Damage Model Bug Stomping - Part 2

[Pursuivant]

The results of my M2 .50 caliber vs. damage model tests are in.

The following report is an expanded and reworked version of my results described in this thread:

http://forum.1cpublishing.eu/showthread.php?t=229427

In the attached file, I've recorded my test results for .50 caliber M2 fire against every plane in the game, both from the front and from the R or L rear quarter at an approximately 45 degree angle.

The report is in a tab-separated text file which can be turned into a proper table using your favorite word processor or spreadsheet.

I've also included my mission files so you can repeat my experiments.

My test method was unchanged from my previous thread.

If any members of TD are reading this thread and care to fix some of the DM problems, I'd suggest the following planes as as being most critically in need of improvements: Bf-109 series, DC-3 series, Ju-88A series, Ki-45 series, Me-163B-1a, Me-321/323, MiG-3U, MiG-9 series, Mosquito series, P.11c, P-38 series, P-39 series, P-63, TBF series, Tu-2S, Wellington III, Yak series.

All I can say is that it's a testament to IL2's many excellent properties that fans of the series have put up with such shoddy and outright incorrect damage modeling work for so many years.

To be fair many planes do have solid damage models, but a significant number of planes seem to have DM based on anecdotal evidence, whining from fans, developer bias, or other non-scientific origins.

Additionally, many planes are missing damage models for certain parts, or have the "hooks" that trigger certain results in the wrong place.

I've noticed the following problems are nearly universal. These are limitations of the game and might not be fixable.

*All propellers are invulnerable to damage.

* Landing gear collapse effects often have plane settling slowly to one side rather than quickly falling.

* Flaps typically don't have damage models and can't be hit or broken due to damage.

* Wheels typically break off rather than going flat and are much less vulnerable to bullet damage than they should be.

* Many planes - especially the oldest planes in the game - have highly unrealistic breaking parts animations that make parts "jump off" or spin wildly - even when these animations aren't appropriate.

* Many planes have damage models which cause damage to an adjacent part, even when that part isn't hit. For example, ailerons might break due to wing damage, or rudder damage might cause damage to the vertical stabilizer.

* Many planes - particularly the oldest planes in the game - have explosion effects for fatal damage to the engine, wing or fuselage - even when there is nothing in those sections to trigger an explosion.

* Most planes don't have pilot/crew bail out following landing gear collapse.

* Wing damage models and damage textures are highly unrealistic. It is usually far too easy to break a plane's wing, and damage to wingtips often breaks the wing at mid-wing or even the wing-root. This makes planes in the game much more vulnerable than they should be.

* Wing breakage causes all outboard wing sections and engine nachelles to break as well. This is unrealistic.

* Most planes are highly vulnerable to breakage of control surfaces.

* There is no consistency in engine damage modeling. Often, the same engine mounted in a different plane will have very different damage properties. Or, engines of equivalent mass, power, compression, etc. will react differently to damage. For example, the Merlin and Packard Merlin engines are incredibly vulnerable, while the DB605 engine is a bit tougher. DB601 or DB605 engines built under license in Japan or Italy behave differently than German-built originals.

* There is no consistency as to the amount of damage which will trigger damage textures or cause breakage. Often, planes of similar size, mass & construction will vary considerably in their ability to absorb punishment. For example, the Ki-43 series is far more durable than the A6M series.

My report also includes a quick ranking of damage ratings, on a scale of 1 (incredibly fragile) to 7 (invulnerable). This is for convenience and as an aid to those whose native language isn't English.

These ratings are somewhat subjective, but are roughly as follows:

1 = Incredibly fragile - very few hits needed to damage/destroy part based on size/mass/construction. Extremely unrealistic.

2 = Very fragile - Just 1 second of sustained fire needed to damage/destroy part or excessively vulnerable to damage given size/mass/construction. Probably unrealistic damage results.

3 = A bit fragile - Just 1-2 seconds of sustained fire required to break/damage part, but results might be realistic for some aircraft.

4 = A bit tough - 3-5 seconds of sustained fire needed to break/damage part, but results might be realistic for some aircraft.

5 = Very tough - 5-10 seconds of sustained fire needed to break/damage part, but results might be realistic for some aircraft.

6 = Almost impossible - 11+ seconds of sustained fire needed to break/damage part. Results might be realistic for certain aircraft.

7 = Invulnerable. Can't be damaged or broken. Highly unrealistic.

[gaunt1]

Thanks for your efforts!
It is possible to upload the file in different format? its a bit hard (tiring) to read it in .txt

BTW, what I was talking about a since a while: obvious bias for some soviet planes, for example the VK-107 powered Yaks: "Hits to oil & coolant tanks & radiators not modeled."
VK-107 was extremely sensitive and fragile engine in RL, its weak points were exactly the oil and coolant. Is it by accident that exaclty these are not modeled? I doubt.

[Furio]

I’ve already expressed my appreciation for Pursuivant’s excellent job. Reading through the results, looks like some of the claims about planes being “too fragile” or “too tough” are not substantiated by facts. Of course, would be great to have the data in a better format, such as a spread sheet. Anyone can give a helping hand to Pursuivant?




VK-107 was extremely sensitive and fragile engine in RL, its weak points were exactly the oil and coolant. Is it by accident that exaclty these are not modeled? I doubt.

:)And now we have our conspiracy theory: 1C altered real data to allow Soviet planes to win WWII, while everyone knows that Germany won.;)

[gaunt1]

:)And now we have our conspiracy theory: 1C altered real data to allow Soviet planes to win WWII, while everyone knows that Germany won.;)

There are only a few soviet planes that are affected by this. For example, all VK-105 Yaks are very well made. (in terms of FM at least) Only La-5/F/FN/7 and Yak-9U are clearly overmodeled, but its for another discussion, it is offtopic here, so dont continue.

[Furio]

There are only a few soviet planes that are affected by this. For example, all VK-105 Yaks are very well made. (in terms of FM at least) Only La-5/F/FN/7 and Yak-9U are clearly overmodeled, but its for another discussion, it is off topic here
You’re right: it’s off topic. Pursuivant just demonstrated that some claims about overmodelling or undermodelling are subjective. Perhaps even yours. Surely not mine, as I don’t claim anything here.


so dont continue.
Agreed on that.:)

[dimlee]

Great work.

But, indeed, it's somewhat difficult to convert current .txt file into easy readable spreadsheet.

[dimlee]

Pursuivant,
One question.
While doing your tests, did you ever see a damage caused by the projectile which MISSED target aircraft?
I recall one "theory" about fragility or P-38 in IL2: allegedly, the whole space between fuselage, tail beams and stabiliser was included in damage model.
It was difficult to prove, obviously.

[majorfailure]

Great work.

But, indeed, it's somewhat difficult to convert current .txt file into easy readable spreadsheet.
At least in Excel it should work like this: Open, select seperared type of data, next select Tab as separator, and that should do it.

[Pursuivant]

Thanks for your efforts!
It is possible to upload the file in different format? its a bit hard (tiring) to read it in .txt

I'm limited in the formats that I can use to upload files to the forum and the amount of data I can upload. Zipped text files is the only way to do it. Not my preferred choice!

To get the files into something you can read, cut and paste the file into MS Word, then use the "convert text to table" function. Alternately, you can just cut and paste the text file into MS Excel.

Things ought to work similarly for other word processors or spreadsheets.

Originally, the table was created in MS Excel 2013.

VK-107 was extremely sensitive and fragile engine in RL, its weak points were exactly the oil and coolant. Is it by accident that exactly these are not modeled? I doubt.

I've seen no evidence of pro-Soviet bias in damage modeling. What I have seen is good evidence that the earliest planes in the game are some of the most sloppily modeled, regardless of nationality.

The Yak series of fighters is incredibly fragile. While the engines arguably are tougher than they "should" be, it hardly matters since their wings fall off so easily. If you're fighting a Yak, aim for the wing - at least if you've got .50 caliber or better guns.

The Yak-1, Yak-7 and Yak-3 series are quite fragile across the board - easy to flame engines, or to break control surfaces or wings. the only place that it's "overmodeled" is the fuselage - and that's probably correct. The Yak-9 series is much tougher - arguably exactly where it should be for all systems other than the engine - except for those incredibly fragile wings.

The only reason that the Yak series stands out as being "tough" is because some of the other important fighters in the game, like the Bf-109 series, are so fragile.

[Pursuivant]

I’ve already expressed my appreciation for Pursuivant’s excellent job. Reading through the results, looks like some of the claims about planes being “too fragile” or “too tough” are not substantiated by facts.

A very important fact to remember.

My ratings for how tough or fragile a particular plane's parts are are somewhat subjective.

They are primarily based on how many .50 caliber bullets are needed to break the part. That's objective. I've also described what parts will or won't break when exposed to unlimited gunfire, which is also objective.

But, my ratings are subjective because I've also included my opinions as to how tough that part "should" be - based on real life and compared to similar planes in the game.

For example, if a small, single-engined plane like the U-2VS requires 5+ seconds of sustained fire to break a wing or fuselage, then it's probably "over modeled" and gets a rating of "Very Tough" or "Incredibly Tough" (5 or 6 rating). But, if the same amount of gunfire is required to break the wing or fuselage of a heavy bomber, then I'd probably give that part a rating of "Very fragile" or "A bit fragile (2 or 3 rating).

[Pursuivant]

While doing your tests, did you ever see a damage caused by the projectile which MISSED target aircraft?

Using arcade mode, it's very easy to see if projectiles which miss the aircraft's physical model hit the damage model. You'll see "arrows" hanging in space where they hit the damage model.

There are a very few cases where the DM doesn't match the physical model, but they are mostly landing gear models and a few cases where the tail surfaces aren't properly modeled.

I saw no evidence that the P-38's damage model was so bad that it modeled the area between the horizontal stabilizer, tail booms and trailing wing edge as part of the plane.

What I did see is incredibly bad damage modeling that makes it far too easy to break the P-38's control surfaces, stabilizers or tail booms, and which doesn't model important engine systems.

[Pursuivant]

The limits on .zip files are far more generous than I realized.

For convenience, I have uploaded my original spreadsheet, plus new work in progress. But, the tabs for Breda-Safat and Breda MG aren't complete or correct.

The files are in MS Excel 2013 and PDF format.

[dimlee]

Using arcade mode, it's very easy to see if projectiles which miss the aircraft's physical model hit the damage model. You'll see "arrows" hanging in space where they hit the damage model.

There are a very few cases where the DM doesn't match the physical model, but they are mostly landing gear models and a few cases where the tail surfaces aren't properly modeled.

I saw no evidence that the P-38's damage model was so bad that it modeled the area between the horizontal stabilizer, tail booms and trailing wing edge as part of the plane.

What I did see is incredibly bad damage modeling that makes it far too easy to break the P-38's control surfaces, stabilizers or tail booms, and which doesn't model important engine systems.

Thanks. Probably "empty space damage" was a myth.
Agree with your last sentence. Those tails...

[majorfailure]

First: Thank you for the enormous effort.
It tells me some of my ingame perceptions are not too far off. Fw190s do obviously suffer from overdone damage from a few hits - while quite alot of hits are survivable, though the landing will be no fun.


The Yak series of fighters is incredibly fragile. While the engines arguably are tougher than they "should" be, it hardly matters since their wings fall off so easily. If you're fighting a Yak, aim for the wing - at least if you've got .50 caliber or better guns.

The Yak-1, Yak-7 and Yak-3 series are quite fragile across the board - easy to flame engines, or to break control surfaces or wings. the only place that it's "overmodeled" is the fuselage - and that's probably correct. The Yak-9 series is much tougher - arguably exactly where it should be for all systems other than the engine - except for those incredibly fragile wings.

The only reason that the Yak series stands out as being "tough" is because some of the other important fighters in the game, like the Bf-109 series, are so fragile.

I don't think the Yak family stands out for being tough because the Bf109 is so fragile. It stands out for being tough because it has an inline engine that is close to indestructible - while other planes, notably the P-40 and P-51 (though Bf109 not far off - and don't let me get started about the MiG-3 series - that thing burns no matter where and what hit it) have engines that regularly die from a few rifle caliber bullet hits. Wings may be too fragile, but Yaks are small and light, so damage should harm them more than bigger, tougher planes.
Bf109 not tough -yes. But maybe that is okay, there is nearly no part of that plane that can be hit without damaging something of value.
And compared to its "natural" counterpart - the Spit - it does not feel overly fragile IMHO.

[Woke Up Dead]

I don't think the Yak family stands out for being tough because the Bf109 is so fragile. It stands out for being tough because it has an inline engine that is close to indestructible - while other planes, notably the P-40 and P-51 (though Bf109 not far off - and don't let me get started about the MiG-3 series - that thing burns no matter where and what hit it) have engines that regularly die from a few rifle caliber bullet hits.

Yak engines don't die, but they do sometimes go to an "almost dead" state when hit where they only get 5-10% of normal power out of them; only good enough to extend your glide towards home a little bit. But to the outside observer it may look like the engine is not damaged because it's not smoking, the prop is still spinning, and the engine is making a little bit of "running" noise.

[Pursuivant]

Fw190s do obviously suffer from overdone damage from a few hits - while quite alot of hits are survivable, though the landing will be no fun.

Agreed. My impression of the FW-190 series is that "they're easy to bend, hard to break."


I don't think the Yak family stands out for being tough because the Bf109 is so fragile. It stands out for being tough because it has an inline engine that is close to indestructible

Respectfully, I disagree. Please play the mission I've uploaded, choosing any of the Yak series as your targets, and you'll see that their engines are actually quite fragile.

Because your posts are generally spot on, you had me worried that my tests were incorrect. So, I went back and tried shooting the Yak series fighters (Yak-1, Yak-3, Yak-9) from different angles. I repeatedly got the same result as I got before - about 1 second of .50 cal MG fire was sufficient to trigger engine failure and/or fire, and to convince the AI pilot to bail out. The Yak-9 is only tougher than the other Yaks because engine fires tend to quickly self-extinguish.

The important trick is to aim for the first third of the front fuselage - just behind the propeller spinner. Hits further back on the front fuselage - just ahead of the pilot - hit the guns or parts that aren't modeled.

Additionally, there seems to be some randomness or error built into damage results, so there will be times when some parts just won't break. You have to "fly" the same mission 4 or 5 times before you can start getting a sense of where the average lies.

other planes, notably the P-40 and P-51 (though Bf109 not far off - and don't let me get started about the MiG-3 series - that thing burns no matter where and what hit it) have engines that regularly die from a few rifle caliber bullet hits.

Agreed. Inline engines are far too vulnerable across the board.

Wings may be too fragile, but Yaks are small and light, so damage should harm them more than bigger, tougher planes.
Bf109 not tough -yes. But maybe that is okay, there is nearly no part of that plane that can be hit without damaging something of value.

But, I'm consistently seeing results where just 5-10 .50 caliber bullets scattered across a Yak's (or Bf-109's or Spitfire's) tail surfaces or wings will trigger collapse when it's sitting on the ground. Basically, the plane falls apart because you punched half a dozen or a dozen random 1 inch/25mm-wide holes into several square yards/meters of aluminum or plywood.

Yes, realistically, those 5-10 bullets might break a part, but only if they all hit in the same place while the aircraft was pulling high-G maneuvers. Scattered across a wing or control surface, that sort of damage result makes no sense.

[Pursuivant]

Yak engines don't die, but they do sometimes go to an "almost dead" state when hit where they only get 5-10% of normal power out of them; only good enough to extend your glide towards home a little bit.

My tests can't test for reduced power output, other than determining if the prop speed is reduced.

But, my impression is that many engines - particularly radials - behave as you describe. It's not just Yaks.

[gaunt1]

A while ago, I did some tests against a Yak-9U. I flown a La-7 (needed something that is really uber, that is superior to Yak9U in every aspect), and I deliberately shot at the engine of the Yak. Just a few hits were enough to make the engine emit a thick black smoke. I continued with the test, and did not shoot down the yak. I got bored following its tail around 8-10 minutes, and eventually downed it, but during this time, I did not notice any decrease in the performance of the Yak despite heavy engine damage!
It was a few years ago, around 4.11, dont know if it was fixed or not since.

[majorfailure]

Respectfully, I disagree. Please play the mission I've uploaded, choosing any of the Yak series as your targets, and you'll see that their engines are actually quite fragile.

Because your posts are generally spot on, you had me worried that my tests were incorrect. So, I went back and tried shooting the Yak series fighters (Yak-1, Yak-3, Yak-9) from different angles. I repeatedly got the same result as I got before - about 1 second of .50 cal MG fire was sufficient to trigger engine failure and/or fire, and to convince the AI pilot to bail out. The Yak-9 is only tougher than the other Yaks because engine fires tend to quickly self-extinguish.

The important trick is to aim for the first third of the front fuselage - just behind the propeller spinner. Hits further back on the front fuselage - just ahead of the pilot - hit the guns or parts that aren't modeled.

Additionally, there seems to be some randomness or error built into damage results, so there will be times when some parts just won't break. You have to "fly" the same mission 4 or 5 times before you can start getting a sense of where the average lies.

Last campaign I played I shot down a few (hundred or so) early war Yaks of all colours - and it may be my aiming, but I rarely got their engine. Pilot dead, controls shot out, structural damage leading to crash were IMHO the most common deaths, structural total failure of fuselage or wings also common.
Fuel tanks punctured and engine losing a little or some power, smoking, total kaboom, seen that happen.

Vs. P-40, engine dead is among the most common deaths, vs. Bf109&Mc202 too. Vs MiG-3 it is about the only death -though I sometimes think there must be a fuel tank behind the engine that burns like that, vs. LaGG it happens at least to be noticed.

Just form feeling, I'd say Yak has highest initial turn rate of all of these, high speed, good roll, small plane. I may be aiming for the same point I would get good hits on a P-40 - but I may end up a little too far and hit pilot/tail section, or wings (if aim not on center but off left/right). So maybe in the end it is ME that is responsible for not getting engine dead regularly on a Yak.


But, I'm consistently seeing results where just 5-10 .50 caliber bullets scattered across a Yak's (or Bf-109's or Spitfire's) tail surfaces or wings will trigger collapse when it's sitting on the ground. Basically, the plane falls apart because you punched half a dozen or a dozen random 1 inch/25mm-wide holes into several square yards/meters of aluminum or plywood.

Yes, realistically, those 5-10 bullets might break a part, but only if they all hit in the same place while the aircraft was pulling high-G maneuvers. Scattered across a wing or control surface, that sort of damage result makes no sense.
Put like that it seems weak, though from ingame performance, sitting in a Yak as well as shooting at Yaks I never percieved them as structurally especially weak except when you try to dive them.

[Pursuivant]

Last campaign I played I shot down a few (hundred or so) early war Yaks of all colours - and it may be my aiming, but I rarely got their engine. Pilot dead, controls shot out, structural damage leading to crash were IMHO the most common deaths, structural total failure of fuselage or wings also common.

Due to the distances and angles at which I was firing, and because I'm fairly confident that IL2 models armor correctly, I didn't test crew vulnerability. (Realistically, at <50 m a .50 caliber AP shell should penetrate at least 26-28mm of armor.) Other than that, your experience squares with my tests for the Yak-1 & Yak-7.

Fuel tanks punctured and engine losing a little or some power, smoking, total kaboom, seen that happen.

In the earlier planes in the game, explosion is a very common "fatal damage" effect, even for parts of the plane that realistically wouldn't explode. Fuel tanks will also explode if they take enough damage - which might or might not be realistic depending on the aircraft.

Vs. P-40, engine dead is among the most common deaths, vs. Bf109&Mc202 too. Vs MiG-3 it is about the only death -though I sometimes think there must be a fuel tank behind the engine that burns like that, vs. LaGG it happens at least to be noticed.

Most inline engines will have a big oil tank just behind the engine. That tank is typically quite vulnerable to gunfire and will sometimes even explode.

My tests show that the MiG-3 series is about as vulnerable to wing and tail damage as the Yak 1, 3 or 7 series. The engine is a bit tougher than the Yaks, but not by much.

So maybe in the end it is ME that is responsible for not getting engine dead regularly on a Yak.

Fly on arcade mode using the QMB to test your gunnery accuracy. In the absence of realistic damage textures which show exactly where each bullet hits, that's the only way to do it.


Put like that it seems weak, though from ingame performance, sitting in a Yak as well as shooting at Yaks I never percieved them as structurally especially weak except when you try to dive them.

Keep in mind that I was shooting at aircraft sitting on the ground -1 G of positive G force, no wind resistance. Other than recording breaking parts, I have no way of knowing how damage effects affect a flying aircraft.

[Furio]

Once, I was shot down by a couple of bullets in the nose of my plane. Tac -tac, and the engine began immediately to emit creaking and squeaking sounds, losing power and stopping after a couple of minutes… The gun was an Mg15 (Stuka gunner) and my plane an armoured Il2… Hardly realistic.

In another occasion, while flying with a cannon armed La5, I set on fire a PZL P11. It burned furiously, and went ahead for a loooong time flying excellent evasive manoeuvres without any hint of diminished performances (or pilot’s cooking).

I think sometimes strange things happen, and are almost certainly result of random errors, having nothing to do with damage models. I think that reports about single events are of little value. What we need are consistent experiments, repeatable by others, like a sort of peer review. Pursuivant demonstrates that can be done, and that results are sometimes surprising.

I believe that further test will bring more surprises. And also think that Il2 remains a fantastic game, still a lot of fun after so many years, a monument to original developers and to the continuing work being done today.

[majorfailure]

Once, I was shot down by a couple of bullets in the nose of my plane. Tac -tac, and the engine began immediately to emit creaking and squeaking sounds, losing power and stopping after a couple of minutes… The gun was an Mg15 (Stuka gunner) and my plane an armoured Il2… Hardly realistic.

Can get even worse. PK from Bf110 rear gunner in an IL-2, that's as impossible as it gets.

I think sometimes strange things happen, and are almost certainly result of random errors, having nothing to do with damage models. I think that reports about single events are of little value. What we need are consistent experiments, repeatable by others, like a sort of peer review. Pursuivant demonstrates that can be done, and that results are sometimes surprising.

Yes. Totally agree. Once is almost never. And data is always better as impressions&conjecture, I think I inadvertently proved that with my last post.

I believe that further test will bring more surprises. And also think that Il2 remains a fantastic game, still a lot of fun after so many years, a monument to original developers and to the continuing work being done today.
I'll drink to that. Cheers!

[Janosch]

Can get even worse. PK from Bf110 rear gunner in an IL-2, that's as impossible as it gets.


So a pilot can die if hit by a machine gun? Madness!

[Pursuivant]

The gun was an Mg15 (Stuka gunner) and my plane an armoured Il2… Hardly realistic.

I beg to disagree. The IL2's armor makes it very tough, but not invulnerable. There are gaps in its armor which you can exploit.

It is quite vulnerable to hits to the oil cooler and coolant radiator. The sort of damage you describes perfectly matches the effects of a hit to either of those.

Additionally, there are gap in the armor where the exhaust stacks emerge from the engine - so a hit there will quickly stop the engine.


In another occasion, while flying with a cannon armed La5, I set on fire a PZL P11. It burned furiously, and went ahead for a loooong time flying excellent evasive manoeuvres without any hint of diminished performances (or pilot’s cooking).

The PZL P.11c is one of the worst modeled planes in the game. Only the Me-323 is worse. While my tests can't test for effects on crew (other than death, injury or bail-out), it wouldn't surprise me in the least if the modeling for the P.11c doesn't include instructions that cause fire damage to damage the pilot.

That said, the P.11c had the unusual ability to jettison the fuel tank in event of a fire. Realistically, if you're in a burning P.11c, you dump the fuel tank and try to glide away from the fight.



I think sometimes strange things happen, and are almost certainly result of random errors, having nothing to do with damage models.

Occasionally, I'll load up the game and discover that parts that broke before refuse to break, or are harder to break. There's some randomness deliberately built into damage models - and there should be - but occasionally I think that there's just an error in the program.

I think that reports about single events are of little value. What we need are consistent experiments, repeatable by others, like a sort of peer review.

Speaking of which, I welcome peer review. Don't just take my word about damage results, please try to reproduce them!

[Pursuivant]

Can get even worse. PK from Bf110 rear gunner in an IL-2, that's as impossible as it gets.

Not impossible, just unlikely.

The IL2 had ~5mm of armor around the cockpit sides and just behind the canopy, plus armor glass at the front and a heavier armor plate behind the pilot's head.

At close range (i.e., 100 m or less) a 7.62mm AP bullet can punch through the side armor, even at an angle.

This thread nicely explains penetration of German 7.62mm bullets vs. armor plate and aircraft aluminum:

http://forum.axishistory.com/viewtopic.php?t=72990

[majorfailure]

So a pilot can die if hit by a machine gun? Madness!
Please. I did nowhere state it is uncommon or unconvincing that a pilot gets hit or killed by rifle caliber rounds. Only in an fully armoured canopy it is.
Not impossible, just unlikely.

The IL2 had ~5mm of armor around the cockpit sides and just behind the canopy, plus armor glass at the front and a heavier armor plate behind the pilot's head.

At close range (i.e., 100 m or less) a 7.62mm AP bullet can punch through the side armor, even at an angle.

This thread nicely explains penetration of German 7.62mm bullets vs. armor plate and aircraft aluminum:

http://forum.axishistory.com/viewtopic.php?t=72990
Should have been clearer at the first try. From around 150-200m, dead ahead. Through a spot were I thought there would be enough armour. And I got hit a few more times, also by Bf110 rear gunner, though not fatally. All from dead ahead. Ther is either a small gap in the armour of real IL-2, that is modeled correctly, or the armour model has a small gap.

But I think we digress at least a little. I hope I have time to do a more in detail review of test results this weekend.

[Pursuivant]

Should have been clearer at the first try. From around 150-200m, dead ahead. Through a spot were I thought there would be enough armour. And I got hit a few more times, also by Bf110 rear gunner, though not fatally. All from dead ahead. There is either a small gap in the armour of real IL-2, that is modeled correctly, or the armour model has a small gap.

I've noticed in previous tests that there can be gaps in coverage of forward armor or armor glass. In particular, otherwise well-armored planes can be vulnerable to damage 12 o'clock high, or occasionally 12 o'clock low.

Sometimes this is realistic, sometimes there is an artificial gap due to the way that IL2 models the intersection of armor plate and armor glass ahead of the cockpit.

Looking at the armor diagrams for the IL2, I think that there is a gap in the protection for the pilot though the canopy from 12 o'clock high and from 1-2 o'clock & 10-11 o'clock level or high. After all, the canopy couldn't be armored that heavily because it had to be light enough to slide or jettison in a hurry.

Additionally, at 100-200m, a 7.62mm AP bullet might have just enough energy to punch through 5 mm of armor plate, especially if it was a head-on shot where the bullet's speed is combined with the speed of the oncoming aircraft.

But, I don't know how sophisticated IL2's armor penetration algorithms are. So, I don't know if it takes things like target speed or angle of impact into account when determining penetration or damage.