Pilot kills

[Lusekofte]

1 hour ago, RedeyeStorm said:

I prefer SP and really enjoy flying as a bomber pilot

I too did SP bomber careers until the new aqmb arrived. 
A few of them was / are good too. Some scripted campaigns like Havocs and Icering are exceptional. But some careers are very repetitive. Still not found a good Heinkel career 

[dbuile]

4 hours ago, LukeFF said:

... tracks ... showing the problem.

Do tracks contain fine-grained detail regarding individual bullet trajectory through components, as well as what was hit?

 

If Stonehouse's material posted earlier in the thread is accurate, it'd be interesting to understand the shape of the pilot's head/body "components".   A single hit to body in many cases supersede the damage points allowed by the body.  My casual interest is if/how legs/arms are modelled, ie, whether they can be hit or not, or whether a compact trunk without extremities (ex, nothing below the knees/elbows) is simulated for damage calculation.    

 

 

[LukeFF]

20 minutes ago, dbuile said:

Do tracks contain fine-grained detail regarding individual bullet trajectory through components, as well as what was hit?

 

If Stonehouse's material posted earlier in the thread is accurate, it'd be interesting to understand the shape of the pilot's head/body "components".   A single hit to body in many cases supersede the damage points allowed by the body.  My casual interest is if/how legs/arms are modelled, ie, whether they can be hit or not, or whether a compact trunk without extremities (ex, nothing below the knees/elbows) is simulated for damage calculation.

 

From what has been posted by the developers both here and in the beta testing forums, posting track files allows them to examine what's going on with internal testing tools. Beyond that, I can't really say what information they yield, other than that they are a superior source of information than a YouTube video.

[Stonehouse]

3 hours ago, SCG_motoadve said:

No , a pilot can get hit and die, its not about that, mostly its about a first bullet laser pilot kills that happens one after the other, even while flying bombers.

 

Just curious, assuming the pilot that killed you was unmodded AI, what range did they open fire at? if possible what skill were they? also, if possible, how many rounds fired?

I think it is logged somewhere as I know tacview displays the info but can't remember where. 

 

 

[dbuile]

47 minutes ago, LukeFF said:

they are a superior source of information than a YouTube video.

No doubt, there... the posted video ~suggests~ a low deflection shot from behind at what could be 100 yards or less;  it really doesn't help advance an argument.

 

@Stonehouse Part of the difficulty trying to understand the argument is that some are pointing at Online play.  I see what you're getting at, and I agree: a potential for AI having uncanny ability to hit a pilot with a burst could have been part of the cause; but that's not what some folks are really saying.  

[Lusekofte]

1 hour ago, LukeFF said:

 

From what has been posted by the developers both here and in the beta testing forums, posting track files allows them to examine what's going on with internal testing tools. Beyond that, I can't really say what information they yield, other than that they are a superior source of information than a YouTube video.

This is why I stopped flying this sim. Because in your eyes and devs all is perfect

[LukeFF]

1 minute ago, Lusekofte said:

This is why I stopped flying this sim. Because in your eyes and devs all is perfect

 

Seriously, that's what you concluded from my post? I can guarantee you that I do not think the game is perfect by any means. I have posted plenty of complaints, both here and in the closed beta testing forums, about issues that I would like to see resolved.

[dbuile]

2 minutes ago, LukeFF said:

that's what you concluded from my post?

I believe he misinterpreted the "theys" as saying the DEVs are a superior source of info, instead of the actual meaning of a TRACK contains better info than video.

[Charon]

Well fine, I'll gather some data about this over the next few days. My planned test schedule is as follows:

 

T0:

Bf 109G-6 vs P-51D (ace)

 

T1:

Bf 109G-6 vs P-51D

 

T2:

Bf 109G-6 (5% fuel) vs P-51D

 

T3:

Il-2M41 vs P-51D

 

T4:

He 111H-16 vs P-51D.

 

T5:

Pe-2 s35 vs P-51D

 

T6:

Hs 129 vs P-51D

 

T7:

C-47 (without cargo) vs P-51D

 

T8:

C-47 (with cargo) vs P-51D

 

T9:

Bf 109 G-6 vs Mig-3 (2x UB)

 

T10:

Bf 109 G-6 vs La 5 (AP only)

 

Defender (novice) vs P-51 (average) unless otherwise noted. In each trial, the defender will be set in auto-level, at maximum continuous power. Both planes at 1000m, "escape" setup with a minimum (2000m) separation. Lapino (winter). Six rounds for each test. I will record only which crew are killed in the initial pass; once the AI breaks off their initial firing pass, I will stop the run.

 

"Average" is chosen because some preliminary testing shows that ace AI tends to kill at such long range that it never gets a chance to penetrate armor, while novice tends to shoot so poorly that it runs out of ammo before killing.

 

Go ahead and place your bets on how many instant-kills we'll see in various trials, as well as propose additions. I don't plan to record tracks, as there would be quite a lot and I think the aggregate statistics are more interesting than any individual track, regardless. Feel free to run your own, if you don't trust me.

 

Edit: Almost forgot weather settings: 935AM, no wind, no turbulence, few clouds at 2500m and 0 haze.

Edited December 21, 2022 by Charon

[6./ZG26_Custard]

1 minute ago, Charon said:

Go ahead and place your bets on how many instant-kills we'll see in various trials, as well as propose additions. I don't plan to record tracks, as there would be quite a lot and I think the aggregate statistics are more interesting than any individual track, regardless. Feel free to run your own, if you don't trust me.

Please, if you can add to those the 110, the Ju 88 and the A-20 and please test online too. I would also recommend that you use the P-51 vs the B-25 and B-26 and than take a pop at the P-47 and compare the results. I would ask anyone who doubts to fly as a bomber pilot online for a few months and then come back with their findings.

And I have quit flying so I'll not be doing any tests, I had 8 years of tests.

[Charon]

2 minutes ago, 6./ZG26_Custard said:

Please, if you can add to those the 110, the Ju 88 and the A-20 and please test online too. I would also recommend that you use the P-51 vs the B-25 and B-26 and than take a pop at the P-47 and compare the results. I would ask anyone who doubts to fly as a bomber pilot online for a few months and then come back with their findings.

T11:

Bf 110 G vs P-51D

 

T12:

Ju 88 A-4 vs P-51D

 

T13:

A-20 vs P-51D

 

Skipping the B-25/26 as I won't be able to get technochat reports on who is killed for AI planes.

 

I will not test this online unless given strong evidence to believe PKills are inherently more common in multiplayer. While I have access to a private server, setting up the missions and running the trials will be far more tedious online than in a local quick mission.

[6./ZG26_Custard]

Just now, Charon said:

I will not test this online unless given strong evidence to believe PKills are inherently more common in multiplayer.

I hardly ever fly single player so I can't and won't comment on SP tests but hopefully someone who flies online would be able to do some tests?

[SCG_motoadve]

9 minutes ago, Charon said:

T11:

Bf 110 G vs P-51D

 

T12:

Ju 88 A-4 vs P-51D

 

T13:

A-20 vs P-51D

 

Skipping the B-25/26 as I won't be able to get technochat reports on who is killed for AI planes.

 

I will not test this online unless given strong evidence to believe PKills are inherently more common in multiplayer. While I have access to a private server, setting up the missions and running the trials will be far more tedious online than in a local quick mission.

They are as common in SP, auto level might not be accurate though.

Try the 190 if you can please.

[Charon]

T4.5:

He 111 H-16 vs P-51D (Ace)

 

T14:

Fw 190A-6 vs P-51D

 

T15:

He 111 H-16 vs La-5 (AP only)

Edited December 21, 2022 by Charon

[Charon]

Okay, I've just run 130ish quick missions. Some results.

 

3 hours ago, Charon said:

T0:

Bf 109G-6 vs P-51D (ace)

This one was shockingly consistent: the right wing came off at around 500m, and as a result the P-51 never got close enough for a pilot kill.

 

3 hours ago, Charon said:

T1:

Bf 109G-6 vs P-51D

Remember, this is "average" AI. With this AI setting, the P-51 opens fire at around 500m. Pilot kill in 1/6 trials, with one additional knock-out (he woke up). 1 fire, 4 wing-off.

 

3 hours ago, Charon said:

T2:

Bf 109G-6 (5% fuel) vs P-51D

I ran this at 10% fuel; turns out that's the lowest allowed. 4/6 pilot kills (one which was preceded by fire). Maybe one of our statisticians can tell me if the the difference between T1 and T2 is statistically significant.

3 hours ago, Charon said:

T3:

Il-2M41 vs P-51D

I ran this at boosted power because I didn't want to mess around with RPM. 1/6 PKill. I'm somewhat amused that the pkill count is the same as against the 109 (although the Sturmovik was in much better shape than the 109 in the rounds where the pilot didn't die).

 

3 hours ago, Charon said:

T4:

He 111H-16 vs P-51D.

Radioman killed 3/6 times. No other crew casualties.

 

(Note: Remember, I'm testing what happens on the first pass, because that's most reproducible. Just because the attacker is driven off without inflicting crew casualties doesn't mean the plane isn't crippled).

 

Quote

T4.5:

He 111 H-16 vs P-51D (Ace)

I was surprised by the above result, so inserted an extra trial against an Ace P-51.

 

Gunner killed: 4/6

Waist Gunner: 4/6

Navigator: 3/6

Radioman: 3/6

Pilot: 0/6

 

2 fires.

 

Also a surprising result to me, frankly, but that seat back armor seemed to do its job.

 

3 hours ago, Charon said:

T5:

Pe-2 s35 vs P-51D

 

Navigator+Pilot killed in 1/6 trials.

 

A very surprising result to me. I think the quick mission format meant the P-51 had very little closure... and because they started at co-altitude, both the dorsal and ventral gunners were able to open fire. They opened up at long range, and so had plenty of chances to land hits before the P-51 opened fire. In the one trial where it was able to open fire, it scored a navigator + pilot kill fairly quickly.

 

3 hours ago, Charon said:

T6:

Hs 129 vs P-51D

Ran this in combat mode, since I didn't want to fiddle with RPM.

 

P-51 broke off without achieving a first-pass kill 6/6 trials.

 

3 hours ago, Charon said:

T7:

C-47 (without cargo) vs P-51D

Co-pilot: killed 2/6

pilot: 1/6

 

I feel like the high rate of closure lead to the P-51 breaking off sooner than it would have against other planes. Obviously the C-47 is toast in the long run.

 

Quote

T7.5:

C-47 (without cargo) vs P-51D (ace)

Co-pilot and Pilot both killed 6/6 trials. Unlike T0 with the 109 where the ace opens up far out and destroys the wing before rounds make it through the armor, this one is very lethal to the unarmored crew.

 

(Isn't there supposed to be a navigator or something? He must be wearing mithril).

 

3 hours ago, Charon said:

T8:

C-47 (with cargo) vs P-51D (average)

 

Just like trial 7, but with cargo. Pilot + Co-Pilot both killed in 3/6 trials. Cargo doesn't appear to provide any protection.

 

I think the high closure rate makes the P-51 vs C-47 test relatively high variance: the average AI is a poor shot, and only really gets once chance before breaking off.

 

3 hours ago, Charon said:

T9:

Bf 109 G-6 vs Mig-3 (2x UB)

Of course I meant BS, not UB.

 

2 fires, no pilot kills.

 

I expected this to be more lethal, but the Mig had trouble catching up and expended most of its ammo at 500m. The Average AI really isn't a very good shot.

 

3 hours ago, Charon said:

T10:

Bf 109 G-6 vs La 5 (AP only)

Some minor damage but no pilot kills, no fires.

 

The La 5 also seemed to struggle to keep up (the AI didn't seem willing to close the rads enough), and only took pot-shots at 500m before backing off. The La 5 AI doesn't seem to like pressing the attack very hard.

Quote

T11:

Bf 110 G vs P-51D

Radioman killed 3/6, pilot killed 2/6. 110 on fire 5/6.

 

Gunner wasn't very effective. The armor seemed to do its job though: in every trial where the crew were killed, the plane was already on fire (and frequently shedding parts) before they were killed.

 

Quote

T12:

Ju 88 A-4 vs P-51D

Gunners did their work well. Ventral gunner killed in 1/6 trials, no other crew kills.

Quote

T13:

A-20 vs P-51D

Gunners vulnerable but very effective for some reason: mustang down in 4/6 trials (even though the A-20 also died in 2 of those 4).

 

Gunner killed 5/6

Radioman killed 3/6

Pilot killed 0/6

 

Quote

T14:

Fw 190A-6 vs P-51D

Pkill 6/6, but not generally instant pilot kills. 2 fires, one knock-out before pilot kills.

 

Interesting to contrast this to T1. The 190 seems much more durable overall, and kept flying even when the 109 would have shed a wing... but that gave the P-51 plenty of time to keep working towards that pilot kill. This does match my personal experience that most of my 190 kills are pilot kills.

 

Quote

T15:

He 111 H-16 vs La-5 (AP only)

No crew kills. La-5 down in 1/6 trials (loss of rudder).

 

I was surprised by this. It seemed like the La-5 suffered from a low rate of closure (I think the quick mission format starts it at the same speed as the He 111?). I hypothesized that it would do better with more starting speed, so added several more trials:

 

Quote

T16:

He 111 H-16 vs La-5 (AP only, +1000m starting altitude)

The AI didn't seem to like this setup. In all 6 trials, the AI La-5 broke off the attack and dove away just before it got within firing range (to avoid colliding?).

 

Quote

T17:

He 111 H-16 vs La-5 (AP only, +1000m starting altitude, 3000m separation)

No crew kills. The La-5 fired very short bursts before breaking away. In the 5th trial, both the dorsal and waist gunner were wounded (not killed). Presumably these were hits to the extremities from 20mm slugs, but interesting to note that even 20mm can be nonlethal in some cases.

 

Quote

T18:

He 111 H-16 vs La-5 (mixed ammo, +1000m starting altitude, 3000m separation, ace AI)

La-5 fired only short bursts. No crew kills -- in 2/6 trials the "ace" La-5 failed to even land any hits.

 

Quote

T19:

He 111 H-16 vs La-5 (AP only, +1000m starting altitude, 3000m separation, ace AI)

Back to AP only. No crew kills, and again in 2/6 trials the La-5 failed to land any hits.

 

----

 

General remarks:

 

These tests are very artificial in many ways. High rates of closure seemed to suppress the number of pilot kills, as the AI would break off the attack quickly; conversely the P-51 had all the time in the world to hammer the Fw 190.

 

The gunners outperformed my expectations. I think a lot of this was due to the nature of the test: the attackers didn't come in with much speed, and the even starting altitudes meant the AI exposed themselves to both dorsal and ventral gunners. Rerunning some of these with more altitude separation might be interesting.

 

I may have read somewhere that AI gunners on the players' plane get a high skill level. Does anyone know if this is true?

 

"Average" AI shoots very poorly, this seemed to make things particularly challenging for some of the Soviet pilots. The La-5 with AP I really expected to be a pilot killing machine, but it didn't live up to my expectations at all.

 

I would of course expect a human to be far more lethal, both for being more accurate but also because they would likely hold their fire until shorter ranges.

 

The La-5 AI generally seemed unwilling to press attacks under any circumstances. The P-51 AI by contrast was out for blood.

 

Armor generally seemed to do its job, and I saw few pilot kills amongst the bombers. The rear of the He 111 and A-20 are both very lethal places to be. The C-47 is of course a sitting duck against anyone capable of aimed fire.

 

Now fight!

Edited December 21, 2022 by Charon

[Stonehouse]

Generally, each weapon has a base accuracy due to the weapon themselves including dispersion and barrel temperature impacts on bullet speed etc. I can see these values in the files and assume they are accurate as the devs can make them barring typos and research sources providing inaccurate data for some reason.  

 

Accuracy beyond the base weapon accuracy for the gunner AI is down to a base default error, skill based error, firer v's target factors error (G loading & velocity vector difference for example) etc. How often the AI searches for target (differs by skill) plus other stuff that determines how an AI shoots their weapon - these values are in the turretcontrollerAI.txt file. In each turret definition there is also a "bracketing" error that reduces each shot at the same target - kind of like the AAA bracketing error but not as severe. As far as I know this is same across all turrets. Ditto with attack distances for different target types (air, ground, balloon etc).

 

So essentially you start with the weapon accuracy and range and degrade the accuracy of it based on AI factors. As I mentioned as well, AI gunners can shoot out to 3x normal range if they have enough ammo. So, in a stock engagement I figure you are going to get shot at more at the start and it will decrease as ammo reduces and they are eventually constrained to their normal range. Bracketing error may explain some of the wild shots although the formula to calculate the final error is based on taking the tan of a derived number and if you start plugging in numbers you can sometimes get some quite odd things out in extreme cases. ie very large errors. I think this is the main reason for the G cut out on gunners as it serves like a boundary condition to stop the error calc getting weird. As the stock G cutout is 5G it is probably still letting the occasional weird thing through. In AI gunnery I set it down to 3Gs, as I doubt you'd be shooting accurately at higher Gs especially in unpowered turrets and I don't want the gunner's wasting ammo. 

 

For fighter pilots I have not so far found the equivalent file describing accuracy modifiers by skill as per gunners not even convergence settings or burst lengths (I think at least this must be there somewhere as AI novices seem to fire longer bursts than aces). It's more about when they can shoot and how well they maneuver and their search ability. I theorize that possibly it is mostly about how far they can shoot from and how often they get to shoot due to their flying skill. Each aircraft does seem to have a GunAirAimSmooth value which can differ but it is not skill based. Generally the values are down about 0.1 or 0.2. It may be a rating of the aircraft as a gun platform. Best translation of Russian comment for the parameter is Smoothing the guidance of guns in aerial combat (necessary when splitting). [0..1], 0 - maximum, 1 - minimum, -1 - off. 

 

Aces open fire at 800m, high 600m and normal 500m and novices at 400m. Looking at say a 109 definition there is nothing defined in the way of armor between rudder and pilot back plate although later models have a 20mm duralumin armor plate behind a fuel tank - this is given an analogue rating of 5mm armor. Presumably the game armor system is based on steel. Bombers do have armor curtains defined sometimes near turrets and things. Pretty much though I think the damage model is a collection of systems and some defined armor positions and everything else is treated as of no effect on a projectile.

 

So, I think unless a projectile actually encounters one of these defined areas it's kind of treated as going through air. I think this is why it is possible to get a round going right through the aircraft from tail to cockpit and things like cargo don't contribute to crew survival.  I assume AI pilots likely aim at generally the centre of their target and don't try to hit wings or some area specifically. What they do to the target is all about how many rounds hit from a burst. Hence my question about the number of rounds fired and how far away they opened fire.

 

As an FYI for the HE111 the defined armor is as below:

Spoiler

//Loss of bullet energy on hit, true - check on entering and exiting the collision box, the first is the equivalent armor on entry and exit, the second is joules per meter of travel inside the collision box

energyloss="~armor0",false,5.0,1 //pilot seat cup - 5mm
energyloss="~armor1",false,10.0,1 //pilot armored back - 10mm
energyloss="~armor2",false,10.0,1 //pilot's head armor - 10mm
energyloss="~armor3",false,8.0,1 //fuselage armor plate behind the top gunner - 8mm
energyloss="~armor4",false,8.0,1 //vertical armored partition behind the left window of the fuselage gunner - 8mm
energyloss="~armor5",false,8.0,1 //vertical armored partition behind the right window of the fuselage gunner - 8mm
energyloss="~armor6",false,8.0,1 //slope armor behind bottom gunner in bathroom - 8mm
energyloss="~armor7",false,8.0,1 //vertical armor behind bottom gunner in bathroom - 8mm
energyloss="~armor8",false,6.0,1 //left bathtub armor wall - 6mm
energyloss="~armor9",false,6.0,1 //bath armor floor - 6mm
energyloss="~armor10",false,6.0,1 //right bathtub armor wall - 6mm
energyloss="~armor11",false,5.0,1 //armor under bombardier's bed - 5mm
energyloss="~armor12",false,5.0,1 //armor plate behind the left water radiator - 5mm
energyloss="~armor13",false,8.0,1 //Armoured damper of the left oil cooler - 8mm
energyloss="~armor14",false,8.0,1 //armored damper of the right oil cooler - 8mm
energyloss="~armor15",false,5.0,1 //armor plate behind the right water radiator - 5mm
energyloss="~armor16",false,5.0,1 //upper armor plate of the left oil cooler - 5mm
energyloss="~armor17",false,5.0,1 //upper armor plate of the right oil cooler - 5mm

 

On top of this the cockpit has the 6 direction cube of armor defined around it as described in my earlier post for shrapnel damage. The energy loss for the cockpit is defined as energyloss="",true,0.23,300

 

So for example a 50 cal round has the damage and penetration values below:

Armor=0,895,        28,191,    11,381,    0,762
Armor=100,852,    26,172,    10,345,   0,690
Armor=500,689,    19,113,      8,226,   0,452    
Armor=1000,512,   12,62,        5,125,   0,250
Armor=2000,301,    6,22,        2,43,    0,86

 

So, at 500m the bullet will penetrate anything less than 19 mm armor equivalent and cause between 113-452 damage.  There is energy loss for entering the hit box only if the boolean in energy loss definition is false and entering and exiting if true. For bullets a head has 50 points as a life rate and a body has 250. The only other energy loss definitions for the He111 are things like prop blades, engines, fuel and oil tanks, water and oil coolers, undercarriage, weapons themselves. So the aircraft skin, minor bulkheads in the fuselage and anything not defined as causing an energy loss I believe has no impact on the projectile passing down the length of the fuselage. 

 

With all the above in mind probably only the armor between the top gunner and window gunners was really involved so it is not surprising you killed the gunners and the pilot was ok when it was pure AP based rear attacks. For HE based shells it is the shrapnel damage that counts and each HE hit generates many pieces of shrapnel each of which can penetrate the 6 direction cube around the crew member and damage them. Talking 20-40 bits of shrapnel per hit. Guessing a spherical pattern to these but I don't know how the game figures out the distribution of the fragments across the target. Guessing that while projectiles may be individually tracked probably shrapnel fragments are abstracted based on position of hit and then a proportion of available fragments allocated randomly to areas that could be hit. Could be wrong but just think that calculating the exact trajectory of each fragment from each HE shell exploding would too big a computational load.

 

Doing testing for the AI gunnery mod I found that most bombers have poor fields of fire and generate a low volume of fire in any one direction and aren't too good at getting turret fields of fire overlapping (particularly the older non-US ones) The only real exceptions to this are things like B25s with powered turrets with multiple guns per turret. Testing balance for the mod I usually take B25s as a reference point. Even so I generally find that late war AI fighters can generate such a volume of fire in a pass they can mostly overwhelm the bomber defenses even if they get damaged or destroyed in the process. This is exacerbated by the AI fighters firing generally accurately from long ranges. I think this is the reason for the fudge allowing gunners to fire out to 3x normal range. 

 

Humans are all outside this and so it is just how many hits, where did each round hit and did it go through one of the defined hitboxes so reducing the energy of the round before reaching the cockpit. Then if it does reach the cockpit, it is how much penetration is available and what damage is transferred if it goes through. More than 50 points cumulative non shrapnel damage to the head or 250 cumulative to the body kills the crew member. For shrapnel it is 100 cumulative points for the head and 500 for the body. It used to be a lot higher points in prior versions of the game and it did not also always have a head hit point value. At one point it was just a life rate. The old vulnerable gunners mod by Reiper_420 used to reduce the stock hit points for gunners so it was easier to kill them as with the stock values it was perceived as too hard to do so.  

 

Probably that the only thing that is going to make people happier is to make the DM and AI much more comprehensive. At the moment it is kind of halfway between things and reducing the hit points without the DM improvements is likely leading to the issue talked about here.

 

For offline players it is pretty straightforward to make a mod that reverts the life points to higher values if that helps you enjoy the game again while the dev team sorts it out. I believe the old values were head 200/600 for AP/shrapnel and body 500/1500 for AP/shrapnel. As you can see these are much higher than they are now - look back at the 50 cal table and think of what that means.  With these values an unarmored torso hit at 500m by a 50 cal would not kill the person (I take non head hits as an abstraction to represent torso hits where vital organs are) - it would take at least two hits. 

 

PS If people feel that bomber gunners are ineffectual, I can easily make a version of AI gunnery that will make you feel a lot more pain when you attack an AI bomber

Edited December 21, 2022 by Stonehouse
added PS

[Oyster_KAI]

 

52 minutes ago, Stonehouse said:

Generally, each weapon has a base accuracy due to the weapon themselves including dispersion and barrel temperate impacts on bullet speed etc. I can see these values in the files and assume they are accurate as the devs can make them barring typos and research sources providing inaccurate data for some reason.  

 

Accuracy beyond the base weapon accuracy for the gunner AI is down to a base default error, skill based error, firer v's target factors error (G loading & velocity vector difference for example) etc. How often the AI searches for target (differs by skill) plus other stuff that determines how an AI shoots their weapon - these values are in the turretcontrollerAI.txt file. In each turret definition there is also a "bracketing" error that reduces each shot at the same target - kind of like the AAA bracketing error but not as severe. As far as I know this is same across all turrets. Ditto with attack distances for different target types (air, ground, balloon etc).

 

So essentially you start with the weapon accuracy and range and degrade the accuracy of it based on AI factors. As I mentioned as well, AI gunners can shoot out to 3x normal range if they have enough ammo. So, in a stock engagement I figure you are going to get shot at more at the start and it will decrease as ammo reduces and they are eventually constrained to their normal range. Bracketing error may explain some of the wild shots although the formula to calculate the final error is based on taking the tan of a derived number and if you start plugging in numbers you can sometimes get some quite odd things out in extreme cases. ie very large errors. I think this is the main reason for the G cut out on gunners as it serves like a boundary condition to stop the error calc getting weird. As the stock G cutout is 5G it is probably still letting the occasional weird thing through. In AI gunnery I set it down to 3Gs, as I doubt you'd be shooting accurately at higher Gs especially in unpowered turrets and I don't want the gunner's wasting ammo. 

 

For fighter pilots I have not so far found the equivalent file describing accuracy modifiers by skill as per gunners not even convergence settings or burst lengths (I think at least this must be there somewhere as AI novices seem to fire longer bursts than aces). It's more about when they can shoot and how well they maneuver and their search ability. I theorize that possibly it is mostly about how far they can shoot from and how often they get to shoot due to their flying skill. Each aircraft does seem to have a GunAirAimSmooth value which can differ but it is not skill based. Generally the values are down about 0.1 or 0.2. It may be a rating of the aircraft as a gun platform. Best translation of Russian comment for the parameter is Smoothing the guidance of guns in aerial combat (necessary when splitting). [0..1], 0 - maximum, 1 - minimum, -1 - off. 

 

Aces open fire at 800m, high 600m and normal 500m and novices at 400m. Looking at say a 109 definition there is nothing defined in the way of armor between rudder and pilot back plate although later models have a 20mm duralumin armor plate behind a fuel tank - this is given an analogue rating of 5mm armor. Presumably the game armor system is based on steel. Bombers do have armor curtains defined sometimes near turrets and things. Pretty much though I think the damage model is a collection of systems and some defined armor positions and everything else is treated as of no effect on a projectile.

 

So, I think unless a projectile actually encounters one of these defined areas it's kind of treated as going through air. I think this is why it is possible to get a round going right through the aircraft from tail to cockpit and things like cargo don't contribute to crew survival.  I assume AI pilots likely aim at generally the centre of their target and don't try to hit wings or some area specifically. What they do to the target is all about how many rounds hit from a burst. Hence my question about the number of rounds fired and how far away they opened fire.

 

As an FYI for the HE111 the defined armor is as below:

  Reveal hidden contents

//Loss of bullet energy on hit, true - check on entering and exiting the collision box, the first is the equivalent armor on entry and exit, the second is joules per meter of travel inside the collision box

energyloss="~armor0",false,5.0,1 //pilot seat cup - 5mm
energyloss="~armor1",false,10.0,1 //pilot armored back - 10mm
energyloss="~armor2",false,10.0,1 //pilot's head armor - 10mm
energyloss="~armor3",false,8.0,1 //fuselage armor plate behind the top gunner - 8mm
energyloss="~armor4",false,8.0,1 //vertical armored partition behind the left window of the fuselage gunner - 8mm
energyloss="~armor5",false,8.0,1 //vertical armored partition behind the right window of the fuselage gunner - 8mm
energyloss="~armor6",false,8.0,1 //slope armor behind bottom gunner in bathroom - 8mm
energyloss="~armor7",false,8.0,1 //vertical armor behind bottom gunner in bathroom - 8mm
energyloss="~armor8",false,6.0,1 //left bathtub armor wall - 6mm
energyloss="~armor9",false,6.0,1 //bath armor floor - 6mm
energyloss="~armor10",false,6.0,1 //right bathtub armor wall - 6mm
energyloss="~armor11",false,5.0,1 //armor under bombardier's bed - 5mm
energyloss="~armor12",false,5.0,1 //armor plate behind the left water radiator - 5mm
energyloss="~armor13",false,8.0,1 //Armoured damper of the left oil cooler - 8mm
energyloss="~armor14",false,8.0,1 //armored damper of the right oil cooler - 8mm
energyloss="~armor15",false,5.0,1 //armor plate behind the right water radiator - 5mm
energyloss="~armor16",false,5.0,1 //upper armor plate of the left oil cooler - 5mm
energyloss="~armor17",false,5.0,1 //upper armor plate of the right oil cooler - 5mm

 

On top of this the cockpit has the 6 direction cube of armor defined around it as described in my earlier post for shrapnel damage. The energy loss for the cockpit is defined as energyloss="",true,0.23,300

 

So for example a 50 cal round has the damage and penetration values below:

Armor=0,895,        28,191,    11,381,    0,762
Armor=100,852,    26,172,    10,345,   0,690
Armor=500,689,    19,113,      8,226,   0,452    
Armor=1000,512,   12,62,        5,125,   0,250
Armor=2000,301,    6,22,        2,43,    0,86

 

So, at 500m the bullet will penetrate anything less than 19 mm armor equivalent and cause between 113-452 damage.  There is energy loss for entering the hit box only if the boolean in energy loss definition is false and entering and exiting if true. For bullets a head has 50 points as a life rate and a body has 250. The only other energy loss definitions for the He111 are things like prop blades, engines, fuel and oil tanks, water and oil coolers, undercarriage, weapons themselves. So the aircraft skin, minor bulkheads in the fuselage and anything not defined as causing an energy loss I believe has no impact on the projectile passing down the length of the fuselage. 

 

With all the above in mind probably only the armor between the top gunner and window gunners was really involved so it is not surprising you killed the gunners and the pilot was ok when it was pure AP based rear attacks. For HE based shells it is the shrapnel damage that counts and each HE hit generates many pieces of shrapnel each of which can penetrate the 6 direction cube around the crew member and damage them. Talking 20-40 bits of shrapnel per hit. Guessing a spherical pattern to these but I don't know how the game figures out the distribution of the fragments across the target. Guessing that while projectiles may be individually tracked probably shrapnel fragments are abstracted based on position of hit and then a proportion of available fragments allocated randomly to areas that could be hit. Could be wrong but just think that calculating the exact trajectory of each fragment from each HE shell exploding would too big a computational load.

 

Doing testing for the AI gunnery mod I found that most bombers have poor fields of fire and generate a low volume of fire in any one direction and aren't too good at getting turret fields of fire overlapping (particularly the older non-US ones) The only real exceptions to this are things like B25s with powered turrets with multiple guns per turret. Testing balance for the mod I usually take B25s as a reference point. Even so I generally find that late war AI fighters can generate such a volume of fire in a pass they can mostly overwhelm the bomber defenses even if they get damaged or destroyed in the process. This is exacerbated by the AI fighters firing generally accurately from long ranges. I think this is the reason for the fudge allowing gunners to fire out to 3x normal range. 

 

Humans are all outside this and so it is just how many hits, where did each round hit and did it go through one of the defined hitboxes so reducing the energy of the round before reaching the cockpit. Then if it does reach the cockpit, it is how much penetration is available and what damage is transferred if it goes through. More than 50 points cumulative non shrapnel damage to the head or 250 cumulative to the body kills the crew member. For shrapnel it is 100 cumulative points for the head and 500 for the body. It used to be a lot higher points in prior versions of the game and it did not also always have a head hit point value. At one point it was just a life rate. The old vulnerable gunners mod by Reiper_420 used to reduce the stock hit points for gunners so it was easier to kill them as with the stock values it was perceived as too hard to do so.  

 

Probably that the only thing that is going to make people happier is to make the DM and AI much more comprehensive. At the moment it is kind of halfway between things and reducing the hit points without the DM improvements is likely leading to the issue talked about here.

 

For offline players it is pretty straightforward to make a mod that reverts the life points to higher values if that helps you enjoy the game again while the dev team sorts it out. I believe the old values were head 200/600 for AP/shrapnel and body 500/1500 for AP/shrapnel. As you can see these are much higher than they are now - look back at the 50 cal table and think of what that means.  With these values an unarmored torso hit at 500m by a 50 cal would not kill the person (I take non head hits as an abstraction to represent torso hits where vital organs are) - it would take at least two hits. 

 

PS If people feel that bomber gunners are ineffectual, I can easily make a version of AI gunnery that will make you feel a lot more pain when you attack an AI bomber

Very good research.
Your AAA/AI Gunnery MOD has greatly improved this game, Keep in faith!!
It's really admirable to do this with the strength of one person.

Edited December 21, 2022 by Oyster_KAI

[Stonehouse]

7 hours ago, Charon said:

the attackers didn't come in with much speed,

By the way - velocity vector difference between firer and target is the largest generator of error on the gunners. Keeping the relative vector difference large = larger gunner error. Crawl up the bomber's six with little speed difference and there is little error. 

[RNAS10_Mitchell]

It's a real pity that there isn't a single beta tester that flys the game and has experienced this first hand.  One the devs have faith in, that could report this and other issues as they occur to the developers for investigation and possible action.  Would certainly be an improvement over the current environment where we are asked to submit tracks over an extended period and HOPE the devs notice and respond.  A real pity.?

[Charon]

4 hours ago, Stonehouse said:

By the way - velocity vector difference between firer and target is the largest generator of error on the gunners. Keeping the relative vector difference large = larger gunner error. Crawl up the bomber's six with little speed difference and there is little error. 

Interesting. I would expect the rate of change in angle to be the largest factor, rather than the raw velocity difference. IMO it feels like a plane that comes screaming in on a bombers rear, but with no change in angle, should get hammered. Can you tell if angular rate of change contributes as well, or is it ignored?

 

Edit: Or maybe just the tangential component of velocity.

Edited December 21, 2022 by Charon

[LukeFF]

3 hours ago, RNAS10_Mitchell said:

It's a real pity that there isn't a single beta tester that flys the game and has experienced this first hand.  One the devs have faith in, that could report this and other issues as they occur to the developers for investigation and possible action.  Would certainly be an improvement over the current environment where we are asked to submit tracks over an extended period and HOPE the devs notice and respond.  A real pity.?

 

How is it a pity? We're not robots, and we don't get paid to do this job. ? If a non-tester sees something and is making a claim that something is wrong, then they should be prepared to put in a small bit of effort and post a track or mission file showing the problem. It's been like that on every single gaming forum I've ever been a member of. 

 

I'll also go a step further here and state that often (but not always, to be sure), if someone posts what they think is an issue and I have the time at the moment to check it out for myself, I'll fire up the game and see if can reproduce the problem. That way the process moves along a bit faster.

[RNAS10_Mitchell]

23 minutes ago, LukeFF said:

 

How is it a pity? We're not robots, and we don't get paid to do this job. ? If a non-tester sees something and is making a claim that something is wrong, then they should be prepared to put in a small bit of effort and post a track or mission file showing the problem. It's been like that on every single gaming forum I've ever been a member of. 

 

I'll also go a step further here and state that often (but not always, to be sure), if someone posts what they think is an issue and I have the time at the moment to check it out for myself, I'll fire up the game and see if can reproduce the problem. That way the process moves along a bit faster.

Did not know you were a beta tester.  Have you experienced this issue?   If so, have you reported it?   Beta testers typically look for bugs and issues, and then report them to the development team.  But, this has been my experience in non gaming software support and development.  Perhaps IL2 doesn't follow this standard approach?  If not, what exactly does a beta tester do?

 

And... it's a pity because that would be the most efficient means to get issues in front of the development team so that they could look into it.  From my experience, beta testers are a known and trusted source for just that purpose.

Edited December 21, 2022 by RNAS10_Mitchell

[dbuile]

7 minutes ago, RNAS10_Mitchell said:

Have you experienced this issue?

I read your post in the thread you started.  Your characterization of "this issue" is different than folks characterize it in this thread.  The descriptions are bound up in rhetorical language, or tangential items (some of which *are* issues, such as modelling of AI gunners) to really understand whether there's a **simulation**/scientific aspect to the issue, or whether it's mostly emotional, ie: not fun gameplay to fade to black so quickly when too many projectiles connect with my virtual pilot.

 

[Lusekofte]

17 hours ago, LukeFF said:

 

Seriously, that's what you concluded from my post? I can guarantee you that I do not think the game is perfect by any means. I have posted plenty of complaints, both here and in the closed beta testing forums, about issues that I would like to see resolved.

I am sorry Luke, you are a great guy. I am just frustrated to a point of no return. 
The only thing I can fly without frustration is carefully tweaked campaigns and fighter in coop. Even there I died with a wheels down landing with a non dramatic ground loop in the end. 

My point is I have not once been allowed to attempt a crash/emergency landing in a 2 engined ac once passed year. Not once. Not in aqmb , coop pat Wilson or servers. Always pk or total explosion in first pass by ai or human. Before this happened but once in a while you got to try to survive for an emergency landing. Not a chance in hell anymore. Not once. How can I film that?  
I fly underdogs like po 2 ju 52 c 47 Stuka and bombers. I find great excitement trying to survive in hostile environments and like that fighters try to get me. 
but the totally useless gunners and that new thing about instant death. Kills all that. If you are spotted, you are dead. 
 

[LukeFF]

1 hour ago, RNAS10_Mitchell said:

Did not know you were a beta tester.  Have you experienced this issue?   If so, have you reported it?   Beta testers typically look for bugs and issues, and then report them to the development team.  But, this has been my experience in non gaming software support and development.  Perhaps IL2 doesn't follow this standard approach?  If not, what exactly does a beta tester do?

 

Yes, I and other testers have seen the issue and it's been reported in the beta testing forums. Apparently it's a complex problem, or otherwise we would've had a fix by now.

 

The beta testers here perform the same sort of tasks you see with other software titles - an alpha or beta version is released by the developers, and we run it through the paces to see if there are any issues. Those issues are reported and discussed in the beta testing forum. I know for myself (and other testers, for sure) that I scan the forums for problems and if I see something that I think needs the developer's attention, I'll pass along the report in the testing forums. 

 

Lastly, I'll just add here as well that all of our testers - too many to list here - are all excellent at finding a wide variety of issues. We're doing about as good as we possibly can. ?

 

54 minutes ago, Lusekofte said:

I am sorry Luke, you are a great guy. I am just frustrated to a point of no return. 
The only thing I can fly without frustration is carefully tweaked campaigns and fighter in coop. Even there I died with a wheels down landing with a non dramatic ground loop in the end. 

My point is I have not once been allowed to attempt a crash/emergency landing in a 2 engined ac once passed year. Not once. Not in aqmb , coop pat Wilson or servers. Always pk or total explosion in first pass by ai or human. Before this happened but once in a while you got to try to survive for an emergency landing. Not a chance in hell anymore. Not once. How can I film that?  
I fly underdogs like po 2 ju 52 c 47 Stuka and bombers. I find great excitement trying to survive in hostile environments and like that fighters try to get me. 
but the totally useless gunners and that new thing about instant death. Kills all that. If you are spotted, you are dead. 
 

 

Understood, it happens to all of us.  

 

As for recording stuff, best thing to do is to just hit the Record command once you've started a mission. Once you've done that, the game will do the rest of the hard work. The great thing with that is that if/when you do then come across a problem, you can then replay the track from multiple camera angles. If then you see something that needs reporting, it's just a matter of uploading the track here.

 

That's a long way of saying, recording a track allows the developers a quick and easy way to see what is going on and where the cause may lie.

 

My apologies if this all comes across as obvious and you already know this, but if nothing else someone else might see this information and find it useful. ?

Edited December 21, 2022 by LukeFF

[RNAS10_Mitchell]

7 minutes ago, LukeFF said:

 

Yes, I and other testers have seen the issue and it's been reported in the beta testing forums. Apparently it's a complex problem, or otherwise we would've had a fix by now.

 

The beta testers here perform the same sort of tasks you see with other software titles - an alpha or beta version is released by the developers, and we run it through the paces to see if there are any issues. Those issues are reported and discussed in the beta testing forum. I know for myself (and other testers, for sure) that I scan the forums for problems and if I see something that I think needs the developer's attention, I'll pass along the report in the testing forums. 

 

Lastly, I'll just add here as well that all of our testers - too many to list here - are all excellent at finding a wide variety of issues. We're doing about as good as we possibly can. ?

Wonderful.  That was all that was asked.

[Noisemaker]

3 hours ago, Lusekofte said:

I am sorry Luke, you are a great guy. I am just frustrated to a point of no return. 
The only thing I can fly without frustration is carefully tweaked campaigns and fighter in coop. Even there I died with a wheels down landing with a non dramatic ground loop in the end. 

My point is I have not once been allowed to attempt a crash/emergency landing in a 2 engined ac once passed year. Not once. Not in aqmb , coop pat Wilson or servers. Always pk or total explosion in first pass by ai or human. Before this happened but once in a while you got to try to survive for an emergency landing. Not a chance in hell anymore. Not once. How can I film that?  
I fly underdogs like po 2 ju 52 c 47 Stuka and bombers. I find great excitement trying to survive in hostile environments and like that fighters try to get me. 
but the totally useless gunners and that new thing about instant death. Kills all that. If you are spotted, you are dead. 
 

I'm sorry, but as I alluded to earlier in this thread, I also fly (SP) bombers and ground attack regularly (A-20, IL-2, and Mosquito to be specific), and I am not seeing any increase in my pilot being killed.  I do note that the gunners are less effective, but I'm bringing my aircraft back every mission, especially since the escort now seems to be doing their job properly, and the mission design doesn't have us overwhelmed by 24 enemy aircraft in 3 flights. 

I sometimes wonder if we're all playing a different game...

[BlitzPig_EL]

7 minutes ago, Noisemaker said:

I sometimes wonder if we're all playing a different game...

 

This is spot on from my perspective as well.  Of course in some ways we are, especially those who only fly single player.  With so many different ways we can configure the difficulty levels, graphics, etc... In effect we are all playing different versions of the same game.  I have great sympathy for the devs, who have to try to sort all this out.

What a mess.

[RNAS10_Mitchell]

14 minutes ago, Noisemaker said:

I'm sorry, but as I alluded to earlier in this thread, I also fly (SP) bombers and ground attack regularly (A-20, IL-2, and Mosquito to be specific), and I am not seeing any increase in my pilot being killed.  I do note that the gunners are less effective, but I'm bringing my aircraft back every mission, especially since the escort now seems to be doing their job properly, and the mission design doesn't have us overwhelmed by 24 enemy aircraft in 3 flights. 

I sometimes wonder if we're all playing a different game...

Different game perhaps?   I've noticed it in multiplayer.  Rarely do off line flights.  You fly off line mostly?

[Stonehouse]

5 hours ago, Charon said:

Interesting. I would expect the rate of change in angle to be the largest factor, rather than the raw velocity difference. IMO it feels like a plane that comes screaming in on a bombers rear, but with no change in angle, should get hammered. Can you tell if angular rate of change contributes as well, or is it ignored?

 

Edit: Or maybe just the tangential component of velocity.

 

The translation of the comment for the error calc formula says speed_difference_in_area_of_aim (in meters per second) but there is also a fixed parameter ProjSpeedCoefficients that is used to multiply the speed difference. This may somehow represent the angular rate of change. Terminology is tricky especially translating Russian to English via google or such. Speed is not a vector, velocity is - so reading the comment at face value it seems like pure relative speed difference but also I do not know how this value is calculated before being handed to turretcontrollerAI to work out the error. It could be something fancy to work out a representation of velocity difference in 3 dimensions - I just would have expected it to be x,y,z values then rather than a single value. All I can really say is that this factor is the biggest one in the error - well other than distance obviously. When tuning the error parameters in gunners part of AI Gunners I ended up deciding the only way I could not misinterpret the formula values was to simplify the situation so the spreadsheet I put together was based on an in-plane stern attack on the bomber thereby making the velocity vectors of both aircraft essentially 1 dimensional and I could just use relative speed. I also note that the result of the calc is axis error so possibly this calc is done 3 times, once for each axis of the relative velocity vector. That would make sense in fact although even if true it doesn't invalidate my spreadsheet.

 

For your easy reference the error for the gunners is:

axis_error (meters) = tan(random_number * (AddCoef + ProjSpeedCoef * speed_difference_in_area_of_aim (in meters per second)) * (current_overload * OverloadCoef + CoefLow/Hi/Ace)) * distance_to_target;

 

random number is between -1 and 1

AddCoef is default aiming error

ProjSpeedCoef is described via translation as "coefficient of conversion of the speed difference of the object on which the turret and targets stand, in the aiming plane, to the error angles"

current overload is G loading on gunner. So this is 0 in level flight I believe

Overload Coef is described via translation as "coefficient for converting overload into aiming units"

CoefLow/Hi/Ace is your skill factor. 

 

However other things are also close to as important as error, eg length of burst, how long the gunner waits to acquire a new target, how often the gunner searches for a target, at what range a gunner can search to, at what range do they begin to aim etc

 

I think but don't know for sure that tan is being calculated in radians. Degrees gave quite large numbers at times for situations where I thought the error should be small, hence my thought to use radians.

 

current_overload * overloadCoef tends to be a small number eg 2*0.05 = .1 

So the main error generator (also the only one you can control as a fighter pilot) is the speed difference eg 30m/s difference * 0.03 = 0.9 

 

Note that zero relative speed (and bomber not maneuvering) means you end up doing tan (random number * (AddCoef * skill factor)) * range.

Edited December 21, 2022 by Stonehouse
typo

[Noisemaker]

2 minutes ago, BlitzPig_EL said:

 

This is spot on from my perspective as well.  Of course in some ways we are, especially those who only fly single player.  With so many different ways we can configure the difficulty levels, graphics, etc... In effect we are all playing different versions of the same game.  I have great sympathy for the devs, who have to try to sort all this out.

What a mess.

 

Medium difficulty ( I figure there's got to be a mix of ace and novice opposition), and high density front (because FLAK!  What brought down most aircraft). I've added my issues in the appropriate forum, and had them addressed, and as far as I'm concerned things have gotten better (Fewer Kobayashi Maru missions, for the non-nerds, wait, who are we kidding... For the younger generations, it's a Star Trek reference).  When something pops up, I report it generally, and it gets fixed (for a while).

 

1 minute ago, RNAS10_Mitchell said:

Different game perhaps?   I've noticed it in multiplayer.  Rarely do off line flights.  You fly off line mostly?

 

Exclusively.

[RNAS10_Mitchell]

1 minute ago, Noisemaker said:

 

Medium difficulty ( I figure there's got to be a mix of ace and novice opposition), and high density front (because FLAK!  What brought down most aircraft). I've added my issues in the appropriate forum, and had them addressed, and as far as I'm concerned things have gotten better (Fewer Kobayashi Maru missions, for the non-nerds, wait, who are we kidding... For the younger generations, it's a Star Trek reference).  When something pops up, I report it generally, and it gets fixed (for a while).

 

 

Exclusively.

Well then...perhaps that might help.   My first thoughts are gunner accuracy setting could be different  and a factor?

[LukeFF]

1 hour ago, Noisemaker said:

Medium difficulty ( I figure there's got to be a mix of ace and novice opposition)

 

To be sure, on Medium difficulty, all enemy AI fighter formations are Regular skill level, while friendly AI fighter formations are Veterans. Unfortunately, there is no skill mix within these flights - only that of the player's.

 

Bomber formations I think are capped at Regular skill level regardless of difficulty level, but I'd have to recheck. 

 

You're also going to see the highest density of flak guns at the target area on the hardest difficulty level. Density only affects ground objects around the front lines and the overall number of aircraft generated.

Edited December 21, 2022 by LukeFF

[III/JG52_Otto_-I-]

On 11/18/2022 at 8:27 PM, ACG_Ginger said:

Hi all

Just a quick observation since the last update. I fly only multiplayer so this might not affect single player ?

As anyone noticed an increase in the amount of PK's when flying lately?  It seems that you get no indication of hit's just killed instantly.  

We had a flight of 9 pilots last night on CB , 8 of them PK'd , so either the enemy have all got very good all of a sudden or there's a problem?

This is from a allied point of view, have also noticed wings coming off 109's like there going out of fashion.

 

@ACG_GingerI was started a thread months ago, about this issue of easy PK, after the patch 5.002b (nov. 2022), but some clap-men started laughing. My squad usually fly Germans, and now i´m amazed to read your post because, i´ve not notice the same easy-to-kill tendency in allied pilot-dolls. ..And i did several test with my squad comrades shooting me and i flying P-51, Yaks, etc and vice versa.
By the way, i am amazing how easy is to kill a pilot in game, inside a birdcage type canopy with several wide steel bars of a Bf-109G or K, compared with the impunity of others allied pilots, inside Plexiglas (methacrylate) bubble canopies, such us; Spifires, P-51, P-47, Yak-9, Tempest, etc.  who still being more difficult to kill, than a Bf-109´s pilot today in game. 
Thank for your post, supporting my claim.

[Charon]

19 minutes ago, III/JG52_Otto_-I- said:

By the way, i am amazing how easy is to kill a pilot in game, inside a birdcage type canopy with several wide steel bars of a Bf-109G or K, compared with the impunity of others allied pilots, inside Plexiglas (methacrylate) bubble canopies, such us; Spifires, P-51, P-47, Yak-9, Tempest, etc.  who still being more difficult to kill, than a Bf-109´s pilot today in game.

Why would the "wide steel bars" of the canopy make any difference? Those bars are thin metal, never designed as armor.

[Charon]

 

Quote

A single Me sat alone and hadn't yet reached the dispersal when my crew chief swing himself on to the wing. I could see by his face that something was wrong. I stopped the machine, opened the canopy and asked: "Where are the others, what's up with the aircraft on the field?"

"It's Fönnekold's machine. He's dead. None of the others have come back yet!"

I couldn't believe what I was hearing and asked again, but I received the same answer. I shut down the engine and climbed out with shaking knees. I sat on the wing and listened to my mechanic's account, according to which two aircraft had suddenly appeared over the airfield. "We thought they were two of ours returning early with engine trouble, realized the second machine was smoking heavily. But then we see the Me begin it's landing approach with the Mustang about 30 meters behind it. The only burst the American fired was on target. The Mustang broke away and the Bf 109 landed smoothly. It rolled to a stop. We hurried over and lifted Fönnekold out. He was dead. A single bullet had gone through the fuselage, pierced the armor plate and struck Fönnekold right in the heart."

The War Diary of Hauptmann Helmut Lipfert, p134

[purK]

14 hours ago, RNAS10_Mitchell said:

It's a real pity that there isn't a single beta tester that flys the game and has experienced this first hand.  One the devs have faith in, that could report this and other issues as they occur to the developers for investigation and possible action.  Would certainly be an improvement over the current environment where we are asked to submit tracks over an extended period and HOPE the devs notice and respond.  A real pity.?

You have no idea. I've spent 6,500 hours of my life playing/testing this game, I can tell you for certain there is little difference in the percentage of pilot kills compared to previous damage model updates. Several other testers also with thousands of hours in this game will tell you the same thing. It's only the few that barely play the game, or limit themselves to only flying one side that think there's an issue.. strange, surely there's no bias there. ?

As I mentioned in your other thread, and for everyone else, feel free to look at some statistics and you'll see it for yourself.

The Pilot Snipe% stat is nearly identical across the board, before and after the dm updates.

https://combatbox.net/en/

http://stats.virtualpilots.fi:8000/en/

http://stats.jasta5.org:8000/en/

[RNAS10_Mitchell]

Never mentioned any bias.  Believe there is no bias in fact.  Not seeing the pilot snipe numbers in your link, nor have I ever heard of it being collected as a statistic.   Help me out here.  Where is it?

Edited December 22, 2022 by RNAS10_Mitchell

[BraveSirRobin]

2 minutes ago, RNAS10_Mitchell said:

Never mentioned any bias.  Not seeing the pilot snipe numbers in your link, nor have I ever heard of it being collected as a statistic.   Help me out here.  Where is it?

He posted them in the other thread.  
 

 

Edited December 22, 2022 by BraveSirRobin

[Charon]

23 hours ago, Charon said:

On 12/20/2022 at 4:59 PM, Charon said:

T4:

He 111H-16 vs P-51D.

Radioman killed 3/6 times. No other crew casualties.

 

(Note: Remember, I'm testing what happens on the first pass, because that's most reproducible. Just because the attacker is driven off without inflicting crew casualties doesn't mean the plane isn't crippled).

 

Quote

T4.5:

He 111 H-16 vs P-51D (Ace)

I was surprised by the above result, so inserted an extra trial against an Ace P-51.

 

Gunner killed: 4/6

Waist Gunner: 4/6

Navigator: 3/6

Radioman: 3/6

Pilot: 0/6

 

2 fires.

 

Also a surprising result to me, frankly, but that seat back armor seemed to do its job.

 

I decided to re-run these two tests, but removing gunners as a factor by ordering them not to fire. Setup is otherwise the same as before.

 

Quote

T20:

He 111H-16 (cease fire) vs P-51D (average).

Waist Gunner killed 1/6. No other crew fatalities.

 

Okay, maybe the "average" AI is just a really bad shot.

 

Quote

T21:

He 111H-16 (cease fire) vs P-51D (ace).

Waist Gunner Killed 5/6

Radioman killed: 3/6

Ventral Gunner killed: 3/6

Navigator Killed: 1/6

Pilot Killed: 0/6 (really!)

 

On fire 3/6 trials.

 

In one trial there were no crew casualties because the P-51 shot out the elevator controls and then proceeded to miss all his shots once the Heinkel went into a dive.