MSFS flight model

[robert young 2,897]

3 hours ago, Scottoest said:

More surface-points being processed in that approximation of reality is certainly always better, but ultimately a good flight model is as much a work of art as it is a work of math.

Yes indeed. Well, if not "art" then at least a facility for observation. Honestly, to make a decent flight model does not require such computing power as that being discussed. Every addition of complexity often results in losing the picture of very basic elements. For example, you do not need multiple "lift points" on the wings. You only need, ideally, an inner and outer wing so you can simulate a lift, drag and stall parameters for inner and out wing parts.

And You really don't need multiple fuselage points in order to simulate other drag forces.

The more complex the algorithms the more scope there is for mistakes or bad tuning. Keeping things simple affords much more control provided there are sufficient numbers of basic parameters. The main goal is to simulate what happens when one or more controls are applied. That is the core of any decent flight simulator. Then add the aircraft's reaction to moving air masses and you are 95% home.

It is not that I am against CFD at all. It is just that you will always need a human to fine tune the basic characteristics so the reaction of an aircraft to control inputs is believable.

[blingthinger 649]

8 minutes ago, robert young said:

It is just that you will always need a human to fine tune the basic characteristics so the reaction of an aircraft to control inputs is believable.

The question then becomes: how is the tuning done? What knowledge is needed to do the tuning? Is it repeated guess-and-check? Or are the adjustments based on some known, expected behavior?

At some point, the guess and check hopefully generates a pattern that might be used to narrow down on the desired product.

Starting with a known, expected behavior will reduce the amount of time spent on subsequent guess and check to fine tune.

If a pattern doesn't appear, the human is in for a world of frustration!

[Alpine Scenery 1,609]

1 hour ago, blingthinger said:

The question then becomes: how is the tuning done? What knowledge is needed to do the tuning? Is it repeated guess-and-check? Or are the adjustments based on some known, expected behavior?

 

That's what I meant earlier when noting "too much theory, they just need to test and rework".

The problem is without real motion and movement that you feel in a real plane, pilots will have a different perception of how a simulator even feels to them. 10 pilots might say 5 different things. I would probably start by just adjusting the model until has the lowest delta error possible given the CPU constraints, but still retains enough turbulence modeling to look somewhat realistic.

We would need so much more CPU, because they don't even have 50% of the CPU available in the game for stuff as the other stuff takes so much of the CPU already. They are probably working with 10% of the CPU for the flight model, and then they have to make it work on peoples slower machines as well (man that's hard).
 

Edited September 13, 2022 by Alpine Scenery

[blingthinger 649]

43 minutes ago, Alpine Scenery said:

too much theory, they just need to test and rework

: ) I get where you're coming from here...but as @robert young pointed out, Asobo got rid of a lot of legacy theory capabilities that would still be very useful now. This meant for example that PMDG needed years to recover, even after starting from well-acclaimed P3D models. Theory is what generates the patterns.

There should probably be a kickstarter to get all 3rd party devs one of these in their living rooms:

 

 

[Krakin 5,743]

1 hour ago, blingthinger said:

PMDG needed years to recover

Did PMDG ever say their hiccups were related to the FM? All I remember them complaining about was stuff related to avionics.

[tweekz 2,165]

This video perfectly demonstrates, what I am missing in MSFS in terms of landing physics. This is how it should look like when you land a loaded plane. In MSFS you still manage to float it by just pulling a little on the yoke.

 

[Alpine Scenery 1,609]

The ground effect does seem silly high in many of the planes in MSFS.

Edited September 13, 2022 by Alpine Scenery

[tweekz 2,165]

40 minutes ago, Alpine Scenery said:

The ground effect does seem silly high in many of the planes in MSFS.

Yeah, lift generally seems to be a bit too pronounced. When coming in for a landing it feels like rowing down a boat.

[jcomm 5,890]

27 minutes ago, tweekz said:

Yeah, lift generally seems to be a bit too pronounced. When coming in for a landing it feels like rowing down a boat.

True, and I think the best so far in this respect are the Fenix A320 and the Black Square Caravan...

[jrw4 624]

I have tried to follow the erudite technical discussion in this thread, but I haven't found the spot where someone asks what the computation costs might be that are associated with evermore detailed real-time computations of aerodynamic behavior. I think we would all agree that Asobo has done a commendable job in providing good performance on fairly modestly equipped computer hardware. That 3PDs have taken good advantage of this behavior is well illustrated by PMDG's implementation of their 737 family entirely within the envelope provided by MSFS, with no need for external processes. Kudos all around for that.

How will further elaboration of the physics models underlying the simulation engine provided by Asobo impact hardware requirements, particularly with regard to CPU frequency, memory, and processor count?

Thanks.

[Alpine Scenery 1,609]

I think it's better in some ways than FSX/P3D and Xplane modeling, and worse in others. I guess none hit the mark perfectly. If I had to pick one, I would still rather have the MSFS out of the 3, it feels more "new age", whereas the others still feel like flying on rails most of the time.

I feel if they toggle down the ground effect and fix the rudder issues in default planes on the ground, it will be the best flight model, by far. They are so close to a great flight model, yet so far away 🙂

 

[ha5mvo 240]

1 hour ago, Alpine Scenery said:

 I would still rather have the MSFS out of the 3, it feels more "new age", whereas the others still feel like flying on rails most of the time.

Not sure what you mean by "feel". Not exactly a tangible parameter. The "feel" using a Brunner yoke is worlds apart from what you'll get using a flimsy T16000 for the exact same plane.

I'm still to see an analysis that demonstrates that fidelity of the MSFS engine. I do see though a lot of "feelings" albeit much less numbers to support them....

I don't think there could be a universal formula which will give you perfect (or even good) results if you only feed it with a given planes data. A good FM involves a lot of voodoo and black magic mastered by the few FM gurus like  @robert young 

I have seen invisible surfaces and drag elements introduced into models to make them fly by "the numbers". I bet Rob could elaborate on that.

At the end of the day, the proof of the pudding is in the eating. While I have not tested ALL available models, those I did were quite far off the mark - leading addons among them. 

My only agenda is improving that aspect of MSFS, contrary of what the cheerleading squad might suggest. I don't see how squelching any form of criticism might be of benefit to either the sim or its users in that regard.  

Edited September 13, 2022 by ha5mvo

[Alpine Scenery 1,609]

It depends on the plane, the Carenado PA-44 actually lands the best as far as not giving too much ground effect over the runway. I just landed it with full flaps at a -200 fpm descent without it jetting upwards on the flare. I landed perfectly pretty much. I would rate the Milviz C310 somewhere in the top 5 as well, but it has a bit more float. It's hard to say how much is real since I've never flown these specific planes before. 

The majority of other planes are hard to get on the ground if you have flaps on, even at 15 or a minimal flap setting, other than the bigger jets. Some planes are so bad, I land with 0 flaps, engine cut, and full spoilers.

The flare in MSFS has to be perfectly timed way more than RL so I hear, it's harder to land is what it amounts to.

Edited September 13, 2022 by Alpine Scenery

[TrafficPilot 121]

I'm assuming that everyone moaning about floating on landing are approaching their destination at the correct airspeed? I know I complain incessantly about the MSFS flight dynamics but I haven't noticed any particular floating issues - at least in the GA aircraft I fly (Turbo Bonanza, 172).

Landing too fast is one of the most common errors with students and low hours PPL holders - probably because we've drummed into their brains how dangerous (and easy) it is to stall an aircraft in the circuit.

I learnt to fly at Denham EGLD which has a relatively short runway so any sort of "floating" on finals was "discouraged". In the past we've had a couple of flyers overshoot and landing on the golf course!

I'm learning a lot of technical computer stuff from this discussion so thank you!

Edited September 13, 2022 by TrafficPilot

[Alpine Scenery 1,609]

7 minutes ago, TrafficPilot said:

I'm assuming that everyone moaning about floating on landing are approaching their destination at the correct airspeed? I know I complain incessantly about the MSFS flight dynamics but I haven't noticed any particular floating issues - at least in the GA aircraft I fly (Turbo Bonanza, 172).

It totally varies from plane to plane, some want to jump right back into the sky even when they are at stall speed. It is harder to hold a Joystick forward on landing than a Yoke, so control issues could be part of it. That said, I think it's the planes mostly.

Yah, the flare issues with flaps happen even when I'm at or near stall speed in some planes. I actually usually land too slow, only land faster than I should with big jets just for fun. What happens is you flare, then the plane just skyrockets upward, stalls, then slams onto the ground. That is one example of what can happen, some of them bounce way too much or uncontrollably fly off the runway due to rudder being hard to control. 

 

Edited September 13, 2022 by Alpine Scenery