tonywob

Not to belittle anybody's hard work, but it's hard for me to understand the end goal with this, or why he's trying to sell the idea to Laminar and others. The idea has been tried, many times since even the XP9 days, and is a step backwards from where things are actually heading with the new tiling scenery system. From the video, it's taking textures and drawing them on top of low resolution aerial imagery tiles, e.g. where there is a forest, it's painting over a polygon of the forest. This isn't going to work because the forest polygons just cover over the actual detail, removing lines in the forest, empty patches, variation. The data inside OSM lacks the detail to do this correctly, and it would need to be much more procedural than just filling in polygons. The other major problem is that we still need the imagery to paint on to, so it still needs the same disk space as normal orthos would. He could achieve exactly the same effect just by draping textured polygons over the terrain like Simheaven does, with the advantage that painted polygons can have higher detail, normal maps, etc, something the aerial imagery can't have Correcting the aerial imagery is a good one though. The idea should be to take multiple legal aerial imagery sources and blend them together nicely to get a seamless mosaic of a large area (Many providers offer imagery from multiple years). That's a lot of work, but would be the best visual pay off for this type of scenery.

It'll be backwards compatible, but I agree, this is one reason I essentially paused the work I was doing waiting on this (especially the tiling system)

Never heard of this. From what I can gather, it's aerial imagery that's been manually colour matched on seams so it looks more seamless (Although the seams are still there from what I can see). He's also taken OSM data and put textures over the top of the aerial imagery where the forests are to try to blend it with the trees that are there, covering up the nice ground detail underneath. A lot of this stuff was done in the TrueEarth sceneries I did a few years ago, especially Florida (i.e. Baked tree/AO, and colour extraction for vegetation/houses). But in the age of AI, I imagine this process is much much easier now and can be done easier on scale, or with faster image processing libraries

It was on my system. It had a tweak to increase the number of cars that ended up ramping up the amount of police vehicles, especially visible at night, so I removed it and the problem went away

That's accurate 😆 In seriousness, yes, this is a bug in the xroads package that still hasn't been fixed

It's still OSM + MS building footprints + heights + the now closed Google Sketchup 3D warehouse. However, now it's just generated again with OSM2Xplane instead of World2XPlane (since that's what eventually turned into the TrueEarth sceneries and couldn't be used commercially as it's now owned by Orbx and not myself). Perhaps people have too high expectations on what it actually does?. At the end of the day, it's similar as before, but now you're paying to support the effort to build and host these, encourage more work, fix problems etc. Considering he's been hosting these for as long as I can remember for free and spending lots of time generating them, I think that's only fair he gets something out of it. In my view (and from painful experience 😊), the scenery could lose most of the unnecessary stuff such as park benches, bus stops, cemeteries etc, it just drags the sim down and makes it take longer to load and uses more memory etc. Nobody will ever see these things once up in the air.

Years ago (Back in the XP10 days) I actually did some experiments of doing the same land-class tile system P3D/FSX used into X-Plane, however because of the way X-Plane's scenery system was (and still is), it was all pre-baked tiles that essentially put fake orthos with realistic objects in tiles all across a large reason. Worked pretty well until we got to sloped terrain (as anyone who has used FSX/P3D will tell you), it even had the classic "roads going over nothing" feel the old sims gave us 😁 If everything they are planning to do in the next-gen scenery system actually happens, it'll open up lots of opportunities for devs to do things like this (if it's even needed).

I rarely bother buying airports these days either (and have also stopped making them). The only ones I care about in high detail are my local ones and I'll certainly buy high quality ones, but honestly, the gateway airports are really good these days. It's a far far cry from the days when we just had a runway and nothing else. A big thanks to those who have been contributing to the WED airports database

I wouldn't have super high expectations over this. It's unlikely to contain the detail MFS does, but it'll still be a big improvement. My take is that it'll be more accurate, better performing and follow real-world data better, including better art. It'll be far easier to author as well since mostly it's image masks and more modern 3D formats. Tools like Ortho4XP which build complicated meshes and try to smooth and flatten airports hopefully won't be needed, and far simpler tools just involving image manipulation can be used. Once I get my hands on the new tile format, I can work on updating my mesh tools to generate ortho tiles + DEM.

I'm looking forward to finally being able to use the tiling engine, it's so badly needed (we've been asking for this abilitly for years) :)

I gave it a go (I don't like just running random exes from the internet without any source-code, but I took one for the team) It's fairly hacky at best, but a decent proof of concept. The major problem is that it can't remove the mesh underneath, so it slightly raises the elevation essentially having one world floating on top of the other. It also converts in raw format, so the sceneries will be extremely heavy. This is not the fault of the tool, it's just a currently limitation of the X-Plane DSF format.

It's not streaming it in, it's downloading the data from Google beforehand and creates a scenery package. You need to select the area you want, generate it and drop it into the sim.

As someone who uses this airport several times a year in real-life, it looks really accurate and instantly brings back the pain of waiting in long queues at the Ryanair checkin desks

I have their Cessna 172s, and for such a basic aircraft, it's extremely heavy on the framerates (although very nice). I can't even begin to imagine how heavy a full-blown 737-Max is going to be.

Be careful with this script, it decompiles the mesh and recompiles it losing height accuracy and causes artifacts and issues at some payware airports. It also won't do much with this scenery since it uses the old style X-Plane 11 trees and water masks.

Sorry, but I'm not sure what you mean? You'll have to give examples. If you've lost any custom scenery it probably means your scenery order has reset and incorrect.

You can get them down to 1/4 the size if opting for 2m/px imagery (ZL16), which I found is absolutely fine above 1500ft or so. I have the Canaries at 60cm (ZL18), but honestly I don't see the benefit too much unless on approach/departure.

The northern coast looks really nice, one of the nicest areas to fly in

The issue is not just limited to disk space, it's to do with the slow loading times and memory requirements, since all that data has to be loaded in large areas due to the legacy 1x1 tile system. If anyone has used TE Florida as an example, then they'll have experienced the super slow loading times caused by this. Whilst it looks good, this scenery is even more detailed than that, so for most users it's just not practical at present, especially for those who just fly liners. There are also lots of hacks I have to do to make features that work in the other sim, e.g. Having matching roof colours and shapes, tree colours matching ortho, etc that X-Plane simply doesn't support easily. IMO, people would just end up disappointed, especially if they can't run the scenery smoothly, even if it does look good. I'm not a fan either, so it would have to support both for those that don't mind the storage requirements.

Since MFS was released a few years ago, interest for such products has hit a brick wall. Since such scenery is essentially free and worldwide in the other sim, it's difficult to compete with that and actually make any money despite some people still wanting to see new and updated regions. Either way, I have most of Europe and large chunks of the US generated, but I'm waiting for the new scenery system before deciding what to do next and how to release this for people to use. The most important factor is being able to reduce the size of the scenery downloads and the regions, i.e. Either by streaming it in or allowing users to only grab the areas they want.

I've been using Macs for years (Since 2003), but I was looking at finally replacing my Macbook with a laptop running Linux (Dual-booting for Windows games, and Linux for work and X-Plane) and this might be the push I need to go for it. I was looking at these custom build laptops that are well supported on Linux https://frame.work/pl/en/laptop16. Sadly, it seems 32GB is the max RAM I can top out at (I currently have 96GB which I make good use of in Photoshop), which considering the price of RAM these days might be a downgrade somewhat.

Yep, I remember needing to clean this up extensively when using the same MS building footprint data to make the TrueEarth US regions. Likely the same thing, i.e. OSM + MS Building data. The effort and bandwidth costs involved to generate and host these files inevitably led to them wanting to charge for it. So this doesn't surprise me, it was going to happen eventually

I've seen a substantial increase in FPS from the scenery changes in this, especially on large scenery areas.

It's been an interesting topic with some nice info, but time to close as it's been taken over by bickering

Yes, despite their limitations such as fixed seasons and baked-in shadows, I still think orthos combined with autogen are the best we currently have. While they can look rough at ground level, the scenery becomes quite convincing from a thousand feet up, especially with good lighting. The long-standing drawback of ortho imagery is that it’s not truly orthographic, building and bridge roofs are often displaced from their actual positions, and large shadows are permanently baked into mountainous terrain. You can suspend disbelief to a point, but the issues can become obvious, e.g., a mountainside covered with trees while the underlying ortho shows the ground completely black due to a shadow stands out like a sore thumb.