German tank vulnerability

[Battlefront.com]

Another example of why an engineered solution works better than design for effect has to do with automatic weapons sound effects.

In CMx1 units fired in abstract "volleys". One sound effect was assigned to each type of volley that we decided needed it's own unique sound. When the unit fired that one sound was used.

In CMx2 every round is explicitly simulated. For automatic weapons this means there is no such thing as a "volley" because the number of shots fired is situationally dependent. Therefore, we have individual shot sound effects which the game plays for each shot. 1000 rpm burst of 25 rounds could very well use the sound effect as used for 500 rpm burst of 5. But the sound to your ears will be quite different and "make sense" in the context of what that weapon did.

Some weapons have different "reports", so we do have a variety of individual shot sound effects. Each weapon in the game uses the one that we feel is the right one. But the game is what determines you actual hear based on rate of fire and number of rounds fired.

Steve

[Rocky Balboa]

Wind, as my previous post pointed out, already changed the paths of shells in CMSF, especially those that have little kinetic energy such as rockets.

As to rain, I doubt it would make a difference. Perhaps it would increase the scatter of bullets at long ranges? For bigger shells, though, the effect of rain drops would be just as marginal as fog is for bullets.

Snow or ice does very little to provide cover unless there is literally a ton of the stuff. It takes several meters of packed snow to provide cover from rifle bullets alone. A tank with enough snow and ice on it to provide additional cover wouldn't be operable by any means! A more important weather effect on penetration is the temperature of the gun and the munitions. A 'hot' grenade may overshoot whereas a 'cold' grenade may undershoot the target, because the propellant doesn't react in the determined time. Which is why you shouldn't first fire a bunch of shots until the gun breach is hot, then load another one and leave it waiting.

I was thinking more along the lines of a slow moving bazooka or piat round skidding off the icy hull/turret when hit at certain angles or reduced fragmentation from shells/grenades landing in mud or flooded field (please tell me flooded fields and marsh will be included in the new terrain types).

I also like your example of thrown grenades being less accurate in ice/snow.

[Michael Emrys]

I also like your example of thrown grenades being less accurate in ice/snow.

He wasn't referring to thrown grenades. In many European languages 'grenade' is a synonym for 'shell'. He's referring to something coming out of a gun.

Michael

[zukkov]

hmmmm.. i'm not usually one for getting down to the nitty gritty of details. but there must be some chance involved here. what told that gunner he was on target and to shoot?

[Gryphonne]

hmmmm.. i'm not usually one for getting down to the nitty gritty of details. but there must be some chance involved here. what told that gunner he was on target and to shoot?

Quite curious about this as well! How does the AI know whether it is on target or not?

[Redwolf]

Uhm, still seems like you are misunderstanding things, Redwolf. There is no hit probability model.

A gun is no longer fired at a target like in CMx, where a hit or miss was determined at the moment of firing, no matter what the target gets up to between shot and impact.

A projectile is now fired along a certain path and it will hit whatever is in it's path, be it a tree, wall or armour, whether that front, side, top or bottom.

So the only "hit probability" is in the path the projectile is given when fired. The rest is all simulated physics. So yes, it could very well hit that side plate.

Yeah but of course picking the path is the same as a picking a hit probability model. You will still have to decide how much variance you have in gun and shell and how much the gunner screwed up. That is picking a hit probability model, even if from then on you go by wind and true 3D model.

In any case the original question hasn't been answered, are there hits at > 45% now?

[Battlefront.com]

hmmmm.. i'm not usually one for getting down to the nitty gritty of details. but there must be some chance involved here. what told that gunner he was on target and to shoot?

Think about this logically... if I throw a ball directly at you, and you step aside after I throw it, where does the ball land? Exactly where I threw it (step one) and where the ball itself wanted to go (step 2).

When a shot is fired Experience, Suppression, target size, target speed, etc. are all calculated to modify the aim point from whatever optimal is. The better the conditions for the shot, the less modification of the aim point. After the shot leaves the barrel the flight path characteristics take over to see how close the round comes to hitting what the adjusted aim point is. That way you can have a well disciplined, undisturbed shooter missing a target by a wide margin because the weapon is inherently accurate, the range is outside it's optimal range, etc.

This is all very basic stuff. Very basic. CMx1 simulated a lot of it, but the flight path and the target were both abstracted. These used "hit probabilities", as Redwolf is asking about, because there was no literal information to key off of.

Steve

[Battlefront.com]

Yeah but of course picking the path is the same as a picking a hit probability model.

Oooo... no, I wouldn't say that. It's treating two entirely different methods of modeling as if they are the same. "Hit probability" means you are calculating the chance of something happening before it happens, then apply that chance to a particular situation to yield a result. That's a "design for effect" approach and it isn't what we're doing at all.

What we do is take the best case accuracy of a particular weapon, modify it based on soft factors and situational stuff, then "fly" the round using a physics model. Where it lands is where it lands. There's no determining this ahead of time, therefore there can be no hit probability.

The closest we have to a "hit probability" is the inherent accuracy of the weapon. This tells the equation how much natural deviation the flight path should have given a particular distance. But this value is so heavily modified in so many different ways, some predictable and some unpredictable, effectively the inherent accuracy value has no real value on its own.

Now, one can go back and quantify misses:hits and sort of reverse engineer a hit probability from it, but that's really an entirely different thing than starting out with a hit probability.

In any case the original question hasn't been answered, are there hits at > 45% now?

I'm really failing to understand why this isn't answering your question. Ask yourself one simple question... "could this happen in real life"? If the answer is "yes" then it can happen in the game, no problem. If the answer is "no" then it can not happen in the game. That's the benefit of simulating something literally instead of abstractly as CMx1 did.

Steve

[zukkov]

good enough for me. thanks for the reply!

[Elmar Bijlsma]

Indeed, calling it hit probability in terms of mechanics is somewhat misleading because whatever calculations are made at the shooting end may be nullified by external factors like the targets stopping, swerving or being obscured by another tank/building.

Thus the shell will hit whatever gets in its path. So yes, a hit could land on any surface of the target.

btw

If you want to be really gamey, there is something you can do to see how much freedom the ballistics model gives.

Bounce small arms fire off a building and into an area out of LOS/LOF. If you get it right you can hit troops behind a wall or in a side street.

Hideously ineffective and wasteful of ammunition but quite a laugh.

[Battlefront.com]

Indeed, calling it hit probability in terms of mechanics is somewhat misleading because whatever calculations are made at the shooting end may be nullified by external factors like the targets stopping, swerving or being obscured by another tank/building.

Which, incidentally, was exactly the problem inherent in the CMx1 system. We had shots go straight through buildings because the hit probability said it hit the vehicle and if the vehicle happened to have moved behind a house during flight path... tough. In CMx2 this could never, ever happen because the chance of hitting the target is never factored into anything. The shell fires and it goes wherever it goes.

Put another way, the shell in CMx1 knew before it fired if it would hit the target or not. In CMx2 the shell isn't even aware it is seeking a target. It just goes along for the ride. Very Buddhist

If you want to be really gamey, there is something you can do to see how much freedom the ballistics model gives.

Bounce small arms fire off a building and into an area out of LOS/LOF. If you get it right you can hit troops behind a wall or in a side street.

Hideously ineffective and wasteful of ammunition but quite a laugh.

Yup! We've all seen some pretty funny stuff like that. Or the opposite. Today I watched a shaken HMG42 gunner plow a long burst into a tree 20m in front of him. The GI that was running 40m on the other side of the tree was quite taken by this turn of events

Steve

[Sardaukar]

BTW, I don't know if it was touched in this topic, but in CMx1, superior optics had quite little effect to hit %.

How is CM:BN in this? And was there any meaningful difference between e.g. Sherman and Pz IV? Germans usually valued long-range shooting and usually had quite decent optics for that.

[LongLeftFlank]

For infantry fire at least, it looks like each shot or burst has some kind of CEP or "dither" assigned to it around the targeted Action Spot rather than always hitting dead center.... this CEP increases with range (as it should). In my CMSF Ramadi map where residential streets are narrow (8m across 2 "squares") and flanked by high walls, infantry are somewhat protected against defilade shots from angles >~10deg off the street direction -- a lot of bursts ricochet off walls (making a lot of odd noises BTW, but that may be my sound card) before they get to the target.

[The_Capt]

Indeed, calling it hit probability in terms of mechanics is somewhat misleading because whatever calculations are made at the shooting end may be nullified by external factors like the targets stopping, swerving or being obscured by another tank/building.

Thus the shell will hit whatever gets in its path. So yes, a hit could land on any surface of the target.

btw

If you want to be really gamey, there is something you can do to see how much freedom the ballistics model gives.

Bounce small arms fire off a building and into an area out of LOS/LOF. If you get it right you can hit troops behind a wall or in a side street.

Hideously ineffective and wasteful of ammunition but quite a laugh.

Nothing "gamey about it. I have seen this done first hand.

For example, Bosnia '94. We were up on a hill building a bunker for our infantry on a beautiful sunny day.

In a valley below Bosnian soldiers were trying to sap a trench towards a Serbian line. In order to do so they had to dig under a rail line. Unfortunately for them they left the rails over the top of the trench and just kept going.

We had lunch up on the hill watching a Serbian HMG bouncing rounds off the rails into the trench and Bosnians carry guys out the back end of the trench...ah good times.

[MengJiao]

Put another way, the shell in CMx1 knew before it fired if it would hit the target or not. In CMx2 the shell isn't even aware it is seeking a target. It just goes along for the ride. Very Buddhist

Wow. One finds Enlightenment in the most intriguing places.

But I guess that's more Zen and the Art of Artillery.

[GSX]

Redwolf,

I see you're still stuck a bit in the CMx1 mode of doing things There is no "hit probability" either. As Elmar and Sergei point out, something is fired and it's trajectory is traced exactly as it is shown on screen. Everything that should go into that trajectory, such as aim point, accuracy, velocity, drop, wind error, etc. is taken into account. Wherever the round hits is where it hits. If it hits the edge of a PzIV's fender... THAT is what it hits. There's no calculation like "well, the probability is it would strike the upper hull armor" because that is not what the round hit. Period.

What happens after that is based on the effect that particular round, with whatever properties it had at the time of the hit, would have on whatever was hit. In the case of striking the fender it would likely be simulated to pass through it and strike the armor directly behind it without much appreciable difference than if it hit that armor straight on (I can't say for sure because that's an extremely technical question). The strike against that armor happens because that's where the trajectory would take it. There's no probability because the game knows EXACTLY where it is.

After striking the armor the game figures out what happens. Penetration? Well, that depends on all the variables being carefully calculated. No probability here either, except in the form of variables within the equations where appropriate. Assuming there is penetration then damage is assessed based on what the round hits and what sort of physics are involved. In theory it could pass through and out the other side (happens to Halftracks quite frequently, for example) and impact something else.

It's all physics, not dice rolling. Which means we have 100% of our bases covered 100% of the time with 0% extra work after the initial modeling.

Steve

Just curious, all of the above being true, which it is in my experience with CMSF.

Where does the gunner aim? If he can see the whole tank that is? And is experience added into where he will aim against a specific vehicle?

[Battlefront.com]

Standard aim point is center mass. Which is one reason a hull down tank, with poor turret armor, is actually at a disadvantage because the only thing that can possibly be aimed for and hit is the turret. Obviously the reduced profile makes the chance of a hit less likely, but if there is a hit it's more likely to be disastrous.

I agree that higher experienced crews should have a chance of aiming for a tank's weaker areas, but that's not something that's got in yet.

Steve

[MikeyD]

Which is one reason a hull down tank, with poor turret armor, is actually at a disadvantage because the only thing that can possibly be aimed for and hit is the turret.

Oh wow, deja-vu! I recall this very topic going round-and-round back on the CMBB board years ago! Hull-down PzIV only presenting their weak turret fronts - there was blood spilt on the floor over that issue! :eek:

[Gryphonne]

Which, incidentally, was exactly the problem inherent in the CMx1 system. We had shots go straight through buildings because the hit probability said it hit the vehicle and if the vehicle happened to have moved behind a house during flight path... tough. In CMx2 this could never, ever happen because the chance of hitting the target is never factored into anything. The shell fires and it goes wherever it goes.

Put another way, the shell in CMx1 knew before it fired if it would hit the target or not. In CMx2 the shell isn't even aware it is seeking a target. It just goes along for the ride. Very Buddhist

Yup! We've all seen some pretty funny stuff like that. Or the opposite. Today I watched a shaken HMG42 gunner plow a long burst into a tree 20m in front of him. The GI that was running 40m on the other side of the tree was quite taken by this turn of events

Steve

Steve, isn't 'hitting the target' mostly a factor of having acquired the correct target distance? It's unlikely a shell would hit too far left or right of the target (save for effects caused by wind and ballistics, or against moving targets).

Thus with stationary targets, how does the AI gunner decide what the range to target is? Obviously, an AI gunner could have this information down to the meter and hit the target every single time. So what makes the round go either high or fall short? And how does the AI gunner then decide what to do? Because this *seems* random, isn't there some kind of "chance" of acquired the correct range to target then? Just curious about mechanics here

[Battlefront.com]

Steve, isn't 'hitting the target' mostly a factor of having acquired the correct target distance? It's unlikely a shell would hit too far left or right of the target (save for effects caused by wind and ballistics, or against moving targets).

Acquiring the correct target distance, aiming the gun correctly based on that information, leading by the right amount if motion is involved, etc. There's a lot to it

Thus with stationary targets, how does the AI gunner decide what the range to target is? Obviously, an AI gunner could have this information down to the meter and hit the target every single time. So what makes the round go either high or fall short? And how does the AI gunner then decide what to do? Because this *seems* random, isn't there some kind of "chance" of acquired the correct range to target then? Just curious about mechanics here

Range estimation is not explicitly simulated. Instead, there are a whole host of factors (including range) that contribute to establishing where the actual aim point is. Some of them are specific to the equipment being used, others are "soft factors" of the crew. Whatever the result, the round is fired and falls wherever it falls.

If the round falls short/over then one of the variables is adjusted so that all else being equal the second shot will be more accurately placed by increasing/decreasing the aimpoint's range. Subsequent misses are similarly adjusted. Experience has a lot to do with the initial delta between optimal and actual aim point, so the worse the Experience the more total shots are likely needed to actually hit the target.

Note that I said "all else being equal". If the tank crew has some sort of change in it's state of being, then those changes are taken into consideration. For example, the tank gets hit by something inbetween shots and the crew gets "Nervous". There's a chance that the next shot it fires will be worse than the first one, or at least not improved, since the crew is no longer in the same state as when it fired the previous shot.

Steve

[Txema]

Wow, Steve, very interesting explanations !!

Of course, the main problem that I see is what you explain here:

Standard aim point is center mass. Which is one reason a hull down tank, with poor turret armor, is actually at a disadvantage because the only thing that can possibly be aimed for and hit is the turret. Obviously the reduced profile makes the chance of a hit less likely, but if there is a hit it's more likely to be disastrous.

I agree that higher experienced crews should have a chance of aiming for a tank's weaker areas, but that's not something that's got in yet.

Steve

And this fact is the final ingredient necessary to explain perfectly what has happened in the AAR with the duel between the Pz IVH and the M4A3 Sherman. (The Pz IVH can not decide to aim for the turret, that is more easy to penetrate. He always try to aim for the center mass - the upper hull in that case, that is more difficult to penetrate).

However you are obviously fully aware of the problem and I am sure that you will fix it in the future.

Steve, thank you very much for all your explanations. It is a pleasure to debate about this topics with you :-)

And thank you for these AARs. I can't wait to play the game !!!

[stikkypixie]

Wow, Steve, very interesting explanations !!

Of course, the main problem that I see is what you explain here:

And this fact is the final ingredient necessary to explain perfectly what has happened in the AAR with the duel between the Pz IVH and the M4A3 Sherman. (The Pz IVH can not decide to aim for the turret, that is more easy to penetrate. He always try to aim for the center mass - the upper hull in that case, that is more difficult to penetrate).

However you are obviously fully aware of the problem and I am sure that you will fix it in the future.

Steve, thank you very much for all your explanations. It is a pleasure to debate about this topics with you :-)

And thank you for these AARs. I can't wait to play the game !!!

Wasn't it standard practice to aim for the center of mass? Did WW2 crews have the information we have now and *know* the weakness of a Sherman?

[MikeyD]

I'm reminded of a report I read from [*somewhere-lost-in-time*] that showed that a trained policemen during an altercation drawing and firing his pistol from [*some relatively close range*] had only a 30% probability of hitting what he was shooting at. You can aim specifically for the arm or the leg, or in tank terms the turret front or lower hull plate. But there's no guarantee that you're going to hit anything at all! Aiming for the center of mass at least increases the chance you're going to clip a piece of it. And of course if the round goes straight through the MG ball mount you can claim that's where you were aiming for all along

[zukkov]

didn't some american tankers scrape off the star(or painted over) because german tanks and atg's kept shooting holes through them? then again, it may have been a case of rounds coincidently punching holes through the star so it must be the star's fault.

[Michael Emrys]

I'm reminded of a report I read from [*somewhere-lost-in-time*] that showed that a trained policemen during an altercation drawing and firing his pistol from [*some relatively close range*] had only a 30% probability of hitting what he was shooting at.

Along those lines, there was an incident in New York a decade or so back where three plainclothesmen stopped a guy to question him, and when he reached for his ID they thought he was going for a gun. They fired something like 33 rounds (at least that's the number I recall) and only hit him 11 times. Range wasn't specified in the report I heard, but I am assuming it was more or less conversational distance, say about three meters. Among other considerations, it sounded to me as if they needed to spend more time on the range.

Michael