Behind the Screen : November 2024

[Stephen]

 

Behind the Screen : November 2024

 

To most Simulator users it is the flying dynamics and the reproduction of iconic aircraft that is their general focus. To relive, train or learn on how to fly a broad spectrum of aviation's greatest developments. From airliners past and present, General Aviation aircraft, Military, Helicopters and even the weird and wonderful contraptions that can fly through the air. But the environment that you fly in is important as well, and yes "I'm Talking about scenery again".

 

"Oh god I really wish this guy would move on from this", I can hear the groans from here, the foreheads banging on the desktops.

 

But it is important as X-Plane is in a critical stage of it's development, it is the biggest issue right now, and an important one to get right. There are other areas that need development in X-Plane, certainly Multi-player functionality and Multi-Threading are major priorities, but even Laminar Research are focusing very heavily on what they call "New Generation Scenery", and over the last few months, trickles of what NGS is now coming out.

 

And it is an interesting debate, and one that could make X-Plane a prime simulator. There is no doubt that the environment is the Microsoft Flight Simulator's 2020, and now 2024 is a main focus, you can't really call it a flying simulator, it just has planes and helicopters in it, but this is a gaming platform, full of experiences...  even now a career if you want one. That is fine, but the biggest attraction and one of which Microsoft promotes is this significant aspect of the environment...  and this is where it does get interesting.

 

If you want X-Plane to grow as a Simulator, then you have to embrace the fact that the environment around your aircraft is a major attraction to people who want to use a Simulator. But this aspect always has been X-Plane's biggest imperfection. Since the start of my Simulation involvement, the default scenery has been, well mostly....  crap.

 

I'm not talking about some of the excellent add-on scenery produced for the X-Plane Simulator, I'm talking about the basics here, and why X-Plane does not have a bigger audience, and some, if most will say, "Well we like it that way", fine, but that way will mean the decrease of investment, not only by user participation, but by the essential developers moving on to better conditions. I'm not going to go all depressed on you again, because actually X-Plane in development has had a very successful year, but the simulator has also lost a lot of talented talent that we couldn't afford to lose, worse it's not attracting talent either, whether you like it or not X-Plane has reached a milestone, that could flip it simply one way or the other. From one perspective it looks absolutely fine if you keep to the basics and use default Global Airports, and yes things at this point are far better than it was 12 months ago. But that crucial line is now more in important than ever, if X-Plane does not grow, and quite significantly over the next few years it will become only a marginal player in Simulation.

 

And we have to be realistic, in that the default Global scenery has always been a hindrance, limiting the expansion of the simulator. The change from X-Plane 9 to X-Plane 10 was a significant step forward, but except for the new excellent trees in X-Plane 12 there has been no progress for over a decade, were as every other component of the Simulator has had significant upgrades, all the way throughout X-Plane 11 and to XP12. Basically we still have the same two tier autogen and tile mapping, and although the mountainous frameworks are very good, at a ground level the basic low-res ground textures are basically the same as they have been for years. Add in the poor insertion of custom scenery, and those horrible wide green spaces between the default and custom elements on the mesh, and I still cringe every time I land at Tampa, Florida, with those wide green spaces were buildings should be, so I never ever really liked it for realism.

 

Worse is if you are flying over empty spaces, say the American Kansas west, plains or worse African Sahara, or in my case over Australia, the repetitive patterns (landclass) were all there to see, for hours on end, and those Saharan landscapes are horrible and actually won't fit into any custom scenery, yes there are always options, but this default aspect is not at all good.

 

I did a demonstration flight in a helicopter hardware simulator with a particularly large audience, it all looked totally brilliant until I landed on those washed out lo-res chunks of textures, I was actually embarrassed of the scene, this was supposed to be a professional simulator, here broadcasting to a crowd that were interested in buying pro hardware, but I could see the reality of X-Plane's weaknesses, and maybe at that moment it has been my overwhelming focus of being on a sort of crusade to fix the "Scenery Problem".

 

Ben Supnik joined Laminar Research to do the very original scenery for X-Plane 6 with the introduction of DSF (Direct Scenery Format) in XP8, and which also introduced global terrain data. Then Supnik created what you could call the 2nd Gen Scenery for X-Plane 10, this version introduced two other clever elements in using both Autogen to place buildings to the modeling of global terrain data and the OpenStreetMap (OSM) data to provide immersive and realistic landscapes. Remember this was 2011, and at that time is was an extremely big step forward. But OSM had limitations, as you were restricted to it's coverage, so in areas like Africa, South America, and certainly in non-western zones like China, suddenly the lights went out and all the buildings disappeared. Yes Laminar added in more autogen, taller buildings and small industrial areas, but overall it stayed very limited for over a decade, with only a few icons added in extra. alpilotx did do a more Hi-Res version of the original XP10 texture tiles, but he left X-Plane in 2016, since then there has been nothing.

 

These two areas are currently expanded in X-Plane by replacing the default autogen by using 3d buildings like in SimHeaven's X-World series, again the limitation is that it relies on the same OSM data as the default scenery. Secondly is the addon to use photorealistic terrain using satellite imagery, known as "Ortho" and the Ortho4XP tool. Believers standby these two tools to replace what X-Plane does not provide, but they do take a huge amount of your processing power and storage, overall they are not a perfect solution to the worldwide scenery issues.

 

Now Supnik is faced with his biggest challenge yet, to recreate the scenery system for a modern Simulator, and that aspect is harder than what you think. It's the "World" and not just the small play area that is recreated in Games of say around 30nm. In that small game area, it is quite easy to recreate sensational scenery for the characters, but when you have the massive expansive world to recreate, then this huge scale has so many different aspects to recreate. Okay I may be over hard on X-Plane scenery for what it currently does, as to a point it does recreate the world quite well, but a decade on the game has now shifted (no pun intended).

 

Although lauded, Microsoft's approach to scenery has it's limitations as well. For landclass it is very good, shapeshifting the world in all it's elements, but it has one massive deficiency, it is when you get close to the ground. The major issue is ground distortion, plus any buildings, but mostly bridges that have space below that are sheared in physically into the ground. If you are flying at a 1,000ft it is not a very realistic scenario, and this is an important aspect of VFR flying, yes the correct buildings are there, but basically they are all generated shapes.

 

X-Plane's approach in this case for VFR is actually better, but not expanded out to cover major detailed areas. So how do you fix this. The Ortho believers think their approach is the best, but Ortho is not great either, as in many areas the buildings are only burnt in, leaving flat photographic images. But at the same 1,000ft point of view the Ortho works best, as it looks photographically realistic, the problem here though is when you get close to the ground, because the flat photo plate and the island looking 3d buildings, again doesn't look realistic.

 

Like myself Ben Supnik does not like Ortho, his argument like mine is the burn effects, trees, buildings will always have shadows, cars are burnt into the roads and so on, so with the daily time scale of moving light it doesn't work. His work is to try to find a compromise between the Ortho detail, but with not the absolute photo effects errors, so like with our current ground textures it is the best compromise in realism.

 

So what new scenery aspects are coming? well better ground textures means new "Terrain Rendering". Laminar wants to cram in more and more stuff on a tile, make it more dense with data. So overall it looks like what Laminar are working on is a completely new and different approach in creating better ground detail, as at the moment the current DSF is not scalable, but NGS or also known as "Direct Scene Snapshot" (DSS) is highly scalable, also highly streamable, plus has better and faster loading properties.

 

But DSS will at first only be an alternative to DSF, as if it replaced DSF, it would render most of the current Custom Scenery library obsolete. One other feature of DSS is that it will allow for Ortho images to be streamed directly into the Simulator if users still want to use that point of view, The interesting aspect of this "Tile Streaming", is the "streaming" is currently a reference in how the data gets into the sim engine. But will that "Streaming" be local (in the computer), or externally like Microsoft does?

 

Well a bit of both. X-Plane Mobile is actually already a streaming Simulator, so X-Plane desktop is probably going to go the same way with cloud streaming (Amazon's AWS), but with the cache to download the areas that you want to fly in off-line. Instead of downloading the current large DSF tiles with data, the streaming DSS files will be far smaller, but broken up into immediate high data quality visuals, but lower res (data) for far distance visuals.

 

The trick is that Laminar are trying to do is have the best of both worlds, the photo aspect of Orthos, but without the problems of the format (flat and burn ins), so the idea is with these smaller DSS tiles is that they will also be non-repeating, or have no Landclass category, meaning that a single image will be in the future just not repeated over and over to represent the area (my pet hate), but to represent the photo aspects without the photograph, so you will get the best of both worlds, great ground detailing, without the shadows and minute that affects Ortho images, plus the tiles will perfectly reflect the area they represent. Is this the mythical solution?

 

The biggest challenge however is filling the world in. X-Plane will still use the autogen model, but probably expanded. The biggest success over the decade was SFD's (ShortFinal Designs) excellent SFD Global, that took the current duel (USA/European) autogen modeling and added in more regions to it around the world, so yes you can currently see Middle-eastern housing in Dubai, and Japanese housing in Japan, and Australia housing in Australia. 

 

Justin Kissling (mister-x, SFD) was already working on an expanded v2 of SFD Global, but has since been recruited by Laminar Research, this certainly means that SFD Global v2 will now be the default in X-Plane, plus in a far more expanded form, but what of those towns and cities?

 

This is still the biggest challenge facing X-Plane as a realistic VFR Simulator. I always like the autogen spread, with central custom city objects as a great way of doing this in a realistic scenario. In payware efforts were cities have been reproduced; New York, Washington, Seattle, London, Paris, even the excellent Orbx Brisbane and Detroit are excellent examples, the idea works well, the trick is to expand it worldwide to cover most if not if all major population centers. The problem is with custom city scenery is that it is very time-consuming to recreate, building by unique building, and annoyingly these city packages don't repay well for the developer's time consuming task. This is an area that could be redefined, could the new tools of A.I. build our custom cities for us? takeaway the overwhelming task of creating custom scenery... ten years ago this aspect was not possible, but it now may be realistic.

 

Ben Supnik notes that the new NGS scenery will be released is stages, and not completely all at once, so access will be available sooner than later, then the development will stretch out over time like with the Vulkan/Metal roadmap, so this a long term project more than the next X-Plane version journey, but certainly the NGS is coming more sooner than later, hopefully in 2025.

 

If done right, the above NGS development has the potential to revolutionise X-Plane as a Prime Simulator, it's an interesting take in that it is using all the best elements, and is discarding the negative ones, it will also make the Simulator even more efficient, but also in delivering an excess of a more visual detailing and a realistic environment for us to fly in. If you understand the concept, you will find it very exciting on what it can achieve... it is a vision of the future of realism in Simulation.

 

As usual there will be no December "Behind the Screen" 2024 issue, but our full yearly round up of the year 2024 review is to be published on 13th December 2024, so watch out for that.

 

Stephen Dutton

4th December 2024

Copyright©2023 X-Plane Reviews

 

 

[Alpeggio]

Excellent Steven, I agree with your points, and improving the default scenery is now the most important thing I want to see.

I have noticed it looking much better at altitude than before, so I have noticed the progress you talk about.

For me, using Orbx UK XP12 version, or auto ortho with X-world were a revelation, though the load times for the Orbx are quite long.

I generally test my mods at RAF Lossiemouth, but the default scenery around there is truly awful.

MSFS though initially took my breath away, at certain altitudes, like you said, about 1,000 to 2,000ft doesn't look that good to me, it certainly isn't a perfect solution.

Thanks for informing us that there may be light at the end of the tunnel for Xplane scenery.

Al

[Medellinexpat]

Releasing NGS scenery in stages makes sense, but perhaps for another reason than its availability. Part of the issue for a flight simulation is providing ‘Global’ scenery. Just how much of that Global scenery (subtracting of course the oceans) is actually ever used? My guess 5-10% of the earth’s surface if that. Then adjust for those remote territories that people only fly over at 38,000 feet so the detail is overkill. Perhaps the answer is have Xplane track where people fly and then use that data to build scenery detail levels. Of course there will always be exceptions cases, but they would continue to have the existing level of detail. There have to be huge areas in the Far North (Canada, Russia), Antartica, Africa, Australia (sorry Stephen), Central Asia and South America that no simmer, at least for VFR has ever visited. Having access to the whole world, like MSFS has, is a nifty trick, but wasteful. Data, and perhaps AI might have you understand what scenery you really need.

[flightwusel]

Thank you for this Behind The Screen to which I am always looking forward to!

Your observations are honest. And your insights are always enlightening.

 

I don't agree with landclass as the saviour, though. Even with more texture variety and some measures to prevent repeating patterns.

Areas with little to no variation in landclass data (pampa like Patagonia, deserts, ...) will stay lame and dull. Yet with ortho they have wonderful colours and you can see many features. I don't know a single Youtuber or streamer not using ortho scenery.

Improved landclass will help to improve the default sim but I can only see it as a stop gap measure, not the ultimate goal. I've seen SFD, and it is better (and I would expect better OSM- and building footprint data to be included) and it might be a good solution for the default sim, but ortho is easily even better than that.

In fact we have most things in place for really good scenery representation. It should receive a little more acknowledgement and support by Laminar in my opinion.

 

I really don't see the baked-in artefacts in ortho imagery as a real problem. Terrain shadows often help in "selling" the terrain better (even if they are not coinciding with the current sun position). Shadows under houses are not a real problem in my opinion. Cars on roads: for smaller roads we use overlays anyways and on highways (xroads) I don't see it a s a problem. The only disturbing bake-in is clouds, but this is solved for many areas of the world. If we don't continue to sleep on the topic we might be able to use AI to get rid of bake-ins anyways.

 

What Laminar could do without too much effort is improve overlay resolution for vector stuff. And coastline resolution. And use higher resolution digital terrain models (elevation). Make it an option so the current lowres is still there for potato PCs.

And communicate more real technical insights about where they are heading with nextgen scenery. What they really mean by "streamable" would be interesting and if it serves the AutoOrtho use case. Maybe even ask the community about it.

[Stephen]

No Landclass is gone, revising photo imagery to take out the nasty artifacts is a clever step forward, and the far more smaller scale of the tiles will give more arid area detail, so those really bad repeatable images will be gone and replaced by more realistic open space textures, for better realism.

[SkyTrooper]

This is my first time posting in this forum so please forgive me if my comment does not suit this thread. I'm a CS major but space flight & aviation has always been my passion. I had my first flight over 2 years ago, then a couple transcontinental ones since then. I've always wanted to fly.

 

About a month ago, I came across this video by Max Lenormand on How Microsoft Accidentally Made the Most Realistic Map which nudged me into researching more on MSFS and how I could also play that and fly. Before this, I had only known that MSFS was a flight simulator, but other than that knew pretty much nothing about it. So on I went learning more and more about this simulator. When I realized that it was only available on PC & Xbox, I was a bit disheartened as I wished it would run on my MacBook (I have the M1 Pro base model).

 

Then somehow going in the rabbit hole, I came across X-Plane. On finding that it was supported cross-platform (thank you X-Plane devs!), I immediately went ahead and downloaded it. It's been the most incredible few days learning about this awesome piece of software!

 

Initially, I had only thought that X-Plane was a slightly "lower" 2^nd option for those who could not have MSFS (like me). But when I learned more about X-Plane, read on these forums, and watched Austin Meyer the creator himself! connect with his audience through YouTube, I came to realize how wrong I was!

 

Even though it has only been a few days, it already feels like I'm a part of this incredible team effort! I'm extremely fortunate to have found this incredible community of builders and flyers and just want to Thank you all for creating something like this for the world to enjoy!

 

Lastly, I'd also like to thank @Stephen. Your article was an amazing read, and I'm sure we'll see many great improvements in X-Plane in the days to come! Will wait for the end of the year update!