Why performance differs per map?

[Panzerlang]

Having gone back to 1941 in PWCG and being on the Moscow map I've noticed what was apparent before, the micro-stuttering that's not present on Kuban (even with a 1080Ti vs the current 3080Ti). I think the Stalingrad map is the same.

Is it simply their greater size that causes the issue or were they made with older tech?

[Gambit21]

I’ve never had stuttering on Moscow with my vanilla 1080. In fact it always performed better than Kuban. After upgrading my rig (but keeping the 1080) they’re about the same, however if pushed with many units etc Moscow will still pull ahead I think.

[kendo]

Same here. Moscow and Stalingrad were always the easiest to run for me. Kuban a little harder and Bodenplatte the heaviest.

[Panzerlang]

Oh, that's weird. I'm going to mess around with some PWCG settings and see if I can isolate the cause.

Thanks guys.

[Goffik]

Too many AI flights clogging up your CPU maybe? Like the others, when I was on a 1060 Kuban always ran noticeably worse than the previous two maps. So much so that I've never really flown on it even now, despite having a better PC.

Edited March 10, 2022 by Goffik

[LizLemon]

Its mostly the 3d water.

[Gambit21]

1 hour ago, LizLemon said:

Its mostly the 3d water.

 

Kuban always performed better for me over the water. Also the water is not 3D. It’s a procedural animation of a normal map on a 2D plane.

[LizLemon]

7 minutes ago, Gambit21 said:

 

Kuban always performed better for me over the water. Also the water is not 3D. It’s a procedural animation of a normal map on a 2D plane.

What are you talking about? Its a 3d mesh.

[Gambit21]

3 minutes ago, LizLemon said:

What are you talking about? Its a 3d mesh.

 

It’s an animated plane - which yes is technically a mesh, but there’s no volume to it.

It’s 2D

 

There’s no subsurface scattering, or refraction etc. Underneath the 2D plane is an un-animated plane.

Both are opaque. There’s  no performance hit to be seen here for this reason.

[IckyATLAS]

I also perform better on the water in Kuban.

[Gambit21]

 

Yeah not sure what it is with Kuban/performance  (trees?) but it’s not the water tech.

 

It has improved however due to Devs tweaking something.

 

 

Edited March 10, 2022 by Gambit21

[LizLemon]

32 minutes ago, Gambit21 said:

 

It’s an animated plane - which yes is technically a mesh, but there’s no volume to it.

It’s 2D

 

There’s no subsurface scattering, or refraction etc. Underneath the 2D plane is an un-animated plane.

Both are opaque. There’s  no performance hit to be seen here for this reason.

There is no volume to any of the 3d models in this game - or pretty much any game. I don't think you really understand what you are talking about. This is not 2D.

 

[Gambit21]

13 minutes ago, LizLemon said:

There is no volume to any of the 3d models in this game - or pretty much any game. I don't think you really understand what you are talking about. This is not 2D.

 

 

No disrespect - but you’re arguing with a 3D game/model developer.

 

Yes one of us doesn’t know what he’s talking about.

 

 

 

 

[LizLemon]

Just now, Gambit21 said:

 

No disrespect - but you’re arguing with a 3D game/model developer.

 

Yes one of us doesn’t know what he’s talking about.

 

 

Same is true here.

 

Please tell me about how the water isn't 3d when its made up of hundreds of polygons. Or how subsurface scattering requires 'volume' in a 3d model rather than 2 dot products.

[Gambit21]

Just now, LizLemon said:

Same is true here.

 

Please tell me about how the water isn't 3d when its made up of hundreds of polygons. Or how subsurface scattering requires 'volume' in a 3d model rather than 2 dot products.

 

If I have to explain any of that to you - then I’m not conversing with anyone who knows his elbow from a hole in the ground where this subject is concerned. You can have a 2D plane made of any number of polys - it’s still 2D. Not to mention I’ve investigated the water mesh and animations up close, underneath etc myself a few years ago.

 

Unless something has changed since then - and I don’t remember a “new water tech” announcement.

 

 

[LizLemon]

Just now, Gambit21 said:

 

If I have to explain any of that to you - then I’m not conversing with anyone who knows his elbow from a hole in the ground where this subject is concerned. You can have a 2D plane made of any number of polys - it’s still 2D. Not to mention I’ve investigated the water mesh and animations up close, underneath etc myself a few years ago.

 

Unless something has changed since then - and I don’t remember a “new water tech” announcement.

 

 

Its not a plane though, and I don't know why you think this when you can clearly see the waves clipping through the pier.

[Gambit21]

20 minutes ago, LizLemon said:

Its not a plane though, and I don't know why you think this when you can clearly see the waves clipping through the pier.

 

Because it’s a procedurally animated  2D plane. Think of a piece of paper, only with  0 dimension in the Y axis (thickness) now wrinkle it.

 

That plane now has wrinkles, but still only 2 dimensions.

 

(my own brain is remembering recalling what a normal map is and isn’t)

 

A normal map provides the illusion of geometry. In the case of the water in IL2 it the plane is actually deforming via the procedural animation (in other words, NOT utilizing a normal map) this is probably the case as the waves do have Y height as you mentioned. However it’s still a 2D plane.

 

3D water would have Y axis depth, with light bouncing around (subsurface scattering) and if really fancy (read resource hog) refraction.

 

However subsurface scattering would require tremendous resources on its own. Plus you’d be able to see the bottom/seafloor in the shallows, because it would have transparency. If you don’t have transparency then there’s no point in the Y axis dimension (thickness/depth) and you might as well just use a 2D plane - which is what they did.

 

 

Put a sub in the water (with waves) and move it’s external camera around.

 

If you’re seeing anything other than what I’ve described then a huge, unannounced change took place.

Edited March 10, 2022 by Gambit21

[LizLemon]

Just now, Gambit21 said:

plane is actually deforming via the procedural animation (in other words, NOT utilizing a normal map) this is probably the case as the waves do have Y height as you mentioned.

 

Displaced geometry is not a 2D plane. If that was the case then just go ahead and claim the terrain is 2D as well because that is a mesh displaced by a heightmap as well.

 

This is most certainly not 2D.

4 minutes ago, Gambit21 said:

 

However subsurface scattering would require tremendous resources on its own. Plus you’d be able to see the bottom/seafloor in the shallows, because it would have transparency. If you don’t have transparency then there’s no point in the Y axis dimension (thickness/depth)

Can you explain why geometry needs 'volume' to create subsurface scattering instead of just using NdotL and NdotV?

15 minutes ago, Gambit21 said:

If you’re seeing anything other than what I’ve described then a huge, unannounced change took place.

I guess you edited your post since I replied.

 

Water on Kuban has been like this since day 1 of that maps release. Nothing has been changed.

[Gambit21]

Land is opaque, water is transparent. Thus claiming “3D water” is quite different...and me needing to point this out is again very telling.

 

The water material has ZERO Y dimension, even though the water PLANE is displaced in the Y.

 

In order to have SSS you have to have an actual “sub-surface” again me needing to point this out tells me who I am, am not talking to here. Not to mention there is no SSS apparent, nor can there be.

 

I’m done.

Good day to you sir. 

[LizLemon]

5 minutes ago, Gambit21 said:

The water material has ZERO Y dimension, even though the water PLANE is displaced in the Y.

So what, the CPU is feeding the GPU geometry. The GPU is rendering waves with actual height. You can try to argue semantics all you want, but it is a 3D mesh.

 

By the way this game isn't using a material system.

6 minutes ago, Gambit21 said:

In order to have SSS you have to have an actual “sub-surface” again me needing to point this out tells me who I am, am not talking to here. Not to mention there is no SSS apparent, nor can there be.

I've written shaders that use subsurface scattering. You don't need any "sub-surface" - nor does such a think actually exist as far as the GPU is concerned. Its view angle, light angle and maybe some textures that are used. Do you want me to link you to papers or demos proving this?

 

So this isn't a completely worthless post, here is a simple mod that disables the 3D water on Kuban Autumn. I have a bit higher FPS with water disabled.

https://www.mediafire.com/file/1w700st3wlmtcsr/no_water_kuban_autumn.zip/file

[Gambit21]

Fine, we’ll call it ‘semantics’

 

That said you’re talking game engine - which I’d a fool to not admit is what matters here regarding shaders and such.

 

Back to the original point - performance over water on Kuban is higher than over land for a reason.Turning off the water animation  - no surprise that helps...as would turning off trees even more so.

[Panzerlang]

Wow, an almost metaphysical debate going on here. 2D or not 3D.

Sorry Bard. ?

[Gambit21]

...and probably pointless.

From a reflection standpoint the animated “3D” water is in fact 3D.

 

So we were talking past each other to some extent.

[Irishratticus72]

"Can't we all just get along"? 

Said no one here, ever. 

[Voidhunger]

I found it interesting

[Art-J]

It's the same in physics. The average layman on the street knows only 3 dimensions (well, 4, if he paid more attention in high school when hearing about relativity theorem), while the qualified physicist will say there are more. Both guys are right.

 

I'm not gfx dev, so for me anything not glass-table flat is "3D". But if 3D means something else and deeper for people "in the industry", so be it. I'm not going to loose any sleep over it ;).

[Panzerlang]

So it was the simple config in PWCG, I had the ground objects set to Med. Put them on Low and now all good. 

[Yogiflight]

Isn't the issue for the Kuban map the caucasus mountains in the south east, and for the Rheinland map the many large towns with plenty of big houses, while the towns on the Russian maps mainly exist of small huts and lots of trees, except on the Moscow map some larger towns like Rshev, which are modelled quite nicely? I remember back in the days of BOS, flying in low altitude over the town of Stalingrad, was a huge problem for the PC, I had back then.