Military AI Works

That's good news. Thanks for re-testing for me. I've done the correction to the mapping in the right hand wing root glove in the Raven model and I'lll do the others tomorrow and then post a temporary update on Rapidshare for you guys here. I'll also re-set the delay in the afterburner to suit the Lakenheath hold to start point distance. I'll wait though for a few days before a full version 2 public release just to make sure the crash problem has gone.

John

Hi all is where away to stop the aircraft jumping around so you may fly it .
Or will it only fly in ai..Here's hoping..

MW

Its AI only, however if you would like a freeware F-111...

http://www.virtavia.com/Freeware/index. ... =date&dir=

thanks Matt..

MW

Anyone else have this problem? (see screenie)

Aircraft bouncing on take off roll before leaping in the air way short of what would be a realistic take off run.

The screenshot is from a take off and you can see the tyre smoke kick up.

Also noticed that they tend to bounce (slightly) during taxi.

Just done a check and measure. The aircraft are going from standing start to wheels off in under 1900 feet (550 metres) which according to this site;

http://www.aircraft.co.za/Encyclopedia/G/75.php

is around half what it should be;

Take-off distance (to clear 50ft) : 950m (3,120ft)

Stew

Thats exactly what I am getting on take-off. Upon landing they hit the deck, have one big bounce and then slow in a cloud of smoke juddering all the time. When the smoke clears they taxi normally with the pilot holding a sign out of the window saying "Wasn't me".

Chris

well for all those no suffering jumps bumps and smoke, there are 29 Lakenheath paints with a crappy flight plan on my site now to finish the UH spark varks.

Chrisb wrote:Stew

Thats exactly what I am getting on take-off. Upon landing they hit the deck, have one big bounce and then slow in a cloud of smoke juddering all the time. When the smoke clears they taxi normally with the pilot holding a sign out of the window saying "Wasn't me".

Chris

No idea what the problem is, but for sure there is no way it should be getting airborne so quick.

Okay - I did a small experiment.

I've been using the Upper Heyford scenery I'm working on.Using John's Traffic_LN bgl, I altered the airport and swapped out the F-111 with a Mirimar F-14 (Nick Black MAIW)
and watched the take off performance. No bouncing - no smoke, so I concluded it was not my scenery......................let's face it, when I've been making changes, that was a distinct possibility.

So - that eliminated, I then swapped the air file from the F-14 into the F-111 folder and changed the entries accordingly.

Back in the sim - bouncing baby time again with loads of smoke and still (imho) taking off too quickly from a standing start.

I'm therefore left with the aircraft cfg file and as I know less than hee-haw about what changes what, perhaps someone can take over and see if this problem can be ironed out.

It's a bloody brilliant model and it's a shame that what appears to be a small thing seems to be causing these issues.

My hope is that a few more people now realize the role that Mike MacIntyre plays in our models and their development. What Mike does for almost every model we use is to make it come to life like that particular aircraft should.

One could model a box and yet make it fly like a fighter with the right knowledge. So relatively speaking, a good model is both a good 3D design and a good flight dynamics model design that work in concert with one another. The model being the visual aspect and the flight dynamics design being the simulation aspect.

As few modelers as there are left these days, there are even fewer who understand how a change in the Aircraft.cfg file or the Aircraft's .AIR file change the way the AI engine then treats the visual model.

The worst part is that the ACES team went and changed the relationships of the values in the flight dynamics model from FS9 to FSX. So what works to tame an AI model in FS9 does not necessarily work to tame the same visual model in FSX.

There are resources on line that describe how to set up the flight dynamics for a model. Most of these resources relate to user/flyable models, not AI models which behave differently. The trick is to take that knowledge and convert it to use with an AI model.

However, one small change in one part of the flight dynamics model usually results in an unwanted change to the model's behavior in another area. It's all trial and error as I'm sure Mike can tell you.

This issue seems to exist in FS9 only. With FSX I don't see any jumps, bumbs, smokes or other strange behavior.

I'm finding this really difficult, not just for the reasons that Mike G mentions, but because I really want to get it right for everyone. I'm going to try again today and rebuild the dynamics from scratch. My starting point was some study this morning of F-111 take off videos. The average time seems to be around 25 seconds (mine use 20 seconds). I found only a few references to take-off distance and the minimum seems to vary from 951m to 1348m. Taking the last figure and converting it to feet (4425ft) that equates to about half the Lakenheath runway length, mine uses about a third. I think that's the easy bit to fix, increasing the empty weight value (AI doesn't use Max weight) should take care of that, so long as the aircraft can then stop in an acceptable distance. Alternatively I can play with the static thrust.

One of the difficulties in all of this is that I can't just grab someone else's cfg and air file that works and tweak them. I don't own the copyright for publication. I have to start with a default pair of files and FS9 being what it is, they are all commercial jets. The Lear Jet however does provide some basis to work from. On top of that AI dynamics are different to user-flyable aircraft dynamics which is why I just can't contend with the latter - the AI is a challenge enough and building for FS9 and FSX complicates it even further.

The biggest problem I have is that I cannot reproduce the bouncing and accompanying smoke that some people are reporting in FS9 at any stage of movement at Lakenheath and I've watched it all again this morning. I do get the intended smoke on landing, but the aircraft does not bounce - the nose wheel goes down and it rolls-out smoothly. It doesn't even occur at 4x speed which usually accentuates problems like that. What I have found though when testing other aircraft in a much fuller environment is that a bounce can occur when the sim loads something that takes the resources for a fraction of a second and causes a hesitation or judder. The more add-ons the user has, the more that can happen I guess. Since I'm running almost clean, I wonder if that's a factor?

Anyway, I'll press on today and see what I can come up with. Difficult though, because I can't tell if I've corrected a problem I don't have.

John

Just out of interest are the people experiencing the bounces and the smoke, connected with TVB? (im purely talking in AI terms of behaviour, its not a flyable model remember)

Mark

I have the same bouncing problem, often also during taxiing. I'm sure it is not related to the scenery where you are testing aircraft, because I have the same problem in every airport I try.
I don't completely agree with John when he says "The more add-ons the user has, the more that can happen I guess" but there should be really something like that.
Many of us has the afterburners problem, he doesn't have
We have bouncing problem, he doesn't have.
On top of this, I already had a long exchange of post with John on ACG forum related to the FDE of the Thunderbolt, Mustang and T-6. (Do you remember john? )
Also in that case, he had not any problem and a very little number of users had, so I sorted it out modifying the FDE by myself. Don't know if I am able to do the same in this case.
BTW John, another big thank for you work.

Giorgio

Yes I do remember that Giorgio. My Thunderbolt, Mustang and T-6 still don't bounce. Can you remember what you did to cure the problems you were experiencing?

John

Hi John, here the old post on ACG Forum:
http://www.airfieldconstructiongroup.or ... 9&start=50

In particular the parameters changed for the P-47 was:
from
point.0= 1, -19.508, 0.000, -3.540, 1800, 0, 0.600, 180.000, 0.100, 2.000, 0.800, 2.000, 2.000, 0
point.1= 1, 1.640, -7.155, -7.477, 2200, 1, 1.450, 0.000, 0.160, 2.000, 0.800, 3.000, 4.000, 2
point.2= 1, 1.640, 7.155, -7.477, 2200, 2, 1.450, 0.000, 0.160, 2.000, 0.800, 6.000, 3.000, 3
to
point.0= 1, -19.508, 0.000, -3.95, 1800, 0, 0.600, 180.000, 0.3, 2.000, 0.800, 2.000, 2.000, 0
point.1= 1, 1.640, -7.155, -7.477, 2200, 1, 1.450, 0.000, 0.3, 2.000, 0.800, 3.000, 4.000, 2
point.2= 1, 1.640, 7.155, -7.477, 2200, 2, 1.450, 0.000, 0.3,2.000, 0.800, 6.000, 3.000, 3

The problem with P-47 was a little bit different, read the original post for further details.
Hope this help.

Ciao, Giorgio

Just been experimenting following Giorgio's post. It seems quite logical that altering the static compression in the undercarriage might help control any bounce in the aircraft, either to give it a bit more to absorb the shock on landing or to reduce it if it's acting like a trampoline. The figure in the aircraft.cfg is currently set at 0.478 which is quite springy (0=rigid). Nick Black, by comparison uses a figure of 0.4 in his F-15E.

The test I have just completed is with the static compression reduced to 0.3. I have also increased the empty weight so that the take-off run is extended. It is now 24 seconds, the same as the video I watched this morning and the aircraft now uses just about half the length of the Lakenheath runway but still stops with room to spare. I've also adjusted the static pitch and static height and verified it with AirWrench.

Finally, I have reduced the delay in the afterburners by 20 seconds. They now light at brakes off on the Lakenheath runway.

I just don't know if I have helped with the bounce that some people are experiencing. The new files, lke the previous set, don't produce any bounces for me. If you would like to try, replacement aircraft.cfg and afterburner files are in the attached zip. If it doesn't help the bounce, try a lower or higher value for the static compression in the [Contact Points] section of the .cfg file. Some of the other figures may need to be changed as a result, but it might at least tell me if I'm on the right track.
John

Spread the aircraft's weight out by moving the fuel values further out towards the wingtips. That will make it less jumpy.

That surprised me a bit Mike. I thought AI aircraft worked on empty weight, so the position of the fuel load I assumed was not important because there isn't one. However, always eager to learn, I tried adjusting the values. This is how they were:

[fuel]
Center1 = -41.4, 0.0, -1.5, 359.0, 0.0 //Longitudinal (feet), Lateral (feet), Vertical (feet), Usable(gallons), Unusable (gallons)
LeftMain = -36.3, -10.0, -1.5, 268.5, 0.0
RightMain = -36.3, 10.0, -1.5, 268.5, 0.0

I then changed them to move the tanks forward and up to the centre of gravity (0 ft longitudinal and 0 ft height) and the main tanks out to 30ft, like this:

[fuel]
Center1 = 0, 0.00, 0, 359.0, 0.0
LeftMain = 0, -30.0, 0, 268.5, 0.0
RightMain = 0, 30.0, 0, 268.5, 0.0

It didn't do any harm in my test, but since I didn't have the bounce problem in the first place, I can't tell if it solves anything. If anyone with the problem could try the values, I'd be interested in the result.

John

Hi John, the new cfg is better than the original. In my setup the 1-11 doesn't bounce anymore during taxi and take-off. It just smokes (without jump) between the Hold short point and the Runway start. More than enogh for me
The bouncing problem persist during landing.
It touches the runway, it bounces at 15-20 feet, then lands bouncing and smoking up to when it stops.
No differences between the new aircraft.cfg and the value indicated in the last post.

I have not so many time in these days, but I'll try to sort it out.

Giorgio