I disagree with a lot of the talk in the flight simulation community that always runs along the lines of "MSFS is a game with nice visuals, use X-Plane if you want a real simulator". I've used X-Plane for many years prior to FS2020 being released, both for legal currency as well as learning the inside out of complex types I haven't flown.
I'm a US certified commercial pilot and have been using simulators since the early 90s.
Regarding #1, the only approaches I've noticed that aren't in the simulator are some RNAV approaches to smaller airports. Are you seeing otherwise? If you need those now, Navigraph seems like a solution.
MSFS is a technological tour de force that has only been out now for 1/2 a year. X-Plane has been around for decades, which is why things like the G1000 are more reliable at the moment.
Aircraft like the newly released Aerosoft CRJ show the potential in the sim, and this is with the developer only having their hands on the SDK for a short period of time.
For developers reading this, a peek at the SDK is interesting. They are using modern techniques throughout the simulator, such as using WASM along with HTML/CSS/JS for gauges, among many other interesting choices.
For anyone like me who immediately scoffed and said "there's no way you can use X-Plane for actual FAA currency requirements," it turns out you can if you have appropriately certified hardware controls to go with it: https://www.faa.gov/documentLibrary/media/Advisory_Circular/...
Be careful with this. I used to give aircraft simulator instruction and had to learn that AC backwards and forwards. Bottom line is it's not enough just to get a certified yoke and throttle quadrant. The entire package, hardware and software together, has to be "authorized," not just "certified." The instrument for this is a Letter of Authorization (LOA) that spells out exactly which kinds of currency/experience can be logged on that specific Aviation Training Device, and the approved aircraft configurations (e.g. SR20, PA-28R, BE-76). No LOA, no hours. If you can get a copy of the manufacturer's LOA, and the experience you want is in that letter, go for it! Otherwise, you've just got a very expensive video game.
Well, at this time XPlane is better for IFR training and everything the original poster mentioned is true.
But MS Flightsim is still a young product, so let's wait and see. But if I had to bet, I'd always bet on Austin Meyers and his enthusiasm for aviation.
As I mentioned in a previous post, my Dad used the 80s MS Flight Simulator for IFR training. The graphics, on our amazing CGA monitor, were Battleship Gray ground (with mixes of solid green, I think) with white lines representing "scenery". I'm guessing that (for the 80s) the physics were a high priority, but graphics -- even for the 80s -- were not.
Microsoft could do the scenery more easily than others due to Bing, and frankly, if it weren't for the scenery, I wouldn't have shelled out the cash. The flight physics, while positively amazing[0], wouldn't have even been able to be described to me in a manner that would have gotten me excited enough to purchase it, but on that -- alone -- it'd be worth it[1].
I can't speak as to how well it simulates IFR/related things -- not a pilot -- though I mentioned in a previous post that my Dad basically got his IFR rating by practicing on ancient MS Flight Simulator.
I am curious, though -- what is it that MS Flight Simulator's latest version gets wrong about IFR that X-Plane gets right?
[0] Admittedly, I've probably used it for a total of 3-4 hours since its release. What I saw was impressive in that I wasn't intending to notice the physics, but it was impossible not to. There were moments where I had (extremely minor) anxiety while angled just right, looking out the "window" trying to get the plane to do something nuts.
[1] Bear in mind, my impressions come from not having played a modern flight simulator (outside of ones that are not intended to be accurate) in over a decade. Perhaps some of the things that I'm so impressed with have been available in other simulators and I'm just seeing them for the first time, here. I'm encouraged that so many people love X-Plane; I hadn't looked at it, yet (and not exactly sure as to why), but I'll be checking it out, now.
> MSFS is a game with nice visuals, use X-Plane if you want a real simulator
I completely agree -- Microsoft Flight Simulator has always been more than a game. I haven't used X-Plane, myself, and wouldn't have the expertise to properly judge the differences, myself.
However, as a kid in the early 90s, we had upgraded from an 8088 to a 80486, and getting time on the shared, single, computer, was difficult while Dad was training for his pilot's license and later IFR rating using the 80s version of Flight Simulator.
The funny thing is that until recently, it never occurred to me that the intention of Flight Simulator[0] was to be "a game". Now, I'm sure 1985 (or so) Flight Simulator on an 8088 left a lot to be desired as simulators go, but even back then it was "far more than a game".
I remember playing with it after my dad purchased it. The pretty picture on the front/box art gave everything the appearance of being a game. But it was missing one, critical, element. Games, I thought, were supposed to be fun. But outside of a small handful of major airports, which had a few lines to represent landscape, the game was basically "make the small airport disappear into large amounts of gray landscape", dial in the frequency, aim the plane, and then the best part -- when you're done, try make it a car[1].
As a child, I flew with my Dad in his Piper Cherokee probably somewhere in high doubles/low triple digits (over about 5 years). As a kid, I remember my Dad saying something along the lines of: Take-off, Landing and Problems are the only time the pilot is doing anything. Most of the time you aim the plane and wait. My Dad used to keep a 500 ct box of Atomic Fireball candies to keep him awake on long, multi-leg flights.
I have only, briefly, played with the latest Flight Simulator, but I felt like they nailed a lot of the physics engine. I used to love taking my friends with us on trips in the small plane. Where we lived, by about my teen years, all of my close friends had been on at least one commercial (jet) flight, and they all acted macho (the boys, anyway) having flown before (none would admit to a fear of heights at this point). The most intelligent of my friends would start to see their anxiety rise shortly after the hangar was opened. My Dad would hand me this pole to attach to the front landing gear -- so that I could pull the thing out of the garage. I was all of 145 pounds soaking wet, so watching me pull a larger-than-a-car-sized object that seats 7 with one arm and no physical exertion is the first sign that it's going to be a bumpy ride.
Shortly after, when my Dad asks for everyone's weight, the new passenger is confronted with the fact that "the plane is so light that everyone's safety partly depends on where I sit in this thing". If they aren't scared, yet, the next chance is about ten minutes after take-off, when my Dad would usually make a comment over the headphones about how smooth the air is -- I'm not sure if he was always being sarcastic, but "a smooth flight" was usually about as bad as the worst turbulence I've experienced on a commercial jet. So depending on how long we had planned on flying that day, my buddy now realizes this is as safe as they're going to feel.
The best description I've heard of it is that it feels like you're sitting in a kite. It's a kite that can be controlled, safely[2], but you feel every gust of wind and you can't help get the feeling like mother nature is getting annoyed with you and might just decide to slap you out of the air, much like we would do to an annoying fly... and that it would take very little effort on her part.
[0] In reference to the original MS Flight simulator from the 80s/90s -- there are many flight simulation games out there, but the point of these games is usually "to use a plane for war/combat/other traditional game purpose". The physics engines of these kinds of games are far simpler and sometimes feel like the plane is flying through loose snow rather than air -- i.e. like a ski/snowboard simulation game that includes "up/down" and reacts more slowly to turns than "being firmly attached to the ground" but not so slowly that you feel completely at the mercy of the wind.
[1] Also known as "landing" -- on that version, if memory serves, a good landing was characterized by a high-pitched "chirp" (implying the wheels hit the ground) and you could roll around on the ground. If you didn't stick the landing, the plane just stopped dead wherever it touched the ground (I think a "game over/start again" screen popped up but I'm not sure).
[2] I've written a few stories about the kinds of fun my Dad has experienced flying -- he was a "private pilot" in that he was not licensed to charge people to take them places, but he flew multiple times per week as a sales manager/owner of his company. He hated flying if he wasn't the pilot (commercial or otherwise) because he'd been through enough trouble in the air (and handled the problems successfully) that he had greater confidence in his abilities to handle problems than "a person he didn't know who was in charge of his life". I had always thought we were trying to save money, but it was far more expensive to fly the small plane (alone) than commercial.
> Please don't post insinuations about astroturfing, shilling, brigading, foreign agents and the like. It degrades discussion and is usually mistaken. If you're worried about abuse, email hn@ycombinator.com and we'll look at the data.
That is not a fair comparison. If MSFS has less features but better graphics it is either moving backwards as a sim or becoming more arcadeish. A sim with less sim is less not more.
The new Flight Simulator is an upgraded version of the engine that dates back to at least 1990s, if not earlier, and traces of the fact are obvious. Legacy aerodynamics simulation from Flight Simulator X (2006) is included and fully functional. Air traffic control system introduced in Flight Simulator 2002 is also there with minor changes throughout the years. Even some artwork is 20+ years old, such as "737-400_scenery_t1.bmp" texture map[1] for a fictional 737-400 that came with Flight Simulator 2000[2]. I guess they forgot to remove it, or left it for backwards compatibility reasons.
"not much" is a bit subjective. In any case, I'd wager that 99% of the ATC code was taken from them. It has the same incorrect terminology and nonsense than in any other version of Microsoft Flight Simulator.
From scratch? No, consequently there is a pretty much automated aircraft converter [1] and the directory structure is incredibly similar and early alphas had the fsx flight physics, and started with fsx airports before the ml scenery was finished.
It's certainly a more ambitious engine upgrade of the engine than p3d or fsw, even excluding the cloud scenery stuff, but "write a flight sim from scratch" is a task on par with "write a Web browser from scratch".
Yes, most of this forum could write a IE3 or Netscape 2 or Lynx clone in a long but manageable amount of time, but if you want to catch up to Chrome/Firefox/Safari, good luck.
I completely agree about avionics, for example the first time I tried FLC mode on the G1000, the "Nose up" and "Nose Down" buttons were inverted ("Nose up" would increase the target speed, hence actually moving the nose down, and vice versa). And many things were missing, even basic ones such as "track up" mode for the MFD. Overall, X-Plane avionics are much better.
I also agree with you that X-Plane still gets an edge in terms of flight model, even though the gap is not as wide as it used to be between X-Plane and FSX/P3D.
The one obvious shortcoming of X-Plane, when compared to MSFS, is the lack of accurate world scenery, on the one hand we could say that it's just eye candy but on the other hand accurate scenery could help students prepare for VFR flights. Yes, you can generate high detail scenery for X-Plane using Ortho4XP, but it's time consuming and it requires a lot of disk space.
VFR students shouldn't be anywhere near a simulator, imo. Primary training is all about stick and rudder skills, which you can't pick up in a simulator. The feel of the controls isn't even close.
(I guess you could use a simulator to help teach pilotage and dead reckoning... but that normally gets sprinkled in with the other lessons along the way.)
On the other hand, simulators are useful for IFR training, when you're focusing on procedures. But you're not looking at the scenery in that case. ;)
> VFR students shouldn't be anywhere near a simulator,
I agree but I wasn't talking about stick and rudder skills, I was talking about cross country VFR preparation (learning landmarks etc.). I guess you can do it just using Google Maps, but it would be fun to do it in sim IMHO.
PS regarding the expression "VFR students": there is a preconception where VFR skills are just something that you learn as a beginner pilot, and then once you have enough experience, you start focusing on IFR. I think this mindset is wrong. There are some very experienced bush pilots who regularly - and sometimes mostly - fly VFR; aviation is not just airliners.
Now that you mention it, I could see MSFS being useful for general area familiarization.
Foreflight has 3D mode now for the same purpose. But MSFS would certainly be more fun, and an instructor could mix in other navigation skills. It would be a good exercise to drop someone into an unknown area to practice VOR navigation.
(And 100% agreed about the "VFR students" thing. Poor choice of words on my part.)
Pair X-Plane with PilotEdge, and you've got something that can help you with your radio skills so that you can effectively communicate with ATC. So many pilots go without flight following (asking ATC to watch your path and notify you of things you should be aware of) simply because they are afraid of the radio.
Completely agree about stick/rudder skills though...
Is anyone doing anything akin to a direct drive system for HOTAS force feedback? That would be incredible.
I have TM's Warthog system which is presently collecting dust due to lack of triple screen support (Nvidia Simultaneous Multi-projection or SMP, AMD has nothing) - in my space/flight titles, but hopefully not forever.
It's not anywhere able to convey forces affecting the plane in the way my DD steering wheel system does for driving. The feedback I get there combined with transducers is truly sensational, especially once you've had the time to become attuned. I'm becoming half competent now and it's an amazing thing to be able to do from the home.
Whilst there might be limited stuff to simulate in higher end fbw planes, for anything hydraulic there ought to be some options. I once had a Microsoft FFB joystick but considerable time had passed since.
There are sticks, yokes and even rudder pedals with FFB, from a couple different vendors (for example the ones made by the Swiss company Brunner) but they cost thousands. Unfortunately, AFAIK, there is nothing in the lower cost segment.
Wouldn’t it be awesome if someone could license brand names like Boeing, Airbus, Embraer, Cirrus, Garmin, or Cessna and sell reasonably priced panels mimicking instrument panels of those makers with their help?
An LCD touchscreen and a plastic overlay would do wonders and still be a reasonable compromise.
I think you've fallen into the uncanny valley, where it is very close to real but not close enough, so it feels off. The rest of us without your reference context continue to find it amazing.
X-Plane has a big advantage by the fact that it has been out much longer. Their built-in Garmin GNS430 / 530 and G1000 are not exact replicas of the real ones (subtle differences in behavior and missing functions). However you can fix that with a simple add-on, because RealityXP sells an add-on that lets you connect the real actual Garmin simulator software that is often used for training (which you can download for free) with the simulator. And if you do that the buttons are all 100% exactly the same as in the real thing.
I'm expecting this type of thing to be released for MS FS as well over the coming years. It just takes time for these third parties to write a version for MS.
I don't understand how it's an excuse that the current version is new or newer than X-plane? Either this is not actually MSFS but just a new game that sells by using (abusing) a known name or it is MSFS that has roots back to the 80's. If it's the last then it is clearly moving backwards if it has missing features and less realism. The logic that every new release is starting from scratch and isn't comparable to the old version is strange. Even if they throw out all the code and start coding from scratch MSFS is still not some un-comparable game to the last release.
To take it to an extreme: If Adobe started from scratch and released a new shiny Photoshop that looked good but had the features of MS Paint would people see it as moving backwards or "give it time, third parties will re-add missing features"?
In case of third party content I think it's a valid point. X Plane did not have these add-ons when it was released, they came a few years later. So it's not that strange to give MS FS a bit of time to catch up in terms of third party add-ons.
Missing features in the game itself can't be excused that way, but guess what... X Plane still after many years also doesn't have a good enough for training built-in simulation of a GPS.
FS2020 is the first simulator Ive played and on a whim Ive been using an add on pack call "FS Academy VFR/ IFR" to learn IFR. It's ok although the PDF manual needs editing because it throws acronyms at you without defining them so I find myself googling terms like VOR, NDB, OBS, etc.
Thanks for your comment regarding X Plane's better IFR realism I'll be sure to check it out next. I enjoy "programmatic and calculated flying" a lot more than I expected, it's a shame FS2020 doesn't encourage it. I think the Hacker News crowd would enjoy this type of flying if they knew more about it. It also makes me feel a lot safer knowing how meticulously pilots have to plan everything. I dont plan on getting an instrument rating but the basic IFR knowledge made me a lot more confident with regular VFR flying because I feel better calculating descent and turns banks rather than just "eyeballing and feeling" them. I got much better at landing than with the in game tutorials to the point where I want to get a PPL.
>3) Planes don't feel "right" at the edges of the envelope.
Is there a simulation that does this well? With regard to racing sims, there's still some argument on what has the best tire model. I can only imagine how complex aero and airplane models must be.
Aero models are actually much simpler than racing tyre models.
Most older flight sims used a lookup table for aerodynamic forces, which is gives a pretty decent model for most of the flight envelope. Newer sims do model the airflow around the wings but it's still relatively simple.
Tyre models have similar history, most models are empirical and based on the Pacejka tyre model with some adjustments, not quite a lookup table but some very basic math formulae with no physics behind it. Notable exception is iRacing which has a physically based tyre model based.
But the physical phenomena going on in a deforming, elastic racing tyre are much more complex than a wing cutting through air.
Not sure about the details, Forza Motorsport 4 has completely redeveloped the tire model at some point, with fresh intel from (IIRC) Pirelli, and they only improved since then.
I could be wrong but it seems to go way beyond Pacejka in my experience (including sidewall deformation). On old cars it's night and day.
I assume most sims have moved beyond vanilla Pacejka model by now, but they are generally still empirical models with their roots in Pajecka. The standard model doesn't account for dynamically changing pressures and temperatures, for example.
When tyre manufacturers provide data, it is usually in the form of coefficients for the Pacejka model or some variant thereof, the model being the de facto standard in the automotive industry.
But I admit I don't know the specifics about Forza 4 tyre model.
At least a few years ago, the iRacing tyre model developers said in their dev log videos that their model is unique in being physically based and not an empirical model. But that may have changed.
Empirical models are generally considered "better" feeling for sim racing, iRacing's model is criticized for being bad in handling extreme slip angles and ratios (read: drifting).
Project CARS described their tire model that they developed for the first game[1], based on three coupled simulations for the carcass, threads and heat. They describe it as a fully dynamic model.
Not sure what the state of the art is though, I just recalled this post from way back.
> The standard model doesn't account for dynamically changing pressures and temperatures, for example.
Forza definitely accounts for these, there are three thermal bands exposed in the telemtry UI: outer, middle, and inner. Pressure is a tunable, and it varies with heat.
There is also axial and radial deformation. Like, on small rim/big wall tires accelerating produces a tire twist along its rotating axis, as the wall itself is elastic, which creates a lag in tire reaction, followed by a tightening (and even a bounce) when it reaches elasticity limit. Same on turns when the tires deform laterally and depress, which creates some additional lag/bounce back effects that tally up to create interesting situations at the limit. This all matches up with my (completely anecdotal and limited) real life experience.
Most people play the cars tuned up to a given perf class slot, which often includes rim/tire upgrades by default (there's an auto-tune feature) when handed by the game for the cars to be sort of competitive in that perf class, so they probably don't get to feel that to the full extent. I myself don't care about the perf, I like to restore the car parts and settings to stock condition and enjoy the "original" feel of cars I will probably never drive or ever come near, and lap the car by myself or with a couple of like-minded friends.
> Empirical models are generally considered "better" feeling for sim racing
Yeah, theory has this way of matching reality, save for the corner cases that are the ones being interesting. I god honest don't know what Forza is using, and certainly don't claim it's a super accurate sim, but they did a fantastic job to make it feel realistic to great detail, and gracefully degrading from a full-fledged sim race setup to a gamepad (where it still manages to convey a ludicrous amount of info given the device limitations)
Live for Speed also had tyre deformation, temperature, pressure etc but it was still based on Pacejka model, applied to different parts of the tyre separately.
I think this might be spot on. Simulators tend to have extremely steep learning curves and turn off a lot of people. MSFS2020 is trying to capture the casual market too, which inherently means lowering that entry barrier.
It's a fantastic game though, and the satellite imagery AI magic they've done is outstanding.
Disclaimer, not a real pilot but I have flown X-plane for many years. So, yes, I am biased. But I do recognize the enormous progress MS has made with their latest sim. The visuals are stunning and the rest got a fairly big upgrade as well. It's hard to not get enthusiastic about it. I don't have the hardware to run it but I definitely want it.
Avionics in the default planes (both products) are not intended to be hyper realistic. Study level plane add ons exist for both X-plane, and all of MS recent simulators. They typically cost more than the base simulator and you can geek out over every little part of the plane's functionality. There's a great ecosystem of add on developers for both simulators that can get you a lot of realism.
That being said, Austin Meyer's primary motivation for building and continuing to build X-plane has always been simulation fidelity whereas MS core customers do include some hardcore users, they are mostly a bit more casual users. However, add-ons exist that do have a some hard core users and they do provide an excellent experience.
In between shipping Vulkan support for the core sim (which landed last year), Austin Meyer actually dedicated an entire release (11.40, 2019) to addressing some long standing edge cases with aerodynamics. There's a great overview of that on Youtube where he explains what he did (https://www.youtube.com/watch?v=v4UfGE-JJY4). He even addresses "that new MS sim" at the end of the presentation.
There's all sorts of subtle behavior that he figured out the mathematics for and built into the sim. I like flying tail draggers in x-plane and the behavior definitely improved. It wasn't bad before (famously) and it got a lot better with that release. That represents the level of attention that goes into the product and it's a reason why lots of people like me continue to enjoy that product. It's also the reason you get more sensible behavior at the edge of the envelope for a lot of planes. Some plane manufacturers actually use plane-maker (which is included with x-plane) to model their new models: it's the easiest way to get a feel for how their plane will behave before they build it.
Likewise, the reason the default G1000 is so nice in X-plane is that they have a German contractor who is a real life IFR pilot obsessing about it for years for the past decade or so. His mission in life is to make this thing as real as it gets. And it's apparently very good. I've not actually touched the real thing ever of course but the simulated one is quite a complex thing to figure out. X-plane also has had ways to interface with real Garmin equipment for many years. But that's just one vendor and of course there are many. Most of the other ones are only available via add-ons.
The recent MS simulator does represent a very huge upgrade in realism with better default planes, improved approach to aerodynamics, weather, etc. It's much more engaging without investing a lot in add ons (which in the past was the only way to not make it look hopelessly outdated). Those add-ons still exist of course. Including scenery add ons, airport add ons, etc. It's good out of the box but not that good that those add-ons are no longer needed. In a way it caught up to X-plane's rendering capabilities (and a bit beyond). Where before x-plane just looked way better out of the box. I use a lot of ortho4xp scenery with osm2xplane objects and a few other goodies (all free btw) and you get quite far with just those two. But what MS did is definitely a few levels better. But for a six year old product, it's not bad.
X-plane 11 is coming to the end of its cycle. I expect v12 betas to start showing up end of this year or early next year. I expect that there will be a lot of progress on the scenery front particularly now that they have the Vulkan upgrade out of the way. They've always been good at making the most of data driven scenery production given that they have a small team. And there is a lot of open data to work with.
Thanks for that -- just out of curiousity, what is "instrument currency training"? My Dad was an instrument rated pilot (among others) and trained on the original MS Flight Simulator. It was obviously missing far more than probably any of the flight simulators released after 2000.
Also curious -- what is meant by your third bullet point? Where is the physics most obviously screwy? As a non-pilot, I was blown away by how the plane felt. Obviously, I don't have the sense for what the plane is actually doing like my pilot father would have, but while playing, it was the dynamics of how the plane moved that really impressed me.
I use simulator (X Plane) to shoot approaches every few days or so. In real life, you don’t really get to go down to minimums every flight. Simulator helps me to stay proficient.
In normal flight, the models are easy. However, in more unusual flying modes (eg stalls, spins, etc) its very hard to model a specific aircraft behavior vs a generic one. I have experience flying several models and a stall in Citabria (tiny aerobatic high wing) is very different than a stall in much bigger Bonanza. X Plane gets some (not all) of it right while MSFS doesn’t even trying it seems.
I'm not a pilot by any means but I did play xplane a lot and flex the 737 with the FMC and autopilot and such. When I tried doing the same in MFS, it was just nonfunctional. The detail wasn't there.
I disagree, I believe that MSFS and XP11 are marketed to a similar niche. MSFS was marketed as a realistic sim and not as arcade, there were multiple statements about that, including in the very first post of the official blog[3]. At the same time, you can buy XP11 on Steam, a platform for gamers[2], and while it's true that there is a version of X-Plane for professional use[3], most people who own it, use it in a home setting.
You may have misinterpreted what the parent meant about a "game" vs a "simulator".
XP11 focuses on the flight models and physics of flying a plane, whereas MSFS2020 focuses on the visuals.
Neither are perfect, but for those studying or familiarizing, XP11 is a far better choice since it goes out of it's way to mimic real life. MSFS2020's basic G1000 (crammed into nearly every cockpit in the game) still has buttons that don't function, just for starters.
Different focuses, one's a "game" and the other is a "simulator".
OK after reading your comment I get what you mean, yes it’s possible that that’s what parent meant. However I still think that MS should spend some effort fixing the avionics, because in their current state, they are not up to scratch, even considering the different focus as explained in your comment.
Completely agree, and am hopeful they will add the missing features and have some sort of "realism" update.
Basic things like the inability to manually set barometric pressure - there's no option in the G1000 implementation, you have to press 'B' on your keyboard and it just sets itself. Little things like that add up and firmly take MSFS2020 out of the "simulator" category... for now.
Some others have mentioned that MSFS2020 might be more akin to a platform, upon which content will be added over years and years. Some enterprising individuals have already figured out how to add helicopters, for example, while officially there's no support for helis by MS/Asobo. So, I remain hopeful.
I would argue that historically MSFS series has indeed been more of a platform and an ecosystem than anything else. It's predecessor Flight Simulator X also had quite shallowly modelled default aircraft, yet over the years third-party (mostly payware) devs modeled almost all commercial airliners in use today to a very high detail.
So, I think MSFS2020 will be a great simulator for those hardcore simmers who are willing to pay 50-150USD for a single aircraft modeled to a very high detail, just like they did in FSX days.
Yes, for sure the add-ons ecosystem will fill every possible niche. However if we look at X-Plane, we can see that having good quality default avionics can actually help add-ons makers, the two things are not mutually exclusive! Many 3rd party X-Plane add-ons improve on the default avionics rather than having to develop a unit from scratch (like it’s often the case with P3D add-ons). This means that developers can focus on other aspects (e.g. failures, systems, walk around, manuals) while reducing dev costs.
What I mean by a "game" is that the goal of the software is to be enjoyable. It's definitely marketed towards a crowd that wants a simulator, but it's not trying to get you ready to fly a real airplane (even if that is possible as a side effect). That's not how it's marketed.
Whereas, the flight simulator that Airbus trains its pilots on is evaluated by a different goal... that simulator's goal is to make sure that you don't kill 400 people on an A380. You're not supposed to enjoy it, using it is paid work and not recreation.
Yes this makes sense, I would say that MSFS is purely focused on the entertainment market, P3D (and many other specialised sims) on the professional market, while X-Plane caters to both segments.
I love planes but am not a pilot or trying to be a pilot. I think MSFS is enough of a simulator to satisfy me. It has enough technical sophistication to be interesting and teach me things. But I dont need to learn to fly with it. And the scenarios I play are completely contrived anyway.
Nope. MSFS and X-Plane may start as games, but you'd be surprised how widely they are used in professional aerospace and aviation for a variety of purposes. In fact, back when MS had the old Flight Simulator (disbanded in 2006,) they had developed and sold a parallel commercial version which was used in a variety of flight simulation and training platforms. It was this version that was sold to Lockheed Martin, which became Prepar3D, which--whether you agree or not--is explicitly licensed as "not a game."
Now, one could argue the differences between X-Plane and MSFS in terms of realism. From a professional point of view (airline pilot or aerospace engineer,) both are very much in the same general field. Truth is that neither of them meet performance requirements to be used in the kind of certified full motion flight simulators airline pilots use on the job, although both products arguably have components that are useable in those. However, both products are also highly extensible. Just because the out of the box models may not be super detailed doesn't mean an add-on product can't come with its own flight and system models--for instance, PDMG and FlightSimLabs produce airliners for the old FSX and Prepar3D that will outperform most everything on X-Plane from a realism perspective. Expect that their products will be on MSFS2020 in time.
Also, it's important to distinguish between flight model realism and systems realism. Both X-Plane and MSFS add-ons can and do come with their own flight and system models which replace out of the box componentry, but arguably it's the flight models that are more dependent on the capabilities of the base sim. Everyone complaining about the lack of realism in the MSFS stock G1000 implementation should just wait 6-9 months, and surely something better will be out.
From my POV, the new MSFS does 2 extremely impressive, arguably revolutionary things, and the aerospace/aviation industry has taken note: a) The photogrammatry rendering of the whole earth in high fidelity 3D detail, and b,) the representation of the real ATC airspace through leveraging ADS-B (i.e. FlightAware and Flight-Radar 24.) Simulating a single airplane to a high degree of fidelity on desktop hardware is easy. Simulating (I mean truly simulating) an airspace of airplanes is not... ATC in most sims--to date--is scripted, not fully simulated. Piping in the real world (and combining it with live multiplayer) is a very interesting alternative.
The visuals are not that relevant for professional training, it's much more about accurate instruments and procedures. On the other hand, in my multi-crew coordination training the first part was sitting in front of a cardboard 737 cockpit. That could also be done with a cardboard cockpit + 2 screens connected to MS FS. Because with cardboard only you have no time pressure, while if something is really "moving" you do have time pressure.
It’s discussed in the article, though it really should be given a different title because of that.
The fact that Asobo is the developer makes it much more impressive. A Plague Tale: Innocence is probably their most technically advanced game (and their...uh...”Rat System” that can handle thousands of rats on screen at once is technically impressive) but it’s nothing like a flight simulator. The article discusses some of the support studios they worked with to develop MFS, which again makes it even more impressive because I don’t think they had much previous experience successfully managing a bunch of support studios to produce a complicated product.
Microsoft Flight Simulator is really a simulator, not just of flying, but also of the entire world.
It would be amazing if they could turn Flight Simulator into a "game engine", so that other developers can build specific scenarios/games on top of it!
Imagine Flight Simulator with all the creativity of Minecraft. That would be awesome.
It's not simulating the world. It's just pulling in elevation and texture data from real world sources (and then people are patching in Google Maps because Bing is not a great data source).
Also if you've ever flown low over somewhere that's not like an airport that is modelled/textured separately to the ortho scenery, you see that the data doesn't hold up, so you'd need stuff with sky high views.
so could be used for flight sims and.... maybe an rts? You're not using that is as for e.g. driving games or fps games, and crunching Street view data into something usable would require processing and storing a lot more data
> It would be amazing if they could turn Flight Simulator into a "game engine", so that other developers can build specific scenarios/games on top of it!
Something _similar_ happens for Digital Combat Simulator in that third party developers can develop modules for the base game engine (so far it is mostly aicraft but there are 1st party ground units and a 3rd party has indicated interest in building a relistic Integrated Air Defense Network simulator on it as well).
Not quite "A world simulator you can do whatever you want with to minecraft levels" though.
> It would be amazing if they could turn Flight Simulator into a "game engine", so that other developers can build specific scenarios/games on top of it!
That's already the case. There's a huge ecosystem of hand-crafted airports, airplanes that are true to every little switch, etc.
Microsoft's flight simulators are an open platform like Windows and are meant to be extended. Hundreds of products have been already released (example[1]) for the new Flight Simulator, and there's a thriving freeware scene[2]. You can of course only use Paint and Internet Explorer, but then you'd be missing out. This also negates criticism towards the simplicity of included airplanes. Not happy with the A320? Install open source A320 expansion[3] and enjoy the ridiculous level of detail[4] it brings.
> "It would be amazing if they could turn Flight Simulator into a "game engine", so that other developers can build specific scenarios/games on top of it!"
They did, back in 2007 in a product called Microsoft ESP, which became Lockheed-Martin's Prepar3D when Microsoft's ACES Studio was shut down in 2009.
“Seem” is the key word here. I’d be fine if they stopped at the scenery, but this quote is just outright misleading and makes it look like a paid hype piece:
> “All aircraft have been reviewed with professional pilots who have many hours flying these aircraft or test pilots from the manufacturers to make sure they are controlled exactly like they should.” … In other words, when sim-pilots try to yank-and-bank an A320, they will get the exact same response as if they were flying the real thing.
For contrast read some takes by professional pilots, here’s one[0]:
> I’ll confess that I don’t know that type of modelling FS2020 is using. But given how it feels, and how it’s the child of FSX, I’m inclined to think it’s using a lookup table. This isn’t necessarily a problem — a good lookup table with many data points could be just as good, if not better than, a blade-element model. Unfortunately, FS2020 doesn’t seem to have a good lookup table. The the linear behaviour unrealistic and the non-linear modelling is terrible.
> Maneuvering is laughably easy — and not because the aircraft handles nicely. It’s easy because the aircraft feels like it has no inertia. Yes, the real plane is very maneuverable and responsive, but there’s a “weight” to direction changes which is completely missing. In FS2020, you can instantly pull 60 degree or more turns, and exit them just as quickly. In XP11, there’s a residual tendency to turn that you have to arrest when exiting a maneuver.
If you care about how the world looks, get MSFS. If you care about the physics of flying, get X-Plane—it’s an excellent project developed by a smaller company and sold for a fraction of the cost of Microsoft’s product. XP strives to provide the actual fine-grained simulation of flight surface behavior based on weight, shape, etc. (Yes, this means your computer will spend most of its resources simulating rather than painting a pretty picture.)
If I had to draw a parallel, X-Plane is more or less the iRacing of flight sims, while MSFS2020 is closer to something like F1 2019.
The article was written by someone who didn't even bother to read/watch Microsoft's promo material. Lookup tables are a thing of the past. Microsoft has published two videos showcasing their new aerodynamics simulation[1][2].
There seems to be a lot of pushback from old-timers who have sunken thousands of $$$ into X-Plane and its expansions (from terrain packages to cloud visuals), many of which now look inferior to what Microsoft's new simulator offers out of the box.
That’s nice progress since FSX. I recall reading that new MSFS is compatible with FSX aircraft, which made me think they must rely on the same lookup tables at least as fallback.
While I stand corrected on lookup tables, the author does make a point on how his personal experience of flying the same aircraft as he does IRL does not compare to in-sim experience with MSFS, and does with XP. I saw similar feedback elsewhere, Wikipedia mentions it too in their reception section. If they actually implemented a sound physics model, this potentially could be addressed by tweaking it in future updates, so there’s hope.
For the record, I only bought XP11 itself for about $60 (no paid add-ons) and am flying the default Cessna. Most pro simmers appear to be invested in third-party aircraft for both sims, so I don’t know if there is so much unfounded bias.
That person is not a professional pilot. They have 160 hours.
Their description of the handling does not match what many other real pilots say about MSFS2020, and, as others have said, they are wrong about how the model is implemented. Their description of the stall characteristics also does not match my experience of playing the game.
> I’ll confess that I don’t know that type of modelling FS2020 is using. But given how it feels, and how it’s the child of FSX, I’m inclined to think it’s using a lookup table
I'll note that this statement is easily disprovable through the two aerodynamics videos that MFS published on Youtube (as another poster here has pointed out).
From what I've seen and read, FS2020's aerodynamics simulation is more advanced than the one in X-Plane.
> > Maneuvering is laughably easy — and not because the aircraft handles nicely. It’s easy because the aircraft feels like it has no inertia. Yes, the real plane is very maneuverable and responsive, but there’s a “weight” to direction changes which is completely missing. In FS2020, you can instantly pull 60 degree or more turns, and exit them just as quickly. In XP11, there’s a residual tendency to turn that you have to arrest when exiting a maneuver.
Uh...what?
Has this person actually played MSFS2020? Because planes definitely have inertia in MSFS2020. You can definitely feel it in the large planes like the A320. If I push the stick hard to the left, the banking begins slowly. And if I then stick right just as suddenly and hard, it keeps banking left for a couple seconds from the rolling inertia.
Even the super-agile Extra 300S has inertia. Doing fast aileron rolls and then trying to revert to straight-and-level flight is tough, as it really wants to keep spinning.
MSFS2020 is the most fun I’ve had with software in a very long time, even with the bugs. It is basically Google Earth with a built-in flight and weather simulator. Bravo, Asobo Studios!
In gaming or technical terms, it's not much good as a game per se (or at least, wasn't when I last tried many years ago) but is still quite fun when playing around familiar areas, getting an idea of various flight times.
There's a pretty amazing documentary from Noclip that goes in to a bit more depth on how they built the world and how they were able to model the physical attributes of each plane more accurately with than previous versions — https://www.youtube.com/watch?v=0w7q1ZFfsxs
Reading this article makes me feel like the team that built this wasn't just building a game. It feels like a passion project where the detail was important, not because it made for a better game, but because it mattered to the people building it.
FS was Bill Gates' hobby project. The game team was disbanded (licensed to Lockheed as Prepar3D) one year after he stepped down.
Around FS2000 an oceanographer was hired to model the ocean waves.
So very focused, deliberate effort has always gone into FS, but that's more to achieve realism than for fun.
My theory is that military sim applications drive the realism, and probably there is military funding, so it's not really a "passionate gamers" angle. This would explain the quick licensing to Lockheed (for govt. contracting requirements), which is unprecedented for a game if you think about it.
(Atari Battlezone was likely also studied and used by the US military. I have some high scores on that. There's one in the SF pier historical arcade, as well as most of the Atari classic machines.)
I've worked with Sebastian Wloch and was so impressed with that guy. If there is such a thing as a 10x engineer I'd nominate him. I didn't realize Asobo was involved with flight simulator. Overall, seems like a big win for European software development.
One thing that was not mentioned in article (searched for "OSM" and "OpenStreetMap"): building data and maybe also some other info is from OpenStreetMap.
BTW, OpenStreetMap data is available for anyone as long they credit it (there are some additional rules about merging it with other databases) - see https://www.openstreetmap.org/ for place where you may see just part of dataset and contribute.
Interesting side effect: It’s been really difficult to find a decent joystick since just before the game was released. Now it looks like they are returning to shelves but still in short supply. Not sure if it’s a coincidence but it meant I haven’t had a chance to try it yet.
There was a run on HOTAS controllers when Star Wars: Squadrons was due for release. Star Wars "space" combat is basically WW2 style dogfighting, so it's not far removed from flight simulator territory.
It's so weird to me that people run out and buy $200 HOTAS controllers for a fun "toy" game like Squadrons or elite dangerous, when something like an Xbox controller is what they're designed for.
The same thing happened early in lockdown with steering wheels when iRacing got very popular due to some real races turning into virtual races on television. I think they upped their subscribers by 5x or something crazy like that.
Microsoft made the force feedback joysticks for a bit; but no one does anymore (not at consumer prices, certainly). MS Combat Flight Sim 2, and Jane's WW2 fighters; could both give you a feel of the wind and had enough terrain to do ridge gliding... that was immense fun.
The fact that they can sometimes struggle with simple things such as reading emails (Gmail, cough) is telling something about some of the modern programming practices...
I'll double down this. Deeper Blue, IBM's machine that crushed Kasparov in the iconic chess match in 1997, has less computing power than an GTX 1070. So my computer, 20 years later, is more powerful and it plays Flight Simulator in glorious ultra details.
That a GPU from 2016 is more powerful than a supercomputer from 1997 is unsurprising.
Does anyone know around which year consumer computers started being more powerful than Deeper Blue? Wikipedia says Deeper Blue was rated at 11 GFLOPS, which is... nothing, by modern standards. The GTX 3080ti is rated at 30000 FP32 GFLOPS.
On another hand, comparing GFLOPS from Deeper Blue vs a modern gpu FP32 GFLOPS is probably meaningless.
Deeper Blue was not a general purpose computer, it had loads of chess-specific hardware. So it's tricky and not very meaningful to compare it to modern computers.
Do you imply that Google is somehow using Gmail as a trojan horse to harvest CPU cycles and do some business with them? Like hashing crypto or SETI@Home?
Demoscene is spending a lot of effort to make things as efficient as possible. Other programmers have other priorities. Clearly the performance is what it is because "running in a browser" was deemed non-negotiable (as it strengthens Google, and then the performance just had to be "good enough" (with Google's lead and market size).
> The size of this data set, which is continuously updated and expanded, is now at three petabytes (and growing)
You're not downloading that much data at installation time, so I assume there's a crucial online component where stuff gets streamed in (partially simulated?) from the server, and your computer is just handling what's currently in view. That seems much more feasible.
Seems like base geometry with base textures are on local hard drive, beyond that higher resolution textures are streamed from cloud, down to 3cm precision, insane.
I bought it when I had a pretty good PC (Ryzen 9, 2070S, memory for days). I’ve since sold it and opted for much less powerful machine. I’d be scared to try it on less than what I was running it on at the time.
The fact that MS did not add mouse yoke to the game, and despite the backlash still hasn't done so (allegedly to push the sale of new flying gear) makes me so sad.
Using the mouse for steering is absolutely superior to a keyboard and connecting a joystick everytime I want to spend a few minutes flying is just non-practical.
As a result of pushing the envelope for realism, they have simultaneously pushed the envelope for PC hardware requirements. It really is the: “Can it run Crysis” benchmark of modern times.
It is very neat to see those kinds of technical leaps.
The most interesting thing about it wasn’t the fact that I could travel the world (in fact I flew out of a small airport 5 miles from me and crashed into my apartment first time I tried it) but the fact that it supposedly simulates real time weather and air traffic conditions. Not sure if that’s technically as impressive but still very cool.
FS2020 is weird in regards to performance. I found it ran okay ("shockingly playable") with reduced settings on my 1050, for instance, before my new card arrived. But at high settings, it's possible to sink very large amounts of money at hardware without ever hitting a consistent 60fps. The performance/$ curve is very flat.
When FS2020 came out, I had an i9-9900K, 32 GB of RAM, and a GTX 1070. With the detail settings I run at (pretty high, and in 2560x1440), I was getting around 25 fps flying over NYC. The FPS display told me it was taking about 20 ms of CPU time per frame, meaning even with unlimited GPU power, I'd cap at 50 fps.
Sure enough, I upgraded to an RTX 3080, which has well over double the horsepower of my GTX 1070, and I only went to 40-45 fps flying over that same area with the same detail settings.
What kind of sucks about being CPU-bound is that it means nVidia's DLSS technology won't improve framerates at all. At best, they could create motion-smoothing to generate frames between rendered frames, similar to what a TV does, but doing so introduces latency, since the software needs to gather 2-3 frames (possibly more) to analyze for motion and generate intermediate frames. And of course, just like a TV, this could break down badly when the algorithm fails.
Going far back in time... In the mid-80's, I remember early MSFS (2.0 and 3.0) were lightyears beyond almost everything else. 4.0 was a major milestone. It wasn't until EA, Spectrum Holobyte, and a few others got into the category in the 90's that there were also other semi-realistic FSes for the time. There were many early ones with flight dynamics more like pure video games ("fakey") that weren't very fun. Another fun sim was Megafortress, which was a fictional, modified B-52 where the player could change crew stations.
One thing though is seriously missing in MSFS is crash physics. So many of other types of games have realistic collisions and destructuble environments (Racing games like Dirt 4 even have that for licenced models). Why can't MSFS can't have something that other games are having for many many years?
Microsoft removed crashes from Flight Simulator following the 9/11 attacks. If you want good destruction physics a combat focused simulator like DCS is probably more relevant than a civilian flight simulator
I find it interesting how the article seems to reference the Bing Maps division as though it isn’t part of Microsoft. There hasn’t been some sort of spin-out that I should be aware of, has there?
I've used MSFS from the early days and here's some random facts:
- the model in the article foto is holding the yoke wrong - you're generally supposed to use one hand, unlike a car steering wheel
- FS2000 was used in some commercial simulators, as seen in various ads in Aviation Week & ST. In the pre-digital era, often sectional maps or video of models were projected for the pilot
- I aced my instrument checkride after practising with FS5, which essentially had no terrain.
Are you implying that it's not actually realistic, it just makes you think so? I'm not sure if the team that actually developed the software used many psychologists. Perhaps they did, but the article didn't mention that.
I tried the latest MS FS when it got out and went back to X-Plane for majority of my instrument “currency” training:
1) MS FS still doesn’t have a lot of US instrument procedures available.
2) Avionics (e.g. G1000) doesn’t behave as in real life: buttonology is just wrong and often the logic is wrong too.
3) Planes don’t feel “right” at the edges of the envelope.
I still checkout MS FS challenges once in a while, but I treat it as an arcade. It is not real by any means thought rendering is pretty nice.