Which DM do you like more - before 3.008 or after?

[unreasonable]

4 hours ago, Panthera said:

 

Once the again the US report really is of little use to us in this case as it only concerns itself with the 1 hit kill probability from one direction, angle & range, and crucially provides no details on the effects or locations of any hits at all. Thus when we record incidents of 3cm shells hitting specific locations on an airframe ingame the US analysis gives us absolutely no reference point with which to compare. Only the British tests in the form of photos, film & descriptions detailing the results of every hit to specific locations from various directions & angles actually provides us with a useable point of reference against which we can compare ingame incidents of hits to the same locations.

 

If you want to do a direct comparison with the US  analysis you're going to have to find a way in which you can perform randomly placed hits from 30 deg above and behind a P-47 or B-25 at a range of 500 yards and repeat this 45 times ingame. If you can manage this then you can say wether the 1 hit kill probability from a single direction, angle and range ingame matches the US estimates or not. I doubt that is a test anyone is likely going to be able to conduct ingame, and even if they do it would be of little value as it's just a probability figure for single scenario, i.e. a MK108 rear attack from 500 yards away at a 30 deg downward angle.

 

More misinformation about the US tests. Actually the P-47 test was front - below, not rear above.  And of course it does provide details on the location and effects of hits: if it did not it would be impossible to generate the probability of a hit generating a kill due to the engine, fuel etc. The most important single piece of output for a DM is the kill probability of a hit. How it generates that result is secondary, but can be tested against the US numbers.   

Front below at 500m is fairly easy to approximate with 20mm and 37mm flak guns, which I have already done. I posted summary results previously. I need to rerun the tests for 3.009 which will take a few hours.  You can test those as often as you like, using entirely objective criteria. From this kind of test you can calculate first hit kill probability, also, if you run it enough times, subsequent hit probabilities to generate the entire distribution and compare with the report's tables as a check for reasonableness.

 

I cannot run this test with the 3cm shell - unless someone finds a way to mod the flak guns - but as I am interested in the realism of the DM as a whole I am not that bothered. The poll is not just about the MK108.   We all know the mechanism of damage for mineshells is an unsatisfactory compromise: this has been discussed since the first appearance of BoS. If, however, the balance between 20mm and 37mm flak shells is broadly right compared to the US tests, that suggests that there is no particular bias against smaller or larger shells as such. 

 

In contrast you have yet to provide any method of direct numerical comparison to either the UK tests or the German numbers. Since you express so much doubt about the US report's numbers, I have ask you (again) for your expected kill probabilities for the P-47, for initial and subsequent hits, for any angle you wish to chose, for at least two shells. If you will not say what you think those should be, how an you make comparisons with the game? 

 

[BubiHUN]

Opinion?
The DM just as accurate as the MP.

Edited December 24, 2018 by -[HRAF]BubiHUN

[Blackhawk_FR]

I think it's also way to early to make a poll about that. We have the new DMs since only few days. 

[sevenless]

26 minutes ago, F/JG300_Faucon said:

I think it's also way to early to make a poll about that. We have the new DMs since only few days. 

 

True, however we already have results from both SP and MP which make it clear that in specific areas the in-game results of the new DM are nothing else than pure fantasy and nowhere near historical reality. So it is clear that either the DM, or the weapons are in desperate need of rework.

 

I recently came across a publication by Steven Zaloga which hints to further US studies addressing weapon effectivity:

 

Quote

The FW-190A-8 had a pair of 20mm MG 151s in the wing root and a pair of the more powerful 30mm MK 108s in the wing. Allied intelligence estimated that the 20mm round had a 6 percent probability of inflicting a kill on a heavy bomber, but that the 30mm round, because of its larger high-explosive fill, had a 25 percent probability.

Zaloga, Steven J. Operation Pointblank 1944

 

He draws heavily on the following US archival studies (most of them immediately post war) I have no access to:

 

US Strategic Bombing Survey reports The Effects of Strategic Bombing on the German War Economy (No. 3, Office of the Chairman)

Aircraft Division Industry Report (No. 4, Aircraft Division)

The Defeat of the German Air Force (No. 59, Military Analysis Division)

US Air Force historical studies Boylan, Bernard, Development of the Long-Range Escort Fighter (No. 136, 1955)

Ferguson, Arthur, The Early Operations of the Eighth Air Force and the Origins of the Combined Bomber Offensive, 17 August 1942 to 10 June 1943 (No. 118, 1946)

Grabmann, Walter, German Air Force Air Defense Operations (No. 164, 1956)

Heinrichs, Waldo, A History of the VIII USAAF Fighter Command (n.d.)

Kammhuber, Josef, Problems in the Conduct of a Day and Night Defensive Air War (No. 179, 1953)

Nielsen, Andreas L. and Grabmann, Walter, Anglo-American Techniques of Strategic Warfare in the Air (No. 183, 1957)

Norris, Joe, The Combined Bomber Offensive: 1 January to 6 June 1944 (No. 122, 1947)

Ramsey, John, The War Against the Luftwaffe: USAAF Counter-Air Operations April 1943–June 1944 (No. 110, 1945)

Renz, Otto von, The Development of German Antiaircraft Weapons and Equipment of All Types up to 1945 (No. 194, 1958)

Schmid, Josef, The German Air Force versus the Allies in the West: German Air Defense (No. 159, 1954)

Schmid, Josef and Grabmann, Walter, The German Air Force versus the Allies in the West: The Air War in the West (No. 158, 1954)

Suchenwirth, Richard, Command and Leadership in the German Air Force (No. 174, 1969)

Stormont, John, The Combined Bomber Offensive, April thru December 1943 (No. 119, 1946)

 

[Operatsiya_Ivy]

41 minutes ago, sevenless said:

the new DM are nothing else than pure fantasy and nowhere near historical reality.

 

Well, that was sadly always the case. Nobody knows how they come up with their values. In theory? yes. In practice? no.

Edited December 24, 2018 by Operation_Ivy

[Panthera]

14 hours ago, unreasonable said:

 

More misinformation about the US tests. Actually the P-47 test was front - below, not rear above.  And of course it does provide details on the location and effects of hits: if it did not it would be impossible to generate the probability of a hit generating a kill due to the engine, fuel etc. The most important single piece of output for a DM is the kill probability of a hit. How it generates that result is secondary, but can be tested against the US numbers.   

Front below at 500m is fairly easy to approximate with 20mm and 37mm flak guns, which I have already done. I posted summary results previously. I need to rerun the tests for 3.009 which will take a few hours.  You can test those as often as you like, using entirely objective criteria. From this kind of test you can calculate first hit kill probability, also, if you run it enough times, subsequent hit probabilities to generate the entire distribution and compare with the report's tables as a check for reasonableness.

 

I cannot run this test with the 3cm shell - unless someone finds a way to mod the flak guns - but as I am interested in the realism of the DM as a whole I am not that bothered. The poll is not just about the MK108.   We all know the mechanism of damage for mineshells is an unsatisfactory compromise: this has been discussed since the first appearance of BoS. If, however, the balance between 20mm and 37mm flak shells is broadly right compared to the US tests, that suggests that there is no particular bias against smaller or larger shells as such. 

 

In contrast you have yet to provide any method of direct numerical comparison to either the UK tests or the German numbers. Since you express so much doubt about the US report's numbers, I have ask you (again) for your expected kill probabilities for the P-47, for initial and subsequent hits, for any angle you wish to chose, for at least two shells. If you will not say what you think those should be, how an you make comparisons with the game? 

 

 

Misinformation? No, like I said it's a 1 hit kill probability estimate for an attack from a single direction, angle and range. That it was from the front and below in the case of the P-47 just makes it even worse as that's even harder to replicate ingame, and even then it still gives no points of reference on what to expect with for example a wing hit vs a fuselage hit. Also it should be added that from the front & below the P-47 would be a lot less likely to get shot down due to that massive engine block infront shielding everything behind it, in other words it's the best possible attack scenario for the P-47. Had the P-47 instead been subjected to attacks from the rear such as the B-25, the 1 hit kill probability would've gone up due to a more vulnerable engine & pilot. The engine alone went from a slim to no chance of being knocked out with a single hit from the front, to basically a 50/50 from the rear. 

 

The only data provided in the US report which us truly useful for our purposes of comparing damage ingame is the knock out probability estimate for direct hits to the engine from 4 different directions, but it's really the only thing from said analysis which is specific enough that we can use it  - even if once again it's a probability estimate which doesn't specify exactly what damage was done with each hit, which is somewhat important when we're talking about an engine that can run all the way home despite having 3 or more cylinders blown off. 

 

By comparison the British tests gives us the detailed results of hits to specific locations via photos, film & damage description, making it easy for us to compare it with ingame results of hits to the same areas and assess wether the damage done is realistic or not - we simply can't do that with the US data.

 

A good example is the ingame photo of the P-47 being hit directly on the rear fuselage from behind at a 10-20 deg angle, not once, not twice but four times without anything coming off. We know from British testing vs the Spitfire & Blenheim that the chances of this happening vs a semi monoque structure of that size would've been very slim, even when we consider that the P-47's rear is likely more sturdy, as even half the damage done to the Spitfire or Blenheim would've been lethal.

Edited December 24, 2018 by Panthera

[unreasonable]

1 hour ago, Panthera said:

Misinformation? No, like I said it's a 1 hit kill probability estimate for an attack from a single direction, angle and range. That it was from the front and below in the case of the P-47 just makes it even worse as that's even harder to replicate ingame, and even then it still gives no points of reference on what to expect with for example a wing hit vs a fuselage hit.

 

No it does not make it worse: it makes it almost trivially easy, for the conventional 20mm and 37mm rounds. 

 

I am in the middle of doing the 37mm right now: and the first shot kill probability looks to be much higher than the US report's estimates, provisionally because the engine/fuel system is more frequently damaged.

 

If the conventional 20mm and 37mm are roughly right, what we are left with is the mineshell issue of how damage is caused, of which we are all aware. But the DM is not only about the Mk108. 

 

edit: I think we have a fundamental disagreement on how the DM should be judged. We all know that with a limitation in graphics and detailing of components the ability of the DM to represent on screen what you see in photographs is also limited. TBH, this does not interest me: what does, is the overall effectiveness of various weapons/shells in destroying aircraft. To assess that you have to quantify, in particular one shot kill probabilities. If the overall probability is reasonable, that is a good first step. The second step is whether the systems that are being damaged in the game correspond to those in RL: and the third is the graphical representation.

 

The game currently cannot show gaping holes in wings: we all know that. For me that is purely a graphical limitation, not one of the DM. 

 

@sevenless You quoted: "The FW-190A-8 had a pair of 20mm MG 151s in the wing root and a pair of the more powerful 30mm MK 108s in the wing. Allied intelligence estimated that the 20mm round had a 6 percent probability of inflicting a kill on a heavy bomber, but that the 30mm round, because of its larger high-explosive fill, had a 25 percent probability."

 

At least that gives an alternative number, so that is something to work on.

 

It may well be the case that these were the allied estimates during the war. However, at that time the allies would know the number of times the surviving planes had been hit, but they would have no clear idea of the number of times the downed planes had been hit.  A detailed 1947 US OR research, with the benefit of actually having German 3cm rounds to fire, came up with a maximum number of 16% for a medium bomber from the rear/above position.  You cannot just wish that fact away; at the very least you need a convincing explanation of why a war time estimate based on partial operational data is better than a test estimate.  

 

 

 

 

 

 

Edited December 24, 2018 by unreasonable

[sevenless]

3 minutes ago, unreasonable said:

You cannot just wish that fact away; at the very least you need a convincing explanation of why a war time estimate based on partial operational data is better than a test estimate.  

 

 

Just to avoid a misunderstanding: It is not my intention to wish away anything. I go by the figures that are published and documented. It is the job of the devs to align all this "hardfacts" and construct a model which delivers in-game results which reflect this. I am not in the business of using a cristal ball and coming up with some fancy model which in the worst case ignores 80% of documented evidence.

 

How the devs reach their results, I don´t care. The end result in-game has to be in-line and consistent with the real life results. I guess most folks think like that.

[unreasonable]

3 minutes ago, sevenless said:

 

Just to avoid a misunderstanding: It is not my intention to wish away anything. I go by the figures that are published and documented. It is the job of the devs to align all this "hardfacts" and construct a model which delivers in-game results which reflect this. I am not in the business of using a cristal ball and coming up with some fancy model which in the worst case ignores 80% of documented evidence.

 

How the devs reach their results, I don´t care. The end result in-game has to be in-line and consistent with the real life results. I guess most folks think like that.

 

My point is that these estimates are not - indeed cannot have been - documented with equal thoroughness, yet the most thorough documentation is the one being downplayed. 

 

For all I know all of the sources you have listed quote exactly the same primary source.  

 

 

 

[sevenless]

45 minutes ago, unreasonable said:

For all I know all of the sources you have listed quote exactly the same primary source.  

 

Then you know more than the authors I quoted.

 

Edited December 24, 2018 by sevenless

[Sgt_Joch]

On 12/21/2018 at 1:01 PM, II/JG17_HerrMurf said:

The new DM is considerably better. I can take most fighters in one or two two second bursts as opposed to one or two 1.5 second bursts in the old version. Considerably fewer wing offs as well.

 

I agree. Flying QMBs with 30 mm equipped 109s, I can usually down AI Yaks and La-5s with just one burst. 

[Panthera]

9 hours ago, unreasonable said:

The game currently cannot show gaping holes in wings: we all know that. For me that is purely a graphical limitation, not one of the DM. 

 

I don't really care about how big holes are painted on the aircraft, what I do care about however is the fashion in which aircraft in general are brought down ingame, which atm overwhelmingly is by damage to the powerplant - which is esp. true in the case of the MK108, which just wasn't the way this weapon usually brought down aircraft in real life. In other words it's not about how effective the weapon is at eventually bringing down the target which is the issue (here esp. the MK108 is still very effective),  it's the type of damage inflicted and HOW the target is brought down which is the issue, because here things just aren't remotely realistic atm - atleast not in the case of HE(M) shells.

 

It really has become an eye sore for me that certain aircraft can take multiple big caliber HE hits to their wings or fuselages and still keep flying, despite the fact that we know hits to these specific places would've been lethal straight away in most cases, and if not then at the very least prevented any further aggressive maneuvering from being possible.  

 

So whilst you might be able to achieve a probability percentage ingame for single hit "A" & "B" kills that closely matches the US probability est. from that single direction, angle & range, the actual cause of those kills will be entirely different, with the far majority in the US estimate being due to structure, whilst ingame it will almost always be due to engine damage - and due to how fast a damaged engine fails ingame it is undoubtedly going to lead to a much higher percentage in both categories..

 

Anyway I'm just repeating what I've been saying all along at this point (the above being pretty much identical to the below post anyway), so I think I'm just going to give this subject a rest now until the next patch, and then we will see.

 

 

 

 

Edited December 25, 2018 by Panthera

[unreasonable]

9 hours ago, sevenless said:

 

Then you know more than the authors I quoted.

 

 

Nice charts: although it is always better to say where a posted extract comes from. 

 

The second one is irrelevant since it mixes shell effect, number of guns and rate of fire. We are here only interested in the first.

 

The last one defines "fire-power" as being based on a "destruction coefficient" presumably also used in the second.  I expect this a rating of the total KE+CE of each munition, since the ratio 6%/25% as per chart one is very close to the ratio of the calculated values. But perhaps not: however, if there is any empirical backing for these numbers in terms of aircraft actually shot down these charts do not say what it is.

 

 

[JtD]

It's based on the same German estimation, 18 hits for 20mm, 4 hits for 30mm. All from the gun cam evaluation document I kept mentioning, which is still available from the old deutscheluftwaffe site:

 

http://www.deutscheluftwaffe.com/archiv/Dokumente/ABC/a/Alliierte/US Air Combat Records WW II.pdf

[Ala13_pienoir]

we all have our vices
Some videos, others flight models, others damage model
Others are rigorously historical

I do not know anything about that, take your plane and get out there.

If you are knocked down, you were worse than the other, it's that simple

I know people who squeeze the performance of the plane and know where and from where it can not go

Sometimes download all the charger on the target and the bad one keeps flying
Other times you shoot a burst at 500m and knock it down !!

It's war!

[unreasonable]

@JtD Thanks for that,  I can see where your comment about ranks came from.     I do not see figures of kills per hit, except for the footnote repeating the German figure from the report already posted. Lots about rounds per pass and hits per pass, mostly not distinguishing between calibres of guns fired, and stating that the latter is hard or impossible to ascertain.

 

I am also not sure how WW2 intelligence estimates would have based their figures on German contemporary reports, unless they were Ultra sure.  

 

This report is about starting fires - which the GAF seems to have used as criterion for awarding a kill (in addition to rank, as you pointed out). Nothing to do with structural damage as such.  Table 31 gives p for fire/shot (not hit) based on range, but we can take the zero range number as being 100% hits: this gives a fire/hit probability of 0.02 for 20mm round vs B-17.

 

That compares to 0.34-0.53 for the 20mm  B kill due to fuel on the B-25 in the US report. That is not wildly inconsistent with an average of 18 20m hits on average, but it does not have anything specifically to do with 30mm hits.

 

 

Edited December 25, 2018 by unreasonable

[sevenless]

6 hours ago, unreasonable said:

 

Nice charts: although it is always better to say where a posted extract comes from.

 

As mentioned above. One of Zalogas primary sources:

 

MILITARY ANALYSIS DIVISION

59. The Defeat of the German Air Force, 1945 and 1947 (both printings are available). 44 p. 

LC CALL NUMBER: D785.U6 No. 59 and No. 59A
Also LC CALL NUMBER: D785.U57 Vol. III

Edited December 25, 2018 by sevenless

[Mac_Messer]

On 12/24/2018 at 6:14 PM, unreasonable said:

If the conventional 20mm and 37mm are roughly right, what we are left with is the mineshell issue of how damage is caused, of which we are all aware. But the DM is not only about the Mk108. 

 

edit: I think we have a fundamental disagreement on how the DM should be judged. We all know that with a limitation in graphics and detailing of components the ability of the DM to represent on screen what you see in photographs is also limited. TBH, this does not interest me: what does, is the overall effectiveness of various weapons/shells in destroying aircraft. To assess that you have to quantify, in particular one shot kill probabilities. If the overall probability is reasonable, that is a good first step. The second step is whether the systems that are being damaged in the game correspond to those in RL: and the third is the graphical representation.

 

The game currently cannot show gaping holes in wings: we all know that. For me that is purely a graphical limitation, not one of the DM.

 

A broad overstatement. Your findings are not anywhere accurate without MG damage ingame. And how MG damage acts now ingame is what MK108 shrapnel acts like. You can cause instant fire to any engine, P47 included, by striking its engine compartment with one shell even from 300m which is what I did with most machinegun ammo available in the game.  Not realistic at all. So right now most times if you get a MK108 kill, it depends on whether its shrapnell hits something vital, fueltank fire and pilot excluded. Don`t forget the minenshell features also on 20mm German weapons so that is even worse.

 

MK108 / MG 151/20 damage mechanics makes no sense since a player can cause the same damage with machinegun fire, available with few hundred rounds each. The ShVak also suffers from much reduced effectiveness aswell.

Edited December 25, 2018 by Mac_Messer

[unreasonable]

11 minutes ago, Mac_Messer said:

 

A broad overstatement. Your findings are not anywhere accurate without MG damage ingame. And how MG damage acts now ingame is what MK108 shrapnel acts like. You can cause instant fire to any engine, P47 included, by striking its engine compartment with one shell even from 300m which is what I did with most machinegun ammo available in the game.  Not realistic at all. So right now most times if you get a MK108 kill, it depends on whether its shrapnell hits something vital, fueltank fire and pilot excluded. Don`t forget the minenshell features also on 20mm German weapons so that is even worse.

 

MK108 / MG 151/20 damage mechanics makes no sense since a player can cause the same damage with machinegun fire, available with few hundred rounds each. The ShVak also suffers from much reduced effectiveness aswell.

 

I am not quite sure what you mean about "my findings": I do not recall claiming that the current DM is accurate, although I do not miss the frequent wing shedding of the prior versions.  The only finding I have posted (provisionally, since I have not yet finished enough runs) is that the kill/hit ratio for conventional 20mm and 37mm vs P-47s appear to be significantly higher than predicted by the US report when I approximate the report's firing conditions.  

 

My criticisms of some other peoples' posts critical of the 3.009 DM are of their methods and arguments in support of some of the more extreme claims being made about the 3cm mineshell.  Reading the US analysis of the German camera film posted by JtD further reinforces my skepticism.   We all know that it was an effective shell: the problem is in quantifying that degree of effectiveness.  When I  disagree with Panthera's claims about the uselessness of the US tests, or with sevenless about the trustworthiness of the estimates in the German report, is about the quality of the evidence, that is all.  

 

 

 

 

 

 

[Ala13_pienoir]

How about a picture of "I do not know, I do not answer"

[Panthera]

Just want to make clear that the US analysis is only useless in our case as we're trying to figure out what effects hits to specific places might have, and since the US analysis provides no data on this it's quite natural that we simply can't use it. Not sure how anyone can disagree with that. If we were trying to figure out the probability of an aircraft going down to a single hit irrespective of shot placement from a single direction, angle & range, then US data would be our only useful source, but that has just never been the goal as it doesn't help explain how the ingame dmg model currently is in error - esp. in terms of how easy it should be to sever a wing or tail. 

 

Hence the only data available which is useful to us atm therefore is recorded test results against specific areas on airframes, and the British MK108 trials fullfill this requirement. In addition to this we do have German test results as well, which unsurprisingly were identical.

 

Finally let's remember that only the German & British examinations take into account the effect of hits by all three types of shells fired by the MK108, the HE, HE-T and INC, the std. MK108 belt load out scheme, whilst the US analysis only concerns itself with the HE shell.

 

 

Edited December 25, 2018 by Panthera

[JtD]

If I was concerned with the graphical presentation of the damage, then I'd consider the US study useless. But as graphics don't bother me, and as all I can test in game is if a plane goes down or not, it is tremendously useful.

 

I'd like to point out again - the US study very well gives results for individual hits to specific areas. I don't quite understand what the importance of this may be, but it's still there.

 

As a side note, the US tested incendiary rounds as well. These results are just not in that report.

[Panthera]

5 hours ago, JtD said:

If I was concerned with the graphical presentation of the damage, then I'd consider the US study useless. But as graphics don't bother me, and as all I can test in game is if a plane goes down or not, it is tremendously useful.

 

The problem once again is that the US study only applies to a single scenario, so how can be it useful for determining wether for example a hit to the wing ingame deals a realistic type & amount of damage?

 

For me it's about how the plane goes down, not so much wether it eventually will go down. Thus the fact that US study uses two criterias for a kill based on time (4 min for an "A" kill), presents a big problem when trying to compare it with ingame shoot downs as it leaves a lot of time for the engine to sieze up ingame (which it does almost invariably quite soon after a MK108 hit almost anywhere), leaving us with ingame results that will most often fall into the "A" category via engine damage alone, which will completely skew the results and raise the probability of an "A" kill ingame significantly above the US estimate - even if we were to go to great lengths of ensuring that all attacks were from the same direction, angle & range for which the US estimate applies.

 

5 hours ago, JtD said:

I'd like to point out again - the US study very well gives results for individual hits to specific areas. I don't quite understand what the importance of this may be, but it's still there.

 

Where? There are no details on damage done by individual hits to the wings or rear fuselage for example, let alone specific areas on these parts.

 

Only the estimates based on direct hits to the engine are truly useful, as here we are atleast talking about hitting a specific part from four different angles. (That's being a lot more specific than a probability estimate vs an entire aircraft from just one direction, angle & range) But even so we are still left without details as to the nature of the damage. But being the only data we have on engine resilience to fire, it's our best source for this.

Edited December 26, 2018 by Panthera

[unreasonable]

5 hours ago, Panthera said:

 

The problem once again is that the US study only applies to a single scenario, so how can be it useful for determining wether for example a hit to the wing ingame deals a realistic type & amount of damage?

 

 

In terms of the amount of damage, the most important fact about a wing hit is simply whether or not the plane is lost as a result. If wing hits cause about the same p of a kill as the estimates, that supports the DM. So you test the DM against that scenario. 

 

5 hours ago, Panthera said:

 

Where? There are no details on damage done by individual hits to the wings or rear fuselage for example, let alone specific areas on these parts.

 

 

Appendix F, Tables F1 and F2 give number of hits and assessed damage (A and B probabilities) by zone for the B25. Table F 7 gives hits on critical components. Appendix G shows the zones: 27 of them. Table H1 shows the projected area by zone.  The is plenty in here about individual hits.   The crucial point that you seem to be objecting to is that the assessment of their implications for the plane has been done by experts, and not by you.  Others may not see this as a problem.

 

5 hours ago, Panthera said:

For me it's about how the plane goes down, not so much wether it eventually will go down. Thus the fact that US study uses two criterias for a kill based on time (4 min for an "A" kill), presents a big problem when trying to compare it with ingame shoot downs as it leaves a lot of time for the engine to sieze up ingame (which it does almost invariably quite soon after a MK108 hit almost anywhere), leaving us with ingame results that will most often fall into the "A" category via engine damage alone, which will completely skew the results and raise the probability of an "A" kill ingame significantly above the US estimate - even if we were to go to great lengths of ensuring that all attacks were from the same direction, angle & range for which the US estimate applies.

 

If you really want to test only hits on the wings and ignore any possibility that over fragile engines are skewing overall results (which I agree looks likely), you can do this. You have to throw away any test result when the engine/fuel tank is damaged, which may be a large proportion depending on the angle of fire. You then have a number of hits on structure of which a certain % cause A or B kills. This could be time consuming, so I certainly do not propose to do it, but not impossible.

 

7 hours ago, Panthera said:

Just want to make clear that the US analysis is only useless in our case as we're trying to figure out what effects hits to specific places might have, and since the US analysis provides no data on this it's quite natural that we simply can't use it. Not sure how anyone can disagree with that.

 

 

They can disagree because your claim is false. See Appendix F as noted above.

 

7 hours ago, Panthera said:

 

Hence the only data available which is useful to us atm therefore is recorded test results against specific areas on airframes, and the British MK108 trials fullfill this requirement. In addition to this we do have German test results as well, which unsurprisingly were identical.

 

 

The only quantitative information we have from "German tests" is a summary of what is apparently mean hits to kill estimated from gun cameras and some testing. The post war US report on GAF gun camera films posted above by JtD that says counting hits is nigh on impossible; assessing the number of kills is also difficult and possibly arbitrary. All one can realistically do is model a probable number of hits given tracking, time and rate of fire, and count visible fires.  So a mean hit/kill number derived from these films is questionable. In the case where the analyst does this for a 20mm hit, he ends up with a single hit probability to cause a fire lower than the B kill due to fuel hits number in the US Weapons report.  I have not seen any hit/kill numbers in the UK reports.

 

Really, if I were wanting to criticize the DM I would use the US report as a base, not dismiss it for all these spurious reasons.  If you make random hits from a similar angle to one of the US tests and get far more engine damage results than the p given in the tables, you have evidence that there is something wrong with the DM, not only in the mix of the results but potentially in the overall hit/kill ratio too. If, as a result of this, the DM gets changed to reduce splinter damage against engines, then it will be much easier to test for structural damage. 

 

  

 

 

Edited December 26, 2018 by unreasonable

[Panthera]

2 hours ago, unreasonable said:

Appendix F, Tables F1 and F2 give number of hits and assessed damage (A and B probabilities) by zone for the B25. Table F 7 gives hits on critical components. Appendix G shows the zones: 27 of them. Table H1 shows the projected area by zone.  The is plenty in here about individual hits.   The crucial point that you seem to be objecting to is that the assessment of their implications for the plane has been done by experts, and not by you.  Others may not see this as a problem.

 

Back to being cheeky I see, and sneaky too by switching to the B-25 when we're talking about the P-47.

 

Still however you didn't manage to show what you claim, as 1) "K", "A" & "B" kill probabilities is not a detailed description of individual hits, there being no description of the damage at all for example and 2) where is the data for the P-47 we're actually discussing?

 

So no I'm not objecting to the US findings, like I said I trust in their figures, but I do object to your notion that the US analysis is our best source for assessing wether the damage dealt to specific parts of aircraft (specifically fighters atm) ingame is realistic or not. Only the British firing trials (and to some extend the German ones), provide data we can actually use for direct comparison as we know the exact details behind every single hit here. 

 

2 hours ago, unreasonable said:

They can disagree because your claim is false. See Appendix F as noted above.

 

You certainly seem to believe that, but once again there's none of that detailed information on damage you claimed is there:

 

F1 shows "K", "A" & "B" kill for hits to the B25 fuselage zones 1-11 from the rear.only (no detailed description of damage, just wether "K", "A" & "B" probabilites were met)

F2 shows "K", "A" & "B" kill for hits to the B25 wing zones 1-11 from the front only (no detailed description of damage, just wether "K", "A" & "B" probabilites were met)

F7 shows hits on components pr. thousand hits on the B-25 ("of 1,000 random hits anywhere you will damage x component x number of times", not exactly useful for our purpose now is it?)

 

So in summary, how exactly are we supposed to use the information in the US analysis when we're trying to assess wether hits to a Spitfire or P-47's wing such as in the cases below should've severed the wing (or atleast made subsequent maneuvering lethal) or not? I'm very intrigued as to how you will prove the US data as being the single best source for figuring this out.

 

Quote

2 hits (right on top of each other due to RoF) to the right wing from directly 6'o'clock, result; engine damaged & right aileron lost. Aircraft still controllable.

 

 

2 hits to the left wing from direct 6'o'clock, result; engine & turbocharger severely damaged. No loss of control to aircraft.

 

4 hits to the right wing from directly 6'o'clock, result; 1st hit did virtually nothing, 2nd hit the engine caught fire immediately, 3rd hit and flap is lost, 4th hit nothing noticable, but probably just another tick to the structure healthbar:

 

2 hits to right wing, with so far the most realistic results: Hit nr.1 to middle of wing did visible damage to wing & stabilizer, but no loss of control (Odd). 2nd hit directly to wing root severed the wing in level flight.

 

1 hit to right wing, results: Damage to cooling system, very little to effect on controllability, little structural damage as evident during subsequent maneuvers. Aircraft still able to dogfight. Takes a high speed & high G pull out to sever the wing:

 

 

Edited December 26, 2018 by Panthera

[unreasonable]

13 minutes ago, Panthera said:

 

So in summary, how exactly are we supposed to use the information in the US analysis when we're trying to assess wether hits to a Spitfire or P-47's wing such as in the cases below should've severed the wing or not? I'm very intrigued as to how you will prove the US data as being the single best source for figuring this out.

 

 

As to being cheeky: how is a general average for shooting down a B-17 supposed to bear on the results of hitting a P-47?  You are hardly consistent in the data you push forwards to support your arguments. This is also a straw man: you may be  "trying to assess wether hits to a Spitfire or P-47's wing such as in the cases below should've severed the wing or not?", I am trying to assess the probability of a hit downing a plane.

 

The reality is that we do not need the details: we need the results, generated by people far more competent that you or I in assessing real damage, presented in a form that allows statistical comparison.  Single shot comparison is simply not meaningful in a DM that uses an RNG to make up for the lack of having a huge number of possible hit locations and angles, subject to several damage mechanisms.  We might wish for a DM that models on a much more detailed level, always assuming that our PCs could actually run it, but we do not have it, so all we can go on is average results.

 

The US report gives you the p of A and B kills from hits, due to damage to the 4 types of component.  You run tests of random hits and assess the A or B kill probability, with the cause. You then compare the statistics generated by the test with those given by the report.  While our game test cannot replicate exactly the US test, and our ability to assess damage in game is cruder than that of the US assessors, we can make a very good approximation.  If the results are very different that suggests a problem to be investigated.

 

I have now run 40 P-47s being hit by 37mm from front and below, currently writing up the results. Probably publish them in a new thread in a day or so.

 

 

[=RS=Stix_09]

Based on the original question from the developer (and not going off topic)

I think both damage models had there problems , before and after update 3.008. All this argument is going in circles. I did not like either myself (something in the middle maybe).

The previous model with plane specific tweaks maybe...

 

DM is highly subjective topic, and no one is ever going to be satisfied with anything the developers do in this regard. Same issue in on agreement in the past , nothing has changed.

 

I think people have lost sight of the fact this is a game and its not 100% historical enactment of the actual events then occurred in ww2 battles covered.

(and i'm not going to even go into where it falls short on that topic, that's another whole thread, and series of circular arguments)

 

It never will be , even historians that specialise in this period of history don't know 100% or agree 100%, and much is still open to interpretation, why would armchair historians be any better?

People can't even agree on the reference point to measure against, or how to measure stuff to make a comparison against that reference , How can you, the reference point itself is subjective.

 

All you can hope for something that works and is playable and fun in a game perspective, and seems somewhat plausible by the vast majority that play.

 

I think the problem is not so much the damage model , but more plane specific damage oddities and things like AI turret gunnery @crazy angles doing impossible snipes in moving plane that no human could possibly react to aim , are more of a problem than damage model is currently.

 

Things that you can capture and maybe do something about, especially when they stands out like dogs balls as silly in a game perspective.

Edited December 26, 2018 by =RS=Stix_09

[Panthera]

13 minutes ago, unreasonable said:

As to being cheeky: how is a general average for shooting down a B-17 supposed to bear on the results of hitting a P-47?

 

If you go back and read when that was initially mentioned it was to illustrate how absurd the notion of a P-47 surviving 10 MK108 hits is if on average it took 3-4 to down a heavy bomber. 

 

17 minutes ago, unreasonable said:

The reality is that we do not need the details: we need the results, generated by people far more competent that you or I in assessing real damage, presented in a form that allows statistical comparison.  Single shot comparison is simply not meaningful in a DM that uses an RNG to make up for the lack of having a huge number of possible hit locations and angles, subject to several damage mechanisms.  We might wish for a DM that models on a much more detailed level, always assuming that our PCs could actually run it, but we do not have it, so all we can go on is average results.

 

The details are the results (loss by fire, structural failure or loss of control etc.) and when you got these all along the wing span at several several angles & directions, then Id' say you got a very good point of reference with which to compare with ingame results with hits to the same part & area.

 

As for the DM using RNG, I would think it's actually using a combination of a simple handgrenade esque spherical fragment spreading simulation, which would be great for simulating random shrapnel damage (eventhough the engine damaging potential of the fragments has definitely been cranked up to 11 ingame), whilst for local blast damage they're probably using a structural HP system where if the HP is depleted said structural part comes off. So to me it merely appears as though what'ever inflicted dmg figures they're using in the case of the latter are too low for certain shells atm, and thus it should be relatively easy to fix so that the damage and cause of the kills are more realistic. 

 

46 minutes ago, unreasonable said:

The US report gives you the p of A and B kills from hits, due to damage to the 4 types of component.  You run tests of random hits and assess the A or B kill probability, with the cause. You then compare the statistics generated by the test with those given by the report.  While our game test cannot replicate exactly the US test, and our ability to assess damage in game is cruder than that of the US assessors, we can make a very good approximation.  If the results are very different that suggests a problem to be investigated.

 

I have now run 40 P-47s being hit by 37mm from front and below, currently writing up the results. Probably publish them in a new thread in a day or so.

 

Problem is that aside from the engine you don't have any component estimates for the P-47, let alone any estimates for when it's being attacked from the rear aspect above or below, or from the side for that matter. You only have an "A" & "B" probability measure for attacks from the front and below, and then once again you don't have details regarding how the "A" & "B" kills in the US estimate were achieved (loss of wing, tail, fire etc..) 

 

My prediction is that in your ingame test you will see the majority of your "A" & "B" losses be due to 1 thing only, engine damage, as that is currently the case when testing the effects of single cannon hits to random places from the rear ingame.

 

[sevenless]

17 hours ago, unreasonable said:

..., or with sevenless about the trustworthiness of the estimates in the German report, is about the quality of the evidence, that is all.  

 

And that is all fine and dandy. My argument is, basically, that it is not up to me to question established primary source material, which has been and still is referenced by a plethora of established and reputable historians, who all know their stuff way better than any hobby armchair historian might ever hope to. That is not to say that those are infallible. But unless proven wrong, they stand as being accepted by the scientific community. Because everyone has to keep in mind that published sources are and have been out there for now decades and I have yet to see one single accepted published work that in earnest calls into question those results. So at the end of the day, for me, it makes very much sense, if the in game results, using whichever DM, stay as close to the published references as possible to remain historically believable.

 

On 12/25/2018 at 6:57 AM, JtD said:

It's based on the same German estimation, 18 hits for 20mm, 4 hits for 30mm. All from the gun cam evaluation document I kept mentioning, which is still available from the old deutscheluftwaffe site:

 

http://www.deutscheluftwaffe.com/archiv/Dokumente/ABC/a/Alliierte/US Air Combat Records WW II.pdf

 

I´m not so sure if that is really the case. The charts I quoted are of a 1945 report by the US Military Analysis Division and part of the study called United States Strategic Bombing Survey dating Sept 30, 1945. The study you quoted is from 1950. If the US had access to the Buehler survey from 9/1944 at the time the US charts were produced in 1945, I am actually unsure about. Furthermore keep in mind that in the 1950s study the Buehler 9/1944 report is only mentioned in passing as a citation.

Edited December 26, 2018 by sevenless

[Operatsiya_Ivy]

17 hours ago, unreasonable said:

we all know that it was an effective shell: the problem is in quantifying that degree of effectiveness.

 

Thats why it makes sense to compare it to other similar rounds/shells. If an arguably weaker shell (if we go by published research papers for example) performs way better compared to other shells, than there need to be questions asked.

 

It might make more sense to look at the individual shells than at the damage model of the aircraft to get a somewhat realistic depiction of it all. I think it is even more important to see that the individual ammunition is accurate (as far as this is possible) in comparison to each other. Something that is very much lacking as of right now.

[unreasonable]

24 minutes ago, sevenless said:

 

And that is all fine and dandy. My argument is, basically, that it is not up to me to question established primary source material, which has been and still is referenced by a plethora of established and reputable historians, who all know their stuff way better than any hobby armchair historian might ever hope to. That is not to say that those are infallible. But unless proven wrong, they stand as being accepted by the scientific community. Because everyone has to keep in mind that published sources are and have been out there for now decades and I have yet to see one single accepted published work that in earnest calls into question those results. So at the end of the day, for me, it makes very much sense, if the in game results, using whichever DM, stay as close to the published references as possible to remain historically believable.

 

I am sorry to say this, sevenless, but since you wish to use the argument from authority, I will throw it back at you: until you at least understand why the distribution of losses/hits is non-normal, you have no business talking about whether a probability analysis is or is not scientific. 

 

We do not need "published references" when we have authentic

primary sources.  The US study JtD posted above very much calls into question any German analysis of  hits/losses based on camera film.  JtD pointed out that the German number of hits estimate was inconsistent with the HE required estimate in the German document posted earlier.  The US OR report gives test based estimates for losses/hit that are somewhat lower than these earlier estimates, that were made in wartime, with less information. There is plenty of credible primary evidence that calls into question "these results". 

 

4 hours ago, Panthera said:

 

 

As for the DM using RNG, I would think it's actually using a combination of a simple handgrenade esque spherical fragment spreading simulation, which would be great for simulating random shrapnel damage (eventhough the engine damaging potential of the fragments has definitely been cranked up to 11 ingame), whilst for local blast damage they're probably using a structural HP system where if the HP is depleted said structural part comes off. So to me it merely appears as though what'ever inflicted dmg figures they're using in the case of the latter are too low for certain shells atm, and thus it should be relatively easy to fix so that the damage and cause of the kills are more realistic. 

 

Problem is that aside from the engine you don't have any component estimates for the P-47, let alone any estimates for when it's being attacked from the rear aspect above or below, or from the side for that matter. You only have an "A" & "B" probability measure for attacks from the front and below, and then once again you don't have details regarding how the "A" & "B" kills in the US estimate were achieved (loss of wing, tail, fire etc..) 

 

My prediction is that in your ingame test you will see the majority of your "A" & "B" losses be due to 1 thing only, engine damage, as that is currently the case when testing the effects of single cannon hits to random places from the rear ingame.

 

 

I do not know if there is a separate blast and splinter estimate in the DM or not: if there were then a tweak to lower splinters and boost blast in line with HE content could well be an improvement.  Modelling blast effects, however, is extremely hard.  

 

We have enough component estimates in the P-47 given how difficult it is to estimate damage in any detail in the game. Kills ps due to engine, fuel cells, pilot and structure give enough texture to assess the damage.  A kills due to fuel cell hits are fires, B kills are leaks. For structure, if the plane falls OOC that is an A kill. etc. It is nothing like as difficult as you are making out to get a useful comparison.

 

As for the mixture of results: that is also my expectation. What is at issue here, apart from the number of kills due to engine/fuel damage, is the number due to structural damage. If the engine/fuel numbers are way too high that suggests some combination of toning down splinter damage or increasing engine/fuel cell damage resistance is in order.  If the number of losses to structural damage is way too low that suggests that either local blast is under-modeled or wings etc are too damage resistant.

 

Having done the test for the 37mm, I will then do it for the 20mm (which will take longer). Then we can see if any biases in outcomes are consistent between the two, or not.    

 

[Operatsiya_Ivy]

9 minutes ago, unreasonable said:

I do not know if there is a separate blast and splinter estimate in the DM or not: if there were then a tweak to lower splinters and boost blast in line with HE content could well be an improvement.  Modelling blast effects, however, is extremely hard.  

 

There are different values in the in-game files of the ammunition for blast, number of fragmentation (shrapnel) and many other factors. 

[sevenless]

11 minutes ago, unreasonable said:

There is plenty of credible primary evidence that calls into question "these results". 

 

Up to you with your cristal ball to wander around in the world.

[Sgt_Joch]

6 hours ago, Panthera said:

 

If you go back and read when that was initially mentioned it was to illustrate how absurd the notion of a P-47 surviving 10 MK108 hits is if on average it took 3-4 to down a heavy bomber. 

 

 

Well no, you are misreading the data. There is no proof that it took an average of 4 30 mm hits to kill a B-17, that is purely a German estimate based on apparently studying gun camera footage.

 

The estimate in the US study is actually that a 30 mm shell hit to a P47 has a 28% chance of resulting in a kill, so yes, it could take 3-4 30 mm hits to kill a P47 based on that estimate.

 

There is no proof that one estimate is more accurate than the other, although the US report is based on actual tests.

 

All the various reports really show is that damage modelling is not as simple as some players would like to believe.

Edited December 26, 2018 by Sgt_Joch

[unreasonable]

1 hour ago, Operation_Ivy said:

 

There are different values in the in-game files of the ammunition for blast, number of fragmentation (shrapnel) and many other factors. 

 

I take your word for it as I have not unpacked them. I assume some parameters are purely empirical, such as shot weight, and others not.  It is hard to say how they all interact and it is not obvious how you would tweak these to get desired results.  So I would rather just have a look at overall game results and make up my own mind on how close to reality they are, in the hope that the developers will make tweaks (or allow SP DM modding  ).

 

I am all in favour of doing this for different shell types, BTW, it is just vastly more time consuming to do that systematically for rounds that cannot be fired by an AA gun. 

Edited December 26, 2018 by unreasonable

[=EXPEND=13SchwarzeHand]

1 hour ago, unreasonable said:

do not know if there is a separate blast and splinter estimate in the DM or not: if there were then a tweak to lower splinters and boost blast in line with HE content could well be an improvement.  Modelling blast effects, however, is extremely hard.  

 

 

I don’t know why you keep on calling for an overly complicated DM, which you or no one else will be able to build from the scarcely available data into the simplified structure of the game anyways. Modeling blast effect? It’s HE or AP, there needs to be a best estimate as to how much of each each type of ammo is. ATM thats not implemented correctly.

 

I would be happy if the damage ranking of weapons would be portrayed reasonably in game. At the moment not even that is happening.

48 minutes ago, Sgt_Joch said:

he estimate in the US study is actually that a 30 mm shell hit to a P47 has a 28% chance of resulting in a kill, so yes, it could take 3-4 30 mm hits to kill a P47 based on that estimate.

 

Wrong

[Sgt_Joch]

38 minutes ago, =EXPEND=SchwarzeDreizehn said:

 

Wrong

 

No, my statement is correct. Look at the chart on page 39 of the report. 28 % of an "A" kill on a P47 from one 30 mm HE hit.

[unreasonable]

1 hour ago, =EXPEND=SchwarzeDreizehn said:

 

I don’t know why you keep on calling for an overly complicated DM, which you or no one else will be able to build from the scarcely available data into the simplified structure of the game anyways. Modeling blast effect? It’s HE or AP, there needs to be a best estimate as to how much of each each type of ammo is. ATM thats not implemented correctly.

 

I would be happy if the damage ranking of weapons would be portrayed reasonably in game. At the moment not even that is happening.

 

It is not me calling for an overly complicated DM - most of the people making a fuss are now, as they have always been, banging on about mineshells not being as other HE shells and not having an in game treatment that reflects how they cause damage. If the game has a blast calculation already, distinct from a splinter calculation, the balance is a matter of tweaking.  If the absolute level of lethality for each round type is reasonable, so will be the relative ranking.  

 

 

 

 

[Operatsiya_Ivy]

Asking for correctly modelled mineshells, which btw are used by roughly 50% of all aircrafts in the game, is hardly asking for an overcomplicated damage model. I think modelling the mineshell in a simplistic way wouldn't  be much more complicated than AP or HE. Even if it is a lot more complicated, which might be very well the case, it is necessary in my eyes, because simply adjusting the HE values for it obviously doesn't work. Even when you don't consider the unrealistic damage depiction (shrapnel focused).

Imagine the outrage if something similar would be the case with the flight model. 

 

A good combat flight sim depends on a good DM as much as on a good FM. I can't emphesize this enough.

Edited December 26, 2018 by Operation_Ivy

[=EXPEND=13SchwarzeHand]

What makes you think the current parameters cannot be tweaked to achieve more realistic results?

I‘m pretty sure they can be.

I am not against a more complicated model, but tweaking what we have is probably more realistic than Devs making a new model.

Edited December 26, 2018 by =EXPEND=SchwarzeDreizehn