AI Corsair, your worst best friends.

[Saxman]

10 hours ago, Laz109G10 said:

Try my mod. After removing the negative values in B4 the beast don't maneuver with a dead engine anymore. I said this in the "Corsair AI Overpowered" thread. Lots of the AI's are over done in other modules too. Same with the P-51D, the AI does crazy maneuvers. Rate of climb for a P-51D is 17.7m/s or 3,575 fpm not 22 m/s. When we fly these aircraft we don't get the same performance as the AI. I have never seen the AI overheat their engines, they fly at max speed with no consequences. Sometimes I get shot up in an aircraft and relinquish controls I go into another aircraft, the aircraft I got out of just keeps flying and some how makes it back to base. Lol.

I'm trying to figure out how to properly calculate that B4 value, because when I try using the formula (B4 = (-Cx0 - B * CL^2 + CD) / CL^4) I seem to get really weird results.

[Lidozin]

3 hours ago, Saxman said:

I'm trying to figure out how to properly calculate that B4 value, because when I try using the formula (B4 = (-Cx0 - B * CL^2 + CD) / CL^4) I seem to get really weird results.

B and B4 are simply constants that define the shape of the L/D polar curve. Cx0 is the drag at zero lift.

[Saxman]

1 hour ago, Lidozin said:

B and B4 are simply constants that define the shape of the L/D polar curve. Cx0 is the drag at zero lift.

I know, like I said, the problem I'm having is the results. I've got a sheet that handles a lot of these calculations and I keep coming up with B4 = 0 across the entire speed range, until about Mach .8 at which point it becomes exponentially more negative.

 

[Lidozin]

36 minutes ago, Saxman said:

I know, like I said, the problem I'm having is the results. I've got a sheet that handles a lot of these calculations and I keep coming up with B4 = 0 across the entire speed range, until about Mach .8 at which point it becomes exponentially more negative.

 

 

Just to be clear — negative values of B4 in the drag polar formulation are physically invalid in most cases. They imply that drag would decrease at high lift coefficients, which contradicts fundamental aerodynamic behavior. If used without extreme care (or worse, by mistake), they can artificially boost L/D in regimes where a real aircraft would be well past its energy limits or approaching stall.

For subsonic aircraft with straight wings, B4 should either be zero or small and positive — and only applied when supported by actual aerodynamic data. Anything else tends to produce unrealistic results and should be treated as a modeling error, not a tuning parameter.

What’s more, estimating CD0, B and even B4 is relatively straightforward if you have access to basic performance benchmarks — such as maximum level speed, climb rate, sustained turn time, or overall energy–maneuverability diagrams. These give you plenty of reference points to constrain the polar in a physically consistent way, even without full wind tunnel data.

Edited July 11 by Lidozin

[LeCuvier]

On 7/11/2025 at 3:59 AM, Laz109G10 said:

Try my mod. After removing the negative values in B4 the beast don't maneuver with a dead engine anymore. I said this in the "Corsair AI Overpowered" thread. Lots of the AI's are over done in other modules too. Same with the P-51D, the AI does crazy maneuvers. Rate of climb for a P-51D is 17.7m/s or 3,575 fpm not 22 m/s. When we fly these aircraft we don't get the same performance as the AI. I have never seen the AI overheat their engines, they fly at max speed with no consequences. Sometimes I get shot up in an aircraft and relinquish controls I go into another aircraft, the aircraft I got out of just keeps flying and some how makes it back to base. Lol.

I have installed your mod, and it seems to change the behaviour of the Corsair significantly. I have leftz the Corsair AI at "Veteran" for the moment, so I can see the difference. 
From a Spit, it ways easy even without  the mod and I haven't re-tried. 
From the Bf-109 it was somewhat challenging to kill the Corsair, now it's almost easy.
With both of these, the kill always started with a dead engine.
From the FW-190D, I found it impossible to defeat the Corsair because I could never get well behind it, and if I tried too hard I would get a wing flip-down at low altitude and low speed, mostly killing me. With the mod, I can kill the Corsair after a lot of turning at low altitude. Three times in a row, I got few or no hits, but it crashed into the ground. Unfortunately, the de-brief does not tell me why it hit the ground.

Edited July 12 by LeCuvier

[Laz109G10]

Try Corsair AI in Ace, more challenging. The FW190D can't turn so well with the Corsair or P-51D, only boom and zoom with FW190D. You can only do barrel rolls with the FW190D, make sure you have lots of speed. Don't turn and burn it loses too much energy quick, then your a sitting duck. Best tactics with FW is stay high and dive onto your target then climb away. 

 

Enjoy

[DSR_T-800]

7 hours ago, Laz109G10 said:

Try Corsair AI in Ace, more challenging. The FW190D can't turn so well with the Corsair or P-51D, only boom and zoom with FW190D. You can only do barrel rolls with the FW190D, make sure you have lots of speed. Don't turn and burn it loses too much energy quick, then your a sitting duck. Best tactics with FW is stay high and dive onto your target then climb away. 

 

Enjoy

Laz, how does the AI P-47 and K4 prefrom? 

[Lidozin]

On 7/11/2025 at 11:10 PM, Saxman said:

I know, like I said, the problem I'm having is the results. I've got a sheet that handles a lot of these calculations and I keep coming up with B4 = 0 across the entire speed range, until about Mach .8 at which point it becomes exponentially more negative.

 

If your goal is to tame the AI's energy performance, one practical approach is to keep the original polar coefficients while simply zeroing out the B4 term. This will already reduce excessive climb and turn rates.

For more accurate tuning, however, it's worth adjusting the values to better reflect real-world aerodynamic characteristics:

 Cx0 ≈ 0.025

 B ≈ 0.07

These values are typical for aircraft of this class and will result in a maximum L/D ratio around 12 — a realistic and well-balanced figure.

[Saxman]

5 hours ago, Lidozin said:

If your goal is to tame the AI's energy performance, one practical approach is to keep the original polar coefficients while simply zeroing out the B4 term. This will already reduce excessive climb and turn rates.

For more accurate tuning, however, it's worth adjusting the values to better reflect real-world aerodynamic characteristics:

 Cx0 ≈ 0.025

 B ≈ 0.07

These values are typical for aircraft of this class and will result in a maximum L/D ratio around 12 — a realistic and well-balanced figure.

Cx0 according to my sheet is .0245 at .1 Mach, falling to .0206 at Mach .29. Then bounces back to .0230 at Mach .3 when compressibility sets in.

B is .0657 until Critical Mach at .50, where CDwave comes into effect, reaching .0677 at Mach 1.1. So far this is yielding some pretty good results; you still don't want to try to turn with a Zero, but if you can keep your speed up and get above him you can control the fight. I'm still fine-tuning these results. My Skin Friction Coefficient throws an error if I try running calculations at altitudes about 8000ft.

[LeCuvier]

20 hours ago, Laz109G10 said:

Try Corsair AI in Ace, more challenging. The FW190D can't turn so well with the Corsair or P-51D, only boom and zoom with FW190D. You can only do barrel rolls with the FW190D, make sure you have lots of speed. Don't turn and burn it loses too much energy quick, then your a sitting duck. Best tactics with FW is stay high and dive onto your target then climb away. 

 

Enjoy

I don't see a very bifg difference with the AI Corsair in "Ace". It's a bit more fun flying the Bf.109 but the outcome is the same. The Corsair catches a few rounds and its engine dies.
Also, flying the FW-109D I see little difference. I turn into the merge and climb with him, and continue turning behind him not too aggressively, just short of flipping a wing down. As we get slower and lower, I set the flaps to landing and keep turning behind him, slowly catching up. Then it looks as if his engine were dying as he slows to a crawl and I try not to overshoot him too much. And as I look back, I see that he has hit the deck. Boom! Pretty much the same as the Veteran.
I wonder if the AI has an engine problem.
Edit: I also tried the P-51D against the Corsair in Ace. I'm not very good with the Mustang and find its tendency to flip wings worse than the Dora, so I was struggling. The Corsair seems to adapt its tactics to the weakness of its adversairy. It flies low and slow turning all the time, and I struggled to keep my Mustang stable in the air, and I didn't always manage that although I used my flaps. So sometimes I win and sometimes I lose. That makes the Corsair an interesting sparring partner, and I will sparr more!

[Laz109G10]

I always try to find real data or pilot reports on actual aircraft performance. I heard of P-51 pilots saying they can out turn a Bf109 and vise versa. Flying a Bf109K I can easily out turn a P51D Ace AI in DCS. How accurate is this in real life? Don't know, also the Bf109K lua is encrypted so can't examine the parameters used by ED. Same goes for P-47D, files are encrypted. My tactics are if I know I am fighting an aircraft that turns better than me, I boom and zoom it to death. A Corsair can out turn a P51D in test by the Navy. F4U has better slow speed turning because it has more wing area and lower wing loading compare to P-51D.

[Saxman]

9 hours ago, LeCuvier said:

That makes the Corsair an interesting sparring partner, and I will sparr more!

What makes the Corsair so dangerous is its flexibility. What it can't out-turn it can usually out run and slash to death in the BnZ. And what it can't outrun it can handily outmaneuver. There's very few opponents the Corsair doesn't have SOME tool it can use to beat it.

[LeCuvier]

I have done some more sparring with the Corsair AI (Ace) flying the Bf-109K4, and this confirms my earlier impressions:
The damage model of the AI seems over-vulnerable. It takes just a few hits with 13mm rounds and the engine dies. I'm fairly sure the Corsair was more robust than that.

[Laz109G10]

The critical damage on the F4U engine is set at 5. On the A-20G it is set at 6 and P51D set at 3. Mosquito set at 5. I will test it at different critical damage numbers to see. Wish I can see the Fw190-A8 which would be close to F4U radial engine, like the A-20G.

 

Enjoy

[DSR_T-800]

17 hours ago, Laz109G10 said:

I always try to find real data or pilot reports on actual aircraft performance. I heard of P-51 pilots saying they can out turn a Bf109 and vise versa. Flying a Bf109K I can easily out turn a P51D Ace AI in DCS. How accurate is this in real life? Don't know, also the Bf109K lua is encrypted so can't examine the parameters used by ED. Same goes for P-47D, files are encrypted. My tactics are if I know I am fighting an aircraft that turns better than me, I boom and zoom it to death. A Corsair can out turn a P51D in test by the Navy. F4U has better slow speed turning because it has more wing area and lower wing loading compare to P-51D.

Alright thanks. Forum users have done the math, and seem to have agreed that the K4 will turn 1s faster than the P-51D@67"Hg and 0.5s faster @75"Hg. Whenever, I'm fighting the AI K4 in my P-47, I find the K4 prefers to out-rate me rather than attempt the tightest turn possible.

Whats strange is in that very same test, the corsair can out climb and out accelerate the P-51B, but its not even on a contest with the DCS P-51 where it readily out paces the F4U. Also, wouldn't the Fw-190D9 be significantly ahead of the P-47 and P-51? The Bf-109K4 is in another dimenson. Honestly, I'm kinda just scratching my head in this one.

 

[GTFreeFlyer]

I have a new one to add:

I taxied out with two AI wingmen in same group.  Got to runway and all was good.  I throttled up slowly as I always do with Warbirds, and exploded before rotation.  #2 rammed into me.  I think perhaps they just go full throttle?  Doesn’t seem like they kept separation on the takeoff roll.  Either that, or AI visibility over the nose is modeled extremely well, although at the speed we collided, my tail was already up.  
Cheers

[Elf1606688794]

On 7/5/2025 at 9:22 AM, LeCuvier said:

The AI version of the Corsair has an incredible flight model. It seems to make turns without losing energy and it climbs like a rocket.When I fly a FW-190 Dora or Anton it kills me every time. I have been able to defeat it when I fly the Bf-109K, but it's tough.
In the reverse situation, when I fly the corsair, I can't even defeat a FW-190 Anton AI.
Something feels wrong. 

I find it very difficult to beat any of the AI aircraft due to their unrealistic flight modeling. And I have yet to beat an AI MiG-21 in the F-4E.

In regards to the bold, in my experience all AI opponents do that to one degree or another. You can watch this happen in a TacView replay. I have a Tacview somewhere that shows a MiG-21 sustaining 8g turns without losing speed in turns or when climbing. (The speed of the MiG flucuates a small amount but anything it loses it regains quickly.)

[LeCuvier]

6 hours ago, Elf1606688794 said:

I find it very difficult to beat any of the AI aircraft due to their unrealistic flight modeling. And I have yet to beat an AI MiG-21 in the F-4E.

In regards to the bold, in my experience all AI opponents do that to one degree or another. You can watch this happen in a TacView replay. I have a Tacview somewhere that shows a MiG-21 sustaining 8g turns without losing speed in turns or when climbing. (The speed of the MiG flucuates a small amount but anything it loses it regains quickly.)

The MiG-21 wasn't always so extraterrestrial. I used to kill the AI flying the Bf-109K4 (no kidding!), and that wasn't even very difficult. That of course was unrealistic too.

Edited July 16 by LeCuvier

[Art-J]

On 7/14/2025 at 7:11 PM, GTFreeFlyer said:

I have a new one to add:

I taxied out with two AI wingmen in same group.  Got to runway and all was good.  I throttled up slowly as I always do with Warbirds, and exploded before rotation.  #2 rammed into me.  I think perhaps they just go full throttle?  Doesn’t seem like they kept separation on the takeoff roll.  Either that, or AI visibility over the nose is modeled extremely well, although at the speed we collided, my tail was already up.  
Cheers

Don't expect that much modelling fidelity . There's a relatively recent YT video by Reflected covering this issue. It's not related to Corsair but it's just a general DCS AI problem. In short, once you're past certain point on the runway, AI considers you "gone" and takes off and rams whatever is on the runway no matter what. The trick is to know where that point is. I'm not at my PC at the moment and can't provide the link, but go to Reflected's channel and you should find that vid easily.

[Elf1606688794]

18 hours ago, LeCuvier said:

The MiG-21 wasn't always so extraterrestrial. I used to kill the AI flying the Bf-109K4 (no kidding!), and that wasn't even very difficult. That of course was unrealistic too.

Maybe ED can find the midway mark between what you used to have and what we all have now. I don't want an AI opponent so lame that fighting them is like fighting a chair, you win every time because, well... it's a frigging CHAIR. AI opponents shouldn't be able to sustain 8g turns without losing speed. They also shouldn't maintain energy across the entire flight envelope.

 

[Gunfreak]

On 7/16/2025 at 9:27 AM, LeCuvier said:

The MiG-21 wasn't always so extraterrestrial. I used to kill the AI flying the Bf-109K4 (no kidding!), and that wasn't even very difficult. That of course was unrealistic too.

The AI has several aspects. All whìch are off.

1. The preformance, AI over G-ing with no effect, energy retention in turn and upwards movements etc etc.

2. Omniscience, AI can see through clouds. Can spot you flying low over the terrain, know when you are behind them etc. ED is working on this. Particularly the cloud thing.

3. AI are stupid, they might be overpowered but they are stupid. That's why you beat it in a 109. A 109 would indeed beat the MIG21 if the 21 pilot decided to get into a turn fight. Even a bad ww2 turn fighter like the Anton out turns any jet(except possibly those with thrust vectoring) 

If you do a ww2 vs jet dog dogfight with a human, and force the human to get into a turn fight. Then indeed you can beat any dcs jet with a ww2 prop plane. In real life the MiG21 pilot would just shoot into the sky or dive away, come back and blast you at 1000kmh.  But AI are stupid. So they don't necessarily do that.

[Laz109G10]

From known stats a Mig-21 should out turn a F-4E Phantom. I have try to kill a F-4E with a Mig-21 with guns only. Every time I am about to get him on my sights the Ai seems to know when I am about to pull the trigger. He does an instantaneous turn at the right moment, and I miss. There is no way a human being can tell in a dogfight when to do that. The Ai does it every time repeatedly. Hopefully ED can change the behavior of the Ai to be more human. When I am at low six there is no way he can see me, especially in warbirds with no radar.

[Laz109G10]

The update has made the Corsair Ai a little more realistic than before. They have fix some of the fm issues. I am glad they heard us. Will test some more against different aircraft.

 

Enjoy