Posts by qwerty42

    I definitely agree with you on that Michael. FSET is quite clunky in a lot of ways like you described. However, it is pretty solid in terms of all its internal routines that download and build the imagery, so my hope was that some of that code might be useful to developers of newer, better tools like Herve's.

    That's not so. Last week a new topic of mine was removed (again), including at least one reply, and there was no notification, no personal message, no trace, no nothing. It was simply gone.

    In that case I retract my attempt to be more fair... maybe they can ninja-edit a redaction into my post above for me :D In all seriousness though, I understand it's a company tool and promotional forum for them. And in general I think it's moderated well by a small team that already has better things to be doing. But, it's also by far the most active and best resource right now for this sim and any user development happening around it. Few things can sour an online community faster than iron-fist modding or not being transparent about it, and this one needs to thrive until other alternatives grow big enough to be useful. I don't think transparency via mod-comments on any deleted or edited posts is too much to ask, and every other good forum out there does this already.

    To revise my post above and be more fair: I think there are visible notifications given when an entire thread or a post within the thread has been deleted. However, I don't recall ever seeing anything on those indicating why it was done, when done by a mod. Showing something was removed is good, but it sure would be nice if a few seconds were taken to at least say why it was done with a short comment line.


    However, when posts are ninja-edited by mods, it shows nothing... except that the post has been locked for further editing by you if you were the original author.

    I couldn't bring myself to click a 'thumbs-up' on a post like this, so I'll just say: I'm so sorry for the pilot, crew, passengers, and their friends and families; and also that you had to witness such a tragic and emotionally devastating event. Fortunately I've never been witness to something like this, but my heart goes out to you all who have.

    I'm sorry to bring negativity in here, and maybe I just get overreactive with these kinds of things, but can the mods here please consider making it a policy to at least add a comment line or notification when they go editing or deleting users' posts? I have had several of my posts silently edited by mods, or outright deleted, and when my words are redacted with no notification to me or reason given for why it was done, it's honestly pretty irritating. One of these was the thread on creating terrain heightmaps which clearly turned out to be quite popular. The most recent case was a Dropbox link I gave out with the FSET source code to help the crew who is working on new tools for scenery geoconversion, which also got ninja-edited.


    If I had at least been notified or knew who the mod was that did this, then I could ask them why they edited the post, and I'd probably find out they thought I was giving out something I wasn't supposed to--in which case I could explain that the code was given freely and unconditionally by the original author for this very purpose. But since posts get edited here and nothing even shows it was done, I can't do anything about it.

    Also, there is something that really bothers me on a visceral level about having a post with my name on it, with my words, and then having those words partially deleted or changed, again without so much as a note on the post visible to other users that it was edited by a mod.

    I understand having rules in place and wanting to keep certain things out of your forums, but silent ninja-edits to our posts or outright deletions without any indication that it has happened starts to feel like blatant censorship. At the very least, this forum should have a visible notice that appears on posts that have been mod-edited like pretty much every other forum on the web.

    Well, I posted a link to it here earlier and it looks like it got silently mod-nuked. This is the third or fourth time I've had posts wiped like this with no notification or explanation and it's really starting to grate my gears, honestly. They did the same thing with my post on how to create terrain heightmaps and I had to plead to have it restored.

    nickhod If it would help, I have the source code for the first version of FSET. It was posted on the web a long time ago when the developer abandoned it. Somebody else fixed some bugs with it later on and released a couple more versions, but I don't have the source for those updates unfortunately. You might be able to re-use pieces of it to manage the imagery downloads:

    hmmm... I thought that it could only load 1 file for any given tile at a certain geoconvert level, where priority is given to paths outside of the default installation directory (e.g. My Documents or any additional external path you've defined in its config file). There is a single unique filename for each scenery tile at each level, and based on the log file it looked to me like it would only pick 1 to load if it saw duplicates. But, I could definitely be wrong! I never actually did any in-depth testing to see if this really is the case. Hopefully it is capable of loading multiples of the same tiles and layering them somehow because that would make it a lot more feasible to seamlessly blend sceneries together.

    I'm afraid not. Hervé doesn't want to make a square carpet of aerial pictures in the middle of nowhere : he wants to make the junction between 2 existing sceneries.

    Therefore the big challenge is to tailor cut masks according to the existing sceneries and then edit the colors of the aerial photo textures he can gather to both match the Swiss DLC and Orbx's LOWI. No automated tool can do that for him unfortunately.


    For the junction with the Swiss DLC, the best would be if IPACS would be neat enough to generate a simple mask outside their scenery. Hervé would just need to invert it and use it as a mask layer for his orthophotos => they would fit exactly to the Swiss DLC border without overlap.

    Without getting such a mask from IPACS, Hervé can spend days hand tailoring the border of his scenery to match the Swiss DLC.

    I don't think this will work either unfortunately. Since AeroFly's imagery is broken into discrete tiles, you can't have two of the same tile in order to overlap them with masks. The only way I think this would work is if the scenery you want to 'blend' with is only converted up to a low-res level, say like level 12, and then you add your overlapping scenery at levels 13-15, for example. But even then you will not see what you want when you are far away and only the level 12 tiles are shown. :/

    Hi everyone, new user here.

    I'm Having trouble with geoconvert. It's not writing anything under scenery/

    My geoconvert folder copied here:

    Please help.

    Tomek

    Hi Tomek, I took a quick look at your files. I think it's a combination of a couple things:


    (1) In your .tmc file, you have mask files globally disabled with the line at the top:

    <[bool] [write_images_with_mask][false]>

    I'm pretty sure this overrides whatever comes below it in terms of enabling or disabling masks.


    (2) Your downloaded imagery from FSET doesn't completely cover an AeroFly tile of any level up to level 13, and since mask tiles are disabled, geoconvert won't write any output at all.


    To further explain, AeroFly has its world map sliced into a grid, and the size of the grid cells get smaller with each increasing 'level' (level 9 is a large grid cell, level 15 is a small grid cell). If your source imagery doesn't completely cover one of these grid cells when you try to convert it, no output will be written unless masks are enabled. If masks are enabled, it will write a partial tile and a second mask file that goes along with it, which 'blocks out' that part of the tile in AeroFly.


    It's a good idea to disable masks when using geoconvert as much as you possibly can. The reason is that masked tiles will make it hard to ever add more scenery around your converted area unless you manually go in and delete the masked tiles later, which can be quite a hassle. This is because those areas that get 'blocked out' by the masked tiles will be permanently blocked out until the mask is deleted. This means you also have to regenerate the file that went with the mask too, or you'll have black stripes appearing in AeroFly.


    The best way to address this is to start with a downloaded imagery area which is already snapped to AeroFly's grid. Then you won't have to worry about enabling mask tiles and you know that your FSET imagery will fully fill AeroFly's grid cells. There are two tools we've created to be able to do this here, one which shows you the cells in Google Earth and another that uses Excel which will calculate the coordinates for your directly and you can copy/paste them into FSET. You can find both tools here:

    Excel tool Image tile coordinates

    Google Earth tool Image tile coordinates


    And there is some more explanation/discussion about this in this thread, starting here and read through to the last page:

    Scenery Loading Distance

    Just wanted to clarify something that I think isn't clear in this thread: for anyone using an Oculus Rift, you'll get better performance if you don't use the SteamVR build at all. IPACS supplies a build with the native Oculus runtime that you should be using, which bypasses SteamVR entirely.

    Agreed on the Oculus tray tool, it's very handy for fine-tuning your performance since you can view the GPU headroom real-time. :thumbup:

    I also would help with this as I'm able. I don't have too much free time to invest, but I've got most of this stuff figured out well enough that things like adding custom scenery, terrain maps, and even custom runways with taxiways and such really aren't too difficult. Most of the time expenditure goes towards letting my computer crunch on it while I do other things. :)

    Hi vogel69 , I could be wrong, but I suspect this issue might be a lot less apparent (or go away altogether) if you used your own data to convert level 7 too instead of using the AeroFly mesh. There might be significant elevation differences between the original data IPACS used to generate level 7 and the data you used for your tiles, which is making this problem visible.

    In addition to that though, it does seem like an issue with how it caches the building elevation without updating them like TomSimMuc explained, which is probably the fundamental problem.

    50 Ghz EXTREMLY NICE :) :) how you keep it cool ? :D=O

    It's water cooled. In fact it was delivered at this speed and I haven't messed with it lol.

    Please tell me too where I can get one of these legendary 50 GHz CPUs!!! :D^^;)=O8o

    Oldar Thanks for the nice write-up on how the Pro works with AeroFly and the visual experience in general. I do wish the price wasn't so absurd, but I have a feeling they won't be able to keep it there for too long -- other competition is going to emerge that forces it down eventually.

    Also very interesting that people with these top-spec'd systems are now having framerate issues with the Pro. I've read the same thing from other users with other games/sims too. The 1080Ti is certainly no slouch, and historically each generation of Nvidia's GPUs isn't *that* big of a performance bump relative to the Ti series of their previous gen card. (FWIW a 980Ti is almost exactly equivalent to a 1070 in performance). Also the Ti version of their cards usually comes about a year after the newest gen, meaning we may not see an 1180Ti until... Sept. 2019?!

    I'm really wondering what Nvidia's next gen of graphics card will really be when they finally give out info and release it. Unless it's a revolutionary leap compared to their current offerings, I think we're going to have a GPU-imposed bottleneck on VR for the foreseeable future... until other technologies like foveated rendering can take some of the load off. :/

    I do very much see your point, and if it were an ideal world where they could realistically just supply us with highly-detailed terrain meshes directly from IPACS, I'd certainly be in favor of that. The need for user-customization stems from the fact that it isn't yet practical to be able to supply a very detailed mesh to cover everywhere that anyone might want it, the same way that having maximum resolution photoscenery isn't practical for a entire large areas either. The available DLC add-ons are pretty good, but they still don't come close to what you can make yourself if you are converting a small area and going for maximum quality. I understand I'm probably not the 'typical user,' but it's hard for me to go back to the default or DLC scenery having seen what is possible!


    It would be great if they'd improve their default mesh that installs with the sim, but having tested all of this I can say I still wouldn't be happy if it wasn't as good as what I'm currently using ^^ The improvement in the visuals, appearance of photoscenery (due to it being mapped to a correctly shaped geometry), and possibilities in the kinds of airstrips (e.g. backcountry or unpaved airstrips in mountainous regions) that can be created with a high-res elevation mesh is huge.

    I can understand why they'd want to control it and not allow the custom stuff. But I'm basically pleading here for them to ignore any requests to do so in my own selfish interests, until they can replace the stuff I've made with something just as good :D

    I hope Ipacs can maintain some overall control over this. Right now, in the interests of better mesh, users are beginning to create their own, and I eventually see a situation arising where there is a crazy-quilt of mutually incompatible user meshes all over the Aerofly world partially because Ipacs never made its own move and headed the proliferation off at the pass a little bit.


    It would be a shame (to me) if the Aerofly world splintered into a confusing kaleidoscope of random user and vendor solutions on this like X-plane.

    I definitely see your point, but I disagree somewhat. I'd like to see it remain as-is with the option for those of us who want to do these things still be able to do them. It's very unlikely that they will give us an ultra-high-res topo mesh that covers the whole globe, or even the remote areas I particularly care about. Just the file sizes on that alone would be enormous. So, being able to improve those areas for my own purposes is very important to me, even if I'm not sharing them with anyone else.

    The way they've implemented custom add-ons is very smart and very easily undone. If you want to restore the original sim as-installed, you just remove the files you've added to the folder in Documents.

    For commercial companies selling expansion areas, then I think they should definitely coordinate with IPACS to ensure compatibility and not create 'kaleidoscopes.' But to take away our individual ability to customize things on our own would take away one of the things I like most about AeroFly (and at this point would break 90% of the areas I fly now because it's all custom stuff I've made). I think the method of letting people customize if they want is great (and obviously those people will know they might make things worse instead of better), while those who want everything to be by-the-book via IPACS can stick to the commercial products and add-ons.

    FWIW, I think in a way it already is this 'kaleidoscope' you describe. The whole sim is built around having varying detail levels, and even with the default installation it changes quite a bit all over the map, including the existing topo mesh. In some areas the mesh is converted up to level 10 right now but for most larger areas it only goes to level 7. The same is true of the scenery. The cool thing about it is that you can add files to refine the 'stock' sim, which (for the most part) seamlessly integrates with everything else.

    IMO, all they really need to do regarding the mesh is fine-tune the algorithm that merges edges of refined levels to ensure it doesn't create any invisible wall artifacts at the boundaries. That's the only thing I've seen so far that risks creating broken behavior in the sim.

    We should check if the airport__outside flatten is mild enough to both keep the terrain natural shape while preventing bumpy landings, without having to feature a fully-modeled 3D runway...

    I have tried this, and I think all the combinations of ideas I've seen mentioned in this thread, and I can say it really depends a lot on what the un-smoothed mesh looks like to start with. If you're dealing with significant faceted lumps or spikes, the airport__outside smoothing usually is not enough to correct it on its own and make the surface land-able.

    The airport__runway does preserve some small 'sloppiness' of runways, like elevation changes. I have some sloped backcountry airstrips where it has worked quite well, but it required me first refining the terrain with a better surface mesh.


    By far the most realistic surface I've gotten for unpaved runways has been using a higher-res topo mesh to start with, and then using airport__outside on top of that. This has given me unpaved airstrips that do make the aircraft appear to bump and vibrate but not so much that you're getting kicked into the air repeatedly.


    My request to IPACS to make my FSCloudPort project more effective is a new option like [decal] but without reflections called, say, [layover] so you could use __runway over an entire airport to flatten it then lay the asphalt at 0.1m using [layover] and finally overlay with [decal] at 0.4m.

    I'd still like to see a way to define areas as __runway while still leaving them transparent so the original imagery shows through. (Or create a new class that we don't call __runway but still uses the same flattening strength, or give __outside an adjustable flattening strength). In some cases where things are really lumpy, you need to define the __runway areas beyond even the actual runways and taxiways to get things smooth enough. In this case, you're stuck with remapping textures to get your original imagery to reappear.

    FWIW, you can make the 'decals' look like you want by using Photoshop, and first adjusting the image with Exposure of -2, and then applying Contrast +10. This is what I've used to re-map orthophotos back onto my large __runway areas and it gives an appearance that is almost indistinguishable to the normal scenery. I realize this probably isn't helpful for an automated tool like FSCloudPort, but for anyone trying to brute-force it right now it's one solution that works.

    Also, you may already be aware, but you don't need to adjust height offsets of decals to change their appearance order when converted. Using the __priorityX tag on the end of the decal name determines its stack order in the sim, where X is the priority number.

    I have the Rift and love it. I do not see any screen door. But to look out the window and look down at 30,000 ft. is amazing. I have never flown a sim without VR But for those that do not have VR try to demo it at a store or a friends house. I can get off work and fly miles from home and I feel great even if I land in Zurich. I still get to work on time the next day.

    I feel the same way, also using a Rift. The screen door is definitely there but I only notice it if I'm intentionally focusing on it. To me this generation of vr devices have clear room for improvement, yet they are good enough that I find the overall experience totally enjoyable and very worthwhile. If it weren't for VR, I never would have purchased AFS2, and I have zero interest in flight simming on a 2D screen. Honestly, I hate to admit it here, but I don't really see the long-term appeal of 2D flight simming ^^ I have plenty of RC planes and toys that I'd much rather use than spend that time simming on a 2D screen. But in VR it's that sense of scale, depth, motion, altitude, speed, and really being inside the cockpit that makes it a completely different experience. It blurs the line between pretend-flying vs. actually being able to fly the aircraft from the same 3D visual cues you'd use in real life. I can't land the Pitts to save my soul in 2D but I can grease it in nice and smooth in VR because in 3D I know exactly how far off the runway I am, as just one example. There's also no substitute for flying beside mountains that genuinely appear several thousand feet tall!

    Also if you'd like, I can see about getting you an improved terrain mesh for that area, which we can probably freely distribute since the USGS data is open to the public. The thing to be careful of though is the short 'invisible wall' that it creates where the refined mesh merges with the default mesh, so we'll have to make sure the refined mesh extends reasonably beyond the airport where nobody will be colliding with the invisible wall on approach.