Are clouds the best ever now? I'm impressed.

[Cpt_Piett 4,651]

17 minutes ago, Noel said:

Kind of a bold statement to make considering this was done after SU5 and after METAR integration.  

That is indeed a very bold statement. Well, they are in the process of smoothing out the differences between metar and meteoblue "zones", which is a good thing. I'd very much like to know what Neumann means by "blow people away". I guess we'll all find out soon enough.  

Edited August 3, 2022 by Cpt_Piett

[March Hare 720]

@NoelI didn't say they couldn't make some kind of improvement. In fact, I literally said in another post that Seb said they could/might try some things with clouds, once other things are done. So, no, I haven't said they can't, or tried to predict exactly how far they can go with it, because I'm not involved in its development.

But just because we might have headroom now with our CPUs or GPUs doesn't mean it can or will be used on cloud depiction (maybe it will be; I'm not saying it definitely won't). We don't know what else will take up that headroom before they return to cloud detail and quality.

But where there is a lot of headroom because it's, say, an i9 or RTX30xx card, that is beyond recommended specs at launch, there is a possibility that, just because there is headroom on that hardware, it won't be tapped into. I am only offering a cautionary tale (opinion based on industry trends) that headroom available doesn't mean it will actually be used, because you seem to think if it's there it definitely will be. Unfortunately that isn't often the reality of development.

So, I've mostly been talking in hypotheticals about the extent to which they will develop specifically for hardware that came after the recommended spec hardware it was originally designed for (i7 and RTX20xx cards). I just thought it would be interesting, to be honest. I don't absolutely disagree with you, I've just been getting the impression you think they will keep significantly redeveloping the code when new hardware is introduced, and that doesn't realistically happen -- usually. They will likely only go so far with rewriting code to push the sim further. They will do it, but there will be some limitation and end point (that doesn't really take into account hardware newer than the initial development period).

As my brother said:

"There is always a goal to perform best on some recommended hardware [launch specs]. It's rare that better or future hardware is considered ... It's not worth the effort and, more importantly, the cost."

But, you know, Microsoft could break trend. Ultimately it's up to Microsoft what Asobo do.

So, let's look forward to seeing what happens.

[Tuskin38 6,506]

11 hours ago, steve310002 said:

They are not going to upstage the Xbox version with better clouds in the PC version.

The PC version already has better graphics than the X-Box version. The X-Box IIRC runs on a combination of high and ultra settings.

Edited August 3, 2022 by Tuskin38

[FishBowlMan 56]

42 minutes ago, Cpt_Piett said:

I'd very much like to know what Neumann means by "blow people away".

Oh don't start people off about the gusts again 🤣

[steve310002 1,198]

29 minutes ago, Tuskin38 said:

The PC version already has better graphics than the X-Box version. The X-Box IIRC runs on a combination of high and ultra settings.

I'd say subtley different though. There are not massive visual differences.

[JRBarrett 3,071]

1 hour ago, Cpt_Piett said:

 I'd very much like to know what Neumann means by "blow people away". I guess we'll all find out soon enough.  

The 100-200 knot surface winds that have manifested in Live Weather a couple of times in the last week!😄

[Krakin 5,767]

1 hour ago, steve310002 said:

I'd say subtley different though. There are not massive visual differences.

Dude just give up. You're making yourself sound silly the more you dig your heels in on this issue.

[Sky_Pilot071 580]

For me I'd like to see some realistic stratus / cirrus formations at altitude that may or may not be engine generated.  IOW pop a few bitmaps up there and we would be good to go.

Cheers

sp

[Noel 4,502]

11 hours ago, March Hare said:

in cloud depiction and quality may entail a more complete code rewrite

You make some wonderful points and as for a complete code rewrite that seems an over the top POV.  It seems doubtful that the code used to get progressively more density/quality between the four user-selectable levels of Low, Medium, High and Ultra somehow stuck at its theoretically maximum already, at Ultra.  Doesn't comport to me. If you take a look at its predecessor FSX they included controls that were far out of reach of the highest end PCs and for a very long time.  With P3D 4.5 where I stopped at I could only run with very mediocre settings despite the 9900K and 2070 super.  So in that realm in several areas complexity stayed way ahead of the hardware and I thought that was genius as it gives the platform long term staying power.  Can't imagine tuning into MSFS say in 2026 only to see almost no change but a few modest trinkets added, like waves or some other gimmick, while my new PC can run circles around it.  Very odd strategy for a 10yr project.

[steve310002 1,198]

9 hours ago, Krakin said:

Dude just give up. You're making yourself sound silly the more you dig your heels in on this issue.

Okay. Thanks for telling me what my opinion should be. Of course no one around here is entitled to an opinion of their own anymore without being told off for it. I had forgotten, dude.

[MrFuzzy 2,118]

11 hours ago, steve310002 said:

I'd say subtley different though. There are not massive visual differences.

When you look at 1080p compressed Youtube videos, yes. 

[FishBowlMan 56]

9 hours ago, Noel said:

So in that realm in several areas complexity stayed way ahead of the hardware and I thought that was genius as it gives the platform long term staying power.

Back in the day, I seem to remember a general rule of thumb was that hardware capabilities would pretty much double every 18 months or so. Most games were released on the assumption that the ultimate graphics performance would need "next year's" hardware.

The majority of games that I bought at the time, in different genres, I could pretty much run at medium settings, and would have to wait until I could afford to upgrade my hardware to be able to bump up the graphics settings.

Even then, I could only afford to upgrade to the "interim" stage... for example when the Pentium was released, I had a 386, and could only afford a mid tier 486, so I could get better graphics, but still not get the maximum.

[Ray Proudfoot 8,956]

A reminder... everyone is entitled to their opinion whether you agree with them or not.

[rka 916]

8 minutes ago, Ray Proudfoot said:

A reminder... everyone is entitled to their opinion whether you agree with them or not.

I'll think about that next time one of my comments is hidden, I receive a time off or a thread gets closed because "it has run its course" 😉

[March Hare 720]

11 hours ago, Noel said:

You make some wonderful points and as for a complete code rewrite that seems an over the top POV. 

I said more complete, meaning a significant change rather than total. Sorry, I should be clearer.

When I refer to cloud detail and quality, I'm not limiting my outlook to resolution (which can be a memory problem in itself to overcome), but thinking more in terms of overall cloud depiction too, i.e. wispy and feathered clouds, maybe hurricane and tornado vortices (they did say they plan to blow us away haha). I don't know for sure what's required for any conceived change in depiction and/or quality, of course. I really am only being hypothetical.

But in terms of code changes, say, if they employ nearest-neighbour interpolation vs trilinear interpolation, that's a significant enough code change. I’d be unsurprised if they are currently using nearest-neighbour, because it takes significantly less rendering time than trilinear and you can make out the individual voxels in a volume, which we can if we look hard enough or zoom in (unless you blur the image via TV or GPU sharpening settings, then you can evidently start to soften the appearance); trilinear softens the appearance to hide the individual voxels, but is more resource expensive and slower to render.

Also, a switch between using a more traditional voxel grid vs a sparse voxel grid is going to be a significant code change, and exactly how they implement a sparse voxel grid could be done in different ways. Although, I suspect they already are using sparse voxel grids, because again it makes sense for resource efficiency.

By the way, the NVIDIA fire and smoke tech demo posted earlier by @Kassu62uses a dynamic sparse voxel grid. It’s all essentially the same base technique (voxel grid-based) and principles/methods that Asobo uses for the clouds in MSFS already, with obvious differences to make the fire and smoke vs cloud appearance. Despite what that linked article suggests, the rendering isn’t revolutionary -- at least not any more -- and it isn’t a feature of DX12; what they mean is the engine running in the demo is implemented using DX12, but the Flow library is Compatible with DX11 and Vulcan too. Flow is a library to make fluid dynamics and related effects easier to implement, to make life easier for developers, not a new revolution. I mean, that tech demo is from around 2017...

As I say, let's see what Asobo can pull off and hopefully pleasantly surprise us.

 

Edited August 4, 2022 by March Hare