Interview With Orbx’ John Venema

[edpatino 1,959]

1 hour ago, GHarrall said:

Also, I have never seen P3D hit 200 FPS no matter what settings you turn down.

I suspect that anyone would do that, otherwise they would get a "naked" simulator.

Cheers, Ed

[Matthew Kane 1,343]

1 hour ago, GHarrall said:

Have you tried it? The buildings in AFS2 certainly aren't simple polygons with photo images on them. Also, the Dash 8 400 that is supplied as a free, default aircraft is very well detailed. I was shocked at just how good it is. Its not quite the MJC q400 but as a stock aircraft you would be surprised with what it can do.

I keep hearing that it will slow down as the complexity ramps up. I have yet to see any evidence of this being the case. So perhaps we get 100 FPS instead of 200?? :))

Also, I have never seen P3D hit 200 FPS no matter what settings you turn down.

Yes I have tried it, AFS2 works great, Here is the difference.

P3D going back to the engine from the guys at Microsoft is using 5 layers, Ground Textures, Land Class which is assigning the Ground Textures, Vectors which is assigning other things like Roads, Railroads, Lakes, and various other things like Beaches, Golf Courses Parks Powerlines, Marshes etc, Then you have your terrain mesh and on top of that Autogen, and Autogen is a performance hog, also other things like advanced weather systems, ATC and AI and other third party items hogging resources.

AFS2 is simply laying out ground photographs over top of terrain mesh and eliminating the need to vector scenery items as they simply appear in the ground image, Then instead of Autogen they are placing polygon buildings and assigning photographic images to them, some generic and some to represent a real building, they are also placing other items like trees or pretty much anything they want. This is the reason for the performance gain is the CPU and GPU are not calculating all those data points to generate scenery items, it is all in the photographic images. Also AFS2 is missing other systems found in P3D like the weather, ATC and AI etc

Reason for the way P3D/FSX works was MS could create the entire globe quicker as they were placing basic data points into the sim to create a global representation in the sim, some areas better then others but they could mass produce the entire globe faster this way, but it made the sim not perform as well. AFS2 will take longer to create the globe as it is more labour intensive but you get better performance. You can also do the same scenery methods in P3D and turn off all those vector points and just have photographic scenery and you will get much better performance in it is as well.

To get 200 fps in P3D you would have to strip out all scenery and create a scenery area similar to the methods in AFS2 and fly around in a basic aircraft, you are basically recreating the same environment of AFS2 in P3D, that would be the true test, also turn off other things like the weather engines, ATC and AI etc that P3D has and AFS2 does not have, at that point the two would be at par with each other

Edited July 16, 2018 by Matthew Kane

[edpatino 1,959]

I've been fying all Microsoft simulators almost since the beginning of the series, FSX and four years ago, by the times of P3Dv2.4, moved completely to P3D (uninstalled all FSX stuff). Last year I started using Aerofly FS 2, and will never regret that decision. Currently I fly in AFS2 70% of the time.

No simulator is perfect, certainly AFS2 currently lacks complexity on the aircraft, even though you can do a lot of things with the existing fleet (the recent Vulkan update added a substantial amount of functionality to them), and several other important features that will hopefully come later. To comment a bit on my personal experience, when I need to do some flying with complex or study-level aircraft I simply go back to P3D for the 30% of the time, to immediately realize that cartoonish look that makes me go back again to AFS2, .... and then the cycle begins again one more time 😂!.

Btw, my experience so far with AFS2 is that without Vsync enabled, and with FPS locked at 120, my worst case scenario is 80 - 100 FPS over very complex areas. I can tell that I've never been as low as 45-60 FPS, even when I turn Vsync ON, and rates are kept at 60 FPS without perceptible variation. And all that happens both on my laptop and desktop PC.

Cheers, Ed

 

Edited July 16, 2018 by edpatino

[HiFlyer 9,336]

In all of this, nobody has brought up Jv's mention of further plans for XP (And its going to Vulkan too, where it seems everyone is hoping for significant FPS increases)

 

[Matthew Kane 1,343]

Vulkan API would be a game changer

[HiFlyer 9,336]

5 minutes ago, Matthew Kane said:

Vulkan API would be a game changer

Potentially, but I'm not completely sure. Certainly its made Aerofly faster, but the changes to the (aerofly) OpenGL engine made it just as fast (faster, actually) So............

Maybe its more of a future thing: once they have the tech working, then they can refine it to take best advantage.

[Matthew Kane 1,343]

2 minutes ago, HiFlyer said:

Maybe its more of a future thing.

Yes a future thing, we are more likely to see a major title adopt it before Flight Sim does but lower voltages means you are raising the envelope performance wise rather then pushing hardware further, which is the benefit of it.

Edited July 16, 2018 by Matthew Kane

[HiFlyer 9,336]

47 minutes ago, Matthew Kane said:

Yes a future thing, we are more likely to see a major title adopt it before Flight Sim does but lower voltages means you are raising the envelope performance wise, which is the benefit of it.

All I really want out of X-plane, is for the VR to work better. I got the impression that was a big part of the push for Vulkan.

The thing I struggle with currently is the generally unacceptable sim performance on my Oculus. 

Since I'm very into that, its a good thing (for me at least, since I know some people couldn't care less) that all sims seem to be racing to enhance their VR experience.

Looking around, I know a lot of the aircraft guys are making a move, but its good to see that some scenery companies like Orbx also have their eye on the (VR) ball.

[Guest]

3 hours ago, HiFlyer said:

Doesn't he tend to keep his frame-rates locked?

I've never been able to get Vsync to work in AF2 (from day 1 of release), but I have not tried external options (NI/NCP) only the native internal option for Vsync.  But yes, on my system AF2 Vulkan Beta has doubled my FPS.

I'm hoping iPACs next step is a weather engine ... as much as people don't seem to like view facing "sprite" clouds (they still work well, don't require huge processing resources, and work on a wide range of hardware).  I haven't seen anything as good as HiFi's AS4 in any "official" simulator yet.  I heard rumors iPACs are looking at TrueSKY, I'm actually NOT a big fan of TrueSKY but my only exposure to TrueSKY was in FSW ... perhaps it wasn't implemented well in FSW?  In FSW, fog didn't look that great and a single cloud layer and cloud motion looked more like a canned WARP effect applied to the cloud sprites and overcast resulted in low performance.  Don't take this the wrong, there were some great "moments" with TrueSKY, but there were many not so great moments.

I haven't seen anything as good as HiFi's AS4 + ENVTEX with great performance on multi-layer clouds (5 layers) and overcast days.  On the XP11 side, I do like how xEnviro weather reflects city lights on the clouds, that's a very nice effect when it works correctly (still some bugs to workout).

But it's hard to say what might impact FPS in AF2, they need to add dynamic lights (aircraft lights on the Q400 don't currently light up anything, not even the runway) and as most are aware that has the potential to drop performance considerably.  If you check the video's I created, you'll notice the shadows are processed in distance radius and there are some other shadow anomalies (looks like "Peter Panning") that might require some adjusts to how shadow maps are rendered which could double shadow rendering work to correct.

iPACs is certainly worth supporting, they have Orbx backing so scenery will come eventually, and they are progressing their SDK and visuals.  AF2 isn't a replacement for XP11 or P3D, it's an augmentation ... for VR users AF2 is able to hit those high FPS required for stutter free VR experience (even at all maxed out settings).

I've always felt that iPACs AF2 had more potential than DTG's FSW ... too bad they couldn't join forces.

Cheers, Rob.

[HiFlyer 9,336]

39 minutes ago, Rob Ainscough said:

In FSW, fog didn't look that great and a single cloud layer and cloud motion looked more like a canned WARP effect applied to the cloud sprites and overcast resulted in low performance.  Don't take this the wrong, there were some great "moments" with TrueSKY, but there were many not so great moments.

I actually had hopes...... Because I had seen TrueSky used in other programs, and knew that it could do better than what was showing up in FSW at the time. I was expecting that as they refined things, more and more of the potential would appear.

I also hoped that the apparently true volumetric nature of Trusky would give us our first glimpse of non-rotating clouds in VR, if FSW ever got that far....

I had actually been very encouraged by this thread:

I eventually went and downloaded the Project Wingman Alpha demo to see TrueSky running in a faster sim. Needless to say I would be happy to see it at least tried in X-plane, P3D or Aerofly which I think could easily shed some FPS for a really modern cloud depiction.

 

By the way! To TonyWob, I hope Barton and Orbx TE England make you a ton of Money!!!

[Guest]

35 minutes ago, Rob Ainscough said:

I'm hoping iPACs next step is a weather engine

I'm hoping so too but I am afraid that next step is still years away. Unless a company like HiFi would create something for AFS2. I've been asking IPACS for a weather engine since day one but so far all that IPACS has said is that they will look at it (read: think about it) in the future. It isn't even on their to do list right now...! The current crappy clouds are placeholders and they haven't even thought about replacing them with something better for the time being. Looking at how long they've been working on ATC already I don't expect anything weather related from IPACS before 2021/2022. Yes, I am serious here. Take note that I am a big AFS2 fan: it's the only sim I use right now. I love the performance, the lighting, the looks, the default planes (which are the best default planes of any sim imho) so I will support them all the way (I can't wait to give TrueEarth Netherlands a try in AFS2... next week maybe!!!) but the lack of weather is a big problem for me... Still, not big enough to let me install P3D. 😉

Anyway, that interview was a nice read, thanks for the link, HiFlyer!

[Guest]

TrueSKY is actually a mix of 2D facing sprites and "3D" clouds ... the 3D clouds are particle based.  It would be impossible for current and future hardware for all of the clouds in a scene to be rendered as particle based clouds in real time.  You could actually do this in P3D today with the FX system but it would be outside the standard weather system ... believe HiFi "cloud motion" effect is essentially particle FX projected in front of the VC ... but don't quote me.

https://docs.simul.co/unrealengine/Clouds.html

I believe 2D sprites are render based on distance from view point but are still "facing" sprites, 3D particle based clouds (similar to P3D's particle based FX system) can be control based on "virtual" distance to view point ... this is sorta similar to how SpeedTrees are implemented in P3D ... you get a facing 2D sprite at XYZ distance and then it transitions to an animated 3D tree as you get closer to it.

My guess is performance will vary based on definition of the 2D to 3D transition radius.  Particle based rendering can be extremely GPU intensive and very AA sensitive and I'd hate to imagine what dynamic lighting might do (lighting up a particle based cloud as you fly thru it or reflecting city lights etc., GPU easy for a sprite, not so easy for particle based).  It's in the implementation and what compromises a developer decides.  Lets hope Vulkan can accelerate particle based clouds.

Cheers, Rob.

[Paraffin 2,197]

6 hours ago, HiFlyer said:

In fact, I suspect as time goes by and more and more is added to Aerofly, (hopefully with nothing near the predicted precipitous FPS drop) this particular argument about achievable framerate will (gradually) give ground until everyone's completely forgotten about it because they're too busy enjoying themselves.

Maybe, but I still think there is "no free lunch" when you start adding more complexity.

An aircraft with only mid-level complexity like the Carenado PC-12 will noticeably drop my frame rate in X-Plane, compared to simpler GA aircraft like a Cessna trainer. And there is nowhere near as much going on with the PC-12 as there is in a modern airliner. Add a few cloud layers in SkyMaxx Pro and my frame rate drops another notch. Although, I can still manage to fly in the 30-40 fps range with most eye candy enabled. That is, unless I turn on tree and building shadows. Then it becomes unflyable on my current mid-range hardware. No free lunch.

It has always been the case in PC flight simulation that more complexity means a hit on frame rate. I don't think AFS2 has any magic pixie dust allowing it to achieve the currently high frame rates. It's just a very stripped-down sim, optimized for high frame rates by simply not modeling many of the things available in other sims.

We won't know how well AFS2 holds up with more depth and complexity until it actually arrives in the sim. 

[HiFlyer 9,336]

1 hour ago, Paraffin said:

I don't think AFS2 has any magic pixie dust allowing it to achieve the currently high frame rates.

I don't think pixie dust is required.

For instance, choose some popular game from about decade ago, then look at a modern iteration of that same franchise. Gta, for example.

What you'll see is....... more. More effects, more polygons, larger areas, higher resolutions, more details. (At the same or probably greater speed)

That generally applies across the technological board.

So why wouldn't a more recent sim not (similarly) take advantage of a decade of advancement in hardware and software development to have at least the potential to do more as well?

1 hour ago, Paraffin said:

We won't know how well AFS2 holds up with more depth and complexity until it actually arrives in the sim. 

Which is a good point. I think we should wait, withhold judgement, and see. Predictions about how slow the sim may or may not become seem premature.

Suppose for instance, that because of Vulkan, X-plane or Aerofly went sli. https://arstechnica.com/gadgets/2018/03/vulkan-1-1-adds-multi-gpu-directx-compatibility-as-khronos-looks-to-the-future/

How would that change our expectations of whats possible, and the available horsepower to our sims? How many immutably concrete assumptions would have to go out the window?

[Paraffin 2,197]

21 hours ago, HiFlyer said:

I don't think pixie dust is required.

For instance, choose some popular game from about decade ago, then look at a modern iteration of that same franchise. Gta, for example.

What you'll see is....... more. More effects, more polygons, larger areas, higher resolutions, more details. (At the same or probably greater speed)

That generally applies across the technological board.

So why wouldn't a more recent sim not (similarly) take advantage of a decade of advancement in hardware and software development to have at least the potential to do more as well?

If the point you're making is that a "more recent" sim should be able to do more than a sim like P3D or XP using a legacy code base, then I'm not sure I buy that (with apologies if that's not the point you're making).

All flight sims share a common need to render 3D graphics at high enough frame rates to be flyable, while also running a bunch of other stuff under the hood like flight models and systems simulation. The "no free lunch" rule always applies when balancing complexity and frame rates.

The fact that the code base of a sim like AFS2 is more recent than say P3D or XP, doesn't mean it has an inherent advantage. If anything, legacy sims have the advantage of more years working out optimization, while adding the features customers want, like dynamic weather, deep systems modeling, AI aircraft, programmed ATC, and so on.

It takes years to build that out. Just look at the state of ATC in X-Plane for example, still a weak area in that sim. By the time a "new" sim manages to include all those desired features, I can't see that it will be miles ahead of these other platforms in overall performance.

Quote

Suppose for instance, that because of Vulkan, X-plane or Aerofly went sli. https://arstechnica.com/gadgets/2018/03/vulkan-1-1-adds-multi-gpu-directx-compatibility-as-khronos-looks-to-the-future/

How would that change our expectations of whats possible, and the available horsepower to our sims? How many immutably concrete assumptions would have to go out the window?

It depends on the ease of integration. We can't just assume this would mean throwing our assumptions out the window.

For example, according to an older post in the dev blog, X-Plane doesn't support SLI because the frame rate would actually slow down when sharing rendering between two GPUs, due to the way graphics are handled in the sim. 

Even in the non-sim, gaming world, the use of multi-GPU setups has been fading in favor of single ultra-fast cards. That's partly due to developers not wanting the extra hassle of supporting it, and also the recent inflation of GPU pricing. For whatever reasons, it does seem to be a trend towards single cards now. I'm not sure Vulkan support for multi-GPUs is enough to reverse that trend. 

Edited July 17, 2018 by Paraffin