8 hours ago, scotchegg said:

Given MS / Asobo’s own publicly stated timelines, that seems like a strong possibility. I’ve often wondered exactly when and how Dovetail found out about MSFS and if that was an influence on their decision to pull the plug on FSW.

The fact that MS have given DTG the task of bringing MS Train Simulator to Game Pass deals were done most likely.

1 hour ago, MrFuzzy said:

I agree, look at P3D: a solid 25% avg increase.

To avoid crashes wouldn't be sufficient to limit the graphics sliders depending on the VRAM and other system resources?

Yeah, I can back the sliders down, and won’t crash, maybe, if I’m lucky. Or I can back them down a good way so that I’m safe.  But it’s dumb.  I don’t want performance just for performance’s sake, I want it so I can run better visual fidelity.  If I have to turn off visuals that my computer has plenty of power to run smoothly, just so I can run the bloody thing at all, what is the point?

Based on this article, it seems that development started in 2017, around one year before Flight Sim World was shut down.

https://www.theverge.com/21347809/microsoft-flight-simulator-2020-preview-interview-hands-on

Quote

In 2017, Neumann got in touch with Asobo and asked if they’d be up for giving the project a shot. Using data of Microsoft’s home city of Seattle, which Bing Maps has rendered down to five-centimeter resolution with photogrammetry, Asobo took a few weeks to put together a demo of a Cessna flying downtown. Neumann then showed it to Phil Spencer, a VP at the time who is now the head of Xbox.

Quote

Neumann is quick to point out that Asobo wasn’t starting from scratch. “I sent them the Flight Simulator X engine,” he laughs. “And the great thing was, they integrated the sim that we had one piece at a time. And pretty much from day one it worked. You know, sometimes you work on a product and for the first two years, nothing works. In this case, we could fly from literally day one and just bring up the different systems, which I think was key.”

Perhaps this is what influenced the closure of Flight Sim World, when Microsoft realised they could have an awesome opportunity in their hands.

4 hours ago, MrFuzzy said:

I agree, look at P3D: a solid 25% avg increase.

To avoid crashes wouldn't be sufficient to limit the graphics sliders depending on the VRAM and other system resources?

 

While I agree there have been some slight performance improvements with v5, this video is not a reliable comparison when you look a bit closer. There's a sequence starting at around 6:40 where you can clearly see v4 draws more autogen (see the bottom left quarter of the screen) and has slightly sharper textures in the distance. I'd love to see some improvements in v5.1, but so far the cons outweigh the pros from where I'm standing...

If I'm reading all this right, you have to run lower settings to keep from getting VRAM OOMs with DX12.  Running with lower settings will increase your frame rate.  Could we just skip the DX12 part and get increased frame rates by lowering our settings? 😄 

Hook

The problems is there are multiple flavors of DX12. to include the DirectX 12 Ultimate. 

3 minutes ago, KenG said:

The problems is there are multiple flavors of DX12. to include the DirectX 12 Ultimate. 

Ah... forgive me but I'm just a simple retired professional programmer with over 40 years of experience and all I really want to do is install a program and run it.  I don't care to jump through hoops to make something work, or to have a lot of specialized knowledge just to get the thing to run.

Is this going to end up being a "conspiracy" to render all our existing graphics cards obsolete and force us to buy new and very expensive ones?  

Hook

40 minutes ago, LHookins said:

If I'm reading all this right, you have to run lower settings to keep from getting VRAM OOMs with DX12.  Running with lower settings will increase your frame rate.  Could we just skip the DX12 part and get increased frame rates by lowering our settings? 😄 

Hook

No. A lower-level api will make so much work for developers to do (don't know frankly about dx12 specifically, but in Vulkan the hello world looks like hell to deal with)

BUT, you gain control over so much stuff. Means for us, fixing stutters and memory paging, once a lower-level api is implemented, can be really straight forward.  

If implemented succesfully (i know the taboo here, but x-plane Vulkan iteration is phenomenal), it's the stablest thing you can currently achieve when it comes to perfroamnce vs graphics.

Simulators tend to be CPU bound, this is the solution, as those api's gives you control to have a much more efficient code + threading safety.

So, it's to other way around, it can make you get with higher settings and overall graphics, while maintaining better performance and stability of you scene.

And frankly, you can't really "feel" a flight model in a simulator once you get micro-stutters.

 

Edited September 5, 20205 yr by mtaxp

44 minutes ago, LHookins said:

If I'm reading all this right, you have to run lower settings to keep from getting VRAM OOMs with DX12.  Running with lower settings will increase your frame rate.  Could we just skip the DX12 part and get increased frame rates by lowering our settings? 😄 

Hook

I give for granted that the video comparison was made with the same settings (seriously)

13 hours ago, LHookins said:

Not true ray tracing like you'd get with DX12, but in one of the videos Asobo said they'd implemented a version of it for MSFS.  Let's not get into semantics here, if it looks like ray tracing and acts like ray tracing, then you might as well call it ray tracing, even if it is emulated.

Hook

No raytracing in msfs.

And no need for it in a global scenery, materials and pbr will achieve the same look in *open scenes* when you have enough light exposure AKA a global scenery in our case.

What it will solve, is the fact that cockpits tend to be darker than what they should; raytraced aircraft interiors can then simulate leaking of light to the interior and then light bounces inside and it results in much better exposure as in real life.

So a hybrid engine (rasterization as global and raytracing for specific effects and closed scenes) can be done, this is the actually what most modern games do.

Edited September 5, 20205 yr by mtaxp

6 minutes ago, MrFuzzy said:

I give for granted that the video comparison was made with the same settings (seriously)

Oh, it may well have been! 😄   Early on they were saying that settings weren't the same in v5 as they were in v4, implying that you'd get the same effect by running lower settings in v5.  I wouldn't know.  I just want to avoid VRAM OOMs without having to buy a new video card.

Hook

4 minutes ago, mtaxp said:

What it will solve, is the fact that cockpits tend to be darker that what they should; raytraced aircraft interiors can then simulate leaking of light to the interior and then light bounces inside and it results in much better exposure as in real life.

Ok, but I'm not having that problem and in any case it can be solved without ray tracing.  

DX12 ray tracing may solve a number of problems and provide improved performance, but are there other costs involved?

Hook

17 minutes ago, LHookins said:

Ok, but I'm not having that problem and in any case it can be solved without ray tracing.  

DX12 ray tracing may solve a number of problems and provide improved performance, but are there other costs involved?

Hook

What do you mean by costs? It'll just result in better graphics, this is how things work, improving tech to achive even more graphics and performance.

You say you don't have this problem  well you don't notice it so you dont really care , but many cases where lightning is just wrong, mainly in closed scenes. That is a fact, right now as without raytracing you are limited to space screen.

Put your cockpit in a condition which has no obvious direct lightning from sun, cockpit is way too dark.

However i do agree it's not urgent, few more years until most people will have fast raytracing capeable graphics cards.

You can bake raytracing on textures, this the best thing to do for cockpit instruments lightning and any other static lights. For dynamic lighning bounces in closed conditions you need to really ray-trace.

Edited September 5, 20205 yr by mtaxp

9 minutes ago, mtaxp said:

Put your cockpit in a condition which has no obvious direct lightning from sun, cockpit is way too dark.

I've been in conditions that not only had no direct sunlight but were under overcast skies, which many people claim makes everything far too dark.  I have no problems, at least with the analog gauge aircraft I fly.  For glass cockpits your mileage may vary.

The only time I've had to use cockpit lighting is during dawn or dusk, and I'd have to use the same lighting in a car in the same conditions.

The only difference between what I do and what everyone else does is that I have my zoom level set to 90 and I use TrackIR which makes use of such a zoom possible.  Overall brightness of the scene adjusts depending on if I'm looking inside the cockpit with no external view visible or looking outside with none of the cockpit visible.  It works like the human eye adjusts.  I have no doubt that the default 50 zoom, with both internal and external views visible at the same time will make the cockpit look too dark, and ray tracing is not going to help with that.  Perhaps look to your HDR settings, but the result will be a bit less than optimal.  You could probably get the effect you want by turning on interior cockpit lighting and it would look better.

Hook

3 hours ago, LHookins said:

I've been in conditions that not only had no direct sunlight but were under overcast skies, which many people claim makes everything far too dark.  I have no problems, at least with the analog gauge aircraft I fly.  For glass cockpits your mileage may vary.

The only time I've had to use cockpit lighting is during dawn or dusk, and I'd have to use the same lighting in a car in the same conditions.

You dont seem to understand what im talking about.