February 20th, 2020 – Development/Insider Update

[Casualcas 830]

1 hour ago, domkle said:

You are welcome. 

Avsim had a dedicated forum to bush flying years back. I don't see it anymore, it seems it went the way of the dodo. Maybe the new sim will make it relevant again. 

In the sim as it is today (P3D), the main area to fly the bush are Alaska, Idaho, Montana, BC and Papua New Guinea because the Orbx "fat" sceneries include scores of nicely made strips.  

What I look forward to is to fly to other regions that were ignored until now in Central America, the Amazon basin, Borneo etc. I was delighted to see the Sirena Ranger Station in Costa Rica in the preview which looks like a b*tch to land into. Good omen ! 

 

 

 

 

Here's a start 

Top 10 Scenic Airports 2019

1. Donegal Airport (Ireland)
2. Barra Airport (Scotland)
3. Nice Airport (France)
4. Orlando Melbourne International (US)
5. St Maarten (Netherlands Antilles)
6. Saba Airport (Netherlands Antilles)
7. Queenstown Airport (New Zealand)
8. Toronto Billy Bishop (Canada)
9. London City Airport (England)
10. Aosta Airport (Italy)

[Scottoest 1,539]

1 hour ago, FlyingInACessna said:

Have we already discussed the fact that at 5:00 of the video it says that the 80 specially chosen popular airports will be given taxiway IDs that match real life, implying that the non-specialized airports will not have accurate taxiway markings? If that’s true (and I hope I’m simply misinterpreting what they said), that would be a huge issue.

Everything else they’ve shown and mentioned is incredible, but it would be incredibly difficult living with incorrect taxiway markings (or having to manually update them for any airport I want to fly to).

Pretty sure they say that the other airports will use generated markings, derived from that pass they do with the waypoints and stuff.  Whereas the 80 airports will all have signage and markings mirroring exactly what they are/look like in real life.

So the other ~37k airports will all have coherent, easily navigable markings and signage - it just won't necessarily be the "true to life" markings.

[royalwin 22]

I guess it's not exactly the right place for this post, but since so much discussion has been around it here maybe it's ok.  I thought it might be worth talking a look at the complexity of the problem of "undulating" runways.  I believe some people look at it as if it were a relatively trivial thing to implement and that not doing so somehow shows ignorance, incompetence, or spite on the behalf of the developer.  Not only is that view rather... uncharitable... I think it shows a poor understanding of the magnitude of this problem.  Let's think about it.  First, there are some constraints.  If they are going to make 37,000 airports they need to all work, as in a plane in the sim can land on them, and most planes are not offroad vehicles, so they need to be flat and smooth enough for that purpose.  And they have to be so despite the wildly varying "quality" of the underlying data.  Additionally, this has to be done as procedurally as possible.  Even if you only averaged 2 days per airport you'd be talking about nearly 300 man years of labor, that's just out of control from a cost and time perspective.  So with those constraints, now we have the technical problem.  As a few people here have correctly pointed out, DEM resolution, or the spatial sampling frequency or density or whatever you want to call it is much less than what would be required to capture runway undulations.  30M is pretty high res these days, though there is spot coverage that's better, but much of the world may be 100M or worse.  This will all get better over time, but this is still way short of what's necessary to capture these imperfections as some sort of ground truth.  I wouldn't be surprised if many - most even - of the runways don't even have a single DEM vertex on the actual runway surface.  The belief that you can just drape the textures over the DEM and have usable runways is based on a misunderstanding of the precision of that data.  Not only, that but the problem gets much worse, because the accuracy of the elevation data isn't precise to the level one might suppose.  Airport data has touch down zone elevation and airport elevation, usually to the foot, which is more accurate than the DEM data.  However, it's critical to get the runway data correct, so you have to merge two conflicting data sets, at very different resolution and accuracy and have the airports be correct with no or minimal artifacts.  That's not easy, and on top of that you have photoscenery textures that aren't exactly georeferenced to the airport data so they don't even line up perfectly which can create very unsightly double airport effects.  So you are merging three conflicting data sets.  Tough problem.  Even if you wanted to put these undulations in and have them perfectly match reality, where would you get the data?  Undulations are mistakes.  They are the errors that occurred because the engineers tried to get perfectly flat runways and were unable, or things settled over time.  I'm not aware of a comprehensive data base of survey points along all the runways in the world.  If someone is, please correct me.  The fact that they manually edited, to some extent, 37,000 airports indicates that they couldn't solve all of these issues procedurally.  It's a certainty that the decision to manually do this wasn't taken lightly or even happily, what a massive undertaking!  That they have done so is pretty awesome, and hopefully thinking about it this way will make it easier to appreciate what they've done, rather than conclude that this sim is unusable without these RL undulations perfectly matching reality.  My two cents.

Edited February 22, 2020 by royalwin

[LHookins 3,076]

I took one look at the screen shot with the grass and thought, "Heh! 'Comically oversized blades of grass.'"  I was amused.  And that's as far as it went.

Hook

[bonchie 2,305]

42 minutes ago, royalwin said:

(snip)...The fact that they manually edited, to some extent, 37,000 airports indicates that they couldn't solve all of these issues procedurally.  It's a certainty that the decision to manually do this wasn't taken lightly or even happily, what a massive undertaking!  That they have done so is pretty awesome, and hopefully thinking about it this way will make it easier to appreciate what they've done, rather than conclude that this sim is unusable without these RL undulations perfectly matching reality.  My two cents.

This is a really great post that lays out the issues with doing it right. There's a reason XP11 has a switch to turn off sloped runways, because in many, many cases it's laughably exaggerated, and in the worst cases it's unusable.

MSFS is simply not going to do that. Their entire philosophy is to do it right or not do it. They will have sloped runways (confirmed) but they may not have undulations. If they don't, it's one of the least important things I'm worried about.

 

 

[LHookins 3,076]

13 minutes ago, bonchie said:

There's a reason XP11 has a switch to turn off sloped runways, because in many, many cases it's laughably exaggerated

It seems a lot of things in XPlane are exaggerated or overdone.  This is actually a good thing as there are a lot of people who simply don't do "subtle."  It only becomes a problem when it is assumed to be the Gold Standard that should apply everywhere.

Everything has its own place.  I trust MSFS to get it right.

Hook

[Scottoest 1,539]

Can we make an "Undulation Discussion" subforum or something?

[royalwin 22]

Sorry 🙂  I won't mention it again.

[Scottoest 1,539]

5 minutes ago, royalwin said:

Sorry 🙂  I won't mention it again.

LOL, it's no problem.  There's just been a kind of mind-blowing amount of "undulation' talk in the last 36 hours.

[Tuskin38 5,926]

Donagal airport looks great, and will probably be one of the first places I fly from. Though it looks like they based it off old imagery, it's missing a parking extension (planes not cars) that was added some time in the last 4 years.

Edited February 22, 2020 by Tuskin38

[azulkb 302]

A little serious for five minutes.

Let's compare any vanilla simulator and not filled with addons to what this simulator without addons will offer. We are facing a real technology gap. 

And maybe in 2040 we'll still have another technology gap by Microsoft or other ... 

And so on and so forth. But current technology does not allow us to simulate to the nearest centimeter every relief. Maybe in 40 years, yes. A terrain as big as France in Asia with 30cm mesh precision, only a Japanese company has the data, and it costs 25 dollars per km2. Imagine the budget...
 

edit : 140 dollars /km2 for 50cm

source :aw3d.jp

Edited February 22, 2020 by azulkb

[MatthewS 761]

2 hours ago, royalwin said:

As a few people here have correctly pointed out, DEM resolution, or the spatial sampling frequency or density or whatever you want to call it is much less than what would be required to capture runway undulations.  30M is pretty high res these days, though there is spot coverage that's better, but much of the world may be 100M or worse. 

Runways are long...

If you have a runway that's 2000m long and a 100m elevation mesh resolution then that's 20 elevation data points along the length of the runway.

Also "merge in" data points from chart data (eg runway thresholds) and give these higher priority than the elevation mesh.

The runway can be rendered at a high resolution in game (1cm?) with those 100+ data points being "smoothed" (think bezier curve).

The result is a smoothly rendered, undulating runway based on the underlying mesh and published data.

This approach can be applied to the entire airport surface (runways, taxiways, aprons etc)... I guess you need to exclude "spikes" in the elevation data from hangers/terminals etc

[Murmur 3,885]

One could make splines/Bezier curves based on the underlying DEM + runway data points from charts. So you have correct runway (average) slope and plausible, smooth runway undulations.

[ChaoticBeauty 6,692]

6 hours ago, sightseer said:

has anyone considered the GPU cycles per grass blade?  would you rather have great grass and a slide show or pretty good grass and good performance?  I don't really know how much of a performance hit each blade is but it can't be no hit at all.

This is most likely going to be tied to the graphics settings anyway. If the grass density, draw distance and quality can be tweaked via a setting, there is no reason not to allow a high-quality effect for appropriate GPUs.

[royalwin 22]

1 hour ago, MatthewS said:

Runways are long...

If you have a runway that's 2000m long and a 100m elevation mesh resolution then that's 20 elevation data points along the length of the runway.

Also "merge in" data points from chart data (eg runway thresholds) and give these higher priority than the elevation mesh.

The runway can be rendered at a high resolution in game (1cm?) with those 100+ data points being "smoothed" (think bezier curve).

The result is a smoothly rendered, undulating runway based on the underlying mesh and published data.

This approach can be applied to the entire airport surface (runways, taxiways, aprons etc)... I guess you need to exclude "spikes" in the elevation data from hangers/terminals etc

You assume the data is perfect, and even correct.  There just isn't reliable data that represents the actual runway and can have this approach work.  I know it sounds reasonable, but I suggest you try it, and have it work for the whole world.  Unfortunately It won't even though it would be awesome if it did.  The subtle runway imperfections are well below the noise floor of the DEM data.  No signal.