Hello there!

Well, this has been a crazy couple of weeks. X-Plane 12.00 is now available for Early Access – in other words, everyone can get X-Plane 12. Over the next few weeks we will post more about ongoing development and get into some of the new features in depth – there’s a ton to talk about in X-Plane 12. For today, here are just a few notes on some issues that have come up over the last few days.

Early Access

X-Plane 12 has been in a private alpha test program with third parties since December (!) – almost nine months. During that time we built 38 (!) official alpha builds, recut the global scenery five times, and committed over 4000 checkins to X-Plane’s source code (plus more to the aircraft, scenery tools and art libraries). The alpha program included completion of major features, lots of debugging, and changing the product in response to early alpha feedback.

So why Early Access now? Not because X-Plane 12.0 is done – we still have over two hundred open bugs and a lot of things we want to do. X-Plane 12 is in Early Access so that the entire X-Plane community can be involved in X-Plane 12’s growth, not just a limited number of testers.

With X-Plane 12.0 in early access, we don’t have to say ‘no’ to users and devs who want to get started with 12, and third parties can get their entire teams using the new sim and run their own test programs.

(We can also finally open up our developer relations program to a wider audience.)

Major Areas of Work

Here are some of the major areas of work ahead of us:

  • Clouds – we are working on the shaping and quality of clouds, improving resolution, fixing artifacts, and improving performance. Clouds are probably the single most expensive part of the renderer, so they are a constant tug-of-war between quality and speed.
  • Lighting – there are quite a few lighting and atmospheric scattering bugs that affect the sim, as well as work to do improving auto-exposure and tone mapping.
  • Philipp is working on an airbus MCDU, which we expect to ship during early access.
  • Third party interfaces – we have a few new SDK and authoring features that are mostly completed that will ship during early access. The elephant in the room is third party access to the weather system.

That’s One Blurry Airbus

X-Plane 12 moves some work that used to be on the CPU to the GPU (looking at you, ocean waves!), and virtually all new computing work in X-Plane 12 is on the GPU. When we discussed this before Early Access, there was a lot of teeth gnashing. “You’re gonna use more GPU power? I can’t buy a 3080, I’d have to sell my kidney!”

We still have a lot of GPU optimization left to do, but we also spent some time before beta 1 working on performance, particularly at intermediate settings. User with high end hardware have been pleasantly surprised to see production-level FPS in beta 1, and a common request is “I have 60 fps and blurry clouds, can I get a higher max setting.”

(I do suspect there is a huge gulf between the haves and have-nots for GPU power – because there’s a huge range of hardware performance amongst our users. We will keep optimizing.)

What we didn’t optimize was VRAM use, and this is why blurry textures is a common problem with the first beta. X-Plane 12 uses Vulkan/Metal as its renderer, always, so it uses our Vulkan/Metal memory management strategy: we dynamically bring the resolution of textures down to fit within your available VRAM, with some guessing as to which textures are most important.

The texture slider in the UI sets the maximum texture resolution X-Plane will attempt – if you have a card without a lot of VRAM, setting this lower can help avoid “thrash” as X-Plane tries to fit 4 GB of textures into 2 GB of VRAM. But X-Plane will further lower res until it fits – X-Plane will not use system memory as backup texture memory, nor will it slow the framerate and stutter by shuffling textures between vRAM and system memory on the fly.

I’m afraid I don’t have any useful information about how much VRAM will get you a better experience – we’re going to do an optimization pass and see what we can tighten up.

I suspect the big driver of VRAM is memory used for effects – X-Plane 12 has HDR always on, but also has extra VRAM reserved for screen space reflections, 3-d water, dynamic weather effects, clouds, etc.

One thing that can help (and I know no one wants to hear this) is to run at a lower resolution. The sim has to internally use VRAM proportional to the size of the winddow or monitor res you fly at. Jumping from 1080p to 4K doubles the resolution in each dimension (making each pixel half as big) but uses 4x the VRAM for surfaces. Full screen anti-aliasing increases VRAM by its factor (4x MSAA = 4x VRAM) for some of those textures, so it’s more efficient than higher res.

What’s All This Magenta

X-Plane renders magenta when it hits a numeric error (a NaN value) inside the rendering engine. Right now there are multiple causes of NaNs – it’s not all one bug because magenta is a symptom, not a cause. A few we know about:

  • We believe there’s some kind of problem specific to the GeForce 900 series. Sidney bought one on eBay so we can debug this.
  • I’ve seen NaNs caused by the traffic debugging lines for ATC – I suspect that particular shader has a bug.
  • We can sometimes get NaNs from the past frame – they get “reflected” by SSR and propagate from one part of the frame to the other.

There’s no easy answer here – each bug has to be squashed one by one. These are high priority bugs and we’re working on them now – hopefully each fix will make things a bit better, but don’t be surprised if some users see less magenta in the next beta and others do not.

Fuzzy Scuzzy Rendering

FSR stands for FidelityFX™ Super Resolution. FSR is AMD’s free open source up-scaling technology. The idea of up-scalers is:

  • Lots of people have 4K monitors.
  • Not as many people have GPUs that can run games and simulators at 4K – they’re expensive.
  • Upscaling a 2K image with a little bit of smarts uses a little bit of GPU and looks a lot better than just running the monitor at low resolution.

When you move the FSR slider to the left, X-Plane renders its 3-d image at a lower resolution and then upscales it to the monitor. This saves GPU time and VRAM at a cost of image quality. The image should look better than running at low resolution but not as good as running at high resolution.

Should you use FSR? I would only recommend using FSR if you want to/need to run at 4K and your GPU is struggling. Support has had a number of complaints about blurry rendering from users with FSR on – FSR is resulting in a less detailed image on purpose just like reducing resolution does. If you are going to use FSR, use full screen anti-aliasing – it helps.

We are still undecided about the future of FSR in the simulator. We added the option of up-scaling based on user requests, and if we didn’t ask for it, we’d probably be asked for it. But we’ve also had lots of “I set this slider low and now everything looks terrible.”

(Why don’t we use FSR2 or DLSS? Both of these upscalers require motion vectors as inputs from the rendering engine, something X-Plane does not provide. We may support them in the future, but adding motion vector generation is not trivial.)

Beta 3 Coming Soon

Over the past weekend part of the team met in person to do planning and roadmapping; beta 3 should be available shortly, with some of the bug fixes we’ve already coded. X-Plane 12 for Steam is in review — hopefully it will be available Real Soon Now™.

About Ben Supnik

Ben is a software engineer who works on X-Plane; he spends most of his days drinking coffee and swearing at the computer -- sometimes at the same time.

149 comments on “X-Plane 12 Early Access Is Here

  1. Any timeframe to have public developers tools, such as dataref editor, wed 2.5 and other tools to the general public?

  2. Operating System: Ubuntu 20.04 (Linux)
    Kernel: 5.4.0-125-lowlatency
    CPU: Intel(R) Core(TM) i7-7700K CPU @ 4.20GHz
    GPU: NVIDIA GeForce RTX 3080
    System Memory: 32 GB, (4 x 8GB), DDR4 SDRAM
    VR Headset: Valve Index @ 120 Hz, Custom Resolution Multiplier 164%
    Flatscreen: Predator widescreen, 3440 x 1440
    Configuration: “X-Plane-x86_64” and “vrmonitor.sh” (i.e. SteamVR) run with “/usr/bin/gamemoderun” for optimal system performance

    Test Scenario:

    Cessna 172 sitting on runway 20 at Portland Hillsboro with no third-party plugins.

    Summary:

    X-Plane 12.00b2 in VR mode uses ~15% of CPU cores and 59% of GPU cores for 17 fps.
    X-Plane 12.00b2 in Flatscreen mode uses ~18% of CPU cores and 48% of GPU cores for 53 fps.

    X-Plane 11.55r2 in VR mode uses ~15% of CPU cores and 56% of GPU cores for 40 fps.
    X-Plane 11.55r2 in Flatscreen mode uses ~21% of CPU cores and 43% of GPU cores for 103 fps.

    X-Plane 11.55r2 fps is 235% of 12.00b2 in VR mode.
    X-Plane 11.55r2 fps is 194% of 12.00b2 in flatscreen.

    Conclusions:

    CPU is less than 20% utilization, which means CPU cores are 80% idle.
    GPU is at 50%-60% utilization, which means GPU cores are 40% to 60% idle.
    VR performance is significantly less than flatscreen.

    My entire report was sent in as a bug report entitled “Linux Performance Woes, VR and Flatscreen” on 9/8/2022. This is just the summary.

  3. I love those updates with brand new X-Plane 12, I see the simulator reaching towards it’s glorious days! There is a question that big majority of community wonders as I see them debating around ”will graphics keep improving aside of clouds and lightning?”

      1. Yes and compare to X-Plane 11, the FPS drops drastically around 15 FPS in VR. It will be very important to optimize frame rate too in VR on X-Plane 12.

  4. Congratulation on releasing xp12.
    Some of the new rain effects are awesome and flight model feels ok but…… this still feels and looks like XP11.

  5. Hi Ben,
    Thanks for providing us with great detail on the progress of XP12. Any updates to VR? particularly as it pertains to WMR(windows Mixed Reality). It appears that frames tank immediately as soon as WMR is launched even when still in 2D mode.

    Thanks

    1. That would be worth a bug report, although it might take us a while to get to it – people in the company have WMR headsets and haven’t seen this.

  6. Hello Ben,

    I would like to mention a few other shortcomings that I noticed by evaluating the demo, all of which have already reported by me as a separate bug report:

    1. Shadows in version 12 are once again only rendered at very close ranges. This was also the situation in the previous version 11. Trees and other scenery objects throwing shadows in close, mid and far distances add a lot of realism and increase the visual fidelity tenfold. For X-Plane 11, there was a third party tech nerd, who programmed ASHA, a FlywithLua plugin, that would create shadow maps of all scenery objects and terrain for up to 10 kilometers of distance as a brute force method. Unfortunately it wasn’t continued and with current versions of 11.55 it flickers too much. Thing is, it was “easily” implemented and framerates were affected, but not so much (maybe 10%). It would be so wonderful if version 12 could include a proper shadow implementation, maybe with a distance slider and quality slider and that shadows can be rendered in close, mid and far distances if one wishes so.

    2. Missing cloud reflections on bodies of water. At this moment, only water puddles do show cloud reflections, though very lowres. I hope that you plan to introduce good quality cloud reflections once more for all bodies of water.

    3. 3D trees transition at very close ranges to their 2d counterparts, which look odd. Also with stronger winds which shake the 3d trees but not the 2d versions, the result is not very good. Additionally 3d trees do look much less bushy than their 2d versions. And missing shadows add to a not optimal visual experience.

    These are my top three concerns with the visual part of the current X-Plane 12 beta.

    Greetings, Ralf Maeder.

    1. The only thing ASHA did was change a few of X-Plane’s datarefs. With a little bit of Lua knowledge (or just using DatarefTool lol) you could edit these same datarefs. I’m considering uploading a simple script that does basically the same thing ASHA did.

      1. Sorry to say that, but this is 100% not true. The ASHA Lua script does a lot of magic to settings files within the x-plane directory, which are most likely designed as closed source from LR.

        The script stopped working at a given update release of XP11(think it was 11.50+). I tried to fix it, as I loved it, but was not trivial at all, or maybe not even possible, as maybe LR removed some DataRefs.

        With this script, 10km+ Shadows were even usable on my midrange PC with nearly no fps drop, as regular shadows dropped my fps by 15 or so.

        Right now, the short range of shadows is one of the biggest immersion killer for me, as it is – as written above – very important for view depth perception.

        1. I agree. It’s not as simple as changing datarefs because I have done that and it slaughters the fps. I really hope someone can pick up that mod and update it for Xplane 12. Missing shadows are a huge immersion breaker for me.

        2. Looking at it again, it does seem to more than setting datarefs.
          It looks like it mostly reads the X-Plane settings though. The only case it seems to write to the settings is if you’re using the aurora borealis or fires option. Otherwise, I think it mostly reads X-Plane settings, does a bunch of complicated math, and in the end sets a few Datarefs. And then there’s extra stuff for walk mode, better shadows, better heat effects, target fps, etc.

          So yeah it is a lot more sophisticated than just changing datarefs but it doesn’t seem to actually write to the X-Plane settings that much. Please correct me if I’m wrong. Seeing how complex it is though, it’s not a surprise that it doesn’t work anymore in X-Plane 12.

          I’ll try to look into what I can do to at least get close to what ASHA did. Hopefully the developer updates it!

        3. Correction to my previous post: Ok yeah so ASHA does actually do some writing to the settings.txt like I mentioned for the aurora borealis and fires options (which seem to be removed from my quick search through the settings.txt), but it also does so for the shadow resolution, but it’s pretty simple and doesn’t really do any calculations or anything (specifically for shadow res), it just changes a few numbers in the settings.txt.

          I _could_ make a script that does that but if you open settings.txt you’d see the huge warning at the top and it’s not a good idea to make a script that modifies it… it’s a fairly simple manual modification anyway: //www.youtube.com/watch?v=l61wL_4wEC0

          But that’s slightly off-topic anyway since that’s about shadow res, not distance. Like I said I’ll try to do something similar to ASHA sometime soon, just don’t have a ton of time at the moment.

          1. In the meanwhile, I also digged a little bit deaper and found an older post in the X-Plane org forum, where somebody edited a setting file of X-Plane 11 in order to increase shadow distance. I already replicated this and it works with X-Plane 11.55, now without the need of ASHA. I also found a solution for the ugly powerline shadows, I simply edited the corresponding PNG image and made its contents completely transparent. Problem solved. Concerning to X-Plane 12, these changes to shadow distance work in principle, but with a lot of terrain artifacts, default 2d trees don’t throw shadows and overall performance decreases to a slideshow. Maybe this might be beneficial once version 12 has more matured.

      2. Tim, if you could provide a script for X-Plane 11/12 that creates shadows similar what ASHA did accomplish but without the problem of flicker and shadows created by powerlines, that would be awesome. I agree with Michael C. about the importance for visual depth input and that shadows have proven to be totally viable without causing too much trouble for the performance. The fact that Laminar did include once again a similar close range solution to version 11 into the new version 12 is somewhat disappointing. I would have been better to implement a native solution instead of relying on third party inventiveness.

          1. There are some things that can and have been easily provided by third party providers, like ATC for example (I have PF3 from Oncourse). Core features of the rendering engine on the other hand are not easily added by third party and I would count under these shadows and reflections.

  7. Thank you guys, job well done and cheering you guy’s on, going forward.
    Will new features arrive with a “beta run” or will there be an official X-Plane 12.00 release which include such things as “weather radar”, ability for third party vendors to access the weather system. Last but at least, where is Austins new Weather program he he…

  8. Hi Ben. Thanks for the write up. One thing I want to request is please don’t change the lighting too much as it is almost in perfect shape and is the single most impressive thing about version 12. Same goes for atmospheric scattering effects and HDR tone mapping. Auto exposure could use some work.
    I am glad performance fixes are coming and hoping to see better cloud tech. I also hope dlss comes soon.

  9. Thanks to all for doing the needful. I know Beta programs are difficult… not only in bug bashing but in prioritising and understanding what’s important to your audience.

    Looking forward to b3!

  10. Thanks for the update Ben. Great job everyone. Enjoying it so far.
    However, the VR experience seems to have regressed. I’m sure it’s on the TODO list but I just wondered if it’s seen as a priority.
    For those of us without cockpit hardware and multiple screens, VR is the best way of using the simulator.

      1. So what about Godfrey’s first question? VR works, but performance is lacking at this point. I’m getting 17 fps in XP12 in places where XP11 gave me a solid 40 fps, and I’m on a RX 3080 with a Core i7. Not the fastest, sure, but no slouch of a system. Any plans to improve VR performance?

          1. Have done so. Please look for bug report entitled “Linux Performance Woes, VR and Flatscreen” on 9/8/2022

      2. Anything on how that will be handled in the future? Sid was working a lot on Zink. Will that eventually ship in early access?

        1. Zink in its own VK context didn’t solve interop problems, so Sidney had to temporarily put it down to work on features. Not sure what the future is – running zink _in_ our context might be a lot more hairy.

      1. If “RSN” becomes much later due to more pressing concerns (i.e. you are swamped), re-compiling and re-releasing the 3.0.3 SDK with arm64 support turned on would at least allow current plugin developers to support M1 Macs without telling users to resort to using Rosetta and losing half their performance. Sounds like something a junior team developer could do really quick (assuming you have any of those).

    1. I believe _all_ exclusion zones from past gateway airports have been removed and need to be put back. This could happen by taking the past gateway ones _if_ they are still valid.

      Our thinking was this: a lot of the exclusion zones no longer make sense due to changes in the road grid alignment, autogen, or including a border that was previously not present in the DSF. If we err on the side of keeping exclusion zones, we end up with no autogen and no obvious way to know the bug is in the airport and not the AG itself.

      If we remove the v11 exclusion zones, the bug is clearly diagnosable – an exclusion zone is needed, and they can be restored.

      I don’t think this strategy is set in stone, and I expect we’ll get some gateway guidance posted shortly.

  11. Regarding the magneta ,I have experienced this while on the water in bad weather , i have rtx 3080 it is not just limited to 900 series . Is brake fade simulation broken/ bugged at the moment because I can not seem to get my brakes to overheats

  12. I am really loving the xplane 12 beta , despite the bugs , glitches ( which of course is expected ) the lighting and atmosphere is really great , it is extremely immersive and incredible when you get stuck in a storm cell , Daniella has done a phernomal job on Audio . I think some people do not realise how much work and how complex it is redo the rendering pipeline , photometric lighting ect

  13. Thanks for the explanation, Ben! I´d say that it is behaving extremely well for an early access beta-ish-almost-alpha release, and it is looking great to me! I do have one question. I remember that at some point it was mentioned that the ground textures would be upgraded, and I seem to remember a mention for a team in Europe working on that. Is that already implemented, or might we see it released/improved in a future beta? I´m asking because (to my untrained eye) the ground textures on XP12 look very similar to XP11 (haven´t had the chance to play…err… I mean, sim a lot), but I thought I would ask.

    Also, regarding Anti alias, I have tested it at 4K (both native and upscaled), and even at max AAn settings (where the sim becomes a powerpoint presentation on my PC haha) I do see serrated edges. Is this being considered to improve on future releases?

    Thanks!!

    1. There was an effort to color correct the ground textures, but 12.0 is not a “redo the ground terrain system” release.

      Aliasing artifacts with AA are aliasing in the _lighting model_ – it’s on my radar.

      1. Thanks for the explanation!! Great to hear from you guys again, amazing work on XP12, thanks!

  14. Thanks for the update. Please keep the FSR feature. Going from no FSR to the “Quality” FSR setting increased my frame-rate by 64% while only minimally reducing the visualy quality (which is to some extent subjective, of course). It’s almost like magic. This allowed me to crank up other settings, like world detail, thus increasing both FPS and overall visual quality. I wish you the best of luck for the next X-Plane updates and look forward to testing them.

    1. PS: AMD has presented FSR 2.1 last week, which aims to further increase image quality and reduce artefacts such as ghosting and shimmering. It would be great if you could integrate it into X-Plane soon.

        1. Thanks for your reply. Yes, I read that in your blog post and understand that you have to prioritise. Nevertheless, these upscaling technologies seem to keep improving in terms of image quality and performance, so adding motion vectors might be a worthwhile investment of development time.

  15. Hi Ben, as a pilot in real life of some of the default aircraft models that are shipped with X-Plane 12, I was thinking about the opportunity to give some feedback on the fidelity of the flight dynamics, taking advantage of my experience on the real thing. Is this something that might be done through the bug reporting web form? And above all, is a kind of feedback that Laminars would welcome? Thanks and keep up the good work!

      1. Please remember: LR doesn’t cull the .org for for bug reports. It’s totally fine to use it or any other forum as an organizing tool! But someene’s gotta make the report. You’d be surprised how often everyone thinks someone else filed the bug.

  16. Thanks, Ben. I know you have read or been told about the many compliments given about the early access. Really well done. We look forward to the future of Xplane 12.

  17. Thanks for letting us know about the beta status.

    Personally, I’m really exited with XP12.

    I prefer the performance optimization over the cloud fix, clouds are nice to me but the truth is, I don’t like the cloud artifacts either.

    As for FSR, I have an AMD 5700XT eGPU on an Intel MacMini, and thick clouds are a real FPS killer, so I thought FSR might help but if I put a lower setting for FSR the graphics is blurry and pixelated. I have a 4k monitor. So how should I set it up for FSR? Have a native 4k resolution on the monitor (that is, set in macOS) and set a smaller one in the simulator, for example fullhd? Thanks for any clarification.

    I look forward to the next beta updates.

    Thanks for the great work LR!

  18. Thank you so much for the update, this has answered a lot of questions I have about the beta. There is one lingering issue I’ve been experiencing that I haven’t seen addressed in the forums, however.

    Regarding the lighting, when zoomed in everything looks beautiful, minus some of my 3rd party sceneries that still may need updated to XP12, as their lighting seems overly bright and washes out a lot of the surrounding scenery. When I zoom out however, the default light sources (Street lights, apron lighting, PAPIs, nav lights on aircraft, ground vehicle lights, etc) they appear a little blurry and kind of blocky looking. I’m just wondering if this is simply a symptom of the early access beta and other users have experienced this as well, or perhaps maybe my system isn’t quite up to the task. This occurs on all slider settings as well, including maximum.

    4.3 GhZ
    GTX1070 8gb
    16Gb RAM
    SSD

  19. Can you please add back SSAA to the sim as an option? Those of us with 3090s have the horsepower for that and we would like to use it. MSAA doesn’t cut it at 1080p.

  20. Have you thought about versioning the flightmodel? That way we could simply enjoy the new engine of XP12 with the well known and tested FMs for existing planes.
    This may also give Austin more room to play around with experimental versions and would allow gradual improvements over the 12 run.

    Cheers and keep up the awesome work

    1. We have, but it’s a structurally difficult thing to do – the FM interacts with systems. I suspect third parties would be annoyed to not be able to get other features because they’re locked on the old FM. The experimental FM run in 11 was sort of a two-stage versioning process.

  21. Thanks for the update.
    My personal view is that this Early Access is purely for developers & not for end-users such as me.
    So, as a pure end-user, I won’t be spending any time or money on XP12 any time soon. I think that there’s at least another 3 months of effort required before it is ready.

  22. X-Plane renders magenta when it hits a numeric error (a NaN value) inside the rendering engine. Right now there are multiple causes of NaNs – it’s not all one bug because magenta is a symptom, not a cause. A few we know about:

    We believe there’s some kind of problem specific to the GeForce 900 series. Sidney bought one on eBay so we can debug this.

    Operating System
    Windows 11 Pro Home 64-bit
    CPU :
    Intel Core i7 10700K @ 4.20GHz.
    RAM :
    32.0GB
    MOTHERBOARD :
    Gigabyte B460MD3H
    BIOS : F5b
    GRAPHICS :
    NVIDIA GeForce GTX 960 4GIG OC
    STORAGE :
    512GB OS Drive NVME SX60000PNP
    512GB NVME SX6000PNP
    128GB SSD SP600

    1. I am running an AMD GPU and seeing a lot of NaNs (magenta rendering). It can be initiated a lot of times by switching to the external view, and rotating. Would it be beneficial for me to do some tests and post a detailed bug report?

      Asus ROG STRIX B550-F WIFI Motherboard
      AMD Ryzen 5 5600x CPU
      TEAMGROUP T-Force Delta 32GB DDR4
      Samsung 980 PRo 1 TB M.2 SSD
      PowerColor RX 6700 XT 12GB Hellhound GPU
      Corsair RMx 750W 80+ Power Supply
      Windows 11
      SAMSUNG Odyssey G7 Series 27-Inch Gaming Monitor

  23. Hi Ben. First of all pay yourself on the back for creating such an insanely good lighting engine. The photometric hdr renderer is magic. Although I have two issues to report
    1) Shadows are only cast for very near objects. This breaks immersion immensely.
    2) There seems to be no light pollution at night ? The atmosphere is pitch black at night. Ground lights should light up the atmosphere a bit if there is humidity present.
    Feature request
    -Visible rain bands please.

  24. Will the XP12 demo be receiving updates during the beta run? I’m holding off buying due to performance and freezing issues in VR (Reverb G2) with the demo (already reported via bug submission).

    I just want to see the same or similar performance in VR that I have with XP11, except with all the cool new stuff you guys have built for 12 🙂

  25. I am a very disappointed customer using high-end hardware: Intel 12900K (OC’ed to 5.4 Ghz on 3 P cores, 5.3 Ghz on the other P cores. E cores disabled) + Nvidia RTX 3090 with 24 GB Vram) and triple 4k screens. In XP11, I am able to adjust the settings to easily run at 40 to 60 fps with dense scenery AND no stutters. In default XP12, I have a hard time reaching 20 fps AND I get constant stutters; the only way for me to increase fps to about 25 is to move the FSR slider to the left (remove that option and I believe you will have lost any user with multiple 4K screens) but still get stutters.

    I have spent some time trying to both increase performance and keep the CPU and GPU times in balance, without success: having moved a lot of the load to the GPU, it is still impossible to get my top-of-the-line CPU to handle the remaining load in XP12 (I suspect that real weather – i.e. clouds – is to blame, but it is not possible to fly online without it).

    I find it hard to believe that in 9 months of alpha no one tried multiple screens at high resolution. You may not want to hear this, but MSFS2020 now supports multiple screens (in a beta release) and I have no trouble running it at ULTRA settings.

    I am prepared to file bug reports about both the low fps and the stutters, what would be helpful to include besides a copy of the text above and the log.txt?

    1. Probably the most useful info would be the relationship between screen count and perf. If the effect of adding screens isn’t strictly linear that would imply something is really wrong.

      With that in mind, 4 screens on 1 computer is asking a lot of X-Plane no matter what hardware because X-Plane (11 and 12) doesn’t scale to multiple screens with good hardware usage. This was less of an issue with 11 because it does a lot less.

      1. Not sure what the “it” refers to in “it does a lot less,” but in XP12, once I move the FSR slider one notch to the left, the CPU becomes the bottleneck (with the FSR on OFF, all the way to the right, the GPU is the one with the higher times, no matter what I do with other settings). I will compare performance with 1, 2 and 3 screens (I do not have 4) and file a bug report.

        1. If the CPU is the bottleneck, it’d be useful to have a matched pair of microprofiler dumps for 11 and 12 – it’s a little surprising that 12 would be eating more CPU time.

          1. Wilco. I assume getting those dumps is obvious within one of the XP menus, if not please tell me how.

          2. Bug report (titled XP12 low fps and stutters, unlike XP11, using high-end CPU and GPU) filed with microprofiler dumps.

          3. I wonder what type of cooling multiple cores running at 5 GHz need. Are you sure they aren’t throttled due to overheating?

      2. As reported below, I filed a bug report comparing XP11 and XP12 with 3 screens and similar graphics settings (medium for CPU, very high for GPU) but did not include relationship between screen count and performance. I just did a comparison using the same graphics settings, with default scenery only and minimal plugins, sitting on the ramp at KBFI, weather CAVOK/clear.

        XP11 3-screen: 65 fps; 2-screen: 90 fps; 1-screen: 133 fps
        XP12 3-screen: 24 fps ; 2-screen: 33 fps; 1-screen: 56 fps

        In all cases the CPU is the bottleneck. All I can say is that there is something VERY wrong with the way XP12 uses the CPU (and I have the top of the line Intel running at 5.4/5.3 GHz).

  26. CLOUDS: At altitude, the aliasing/cloud dancing/pixelated look is quite obvious and probably my biggest issue with XP12.

    GFX SETTINGS: I wish there was more granular control restored in settings to allow people the option to dial things back and save a few fps. For example flying an airliner, I’d like to be able to dial down trees, vegetation, ground traffic individually. Flying a piper cub, I’d want all the trees/vegetation set up to max. In addition, to allow a greater user base, allowing users with mid to older systems/laptops would have the opportunity to play, albeit lower settings but enough to have fluid movement.

    PUSHBACK: Only works when requesting straight back. As a hack, I’ve been using tiller during push to get some limited movement to left or right. Is there a way to force the tug to continue to push until we push the stop button? Half the time it doesn’t even clear the jetway centerline.

    AIRPORT LIGHTING: **AMAZING** The runway lights, the ramp/apron areas are well lit, yet when you leave the area on the taxiways it becomes much darker as it should. Really well done.

    DEFAULT RUNWAYS: Tire marks, intensity, dirt, cracks, ect. Scored an “A”. Way better than the other sim.

    PUDDLE EFFECTS: Again, another amazing addition. Compared with XP11 default, XP12 beta is leaps and bounds better, and will continue to improve as time goes on.

    GROUND SERVICES: After pushing the request button, there is no confirmation that it was activated or the services are enroute. Would be nice to have a flashing button or a (stop) button appear below to indicate the button press is active.

    Boeing 737-800 “ZIBO MOD”: Thanks for allowing 3rd party devs access to the base model for improvement. I hope Zibo pilots have “report anonymous data” checked because the user base for this aircraft is Huge and is likely a significant percentage of the most used aircraft in X Plane.

    Thanks for all the hard work you guys do and thank Austin for being so passionate about aviation and flight sims! FYI- Flight Sim Expo 2023 is in Houston, TX next June.

    1. > CLOUDS: At altitude, the aliasing/cloud dancing/pixelated look is quite obvious and probably my biggest issue with XP12.

      I don’t think this is only at altitude. It’s the first thing I noticed when I spawned on the ground for the first time in XP12.

  27. Thank you for the updates.

    I have a GTX 980ti, and I encounter magninta 70% of the times. Sometimes to the point where the whole screen is magenta.

    Its triggered mostly when I change the weather (specifically going from no clouds to clouds)

    Is there any test methodologies we should follow to better file a bug report?

  28. Since X-plane 12 is all GPU, wouldn’t that put macs at a real disadvantage?
    What’s the future for Macs on M chips?

    1. It certainly hurts that you can’t get a 3090 on OS X. I think it’s too soon to tell on OS X – our renderer isn’t architected for tiling GPUs, so we see the pain from that, which hides the true capabilities of the chip from us. E.g. my M1 laptop will do all the autogen at 60 fps but struggles with clouds.

      1. FWIW my MacBook Pro M1 Max 64GB runs smooth both autogen AND clouds @ 40 fps on dense (almost all settings at max). I suspect the M1 Max plus the 64GB helps a ton. This on a non-Roseta emulation environment. Enabling Rosetta (for plugins) eats some around 10/15 fps easily. I just wish plug-ins can be used without Rosetta in a future…..

  29. Much good work going on here and I’m proud to be a user of this wonderful development. My only request is for the Citation X. This is a wonderful aircraft so far but it has an incessant hunt left and right at cruise. Can we make it simply cruise straight and true??? Thanks for all you do.

  30. Hi,

    Great job on XP12! I only have the demo for now, but it is looking great!

    I read the other comments on 4K displays, and wanted to add my voice as another 4K user.

    I have modest hardware by today’s standards (i7 9700K @ 5 GHz, GeForce 1080 (not Ti), and 16 Gb DDR4 3466 MHz memory).

    I see practically no difference in frame rate between 1920×1080 and 3840×2160. This is quite a surprise, and unsure if you’re aware of this?

    With all sliders at minimum I see 60 FPS. With FSR set to OFF and most sliders maxed (where sensible – some settings I can’t see a difference) I get 30-45.

    Yes, the latest GPUs today are super-expensive, but please don’t think we are not running 4K out here – we most certainly are.

  31. On the Mac build, specifically the M1/M2 native build will you be further optimising or implementing metal 3 upscaling for those on MacOS Ventura?
    On my M1 Pro with 32GB ram performance is barely playable

  32. I noticed some plugins like AVITAB for Xplane 12 pop out window works sometimes when you click on the item in the plugin list. for me it does not pop out in full screen mode at all but does in window mode . It sometimes stop working in window mode as well. Same thing for Xchecklist (although this is from XP11 so that may be a version issue). Please advise what info should i send to you to help .
    System

    AMD RX 5500 XT 22.8.2
    Ryzen 5 3600 6 core
    16 GB 2400 MHZ DDR4 Memory
    Default scenery

  33. Thank you! Thank you! thank you!!!
    XP12 is such a step forward in VR, compared to other sims, there is no better immersion than with XP, the interaction is still as it was in XP11 which is quite perfectly done, I’m able to fly as I would in real world flying! happy with that, of course much work i suppose is ongoing to polish the graphics, shadow and clouds mainly!
    ..I’m having very good performance in VR with RTX3080 and i7-12700, I’d like to know if I squeeze in a few more FPS (to cap at 45 instead of 30 as doing now) by moving the fsr slider, I have no interest in rendering high res image in my flat screen, would that make any sense? or reduce also the VR headset resolution?

  34. Would you mind clarifying what should be reported as a bug?
    The new lighting is great and very realistic, but the realism creates a problem.
    I find the light to be too bright in VR using the Quest 2. In XP11, the Shift-S virtual sunglasses helped. In XP12, the virtual sunglasses don’t do much. I’d like to know if there’s a way of increasing the virtual tint on the virtual sunglasses. Should I report this as a bug?

    1. Try the sun visor on aircraft that have them. I was flying into the sunset in the Lancair Evolution last night and the sun visor really helped my view in VR.

  35. Ben, and Team X-plane.
    I want to thank you and all the Airport Scenery Developers, All the X-Plane Payware and Freeware Developers that give their precious time for all of us so that we can enjoy at least the illusion of flight.
    For those of us that can no longer fly in real life whether it’s a Cessna 172 or an airliner and use X-plane because the flight model feels real. That’s a big deal.
    It’s the reason I use X-Plane as my only flight Simulator.
    Keep up the great work.

  36. Regarding Plane Maker:

    Why has the ability to precisely manipulate the xy coordinates of the nodes of aircraft bodies and nacelles been removed from PlaneMaker 12?

    This doesn’t appear to me to be a net benefit.

  37. Glad it’s finally in early access, been enjoying it a lot so far! Happy to hear that your main points on your roadmap are clouds and lighting, hope this also includes the (most of the time) too dark cockpit.
    In addition I am hoping for Shadows ( especially shadow distance) and Anti-aliasing (even on highest setting no real crisp and sharp edges) can be improved in the future!

  38. Hello Ben and congratulations to you and the LR team.

    Like many others, I didn’t have the biggest “WOW” reaction the first time I ran XP12, but then it grew on me day after day, and now it’s a continuous “WOW” every time I run it. You did an incredible job.

    My observation relates to the lack of the well known “blue hour”. In real life, reddish sunsets/sunrises are usually followed/preceded by about 30 minutes of a pure bluish light. In all flight sims instead, there is this dramatic reddish sunset light that is directly followed by almost total darkness.

    I read this “blue hour” is dominated by something called “Chappuis absorption”, distinct from Rayleigh scattering and prevalent at dawn and dusk. Maybe you could consider implementing it, to complement this new FANTASTIC photometric engine.

    1. We have an ozone absorbtion model in the current scattering – it helps make the top of the sky a deeper blue. I think the main problem with the blue hour is that the assumptions and simplifications of the scattering model that let it run on real time break down in that lighting environment (where there’s no direct light and therefore all light is secondarily scattered). I have a TODO to look into this more, but I don’t know if we’ll find dumb bugs or fundamental limitations.

      1. Thanks. (Not OP here, but noticed the same thing.) Its ultimately not a huge deal but if its an easy enough of a fix, a much appreciated one!

      2. For Android there is a “suntimes” app that says holden hour starts when the sun is less than 6° above horizon, and blue hour is 4 to 8° below horizon. Just changing the color of the dun would be cheating of course, but maybe it’s worth trying

        BTW: A meta-bug: Whenever I write a comment here, the browser scrolls to the start, making me scroll like mad; maybe that can be fixed, too.

  39. Ben,

    Just wondering where one makes suggestions about improving “X-Plane Landmarks”

    IMHO “X-Plane Landmarks – Las Vegas” is missing a key land mark that the whole South Western USA (as we know it would not exist if it were not built),… basically just suggesting LR add the Hoover dam and Mike O’Callaghan–Pat Tillman Memorial Bridge (since it is close to Vegas, scenic flights to see the landmark are flown from near by KBVU (Boulder City Municipal Airport) and the Hoover dam provides water for 40+ million people who live in “Lost Wages,” Los Angeles, San Diego, Phoenix, etc.

    //en.wikipedia.org/wiki/Hoover_Dam

    //en.wikipedia.org/wiki/Mike_O%27Callaghan–Pat_Tillman_Memorial_Bridge

    Also have to say really like the new and improved default airport runway graphics,… BUT with the new an updated airports “boundaries” along with new forest scene generation, is there anything that can be does to get a virtual maintenance crew to trim the trees that block the approach end of runways (for example trees blocking runway 32 @ 2NC0)

    //ibb.co/xGgnVsm

    1. Well, I think if you CAN add the landmark, then it’s not a bug; it just has to be done.
      Maybe offer some $$$ to the community to fo it!
      While it may look important to you, it’s still just a tiny spot on earth, like millions of others.

  40. Ben and LR team, any plans to make the reflect sunlight from the moon the lit the sky too?

      1. What about night lighting? Will be some kind of night cloud reflection? (I guess not light by light, but maybe something related to urban areas were light pollution is usually more frequent?).

        Also, will lights visibility range for far lights be changed? (We always have the ENL addon, but it’s always better if implemented by default 🙂 )

        Thank you for XP12.00. Really excited about what the future can bring us!

        1. It would be interesting to see what type of “map” for light pollution will be used, because light pollution play a interesting role on night flight, specially if you coming from the middle of the ocean and start to see the city glow in the horizon

  41. All the years you were supporting TrackIR, that was great – but it is a little bit old. It would be great to see support for Tobii Eye Tracking. Thank you very much.

  42. LR said the lighting engine has been redone but the cockpit is still very dark during dayflights, it doesn’t seem to be interacting with the external lighting.

    Despite this problem, the terrain/terrain textures, the performance that could be better and the obvious bugs (which are completely acceptable in the Beta phase), the simulator is very good for those looking to improve their flying skills.

    Keep up the good work

    1. The cockpit IS dark in real life. Compared to the outside, at least! I have a pic of my son when he was 2 in the cockpit of an A320 at Charlotte and it’s white windows and a really dark cockpit – the light levels are just really wide, and we now have an engine that can capture that.

      With that in mind…is it the exact right amount too dark?
      1. We don’t know! I have a TODO to get better checking of the absolute light levels in the cockpit in the HDR buffer. We want this to be right.
      2. How it looks to YOU, the user, depends on how we tone map the photometric values back to something your poor monitor can show. Even an HDR monitor has a lot less dynamic range than real life, so we have to map this down, and we need to map it in a way that looks good and makes the sim usable.

      But this mapping can’t be tuned until the lighting is confirmed, so there’s a few items to do here.

      With that in mind, if you were to show me an image where the cockpit and outside were the ‘same’ amount bright on a computer monitor, personally, I’d have to say “that looks totally fake” because it’s so far off from reality. A more real image might be *harder to use* – so I’m not sure how we’ll balance that. The rest of the team will have strong opinions on that but first I have to make sure tech isn’t just _buggy_ i some way.

      (The exterior light affects the interior brightness via the windows – the pre-computed radiance environment maps we build on the fly understand this and take it into account. Since we don’t have a generalized global-illumination model in the cockpit, the effects can be over-or-under-counted based on the materials in the cockpit, just like in v11, hence the need to go validate our data.)

      1. But Ben, some cockpits in XP12 are pitch black. In some situations you can’t read anything. This is definitely not the case in real life. I hope you will be able to find a short term workaround until a final solution has been found, which will take time if I understand you correctly.

        AND thank you and your team for this fantastic new sim!

          1. In real life in real cockpits when you go from looking outside to inside our eyes dilate to compensate can you code this effect into the sim? Maybe easier with VR eye tracking than we users with GTX960s and 42″ 4K monitors.

        1. I agree. The cockpits seem much too dark, making it almost impossible to read anything, even if it is bright outside. I never sat in the cockpit of an airplane, but if I sit in a car during a bright day, the interior is also quite bright. Therefore, the dark interior of the cockpit not only feels unrealistic but also bad from a usability point of view.

      2. Another variable is the user’s monitor. I fly only GA but noticed the cockpit was darker than was comfortably useable. I calibrated my monitor and it made the cockpits much more natural and useable. With the new photometric lighting, it seems to me that users really need to consider adjusting their monitors to enjoy the full effect.

      3. I think the cockpit lighting does NOT require a fix. You guys have nailed the lighting inside the cockpit. Speaking from my flying experience, Cockpits are dark specially when you compare them from the outside. Right now the lighting inside cockpit looks extremely realistic.

        1. I agree on what Babar says. In some scenarios I have to turn on the panel lighting, otherwise it gets too dark. Feels very realistic to me.

          I had manually calibrated my monitor though, a long time ago for photo and video editing/grading. It still shows the nuances in very dark colours. I think that might be important.

          Imho the lighting is already very close to fantastic. Very immersive, only beaten by the puddles, which are just awesome. : D

        2. The issue is that “Cockpits are dark specially when you compare them from the outside.” but only when compared with outside and only when you were looking outside and than shift your focus to the dashboard. Nothing different with car dashboard when one drives on the highway in a sunny summer day. But when you shift your focus to the dashboard your retina would react and expand so you are able to see the panel within second of looking at it. This biological process can not be replicated on a 2d monitor unless there is some kind of super fast eye tracking. So i think it should not be replicated at all. Just the cockpits should be lighter regardless of if it is realistic in all conditions or not. Dark cockpits are not realistic in all conditions too.

      4. Hi Ben,
        1. The human eye can distinguish about 16-18 stops of luminosity. The electronic devices can 4-4.5. Hence, a photo taken under uncontrolled conditions and developed (even digital images are developed!) under not controlled conditions can not be a reference for cockpit appearance.
        2. Dark, or better dynamic range of a scenery, can be assessed in a Gray 18 room and rendered accordingly to provide a similar physiological reaction from the viewer’s eyes.
        3. I use a calibrated monitor for photo work. I don’t believe that many of the users use a calibrated monitor. so my suggestion is to provide some setting tool, per aircraft, to tune the black, white and mid points. This might free you from luminosity tuning and allow the users to set own preferences, like, for example, response curves for the yoke.

        Good luck!

        1. The point of the photo is not to reference cockpit _appearance_ (e.g. the perceptual experience of being in the cockpit). It’s to point out that the actual difference in luminance is large compared to what electronics comfortably deal with.

          In particular, your computer monitor is 8-bit, and it’d be 10-bit if we supported “HDR”. (Scare quotes doing a lot of work – HDR is better than LDR but it’s not close to the real world and doesn’t make tone mapping unnecessary.)

          The idea of a photometric renderer is to get the _internal_ light levels to be accurate to real world physics and then solve the perceptual problem with tone mapping.

          I agree that there’s a lot of uncalibrated monitors out there – I could see us providing a generalized brightness control, but I do’nt think it makes sense to let people tune black/mid/white…at some point they should just tune their display itself.

      5. My perception is: The balance of light and darkness in a airliner cockpit is spot on.

        The only issue that remains is that the sun is still being reflected by glass instruments as the aircraft body was translucent.

  43. Will Laminar provide the roadmap for the main plans after the Beta phase?
    We – users – would be very happy…

  44. Is there a problem on the Bug Reporter?
    I tried to file a bug, but i am getting a Server Timeout when submitting it.
    It looks like a certificate is not renewed…

  45. Great job, love it, but I hope the roads without houses in the middle of nowhere disappear. They look like they are part of a village but without the houses. This also was the case in V11. Objects are set at the max already.

  46. Hello Ben, what will be the focus of the development after leaving early access? Above, you stated: “There was an effort to color correct the ground textures, but 12.0 is not a “redo the ground terrain system” release.”

    Will you redo the ground terrrain system in 12.1 or 12.2? This aspect seems to be among the most urgent now that you have moved to Vulkan/Metal in 11.50 and greatly improved many visual aspects in 12.0.

  47. Can you please do something about all sounds stopping working when changing sound output on windows? It’s quite annoying. Loving XP12 though, great work by all!

  48. Hi,
    First congratulations on the X-Plane 12 early access publication.
    Although there are chalanges ahead, I think that the in plane textures, especially the gauges must be clear and with the highest quality possible.
    Currently xp11 produces sharper and clearer gauges textures vs xp12 (I already filed bug report).

    Oh, and the cloud pixelation is still there in VR (after xp12 b3 patch)

    Cheers

  49. I have to use the flashlight in all cockpits because they are so dark I can’t read anything.Even changing the FOV
    setting,as suggested by some, doesn’t work.
    Any other fixes,temporary if necessary,will be much appreciated.Can’t use it in its current state.

  50. Please upgrade ground. It renders like 2006 due to outdated models for trees, lack of grass and old textures.

  51. Hi All,
    Great expectations. XP 11 is a very intelligent product. Probably the most intelligent flight simulator on the market. Definitely not the most beautiful. I purchased the XP 12 and this is what I feel at this stage: XP11 in a stiff competition with FS2020. Please do not. Garmin 430/530 FP works now. This is important. More clouds, less clouds irrelevant. Better aircraft dynamics, this is important. A few rain drops, irrelevant. And so on. IMHO.
    Florin.

  52. Does this version “12” has an option to install an external aircraft model to this software? My intentions are to install a Quad City Aircraft Challenger II model to practice crosswind landings. Thank you!

Comments are closed.