11 hours ago, robert young said:

Hi Matt,

Nice to hear from you (enjoying the CJ4 immensely). If you PM'd me I can take you through it.

Why all the secrecy?

11 hours ago, Donstim said:

Why all the secrecy?

Until I know which issue needs solving on which aircraft, what size it is, what its wingspan is and other info, there isn't a single tweak solution but a combination - so if you don't mind if anyone needs help or input I'd rather wait and deal with the person one to one.

What is "CFD", and where did this term come from?

2 hours ago, Mace said:

What is "CFD"

"CFD, or Computational Fluid Dynamics, is a type of computer modeling researchers use to show where air molecules are pushed out of the way by a moving aircraft and generate noise. These models can help engineers pinpoint the best places to alter an aircraft's design in order to reduce noise."

https://en.wikipedia.org/wiki/Computational_fluid_dynamics

since CFD runs hundreds times faster on a GPU and Asobo uses GPU programming for CFD this is yet another reason why it is a good idea to use a powerful GPU for MSFS, it makes much better use of GPU programming than the other simulators. hence MSFS 99% GPU load versus only ca. 33% in x-plane/P3D.

how CFD can run 100 x faster on a GPU vs. CPU, 10.000 CUDA cores running in parallel vs. a CPU with only 8 cores:

 

Sebastian Wloch from Asobo experiments with CFD on GPU with Shadertoy:

 

Edited February 24, 20224 yr by turbomax

Interesting video to watch, if you find the patience & time:

https://www.youtube.com/watch?v=GKMBoGwLG1g

based on SimScale:

https://www.simscale.com/docs/simwiki/cfd-computational-fluid-dynamics/what-is-cfd-computational-fluid-dynamics/

Another pair of interesting videos using OpenFoam, a free / opensource CFD workbench:

https://www.youtube.com/watch?v=CDzaksIF3sI

https://www.youtube.com/watch?v=9Ex9yf93w8Y

and here a good reference for oother free tools:

http://aircraft-design-software.experimentals.de/cfd/computational-fluid-dynamics.html

Problem is, it takes a very powerful machine and a good deal of time to calculate "just" the flows over a finite wing, so, not feasible until we get, maybe, quantum desktops, to enable the beauty of abrams dream 🙂

 

Edited February 24, 20224 yr by jcomm

jcomm ya until desktop computing becomes powerful enough to do true CFD calculations (near future? :), I'm guessing it's all about efficient approximation algorithms to produce feasible or "good enough" results/calculations that translate to useful impacts on the aerodynamic simulation. Really hoping Asobo provides more details by the SU9 timeframe as to how exactly they are implementing and taking advantage of whatever CFD techniques they are employing, and the interplay with good ole parametrized flight modelling.

1 hour ago, jcomm said:

Problem is, it takes a very powerful machine and a good deal of time to calculate "just" the flows over a finite wing, so, not feasible until we get, maybe, quantum desktops, to enable the beauty of abrams dream 🙂

..And this is almost precisely my point. It's all interesting stuff but it tells me that people are running before they have learned to walk, or even crawl. There is no point in studying this branch of flight modelling except for personal pleasure. The CHASM between computational sophistication and simply getting a sim aircraft to roll with a degree of elegance and believability is enormous. You really do not need quantum power or nano technology or even "advanced" calculations. You just need to FLY THE AIRCRAFT with critical pilot faculties switched on and OBSERVE what the sim aircraft is doing, then by any available means correct the behaviour.

Some of the current very poor responses in standard FS2020 aircraft, and more importantly in ENCRYPTED aircraft which are extremely difficult to access for even a minor user-tweak, suggest that too much time has been spent ambitiously investigating advanced aerodynamics at the expense of very basic behaviour, like not bobbing up and down like a yoyo at the slightest stick input, or like displaying hopelessly wallowing, over exaggerated inertia (standard B747). All these examples indicate that whoever is in charge of this stuff simply has no understanding about how aircraft fly. Sorry to say this, but I believe that to be true. I believed it from day one of release but got an avalanche of criticism for saying so. More recently there has been a distinct change of mood on this and I see that I am far from alone.

1 minute ago, robert young said:

Some of the current very poor responses in standard FS2020 aircraft, and more importantly in ENCRYPTED aircraft which are extremely difficult to access for even a minor user-tweak, suggest that too much time has been spent ambitiously investigating advanced aerodynamics at the expense of very basic behaviour, like not bobbing up and down like a yoyo at the slightest stick input....

Robert would it be fair to say this just speaks more to Asobo not spending enough resources in fine tuning and providing properly parametrized models for their default aircraft, but that doesn't mean the core sim's aeorodynamics/modelling capabilities are lacking per se?, as seen with the FBW A320, Kodiak, PMDG DC6, etc and also your mod.

3 hours ago, lwt1971 said:

Robert would it be fair to say this just speaks more to Asobo not spending enough resources in fine tuning and providing properly parametrized models for their default aircraft, but that doesn't mean the core sim's aeorodynamics/modelling capabilities are lacking per se?, as seen with the FBW A320, Kodiak, PMDG DC6, etc and also your mod.

Yes, I think that is right. Although quite a few ex-FSX params have been stripped away it is possible to get a decent flight model as with the examples you give. But even the humble Turbo G36 took many, many months of work because half the time was spent trying to get to grips with what Asobo had included and what they had stripped out or simplified. I agree that the core FDE engine is ok, BUT it is not any better than FSX in its current state, and as I've pointed out, many or most params look to be carbon copies of the table based FSX routines but now all expressed as text/number arrays. Actually the graph based access through a couple of well known FDE applications was more useful than a text file, which used to be confined to a config file separated from the flight model file.

Edited February 24, 20224 yr by robert young

I think Robert has hit many nails on the head in the course of this thread, but the latest post sums up a lot of what I find -

1) many, many months of work, I estimate three times as long based on the aircraft I have created so far. Partly down to the confusion and incompleteness of the SDK, partly down to . . .

2) . . . the removal of a large proportion of the tools available to us in FSX. It would have been much easier and clearer had they maintained the FSX parameters in their entirity to allow us to adjust for any weaknesses in the core FM.

There seemed to be a move towards 'one size fits all' with the core flight model having overall control of how an aircraft flies and the removal of the ability to carefully adjust from the developer. Once we move away from straightforward shapes, however, weaknesses have become apparent and rather than having a surgeon's scalpel to gently nudge things in the right direction without affecting other areas of the flight model, we now have to use a sledgehammer.

 

14 hours ago, robert young said:

over exaggerated inertia (standard B747)

In FSX times I once had a session in a MD-11 Level-D simulator, and the single thing that struck me as the most different was the inertia of the real heavy aircraft. FSX was on rails. The real MD-11 would not stop rolling just by centering the yoke. The difference was glaring. And the MSFS default 747 is the first aircraft in a sim, I ever experienced, that gave me the feeling I had in the Level-D sim. Because of the inertia. So, are you sure, you know the real 747 flight handling characteristics good enough to judge?

Second point, without having tested it, I assume by two or three tweaks in the cfg file you can give it whatever inertia you want it to have. So, even if the inertia is wrong, the issue is not the flight model but the configuration of the flight model. After all a cfg file is called cfg because cfg means configuration.

9 hours ago, mrueedi said:

In FSX times I once had a session in a MD-11 Level-D simulator, and the single thing that struck me as the most different was the inertia of the real heavy aircraft. FSX was on rails. The real MD-11 would not stop rolling just by centering the yoke. The difference was glaring. And the MSFS default 747 is the first aircraft in a sim, I ever experienced, that gave me the feeling I had in the Level-D sim. Because of the inertia. So, are you sure, you know the real 747 flight handling characteristics good enough to judge?

Second point, without having tested it, I assume by two or three tweaks in the cfg file you can give it whatever inertia you want it to have. So, even if the inertia is wrong, the issue is not the flight model but the configuration of the flight model. After all a cfg file is called cfg because cfg means configuration.

Yes the 747 has a lot of inertia compared with smaller aircraft. And the controls are sluggish compared with smaller aircraft. But not THAT sluggish. And yes you are right, roll doesn't stop suddenly by centering the yoke, especially after a large aileron input, but these things are NAUNCED, not black and white. There is a balance to be struck. Talking more of roll - it is not just inertia, it is also stability factor, and then authority. To get a harmonised overall control feel for a given size of aircraft, it is the combination of all three (and some other params) that create the balance. FSX is not on rails. I think you mean some individual flight models in some aircraft made for FSX feel like they are on rails. As you have just pointed out it is how you tune the individual flight model that counts, not the system itself. If you like the MSFS 747 it will be because you like the way it was tuned. Personally I find it over the top in slow reaction time in every axis.

20 minutes ago, robert young said:

Yes the 747 has a lot of inertia compared with smaller aircraft. And the controls are sluggish compared with smaller aircraft. But not THAT sluggish. And yes you are right, roll doesn't stop suddenly by centering the yoke, especially after a large aileron input, but these things are NAUNCED, not black and white. There is a balance to be struck. Talking more of roll - it is not just inertia, it is also stability factor, and then authority. To get a harmonised overall control feel for a given size of aircraft, it is the combination of all three (and some other params) that create the balance. FSX is not on rails. I think you mean some individual flight models in some aircraft made for FSX feel like they are on rails. As you have just pointed out it is how you tune the individual flight model that counts, not the system itself. If you like the MSFS 747 it will be because you like the way it was tuned. Personally I find it over the top in slow reaction time in every axis.

Like I said before, if there are specific things you would like to tune but you cannot tune at the moment, then compose a question for the next Twitch Q&A, in the Twitch Q&A forums.  Let us know the question you have composed, and many of use will help upvote your question so that it's put in front of Jorg/Seb/Martial at the Twitch Q&A.  This is the easiest way to get your question addressed, and for Jorg/Seb/Martial to take notice of your question.  And it won't even take that many upvotes, probably 100 or so upvotes will get your question asked at the next Twitch Q&A.

The FBW team was also not happy with the weather API for WASM in MSFS.  So the FBW team, who even had a good relationship with Microsoft/Asobo (probably a good relationship with Jorg) at the time, galvanized the community to upvote a Wishlist item to implement a weather and terrain API. It got enough upvotes, Jorg/Seb/Martial got the message, and now you can see that more of the weather API is being exposed on the roadmap.  Granted, this Wishlist item got over 2100 upvotes, so getting a Wishlist item to the attention of Jorg/Seb/Martial is harder. That's why I suggest that you write a question for the next Twitch Q&A and let us here at Avsim know you have written the question, and we can upvote it for you - it will probably take 100 upvotes for so to make the top 5 upvoted questions, and Jane will ask Jorg/Seb/Martial your question at the next Twitch Q&A .  The next Twitch Q&A will probably be in a few weeks, and the forum to post questions for the Twitch Q&A usually opens up 2 weeks before the Twitch Q&A happens.

 

Edited February 25, 20224 yr by abrams_tank

1 hour ago, abrams_tank said:

Like I said before, if there are specific things you would like to tune but you cannot tune at the moment, then compose a question for the next Twitch Q&A, in the Twitch Q&A forums.  Let us know the question you have composed, and many of use will help upvote your question so that it's put in front of Jorg/Seb/Martial at the Twitch Q&A.  This is the easiest way to get your question addressed, and for Jorg/Seb/Martial to take notice of your question.  And it won't even take that many upvotes, probably 100 or so upvotes will get your question asked at the next Twitch Q&A.

The FBW team was also not happy with the weather API for WASM in MSFS.  So the FBW team, who even had a good relationship with Microsoft/Asobo (probably a good relationship with Jorg) at the time, galvanized the community to upvote a Wishlist item to implement a weather and terrain API. It got enough upvotes, Jorg/Seb/Martial got the message, and now you can see that more of the weather API is being exposed on the roadmap.  Granted, this Wishlist item got over 2100 upvotes, so getting a Wishlist item to the attention of Jorg/Seb/Martial is harder. That's why I suggest that you write a question for the next Twitch Q&A and let us here at Avsim know you have written the question, and we can upvote it for you - it will probably take 100 upvotes for so to make the top 5 upvoted questions, and Jane will ask Jorg/Seb/Martial your question at the next Twitch Q&A .  The next Twitch Q&A will probably be in a few weeks, and the forum to post questions for the Twitch Q&A usually opens up 2 weeks before the Twitch Q&A happens.

 

As much as I appreciate this, I really feel that if basic aerodynamic issues need to be submitted as a wish list on an already crowded list of to-do items on a forum that has thousands of competing requests, in order for it to be "listened to" I'm a bit concerned about the process.

Aerodynamics and individual stock aircraft flight models should be, and should have been, right from the inception of the sim, the principle priority. After all, how can you call yourself a "simulator" if it doesn't simulate anything with reasonable accuracy. But it is not only that.... the fact is that these very basic issues have been largely unaddressed, along with a host of other very fundamental subjects, in favour of endless updates relating to scenery (themselves not altogether warmly received). One example equivalent is to design and manufacture a splendid looking grand piano and to constantly "update" it with beautiful inlay, polish, a better lid, and countless other things, but continually ignore the fact that it is badly out of tune.

Another example is that I designed a car that doesn't have a functioning gear box, then suggesting that if people want a proper gear box, they should go onto a forum and request votes for something that should have been present in the first place!

6 minutes ago, robert young said:

As much as I appreciate this, I really feel that if basic aerodynamic issues need to be submitted as a wish list on an already crowded list of to-do items on a forum that has thousands of competing requests, in order for it to be "listened to" I'm a bit concerned about the process.

Aerodynamics and individual stock aircraft flight models should be, and should have been, right from the inception of the sim, the principle priority. After all, how can you call yourself a "simulator" if it doesn't simulate anything with reasonable accuracy. But it is not only that.... the fact is that these very basic issues have been largely unaddressed, along with a host of other very fundamental subjects, in favour of endless updates relating to scenery (themselves not altogether warmly received). One example equivalent is to design and manufacture a splendid looking grand piano and to constantly "update" it with beautiful inlay, polish, a better lid, and countless other things, but continually ignore the fact that it is badly out of tune.

Another example is that I designed a car that doesn't have a functioning gear box, then suggesting that if people want a proper gear box, they should go onto a forum and request votes for something that should have been present in the first place!

<sigh>.  We have the ability to do something about this, but you are just dragging your feet.  No offense, but while I originally thought that you had valid points, I am beginning to doubt it.  You come across as someone who complains but when offered the opportunity to fix the problem, you then run away.

Not everything is going to be perfect in the first 1.5 years of MSFS.  But we can prod/nudge Microsoft/Asobot to fix stuff, step by step.  And they have been fixing stuff step by step over the last 1.5 years.  What disappoints me in your comment is that you can also take action to get them to fix the stuff you want, and now there are other people that will even help you get your complaints in front of Jorg/Seb/Martial, who can respond to your concerns.  But you don't even want to do that.  SMH.

 

Edited February 25, 20224 yr by abrams_tank