First test with Fireflies vs King Tiger

[PhantomApple]

Edited out - too slow of a typer

[sburke]

TRPs are needed for artillery and are much to scarce to model this advantage of defending guns.

They can play a dual role. As a TRP for artillery and as a bore sighted location for an AT gun.

[Steiner14]

They can play a dual role. As a TRP for artillery and as a bore sighted location for an AT gun.

You are aiming with your guns only, if enemy tanks are close to TRPs for arty?

[sburke]

You are aiming with your guns only, if enemy tanks are close to TRPs for arty?

Not sure what that is supposed to mean. I suspect however you are under utilizing your TRPs.

[slysniper]

I am glad I was too busy at work to see this thread today.

What I get out of all this is, they seem to not know what their own game is programmed to do, so asking if this is what is intended cannot be answered because they are not sure.

They are going to look into it only if you test it to a point to prove that in fact what appears commonly happening, is happening.

They do not have the time to run the testing, that makes sence, but should they not know if its programmed to have a varience. If they just answer the question, then we can look for ways as to why we might not be seeing it.

Your test shows no varience between the ammo type, all we want to know, does the game have one.

I have just placed this game back down to the level of nothing more than a game, I need to forget about it as to trying to be a realistic model. If they cannot tell what the intended results should be without someone else testing it within the game. Then making adjustments must be a nightmare for them.

How do you adjust a formula unless you know what the intended results should be. But you never hear them tell you that.

Like we intended the tanks to have a accuracy level of 80% on judging ranges at approx. 1000m. then that increasing by 10% with each shot until range is known.

I think I found the game will allow me to kill King Tigers out to 1000m from the front with Ammo that likely should not be hitting often at that range but does within the game. Is that historically correct, depends on who you ask, but it appears it does not matter because it is not changing. So on them few rare occasions where it happens within the game and people wonder, let them. I just hope I am the one killing the beast.

History is twisted all the time anyway, each generation writes their own version of it.

So I will write mine own and stop expecting the game to be the same.

This game will write its version (which my favorite at the moment is them sharp shooting pistols hitting at 100m) makes the tanks super ability pale in comparison.

Dont, worry Phil, no more comments from me for awhile about any unrealistic traits the game might have. I dont want to hurt sales or image

[Guest]

I am glad I was too busy at work to see this thread today.

What I get out of all this is, they seem to not know what their own game is programmed to do, so asking if this is what is intended cannot be answered because they are not sure.

They are going to look into it only if you test it to a point to prove that in fact what appears commonly happening, is happening.

They do not have the time to run the testing, that makes sence, but should they not know if its programmed to have a varience. If they just answer the question, then we can look for ways as to why we might not be seeing it.

You seem to have a very simplistic understanding of how this game works. That's all right. But projecting that simplicity onto the game itself - or BFC - is severely flawed. This seems to be a fairly nuanced question people are asking. "Is a particular type of ammunition accurately modeled?" is something that can only be answered with testing. We model it, yes, and with the best data we can get. There are a huge number of other variables that come into play, though. The end results are best shown through significant testing.

Your test shows no varience between the ammo type, all we want to know, does the game have one.

This was asked and answered back when the same threads came up after CMBN came out. Yes, different ammunition is modeled differently. I thought what we were doing here was working out whether there was a bug or serious inaccuracy in how it was modeled.

I have just placed this game back down to the level of nothing more than a game, I need to forget about it as to trying to be a realistic model. If they cannot tell what the intended results should be without someone else testing it within the game. Then making adjustments must be a nightmare for them.

How do you adjust a formula unless you know what the intended results should be. But you never hear them tell you that.

Like we intended the tanks to have a accuracy level of 80% on judging ranges at approx. 1000m. then that increasing by 10% with each shot until range is known.

There are heaps of variables that go into shooting. Ballistics in this game are fully modeled, as are a number of other factors. If there was a simple answer I would have given it.

Dont, worry Phil, no more comments from me for awhile about any unrealistic traits the game might have. I dont want to hurt sales or image

Well, you can feel free to keep posting then. I'm sure this sort of argument - "is X historical thing accurate / flawed?" - is going to come up very, very often, just as it always has. And rest assured, if there ARE flaws we'll do our best to correct them.

[akd]

Your test shows no varience between the ammo type.

Although that may be the case, Vanir's test does not conclusively show that. A relatively small sample shows that there is no or little variance in the first round chance to hit a target of a certain size at a certain distance.

Think about this simplified scenario:

Assume range is known and target is at 1000m. Target is a 100' x 100' square. A gun fires 100 rounds each of two different types of ammunition at this target and every round is a hit. Can we then assume there is no variance? No, we can only conclude that neither round has dispersion great enough to cause a miss on a 100' x 100' target at 1000m.

[slysniper]

Thanks for the reply, Phil

Yes I agree that my view is simplistic.

But let me put it in perspective on how I see things.

I work in an engineering Firm, which we design structures.

I can promise you, that the formulas are just as complex in our programs as they are within yours.

The point I want to share though is this. We start with a base of design, we know our loads, we know our forces, and we know how we want the structure to react. We know what we want the finished product to do. We do not delve into the complicated math the machines now does unless we see results that do not come close to a basic desired concepts of the design we were anticipating.

When we see unusual results, we then investigate the calcs. To see what is causing the unexpected results. Most of the times it is an error or incorrect selection of a factor.

So in my world, I would expect you to have set certain goals as to how the end results of the game would react. At times I think it appears to me it’s more of here is the imputed math and the formula’s that we believe will give correct ballistics and the results are what they will be without having any base plan as to what they should match.

So I just point out things that appear wrong, figuring that. Hey, have you noticed this might not be reacting as to the concept you want for your game.

Only you know or can check if the math is correct and that the output is what you want. Since I know neither you anticipated end results desired or the way you are programming the game to do what it does.

All the testing in the world on our end is worthless unless there is direction as to what BF wants looked into and feels that testing could help in making possible adjustments to. People are not going to test away on aspects of the game if they feel BF could care less about how the game is presently acting and respond as such.

This example, you have someone that points out that special ammo should likely have a poor hit ratio because of flight issues with the round. OK, you point out the game has a variance for that, there should be a difference. His limited test shows otherwise, so you inform him that the testing needs to be much more intense and looked at as to if it is giving good results. Nothing wrong with that.

But if I see no change at all, even in the limited test. I would say. Your efforts might be showing something there that we might need to address in the game. But I need you to do more testing to help show that there is indeed a flaw in what we would like to see the game preform at. Giving guidance as to what BF would expect the variance might be. Getting help for free, takes working with people. Treating their efforts as nothing is not going to lead to much positive help.

I personally would love to get the super pistol killers toned down within the game.

But I will not waste my time, unless I felt that my testing was going to help you, that I was doing it in a way that would be of value to you and that you would actually look at it and have a known goal as to what we thought should be getting portrayed within the game.

I test presently for my own enjoyment as to know what to expect when I play the game and to have a good feel on how I want to play it. I will never be throwing hours of my life away in vain efforts without directions that make it worth the effort. I would not expect any other user to do so either.

But I would also likely be very helpful if I had direction on what testing could be done to make possible corrections to things BF wanted looked into if it could improve the game.

[Vanir Ausf B]

We model it, yes, and with the best data we can get. There are a huge number of other variables that come into play, though. The end results are best shown through significant testing.

This was asked and answered back when the same threads came up after CMBN came out. Yes, different ammunition is modeled differently. I thought what we were doing here was working out whether there was a bug or serious inaccuracy in how it was modeled.

Here is what I'm not understanding. If you don't know what the intended behavior is, how would you know if the test results showed a bug or intended behavior?

[Vanir Ausf B]

Getting help for free, takes working with people. Treating their efforts as nothing is not going to lead to much positive help.

Hey slysniper, I'm glad to see someone understands why I gave up on this headache.

But I'm ok with it. Not having any more testing in front of me is liberating.

As for the game... it is what it is. Whether or not APDS is the same or less accurate as AP will have to remain a mystery to all, including BFC oddly enough. I'm personally convinced that APDS dud rate is not modeled, but it won't impact my enjoyment of the game enough to lose sleep over

[JonS]

APDS dud rate

One of the issues is that the dud-rate is not the same as the dud-rate-that-has-a-noticeable-effect.

Take uneven petal separation of APDS. It's a known issue, and the increased drag on one side of the round would tend to cause the round to slew unpredictability. All the more so because it's spinning like a banshee due to the barrel rifling.

However, the mere fact of uneven petal separation does not intrinsically cause a miss. Range to target and the length of time the sticky petal remains stuck are two obvious factors that spring to mind which would affect whether a hit was achieved, even when one petal sticks.

For example, Radley-Walters might have had major problems with his APDS rounds on the morning of 8 August, but he still would have killed Wittman even if one petal had stuck all the way till impact because the range was only 140m.

[Vanir Ausf B]

Jon, that is true. However, the US tests at Isigny -- conducted against actual Panther tanks at ranges not exceeding 600m -- are suggestive of a dud-rate-that-has-a-noticeable-effect.

[StellarRat]

Vanir, you might be on to something. To make your test a lot more valid you need to have them fire at a vehicle/target that is 6' x 6' cross section as viewed from the shooter. If they are still hitting 100% of the time then you will have a case. The Tiger II is much larger then the 6' x 6' test target in your documentation so dispersion isn't as a big a factor. Alternately, you could increase the range enough to account for the difference in sizes, but that wouldn't be as good a test. If you still get 41 out 41 hits the odds of that happening are roughly 400,000,000,000 to 1 and I'd say you have a very good case for a bug especially since your "14% @ 1000m" documents are the UPPER bound on accuracy.

[JonS]

Jon, that is true. However, the US tests at Isigny -- conducted against actual Panther tanks at ranges not exceeding 600m -- are suggestive of a dud-rate-that-has-a-noticeable-effect.

Acknowledged.

[JonS]

Alternately, you could increase the range enough to account for the difference in sizes

That'd be a really bad idea since range estimation errors are usually the single largest element of the error budget.

[StellarRat]

That'd be a really bad idea since range estimation errors are usually the single largest element of the error budget.

True, but range is only truly important when the round is far enough out to drop significantly or you are firing low velocity weapons. The 17 lb. gun with APDS has muzzle velocity of 3,950 feet per second and good aerodynamics. The resulting drop at 1000 meters (less than one second of flight time) is not very significant on a 6' x 6' target. Battlesight is usually good enough at ranges less than >= 1500 m for most hyper velocity rounds (assuming the ground is flat, your not moving, not much wind, etc...) Obviously, good optics properly aligned with the gun are extremely important under all conditions even battlesight. In theory it should be a VERY accurate round compared to a standard round due to the higher velocity, but if it's unstable, has a tendency to hit something coming out of the gun, or doesn't discard the sabot properly then all kinds of bad crap can happen.

[Vanir Ausf B]

Vanir, you might be on to something. To make your test a lot more valid you need to have them fire at a vehicle/target that is 6' x 6' cross section as viewed from the shooter.

I haven't tried that because it was my assumption that any target that small would not have enough armor protection to make the TacAI feel the need to use APDS.

Alternately, you could increase the range enough to account for the difference in sizes, but that wouldn't be as good a test.

My first attempt at testing was at 1000m. But I abandoned that effort early on after only 4 of the first 20 first shots were APDS. It seems that when the first round hit % is low the AI will most often use AP for a ranging round and then switch to APDS. That's smart, but it makes testing APDS at that range prohibitive.

[hoolaman]

The game engine is now so stable that we are into the territory of tweaking fairly esoteric and sometimes even rather subjective variables.

I get the feeling that the work that could be put in to test out some of these scenarios far outweighs the results of the eventual changes, and it is far from guaranteed that BFC will even bother tinkering with this stuff on a pretty stable and complete game engine.

Its a shame because sometimes these arguably minor issues can change the flavour and even the results of particular games, but none by itself is the sort of thing that will have anyone ragequitting.

[StellarRat]

I haven't tried that because it was my assumption that any target that small would not have enough armor protection to make the TacAI feel the need to use APDS.

My first attempt at testing was at 1000m. But I abandoned that effort early on after only 4 of the first 20 first shots were APDS. It seems that when the first round hit % is low the AI will most often use AP for a ranging round and then switch to APDS. That's smart, but it makes testing APDS at that range prohibitive.

Good points. I forgot that you have to deal the AI ammo selection. That does complicate the testing substantially. :-(

[hoolaman]

If you still get 41 out 41 hits the odds of that happening are roughly 400,000,000,000 to 1 and I'd say you have a very good case for a bug especially since your "41%" documents are the UPPER bound on accuracy.

It all depends on what the expected or "true" accuracy is in that situation.

If there is a 50% chance to hit that result is spectacularly unlikely.

[akd]

I haven't tried that because it was my assumption that any target that small would not have enough armor protection to make the TacAI feel the need to use APDS.

Hull-down tank. The test cited in this thread was conducted against a 5' x 2' target meant to represent a hull-down Panther.

[Guest]

Here is what I'm not understanding. If you don't know what the intended behavior is, how would you know if the test results showed a bug or intended behavior?

The *data* is known. What we're trying to replicate is known. But that's very different from in-game behavior, which has dozens of influencing variables. So... we do lots of tests to see if, including all of that in-game behavior, what we're expecting is actually happening. If not we try and control some of those behaviors to reduce variables - so far I haven't seen anything to indicate that there *is* a problem, let alone what variables would need to be controlled.

A good example would be the "tanks bog down on dirt roads" bug. It was only after extensive testing that a common thread appeared: ditches *always* seemed to be involved when bogging rates were VERY high. Once we had that, it was a pretty quick fix. But we needed a lot of data for that to happen, mostly to cancel out variables, and the first bunch of pages of that thread where people accused us of not caring about the bogging problem didn't help at all.

There are heaps of variables we have to deal with. Narrowing down what is a problem, and where that problem is actually happening, is a huge undertaking and your taking the first steps really, really helps us.

TBut if I see no change at all, even in the limited test. I would say. Your efforts might be showing something there that we might need to address in the game. But I need you to do more testing to help show that there is indeed a flaw in what we would like to see the game preform at. Giving guidance as to what BF would expect the variance might be. Getting help for free, takes working with people. Treating their efforts as nothing is not going to lead to much positive help.

I stated what I needed to see. Precisely what I needed to see. Telling people their sample size needs to be larger is not "treating their efforts as nothing". I took great pains to say "thank you" for the testing, very clearly. Of course we appreciate people doing testing so that we can spend dev time on stuff we're pretty sure are bugs.

Seeing no change at all across 40 shots is nothing. Like I pointed out, that could be down to methodology, or the fact that a KT at 500 meters is actually a pretty honking big target... I told them what test needed to be done.

If you guys don't want to do this testing, that's okay. There's always a chance that c3k will come along and take up the cause, or that one of us will become convinced there's a problem and do it ourselves. The chances of that go way UP when at least *some* substantial testing has been done and common threads have appeared. I love to fix problems in the game when they appear. But we have to find and confirm them first.

[StellarRat]

It all depends on what the expected or "true" accuracy is in that situation.

If there is a 50% chance to hit that result is spectacularly unlikely.

Couldn't we reasonably expect that 14% hits @ 1000m are the best case scenario? Like I said, I think this all comes down to target size. A Tiger II is far larger than 6' x 6'. So, the only way to really do a good test is fire at similar sized target (6' x 6') as the test range or try to compensate by moving the Tiger back so it has the same relative cross section, but as Jon pointed out that isn't the best option. One thing I'm pretty sure of is that if the accuracy remains above 14% @ 1000m there is VERY likely a problem. I imagine that nearly every other factor that BF uses LOWERS accuracy. Even super crew is unlikely to fire better under battle conditons of any kind than a gun at test range against a stationary target with 23 rounds to work with.

[Vanir Ausf B]

The *data* is known. What we're trying to replicate is known. But that's very different from in-game behavior, which has dozens of influencing variables. So... we do lots of tests to see if, including all of that in-game behavior, what we're expecting is actually happening.

Yes, but what are you expecting? We don't know if CMBN even attempts to model ammo dud rates. There are many things that happened in the Real War than CM does not try to replicate for various and sundry reason, most of them very legitimate reasons. I have no motivation to spend a huge chunk of time proving that there is no real difference in-game accuracy between AP and APDS when it is entirely possible we will hear back that this is WAD.

[slysniper]

I haven't tried that because it was my assumption that any target that small would not have enough armor protection to make the TacAI feel the need to use APDS.

My first attempt at testing was at 1000m. But I abandoned that effort early on after only 4 of the first 20 first shots were APDS. It seems that when the first round hit % is low the AI will most often use AP for a ranging round and then switch to APDS. That's smart, but it makes testing APDS at that range prohibitive.

In my test, where the shermans are firing away, most rounds were bouncing off at a 1000, thus when I was still getting penetrations once in awhile I started to wonder if it was APdS rounds making the kills. I likely could find that out to verify it for myself, but I will not be doing 500 test samples of it:)