reported Gunner Issues - Accuracy and engagement problems

[[HOUNDS] CptTrips]

I was wondering if the OP could look at this thread to see if it is a variation of their issue: 

Thanks in advance.

[[HOUNDS] CptTrips]

BTW, I modified my top gunner mission test to the A-20.  Same as the top gunner on the B-17.  Will not fire a round. Helpless. (See attached miz.)

 

 

 

TestA20Gunners_top.miz

[[HOUNDS] CptTrips]

16 hours ago, [16AGR] CptTrips said:

I'll amend my complaint on the waist gunner. 

 

Nope.  I'll have to re-grade my re-grade of the side gunner.  Fail. 

Moved the target out to 600ft.  Closer he was shooting just in off the front of the wind shield like a foot away from the cockpit.  I figure further out with dispersion that would more often hit.  But the error just increased as well. Now the aim point is well off the nose.

 

At 800ft he stopped shooting all together.  That's what I mean by lack of aggressiveness.  Anything under 400 yd (1200ft) he ought to be lighting up.

It's like he might be doing a blind lookup on angle of lead based on the targets airspeed and distance and aspect angle.  Not taking into account the relative velocity which in these tests is near zero.

 

Seems like it ought to calc the perfect aim point based of relative motion and distance and then apply a random offset from perfect aim based on skill level.

Even a day 1 recruit would shoot better than this, because a human would see it is miss and adjust his aim point.  Even a rank novice human gunner would hit in a couple of squirts because he would intuitively adjust based on where he saw his tracers go.

At 800 ft he should certainly still be shooting.  That is still prime kill zone.  If you were a human gunner, would you not shoot at this target?

 

 

 

 

[[HOUNDS] CptTrips]

 

Some gunners won't shoot at all.

Some gunners fail to shoot at ranges they should.  

Gunners that shoot, seem unable to calculate proper deflection lead and fail to correct their mistake.  (Probably why the chin and tail gunners seem better, because velocity vectors are mostly aligned and little to no deflection is needed.)

 

 

 

[[HOUNDS] CptTrips]

The ball turret is the worst gun position to lose effectiveness on.  It was the most fear by LW pilots.  (I would have said the tail gun, but what idiot would approach from dead 6 I guess. Maybe in RL the sun glare was more of a problem for top rather than belly gunner.)

The ball turret may have had up to 57% kill effectiveness of engaged targets.

Gunners would start shooting out to 600 yd (1800 ft vs 800ft in DCS).  Especially if the fighter is closing on them.  If a fighter is diving on you, if you pull the trigger at 600 yd the target will be at 400 yd by the time the bullets connect.

 

I fear the top turret as much for my normal pass.  If he would pull the freakin trigger, the top turret is shooting at me at the start of my dive before I've built up max speed.  The ball turret would be shooting me on egress after I have converted the dive into max speed. 

 

But gunners should be shooting out to 600 yd and should be achieving reasonably good hit percentage in their prime kill zone.

 

 

Edited June 19, 2023 by [16AGR] CptTrips

[[HOUNDS] CptTrips]

If he would shoot....

Effective range 600 yd (1800 ft) except high rate of closure targets then effective out to 1000yd. 

 

[[HOUNDS] CptTrips]

Sorry for the spam but I want to give you as many data points as I can for debugging.

 

Not sure what the difference on this one was.  I was piddleing around with the alt and suddenly the top gunner woke up and shot some rounds. 

Of course they were horribly mis-aimed, but at least he pulled the trigger.  Maybe debugging these different miz will reveal what is going on.

 

Miss.  Miss.  Miss. (See attached miz.)

 

 

TestBuffGunners_top_didshoot.miz

Edited June 19, 2023 by [16AGR] CptTrips

[Skewgear]

Tagging @NineLine and @BIGNEWY for this important bug report thread that includes tracks demonstrating the reported problem.

[Nealius]

A few years ago the B-17 gunners were too effective, shooting down everything you threw at them with laser accuracy. I assume the current result is a "nerf" to make them less formidable until at some point wider AI improvements can balance it out.

[[HOUNDS] CptTrips]

11 minutes ago, Nealius said:

A few years ago the B-17 gunners were too effective, shooting down everything you threw at them with laser accuracy. I assume the current result is a "nerf" to make them less formidable until at some point wider AI improvements can balance it out.

 

Soooo, Two Weeks?

[SUNTSAG]

Tested in latest OpenBeta using the locations you mentioned and no issues on my end. Simple example miz attached. Cheers.

wwiiembark.miz

[NineLine]

Thanks I will take a look. The gunners were tuned a while back based on range and a form of human delay. Its possible something has broken recently so I will review.

Understand you reported this when most of the team was out of the office, in many cases we are all still playing catch up. 

[NineLine]

On 6/13/2023 at 2:52 PM, [16AGR] CptTrips said:

I recently purchased the WWII asset pack and FW-190D and I was trying to set up a bomber intercept\escort mission but the results were very poor. It appeared the bombers were completely helpless and made little to no effort to defend themselves.

I ran some tests. I setup a lone bomber on a level course. Bluefor. I tried both Veteran and Ace but with similar results. Just run the mission and watch from external view. AI has control.

I positioned a Redfor 190 about 500ft off the bomber in a stable formation with the bomber. The fighter was set to not react to threat so he just maintained course.

I tried a couple of Tests:

1. Target at 6 o’clock level.

Result – Tail gunner was a little shy with the ammo but did fire and easily killed the target. Fine. (But it would be nice to be able to dial in some more aggressiveness for Veteran and Ace skill level.)

2. Target 12 o’clock level.

Result –Chin gunner was a little shy with the ammo but did fire and easily killed the target. Fine.

 

3. Target 9 o’clock level.

Result – Side gunner shot but only little squirts occasionally. He should not have even had to aim. Target was close enough to almost fill his view, yet some how he couldn’t hit. I watched for 15 minutes before he actually pinged it once. It made me sad to watch.

 

4. Target directly above 500 ft level.

Result – Top gunner never would fire. Perfect target. He should have shredded him. Never pulled the trigger. He didn't even swivel his turret to point at him.  It made me sad to watch.

 

6. Target directly below 500 ft level.

Result – Belly gunner never would fire. Perfect target. He should have shredded him. Never pulled the trigger. He did swivel the turret and point at him, but would not pull the trigger.  It made me sad to watch.

 

From what I see, my theory is the AI is refusing to shoot at targets with high angle of deflection. Even when the target is just sitting there. In terms of relative motion, it is motionless. A perfect target. Close enough to open the window and spit at.

In practical terms, it means the AI bombers will virtually ignore targets unless they are crawling directly up their 6 or coming directly head-on.  In virtually any other approach, they mostly refuse to shoot.

 

(See attached test missions.)

Thanks in advance.

 

TestBuffGunners_side.miz 7.39 kB · 4 downloads TestBuffGunners_tail.miz 7.38 kB · 3 downloads TestBuffGunners_chin.miz 7.38 kB · 3 downloads TestBuffGunners_top.miz 7.39 kB · 4 downloads TestBuffGunners_belly.miz 7.39 kB · 3 downloads

So I tried a few tests, but they are not great as you are starting 1 fighter and one bomber right up next to each other. Scenarios should be more realistic to get a better idea of the intent of the AI. For example, I tried the tail gunner mission and placed the aircraft farther back at greater speed so it was approaching the bomber. The fighter took some shots right to the face and peeled off. 

The AI is generally tuned for more realistic scenarios, in most cases fighters were going up against a large group of bombers and as such you would see a more saturated firing. Previously gunners would try to shoot too far out and just waste ammo. Now they are a little more conservative and tend to fire at good targets. If you don't fly evasively you will get shot in the face. I will check out the A-20 report next but I don't see a major issue here. 

If you want try and give a more realistic scenario and if you are seeing major issues, feel free to add the tracks here. 

[[HOUNDS] CptTrips]

17 minutes ago, NineLine said:

For example, I tried the tail gunner mission and placed the aircraft farther back at greater speed so it was approaching the bomber. 

You need to look a little deeper.

I graded the tail gun as "Fine".

Try the top and belly gun examples.  Are you telling me that is intended behavior?

If so, you might want to tell these guys it's correct as-is.  

 

Edited June 28, 2023 by [16AGR] CptTrips

[NineLine]

On 6/20/2023 at 10:08 AM, Skewgear said:

Tagging @NineLine and @BIGNEWY for this important bug report thread that includes tracks demonstrating the reported problem.

Did you report it in the Closed Beta section if you checked it out?

[NineLine]

7 minutes ago, [16AGR] CptTrips said:

You need to look a little deeper.

I graded the tail gun as "Fine".

Try the top and belly gun examples.  Are you telling me that is intended behavior?

If so, you might want to tell these guys it's correct as-is.  

 

 

Your tests are not done well, sorry. That is not a slight. I tried different tests from different angles with a more realistic scenario and I am not seeing an issue. Are you creating missions where you start a few feet from a bomber and its not working right for you?

The A-20G I have not checked but doesn't sound the same as what you are describing. 

Anyways, if the 190 takes shots to the face and peels off, do you not feel this is correct?

[[HOUNDS] CptTrips]

2 minutes ago, NineLine said:

Anyways, if the 190 takes shots to the face and peels off, do you not feel this is correct?

 

From the top and belly guns?  Can you post a track.  I've only ever seen the top gun shoot once.  It was off by 50 yards on a perfect target.  Are you saying that is intended behavior?

Have you tried the top and belly gun examples I posted.  Regardless of whether you like them.  If it's not too much trouble.  Just to make sure you understand what I am posting.

This thread also has a A-20 example.  Demonstrating a similar behavior to the b-17.  Is that intended behavior?

On 6/18/2023 at 11:23 PM, [16AGR] CptTrips said:

TestA20Gunners_top.miz 7.63 kB · 1 download

 

[NineLine]

Looks like you had the bomber set to passive defence, that seems to be impacting the rounds fired. 

See if you can open this mission and see what I have here. 

TestBuffGunners_tail_NLtop.miz

[[HOUNDS] CptTrips]

1 minute ago, NineLine said:

Looks like you had the bomber set to passive defence, that seems to be impacting the rounds fired. 

See if you can open this mission and see what I have here. 

TestBuffGunners_tail_NLtop.miz 8.88 kB · 0 downloads

 

I will be glad to,  but a couple of points.

1.  Passive defense is the b-17 default when placed.  I assume that is to tell it to maintain formation and defend with its guns instead of make crazy evasions.

2.  I had graded the tail gun as "fine".  I also noted he killed the target. Regardless of whether you like them, before responding please try the top, belly and side gun examples I posted and simply tell me if that behavior is correct as-is.

 

On 6/13/2023 at 4:52 PM, [16AGR] CptTrips said:

 

 

TestBuffGunners_side.miz 7.39 kB · 5 downloads   TestBuffGunners_top.miz 7.39 kB · 4 downloads TestBuffGunners_belly.miz 7.39 kB · 3 downloads

 

[NineLine]

The missions are not set as proper intercept missions, I cannot tell you if something like this is correct as is or not as it's just not a real situation. 

Please try my mission and let me know if it works better. 

The AI is not a super computer AI, it listens to what the Mission Editor is telling it to do in most cases, putting it in odd situations can yield odd reactions.

I will check if the default action is Passive defence, that probably isn't correct, but the B-17s are meant to fly in a bomber formation, this is how the are intended to function. 

[[HOUNDS] CptTrips]

1 minute ago, NineLine said:

The missions are not set as proper intercept missions, I cannot tell you if something like this is correct as is or not as it's just not a real situation. 

HAVE.  YOU. TRIED. THE. TOP. AND. BELLY. AND. SIDE. GUN. EXAMPLES. I. POSTED?

Please show me the courtesy of looking at what I posted.

Then we can discuss further.

 

[[HOUNDS] CptTrips]

 

OK.  Tried your mission.  

You showed the same results I did in my "incorrect" test.  Which is why I grade the tail gun as "fine" in the first post I am not sure you read.  

So now where are we?

 

[NineLine]

As I explained, you are setting up unrealistic expectations for AI behaviour. Your bomber is set to passive defence and your fighter is right feet from the bomber, this is not going to work as intended, sorry.  

[[HOUNDS] CptTrips]

1 minute ago, NineLine said:

As I explained, you are setting up unrealistic expectations for AI behaviour. Your bomber is set to passive defence and your fighter is right feet from the bomber, this is not going to work as intended, sorry.  

 

Are you refusing to look at the mission I posted?  Really? Seriously?  Just flat out refusing to even look?

[NineLine]

Just now, [16AGR] CptTrips said:

 

Are you refusing to look at the mission I posted?  Really? Seriously?  Just flat out refusing to even look?

I looked and I am explaining what is wrong with your missions.

null