PDA

View Full Version : AI roll rate



robban75
01-11-2006, 11:07 AM
Hello Mr. Maddox!

This is nothing new, as it has been around for as long as I can remember. But it's probably more appearant now because the AI is fighting much more at high speed in 4.02(which is great!).

Fighting against the AI is fun, except for one thing. The AI doesn't suffer from heavy ailerons. I'm guessing I'm not the only one who's been outrolled by a Spitfire at 800km/h IAS in a Mustang or Fw 190. 180deg/sec roll rates seems to be avaliable to the AI throughout the flight envelope.
This means that the AI cannot be outmanouvered at high speed, nor can one hope to manouver with it.

Can this be fixed? It would be great! http://forums.ubi.com/images/smilies/16x16_smiley-happy.gif

Thanks!

robban75
01-11-2006, 11:07 AM
Hello Mr. Maddox!

This is nothing new, as it has been around for as long as I can remember. But it's probably more appearant now because the AI is fighting much more at high speed in 4.02(which is great!).

Fighting against the AI is fun, except for one thing. The AI doesn't suffer from heavy ailerons. I'm guessing I'm not the only one who's been outrolled by a Spitfire at 800km/h IAS in a Mustang or Fw 190. 180deg/sec roll rates seems to be avaliable to the AI throughout the flight envelope.
This means that the AI cannot be outmanouvered at high speed, nor can one hope to manouver with it.

Can this be fixed? It would be great! http://forums.ubi.com/images/smilies/16x16_smiley-happy.gif

Thanks!

Kwiatos
01-11-2006, 01:58 PM
Yea i found the same AI have too good roll rate at high speed no metter of plane.

RegRag1977
01-12-2006, 10:06 AM
Yes, Robban is definitely right http://forums.ubi.com/images/smilies/25.gif!

That would be cool if AI rolling rate was realistic in this sim! We would like our Fw more! http://forums.ubi.com/groupee_common/emoticons/icon_cool.gif

I want that to change too! Hope Oleg will read his post! http://forums.ubi.com/images/smilies/25.gif

Kwiatos
01-12-2006, 11:34 AM
I think Oleg dont care too much FB yet - dont see any his post for long time.

OldMan____
01-13-2006, 08:17 AM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Kwiatos:
I think Oleg dont care too much FB yet - dont see any his post for long time. </div></BLOCKQUOTE>

Not having time does not mean he does not care for it.

AKA_TAGERT
01-19-2006, 08:32 AM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by robban75:
Hello Mr. Maddox!

This is nothing new, as it has been around for as long as I can remember. But it's probably more appearant now because the AI is fighting much more at high speed in 4.02(which is great!).

Fighting against the AI is fun, except for one thing. The AI doesn't suffer from heavy ailerons. I'm guessing I'm not the only one who's been outrolled by a Spitfire at 800km/h IAS in a Mustang or Fw 190. 180deg/sec roll rates seems to be avaliable to the AI throughout the flight envelope.
This means that the AI cannot be outmanouvered at high speed, nor can one hope to manouver with it.

Can this be fixed? It would be great! http://forums.ubi.com/images/smilies/16x16_smiley-happy.gif

Thanks! </div></BLOCKQUOTE>
It's not that the AI can outmanouver real people. Real people can out manuver the Fw190 at high speeds too.

The problem is with the roll-rates. The in-game Fw190 has one of the highest, if not the highest roll rate in the game, but, the problem is it happens at a much lower speed than it did in real life.

Note here is a graph of the real life roll rates.
http://www.geocities.com/grantsenn/NACA_RESULTS/ROLL_RATES/402/V402_ALL_SUMMARY_NACA868.JPG

And here is a graph of the in-game roll rates for the same (almost) planes.
http://www.geocities.com/grantsenn/NACA_RESULTS/ROLL_RATES/402/V402_ALL_SUMMARY_INGAME.JPG

As you can see, the roll rate for the Fw190 is the fastest, but note it happens at a much lower speed (~195mph vs. ~260mph). But that is not the worst part, note that in the real life roll rates the Fw190 maintains it's advantage from ~260mph out to ~360mph, i.e. the HIGH SPEED ZONE your talking about. But, that is NOT the case for the in-game Fw190! Note that the P40E has an advantage over the Fw190 from ~250 onward, and the P51B has an advantage over the Fw190 from ~270 onwards. That is the area where the real life Fw190 had it's advantage, and in-game it does not have an advantage.

So, what your seeing is not an AI error, it is a roll rate error. The only way to take advantage of the Fw190s high roll rates is to fly slow.. i.e. ~190mph, if you fly fast in-game, your advantage is gone.

RocketDog
01-19-2006, 01:21 PM
Nicely explained TAGERT.

Do you know if there is any indication that this is in line for a fix in a future patch, or are we stuck with it?

In fact, do you know if this error cropped up in the new FM, or has it always been in the game?

Cheers,

RocketDog.

AKA_TAGERT
01-19-2006, 04:09 PM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by RocketDog:
Nicely explained TAGERT. </div></BLOCKQUOTE>
Thanks! If you€re interested in the full write up you can find it here

AIRWARFARE.COM (http://airwarfare.com/phpBB2/viewtopic.php?t=771)

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by RocketDog:
Do you know if there is any indication that this is in line for a fix in a future patch, or are we stuck with it? </div></BLOCKQUOTE>They have not even acknowledged that it is a bug, thus I doubt there will be a fix. But, than again, if they did, it wouldn€t be the first time they tweaked/fixed something without any prior indication. So we can keep our fingers crossed.

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by RocketDog:
In fact, do you know if this error cropped up in the new FM, or has it always been in the game? </div></BLOCKQUOTE>That I don€t know, in that I never tested the RR in older versions of the game.

Vrabac
01-19-2006, 04:27 PM
If you take a look at the mere shape of the graphs, you'll probably come to conclusion that the bug is quite deeply embedded in the flight model. Basically all planes have wrong roll rates, they peak at wrong speeds etc etc. So i doubt it will be fixed. http://forums.ubi.com/images/smilies/16x16_smiley-sad.gif FW really suffers from this, but so does the P47, it's roll at high speed is pathetic, and for plane that is supposed to count on B&Z, this is really a bad handicap. But... I'm afraid we'll just have to live with it.

The probelm is however that AI has rediculously high roll rate at high speeds regardles the plane. Look at their evasions, when they start rolling. They go towards the ground and at 700kmh start rolling with for example 109 5 times better than human 190 can. It's quite irritating. Also I got the feeling that they retain energy in these barrel rolls much better than they should. i did no tests, but it seems wrong, one would expect a plane to rapidly lose speed that way, and fighting human opponents made me think so, but they don't. I guess this might be fixed, as it's AI and not strictly FM related.

VW-IceFire
01-19-2006, 05:36 PM
Yeah the AI definately seems to be able to outroll and do several things at higher speeds that no human can even try and do. Simplified flight models for the AI...its that simple.

Kind of wish there was a bit more added to the simplified model so they wouldn't do anything blatant.

AKA_TAGERT
01-19-2006, 05:57 PM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Vrabac:
If you take a look at the mere shape of the graphs, you'll probably come to conclusion that the bug is quite deeply embedded in the flight model. Basically all planes have wrong roll rates, they peak at wrong speeds etc etc. So i doubt it will be fixed. http://forums.ubi.com/images/smilies/16x16_smiley-sad.gif </div></BLOCKQUOTE>
I do to, but for different reasons. It is not a limitation of the FM, in that it can clearly simulate the classically ramp.up-peak-ramp.down curve. That is to say the *shape* is fine, the problem is magnitude and location of the peak. Where most in-game planes have the same RR problem.

1) Thier peak is larger than it should be
2) Thier peak happens at a lower speed than it should

Put another way, TOO FAST of a RR at TOO SLOW of a speed.

The only part of the FM that is wrong, is that in real life, once you passed the peak, it was down hill from there onward. But look at the in-game RR.. Some of them go down, hit bottom, then start to come back up again. Take the in-game zero for example (black dotted line near bottom) it has the same RR of 40?/sec at ~239mph and ~280mph with the min RR in-between the two at ~255mph. That it totally wrong! No plane should ever increase RR on the back side of the curve, i.e. the semi under curve decrease side after the peak. A few in-game planes have this error.

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Vrabac:
FW really suffers from this, but so does the P47, it's roll at high speed is pathetic, and for plane that is supposed to count on B&Z, this is really a bad handicap. </div></BLOCKQUOTE>
Note, I didn€t test all 150+ planes, just the ones I had real life data for, and out of those, I agree the Fw190 is getting robbed!

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Vrabac:
But... I'm afraid we'll just have to live with it. </div></BLOCKQUOTE>
In light of the fact they have not said one word about it, can only mean two things.. they are going to ignore it or fix it and surprise us. The good news is Oleg might.. in that this is the new FM for BoB. Thus, he just might, in that since the introduction of the new FM (4.0+) the Il2-PF sim has become a sort of test bed for BoB.

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Vrabac:
The probelm is however that AI has rediculously high roll rate at high speeds regardles the plane. </div></BLOCKQUOTE>
Disagree 100%! In that if Oleg has said it once, he has said it 1000 times. The AI fly the same FM we do. So, *if* the AI can roll their plane faster than you can, it can only mean one thing. The stick force limits for the AI is greater than ours. It has been said the the user (real player) stick force is limited to 50lbs, Which basically determines where that *peak* happens. So, if the AI can roll the plane faster, and they do have the same FM, then the only way is if they can apply more force, say 60lb, or 70lb, or more. But, I have yet to see any real proof that the AI can roll the plane faster than I can. But problem would only happen post peak, all RR prior to the peak does not mater if you have a stick force limit of 10,000lbs it would not increase the RR, only post peak would it mater.

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Vrabac:
Look at their evasions, when they start rolling. They go towards the ground and at 700kmh start rolling with for example 109 5 times better than human 190 can. It's quite irritating. </div></BLOCKQUOTE>
Got Track? In that I have never seen anything close to 1 times better let alone 5 times better.

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Vrabac:
Also I got the feeling that they retain energy in these barrel rolls much better than they should. i did no tests, but it seems wrong, one would expect a plane to rapidly lose speed that way, and fighting human opponents made me think so, but they don't. I guess this might be fixed, as it's AI and not strictly FM related. </div></BLOCKQUOTE>
That is the problem with *feelings*, so many of them, like opinions! http://forums.ubi.com/groupee_common/emoticons/icon_wink.gif

AKA_TAGERT
01-19-2006, 06:04 PM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by VW-IceFire:
Yeah the AI definately seems to be able to outroll and do several things at higher speeds that no human can even try and do. </div></BLOCKQUOTE>I had someone tell me once they *felt* the Ki61 could not get close to it's rated top speed. I asked them to send me a track file of it, turned out thier *preception* of the event was wrong, thus thier *feelings* were wrong too. Any cop will tell you, humans make the worst recording and measuring devices. Interview 10 people who saw a guy rob a bank, and you get 10 different discriptions.

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by VW-IceFire:
Simplified flight models for the AI...its that simple. Kind of wish there was a bit more added to the simplified model so they wouldn't do anything blatant. </div></BLOCKQUOTE>
Disagree 100% Because Oleg has stated the AI flys the same FM we (real people) do. That is to say there is NOT 150 FM's for real people and another 150 different FM's for AI to fly. We all fly the same FM, even when we are flying the AI only planes with the simple FMs, there too, we both fly the same FM.

VW-IceFire
01-19-2006, 09:01 PM
Trust me...I don't trust my own recollections all that well. I'm a healthy skeptic.

So what I've done is do 1 vs 1 with the same aircraft. My job is to follow the Veteran or Ace AI as a trail chaser replicating their manuvers as best I can. Not a perfect test either...but how an AI Zero pilot can outroll me with ease suggests that they play by different rules.

I know what Oleg said...and its contradictory to what I see. Thats all I can speak to...just my version of how the bank was robbed I suppose http://forums.ubi.com/groupee_common/emoticons/icon_smile.gif

AKA_TAGERT
01-19-2006, 10:07 PM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by VW-IceFire:
Trust me...I don't trust my own recollections all that well. I'm a healthy skeptic.

So what I've done is do 1 vs 1 with the same aircraft. My job is to follow the Veteran or Ace AI as a trail chaser replicating their manuvers as best I can. Not a perfect test either.. </div></BLOCKQUOTE>Roger, not perfect, but about as good as we can do in light of the fact we can not get any DeviceLink info from other planes.

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by VW-IceFire:
but how an AI Zero pilot can outroll me with ease suggests that they play by different rules. </div></BLOCKQUOTE>Now your talking.. In that as I pointed out, Oleg has said the FM is the same for both.. So all that leaves is the stick force.. *IF* note, not agreeing here, but *IF* the AI is out rolling you with ease.. that only leaves the stick force *AND* it can only happen at a point above the RR peak.

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by VW-IceFire:
I know what Oleg said...and its contradictory to what I see. </div></BLOCKQUOTE>Well, like I also said, humans make for the worst recording and measurement devices.. TWO people can watch the same event and have TWO totally different takes on it. Not saying your wrong.. not saying your right.. All Im saying is nobody has provided anything up to now to even being to prove it is so.

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by VW-IceFire:
Thats all I can speak to...just my version of how the bank was robbed I suppose http://forums.ubi.com/groupee_common/emoticons/icon_smile.gif </div></BLOCKQUOTE>Roger! http://forums.ubi.com/groupee_common/emoticons/icon_wink.gif

I do have any idea though.. never tried it.. but.. can you set up a mission.. 1 vs 1 and then enable the autopilot and allow him to fight the fight? That is to say AI vs AI where the AI is flying your plane? If so, then record the whole thing in a track file. Then play it back and note the TIME you see what LOOKS LIKE your autopilot (ie AI) rolls your plane faster than you think it can go. Then, I could process the track file at that moment and see what the roll rate is.. and if it is higher than it should be.

The closer to 10kft this event happens the beter.. because then we can comp it to the NACA RR data. Oh, and pick a plane from the NACA list too!

Now.. *IF* the AI can out roll you with EASE.. THAN it should be a very Very VERY much larger RR than what the plane can do and thus show up in the analysis as a RR that is at least 50% higher than the max roll rate.

If and when somebody does that.. Than I will be sold on the idea.. But until then, not so much! http://forums.ubi.com/groupee_common/emoticons/icon_wink.gif

Blackjack174
01-20-2006, 09:06 AM
Its much simpler than that, try fight an veteran / ace hurricaneat altitudes above 3000 meters in a bf109.
The AI will actually dive very often (exceeding maximum dive speed for us humans of course) and then doing evasive maneuvers when followed with so fast turns it looks like a aerobaticplane.
So maybe the fact that stickforces are somehow less for AI (or not modelled) AND that AI can go beyond max dive speed make them pronounce the faster rolling even more.
Maybe some of you can turn "invincible" on in difficult setting and test if a hurricane at 800kph can roll so fast with player input (cant test as my gfx card is toast)

AKA_TAGERT
01-20-2006, 09:32 AM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Blackjack174:
Its much simpler than that, try fight an veteran / ace hurricaneat altitudes above 3000 meters in a bf109.
The AI will actually dive very often (exceeding maximum dive speed for us humans of course) and then doing evasive maneuvers when followed with so fast turns it looks like a aerobaticplane. </div></BLOCKQUOTE>That is a neat feeling to have.. but where is the proof? As I said, humans make the worst measuring and recording devices. Ask 10 people who saw a guy rob a bank, and you will get 10 different descriptions. Same thing goes for what people *see* happening in this or any other flight sim.

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Blackjack174:
So maybe the fact that stickforces are somehow less for AI (or not modelled) AND that AI can go beyond max dive speed make them pronounce the faster rolling even more. </div></BLOCKQUOTE>Oleg may be alot of things.. like human.. thus prone to making a mistake from time to time, but I don€t think I would call him a bold face lire.

Oleg has said on many occasions that the AI and real people both fly the same FM. Therefore there is no chance of the stick forces being less for the AI, or that no stick force is modeled at all. The only thing I can come up with that might explain what people *think* they are seeing is that the AI has the ability to apply MORE FORCE than real people. Real people have this stick force limited to 50lbs, maybe the AI can go as high ast 60lbs?

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Blackjack174:
Maybe some of you can turn "invincible" on in difficult setting and test if a hurricane at 800kph can roll so fast with player input (cant test as my gfx card is toast) </div></BLOCKQUOTE>Roger, if you want to convince me, I will need some proof. To prove it to me, set up a 1 on 1. Pick a plane from the NACA RR test chart I provided. Set your DM to invincible. This will allow the AI pilot in the other plane to get on your six and shoot at you and not cause any damage to your plane. Now, during the 1 on 1 allow the AI plane to get on your six, then switch to autopilot, now it is AI vs. AI. Start recording a track file and pay att to your plane, once you *see* it perform a RR that you *feel* is TOO HIGH stop the recording and send me the track file at naca_tesing@yahoo.com where I will then graph the RR from your aircraft. From what you guys are saying, the RR should be alot bigger than the max I obtained during my testing.. ALOT BIGGER! Because for the human eye to detect it, it would have to be allot bigger.. not 5?/sec more, not 10?/sec, not 20?/sec but at least 50% higher than the max I obtained.

Until than, I will have to be like Oleg and disregard this until someone takes the time to show me, what with all the whining that goes on around here I don€t blame him one bit.

Vrabac
01-20-2006, 09:34 AM
Tagert, if you want I can send you the track, but it would be much simpler if you just make a quickmission against AI, get on his six and observe what he does. http://forums.ubi.com/groupee_common/emoticons/icon_smile.gif The roll it makes is unreal. And this isn't my feeling, it's my observation. The speed retaining during a barrel roll is feeling, and I admit it.

AKA_TAGERT
01-20-2006, 09:38 AM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Vrabac:
Tagert, if you want I can send you the track, but it would be much simpler if you just make a quickmission against AI, get on his six and observe what he does. http://forums.ubi.com/groupee_common/emoticons/icon_smile.gif The roll it makes is unreal. </div></BLOCKQUOTE>Nah, dont bother in that it wont prove anything for me to see it either, just like it does not prove anything for you to see it. Because like I have said, humans make the worst measuring and recording devices.

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Vrabac:
And this isn't my feeling, it's my observation. </div></BLOCKQUOTE>Observations that are based on your feelings.

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Vrabac:
The speed retaining during a barrel roll is feeling, and I admit it. </div></BLOCKQUOTE>Maybe, maybe not, note, not saying you are wrong, only that nobody has providing anything to prove it.

Blackjack174
01-20-2006, 08:35 PM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by AKA_TAGERT:
That is a neat feeling to have.. but where is the proof? As I said, humans make the worst measuring and recording devices. Ask 10 people who saw a guy rob a bank, and you will get 10 different descriptions. Same thing goes for what people *see* happening in this or any other flight sim. </div></BLOCKQUOTE>

Ok , scratch that about taking a bf109 , take a hurricane yourself , stay on the 6 of an AI Hurricane , you simply cant roll that fast http://forums.ubi.com/groupee_common/emoticons/icon_biggrin.gif

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">
Oleg may be alot of things.. like human.. thus prone to making a mistake from time to time, but I don€t think I would call him a bold face lire.

Oleg has said on many occasions that the AI and real people both fly the same FM. Therefore there is no chance of the stick forces being less for the AI, or that no stick force is modeled at all. The only thing I can come up with that might explain what people *think* they are seeing is that the AI has the ability to apply MORE FORCE than real people. Real people have this stick force limited to 50lbs, maybe the AI can go as high ast 60lbs? </div></BLOCKQUOTE>

IMHO saying the stickforces for AI are the same but they have more muscle is the same like having less stick force for AI , dont you agree ? At least effect ingame is the same, the can do maneuver you cant cope with.

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Roger, if you want to convince me, I will need some proof. To prove it to me, set up a 1 on 1. Pick a plane from the NACA RR test chart I provided. Set your DM to invincible. This will allow the AI pilot in the other plane to get on your six and shoot at you and not cause any damage to your plane. Now, during the 1 on 1 allow the AI plane to get on your six, then switch to autopilot, now it is AI vs. AI. Start recording a track file and pay att to your plane, once you *see* it perform a RR that you *feel* is TOO HIGH stop the recording and send me the track file at naca_tesing@yahoo.com where I will then graph the RR from your aircraft. From what you guys are saying, the RR should be alot bigger than the max I obtained during my testing.. ALOT BIGGER! Because for the human eye to detect it, it would have to be allot bigger.. not 5?/sec more, not 10?/sec, not 20?/sec but at least 50% higher than the max I obtained.

Until than, I will have to be like Oleg and disregard this until someone takes the time to show me, what with all the whining that goes on around here I don€t blame him one bit. </div></BLOCKQUOTE>

Ok , so I have to do my tersting with an 4mb gfx card , and track will be postet here.
But maybe you do the easiest test with autopilot yourself:
Start QMB (make sure all settings except externals and speedbar are turned on in difficulty menu), select a lagg3 (or similar crappy dive plane http://forums.ubi.com/groupee_common/emoticons/icon_biggrin.gif ), then dive straight down from 5000+meters , note that the airframe desintegrates around ~700kp/h , now repeat the test and switch autopilot on at around 690 kph, the autopilot will immediately try to recover from the dive but while doing so it will exceed 750km/h , a speed no one can reach because of the breaking airframe.
So I dont doubt Oleg when saying that the AI has the same FLIGHT model, but isnt maximum dive speed also a part of it ?
As I said earlier I cant see **** with my current gfx card, dogfighting is out of the question, someone else please supply a nice track with that spiffy rolling hurricanes at 800kph http://forums.ubi.com/groupee_common/emoticons/icon_biggrin.gif

so here is the divetest that clearly shows how the AI exceeds speeds I cant (in the end of the track I do the dive http://forums.ubi.com/images/smilies/34.gif )
Lagg3@750km/h (http://s46.yousendit.com/d.aspx?id=0PX7Q66R2AO1T1TKWV5OPA3VKI)

arrow80
01-21-2006, 02:00 AM
AI still use SIMPLIFIED models at least at high speeds. I got my campaign scrapped because of that, after 54 missions. I had autopilot on and it was diving down the plane, I didn't realize that the speed was 780 km/h and switched off the autopilot, the plane (LaGG 3 S4) immediately disintegrated...

AKA_TAGERT
01-21-2006, 10:03 AM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Arrow80:
AI still use SIMPLIFIED models at least at high speeds. </div></BLOCKQUOTE>Oleg may be alot of things.. like human.. thus prone to making a mistake from time to time, but I don€t think I would call him a bold face lire? In that Oleg has said on many occasions that the AI and real people both fly the same FM!

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Arrow80:
I got my campaign scrapped because of that, after 54 missions. I had autopilot on and it was diving down the plane, I didn't realize that the speed was 780 km/h and switched off the autopilot, the plane (LaGG 3 S4) immediately disintegrated... </div></BLOCKQUOTE>Which proves nothing really, in that the act of switching off the AI is most liky what caused the disintegrtion, in that the controls were *suddenly* switched from the AI to your controls.. The effect would be like jerking the stick very hard.. Which we know will take the wings off of a P51

AKA_TAGERT
01-21-2006, 10:13 AM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Blackjack174:
Ok , scratch that about taking a bf109 , take a hurricane yourself , stay on the 6 of an AI Hurricane , you simply cant roll that fast http://forums.ubi.com/groupee_common/emoticons/icon_biggrin.gif </div></BLOCKQUOTE>That is a neat feeling to have.. but where is the proof? As I said, humans make the worst measuring and recording devices. Ask 10 people who saw a guy rob a bank, and you will get 10 different descriptions. Same thing goes for what people *see* happening in this or any other flight sim.

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Blackjack174:
IMHO saying the stickforces for AI are the same but they have more muscle is the same like having less stick force for AI , dont you agree ? </div></BLOCKQUOTE>No, because if I push on a wall with 50lb or 60lb the wall does not move. Same thing for alirons, elevators, rudders.. Once they are at thier max deflection they are not going to move any farther than that. The only time the 50lb vs 60lb comes into play, wrt RR, is the ponit past the *peak*. Basically having more force allows you to keep the alirons fully deflected, thus you can move the *peak* out a little farther. But, prior to the *peak* it wouldnt mater if the AI was 1000 times stronger than were are, the aleirons are not going to go past full deflection.

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Blackjack174:
At least effect ingame is the same, the can do maneuver you cant cope with. </div></BLOCKQUOTE>That is a neat feeling to have.. but where is the proof? As I said, humans make the worst measuring and recording devices. Ask 10 people who saw a guy rob a bank, and you will get 10 different descriptions. Same thing goes for what people *see* happening in this or any other flight sim.

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Blackjack174:
Ok , so I have to do my tersting with an 4mb gfx card , and track will be postet here. </div></BLOCKQUOTE>Or send it to naca_testing@yahoo.com

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Blackjack174:
But maybe you do the easiest test with autopilot yourself:
Start QMB (make sure all settings except externals and speedbar are turned on in difficulty menu), select a lagg3 (or similar crappy dive plane http://forums.ubi.com/groupee_common/emoticons/icon_biggrin.gif ), then dive straight down from 5000+meters , note that the airframe desintegrates around ~700kp/h , now repeat the test and switch autopilot on at around 690 kph, the autopilot will immediately try to recover from the dive but while doing so it will exceed 750km/h , a speed no one can reach because of the breaking airframe. So I dont doubt Oleg when saying that the AI has the same FLIGHT model, but isnt maximum dive speed also a part of it ? </div></BLOCKQUOTE>Not sure if what you saw was a FM or DM thing, but in either case it can be explaned away very easly, as I did above

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Blackjack174:
As I said earlier I cant see **** with my current gfx card, dogfighting is out of the question, someone else please supply a nice track with that spiffy rolling hurricanes at 800kph http://forums.ubi.com/groupee_common/emoticons/icon_biggrin.gif </div></BLOCKQUOTE>Yes, anyone! Stil waiting.. nothing yet.. why? Because it is most likly not true. If it was as bad as everyone is saying, than it should be a simple mater to make a track showing it.. Yet no one has.. Makes you wonder, dont it?

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Blackjack174:
so here is the divetest that clearly shows how the AI exceeds speeds I cant (in the end of the track I do the dive http://forums.ubi.com/images/smilies/34.gif ) </div></BLOCKQUOTE>Dont get mad at the AI because he is lighter on the controls than you are while the plane is under stress! http://forums.ubi.com/groupee_common/emoticons/icon_wink.gif Also, I need *.ntrk files not *.trk files, use the Quick Record option not the save trak at the end option

Blackjack174
01-21-2006, 09:40 PM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by AKA_TAGERT:
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Arrow80:
I got my campaign scrapped because of that, after 54 missions. I had autopilot on and it was diving down the plane, I didn't realize that the speed was 780 km/h and switched off the autopilot, the plane (LaGG 3 S4) immediately disintegrated... </div></BLOCKQUOTE>Which proves nothing really, in that the act of switching off the AI is most liky what caused the disintegrtion, in that the controls were *suddenly* switched from the AI to your controls.. The effect would be like jerking the stick very hard.. Which we know will take the wings off of a P51 </div></BLOCKQUOTE>
Ok, as you fail to see the point clearly proven by now:

You can NEVER EVER reach 780km/h in a LaGG3, even if you hold the controls as steady as you like (thus not exceeding maximum g´s for airframe)
AI can because there is NO limit for it.
As long as you deny that,there is no point arguing any further, as my instructions for QMB shows you that fact in 1 minute.

over and out as long as I have to become repetive, too http://forums.ubi.com/images/smilies/88.gif

AKA_TAGERT
01-22-2006, 12:22 AM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Blackjack174:
You can NEVER EVER reach 780km/h in a LaGG3, even if you hold the controls as steady as you like (thus not exceeding maximum g´s for airframe)
AI can because there is NO limit for it.
As long as you deny that,there is no point arguing any further, as my instructions for QMB shows you that fact in 1 minute.

over and out as long as I have to become repetive, too http://forums.ubi.com/images/smilies/88.gif </div></BLOCKQUOTE>Not denying it, just saying that I need a *.ntrk file to check it out, the *.trk files are not good enough.. You can make a *.ntrk out of your *.trk by simply playing back the *.trk and pressing the quick record button to make a *.ntrk file out of it.

As for top speed vs. roll rate.. All my statments were wrt the RR and FM. What you describing may be part of the FM or the DM, not sure where the line between the two is drawn. Oleg has never commented on the DM aspects, but has said both real and AI fly the same FMs. So, your guess is as good as mine.

As for the RR and this top speed thing your describing, Again, not saying your wrong, just saying nobody has provided any proof of it yet. Even though I have described a very simple method to do so.. Yet nobody has done it yet.. Yet they all claim it happens all the time and is noticable to the nakid eye.. yet nothing. Kind of makes one wonder, dont it?

lowfighter
01-22-2006, 03:32 AM
I think some cleverly thought recorded tests would be extremely enlightening! As Tagert said: the opinions about AI flight model are sometimes strikingly divergent. One note: my feeling http://forums.ubi.com/groupee_common/emoticons/icon_smile.gif is that AI behaviour changed drastically from FB v3. to FBv4.01 and than once again from v4.01 to v4.02. Remains to be proved though if AI cheats with the flight model.

Brain32
01-22-2006, 04:02 AM
There is a simple test, in QMB put ACE for yourself and ACE for your opponent, when game loads, switch it to autopilot and watch the fight from the cockpit. I took 109 vs Spitfire, 109 rolled like mad, what human can't achieve at 450KM/h(when 109 becomes like a brick http://forums.ubi.com/groupee_common/emoticons/icon_rolleyes.gif) AI does at much higher speeds...

lowfighter
01-22-2006, 05:36 AM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Brain32:
There is a simple test, in QMB put ACE for yourself and ACE for your opponent, when game loads, switch it to autopilot and watch the fight from the cockpit. I took 109 vs Spitfire, 109 rolled like mad, what human can't achieve at 450KM/h(when 109 becomes like a brick http://forums.ubi.com/groupee_common/emoticons/icon_rolleyes.gif) AI does at much higher speeds... </div></BLOCKQUOTE>

Brain we need numbers, we have to prove that a AI 109 rolls (or doesn't!) faster than the real life maximum roll rate at given speed! Same goes with AI diving, AI overheating etc. But it's worth discussing http://forums.ubi.com/images/smilies/25.gif

Brain32
01-22-2006, 06:06 AM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content"> Brain we need numbers, we have to prove that a AI 109 rolls (or doesn't!) faster than the real life maximum roll rate at given speed! Same goes with AI diving, AI overheating etc. </div></BLOCKQUOTE>
Numbers are going to be hard to get, but I have an idea how to test it(maybe). Now please understand that I don't know nothing about DeviceLink, but couldn't we measure it that way? I mean take a plane select autopilot and collect data with help of DeviceLink...

lowfighter
01-22-2006, 06:52 AM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Brain32:
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content"> Brain we need numbers, we have to prove that a AI 109 rolls (or doesn't!) faster than the real life maximum roll rate at given speed! Same goes with AI diving, AI overheating etc. </div></BLOCKQUOTE>
Numbers are going to be hard to get, but I have an idea how to test it(maybe). Now please understand that I don't know nothing about DeviceLink, but couldn't we measure it that way? I mean take a plane select autopilot and collect data with help of DeviceLink... </div></BLOCKQUOTE>
I don't know about devicelink either. The point is anyway to come with relevant tests. I was thinking that perhaps doing tests with airplanes which have very different roll rates would be interesting. Say match a zero and a FW190 at high speed. The AI zero is behind the AI 190, very close, the 190 will go into a fast roll as usually and dive. The zero will try to roll as fast as possible too, perhaps playing the track with 1/4 speed and measuring the zero's roll time would do. Though the roll will not be complete!

VW-IceFire
01-22-2006, 07:51 AM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by AKA_TAGERT:
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by VW-IceFire:
Trust me...I don't trust my own recollections all that well. I'm a healthy skeptic.

So what I've done is do 1 vs 1 with the same aircraft. My job is to follow the Veteran or Ace AI as a trail chaser replicating their manuvers as best I can. Not a perfect test either.. </div></BLOCKQUOTE>Roger, not perfect, but about as good as we can do in light of the fact we can not get any DeviceLink info from other planes.

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by VW-IceFire:
but how an AI Zero pilot can outroll me with ease suggests that they play by different rules. </div></BLOCKQUOTE>Now your talking.. In that as I pointed out, Oleg has said the FM is the same for both.. So all that leaves is the stick force.. *IF* note, not agreeing here, but *IF* the AI is out rolling you with ease.. that only leaves the stick force *AND* it can only happen at a point above the RR peak.

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by VW-IceFire:
I know what Oleg said...and its contradictory to what I see. </div></BLOCKQUOTE>Well, like I also said, humans make for the worst recording and measurement devices.. TWO people can watch the same event and have TWO totally different takes on it. Not saying your wrong.. not saying your right.. All Im saying is nobody has provided anything up to now to even being to prove it is so.

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by VW-IceFire:
Thats all I can speak to...just my version of how the bank was robbed I suppose http://forums.ubi.com/groupee_common/emoticons/icon_smile.gif </div></BLOCKQUOTE>Roger! http://forums.ubi.com/groupee_common/emoticons/icon_wink.gif

I do have any idea though.. never tried it.. but.. can you set up a mission.. 1 vs 1 and then enable the autopilot and allow him to fight the fight? That is to say AI vs AI where the AI is flying your plane? If so, then record the whole thing in a track file. Then play it back and note the TIME you see what LOOKS LIKE your autopilot (ie AI) rolls your plane faster than you think it can go. Then, I could process the track file at that moment and see what the roll rate is.. and if it is higher than it should be.

The closer to 10kft this event happens the beter.. because then we can comp it to the NACA RR data. Oh, and pick a plane from the NACA list too!

Now.. *IF* the AI can out roll you with EASE.. THAN it should be a very Very VERY much larger RR than what the plane can do and thus show up in the analysis as a RR that is at least 50% higher than the max roll rate.

If and when somebody does that.. Than I will be sold on the idea.. But until then, not so much! http://forums.ubi.com/groupee_common/emoticons/icon_wink.gif </div></BLOCKQUOTE>
Interesting concept actually...I'll look a little further into that and see what I come up with. Cheers!

AKA_TAGERT
01-22-2006, 09:56 AM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Brain32:
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content"> Brain we need numbers, we have to prove that a AI 109 rolls (or doesn't!) faster than the real life maximum roll rate at given speed! Same goes with AI diving, AI overheating etc. </div></BLOCKQUOTE>
Numbers are going to be hard to get, but I have an idea how to test it(maybe). Now please understand that I don't know nothing about DeviceLink, but couldn't we measure it that way? I mean take a plane select autopilot and collect data with help of DeviceLink... </div></BLOCKQUOTE>Yes it can be measured via devicelink. On the previous page I described a method on what to do. Basiclly set up a 1 on 1. Pick a plane from the NACA list so we have some numbers to compare to. Also turn the "VULNERABILITY" option OFF. This will allow your plane to take hits and never get any damage. Now, start the 1 on 1, allow the other AI plane to get on your six. Then start recording a track file (use the quickrecord). Then enable auto pilot (ie allow the AI to fly your plane) and then watch. Let the two AI planes duke it out until you see your plane (that the AI is currently flying) preform one of these unrealisticly high roll rates. After which, wait a second or two then stop the recording.

Then send me your track file and I can use devicelink to measue the roll rate of your plane.. ie the roll rate the AI was able to do. Then we can compare that to the roll rate I was able to do. Now if what you guys are saying is true, the AI roll rate should be much higher.

The problem here is there is going to be a small error the farther this happens away from 10kft. In that all the roll rate data I have is at 10kft. So, I would also sugest that you start the 1 on 1 at about 15kft to 20kft, in that the AI will surly dive down a bit.

PS you may want to make the other AI and ACE and your AI group novice.. Otherwise your plane might end up on the six of the other plane.. and we dont want that.. we want the other AI to chase YOUR plane so that YOUR AI will preform one of these majic roll rates. Because Devicelink only allows you to look at the data of YOUR plane not any other plane in the track file.

lowfighter
01-22-2006, 10:45 AM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by AKA_TAGERT:

PS you may want to make the other AI and ACE and your AI group novice.. Otherwise your plane might end up on the six of the other plane.. and we dont want that.. we want the other AI to chase YOUR plane so that YOUR AI will preform one of these majic roll rates. Because Devicelink only allows you to look at the data of YOUR plane not any other plane in the track file. </div></BLOCKQUOTE>

Tagert, I think we all agree that rookie AI is doing very moderate maneuvers? I think in case the rookie test reveals nothing it's worth testing higher AI level.

AKA_TAGERT
01-22-2006, 11:09 AM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by lowfighter:
Tagert, I think we all agree that rookie AI is doing very moderate maneuvers? I think in case the rookie test reveals nothing it's worth testing higher AI level. </div></BLOCKQUOTE>Sure, no prob, I just thought it might save you some time in that maybe your AI might get on the six of the other AI and shoot him down before the test is done.

Basiclly do what ever it takes to cause the AI flying YOUR PLANE to do this majic RR your all seeing. And the closer you get this to happen at 10kft the better! Speed is not important, as long as it is not over 400mph IAS.

Just dont send me a track file of your plane with no wings on it because the AI shot them off and your spinning to earth at 600mph! That wont prove anything! http://forums.ubi.com/groupee_common/emoticons/icon_wink.gif

jeroen_R90S
01-23-2006, 12:35 PM
Oké, did some testing after watching AI autopilot going 700km/h with my La-5 in a campaign while mine loses parts at 630km/h. (speedbar) and just read this thread.

I'm a real n00b as far as this is concerned, but I made 2 tracks. (converted to .NTRK)

Setup:

La-5, 100% fuel, crimea map, 12:00 etc, starting @5000m.

Placed AI Fw-190A4 in FMB @ 500m to make AI autopilot dive at it, which it does, reaching ~800kp/h without wing loss, blackouts and other parts falling off.

I then sort of did the same thing, but losing parts and eventually lawndarting @800km/h.

I don't know if they're of any use, but they might provide a starting point for someone else to make better ones or perhaps use these.

Since my webspace is still down I used savefile.com but can e-mail if requested as well... (if needed...)

http://www.savefile.com/files.php?fid=1412260

Jeroen

jeroen_R90S
02-09-2006, 06:30 AM
?

Nobody did any testing to make better tracks than my crappy attempt? http://forums.ubi.com/images/smilies/16x16_smiley-indifferent.gif

Or is it better to wait for the new patch?

LEXX_Luthor
02-09-2006, 05:27 PM
Arrow (last page):: <BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">AI still use SIMPLIFIED models at least at high speeds. I got my campaign scrapped because of that, after 54 missions. I had autopilot on and it was diving down the plane, I didn't realize that the speed was 780 km/h and switched off the autopilot, the plane (LaGG 3 S4) immediately disintegrated... </div></BLOCKQUOTE>
Arrow, maybe disintegration is not considered part of the flight model -- think of tailplane flutter at high speed that is fixed with structural strengthening thus allowing higher speed without breakup. The beefing up is NOT flight model, but more like damage model with respect to airflow, but the Beef allows high speeds with safety. Thus AI and humoid use the same flight model but the AI can dive faster without breaking up. This is just a hunch, or a feeling.

TAGERT, whenever I roll as much as possible, in any fast rolling plane (-190, -16, etc...) the external view roll rate "looks" so much faster than cockpit view roll rate. They are the same roll rates obviously, but this could make people *think* they see AI roll "faster."