A "generation after next" flight sim

[Hook]

MSFS looks to be the next generation of flight sim. The world "incredible" doesn't even start to cover it. It is a whole new sim that looks very good indeed.

But... I have a dream. The next generation of flight sim after MSFS.

Imagine Lockheed Martin partnering with Google and possibly the Active Sky people to make P3Dv6, with similar streaming scenery technology as is used in MSFS, but using Google assets, with Active Sky providing accurate real world weather. I would hope for the ability to import all your current P3D aircraft, possibly with new installers.

With properly realistic world wide ground scenery, Prepar3D could become the training simulation for ground operations without necessarily having to create custom scenery for each scenario. Air operations would be flying over correct terrain features everywhere in the world. Without the necessity to start from scratch as MSFS is doing, existing scenery assets could be imported from earlier versions of P3D, possibly with some updates to improve them to modern standards. People training for private pilot could fly over accurate local landmarks.

Would Google be willing to invest in such a venture? Could they outdo Bing? Would they want to?

Some minor changes in the default flight model coupled with an improved atmospheric model from Active Sky could produce the best flight experience available. It is already very good, especially with A2A aircraft which inject their own processing into the flight model. It could be tailored, not for the beginning simmer crowd, but for the more serious simmers like A2A customers.

I don't know if all this is possible or how long it would take, but I'm hoping to see it eventually.

Would you buy such a sim?

Hook

[AKar]

Now that you mentioned, I'm already thinking about something that would perhaps sound like a return to the old style.

In fact, what I'd be wanting is better landscape generating engine using procedural methods, to the visual quality similar to "some other games". Non-mandatory satellite and/or aerial imagery aside, there would be a landclass-style layer as well, that would include both static terrain parameters, such as azimuth and zenith angles, surface type and roughness, and time variable terrain parameters, such as coverage, albedo and so on. The latter ones would change with time along the seasons, but would also integrate the weather data, both real-time and historical from the past months, such as rain and snow accumulation.

There are two reasons for this.

First, visually some procedural terrain engines produce very believable results that require no such huge data volumes as streaming imagery do and, more importantly, are able to do that where no such imagery exists in good quality. Also, this would allow for true seasons, not only four plus one of them but proper transitions as well, if the engine is built with care. There are corners of the Earth where these really matter, here above 60°N included.

Second, this has to do with the weather. The weather should also be procedurally generated to larger extent, not 'forcibly' METAR-matched as it is with current weather engines - even to wrongly match the designed limitations of METARs to the weather synthesis! Now, 'micro weather' is heavily influenced by terrain. One could say it is created by the terrain. Parameter-driven procedural terrain engine allows for direct interaction with the weather engine. This would allow for Condor-like micro weather synthesis, suitable for proper gliding by reading the terrain for instance.

Obviously, the terrain depiction should be an open format to allow for data translation from various sources, such as land cover and terrain maps, or numerical measurements, and for manual adjustments where the available source data does not produce adequate results.

-Esa

[Hook]

Good info.

I fly with real world METARs so I want to see that in the sim. Active Sky already does a good job of interpolation and handling out of date METARS. For soaring, CumulusX worked in conjunction with Active Sky to produce proper and realistic thermals and even produced its own cumulus clouds to mark thermals. The method MSFS uses for wind over hills causing ridge lift seems to work. I haven't encountered any rotors yet but I haven't looked for them either. Turbulence is appropriate depending on terrain, for example flying in valleys or on final approach to some airports.

If we're going to have procedural land classes, then I'm happy with current P3D except for desert terrain. The whole idea of partnering with Google is to get photo real ground with appropriate buildings and vegetation. I'm really enjoying comparing what I see in MSFS with Google Earth... everywhere I fly.

I have no problems with P3Dv4 lighting. I'm assuming there are ways to produce proper seasons based on the same techniques used in procedural landscapes.

I have no interest in flying over "fantasy" terrain in "fantasy" conditions though.

Hook

[AKar]

I fly with real world METARs so I want to see that in the sim. Active Sky already does a good job of interpolation and handling out of date METARS.

The issue here is that the METARs do fairly poor job in presenting the weather as a whole. For instance, let us say that the METAR is CAVOK with no cloud observations. With AS, this results in clear blue skies and shining sun. In reality, there could be (and often is) a mid-level overcast, and the day really is a dark gray one with almost zero convective activity. Also, auto-generated weather observations are often notoriously inaccurate in reality.

As another example, in many places the sparse grid of weather stations is directly causing implausible situations. Just to throw in something as an example, say there is a weather station at the airport on the island's coast. As the weather is 'locked' to the METAR from that station, I've often ran into apparent sea breezes running into wrong direction on the opposite site of the island.

It is important to note that these issues I've picked as examples are fundamental limitations to the method of creating the weather the way it is done, not 'faults' that are fixed with some tuning. That's why I'm wishing that dynamic methods to create weather synthesis with a realistic 'big picture' running the show would be developed.

So far the only weather engine I'm finding 'adequate' for proper gliding is the one in Condor2 (I haven't tried the Silent Wings enough to form an opinion). Obviously it does not do 'live real weather', and is far from perfect, but it is good enough that the criticism and discussion can often be on the level of very specific issues and phenomena that not many but glider pilots are even aware in aviation context.

In regards of turbulence, I've recently started trying out the RealTurb addons. They do something to analyze the terrain against the prevailing weather synthesis (provided preferably by AS), and inject some turbulence and lifts and sinks accordingly. I've not tried it for long enough to have a 'final opinion', but so far I'm seeing it as a definite improvement over the 'default' AS turbulence model. It is not, nor does it pretend to be, a "soaring grade" simulation of fine details of thermals or such, but any improvement is an improvement, and I'm finding take-offs and approaches in faster airplanes particularly improved.

As a closing remark, I often speak in terms of gliding, because that's how I learned to fly and as a result I try to read the sky with the eyes of a glider pilot regardless of what I'm flying or flying in (unfortunately primarily the latter nowadays). With these remarks, I don't intent to claim that "soaring grade" fine details of weather are needed for simulated airliner flying, or even for light GA flying.

-Esa

[Hook]

I spent a lot of time in FSX soaring with CumulusX, so I know where you're coming from. I had to do a lot of real world research to learn how to do thermals properly, and everything I read worked in the sim. I was using the DiscusX sailplane from Aerosoft, I think. I'd like to be able to have the same experience again.

Active Sky already does a good job with weather, often injecting what it thinks should be there rather than simple METAR data and does considerable interpolation where necessary. Inaccuracies shouldn't be much different from those encountered by real pilots.

Prepar3D has never been better than it needed to be, but MSFS has raised the bar considerably and has provided a proof of concept. MSFS is being made for a mass market, P3D for a more serious one. Lockheed Martin couldn't match MSFS by themselves, but in combination with Google I think they can exceed it.

As for AI processing, the people with the capability to write the Google search engine front end have capabilities that exceed anyone Microsoft could hire.

Hook

[AKar]

I did test CumulusX several years back, albeit not on my own setup so I could not dissect it fully. In context of FSX, at least back then, it was a definite improvement over what else was available. However, I did not find it to be as good as purpose-built gliding simulations (nor would it be realistic to expect it to). As far as I remember, it felt too easy and too predictable; whether that was an issue with the terrain depiction of the host sim, I don't know. Thermals were definitely better than today's stupid steady lift/sink areas of the AS, but still not quite there yet. In fact, I'd even argue that I don't find the current thermal simulation in Condor 2 to be nearly finished either; effects such as "bubbling" of the thermals not being nearly as evident as they often are in real life in such conditions. Hence, they are too easy to center "perfectly".

Personally, I'm less impressed of the accuracy of the weather results produced by AS than you are. Of course, it is absolutely a huge improvement over what the weather used to be, but that's even bigger reason for me to get distracted over the things it gets wrong instead of simply looking elsewhere and enjoying what it gets right. At the rate I get something that doesn't match my experiences or that I find physically improbable or implausible, or that simply does not match what I see out of the window, I feel the journey is still far from finished. Which is good, actually.

Plus the RealTurb, by quick testing, seems to do the trick of necessitating some actual flying when on the approach.

-Esa