Accuracy vs Realism – The Major Problem for Asobo

Discussion Hub General Discussion
1

There several areas of discussion at the moment where the basic question boils down to: exactly how “accurate” do we want the simulator to be, or would we be satisfied with what looks and feels to be “realistic”.

A few examples:

***Scenery. I fly over many parts of the world, and think “this looks great, very realistic”. I then fly over my home town (Chelmsford, UK) and think “well they haven’t done this at all well, it’s nothing like the real thing”. And my house does not actually exist.

***Live weather. I generally choose live weather for my flights, as it gives a good variety, an forces me to fly in all conditions. Very enjoyable. When I fly near home, I think “ where’s the rain I can see out of my window?”.

*** Weather simulation. As I said above, it’s great flying in the various scenarios, but it is certainly not modelled anything like perfectly. This is not surprising, as weather is one of the most chaotic things to model, and normally requires super-computers at meteorological centres to achieve a best-guess.

*** Taxiway numbering looks great to me, and serves it’s purpose – to those in the know, many of these are inaccurate and therefore confusing.

*** Aircraft liveries look okay at airports in other countries, but I don’t feel at home never seeing an EasyJet flight at Stansted, London. (I’ll have to get used to it though – they’ve now sadly moved out of both Stansted and Southend.)

*** Ground services. It certainly adds to the immersion, being able to call the baggage cart, the catering truck etc when preparing for a flight. But when you look out of the cockpit at all these vehicles waiting to pounce, it is a bit contrived to say the least.

*** Fauna. It’s fun looking for the wildlife, and they are extremely well modelled and animated. So very realistic, but the distribution is hardly accurate – I don’t need to go to Canada to see seagulls in real life.

*** Aircraft dynamics, To me, a non-pilot, it feels good as well as challenging trying to fly a plane “properly” in the sim. To a pilot, there are lots of things not modelled correctly, and I can understand how it irks them. Being a scientist, I cannot watch science fiction films for the same reason.

Looking at the list, and the fact that Asobo are trying to implement this over the whole world, it’s obviously an impossible task. They need to be pragmatic over what they can and can’t implement, and where they accept they can’t be 100% accurate (which tbh is everything in the list, and more), they need to decide what level of accuracy they can achieve (or as some people crudely put it, how much they need to “dumbdown”), while still providing a good experience for the user. And, despite all the complaints I’ve got, it generally works for me.

Finally, I don’t think it is helpful to classify FS2020 as either a game or a simulator, as it is clearly both. No simulators are 100% accurate, and no games are 0% realistic. Perhaps we should just call it a “Flight Experience”.

15 Likes
2

I like this. Pragmatism is key. I think the old adage of “you can’t please all of the people all of the time” is very true here, and many people want many different things from the sim. Personal view is that Asobo should get the realism up to a point where it “sufficiently realistic” across all the points you raise above (albeit “sufficiently realistic” remains subjective) for the (vast) majority of users before starting to push into added levels of accuracy in some areas, if at all. This would enable 3rd parties to add the accuracy where people specifically wanted it and therefore enable people to have the choice to “fill in the gaps” that there might be.

An example is the G1000. Asobo planned to do a basic job with this, but based on feedback, they’ve realised that this isn’t “sufficiently realistic” and so are going to improve it. I’m going to assume they will never recreate all the screens and functionality of the real thing - so that is where a 3rd party (or community team) might step in with a mod that takes it to the next level. Those who want this level of accuracy can then access it. “The rest of us” who might want extra accuracy elsewhere don’t need to tap into that if we don’t want to.

The key here is probably that Asobo need to define (and then publicise) how far they will take the base sim in these (and other) areas. Given the Q&A yesterday, I got the distinct impression that if the server cache issue with METARs is now fixed and once they fix the lightning issue, that might be “it” for live weather. For most of us (per your post above) that is probably more than adequate. For those using VATSIM or other external “stuff” they may need/want more.

Clarity from Asobo would be really helpful here. Which is why I think they could look to provide written responses to some of the questions that we all wanted to ask at the Q&A enhanced by explanation at future Q&As rather than trying to explain a complex topic like live weather ot AI traffic behaviour in a live Q&A

EDIT - I wonder if the thread title should really read “Accuracy v Realism - The Major Opportunity for Asobo” - it’s their chance to set out how far they want to take the sim in the many areas highlighted

3 Likes
3

A good post. Though in my opinion, it says “simulator” on the box. To me that means that flight and weather modelling are key to a good experience.

What they have is not bad and often it is realistic, not counting obvious bugs. But flying through storms is a huge disappointment and apparently, if quoted in context, it is nerfed on purpose. To me that is completely bizarre. no flight simmer will appreciate that.

I think they are on to something special with this sim. But if they are in fact dumbing down modelling they know is wrong then it is a massive kick in the guts to me. I though they were bugs that would get fixed and not, “people don’t understand weather so we will disable most of it”.

It can be salvaged though if they are able to maybe make an “advanced” mode that removes the limitations.

The above opinion is largely based on quotes from another message. It may not be right and I hope it isn’t so I won’t criticise them until I see evidence that is really their path.

Of course, future pay ware aircraft will fill some gaps. But the weather, it is unknown if that vital and most interesting part of flying will be as well rounded as it can be. As I say, most of the time it is modelled ok, leaving out the bugs, but updrafts, down drafts and severe turbulence are sorely lacking.

I had one flight in x-plane where my aircraft literally went upside down in a storm. It was was extremely challenging to right myself and not stall and fall into the ocean. It was the most exciting and rewarding flight ever, if that will never be possible in MSFS then it really will just be for sight seeing and fair weather practise.

3 Likes
4

Any flight simulator will have to have compromises to allow for the fact you aren’t in a real aircraft. You can’t feel the movement of the aicraft, g forces etc, so some method has to be introduced to allow for that sensory deficiency. The camera shake is a good example. Also fonts generally need to be bigger than real life, to cater for the resolution limitations of a computer monitor. Anywhere Asobo hasn’t done this makes flight simulation difficult, unless you have a very large and very high resolution monitor. Field of view is another area that requires less accuracy in order to make flight simulation appear more realistic using a computer monitor.

5

You should listen to the q&a. I think it makes sense in the context of where the sim is and the huge level of abuse that Asobo have been the subject of from many in this forum. We reap what we sow and so we only have ourselves to blame for this.

1 Like
6

Working Title already has:

7

Yes I know, but what we don’t know is how far Asobo might take it and hence the work coukd end up getting done twice

8

Agree totally with almost all of what ExtraPilot6480 says. About my only quibble might be with the “No simulators are 100% accurate” bit, since it rather implies that you can put a percentage to overall accuracy, though I don’t think that was really meant anyway. In any case, you can’t. At least, you can’t come up with a meaningful number when doing something as complicated as modelling the entire world, it’s weather systems, and the billions-of-dollars-to-design-build-and-operate aviation infrastructure that we are interested in. There are some things within the sim that could meaningfully be measured in terms of ‘percentage accuracy’, but almost almost of them can’t. Not without inventing arbitrary measures that may tell you nothing of any real consequence.

Simulators only ever replicate part of whatever it is they are modelling, and almost always do so with less than perfect results. That doesn’t matter, as long as they are good enough for the purpose they are built for. The first ‘aircraft simulator’ built for real-world pilot teaching was probably the Link Trainer: basically a crude model ‘aircraft’ you sat in that pitched and rolled around on a universal joint as you moved the controls. Not very realistic, but useful for giving trainees an idea of what to expect.

So what constitutes ‘good enough’ for MSFS? Well, avoiding the rather pointless ‘game or simulator’ debate, I’ll just note that (for now at least) the sim is being sold only for entertainment purposes. Clearly, some (almost all?) people using a flight simulator expect their entertainment to come from interacting with something that at least vaguely resembles real-world flying. And most, will probably think that the more it resembles the real world, the better it is. Which makes ‘good enough’ more or less undefinable. All a flight sim developer can ever do is produce something that makes a fair crack at it, using the technology available at the time, for a price the market is prepared to pay. If someone doesn’t find a particular simulation ‘accurate’ enough to entertain them, they should perhaps consider seeking entertainment somewhere else. Or lower their expectations a bit.

As I’ve written elsewhere, I think it is fairly obvious that MSFS was released prematurely. And with rather more hype than was justified by what was actually achievable - though that wasn’t helped by some people taking MS marketing-speak, and inflating it further by multiplying it by whatever their fertile imaginations could come up with. Unrealistic expectations haven’t been met. Because they can’t be. What is probably realistic though is to start from where we are now, and say that there is enough evidence to suggest that things can still be improved. And frankly, if back in the 1980’s, when my first computer ‘flight simulator’ came on a cassette tape, and updated the four-colour low-resolution wireframe screen image at maybe two frames a second, I’d have been shown what was being achieved now, I’d have been astonished. And would have taken some of the complaints about the ‘lack of realism’ on this forum as the ravings of lunatics. Things have got a whole lot better. Almost unbelievably so. And they can go a whole lot further. Just not all the way to an impossible ‘100%’.

1 Like
9

I have the same thoughts in ref to Northeast Ohio, USA looks. In FSX you could see the Perry Nuclear Power plant as well as other Cleveland landmarks that just aren’t. I wonder why when it’s supposed to be Satellite based. (I could be wrong the satellite thing though.

10

Maybe coming in the US upgrade in update 7?

11

Its analogous to the Frame Rate discussion.

Your eye can be happily fooled (for most purposes) at 24 FPS. Does more FPS add to realism?
Realism (without G forces or stick forces) would be smoothness of movement, responsiveness of the control to input, sensitivity varying across planes, the dynamics of reacting to changing wind etc… which ultimately results in the “visual” experience.

Required Fidelity here … an 80/20 approach is good enough for immersion, IMHO. and is most related to core flight simulation programming.

Where it gets interesting is the other aspects of immersion … scenery fidelity to what you know, accuracy to real world navaids and signage. This is programming, but more of a data quality issue especially if the scenery is being auto-generated.

So, if I were to prioritize functionality for a flight sim - heres my order of priorities - a flight/weather model, graphics/control performance, scenery model, nav-data/auto-gen/airports etc, and then real-time stuff like auto gen cars, ai traffic, airport traffic , live traffic etc. You should be able to dial these elements up or down to tailor the sim experience (or have it be done for you) based on what is important to the player… A multiplayer experience will be different from solo hard IFR or virtual airline pilot duties. Realism will mean different things.

The platform has good capabilities, how well they are managed is a different topic altogether.

12

A simulator at it’s heart is in the case of most aviation regulation agencies is - a replica of a specific type, make, model, or series aircraft.” A flight simulator duplicates the performance and flying characteristics of a particular airplane, and it must re-create an airplane’s cockpit with great fidelity, including exact reproductions of the real aircraft’s physical controls, instrumentation, and switches. It must reproduce the aircraft’s flight characteristics with high fidelity. (mostly lifted from Bruceair.com)

And here we are with a base product that presupposes multiple aircraft. So by definition, it’s intended for a purpose beyond pure study, although some customers bought it for exactly that.

You can’t recreate the experience of flying without both the environment and the subsystems. An actual G1000 trainer panel by casual market canvass is around 1000USD - and presumably recreates down to exacting detail some baseline version of the avionics package.

It’s funny - at the beginning of FSX, I don’t think anyone would have thought it could rise to being good enough to qualify as an FAA Approved ATDs, but they are. At least the base code anyway - as most of them are P3D, some XP.

So there is a definite question of what is this product intended to do? And to what level is the detail necessary to facilitate the intention?

The unspoken messages coming out of the Twitch (as well as the body language of the panel) is that they intended (and still do) to deliver a more comprehensive product. They are grateful for the enthusiasm and expertise of a community, but now must come to a clear reconciliation of how to capture and include those changes that are clear improvements, without conflicting with base product improvements and features.

I don’t envy MS-Asobo’s task at all. Yet, they’re still cranking at it despite all the up-down feedback from this forum and other channels. I’m cautiously excited to see all these changes that are projected before year-end. To me, improvements are both big and small. CPU efficiency with more FPS, you bet. Bring it. More planes? Hopefully ones that I would like to fly (I’m not a big iron or bizjet fan). Can I really get closer fidelity with Live Weather now? Let’s see it!

When authoritative or empowered panelists make commitments, people will hold them to it. Let’s acknowledge the difficult road ahead, but be cautiously optimistic that over time, it will get better. Useful, productive feedback with suggestions, and constructive criticism remain one of the best ways we as a community can help to improve it.

1 Like
13

well-written. not casual at all :slight_smile:

14

Ignorance is definitely bliss as far as scenery is concerned. The problem in my opinion though is that if everybody (not including people who live in areas represented by Photogrammetry) says the areas they know are not realistic then a great deal of what we are seeing is inaccurate.

Ultimately Asobo and MS have to decide what they want the product to be. If its a sim that accurately represents flight, they need to focus on the aircraft functions, behaviour and features along with weather. If an accurate digital representation of the earth is also high on the list of priorities they need to work on the systems that they use to populate the orthos with 3D objects. Autogen is in some way very good, in others rather poor and I base this on just how much work I am having to do to make the area I am most fond of flying in, look accurate..

15

I’ve mentioned this elsewhere. You familiarity breaks the illusion. I think the same when I visit my home. The house doesn’t look the same, the trees are far too prevalent, and too large etc.

I don’t think the same when I fly over San Francisco though, and I think it looks amazing.

16

Well written OP. 100% agree, especially on fauna and maybe some more basic improvements to flight dynamics (because other sims seem to have this nailed down a bit better than Asobo).

Graphics look amazing in the country-side or remote areas, look terrible in US citites. I’ve posted 15,000 examples of LOD and photogrammetry looking awful in Chicago, NYC, San Francisco, Seattle… I primarily only fly airliners and occasionally the Longitude, and it always seems this makes the LOD terrible vs when flying a slower propeller plane (have no idea if this is factual, just my observation)

On to the point, Asobo and Microsoft made it clear this is meant for avid simmers and casual simmers, and that balance won’t be struck for quite some time, in my humble opinion.

2 Likes
17

As the OP, I don’t think I can argue with any of the points raised in the above answers, so thank you all for your comments.

There is no doubt much is yet to be done to correct existing problems, as well as to move closer to the original objectives. Constructive criticism/advice can only help Asobo in achieving this.

1 Like
18

I like to think of myself as being a pragmatic person; as such, I really don’t mind that things like live weather or scenery isn’t particularly accurate. What does “grind my gears” though is glaring errors and omissions that really should have been picked up before the product was released; basic things like not being able to select altimeter pressures in millibars. And now that they’ve given us a setting for that there is exactly ZERO change in the TBM … or lighting that doesn’t work.

I have a whole list of things like this; they’re not things that require massive time for them to overhaul - they’re things that should have been available on day one - and they’re things that should have been tested. I’m left with the feeling that the software is more of a public beta at the moment than something up to the quality of a release candidate.

I’d love to help them with feedback about specific things that just aren’t right … but they even make that process impossible. I’m still partially enjoying the product - and I’m sure than in a year it’ll be a much better product - but I for one can’t wait for one of the 3rd parties to create some far more accurate models. I just can’t help but feel that it’s a significant quality control failure when someone like me can install a new product like this and discover 10 bugs in the first hour.