Military AI Works

Hey Drew,
Where you been brother! Long time no hear.

Does the cross hair underneath the model not need to be in the centre line of the model? I don't know I'm just curious.

Normally the answer is yes. However I found while working on the Tucano and L-39 it helps improve the stability of the AI aircraft in game when the centre of gavity is made lower. It helps to remove those wobbly wings and lifted wheel when doing a tight turn. Also makes the FDE production a little less stressful.
Typically the cross hair would be just in front of the rear gear so the weight balance is slightly over the forwad fuselage selection.

Nice one Mark,I am glad to see you are back.Talking about cross hairs,I think when gear down they should be in negative,other than that it can affect FDE when aircraft is on the apron and when AI engine is on.It appears that suddenly aircraft drop down little bit.
I hope you understand what I want to say


Miljan

Hi Miljan,
I think with the L-39 the gear was only in negative latitude when gear compression was extended, other than that I think the L-39 FDE has it flaring and touching down OK, but I never really looked that hard at it to tell you the truth but I get your point!
With the F-35 in its current position it may be a little nose heavy as the main gear is situated a fair way back under the wings but all can be moved to a better balance position prior to doing the animation key frames ( if I remember that is).

I'm still trying to find a balance in poly usage verses the quality of animated parts. As you can see I have the V-toll inlet cover very basic in design, that may change in the final main model depending on where I am in poly usage.

I'm struggling a bit with the underneath design so that may take a while to mould into something kinda like what it should be. Sure wish I had Andrew to guide me through some of the rough patches. His help was always a boost to my motivation.
PS: I think it kind-a looks, resembles an F-35B.

fishlips wrote:

Does the cross hair underneath the model not need to be in the centre line of the model? I don't know I'm just curious.

Normally the answer is yes. However I found while working on the Tucano and L-39 it helps improve the stability of the AI aircraft in game when the centre of gavity is made lower. It helps to remove those wobbly wings and lifted wheel when doing a tight turn. Also makes the FDE production a little less stressful.
Typically the cross hair would be just in front of the rear gear so the weight balance is slightly over the forwad fuselage selection.

miljan wrote:Nice one Mark,I am glad to see you are back.Talking about cross hairs,I think when gear down they should be in negative,other than that it can affect FDE when aircraft is on the apron and when AI engine is on.It appears that suddenly aircraft drop down little bit.
I hope you understand what I want to say

Miljan

fishlips wrote:Hi Miljan,
I think with the L-39 the gear was only in negative latitude when gear compression was extended, other than that I think the L-39 FDE has it flaring and touching down OK, but I never really looked that hard at it to tell you the truth but I get your point!
With the F-35 in its current position it may be a little nose heavy as the main gear is situated a fair way back under the wings but all can be moved to a better balance position prior to doing the animation key frames ( if I remember that is).

Mark,

I hesitate to interject my thoughts here, but what Miljan is referring to is some problems we had with the MGMK AI KAI KT-1 model. I found that the zero datum reference point was actually located slightly below ground level. This created a situation where the static cg height had to be set at a slightly negative number in order to have the tires in contact with the ground when FS9 first started up. Unfortunately this setting prevented this model from taxiing out properly and I found that I had to raise the static cg height to 0.35 in order to correct this. This created a situation where the MGAI KAI KT-1 model's tires were elevated slightly above the ground when FS9 first started at which point the model immediately dropped down so as to have the tires in contact with the ground as the contact point settings took over control of its height above the ground.

Neither of us was satisfied with this compromise, so Miljan found a way to adjust the model so that the zero datum reference point was back up in its normal middle of the fuselage position and after that I could use more typical static cg height and contact point settings so that the model's tires were always in contact with the ground, the model didn't drop down to the ground at FS9 startup and it taxied out properly.

Personally I have found your models with the lower zero datum reference point, the MGAI Shorts Tucano, MGAI Aero L-39 & MGMK AI KAI KT-1, difficult to work with and especially with finding the contact point settings necessary to keep their tires properly in contact with the ground when they are parked, taxiing, taking off and/or landing.

I'm certainly biased, but my suggestion would be to always place the zero datum reference point in its normal centrally located position and then adjust the empty weight cg position setting up or down to get the AI FDE performance that you desire. However, I also realize that these are your models and you can place the zero datum reference point wherever is most useful and beneficial to you.

Mike M.

No worries Mike, I thought we were on the same page with the Tucano but I guess not.

I'm happy to place the centre cross hairs in the same position as used by Nick Black if that helps to make the job of the FDE easier. I do recall we had some difficulties with models that had suspenion that otherwise moved on an angle rather than straight up and down...

Some of those models also use the new custom xml coding that made other issues a factor in the FDE, such as landing aproach speed, drag, etc to achieve smooth flap movement / position.

The Kia given to Miljan and Mark C was basically an unpolished project. While it was supplied to Miljan as main model and five lods the model required more refinement. For example the nose wheel strut was given animated suspenion but the two main gear wheel struts were not. (as the screenie shows below).
This could have upset the positioning on the afcad, however as you can see the main gear wheels are sitting dead on ground level while the front wheel is under the ground level freshhold line. The front wheel should be level with the ground line in the simulator with suspenion operating but I guess for some reason that is not occuring. It does however sit level on the ground when made into a static scenery object which means there are other factors at play to do with AI aircraft.

I'm happy to work it any way that makes life easier for others to do there magic when I'm made aware of the issues there having!
I should also mention that my paintkits are crap and not professional but hey, the models are free and people can take them or leave them.

PS: I believe it was the PC-21 that gave a fair bit of greif however this model was built normally on the cross hairs but he suspension had angled movement like a Hawk.

Hi Mark,
My own personal choice....
I put the cross hairs 1/3 to midway back on the wings on the Z axis, and centred on the wings in the Y axis.

If an aircraft jumps when it becomes active - the Static C of G in the aircraft.cfg file is off.

Doing the nozzle on the VTOL one is going to be a pain.
I think you are going to need two cylinders where the meeting faces are at an angle.
Both would need to be animated to get the 'swivel' motion.

You will also need to use the default FSDS/MSFS ugly lights on it

Thanks Kev, I'll do that in the future.

I just thought after some of the drama we had agers ago with wobbly wings and lifting wheels on some of my early models that it was easier to lower the centre of balance in FSDS rather than in the FDE. Several models on Mike M has now made me aware that I infact made it more difficult to achieve a good FDE. Armed with that knowlwdge I can now do things differently to assist others who are involved in the process.

I have some ideas for making the animated nozzle parts but whether I can keep the poly count down is the real trick. The xml will be the tricky part to make it all fit in smoothly on take-off and landing. Could be a few of those triangle bits go in. lol

This is a demanding AI, unlike anything done in the past so whether it can be pulled off will come down to a lot of people much smarter than me.

Hehe it is demanding airplane on RL why it woldnt be demanding as AI

At this point in development, I'm at 500 ACM polys. It's looking like this model will be too heavy to be of use an an AI aircraft. It's going to take a fair whack of polys to do the turbine animations, landing gear, flaps and bay doors, etc.
It will of course come in at an acceptable poly count for an on-line multi-player model.

May be the A and C model could be made with an acceptable poly count for AI use ???

Cheer's,

Mark even if you finish in 7000+ poly in main LOD still you can use LODs to save fps.Better computer are coming and this should not be a problem,just keep going

Mark,
I would agree with Miljan here. The target for polys is just that. It is a nice to have and timely reminder to keep the weight down, just like real aircraft.
However, not to do it because it is 1000 polys over the limit would seem counter productive. As Miljan says, even if you have an ASP full of 40 of them not all of them will be viewed as LOD1, the most detailed so it shouldn't be to much of an issue.

I'm attempting to keep the F-35B within 500 polys of Nicks Hornet - RAAF upgrade version. The aircraft will be finished but don't expect good frames from it in FSX if used as AI.
I'm tending to learn more toward doing an un-animated version which will strip away the need for much of the nozzel parts. I can also dump the boarding ladder and hatch and fuel boom and hatch if need be. Lets just see where it ends up but I know the on-line Vatsim guys should fancy it.
I want the main model under ACM 3000 polys, that's about 1400 FSDS polys. It's now at 337 FSDS polys with the elevator now attached.
Got my orders so I'm off for now!
Cheer's

You remember number of poly in PC-9/T-6 and everything is working well even more than 30 of them are in one place.Quality Mark,that's who,AI modelers,we are
Push it put to 6000 if needed but don't lose the shape of it.

I have to ask, Mark who is she?

Ha ha, Sammy is not a porn star if that's what come to mind.
Click the link in my signature and all will be revieled.
nighty_night y'all,

I wouldn't worry about the poly count for LOD 1.
Where you really save on the frame rates is the further away LOD's.

Some aircraft are always going to be heavier on polys than others, helo's are a perfect example.

We're 8+ years from the release of FS9 and many years now removed from even what the AI Aardvark guys thought were possible at the time. The original ideas on poly counts are not as critical as they were in the old days. The processors most people have these days are more than adequate.

Besides when was the last time you saw the sim stutter to actually render a model, it's always, always the textures that are the last thing to load up.

Guy's, I can appreciate all that but I'm also keeping FSX use in mind also. As I know nothing about FSX native conversion at this stage I'm trying to keep everything as low as posible. Also I'm keeping the vatsim multi-player guys in mind where the Vatsim server exchange can only handle so much info.
It's also more chalenging to keep it low and tight.
I'm thinking of giving FSDS the flick and moving to 3D max. Although its more complex to use there are many more skilled people out there to offer advice and hands on help when you get stuck and the program isn't glitchy like FSDS 3.5.1 and it can interface with G-max.

I kind of like the screenies below!

Houston, we have a problem. I've run into a design quagmire. I have several photos showing different shaped gear bay compartments. I'm not sure which one is the current production model. bugger!