.

Why? Because of the simulatenous hands-off movement execution phase occurring in CM. Randomize spotting ability during the execution phase. Just because one tank "sees" an ATG doesn't mean every tank automatically "sees" that ATG. Course this limited to only one execution phase, as player control will kick again on the next order phase.

[Freak]

I would like to second a new command that allows ATG's to open up fire on a specific target and then go directly back to hide status if the ATG makes a hit.

Example:

Say an 88 flak is in hide in some woods with a flank shot on a sherman. I issue the 88 flak to target the sherman. The sherman gets knocked out on the first shot, and then goes right back to hide. This could help in the presevation of ATG's a bit. If the ATG is spotted by the enemy, well then there's no pont in hiding so the ATG will fire back.

So I guess the proposal would be two seperate firing commands for ATG's or guns in general.

These would be:

1)Regular target as in CM.

2)Stealth target, or simply fire and hide.

This could be a command just for ATG's or guns in general.

What do you think?

[Battlefront.com]

Hi Jeff,

<BLOCKQUOTE>quote:</font><HR> Because of the simulatenous hands-off movement execution phase occurring in CM. Randomize spotting ability during the execution phase. Just because one tank "sees" an ATG doesn't mean every tank automatically "sees" that ATG.<HR></BLOCKQUOTE>

As the CM spotting system currently works, a unit "spots" another one and that triggers a "flag" which mean all units can see it. This is the typical way spotting is done and is what we call the Absolute Spotting system. There is no way to randomize this other than putting in some retargeting delays (which we have done) so that when unit A spots Target it doesn't mean that unit B imediately acts on that information. But there this work around can not be extended too much or there will be negative side effects.

The Relative Spotting system works the other way around. Instead of the Target saying "hi, I am the target. Someone spotted me, so bring it on!!" the individual units keep their own unique list of what they spotted or didnt (and using CM's system, the degree spotted). This means that there is no inherent broadcasting of spotting information since each unit can only see what is on its own list. When we implement this system (after CMBB) there will be a information "hand off" based on a bunch of factors like C&C so there will be a chance of passing on learned information to other units. Obviously, with WWII technology this will be limited at best.

When the player gets into the Orders Phase he will NOT be able to order a direct fire mission for a unit unless the target is in fact already on its "spotting list". The player will NOT be able to override this. Indirect Fire and Area Fire are impossible to prevent being used by the player. Although we might be able to reduce the effectiveness of such actions (i.e. firing blind into the woods won't be as effective as if the target were spotted by that unit).

Steve

[WWB]

One other point: ATGs get exponentially harder to spot if placed deep within woods, just at point where they can barley see out.

I am currently playing a scenario where a large british armored force rushes a number of ATGs placed on a woody ridgeline. I have noticed that the guns tend to last alot longer if placed deeper within the woods.

WWB

[CMplayer]

Steve,

Thanks for replying. I mean a big

thanks.

In that case another way of trying to

solve the problem would be that you

_retain_ all the chaos and spotting uncertainty which you already have

but that _on any given second_ of the game,

if several units have a chance of spotting

the same gun (or whatever) then the one that

acc. to your system has the highest

chance of spotting gets to do its resolution

normally, but the others suffer an additional

_reduction_ in their chance. The

result would be that the curve of overall likelihood of being spotted would taper

off as you add more spotting units. It

would keep the guns from becoming too easy

to see, when there are many eyes, without

making them too hard to see when there

are fewer eyes.

What is the maximum number of times units

can spot? Once per second, once every three

or six seconds? Perhaps you could bunch

all resolutions occuring within 3 seconds

of each other, and apply this spotting 'hit'

to units doing their spotting resolutions

within a close time proximity of each other.

Obviously I don't know how the game is

designed, but some way of getting

diminishing returns for the 'extra eyes'

could be exactly the fix needed, until

the day when relative spotting can come

into play.

regards,

--Rett

[Battlefront.com]

Hi Freak,

It ain't that simple When we rewrite the core game engine (after CMBB) we will include SOP (Standard Operating Proceedures) for units. But this will require a ton of programming, testing, tweaking, programming, testing, tweaking, etc.

Steve

[Freak]

Ahhh, Thanks Steve. Looks like I'll be waiting for CMII the rewrite for sure. Seriously, I am way way excited about CM2 and the CM family of computer games! Thanks for all of your support and your amazing work for making CM the game and place it is.

Thanks again Steve.

[ 06-27-2001: Message edited by: Freak ]

[ 06-27-2001: Message edited by: Freak ]