X-Plane 11.0 public beta 8 went up over the weekend. If you are seeing problems with the flight model, please report bugs ASAP so Austin can look at them! I think we are approaching the end of some major flight model updates; the last thing Austin is looking at is better body shadowing, which he will probably write up shortly.
Over the weekend, with the help of some very patient users, we found the cause of poor performance on some AMD hardware*. I have a prototype fix, and I hope to have it in X-Plane some time this week. This fix will only affect users who were seeing super-lousy performance on very light configurations.
In the most recent betas, the threaded driver no longer totally kills X-Plane performance. But it does still slow things down a little bit on some computers – I see 5% fps loss with the threaded driver on. My suggestion for now is: try it both ways and run with whatever works best for your machine; this bug is affected by both your particular CPU and the kind of work-load your settings induce in X-Plane.
I am still working on improving cloud performance, and a recut of the DSF tiles is rendering as I type. The first priority for the new DSF tiles is to fix blatantly broken tiles (e.g. YSSY) and get the file size down so we can put Alaska and friends back.
* I wouldn’t call the AMD problem a bug on either our part or AMD’s – it’s more in the category of “OpenGL makes no promises as to what might be fast, so app developers have to debug on all shipping hardware and try not to do painful things.
Will the AMD “fix” make it back into X-Plane 10 as well? I found that to have about the same level of performance (or lack thereof) as previous beta versions of 11. (And yes, I filed a report about that.)
I am examining this now.
Well done Ben, glad it’s starting to come together.
One question if I may: During the beta release, will we see the return of certain rendering options in the graphics tab? The ability to control the amount of cars, draw distance etc is sorely missed by myself and quite a few others.
Keep up the great work and don’t forget to ask for that pay rise!!!
Cheers
Dom
Probably not. I meant to write up a big post on this but got bogged down trying to make it coherent, since the issue was an important one.
The VERY short version is that MOST of the “bring back X” settings requests I’ve seen have been aimed at solving a problem that is both:
– Badly or not at all solved by bringing the settings back and
– On our radar of things to fix in the sim itself.
For example, in pretty much every case of drawing distance I’ve heard about, the goal is to tune peformance. But here’s the thing: there just _isn’t_ one drawing distance anymore. That may have been true in X-plane 9, but in X-Plane 11, there are interactions between several categories of draw distances, several rendering modes (e.g. normal, shadow, reflection), the type of scenery, the visibility, and the distance at which we build stuff.
Given so many variables, I am not looking to try to break out drawing distance…we’d have to do a ton of work to make the drawing distance work in a way that was not astonishing; if the goal is just to get performance back, what we need to do is to address some of the problems that exist at the art asset level (e.g. bad LOD tagging in the art) and then tune both what we draw and when we draw it at the same time for all conditions.
Same thing for the cars – the usual complaint I hear is “cars hurt performance”. That’s why they are turned down so far right now. The fix for this isn’t more settings, it’s to fix the performance bottleneck in the car code.
In other words, I am _not_ in favor of the settings turning into a giant control panel of check-boxes where users try to band-aid around sim problems…we’ve tried that in the past and it works badly. I know this because users send me their settings and they are often totally non-optimal.
There are a few settings that may have to come back for specific operational reasons. Those are few and far between and being considered now.
Finally, if you are going “stop calling me a noob, I can TOTALLY tune your engine for you”…then go hack up settings.txt, which will give you access to ALL of the actual controls in the engine.
would you consider brining back a toggle for Slopped Runways
theres a fix by modifying xplane.prf , but many folks dont know that and or is tedious for them.
That’s the most likely one to come back.
I am definitely _not_ happy about bringing it back, because I think the community needs to start getting flattening into a per-airport basis, and putting it as a user setting stymies this effort. But we do have some pro customers who need to be able to control this globally.
Thanks Ben.
Hi Ben:
The sloped runway option is important for those of us who are flying online ATC (such as PilotEdge) where you have so many different users on different sim platforms flying together. I personally love the look and accuracy of the sloped runways and you are correct the X-Plane community should get behind the per-airport flattening, but in this case we are trying to play nice in a sandbox with lots of people on other sims.
Your concept makes sense re: setup options to “tune” performance – the one item I would hope we would have a slider for is the bird flocks – not for performance, but they are very inconvenient when trying to setup certain flight training scenarios (that you might not want interrupted by inopportune bird flocks showing up; this has happened a few times for me already).
I have to say so far the smooth performance I am seeing in PB8, if indicative of the final product, will seem to lead to a refreshing absence of the need for tweaking and tuning. Kudos to the team for that!
Our current thinking for birds is that they need to be managed by the failures system, and there need to be failure modes where the sim fails only on command for flight training.
If you assume that the main reason people turn off features like cars is to improve performance then your philosophy of improving the performance makes sense. I think, however, that some users want to turn them off simply because they don’t like them. I personally think their behaviour looks cartoonish and they detract badly from the experience of realism. The same with features like deer and birds.
I like your new approach about per airport flattening (as I suggested you a while ago), but I would have liked it to function in the opposite way, too, ie you could also specify in WED if an airport mandatory needs a sloped runway: in that way with a global setting you could both tell the sim to unflat everything except the airports that mandatory needs flattening, or flat everything except the airports that mandatory needs a sloped runway. Airports that need flattening are much more than those that mandatory need a sloped runway, and currently it’s a bit of a pain if you already have a ton of sceneries that need flattening to check each one or wait an update for each one.
The problem with this is that the users who need to control the _general_ state of flattening all need flattening, so your proposed logic kind of goes against what users are looking for. IF a user NEEDS flattening (e.g. they are integrating with a separate flight model and using x-plane for visualization and they need the airport to be at a published MSL) then the fact that the author goes “no no it looks FANTASTIC with sloped runways” isn’t important to the customer.
My view is that an airport needing flattening represents a bug in the base mesh; our goal should be to condition the base mesh such that flattening is ALMOST never needed.
Also, bear in mind that v10 is a poor sample set of how this can work – there was a huge bug in the v10 DSFS — a large number of airports had their base mesh rendered WITHOUT the airport..hence all of the trees on the runway, bumpy areas, wrong texture under the runway. These errors are being fixed in v11.
Ben. This is good news. In my case, in aerodrome near me, have wrong mesh altitude. and runway was wrong.
I hope the mesh is corrected, for best use of this fantastic feature.
Continue with good work!
Hi Ben,
I’ve been studying and experimenting with X-Plane clouds for years (my mods are quite known and used in the community). If I may give a few suggestions coming from experience:
.) Using .dds (DXT5) instead of .png should improve performance without visual degradation;
.) 6 horizontal cloud layers (layer_0 to layer_5) are way too much, lead to bad performance, and create unpleasant visible boundaries; 2 or 3 are sufficient, but you would need to re-code the altitude at which they are drawn;
.) for the same total cloud volume, a small number of bigger puffs have better performance than a greater number of smaller puffs; I understand this can highlight the 2D nature of near cloud puffs, but maybe this can be worked out or tuned;
.) the cloud opacity to diffuse light (sim/private/controls/cloud/sample_opacity) is too much and produces unpleasant dark clouds. I use 0.02 instead of 0.05;
Here you can see some results of these tweaks (plus others):
http://i.imgur.com/dGhARdB.jpg
http://i.imgur.com/XPuK0Zx.jpg
http://i.imgur.com/SOc205S.jpg
In case you’d think I could be of help, you can contact me. 🙂
Hi Marco,
Max is working on the clouds now. Point 3 is definitely true – cloud performance is limited by over-draw, which is directly proportional to the number of puffs, because this is what causes depth stacking. Those pics look good!
What about city lights being visible when you are flying above and in the clouds ? I don’t remember that being a issue in XPX-10
+1
Always an immersion killer but maybe there are technical reasons why they cannot be skipped, i.e. the drawing engine of XP11.
I agree about cloud color, they are too dark for cumulus-like clouds. Heavy clouds and overcast looks great in moody look.
I see another problem (was in prev versions also), when you fly into cloud layer, when you’re blinded and see uniform color, it’s too light. Often near sunset/sunrise, clouds from above/below have distinc color, like orange or yellow-ish, inside they are white or light gray.
Other thing, cloud layer may be little more undulated, i get filling that clouds are often to uniform and have too even surface.
But anyway, i love changes in weather model in X11, it’s so much fun flying through clouds and in low visibility!! Great work Team!
Hi Marco,
Where would one be able to find your mods? Also, which weather engine would you be using? It doesn’t look like default x-plane.
Cheers!
Hi Marco,
Could you tell which are the mods you created? I was impressed with the pictures and would like to test them.
Thanks
hi Ben,
i submitted bugs this weekend re beta 8 about lingering smoke.
I dont see that listed above as a topic, is this being looked at. thanks
This is NOT an exhaustive list of every open bug.
ok cool
Hi, Ben!
Thanks for all the good stuff so far and coming down the pike!
Should we expect some info/instruction on integrating FMOD sounds into XP sometime after the Beta cycle has ended, or maybe sooner? We are working on a new model right now, and curious about whether to change direction for our custom sounds.
Thanks!
I think FMOD instructions may be after the beta cycle.
WHY? devs need that now to have stuff out in time
FMOD is an entirely new piece of technology to X-Plane 11, and it is not _ready_ for third party authoring yet. There is a difference between “we can run a plane and you can hear it and it sounds good and it is fun to fly” and “this tech is stable for third party use and won’t need to change and break everyone’s work one week after they implement it.”
We do want X-Plane 11 to be at a point where an existing v10 can be updated to v11 without a need for do-overs after we ship. This means:
– The requirements for mandatory changes are known and will not shift later after we ship.
– Compatibility bugs where we should “just work” and don’t are fixed.
This does not mean that NEW v11 technology is third party accessible.
I bought XP11 I do not have and I do not want XP10, I’m patient and I’m still waiting, I’m a P3D user, my only reason to migrate to XP was that XP11 would work with Vulkan, leaving behind the obsolete and horrible OpenGL, I have an ATI R9 290 4GB DDR5 286Bits and the Vulkan is developed by AMD, if XP does not have better performance with a board like that, 20FPS at the moment with setting low … well !!! Welcome P3D with 35FPS, and just for the simple fact that I can fly it. I still wait and the ones you just gave are good news and renewed my hopes.
I understand why you would want the highest framerate possible.
Why do you care _which_ API your flight simulator uses? Is there something -specific- about OpenGL you don’t like? Do you care if P3D uses DX11 or DX12?
I understand and if not, please correct me if I am wrong … The API import without the same performance you get bigger effects and better graphics, if you have an API 2006 your graphics and effects were vintage … example FSX, with new API technology takes better advantage of the graphics cards … which is my case … I did different tests in XP11 in a default airport with the default cesna in 42FPS but when turning on the cesna avionics the FPS dropped to 20, I do not know what does not work on my PC and with XP11.
Ah, I understand! Well, there are TWO separate issues:
– Is an API old and lacking in support for new features? Here, the answer ins no. OpenGL is continuously updated; 4.0 was released in early 2010 and it was then updated repeatedly for the next 5 years. D3D11 was released in mid 2009, and has been updated for at least six years. I am not aware of any hardware features accessible via D3D11 that are not accessible via GL4.5 + vendor extensions.
– Does the API have quality control problems? Here the answer is probably yes. Most developers (like me) have spent almost all of their dev time using only D3D or OGL, and aren’t in a position to make good comparisons, but the comments I have written from the few who have done serious app on development imply that D3D drivers match the spec more reliably than OGL drivers on Windows.
I am unaware of any cases where FSX is doing something with a D3D API that we can’t/don’t do in X-Xlane; we’ve had hardware instancing for five years now, and FS X is not using the tessellator (which is available in OpenGL anyway). There isn’t really an OpenGL gap here.
A comparison of X-Plane and FSX’s framerate tells you nothing about the underlying APIs; what they actually do is so different that it’s not a valid comparison. To compare the APIs, you’d need to find a single app capable of using _both_ APIs and viewing the relative performance. In the case of AMD + X-Plane, the FPS drop is an indication of something going wrong between X-Plane and the AMD driver. We can fix this, we are fixing this, and at no point does OpenGL stop us from doing so.
Both D3D11 and OGL have the same fundamental design problem: they’re not friendly to apps that want to be heavily multi-threaded and use a predictable low level API for guaranteed high speed reliable performance. Both APIs have an alternative (D3D12 and Vulcan), and in both cases the new API is not a replacement, because the new APIs are much more low level. I would argue that OpenGL’s threading model is _more_ screwed up than D3D11’s, but I don’t think either deliver what D3D12, Vulcan and Metal provide. So I would argue that Vulcan does not make OpenGL “obsolete” because it doesn’t solve the same problems.
Thanks a lot for answers and clarifications, now I’m a little clear about the APIs, which is not clear to me what my problem is with XP, maybe with your new fix will correct my problem. I like XP because I can buy any plane and this will have the flight physics, but in P3D I can only resort to A2A and Majestic and not many more. I reiterate my thanks to your good predisposition…. It is good to know that here any user is heard by developers.
Anything happening about the Null Texture error which still seems to exist in pb8? This is still an issue in my set up for some reason. Many thanks.
We found it today – will be fixed in beta 9 soon. You can work around it by plugging in your joystick hardware after starting -if- you are affected by it, at least, we think.
The workaround you mention does not work for me. I am now (as of today) using 1 powerful computer and 3 independent views on 3 displays.. Up to last week, running 3 lame computers/displays, I never saw this error.
I sent in the “automatic” crash reports on my 2 attempts to fly today; are we also supposed to file bug reports?
Please _do_ file a bug – it’s possible you have a different case of this than the other users who saw this.
“null texture” isn’t one error – it’s the message we get any time any code path ends up trying to use a texture that is not yet loaded, so one symptom has many causes.
Fantastic, thanks for all the great work you are doing!
I tried leaving my joystick unplugged until after starting, but it didn’t fix my particular issue (something to do with the pref files and possibly the toe brakes – bug reported and pref files sent as requested on followup).
Looking forward to the next beta to see if this one got quashed.
Keep up the great work everyone!
Hi Ben, did you consider to increase the global scenery size to include Alaska and northern latitudes again instead of cutting down on data? I believe that with all the new osm data available dsf tiles size should inevitably increase over time
(unless you find other comprassion ways) and it’s a pity to lose some data to fit in the same size as before: you either have less osm data or less base mesh or coastlines data. I know that the DVDs number is a constrain, have you thought about the possibility to let people download the extra tiles that don’t fit in the DVDs?
I think you’re suggesting two different things. We did two things to cut data size:
1. Remove MORE detail from the OSM coastlines. This is necessary to counter-act increasing OSM detail and a TON of new OSM data that has been imported.
2. Cut down the scenery set from 72N to 60N.
Even with this, we were over size in the first render. Thanks to some really useful tips we got, we fixed a bug and the DSF size is smaller now. So:
1. Did we think about putting Alaska back? YES! In fact, our first priority is to get back “high value” northern scenery, e.g. Alaska and Scandanavia.
2. Did we consider dropping scenery areas to raise the detail of the areas we ship and letting the rest be downloadable? Yes, but I think this is problematic – any time we make the DVD vs online version different, someone will be unhappy. And users buying the DVD these days probably don’t have great internet.
So my expectation is we’ll keep the detail level we have now and try to get at least some northern territory back.
Isn’t more detail better ? Ship more DVDs ! Charge a few more dollars…
Obvious answer (at least, in my mind) make more DVD’s? More DVD’s not an option?
I see. I think you could also consider the possibility to ship extra payware DVDs for the missing DSF areas for people that don’t have a great connection. In this way you could really ship a big step forward in the global scenery and make everyone happy. I fear that we are going to see many missing areas in northern Russia and Canada this way: I would be really happy if you could let people download those dsf.
I second Eric.
I don’t know if there are some specific reasons not to do that and for some time now I was going to ask you this, why you don’t simply add one or even two dvd disks?
Could this have to do with the potential risk to have different DVD and online versions?
Ben, you make no sense here: you can keep the digital download via the installer and the DVD version identical with as much level of detail as possible AND make additional tiles available to download separately (requiring authentication in some fashion) and with the user installing it (and not necessarily in Global Scenery). The updates/installer would just ignore those additional tiles; should they be improved, the user would have to download them separately again.
What problem would this create? What valid reason would someone have to complain?
Like the others above, I don’t understand why 8 DVDs is sacrosanct. What’s wrong with 10, or 12? I want the Scandinavian countries back, but I don’t want to lose coastline details.
I wonder why the number of DVDs is a constraint, except if it was to increase by a multiple. Even so, anyone who remembers software shipping on 5 1/4″ floppies wouldn’t be surprised by a box as big as your head. Hey I have the complete set of “The West Wing” on DVDs and that box is huge!
There’s always Blu-ray too, although some people may not have support.
I am waiting on the DVD version to ship myself rather than go digital download.
Good luck Ben and team with the run-up to release-candidate and 11.0
I have to agree, Ben. Limiting the DVD set to 8 is artificial, and sacrifices the improvement in quality that one would expect with a next generation release. Bumping up to 10 should not be a big deal or an inordinate expense, and you probably get all of the space you need to maintain +/- 72 degrees as well as the improved coastlines that are frankly pretty important too. Jagged coastlines in XP10 are horrible, and not really that much better in the HD variations. An extra $5 per set increase would probably keep everybody happy, if money is all that tight. Give it more thought, if you would, please. X-Plane 11 should improve all the way around, not be a bundle of trade-offs.
+1
Lots of details AND sweden please!
Dont let an old medium hinder your modern product!
Mostly replying to say that I, as someone who buys the physical release simply because I like having the physical copy, would really like to see a Bluray release of X-Plane 11. I am kind of surprised that the idea of cutting down content comes before switching to Bluray media. I suppose there’s probably evidence to suggest that there’s not enough market penetration on Bluray drives to make it worth while, but I would imagine that’s largely since people are avoiding releasing things on the medium.
I think there’s about zero chance of us doing blue-ray, for exactly the reason you said – there just isn’t market penetration to make it worth it.
As cheap as flash drives have gotten (128gb for around $25), would it be feasible to replace scenery DVDs with one?
There are two issues:
1. The pricing has to match for cost of goods, so $25 is < $60 but it's still more than we pay in materials for DVDs. 2. When we get DVDs, we get them manufactured in bulk. Even if you can find that a box of OEM wholesale flash drives are cheap, you'd still have to get the data onto them. That might be cheap - I don't know - but it's another service that will add price.
Hi
I keep seeing crooked taxy lines on the ground of the airport. It doesn’t matter wether or not my settings are on max or high, they keep showing. Is it a bug? Something you are working on?
Thanks
Are you speaking of teh arrows that show you teh way?? They follow ATC lines “IF” the airport has them and many wont as it hasnt being around long or they follow the xplane generated route which is not perfect. Best thing to do if you fly out of an airport constantly ifs fix it and upload it for others. Thats what im doing on some Aussie airports
Ooooooooh…right! If the airport doesn’t have human-drawn taxi routes, the auto-gen ones are often quite crooked, and this is visible when ATC draws ground arrows for your routing. So you can see how goofy the auto-gen taxi routes are.
Drawing the routes yourself is the best bet – with careful use of WED’s features it can be done relatively quickly, and provides clean results.
Hi Ben, i have a ask / request / question.
I m not sure if you prefer to email it or how can do to be fine, but i will little explain here to tell me something are really 3 points
– From XP10 i detected in some regions when load a library objects win EXPORT i uderstand that this will remplace all definition with this last definition right? after this can use EXPORT_EXTEND to improve a definition with new objects for example. Well… some times work good, but anothers time EXPORT work like EXPORT_EXTEND and vice versa, i dont know what i can send you to check but i sure that here are somthing that is not work fine.
– About ground vehicles wich are referenced to library, i want to add more variety of trails or traktor, and EXPORT_EXTEND is not working, only get the first defeinition. Will be fine to make use of extend to add new types of vechicles.
– The las one: will be more category ground vehicles avaible shorthly? I personaly miss: stairs, security/administation cars and firefighters, at least!
Hi Alex,
1. EXPORT should replace all objects from other scenery packs but mix with exports in this pack; EXPORT_EXTEND mixes your scenery pack with others.
http://developer.x-plane.com/?article=library-library-txt-file-format-specification
If you find a bug I will need a detailed reproduction test case.
2. The ground vehicles may not be library-ready yet; some global parts of the airport system aren’t regional yet. This is something we want to do some day.
3. Not shortly, but yes – we do intend to add more vehicle types. The only one coming shortly is a limousine, which provides another way to get picked up from your airplane; the artists are working on it now.
Our intention is to be very specific in WED about which vehicle you get, rather than randomize, so WED authors can control exactly what they get.
Thanks for the update Ben.
I have flown VFR in Germany last night and it felt great. Performance and the “reworked” flight model are feeling just awesome.
Just the real world weather is quite quirky and for now I’m just turning it off and using constant weather. However, this is a bit inconvenient when flying online (IVAO or VATSIM).
Any plans on improving the real world weather a little bit?
MANY THANKS!
Nice. I really notice some performance decrease since some version on x-plane 10. I also got a little performance decrease since beta 5 on my AMD.
I have an AMD 7970M running on a clevo.
Ben,
Thanks for all the updates , getting some good performance on the Gt525m.
BTW why is the CPU load high ? Is that going to be balanced ?
As to the missing DSF’s , MENA regions seems to have some missing like Cairo. Even the areas in California seems to be missing tiles. Asia seems to be OK not yet tested anywhere else other than India. Australia , Queensland , Sydney not there.
So once you have the missing regions rendered do we have to re-download the whole area ?
Rgds.
Hi Ben
First off, thank you for the FPS fix, running NVIDIA and updated from PB5 to PB7 and the FPS dropped by half, then updated to PB8 and it came back, excellent.
I do hope you work on bugfixes, and furthermore hope you bring back some of the XP10 graphics settings, its way to simple and hard as is to tweak XP11 to perform, also I miss the runway smooth option, shadow tweak, ability to control water detail etc.
I know this may be inteded for the released version, but most of us are enjoying XP11 as is, and while estimated 70% of the userbase are pretty much running on the GTXxxx series still, we need to have options to control performance. I do not complain when getting 20-25 minimum and 35-40 maksimum fps even with a third party vender sky texture. But in XP10 as is, I get more than 60 and in some rare cases over 90 and 100 depending on the angle of attack. So with my GTX770 by Nvidia and latest drivers I would love to see and PB with some extra handles for us power users. Simplicity killed the cat is a known GUI advice in some geek communities, which I think applies to X-Plane to. Though the new gui looks and feels fantastic, we are still missing detailes which as I said may come in the official release of version 1.0.
hi Tom,
We have no plans to extend the X-Plane 11 rendering settings to make it “tweakable” for performance, the way xp10 was. This was a strategy that failed in x-plane 10, and we are not going to repeat it in X-Plane 11.
Thank for the reply Ben
Understandable from a developers point of view. I do have an follow up question if you can answer at this point. Are you or do you have a strategy for tacling the blending draw order in XP11?
I’m not sure what you’re referring to by “blending draw order” in this thread.
Sorry, forgot to elaborate.. Blending draw order in the sense of transparent clouds. I think it has something to do with z-buffering and the draw order for rendering. I talked back and forth to Andrey Sheybak the developer for xEnviro and he and I discussed this perticular issue regarding an issue were of ground was visible through the clouds (thick scattering clouds) Does this sound familiar, I believe there was mention of some issues in XP10 as well. How does X-Plane use z-buffering if any in case of clouds.
My question to this is then WHY did Austin say what he did about settings and ability in XP11? from your answers here its VERY clear we are going to be. Being told by the “face” of XP one thing and having another done, feels like we’ve been sold a bill of goods. While I understand you want to optimize and fix the things you can so that we don’t need those settings, but in the mean time, what we have and what we were told. do not match.
Austin does not want to lose functionality from X-Plane.
I do not view “having a GUI to tweak the rendering engine internals” as functionality, I view it as bad product design that causes poor results in the field.
Soooo…there’s your disconnect. If you really think that “having a lot of rendering settings” is a key product feature…I don’t know what to tell you. In almost every case where a user wants a setting back, the back story is “this setting worked really badly and shouldn’t be used in the way you were trying to use it.”
Having “a lot of rendering settings” is definitely not important.
Having rendering settings that, to the extent that the code permits, can alleviate workload on the CPU and/or GPU is. These settings need a certain level of granularity and specificity, since there is a better level of control in certain areas.
I still see a certain dumbing down of the render settings that goes too far the other way, Ben. Try to be more neutral. 😉
Hi,
Please keep it as it is.
Many users, even active in the forums, do hardly understand what they are switching on or off – just complaining ‘I need 100 fps’.
However, AI Aicrafts are still one of the few left examples where something needs to be done. who could imagine a 50% fps drop by activating a few ai models. This may have to be fixed in the code in the future.
Right – AI perf is like a lot of perf issues – the real work to be done is not just in turning some setting up and down, but in finding smarter ways for the AI to work.
Considering the settings that I “want back” are controls for things that I had (ie cars, shadows)
I’ve indeed lost functionality. I had it, now I don’t, that by definition is loss of functionality.
Having the ability to raise and lower the amount of trees/cars/shadows within the UI was pretty simple, I don’t want trees I lower the setting. same for cars and shadows. I can’t do that now.
I have no other way to explain that than that, its a straight loss of function.
But Neely, that’s the point Ben is trying to make. Why would you want less cars, or less shadows? Ideally, we always want full realism. The objective of reducing the number of cars or shadows is to fix an “issue” that shouldn’t be there in the first place, lack of performance.
If you had perfect performance, would you still care about reducing cars or shadows?
Cheers.
There are usually two issues: performance and realism.
For performance, my view is that we need a small number of sliders that make real changes to performance by affecting a lot of settings at once so that EVERY setting makes the sim look as good as possible for a given performance level, and every setting is as _efficient_ as possible for a given performance level.
For the look of the sim, my view is that final judgment on the look of the sim needs to come from our art team. So if you think the cars look bad and our art team agrees, they should go fix them. If you think the cars look bad but our art team likes them…too bad I guess?
I certainly think that a “mix and match” world is not a great idea. For example: we have a check box for “cars”. But some cars are dynamic from the road file and…some are just OBJs. So a user turns off cars and … there are still cars everywhere. A user turns off forests, but that affects only .for files, not OBJs, and so there are some trees in some places. Consider the “roads” setting in v10 — we eliminated road by road type, which is dreadful because you end up with highways where the off-ramps are gone…but sometimes a complex interchange is _only_ made up of off ramps and the road just has a massive hole in it.
My view is that we could, with _considerable_ engineering effort, modify the scenery system so that these kinds of “category edits” actually do work. But it’s not free at all – it would require very careful coordination of programming and art to make them work. So my view is that this is not a good use of our time, and that we should focus on _one_ X-Plane world, modulated only by performance requirements, that meets our art teams goals as much as possible. Most of the time this is accomplished by a carefully integrated mix of clever programming and art asset modification.
The ideal is what the customer wants. We had specific controls for certain things in X-Plane 10 and previous that some users simply did not want. Like deer on the runway, birds, balloons, car and tree volume etc. Grouping things together is clearly not what the customer wants.
Many, many people have requested a two tier interface – the simpler one that we see now, and one that permits them the same level of utility as they had in X-Plane 10. Yes, you may have people that set things up crazy. If you don’t want to support the advanced menu system, then don’t. No complaint from me, and that gets you the support streamlining that it seems is more important here than design philosophy.
And if people like the simpler system, fine too. But when the pushback is as consistent here as it is in the forum(s), and you want to sell the sim to as many people as possible, why fight back so hard, Ben? We’re the ones that are telling you what we feel we want in the product that we’ve had before. Yes, there would need to be more time spent on the UI, time better spent than on philosophy that’s telling people to accept what you think they really want.
People that don’t want the birds, cars, deer, etc., are now left with having to replace the art assets with clear .pngs and still suffer whatever impact that code has on frame rates.
The other side of the scale from “some users like having lots of choices” is the opportunity cost to develop and maintain them, which is a trade-off between have them work nicely (and suck down dev time) and have them work badly (and have users use them to get bad results).
In response to Bruno, I take your point about shadows. They look fairly realistic already so ideally you want them improved further rather than turned off.
The behaviour of the AI cars, on the other hand, looks like a Disney cartoon! Their jerky stop-start, instant turning performance looks utterly ridiculous and since they are already a performance hit I imagine that making them more realistic would be even worse.
Surely Ben could save himself a lot of trouble by simply letting people turn them off (along with the deer, birds and balloons).
And … look at the assumption: “since they are already a performance hit I imagine that making them more realistic would be even worse.”
This is actually not true at all! There are two big costs of the cars:
1. In HDR mode, computing the HDR spill light volumes per car. This cost is constant no matter how good or bad the car animation is – it is a cost per car on screen per frame per light. We could have better animation and pay the same here. The fix is in code – thread car headlight volume generation and/or optimize the code and/or push the volumes to the GPU and/or use cheaper volumes (CPU wise) that cost more GPU-wise. That last trade-off is good because car headlights are tiny in screen space most of the time.
2. In all modes (and this is the BIG cost) when not paused, each time a car reaches a navigation junction, we recompute its culling sphere for the next road segment and the car has to be removed and replaced in the quad tree.
THAT step is embarrassingly expensive due to bad use of the CPU cache…the data structure in use was great 8 years ago and is stale now and is scheduled to be replaced. When we replace this data structure, car performance should get a lot better, and we’ll be able to have a lot more cars for not much more CPU – a big win.
I am sympathetic to “the cars animate badly, I’d rather not have them at all.” But even if this is the case, and we say “yeah, these are so stupid that we should hide them (and get a little perf back) until they animate nicely” there are still smarter things to do than just a user-set on-off switch.
For example, we could put cars ONLY on the highways and refuse to let them go on an on-off ramp. Pretty much every car artifact comes from lane changes and sharp turns…we could simply program the cars to NOT do that and still get the benefit of seeing cars on highways at night for VFR flight.
I’d like to solve the real problems that drive the demand for settings, not just put the settings back without asking questions first.
In the case of flattening the earth, I think we may HAVE to put that one back, and we can be pretty sure we have good reason to because we’ve heard extensively about the specific problems that users are trying to solve and the setting solves the problem in a way other solutions do not.
If you really want the cars gone, go whack the art controls.
But I think the issue here is that there is a BUG: “car animation looks bad.”
My view is that we (LR) need to make some real decisions, e.g.
1. Is the car animation bad enough that cars should not be in the product.
2. Where in our list of priorities should we place making the cars animate more realistically.
Look at the amount of comments spilled on this to get to an -actual- problem (car animation is bad). To me, a huge amount of “we want are settings back” is the patient asking the Dr. for pain pills without saying what the injury is.
To be sure, Ben. Dev costs are everywhere, and I realize you’re speaking of “balance.” But if you get enough demand, isn’t it worthwhile to throw those that feel they can manage their settings at least some sort of compromise? Clearly we’re not going away. 😉
Think of it this way: the advanced interface can be added, or people will risk those crazy trained monkeys that will steal all their underwear. LR can control this, or some enterprising author will put together a plugin that does. Lots of what we’re asking for is probably already in the sim. I just checked the art asset datarefs and found many I could change that reflect just the sort of control I think we seek – at least at a modest level.
Methinks that while your philosophical arguments are indeed sound for a great many users, the real dev cost resistance may be more tied to developing the secondary interface. Toggling whether or not something draws, or the quantity of which it draws, is something that seems to be a zero dev cost at the moment.
The next time I see deer or birds in X-Plane 11, DRE gets opened up and I validate my assertion. Right now, LR is in the driver seat. Do we or don’t we? If “we” do, then “we” control what else we add to controls, rather than the users. Those that want greater asset and rendering control will indeed hack setting.txt, and while that’s clearly *not* supported, it’s been easy to avoid to this point with the bulk of the settings being exposed within the sim. I think it would be interesting to learn settings.txt by experimentation – but I’m just nutty that way. But if one person is successful, then that spreads. In fact, there are a number of threads on this sort thing springing up on at least one forum. X-Plane users are an inquisitive and intelligent – and determined – lot. 😀
Well I apologise for being so hard on you, but you see my point about he cars. (The trains are even worse because of the way they bump over the undulations in the track – I’m an ex railway modeller).
The thing is there are also plenty of things that are modelled perfectly well, but some people would still prefer to be able to reduce them. For example I know quite a few people who like to keep the tree numbers low because they feel that they detract from the photoscenery they are using.
Like Steve Wilson says, there is an element of customer preference here. I think a lot of people will accept your reasoning behind giving less control over rendering sttinngs in the technical sense, but they will be less understanding about the inability to turn off features they just don’t like.
An option to adjust the trees is probably the big one for me. Some areas I love lot’s of trees but other areas I may not want as many and when using photoscenery sometimes I don’t want any trees.
Apart from that I am very happy with the decisions that are and have been made.
Keep up the great work Austin, Ben and Co
How are the trees getting _on_ the photo-scenery in the first place?
– If the photo-scenery is a base mesh, it would exclude our base mesh trees.
– If the photo-scenery is “.pol” based (and I strongly do not like this because of the performance problems, but I know it gets used) then the photo-scenery could and probably should exclude 3-d from orthophotos that need protection).
I do agree with Ben here. I am X-Plane customer since about 10.20.
I prefer to have less sliders and less rendering options unless they provide scalability to present hardware power or as long as they are not needed to fix the present issue in the simulator. Good example are deers and birds – I want them, but I need to turn them off right now as it is incorrect to see deer on big airport behind the fence or meet flock of birds on taxiway staring at my big noisy airliner approaching on them.
When I open new game, a lot of rendering options just confuses me. They are often poorly described and effect of changing them is usually unpredictable, negligible and/or invisible. Maybe 5 years ago it made sense to decide whether to render 10 or 100 cars but with today hardware it likely doesn’t make sense. Also, the sheer amount of options just adds complexity and very floating baseline which is not beneficial for customer.
I don’t want 3rd party aircraft/scenery developers to dictate me rendering settings to use justifying their inability to balance optimization/programming/performance in their products.
I don’t want 3rd party scenery developer to be able to force me use of flatten runways if he/she is not capable of developing airport with correct mesh. I see the point with online flying and although I don’t fly online yet, I will not care if other airplane is buried in the taxiway or hovering 10 meters above runway. It may be on different horizontal position anyway due to differences in sceneries. Hopefully future flight simulators will all have sloped runways by default.
For very special scenarios like development/debugging/instructional and teaching I am OK if whatever option is available, but it should not be easily accessible just to avoid little chaos we have with X-Plane 10 rendering options now.
Two comments on this:
“I don’t want 3rd party aircraft/scenery developers to dictate me rendering settings to use justifying their inability to balance optimization/programming/performance in their products.”
I strongly agree with this, and it’s _my_ job to make sure third party developers have the tools to meet their performance targets without having to dictate rendering settings. E.g. a third party scenery pack should have the same property of scaling performance with settings that we (LR) have with our default scenery.
The ability to let the author pick flattening on a custom airport specifically exists to make it unnecessary for them to demand a particular flattening setting in the UI, which was the -only- choice for third party scenery in 10.40.
“I don’t want 3rd party scenery developer to be able to force me use of flatten runways if he/she is not capable of developing airport with correct mesh.”
I do expect that a third party scenery developer can do that for the _one specific airport_ that the custom scenery pack ships for. There may be things going on in the airport that require this, and given that there are multiple base meshes for X-Plane, you can’t really demand that a custom scenery pack look good for -any possible base mesh you install-.
You make a good point about the trees Ben, but I wasn’t complaining myself. Just passing on a comment I have heard very often. In fact I have seen more complaints in various forums about not being able to turn down the trees than about the cars. The deer seem to be universally disliked, although one enterprising soul has made productive use of them by turning them into running security guards 🙂
That’s a nice mod…I’d like to see them turned into Daleks. 🙂
Hi,
Great, looking forward to be able to make the U-turn approach in the fjord at BIIS, Iceland again! (https://www.youtube.com/watch?v=Fv_c6vA8DXE)
But how is it progressing on the data side?
There are plenty of automatic scenery bug reports for missing runways because the new GNS/FMS can’t find some runways in the older apt.dat. (The magnetic poles wandering around…)
Even without these, and despite Philipp’s last modification for misinterpreting procedures ending in -W, there are still SID, STAR and APPCH that are present in the CIFP files, but not displayed on the GNS or FMS, only the runways with a point on the extended centerline (yes, bug report open).
Thanks, Marc.
Actually, I need a mouse whenever I use the ATC-window. I would like to ask, if it’s possible to use buttons or keys instead for navigating up/down and selecting. While it was possible to use up/down key and enter in X-Plane 10 this doesn’t seem to work in X-Plane 11 yet. Is this correct – maybe I haven’t found the correct name to set a key.
Thank you for your help and this great flight simulator!
Our plan, not for 11.0, but for the v11 run, is to make the entire UI accessible by commands, e.g. either the hat switch + joystick or keyboard keys. We have this on the mobile product and the UI code is shared, but we didn’t have time to do key bindings for all of the v11 desktop UI elements, but it is coming.
Yes, would be cool if every menu command could be accessed via a custom keyboard shortcut/ joystick button! 🙂
this is great – thank you very much
To keep my comment short, you guys are doing a great job. I’m getting BETTER performance from XP11 than 10.
Take care Ben.
I can say what you do is just unbelievable. I stumbled across X-Plane right before Christmas. Did not know anything like this even existed. I came in at a time where your at a juncture in your business building a new system from 10 to 11. I have flown for over 50 years in my little pipers. Now I can fly again, and no one can remove my license for a medical reason. You do give back to the flying community a great sense of dignity.
God Bless you all.
1000%
That’s great to hear, John! Stories like yours are why we do this. 🙂
Fantastic hats off to the dev team on XP11.
My current questions are,
1. ‘Is there a planned update to increase the font size of the waypoints on the ND? ‘ Right now it’s rather annoying having to squint/zoom to try and read thes.
2. This one is aimed at Ben mostly, and probably not the highest priority. in the v10 run, you mentioned about jetways and patching in dynamic functionality to the sim/WED. Any chance we’ll see this during the v11 run? I only ask as I highly believe in ‘add in functionality now, and less updates by everyone later’ concept. Particuarly with the introduction of ground-service traffic and everybody rebuilding their scenery files, might be simpler to get this out of the way?
Many Thanks and all the best!
Hi,
If you’re asking about dynamic jetway support in the apt.dat file: we have intentionally _not_ put features into WED that aren’t supported in the sim. We used to do this a bit and I am trying to back away from it.
The problem with being “forward looking” with the scenery creation tools and letting them get ahead of the sim is that no one has any idea what their data in WED is _doing_ if they can’t test it in the sim. We run a risk of:
– The data design we picked not actually being useful in practice. Without having the entire ecosystem built, it’s virtually impossible to predict this. (The incredibly stupid Bezier curve encoding that we are stuck with in the apt.dat file format, one of my big file format mistakes at LR, is the result of designing the format on paper and shipping it first, then creating practical editing tools only after the format was locked down. If I had had WED, the mistake in the format would have been totally obvious.)
– Users entering junk data. We find the reliability of data across all parts of the X-Plane ecosystem is entirely a function of how heavily it is used and how easily it is validated in the sim. So we’d rather not have the data and then when users add it later, we know it’s good.
should we be seeing improvements to the FPS with NVIDIA cards yet? A GTX1050 isn’t giving good results for me even with low settings.
That’s pretty strange – NV cards should just work, although what you see may be specific to your system. E.g. if you drive a 4K monitor with a 1050 you may not get awesome results.
Nope. Standard monitor 1920×1080. Single screen.
Framerate is less that 30 parked up on the apron. Compared to XP10 on the same machine its pretty crap.
Windows 10
i3 with 16GB ram
It is very easy to get XP10 low on Fps, maybe your settings are way to high. With my gtx770, i5 3570k @4.5ghz, I have regarly fps between 20-30 at high density regions. For me this is not a problem. A flightsim is not a shooter.
I mean both XP10 and XP11
Yup, seeing quite a difference between Windows and Linux. In Windows I see 25-30FPS on a consistent basis with sliders one notch off max. In Linux, on the other hand, the very same settings produce 50+FPS. This is on the very same hardware each time (identical, swappable drives). I’m also seeing an old bug (will be filed shortly) crept back in in Linux. All planes are being followed by a huge shadow, visible in external views from certain angles.
Thats interesting. Maybe I’ll look into dual booting my machine to run Xplane on Linux. What disto are you using?
I run Gentoo. I’ve spent copious quantities of time optimizing for performance. Don’t know how the other distros perform.
“we found the cause of poor performance on some AMD hardware”
Ben, this will FIX the FPS drop caused on AMD hardware, when turning ON the avionics or using default FMC?
I also answered support e-mail with my testing.
Yes.
It will be available to X plane 10 ?
Gracias
after beta 8, it’s still present the FMC VNAV error ( unable flpn alt), and it can’t be deleted. Is it a common problem or only mine?