8 minutes ago, turbomax said:

can we ever trust our approach charts and navigation instruments again with DLSS5?

Only if you hit 1000 FPS ... or just take the blue pill.

Just to think not that long ago all we wanted was a Cessna 172 to fly no matter how bad it looked or sounded, Was fun than, how things have changed, pick pick pick, all the fun seems to have disappeared with new tech. Always finding something new to complain about. 

2 hours ago, SayAgain said:

going to take you down a path of thread lock or deletion here on AVSIM

Me? No. Only if someone else wants it to. We've had plenty of off-thread-topic discussions in the past that didn't get locked because they remained science/engineering-focused. CFD-based weather prediction, NN discussions, etc.

Climate, weather, and cgi are each rougly equidistant as far as algorithm differences go. Under the covers they're all neural networks. Huge matrices of weights waiting for a GPU to multiply them together. "AI" is a silly wallstreet buzzword.

The initial convolutions will be different, but ultimately they're the same concept: aggregate massive quantities of data and spit out the most likely outcome. In the case of weather (climate didn't even need NN models...well they kind of did an aggregate prediction with lots of approaches, so still a statistical conclusion) you have tons and tons of historical sensor data and CFD predictions that are all in a perfect archived state for data mining and subsequent model training. Much like the process I described earlier with automating camera view and object motion to train a RT NN. There's still lots of non-linear behavior though, so it won't be as precise as dlss5 is right now (RT is just a bunch of straight lines and reflections). But compared with what we had before, yeah, it's a significant improvement.

Jcomm and feline are career weather experts, hence my ask of him.

12 hours ago, blingthinger said:

Do you folks use Deepmind at all? Or is it being looked at?

https://www.noaa.gov/news-release/noaa-deploys-new-generation-of-ai-driven-global-weather-models

Not yet, but we at the "Portuguese MetOffice" (the BEST in the Whole World, btw...) are deeply investing in AI-added forecast models. ECMWF, our main model source even if we also use ALADIN, actually has it available and you can get it from here:

ECMWF’s AI forecasts become operational | ECMWF

 

Edited March 21Mar 21 by jcomm

21 hours ago, blingthinger said:

"AI" is a silly wallstreet buzzword.

 

Absolutely, it’s what I would NOT consider real AI nor anyone really in the field would consider it real AI.  An abused term similar to the use of “Cloud”.

As far as real application to MSFS specifically, I don’t see how DLSS5 could transform a PG tree blob into actual trees … the source data that drives AI is just too far outside of parameters and no amount of training could resolve it to a tree or group of trees (which is what they really are) … we might get a better looking blob, but it will still be a blob.

As far as MSFS wall bridges with PG, that needs some serious geometry change which according to nVidia engineer isn’t what DLSS5 will do.

About the only benefit I can see is that DLSS5 might improve on AG objects and maybe some airport objects if it can interpret transparency correctly.

As has been suggested, so long as we can turn it off … my concern is the development resources being pooped into this process that could be used in what might be more beneficial to flight simulation (aka weather, flight physic, ground physics, collision, etc.)  Losing sight of other more important objectives due to the AI hype.

Edited March 21Mar 21 by SayAgain

5 hours ago, SayAgain said:

I don’t see how DLSS5 could transform a PG tree blob into actual trees

Just need train it with enough data and any blob can be whatever your heart desires. 2d. 3d. red trees. blue trees. It doesn't matter. Replace them with perfectly manicured shrub art of cats. I'm really curious what details the devs have to provide to train the NN. It rings a lot like dlss 1.0-style requirements.

5 hours ago, SayAgain said:

interpret transparency correctly

There were a couple scenes with windows/glass. Hogwarts and Starfield. Not a lot of attention paid there though. I agree that one will be interesting to watch out for. Though again...it's all in the training set. The reflections of off-screen geometry were in place. I struggle to see transparency being a problem. Just give it enough data and you'll have Pitt and Cruise dancing on a rooftop.

5 hours ago, SayAgain said:

aka weather, flight physic, ground physics, collision, etc.

Meteoblue tried. I'm sure it would be more successful now with current methods.

Generalized flight model is where I draw the line for a consumer sim using NN simply because of the randomness and non-linearity (worse than short term wx forecast) and amount of quality data needed to build it. I'll eat my hat if it ever shows up in XP. XP's already really computationally efficient. I struggle to see a NN buying in especially when the GPU is already blowing out smoke on the graphics calcs.

Ground physics? That's just a glorified mass-damper model. No need for NN there. Eh... maybe tire friction/flex models would benefit with a small NN but again...need lots of either high quality numerical simulation results or experimental data. Not easy to come by.

10 hours ago, blingthinger said:

 

Ground physics? That's just a glorified mass-damper model. No need for NN there. Eh... maybe tire friction/flex models would benefit with a small NN but again...need lots of either high quality numerical simulation results or experimental data. Not easy to come by.

Depends on how many NACA reports and the like it assimilated ...

Seeing and reading what I read these days about NN, I honestly believe a good, top, FM could be expected... 

Maybe we can have it sooner than later when FS 2028 or the like get's announced 😁

Edited March 22Mar 22 by jcomm

I don't see how anything is missing from the flight model possibilities, it is up to the 3rd party airplane developer to undertake the huge task of defining and fine tuning scalars, parameters, look up tables etc. WorkingTitle have shown that a flightmodel (for example their CJ4) within 2% of book value is indeed possible with the current SDK.

https://docs.flightsimulator.com/html/Content_Configuration/SimObjects/Aircraft_SimO/flight_model_cfg.htm

....

ooops, wrong forum, sorry  😀

Edited March 22Mar 22 by turbomax

Oh wow I'm sure that's really big for you to be able say... it uhhh, it better be able to do that. FSX/P3d could do that and it's at the core of asobo's model.

What's amusing is to hear Toliss describe interacting with the fs2020/4 dev ecosystem. He took one look at the FM and said "uhhh nope" and wrote his own. And that's just a drop in that bucket. All of his dev work was done in XP because it's so flexible and intuitive. The only part of the XP model he was overriding was the turbine engine. He trusted the default aerodynamics enough to just let it be as is. Not so with asobo's concoction.

But this conversation isn't about what's potentially "missing". Nobody declared that it's missing anything. It's if one could create a generalized NN model to be able to simulate any airframe in any flight condition with any level of accuracy.

54 minutes ago, blingthinger said:

Toliss describe interacting with the fs2020/4 dev ecosystem. He took one look at the FM and said "uhhh nope"

that's just him and his private opinion.

Working Title’s view:

MSFS flight model: fundamentally solid and capable of realism
Default aircraft: often poorly tuned or simplified
Realism comes from: accurate data, systems, and careful implementation

They’ve consistently shown (rather than just said) that you can get very realistic results out of MSFS without replacing its physics—just by using it correctly.

Flight model realism alone doesn’t define realism Systems depth and correct data matter just as much

And that's their private opinion. Like I said...it better get there in the aspects they're bragging about. That it's so much more difficult to get there than it used to be, is telling.

He's used both models now and out of the 2 groups, he's the engineering PhD. I'll go with that "private opinion" over a few web devs any day.

What's funny is you trying to bring "systems depth" into a neural network discussion. Such fight-picker, much surprise.

Edited March 22Mar 22 by blingthinger

15 hours ago, blingthinger said:

I'm really curious what details the devs have to provide to train the NN.

Sorta my point, there ain’t no additional data from imagery so it will have to come from estimated Biome data which brings us full circle to AG and more “probability” … not “as real as it gets” to “as AI as it gets”.  Anyway, let’s see where the latest nVidia hype takes folks, maybe I’ll be surprised and it’s not just another sales gimmick  … but to be clear, I don’t want to see “anything I want”, I want to see reality of how it actually looks on Earth.

Honestly I do not understand why we need AI in every thing, its in just about every new app, new processes all doing the same thing stealing CPU cycles I do not wish to give them. When is enough enough

18 minutes ago, mjrhealth said:

CPU cycles I do not wish to give them.

just turn it off then. besides, this can run on the GPU instead of CPU

1 hour ago, turbomax said:

just turn it off then. besides, this can run on the GPU instead of CPU

And you know its turned of because... Ahhh that honesty system that got forgotten a long time ago by big tech.