Will the AAA be nerfed for the Pacific

[Gunfreak]

Reading shattered sword about Midway. 

It appears Japanese AAA was close to useless, during American attack on the Japanese ships the anti aircraft fire only shot down like 3 planes(out of 150 attackers over several hours) 

Now anyone who has played ww2 DCS knows, that's not how AAA works in DCS. You add 150 attacking planes against a dozen ships with AAA and you'll have about 150 dead planes.

Edited December 15, 2022 by Gunfreak

[Mr_sukebe]

There's plenty of other similar wishlist requests on the subject, but you're not wrong.

[Gambit21]

PTO is a priority for the Devs...so maybe we should just see what shakes out.

AAA being too accurate is a common problem since the AI always knows where you are, and programming it to miss is not the easiest thing to do.

[rkk01]

4 hours ago, Gambit21 said:

AAA being too accurate is a common problem since the AI always knows where you are, and programming it to miss is not the easiest thing to do.

This - precisely 

 

[Lurker]

5 hours ago, Gambit21 said:

PTO is a priority for the Devs...so maybe we should just see what shakes out.

AAA being too accurate is a common problem since the AI always knows where you are, and programming it to miss is not the easiest thing to do.

Actually it is very easy, it's just that Eagle Dynamics will not do it. There is a big difference between the two. 

Edited December 19, 2022 by Lurker

[Gambit21]

On 12/19/2022 at 12:29 AM, Lurker said:

Actually it is very easy, it's just that Eagle Dynamics will not do it. There is a big difference between the two. 

 

Says you...I know differently.

 

 

[peachmonkey]

2 hours ago, Gambit21 said:

Says you...I know differently.

 

 

so, what is your take on it ? Why does AA continue to be an aimbot?

[Lurker]

9 hours ago, Gambit21 said:

Says you...I know differently.

 

 

Have you done any programming or probability mathematics? You are right, I have no idea how Eagle Dynamics have programmed their AI, but if this kind of thing is hardcoded into their engine and can't be changed easily then this is extremely poor coding.  

Edited December 21, 2022 by Lurker

[Gambit21]

13 hours ago, Lurker said:

Have you done any programming or probability mathematics? You are right, I have no idea how Eagle Dynamics have programmed their AI, but if this kind of thing is hardcoded into their engine and can't be changed easily then this is extremely poor coding.  

 

I've worked with flight sim developers...I'll leave it there.

The computer always knows where you're at...getting it to pretend like it doesn't in a believable 'hit sometimes, miss sometimes...how much, when, under what conditions" etc etc is not easy. Go ahead and PM then with the solution if you have it.

 

 

 

[peachmonkey]

13 minutes ago, Gambit21 said:

I've worked with flight sim developers...I'll leave it there.

The computer always knows where you're at...getting it to pretend like it doesn't in a believable 'hit sometimes, miss sometimes...how much, when, under what conditions" etc etc is not easy. Go ahead and PM then with the solution if you have it.

 

the computer still needs to send a round to your location, regardless if it knows where you are. It still needs to calculate the trajectory of the user's AC vs. the round's and 'lead' the AC. How many variables go in to the calculation depends on the simulated world and its physics. Correcting for the wind is one of those variables. A human tending to the AA wouldn't know the exact wind speed at the location of the AC, a human would only know what's in his direct reach, i.e. the wind speed at the ground. Yes, there are forecasting reports that highlight the wind speed at different altitudes, but those are all estimates. So, how difficult is it to introduce a round calculation that is done based on the what the human would know instead of knowing 'everything' ? Instead of knowing the AC's ground/air speed at all times/locations introduce an estimation table (just like in the old days), etc etc.. There is sooo much fuzziness that could be added to the AA in DCS imho that I find your response skeptical.  Why are there so little fuzziness in the calcs? I may have dumbed it down too much, so forgive me if I miss some critical factor, but i'm definitely willing to learn about it...

[Mr_sukebe]

+1 on the above.

A few thoughts:

•  Have the skill level change the multiplication factor used with the probability bell curve standard deviation
•  Includes variables such as:
  • Is the spotter asleep, just looking in the wrong direction
  • as already mentioned, wind, which is itself NOT consistent in it's speed
  • Estimation of range
  • Estimation of height
  • Estimation of the targets speed/direction
  • Delay time between spotting and firing.  This should be small for a simple unit.  However, if you read about the bigger systems, that might include:
    • Multiple spotters, who'll hand off to,
    • Someone to conduct the calculations
    • Someone to turn the handles to make the gun turn/elevate (sometimes different people)
    • Someone to make the decision on when to fire

I've read of some units having 10 different persons involved, some of whom might be sleepy, poorly trained etc, resulting in errors and then delays between spotting and firing of 30-90 seconds

 

From what I've read about the big AA guns, e.g. FLAK41, I've taken away the impression that they tended not to shoot below probably 10,000' and when they did, to shoot at a static point in the sky, where they believe that a large formation is likely to fly through.  

The way we have FLAK guns shooting at lone fighters that are on the deck with seemingly God like ability to predict their flightpath is IMO one of the worst elements of DCS right now.

[Gunfreak]

9 minutes ago, Mr_sukebe said:

+1 on the above.

A few thoughts:

•  Have the skill level change the multiplication factor used with the probability bell curve standard deviation
•  Includes variables such as:
  • Is the spotter asleep, just looking in the wrong direction
  • as already mentioned, wind, which is itself NOT consistent in it's speed
  • Estimation of range
  • Estimation of height
  • Estimation of the targets speed/direction
  • Delay time between spotting and firing.  This should be small for a simple unit.  However, if you read about the bigger systems, that might include:
    • Multiple spotters, who'll hand off to,
    • Someone to conduct the calculations
    • Someone to turn the handles to make the gun turn/elevate (sometimes different people)
    • Someone to make the decision on when to fire

I've read of some units having 10 different persons involved, some of whom might be sleepy, poorly trained etc, resulting in errors and then delays between spotting and firing of 30-90 seconds

 

From what I've read about the big AA guns, e.g. FLAK41, I've taken away the impression that they tended not to shoot below probably 10,000' and when they did, to shoot at a static point in the sky, where they believe that a large formation is likely to fly through.  

The way we have FLAK guns shooting at lone fighters that are on the deck with seemingly God like ability to predict their flightpath is IMO one of the worst elements of DCS right now.

It's possible to set min and max altitude on AAA now.

From what I understand the 88mm generally only fired from 3000 meters and above (so 10 000k feet as you say) and it's up to mission/campaign editors to make sure that's correct. 

But the all ground fire is still way too accurate, be it flak, automatic cannons, MGs or even rifles. 

[Lurker]

Sure the application always knows your position, but you can always artificially introduce errors, or rather probabilities into every known variable. It's could be very similar to the "dice roll" of missile chaff resistance. In fact it would be more robust because you can "fudge" every variable. I mean even a simple probability equation of: this cannon will intentionally shoot  (x)*5m in the X,Y,Z direction from the target for 20% of the time, (where x could also be a random number between 0.1 and 1) would be better than: this cannon leads the target perfectly all the time. 

I will say that a lot of players forget to jink every few seconds or so, but even so the AAA fire in this game is deadly accurate. 

Edited December 22, 2022 by Lurker

[Mr_sukebe]

8 hours ago, Gunfreak said:

It's possible to set min and max altitude on AAA now.

From what I understand the 88mm generally only fired from 3000 meters and above (so 10 000k feet as you say) and it's up to mission/campaign editors to make sure that's correct. 

But the all ground fire is still way too accurate, be it flak, automatic cannons, MGs or even rifles. 

Agreed.

I’m not sure that the min altitude works 100% correctly.  If I remember correctly, when you’re within a certain range, those restrictions are removed

[Nealius]

Recently there's been advanced waypoint options for max and min AAA engagement range which can help nerf the AAA to the mission creators' pleasing. The problem is that only 8.8cm flak guns respect this setting. The 20mm AAA guns ignore it and continue to shoot through trees at fast-moving fighters 30ft off the ground.

However, I've noticed their accuracy isn't that good unless you're at a very low energy state, so there has been some improvement. 

[Gambit21]

A lot of it comes down to mission design as well. I have a Normandy map set up with some smart AI/AA group logic that cause them to wake up at different times based on random logic etc. Plus setting the AI levels etc. 

IME mission design accounts for a lot of what many consider AI issues. I'm not saying all...but many.

 

 

[Nealius]

That could be the case, since the DCS mission editor requires the user to micromanage a lot of things, however much of that should be default without forcing the user to dig so deep just to get basic functions working.

[Gambit21]

Average, Good, High, Excellent, Random.

Average is set by default, which on it's own I find manageable enough in my tests.

If I want to dig deeper I can design logic that is based on all sorts of factors and randomizations.

I'm guessing that would take a time-consuming re-work of core AI logic for these ground units to make them by default, account for so many possible factors, for instance being able to sneak up on them (not see you at certain times...able to surprise them etc) 

However the functionality to do this is already in the editor. With various triggers, AI On/Off functionality.

 

 

 

 

[cordite]

I hope some sort of nerf is built in to EDDCE. It would defeat the purpose to have to micromanage the AAA logic.

[Nealius]

3 hours ago, Gambit21 said:

Average is set by default, which on it's own I find manageable enough in my tests.

In my tests Average is still quite accurate at low altitudes, high speeds, shooting through trees, when a hand-cranked gun shouldn't physically be able to keep up with such high lateral motion. The AAA minimum engagement altitude advanced waypoint action should help mitigate that but the 20mm AAA ignores it, still firing well below the minimum engagement altitude set in the ME. (88s work properly, though).

Edited January 7, 2023 by Nealius

[AG-51_Razor]

Part of this issue would take care of itself if ED would, as they said they would so long ago, just make the clouds as hard to see through for the AI as they are for us humans.

[Nodak]

There should be a distinct difference in quality of AAA, because radar directed guns with VT fused shells are magnitudes more efficient, allowing higher rates of fire, higher hit rates with vastly increased lethality over any mk1 eyeball manually ranged and directed, mechanically fused shell guns.  Radar directed guns could engage within 1.1 mil out to max range and do it in the blind without visual contact, and the VT fuse would detonate the shell when in proximity, no best guess for fusing, visuals, or math required.  A battery could set staggered elevations in the blind and pretty much make the kill.  They got so good even the V-1's had difficulty getting past requiring the more complex and expensive V-2.

The Axis never had anything in its class and went backwards as the war progressed, supply, training, and manpower difficulties ever increasing.  Allied batteries as constituted continually shrank in size as they simultaneously increased in effectiveness.