He's already modeling the systems. Power, hydraulics, etc. That's not flight model.

Ground handling is also not flight model. 

You don't want to override_planepath because then you're needlessly throwing away expensive terrain collision checks. You create a stub/placeholder .acf in planemaker with no wings/bodies and use _plug_acf. This retains ground collision and bypasses all of XP's force calcs. I vaguely recall seeing similar arbitrary force-injection functionality in asobo's SDK but can't remember the names at the moment.

What he would miss with either technique though is environmental feedback: weather, runway surface conditions, etc. Weather could probably be derived from the velocity vectors and state angles but runway surface.... there I dunno. I bet ground handling would remain native to the respective sim.

52 minutes ago, blingthinger said:

You don't want to override_planepath

If override_planepath is not set to 1, then it is 100% using X-Planes flight model, and will 100% need to use the MSFS flight model on that platform, behaviour and quirks on MSFS will be the same as every other plane on MSFS

52 minutes ago, blingthinger said:

you're needlessly throwing away expensive terrain collision checks.

afaik, you still get full ground interaction/collisions in xplane with override_planepath set to 1, to disable that you need to set override_groundplane to 1.

52 minutes ago, blingthinger said:

modeling the systems. Power, hydraulics, etc. That's not flight model.

Kinda, they are the first step, but in X-Plane lingo "defining the flight model" also refers to defining the interaction between the flight surfaces and physics model (of which they are part), but in common use "defining the flight model" is almost exclusively used to refer only to the physics model.

52 minutes ago, blingthinger said:

there I dunno. I bet ground handling would remain native to the respective sim.

"ground handling" doesn't really happen with override_planepath set to 1, because all the movement is still set outside the sim, stuff like suspension animation and crash detection do still work as long as override_groundplane is set to zero, the way to transition from one to the other is described in the next section of your link:

Quote

Transitioning From Disabled to Enabled Flight Model

When you re-enable the flight model, the sim will continue based on the velocity vector and quaternion q. So if you have set up the aircraft based on phi, psi, and theta for orientation, and some kind of speed, you will need to set the velocity vector to give the aircraft speed along its current heading, and conform the quaternion q to the current rotations. Once you do this, re-enabling the flight model should be relatively smooth.

Do that just after touch down and just after take off for XP provided ground handling, but then you still have to duplicate a lot for XP to handle the ground handling (and that wont be the same in MSFS).

use the "onground" datarefs to know when to switch.

In aviation a flight starts the moment the engines start, and doesnt finish until the blades stop turning. a complete "flight model" covers that entire period.

52 minutes ago, blingthinger said:

though is environmental feedback: weather, runway surface conditions, etc.

absolutely everything, "X-Plane as visuals" means exactly that, xplane is "input only" and just draws stuff based on pilot_head and aircraft position data refs.

Hence my "I'm not really sure what to make of this." comment wrt to "Toliss developed their own flight model for the A340 in collaboration with McGill University in Canada." - suspect that means something different than people interpret it to mean. That kind of miscommuncation causes problems, if that is indeed the case.

Edited July 5, 2025Jul 5 by mSparks

15 hours ago, blingthinger said:

Cheap and eager labor supervised by some of the top minds in their field with access to the best aerodynamic analysis software/hardware that money can buy.

Good one, chapeau! 😂

 

15 hours ago, blingthinger said:

XP is a generalized model. IF they were running CFD on the actual airframe geometry to acquire the force coefficients and then able to tune the 6DOF behaviors further with additional optimization techniques, they would end up with a better specialized model. "Fly by wire" is not "flight model". Fly by wire sits on top of it. It's an autopilot that's active all the time. For asobo20/4 at least, he's replaced both of those elements and claims to be on par if not slightly better than XP's default aero.

Running CFD on an entire A340 airframe under various flight conditions within two years time on a university cluster with people who know their toolboxes and its limitations and manage to get plausible results? Sorry, that's more unrealistic than a manned moon landing by 2030. A BET approach is more plausible.

FBW's envelope protections allow handwaving the need to explore edge cases because one doesn't enter them anyway.

 

15 hours ago, blingthinger said:

The engine model has already long been external. It is already more specialized than XP's generalized turbine could ever be. Further improvements would only expand that gap. Not shrink it.

Of course it's external, otherwise you couldn't switch engine types without switching to another aircraft.

25 minutes ago, Bjoern said:

that's more unrealistic

Incorrect. Very realistic given their compute cluster. Even more plausible with their GPU compute capabilities. Nor do you need to run at every.single.condition to fill up a lift/drag/moment polar. Worst case scenario is that you run a handful of high res simulations at some strategic points in the curves and use those as anchors (scale/shift factors) for the rest. Even if he is doing his own BET instead of a 1D table model, he'd still benefit significantly from using high resolution 2D or even quasi-2D CFD. All with tools he wouldn't necessarily have access to otherwise.

If he didn't run any CFD, the access to advanced optimization routines could help reverse engineer the Airbus aero for XP. But that alone doesn't yield a model that would be behave identically in asobo24. There's no question that he wrote his own 6DOF solver. That or dragged in something akin to JBsim.

Don't downplay mentored students. They're more capable than you give them credit for under the hands of the right guidance. Especially if they were grad level. He's obviously pleased enough with the result to sell it for big bucks and claim that he's equal to or better than native XP. He made it sound like his pilot testers are just as convinced.

Plus there's this notion that he's only been working on this for 2 years. I don't buy that at all. His Phd mind has been churning on this for a long time. 2 yrs is merely the admitted hand-shaking time frame with Aerosoft. Who knows how long he's been chatting with the university.

Edited July 5, 2025Jul 5 by blingthinger

34 minutes ago, blingthinger said:

Incorrect. Very realistic given their compute cluster. Even more plausible with their GPU compute capabilities. Nor do you need to run at every.single.condition to fill up a lift/drag/moment polar. Worst case scenario is that you run a handful of high res simulations at some strategic points in the curves and use those as anchors (scale/shift factors) for the rest. Even if he is doing his own BET instead of a 1D table model, he'd still benefit significantly from using high resolution 2D or even quasi-2D CFD. All with tools he wouldn't necessarily have access to otherwise.

Right, and the week of time on the cluster to run the simulations was afforded under a "little John Doe here writes his Master's Thesis and needs to reinvent the wheel for a commercial flight simulation product" premise?

 

34 minutes ago, blingthinger said:

If he didn't run any CFD, the access to advanced optimization routines could help reverse engineer the Airbus aero for XP.

Something like that is more plausible. If papers and a thesis are to be byproducts of the cooperation, some cutting edge approaches (machine learning, I bet) have to be involved in turning external data into FMs.

 

Quote

But that alone doesn't yield a model that would be behave identically in asobo24. 

Does it have to, though? TolLiss is mission accomplished when they provide the same 6 DOF outputs in both sims and maybe tweak each a little for feel. If sim x does not model something wake turbulence or down- and updrafts as inputs to the FM to provide some dynamics, it's out of ToLiss' control.

 

34 minutes ago, blingthinger said:

Don't downplay mentored students. They're more capable than you give them credit for under the hands of the right guidance. Especially if they were grad level.

I'm inclined to agree if "guidance" means "whip and stick". And you won't hear any praise for academia from me, because I've been there and:

 

Edited July 5, 2025Jul 5 by Bjoern

4 minutes ago, Bjoern said:

reinvent the wheel

Who said they're reinventing? Tweaking a JBsim-like algorithm and using supercomputer-CFD-based forces would absolutely count. But overall to that entire paragraph...uhhhh yeah. Especially with Airbus' name officially in the mix. And even more especially given Airbus now lives just down the road at Bombardier's house, A340 or not.

 

6 minutes ago, Bjoern said:

Does it have to, though? TolLiss is mission accomplished when they provide the same 6 DOF outputs

Have to? He's claiming that it IS. How else are you going to arrive at those 6DOF outputs? You need equations of motion and forces. Panel code and potential flow solvers for the latter only get you so far especially up in the transonic range. Flexing your friendly local supercomputer will get you all the way there, plus applause.

 

8 minutes ago, Bjoern said:

You won't hear any praise for academia from me

Sorry you had a bad experience with that, in whatever form it took. Regardless of your unfortunately-achieved bias, he used McGill somehow. I'm just spitballing on how.

1 hour ago, mSparks said:

afaik, you still get full ground interaction/collisions in xplane with override_planepath set to 1

Fire XP up and tell us what you find there. It would be even more impressive if Toliss was using that dref.

 

1 hour ago, mSparks said:

use the "onground" datarefs to know when to switch.

Yeah that's a blending detail you don't want to deal with for real-time physics. Especially right at touchdown. Respectfully decline.

 

1 hour ago, mSparks said:

If override_planepath is not set to 1, then it is 100% using X-Planes flight model

Hence my description of the .acf file with no bodies/wings. No (or minimal, as in a nearly infinitely-thin cylinder) airframe means zero physics other than the landing gear. 100% * 0 = ...

 

1 hour ago, mSparks said:

That kind of miscommuncation causes problems

Austin's had various flavors of force-replacement capability in the sim since a long time ago. I can't imagine why folks like you would squirm any. And what I'm describing is only applicable to whatever A340 model Toliss has conjured up. Very customized and specific to this airframe. Plus I'm sure we'll find glitches in his Matrix!

10 minutes ago, blingthinger said:

Fire XP up and tell us what you find there. It would be even more impressive if Toliss was using that dref.

It is the bare minimum required for a "custom flight model" that would enable transfer any kind of inflight behaviour over to MSFS, and yes, you still get all the ground stuff:

 

5 hours ago, mSparks said:

I assert people interpret "Toliss developed their own flight model for the A340 in collaboration with McGill University in Canada." to mean the A340 will behave equally under MSFS and XP12. 

There may be things beyond flight dynamics that could be different. But in regard to flight dynamics, Toliss is promising the same experience across platforms. It was tested and runs in Xplane, and now it runs in MSFS, according to them. But the implications go beyond just these two SIM. Its sounds like what they've done could be used on any platform, commercial or civilian. It would be very interesting if they discussed the process in more detail.

 

21 hours ago, Bjoern said:

The statement about the XP.org store is confusing as the A340-600 is already on offer there. Or are we talking a release 2.0 here or an A340-200/300?

I think Aerosoft's comment was about the MSFS version of the plane. It appears Aerosoft would like Toliss to sell it on the xplane org store as well. The org will certainly lose some sales to the built-in xplane store once released. The question in my mind is will the org make an exception and sell non-platform products. Also, not sure if I misunderstood but it sounds like Aerosoft would like to get this on Xbox also if possible. They certainly are trying to recoup every all their money.

34 minutes ago, brinx said:

It appears Aerosoft would like Toliss to sell it on the xplane org store as well

I remembered hearing this as well but couldn't recall where it was said. Was it in that FSelite-stage presentation?

18 minutes ago, blingthinger said:

I remembered hearing this as well but couldn't recall where it was said. Was it in that FSelite-stage presentation?

Yes, it was said at the FSElite-stage presentation between aerosoft/toliss.

 

 

19 hours ago, mSparks said:

bare minimum required for a "custom flight model" that would enable transfer any kind of inflight behaviour over to MSFS, and yes, you still get all the ground stuff:

Just got the notice that you posted. For slewing around, sure, but slewing isn't flight model. Given what I described earlier, I have no reason to agree with "bare minimum".

Your video is interesting. I wonder what is different in the landing gear configuration because I can sink the 172 into the ground with local_y + override_planepath. Maybe it's your use of the elevation slider in the map? The LR website describes setting the Cartesian coordinates directly. You wouldn't be setting AGL in a custom flight model.

37 minutes ago, blingthinger said:

Just got the notice that you posted. For slewing around, sure, but slewing isn't flight model. Given what I described earlier, I have no reason to agree with "bare minimum".

Your video is interesting. I wonder what is different in the landing gear configuration because I can sink the 172 into the ground with local_y + override_planepath. Maybe it's your use of the elevation slider in the map? The LR website describes setting the Cartesian coordinates directly. You wouldn't be setting AGL in a custom flight model.

I had the dref editor opening showing planepath set to 1. when that is set xplane does not move the aircraft (it will just hang in the air - except for ground interaction), the map sliders just change what you need to change with the custom flight model to get the aircraft to move.

Here he is very much saying that is what they have done:

 

Edited July 6, 2025Jul 6 by mSparks

Huh...that's strange. He doesn't say a thing about maps or datarefs in there.

22 minutes ago, blingthinger said:

Huh...that's strange. He doesn't say a thing about maps or datarefs in there.

he doesn't give much detail at all, which is why I said I don't know what to make of it.

It does very much sound like the uni project connected their plane outputs to a "position x y z" model, which suggests sim specific ground handling and toliss specific everything else. that flight model could just be old fsx style lookup tables to fill that gap.

Edited July 6, 2025Jul 6 by mSparks