cannot reproduce WW2 AI Gunners...

[rkk01]

Well, at least DCS AI is consistent... consistently spoils the sim in many respects...!

 

Both aerial gunners and land / ship based gunners are way too accurate - especially as radar controlled gunnery was either absent (eg on merchant ships) or in its infancy.

 

Their ability to lay fire onto fast moving and maneuvering targets is beyond uncanny

 

[Mr_sukebe]

Agreed, they’re still far too good, despite a bit of a downgrade not that long ago.

[lt_d4n]

 

Not just aerial gunners, all AI has excessive precision, even in tight maneuvers. needs to be balanced.

in a dogfight, ia always be in your six

Edited February 2, 2021 by lt_d4n

[NineLine]

Sorry I need a track, indeed if you fly in the wrong way to attack a bomber, they will rip you up, but if you attack appropriately and not 1 vs 1 or 1 vs many, its generally not an issue.

[Nealius]

Even 1 vs 1 or 1 vs a few shouldn't rip you up. Leading an aircraft of unknown distance and speed at high angles of motion with nothining but an eyeball and arm muscles is not as easy as our AI gunners appear to make it seem. Historically bomber gunners were never this effective in real life. 

Edited February 11, 2021 by Nealius

[DD_Fenrir]

This is a tricky subject as there seems to be very very little data to actual model from; I have not seen one single document from any air force that specifically analyzes the effectiveness of the air-to-air gunner in WW2.

 

The truth I feel is that they could more effective than sometimes given credit; many a Hurricane or Spitfire pilot rolling in behind a German bomber confidently thinking "you've got one gun, but I have eight!" had their balloon of assuredness well and truly popped by a well trained gunner with armed with only a single 7.92mm.

 

Bob Tuck had his aircraft hit by a Bf110 rear gunner; Peter Townsend's Hurricane was shot down attacking a Do-17 from the rear. Ginger Lacey was obliged to bailed out after being hit by return fire from a Heinkel He 111. So too Paul Richey during the Battle of France. 

 

Needless to say, the Luftwaffe during the Defence of the Reich Campaign didn't refine the Head-on attack technique or develop Sturmböcke for no reason; the firepower of the B-17s was at the very least respected and apparently enough to drive the Jadgwaffe to some fairly serious thinking.

 

That being said, I believe the OP is correct. I suspect at longer ranges, high angle off and high speeds the gunners are too accurate, particularly at the "average" skill level. The trouble is proving it.

 

It might be interesting to run an analysis of the aerial gunnery hit rates in sims for human controlled gunners against aerial targets to provide some kind of benchmark; admittedly they do not suffer some of the vestibular and 'G' effects that would a real gunner, but for level bombers these are minimised. Also the sim gunner, unless in VR, has little/no depth perception, though out to 2-300 yards this would probably be of little value in real life. 

 

Have them work through several scenarios of aerial targets attacking from various aspects and various speeds and see what their hit percentages are; then see whose are best and use any apparent stratification to begin to work out a model, based on target parameters and gunner skill level.

 

This, I feel, is the only way that this problem could be adequately addressed.

[NineLine]

14 hours ago, Nealius said:

Even 1 vs 1 or 1 vs a few shouldn't rip you up. Leading an aircraft of unknown distance and speed at high angles of motion with nothining but an eyeball and arm muscles is not as easy as our AI gunners appear to make it seem. Historically bomber gunners were never this effective in real life. 

 

No, but if you don't fly defensively, and the turrets get a bead on you, then yes. Understand I said if you are the lone aircraft AND flying in a way that the gunners can get a good track on you. They are quite a bit less accurate than they were before.

[Nealius]

20 hours ago, DD_Fenrir said:

This is a tricky subject as there seems to be very very little data to actual model from; I have not seen one single document from any air force that specifically analyzes the effectiveness of the air-to-air gunner in WW2.

 

 

I would think documents on confirmed/claimed kills by B-17s would suffice. I was watching a period documentary the other day, I think about the Schweinfurt raids, where the B-17 crews claimed upwards of a hundred fighters downed, but in reality was 20 or fewer. Among hundreds of bombers over a multi-hour mission. That's a low percentage of accuracy.

Edited February 12, 2021 by Nealius

[DD_Fenrir]

But Nealius, that's not useful data.

 

How many other German aircraft engaging the bombers suffered damage but landed, or crew injury that didn't render the aircraft a loss?  Those would still count as percentile hits, no?

 

At some point, you want to calculate the probability of whether the gunner will hit the target; whether that target is brought down or not depends on (a) the number of rounds landed and (b) the Damage Models calculations on the damage those rounds have done. Ergo whether the target is downed or not is kind of irrelevant when discussing calculations for gunner accuracy.

 

You want to do it right you need a controlled test. 

 

 

[Nealius]

I thought we were discussing percentile kills, not hits. Going up against AI gunners in DCS typically results in the former, unless you're flying something extremely rugged like the Anton. Even then you'll be dead on your second or third pass. This data is also more useful than zero data and 100% game developer guesswork. Furthermore the whole idea of the effectiveness of bombers formations to maximize accurate gunnery was disproven by the fact that there were heavy losses until escorts were finally used. It's a logical conclusion that they weren't that effective. 

Edited February 12, 2021 by Nealius

[122sqn Bruv]

I'd also like to know how people are attacking them.   If your going in solo vs a formation every single gunner is going to be pointing at you.  Multiple attackers and things get much easier. 

 

Before the changes It was near impossible to come out unscathed,  at the moment things seem more realistic IMO.    

[philstyle]

On 2/12/2021 at 12:18 PM, Nealius said:

 

I would think documents on confirmed/claimed kills by B-17s would suffice.

 

Not unless you know how many fighters got within firing range of the B17s.
If a total of 40 fighters got to the B17s, and 20 got shot down that's a 50% kill rate., i.e. half of all the fighters that could have been shot down, were.
If 400 fighters got withing firing range, and 20 were lost, that's a very different statistical story.

As it was, a total of around 350 to 400 fighters were sent up to intercept Schweinfurt-Regesnburg II. Around 25 were lost. But there is no information regarding how many interceptors actually made contact with the bombers.

If anything the massive over-claiming by B17 gunners does suggest that many, many gunners were often firing at the same aircraft and all making a claim when one went down. There is precious little information available regarding statistical changes of getting hit by a gunner when attacking B17 formations.

However, Battle of Britain anecdotes vary wildly, with some accounts where a single bomber was able to fight off pairs of fighters, or other examples where bomber formations were decimated by a single squadron of interceptors.

Finally, we almost NEVER get to test a realistic WW2 bomber v fighter scenario with 40 or 50 fighters intercepting a 5 or 10 mile long stream of 150 bombers. Most of the time in DCS it is 1 or 2 players taking on a group of 6 to 10 bombers.
 

On 2/15/2021 at 9:07 PM, No.119_Bruv VK-B said:

I'd also like to know how people are attacking them.   If your going in solo vs a formation every single gunner is going to be pointing at you.  Multiple attackers and things get much easier. 

 

Before the changes It was near impossible to come out unscathed,  at the moment things seem more realistic IMO.    

 

 Head on attacks against bombers are almost successful 100% of the time for me. It is no surprise that this tactic was favoured by interceptors.

Edited February 17, 2021 by philstyle

[NineLine]

Its always going to be a balance of realism and gameplay, there is no way around it. But we cant really tune based on 1 vs 1 or 1 vs 20 scenarios. 

[StevanJ]

On 2/16/2021 at 10:53 PM, philstyle said:

Not unless you know how many fighters got within firing range of the B17s.
If a total of 40 fighters got to the B17s, and 20 got shot down that's a 50% kill rate., i.e. half of all the fighters that could have been shot down, were.
If 400 fighters got withing firing range, and 20 were lost, that's a very different statistical story.

As it was, a total of around 350 to 400 fighters were sent up to intercept Schweinfurt-Regesnburg II. Around 25 were lost. But there is no information regarding how many interceptors actually made contact with the bombers.

If anything the massive over-claiming by B17 gunners does suggest that many, many gunners were often firing at the same aircraft and all making a claim when one went down. There is precious little information available regarding statistical changes of getting hit by a gunner when attacking B17 formations.

However, Battle of Britain anecdotes vary wildly, with some accounts where a single bomber was able to fight off pairs of fighters, or other examples where bomber formations were decimated by a single squadron of interceptors.

Finally, we almost NEVER get to test a realistic WW2 bomber v fighter scenario with 40 or 50 fighters intercepting a 5 or 10 mile long stream of 150 bombers. Most of the time in DCS it is 1 or 2 players taking on a group of 6 to 10 bombers.
 

 

 Head on attacks against bombers are almost successful 100% of the time for me. It is no surprise that this tactic was favoured by interceptors.

 

https://www.quora.com/Did-the-gunners-aboard-B-17-bombers-in-WWII-manage-to-shoot-down-enemy-fighters
 

Warthunder and Il2 had the same problem at the start before bombers gunners were 'buffed' by complaints in the Forums.

I think the only time we'll ever know is if we get a chance to use the gunners positions and 'try to shootdown planes' ourselves.
Then we can get the best analytics towards real figures.
Right now, there doesnt seem to be any research towards accuracy.

Although I 100% agree that the gunners seem way too high in accuracy.

During a mission I had 1 'Rookie' A-20 utterly destroy my squad of 'Ace' 109's (3 of them strafing at 400 knots).
And ill i did, was drift low and behind it, and take out the engines with a few rounds each engine.

 

[Mr_sukebe]

Well one solution would be to allow players to enable a variable setting about gunner accuracy, from say the current "ace" level, down to "completely hopeless", i.e. significantly worse than the rookie level.  You can then leave the decision to the mission maker. 

[NineLine]

First I need to see these issues with tracks before I go to the team and make the gunners dumber yet. 

[philstyle]

3 hours ago, StevanJ said:

 I think the only time we'll ever know is if we get a chance to use the gunners positions and 'try to shootdown planes' ourselves.
Then we can get the best analytics towards real figures.
Right now, there doesnt seem to be any research towards accuracy.
 

My experience from manning the AA guns online (using combined arms, on Storm of War server) is that as a player I can be far more accurate than the AI is. Aircraft are really not that hard to track if they attempt multiple passes over a ground target.
Despite the howls of displeasure about the AA being too tough, in my opinion the AI gunners on the ground are not as smart or accurate as a human player.

3 hours ago, StevanJ said:

During a mission I had 1 'Rookie' A-20 utterly destroy my squad of 'Ace' 109's (3 of them strafing at 400 knots).
And ill i did, was drift low and behind it, and take out the engines with a few rounds each engine.

 

I've seen how the AI set up for attacks on bombers. They fly straight and level at  the bomber's 6 O'clock at moderate speeds. No jinking, no evasive maneuvers - just target presentation on a predictable course.
That's not a demonstration of ace gunnery. That's just evidence of crap attacking play by the AI. You'd have to be a donkey to miss those AI.

Edited February 22, 2021 by philstyle

[Mogster]

1 hour ago, philstyle said:

My experience from manning the AA guns online (using combined arms, on Storm of War server) is that as a player I can be far more accurate than the AI is. Aircraft are really not that hard to track if they attempt multiple passes over a ground target.
Despite the howls of displeasure about the AA being too tough, in my opinion the AI gunners on the ground are not as smart or accurate as a human player.

I've seen how the AI set up for attacks on bombers. They fly straight and level at  the bomber's 6 O'clock at moderate speeds. No jinking, no evasive maneuvers - just target presentation on a predictable course.
That's not a demonstration of ace gunnery. That's just evidence of crap attacking play by the AI. You'd have to be a donkey to miss those AI.

 

if you read pilots accounts like “The Big Show” mostly they don’t admit to being scared by encounters with enemies in aircraft but they are terrified of enemies on the ground pointing a 20mm light Flak 38 at them. Light flak was very dangerous and pretty much impossible to counter.

[NineLine]

9 hours ago, Mogster said:

if you read pilots accounts like “The Big Show” mostly they don’t admit to being scared by encounters with enemies in aircraft but they are terrified of enemies on the ground pointing a 20mm light Flak 38 at them. Light flak was very dangerous and pretty much impossible to counter.

Sure, but if you had some cannon rounds land next to your head from a fighter, I imagine you would be a little shook

[Nealius]

11 hours ago, philstyle said:

I've seen how the AI set up for attacks on bombers. They fly straight and level at  the bomber's 6 O'clock at moderate speeds. No jinking, no evasive maneuvers - just target presentation on a predictable course.
That's not a demonstration of ace gunnery. That's just evidence of crap attacking play by the AI. You'd have to be a donkey to miss those AI.

 

 

Which AI? I've seen the Antons do boom and zoom, doing corkscrews around the bomber formations. Granted the last I saw this was way back when the A-8 was first released. I've seen K-4s come up on their dead six and take massive losses, but I also had those AI set to the dumbest possible level.

Edited February 23, 2021 by Nealius