#3675822 - 11/05/12 07:14 PM
Re: The almighty Flight Model thread
[Re: PanzerMeyer]
|
Joined: Aug 2012
Posts: 763
WhoCares
Member
|
Member
Joined: Aug 2012
Posts: 763
|
Please correct me if I'm wrong but the issue (at least for me) is not whether the FM's are too inaccurate but whether AI controlled planes use a different FM from what human players get. I have been suspecting this for a while because over the course of flying many missions in career mode I've noticed AI controlled planes pulling maneuvers that I myself have never been able to do in the same plane without tearing off my wings or spinning the plane. Consider yourself corrected In context with a different topic, namely with the request to introduce "pure" AI planes, it is regularly stated by Jason that this does not make sense from a development perspective, as the main effort in the development process is the flight model, and the AI gets the same flight model as the player. What you might experience is an (ace) AI perfectly handling the plane at the edge of its envelope, in a way that only few players might ever be able to do. E.g. (I don't really know, but I assume) in any environment, altitude, temperature the (ace) AI will always have a perfect engine setup. As has been stated, there are so many factors regarding FM parameters and the way they were measured in different tests, we will never get the perfect model. Subjective pilot accounts are also only of limited value, as, well, there is a reason why they are call subjective...
|
|
#3675964 - 11/05/12 10:38 PM
Re: The almighty Flight Model thread
[Re: TheBlackPenguin]
|
Joined: Feb 2000
Posts: 49,716
Jedi Master
Entil'zha
|
Entil'zha
Sierra Hotel
Joined: Feb 2000
Posts: 49,716
Space Coast, USA
|
Yes, it doesn't matter that they have the same FM because they are capable of keeping it right on that knife's edge the entire time. They never accidentally go over that line, or back off a bit to gain more E before tackling it again. It's like throwing a bullseye playing darts. Sure, lots of people have done it before, but how many can do it EVERY time? The AI is doing the equivalent of that.
The Jedi Master
The anteater is wearing the bagel because he's a reindeer princess. -- my 4 yr old daughter
|
|
#3676372 - 11/06/12 01:47 PM
Re: The almighty Flight Model thread
[Re: PatrickAWilson]
|
Joined: Apr 2001
Posts: 121,383
PanzerMeyer
Pro-Consul of Florida
|
Pro-Consul of Florida
King Crimson - SimHQ's Top Poster
Joined: Apr 2001
Posts: 121,383
Miami, FL USA
|
Sorry. Engineer by trade. No sense of humor or irony I think CPA's fall under that category too. @ LukeFF - Yeah, I've seen that too in some of my career missions where an AI plane would start spinning and then not recover in time before it hit the ground.
Last edited by PanzerMeyer; 11/06/12 01:49 PM.
“Whoever fights monsters should see to it that in the process he does not become a monster. And if you gaze long enough into an abyss, the abyss will gaze back into you.”
|
|
#3676456 - 11/06/12 03:57 PM
Re: The almighty Flight Model thread
[Re: TheBlackPenguin]
|
Joined: Feb 2000
Posts: 49,716
Jedi Master
Entil'zha
|
Entil'zha
Sierra Hotel
Joined: Feb 2000
Posts: 49,716
Space Coast, USA
|
Perhaps the problem then is some of the mission makers deciding to put unrealistically high AI settings in their missions (ie "the AI is too easy on anything but uber-Ace, so every plane will be that!")
The Jedi Master
The anteater is wearing the bagel because he's a reindeer princess. -- my 4 yr old daughter
|
|
#3676523 - 11/06/12 05:31 PM
Re: The almighty Flight Model thread
[Re: LukeFF]
|
Joined: Jul 2001
Posts: 419
2Lt_Joch
Member
|
Member
Joined: Jul 2001
Posts: 419
Montreal, Canada.
|
Yes, it doesn't matter that they have the same FM because they are capable of keeping it right on that knife's edge the entire time. They never accidentally go over that line, or back off a bit to gain more E before tackling it again. Not so. I've seen undamaged AI fighters go into a spin and in some cases also crash due to them not being careful enough. In fact, that's how I finally completed the final training mission, which pits the player's Spad against two Fokker D.VIIs. I got one to spin and crash from a low altitude without putting a single bullet into it, and then the second one I downed by the more traditional means of gunfire. yes, I have seen the AI lose control of and spin DVIIs, Spads, Se5as and DR.1s. It seems to have problems with some FMs. The AI does better with turn fighters. Is it clear they use the same model we do? I had read that they have less mixture settings?
Intel Q9550, Gyga P35-DS3R, XFX 6950 XXX, 27" widescreen, 8 g. DDR2 @800, 2xWDRaptor 36g HD @ RAID 0, 1tb WD Caviar black HD, X-Fi Fatal1ty, win 7 64bit ultimate, Cougar/FSSB/HS1, Tir4.
|
|
#3676623 - 11/06/12 07:23 PM
Re: The almighty Flight Model thread
[Re: PatrickAWilson]
|
Joined: Apr 2004
Posts: 2,426
Wodin
Member
|
Member
Joined: Apr 2004
Posts: 2,426
Liverpool
|
Well said...the only issue I've seen a few occasions is a DR1 with two wings missing on one side still fly perfectly and land no problems..seems odd to me..weight balance alone surely should make that difficult. Apart from that really I have no real issues with the FM...it's the lack of WW1 atmosphere thats the main drawback for me..like alot of the better Sims that do the sim side really well (DCS) the actual game is very dry and lacks immersion\atmosphere outside the cockpit. What I will say is dismissing the views of the actual pilots that flew the things in combat feels like a huge disservice. Too say we know better because we have a PC game...come on now. The case for hard data: human perception is off. No two people see things the same way.
Why pilot perception is not good: - Pilot quality is not taken into account, as it is unknowable, and yet vital for getting the best fron an airplane. - There are variables in real life scenarios that alter the outcome but are not taken into account by the participants. - Example: That plane is faster than this one - well maybe that plane was already in a shallow dive and had built up a head of steam. - In the heat of combat, flawed perception is the norm and not the exception.
The case for pilot accounts: WWI data is notoriously bad. The hard data simply does not exist. Therefore pilot perception is key.
Why hard data is not good: - Variations in rigging could affect performance. - Airframe aging could affect performance. - Engine quality could affect performance. - Pilot ability could affect performance. - Sometimes tests were performed on specially rigged factory planes. - Sometimes tests were performed on war weary captures with unfamiliar pilots and incorrect parts (fuel, propeller, just to name a few). - German HP results were at fixed RPMs - Allied results were recorded at optimal RPMs.
Argue away, but anybody that believes that they have the answer is wrong from the start. As soon as you start quoting performance figures I tune out. That was maybe one plane, one time. It is a data point only, not TRUTH.
So, where do we go? The TRUTH is not known and will be - probably because a single TRUTH never existed. However, if we combine pilot accounts with as many data points as possible we can get something that is sort of reasonable. With that in mind, lets ask for that which is unreasonable to be fixed, but limit things to that.
Last edited by Wodin; 11/06/12 07:27 PM.
|
|
#3676713 - 11/06/12 09:01 PM
Re: The almighty Flight Model thread
[Re: TheBlackPenguin]
|
Joined: Jan 2002
Posts: 6,623
Mogster
Hotshot
|
Hotshot
Joined: Jan 2002
Posts: 6,623
England
|
The damage model is very complex, sometimes you get a miss-match between the visual and physics going on under the bonnet. It seems to happen less often than it did a couple of years ago.
I'd rather have ROF's dynamic damage and physics and see the occasional screw up than canned damage effects being added over the flight models
WAS C2D 8500 3.16ghz, 285gtx 1gb, 4gig ram, XP NOW Win7 64, I5 2500K, SSD, 8Gig ram, GTX 570
|
|
#3676718 - 11/06/12 09:06 PM
Re: The almighty Flight Model thread
[Re: 2Lt_Joch]
|
Joined: Jun 2001
Posts: 11,234
LukeFF
Veteran
|
Veteran
Joined: Jun 2001
Posts: 11,234
Redlands, California
|
Is it clear they use the same model we do? Yes
|
|
#3677490 - 11/07/12 09:15 PM
Re: The almighty Flight Model thread
[Re: TheBlackPenguin]
|
Joined: Jul 2001
Posts: 419
2Lt_Joch
Member
|
Member
Joined: Jul 2001
Posts: 419
Montreal, Canada.
|
The other problem with pilot quotes is how do you interpret them? Adolf Von Tutschek flew Alb. D.III-D.Vs in 1917 until he was wounded in august and went home for 6 months. While there, he wrote his memoirs, an exceprt of which reads: In the past Four weeks three new types of enemy aircraft have appeared. They are without a doubt far superior in their ability to climb than the best D V. They are the new English SE 5 single-seater, the 200hp SPAD and the very outstanding Bristol Fighter two-seater. While the Albatros D III and D V come near in their ability to climb with the Sopwith and Nieuport, and even surpass them in speed, it is almost impossible for them to force an SE5 or a 200hp SPAD to fight because the enemy is able to avoid it by the ability of his craft to outclimb the Albatros.
From his quote, we can glean: 1. there were performance differences between Albs since he refers to "best D.V"; 2. The Alb D III and D V had similar performances; 3. The Alb D III and D V had a slower climb rate than the "Sopwith" and the "Nieuport", probably a reference to the Sopwith Pup and the N.17 since he fought both. The ROF Pup has a climb rate of 13 mins to 3 km in ROF, but everyone agrees it is about 10% too high, it should be closer to 14 1/2 mins. The ROF Alb D III and D V have a climb rate of, respectively, 14 1/2 mins and 13 1/2 mins to 3 km. So they already pretty much match up. 4. The Alb D III and D V were faster than the Pup and N17, but it does not sound by much (..."even surpass them in speed"...). The ROF Alb D III and D V are already about 10-25 km faster than the ROF N 17 depending on the altitude. The ROF Alb D III is already 5-10 kmh faster than the Pup between 2-4 km altitude. So they already pretty much match up. So if I read this "pilot note" of an actual German pilot who flew actual Albatros scouts for 6 moths in 1917 and shot down 23 actual planes in that time, it sounds like the ROF Albatros is already in the ballpark.
Intel Q9550, Gyga P35-DS3R, XFX 6950 XXX, 27" widescreen, 8 g. DDR2 @800, 2xWDRaptor 36g HD @ RAID 0, 1tb WD Caviar black HD, X-Fi Fatal1ty, win 7 64bit ultimate, Cougar/FSSB/HS1, Tir4.
|
|
#3677748 - 11/08/12 06:22 AM
Re: The almighty Flight Model thread
[Re: Tarnsman]
|
Joined: May 2007
Posts: 89
imaca
Junior Member
|
Junior Member
Joined: May 2007
Posts: 89
|
The answer is obvious: improved modeling. Forget putting in "it could go X kts at Y altitude!" We should put in the plane's schematics and engine design and the sim should put out a plane that flies exactly like the real one did!! It's just number crunching, put enough numbers in and you'll get the proper outcome!!!
The drag of every guide wire, the weight of different paint schemes, the airfoil of the prop, the power generated in each piston...all of this can lead to the performance being dynamically ascertained with the proper design input into the sim!!!! Add in accurate modeling of temperature, humidity, and oxygen variations with altitude, how the engine oil and fuel reacts to those changes, historical notes on how well each squadron kept its planes maintained along with the general level of quality expected from the factories, and it will all just write itself!!!!!!!!!!!
You won't have to wonder "should the Camel out-dive the Alb at 1000m?" because you'll know from the sim the answer to that! What's so hard about this idea??!?!??!?!?!?!?!?!?!?!?!?!
The Jedi Master I wish that was how it could be done. Perhaps in the not so distant future. This is what Xplane claims to do and RoF to some degree. I dont know how far along flight modeling has come, but I would think air and physics are known quantities that could be modeled in a flight sim. Then plug in the appropriate airfoil shapes, masses, and engine performances like you point out. With enough processing power I don't see why it cant be done. It could be done, with enough processing power. There's no way that it is done in real time in any flight sim. What x-plane and ROF may do is some kind of simulation using all kind of simplifications and rules of thumb. Real simulation of airflow requires expensive software ($20000+/ seat), and from experience, the object being simulated is typically simplified to speed up processing time. In my case the object being simulated is probably about 1000x simpler than a WW1 aircraft. In each particular case a myriad of parameters are tweaked (eg trying to determine areas of laminar vs turbulent airflow) to try and get something approaching reality. In the case of WW1 aircraft some of the things that would have to be modeled include deformation of airframe in response to accelerations and aerodynamic pressure and deformation of the fabric skin (aeroelastics). For example the Albatross D3 had a problem with flutter of the lower wing caused by the wing twisting under aerodynamic pressure. The resulting aerodynamic force on the wing could rip them off. When tested under static loads (the only way known at the time) the wing seemed to have ample strength. The day when the physics of an accurate 3d model of an aircraft can be accurately simulated in real time on your PC is probably many decades away.
Last edited by imaca; 11/08/12 06:26 AM.
|
|
#3677917 - 11/08/12 02:29 PM
Re: The almighty Flight Model thread
[Re: 2Lt_Joch]
|
Joined: Jun 2010
Posts: 615
RoFfan
Member
|
Member
Joined: Jun 2010
Posts: 615
|
4. The Alb D III and D V were faster than the Pup and N17, but it does not sound by much (..."even surpass them in speed"...). The ROF Alb D III and D V are already about 10-25 km faster than the ROF N 17 depending on the altitude. The ROF Alb D III is already 5-10 kmh faster than the Pup between 2-4 km altitude. So they already pretty much match up.
Where do you get that information? The RoF store page shows the Pup and D.III having the same airspeed at 3km. The Albatros was also faster than the Sopwith Camel and Fokker Dr1, but in the game it is the other way around.
|
|
#3677929 - 11/08/12 02:51 PM
Re: The almighty Flight Model thread
[Re: RoFfan]
|
Joined: Jul 2001
Posts: 419
2Lt_Joch
Member
|
Member
Joined: Jul 2001
Posts: 419
Montreal, Canada.
|
Where do you get that information? The RoF store page shows the Pup and D.III having the same airspeed at 3km.
From the TAS charts Gavagai had prepared. http://riseofflight.com/Forum/viewtopic.php?f=49&t=33269 The Albatros was also faster than the Sopwith Camel and Fokker Dr1, but in the game it is the other way around. That is two separate issue, it is far from clear that the Camel was slower than the Albatros. That depends on whether you believe the hard data or what some Camel pilots thought. However, the fact that some Camels may have been slower than some Albatros does not mean that the speed of the Albatros in game is too slow. Here we have an actual quote of an Albatros pilot on the speed of his plane in relation to other planes. That is a lot more pertinent to the speed of the Albatros than what Camel pilots thought about the speed of the Camel. The DR.1 is another issue altogether. I think everyone pretty much agrees that it is too fast in game, but the answer for that is to slow down the DR.1.
Last edited by 2Lt_Joch; 11/08/12 03:07 PM. Reason: additions
Intel Q9550, Gyga P35-DS3R, XFX 6950 XXX, 27" widescreen, 8 g. DDR2 @800, 2xWDRaptor 36g HD @ RAID 0, 1tb WD Caviar black HD, X-Fi Fatal1ty, win 7 64bit ultimate, Cougar/FSSB/HS1, Tir4.
|
|
#3677975 - 11/08/12 04:11 PM
Re: The almighty Flight Model thread
[Re: RoFfan]
|
Joined: Apr 2008
Posts: 788
NattyIced
Member
|
Member
Joined: Apr 2008
Posts: 788
|
The Albatros was also faster than the Sopwith Camel and Fokker Dr1, but in the game it is the other way around. Based on two books that have no data just impressions. One of which is a work of fiction.
|
|
#3678147 - 11/08/12 08:12 PM
Re: The almighty Flight Model thread
[Re: imaca]
|
Joined: Jan 2002
Posts: 6,623
Mogster
Hotshot
|
Hotshot
Joined: Jan 2002
Posts: 6,623
England
|
The answer is obvious: improved modeling. Forget putting in "it could go X kts at Y altitude!" We should put in the plane's schematics and engine design and the sim should put out a plane that flies exactly like the real one did!! It's just number crunching, put enough numbers in and you'll get the proper outcome!!!
The drag of every guide wire, the weight of different paint schemes, the airfoil of the prop, the power generated in each piston...all of this can lead to the performance being dynamically ascertained with the proper design input into the sim!!!! Add in accurate modeling of temperature, humidity, and oxygen variations with altitude, how the engine oil and fuel reacts to those changes, historical notes on how well each squadron kept its planes maintained along with the general level of quality expected from the factories, and it will all just write itself!!!!!!!!!!!
You won't have to wonder "should the Camel out-dive the Alb at 1000m?" because you'll know from the sim the answer to that! What's so hard about this idea??!?!??!?!?!?!?!?!?!?!?!?!
The Jedi Master I wish that was how it could be done. Perhaps in the not so distant future. This is what Xplane claims to do and RoF to some degree. I dont know how far along flight modeling has come, but I would think air and physics are known quantities that could be modeled in a flight sim. Then plug in the appropriate airfoil shapes, masses, and engine performances like you point out. With enough processing power I don't see why it cant be done. It could be done, with enough processing power. There's no way that it is done in real time in any flight sim. What x-plane and ROF may do is some kind of simulation using all kind of simplifications and rules of thumb. Real simulation of airflow requires expensive software ($20000+/ seat), and from experience, the object being simulated is typically simplified to speed up processing time. In my case the object being simulated is probably about 1000x simpler than a WW1 aircraft. In each particular case a myriad of parameters are tweaked (eg trying to determine areas of laminar vs turbulent airflow) to try and get something approaching reality. In the case of WW1 aircraft some of the things that would have to be modeled include deformation of airframe in response to accelerations and aerodynamic pressure and deformation of the fabric skin (aeroelastics). For example the Albatross D3 had a problem with flutter of the lower wing caused by the wing twisting under aerodynamic pressure. The resulting aerodynamic force on the wing could rip them off. When tested under static loads (the only way known at the time) the wing seemed to have ample strength. The day when the physics of an accurate 3d model of an aircraft can be accurately simulated in real time on your PC is probably many decades away. Some of these WW1 era aircraft use strange airofoil shapes that have never been properly wind tunnel tested. In that case it's next to impossible for a computer to model them correctly, that is you can make assumptions but none of your computer results can be validated using real world data.
WAS C2D 8500 3.16ghz, 285gtx 1gb, 4gig ram, XP NOW Win7 64, I5 2500K, SSD, 8Gig ram, GTX 570
|
|
|
|