the future of Clouds .

[Jotaele]

I would like to see that in any flight sim in the future.
Isaw a workmate working in houdini with real 3d volumetrics clouds(in the visors) vased on voxels, so i did search for real time cloud engines on youtube, and i found that.
http://www.youtube.com/watch?v=pLfHDul5XGw

and this one.

http://www.youtube.com/watch?v=yhBhOPLIuDY

I think this is the future of cluds in flight sims.

[Jotaele]

Do they leave enough resources for the core of the game to use? I'd hate to suddenly loose features of my plane just to see pretty clouds.

I dont know, but it is worth a try.After all the used clouds use lots of resources too.

[guiltyspark]

honestly both tech demos looked pretty bad

[Cobra8472]

I have used both Nimble and SilverLining, and unfortunately- both are poor for flightsim usage.

There are many whitepapers available on the implementation and coding of clouds based on the simulation of light scattering, seen implemented in Wings of Prey, Ace Combat 5, HAWX, Rise of Flight, etc etc..

Here is the most recent example I can remember seeing, but there are many many third party and proprietary examples of this stuff..

http://www.youtube.com/watch?v=pnvBF5Nu5gw

[PeterPanPan]

This looks pretty amazing for FSX, so it must be possible to pimp up clouds for the flight sim genre.

The developer says the clouds are "Hi-definition, ultra-realistic photo-real 3D clouds (stratus, wispy, cumulus and highly realistic rendering of cirrus clouds as seen in the real world).

http://www.aerosoft.de/_php_projekte/_php_screenshots/screenshots.php?sp=fsx&p=rex

[Jotaele]

I have used both Nimble and SilverLining, and unfortunately- both are poor for flightsim usage.

There are many whitepapers available on the implementation and coding of clouds based on the simulation of light scattering, seen implemented in Wings of Prey, Ace Combat 5, HAWX, Rise of Flight, etc etc..

Here is the most recent example I can remember seeing, but there are many many third party and proprietary examples of this stuff..

http://www.youtube.com/watch?v=pnvBF5Nu5gw

Wow, that vid is outstanding.I think its a matter of time the flight sim will go in this direction.

Peter, rex its a great extension for FSX, it has raised the bitmap based clouds to its limits, based on photos .Im sure someone will make som mods for CloD soon , has the one that was made for il2, that becomes quickly in a must have. Anyway i cant talk about CloD until i saw the game ,only trying to share some thoughts and wishes but what i have seen in the videos are the tipical pluffy clouds that works great or bad depending from the camera angles or light conditions.

[Jotaele]

BTW, Cobra, what do you mean what you described like a poor usage?Do you mean graphically? ore process consuming? or implementation? I have read the yutube video descriptión and looks very promising.
trueSKY™ is a C++ library which generates weather system data and updates it in real time. trueSKY™ creates volumetric cloud data, and provides realtime access to that data via a lightweight API. The cloud system generates pure volumetric data, it is cross-platform and renderer-independent. The sample applications that come with the SDK show how realtime clouds can be rendered using the generated cloud data, in various graphics API's

[Cobra8472]

BTW, Cobra, what do you mean what you described like a poor usage?Do you mean graphically? ore process consuming? or implementation? I have read the yutube video descriptión and looks very promising.
trueSKY™ is a C++ library which generates weather system data and updates it in real time. trueSKY™ creates volumetric cloud data, and provides realtime access to that data via a lightweight API. The cloud system generates pure volumetric data, it is cross-platform and renderer-independent. The sample applications that come with the SDK show how realtime clouds can be rendered using the generated cloud data, in various graphics API's



Both SilverLining and Nimble are simply not good looking enough for Flightsim usage. They are very easy to implement and use, but both are very lackluster for the area of utilization required of a Sim.

They are, however, good solutions for sky rendering for a ground-based game.

I have personally not used TrueSky, alas from what I can see it is a rather decent alternative. There are, as mentioned, many many examples of solutions like this, and creating your own is not too difficult due to the amount of research and public material available on the implementation.

[Jotaele]

Both SilverLining and Nimble are simply not good looking enough for Flightsim usage. They are very easy to implement and use, but both are very lackluster for the area of utilization required of a Sim.

They are, however, good solutions for sky rendering for a ground-based game.

I have personally not used TrueSky, alas from what I can see it is a rather decent alternative. There are, as mentioned, many many examples of solutions like this, and creating your own is not too difficult due to the amount of research and public material available on the implementation.

Thx for the answer!

[CRO_Adriatic]

Nice topic...

[Heliocon]

Second video looks bad, first one looks ok because of the advanced lighting features. I am going to say this again and I know I know... DX11

Why? Because volumetric clouds that interact with light and weather physics would only be possible using DX11 tech for both cpu and gpu reasons. (it would also run faster).

[Cobra8472]

Second video looks bad, first one looks ok because of the advanced lighting features. I am going to say this again and I know I know... DX11

Why? Because volumetric clouds that interact with light and weather physics would only be possible using DX11 tech for both cpu and gpu reasons. (it would also run faster).


This is certainly NOT true.

An implementation of cloud rendering based on basic light scattering through clouds (or particles.. basically), is possible, and perfectly viable using DX9.
(and older versions too)
on a sidenote, I believe I have an old demo on my other PC from a whitepaper implementation from 2002, that looks amazing.

Again and again I see people overestimating the capabilities of DX11 or DX10 vs DX9.

It MUST be noted time and time again that while these DX iterations are more powerful, they do not drastically improve the amount of things that you can actually do on the GPU.

[Heliocon]

This is certainly NOT true.

An implementation of cloud rendering based on basic light scattering through clouds (or particles.. basically), is possible, and perfectly viable using DX9.
(and older versions too)
on a sidenote, I believe I have an old demo on my other PC from a whitepaper implementation from 2002, that looks amazing.

Again and again I see people overestimating the capabilities of DX11 or DX10 vs DX9.

It MUST be noted time and time again that while these DX iterations are more powerful, they do not drastically improve the amount of things that you can actually do on the GPU.

DX9 vs DX11 does. I have already explained why in many post from a base level pov with pipelines all the way to a graphics designer pov that makes it more efficient, better looking and easier to create.

So DX8 had sub surface scattering, and advanced particle physics/light interaction? Care to back that up with pics/evidence?

[CharveL]

Been reading some more wiki's I see.

DX11 is a great incremental improvement in many different areas, especially visual quality and efficiency of certain effects (among other stuff) but much like DX10 it's hardly the panacea you always make it out to be.

Good topic though, and I'd love to see some developer input although I'm also just as sure you'd be telling Ilya what he really should be doing with DX11, or assume he's just lying. ;)

If we're not seeing a lot of implementation of cloud systems like that first video by now after 4 years it's a pretty good indication it's not all it's cracked up to be.

Other systems like SpeedTree have caught on and are being used in CoD (I'm pretty sure) and many other games.

[Heliocon]

Been reading some more wiki's I see.

DX11 is a great incremental improvement in many different areas, especially visual quality and efficiency of certain effects (among other stuff) but much like DX10 it's hardly the panacea you always make it out to be.

Good topic though, and I'd love to see some developer input although I'm also just as sure you'd be telling Ilya what he really should be doing with DX11, or assume he's just lying. ;)

If we're not seeing a lot of implementation of cloud systems like that first video by now after 4 years it's a pretty good indication it's not all it's cracked up to be.

Other systems like SpeedTree have caught on and are being used in CoD (I'm pretty sure) and many other games.

As usual charvel you have to butt in like a moron to smack talk and make yourself look ignorrant. I am getting really tired of you idiotic canned replies to whenever I post.

http://translate.google.de/translate?js=n&prev=_t&hl=de&ie=UTF-8&layout=2&eotf=1&sl=de&tl=en&u=http%3A%2F%2Fwww.pcgameshardware.de%2Faid%2C8155 44%2FTechnikinterview-zu-IL-2-Sturmovik-Cliffs-of-Dover-DirectX-11-noch-zu-instabil-bis-zu-1000-Kilometer-Sichtweite%2FRennspiel-Sportspiel-Simulation%2FNews%2F&act=url

http://forum.1cpublishing.eu/showthread.php?p=233795#post233795

"PCGH: Would the use of DX11 will only accelerate the performance when rendering or DX11 provides for a clear appreciation of the optics?
Oleg Maddox: Beide Aspekte sind sehr wichtig. Oleg Maddox: Both aspects are very important. Zum einen ist eine verbesserte Performance hilfreich. First, an improved performance is helpful. Zum anderen planen wir, die Landschaft per Tesselation optisch aufzuwerten und Post-Effekte wie einen Local Tonemap Operator oder High Qualitiy Screen Space Ambient Occlusion (HQSSAO) einzusetzen. Second, we plan to enhance the landscape visually by tessellation and post-effects such as a Local Tonemap operator or high Qualitiy Screen Space Ambient Occlusion (HQSSAO) use."

As always if you post like an idiot - I reserve the right to call you one, especially because its the only type of post you ever write in response. I guess your reading skills arent very advanced because this is not the first time I have had to link you to info like this.

[Cobra8472]

As usual charvel you have to butt in like a moron to smack talk and make yourself look ignorrant. I am getting really tired of you idiotic canned replies to whenever I post.

http://translate.google.de/translate?js=n&prev=_t&hl=de&ie=UTF-8&layout=2&eotf=1&sl=de&tl=en&u=http%3A%2F%2Fwww.pcgameshardware.de%2Faid%2C8155 44%2FTechnikinterview-zu-IL-2-Sturmovik-Cliffs-of-Dover-DirectX-11-noch-zu-instabil-bis-zu-1000-Kilometer-Sichtweite%2FRennspiel-Sportspiel-Simulation%2FNews%2F&act=url

http://forum.1cpublishing.eu/showthread.php?p=233795#post233795

"PCGH: Would the use of DX11 will only accelerate the performance when rendering or DX11 provides for a clear appreciation of the optics?
Oleg Maddox: Beide Aspekte sind sehr wichtig. Oleg Maddox: Both aspects are very important. Zum einen ist eine verbesserte Performance hilfreich. First, an improved performance is helpful. Zum anderen planen wir, die Landschaft per Tesselation optisch aufzuwerten und Post-Effekte wie einen Local Tonemap Operator oder High Qualitiy Screen Space Ambient Occlusion (HQSSAO) einzusetzen. Second, we plan to enhance the landscape visually by tessellation and post-effects such as a Local Tonemap operator or high Qualitiy Screen Space Ambient Occlusion (HQSSAO) use."

As always if you post like an idiot - I reserve the right to call you one, especially because its the only type of post you ever write in response. I guess your reading skills arent very advanced because this is not the first time I have had to link you to info like this.

Actually - you are making yourself the fool.

Both SSAO and Tonemaps (+ a WIDE variety of other post-processing shaders & filters) are fully available in DX9, and with good implementation, are perfectly viable performance wise.


The BIGGEST example of this is CryEngine 3, which is running under DX9 (on consoles and PCs) and is an amazingally beautiful engine.

http://gaming.icrontic.com/files/2009/10/Dynamic-Volumetric-Light-Beams-and-Light-Shaft-Effects.jpg

http://get-2.wpapi.wp.pl/a,61764431,f,upload/f/a/d/fadb9281733a4269104ff47cd3b9c31e/crysis_2.jpg

[Cobra8472]

Been reading some more wiki's I see.

DX11 is a great incremental improvement in many different areas, especially visual quality and efficiency of certain effects (among other stuff) but much like DX10 it's hardly the panacea you always make it out to be.

Good topic though, and I'd love to see some developer input although I'm also just as sure you'd be telling Ilya what he really should be doing with DX11, or assume he's just lying. ;)

If we're not seeing a lot of implementation of cloud systems like that first video by now after 4 years it's a pretty good indication it's not all it's cracked up to be.

Other systems like SpeedTree have caught on and are being used in CoD (I'm pretty sure) and many other games.


I'd rather they take their time and properly implement a cloud rendering solution for the next expansion or something as such.

Clouds are by far one of the most important factors when it comes to the look and graphics of the game.

[CharveL]

I'd rather they take their time and properly implement a cloud rendering solution for the next expansion or something as such.

Clouds are by far one of the most important factors when it comes to the look and graphics of the game.

Agreed. I think they already have better clouds on tap but I suspect they realized some time ago the performance hit in it's current state is too much at this point, much like the dynamic weather. I guess we'll see.

Helicon the only issue I have with your posts is how you draw conclusions from the technical aspects whose purpose you know but have little clue as to their real-life implementations.

It's just that you end up spreading a lot of disinformation that confuses people unless someone calls you on it, which has happened a few times now, once by Luthier himself. I'll be happy to support your posts that are constructive or even just posited as opinion.

[Heliocon]

Actually - you are making yourself the fool.

Both SSAO and Tonemaps (+ a WIDE variety of other post-processing shaders & filters) are fully available in DX9, and with good implementation, are perfectly viable performance wise.


The BIGGEST example of this is CryEngine 3, which is running under DX9 (on consoles and PCs) and is an amazingally beautiful engine.

http://gaming.icrontic.com/files/2009/10/Dynamic-Volumetric-Light-Beams-and-Light-Shaft-Effects.jpg

http://get-2.wpapi.wp.pl/a,61764431,f,upload/f/a/d/fadb9281733a4269104ff47cd3b9c31e/crysis_2.jpg

I never said they cannot be done - they are done alot better and with less of a performance hit in DX11. Crysis 3 is a CONSOLE oreinted engine, its sacrfices alot of Cryengine2 features and in fact is less powerful. They cut down on its draw distances, physics, lighting effects among other things to gain performance. ALso you just called Oleg Madox the fool because that was a quote from him on DX11 use vs DX9 in COD! So instead of criticising me on the hssao which I never once wrote (until right now) maybe you should pay attention to something called a quotation mark (") and a link, but as someone who thinks cryengine 3 is the best thing since sliced bread I should not of expected more (because its a crappy engine and game, not to mention that there is only three graphics setting on the computer (discluding res) which is "gamer" "advanced" "hardcore" and absolutely no info on what each entails. But I guess thats how you like your options, clean cut and just what you see right infront of you?
But I suppose if you know squat about what goes on under the hood you might think it looks better.

Now remembering that cryengine 2 is many years older then 3, and that its running in DX10 not DX11 (10 is not as efficient and is harder as a graphics designer to use) this shows the differance well: http://www.youtube.com/watch?v=2WJG14uLA3k

[Heliocon]

Agreed. I think they already have better clouds on tap but I suspect they realized some time ago the performance hit in it's current state is too much at this point, much like the dynamic weather. I guess we'll see.

Helicon the only issue I have with your posts is how you draw conclusions from the technical aspects whose purpose you know but have little clue as to their real-life implementations.

It's just that you end up spreading a lot of disinformation that confuses people unless someone calls you on it, which has happened a few times now, once by Luthier himself. I'll be happy to support your posts that are constructive or even just posited as opinion.

Wait Luthier called me on what? Did you read my linked posts? I was right and saying what they should do 4 months before they even announced that it was infact what they were going to do. I also have a decent amount of experience in Maya/Autodesk which is what the majority of games and CGI graphics are modeled and textured/animated in.

So what disinformation exactly? As usual you completely fail to support any claims you make, its a habit you seem to refuse to break and I wouldnt respond to you this way if you would 1. stop trolling about 1/2 my posts, and 2. actually provide evidence or a cogent argument.

Edit: A few hours after making this post, luthier posted this: http://forum.1cpublishing.eu/showthread.php?t=19819
Under E for how to improve performance: "E. NOTE: DirectX 9 offers slower performance and lower graphic fidelity. If your system meets Recommended system requirements, upgrading to a DX10-capable operating system will offer better performance boost in Cliffs of Dover than perhaps any hardware upgrade."

Not to sound arrogant - but yet again I am verified indirectly by a dev statement.

[CharveL]

Edit: A few hours after making this post, luthier posted this: http://forum.1cpublishing.eu/showthread.php?t=19819
Under E for how to improve performance: "E. NOTE: DirectX 9 offers slower performance and lower graphic fidelity. If your system meets Recommended system requirements, upgrading to a DX10-capable operating system will offer better performance boost in Cliffs of Dover than perhaps any hardware upgrade."

Not to sound arrogant - but yet again I am verified indirectly by a dev statement.

What?? I don't see anything about DX11 in there anywhere or how it makes DX11 the holy grail of performance for CoD. Also, I'm not contending that everything you've copy/pasted or rambled on about is wrong by any means, just some of your conclusions on how it should be applied to CoD.

I also got a kick out of your suggestion to Luthier that maybe he should try using simple "hit boxes" around trees as a solution to performance. That's like me telling my mechanic to check the engine as a solution to why my car won't start.

If I had the inclination to joust with you further I'd bring up the post where you shouted about how CoD doesn't use more than one core, showing pics of apparently idle cores from task manager, which further illustrated your lack of understanding of how tasks tend to be distributed to additional cores in games. Then, while you were obnoxiously insulting another poster that came in to correct some of your misunderstandings, Luthier dropped in to explain that yes indeed CoD uses other cores as needed.

I'm not going to follow you down your preferred path of debating by hurling insults, although I've got a thick enough skin that it doesn't really bother me. People really aren't interested in wading through our back and forth banter.

You do bring some good information to the table and if you just stuck to that, instead of pretending you know better than the developers, and getting pouty and butthurt when someone steps in to correct you, you wouldn't find yourself being embarrassed when it does happen.

If you can stick to that, I'll look forward to your posts and will learn a few things myself from you and the discussion.

[Cobra8472]

When you say that CryEngine 3 is a step back from CryEngine 2, I'll just leave you with this .pdf:

http://www.crytek.com/sites/default/files/Crysis%202%20Key%20Rendering%20Features.pdf

Just because they cut back some of the settings and numbers of dynamic lights etc etc for console versions, does not mean the engine is worse...


and for that matter, ALL respect to Oleg and Luthier for amazing skills in various areas of game development, but I'm unsure as to how many HLSL shaders they have written....

[Cobra8472]

Here's another SDK to get those tastebuds working ;)

Easily implementable,

http://www.youtube.com/watch?v=WluLUXMg604

[CharveL]

Clouds don't need to move.

Moving clouds is a huge FPS killer.

Fact is, in an aircombat simulation the player is always moving. Usually at a speed far exceeding any cloud movement. Cloud movement would be practically non-discernible to player.

Also, when programming for AI:

Not being able to see in the clouds it would be much more difficult for tracking cloud positions along with AI movement. This would affect the FPS as well.

It can all be done I'm sure, but to be honest with you. I would rather have the ability to have more moving objects that really mean something to gameplay.

So, give me static clouds that don't move, that AI cannot see through.

Above all... I want the best FPS possible. So give me eye-candy taht doesn't affect FPS.

Agreed. It's a shame we can't have it all but frame rate really is king.

Then again, they might be able to do some clever work to get them in without as much impact or offloading to another core but that would raise the minimum system requirements to 4 cores, much like the tree collisions Luthier mentioned. The other factor would be the effect on online play keeping track of them in relation to all players and AI.

Who knows, maybe DX11 will give us free clouds. ;)

[Heliocon]

What?? I don't see anything about DX11 in there anywhere or how it makes DX11 the holy grail of performance for CoD. Also, I'm not contending that everything you've copy/pasted or rambled on about is wrong by any means, just some of your conclusions on how it should be applied to CoD.

I also got a kick out of your suggestion to Luthier that maybe he should try using simple "hit boxes" around trees as a solution to performance. That's like me telling my mechanic to check the engine as a solution to why my car won't start.

If I had the inclination to joust with you further I'd bring up the post where you shouted about how CoD doesn't use more than one core, showing pics of apparently idle cores from task manager, which further illustrated your lack of understanding of how tasks tend to be distributed to additional cores in games. Then, while you were obnoxiously insulting another poster that came in to correct some of your misunderstandings, Luthier dropped in to explain that yes indeed CoD uses other cores as needed.

I'm not going to follow you down your preferred path of debating by hurling insults, although I've got a thick enough skin that it doesn't really bother me. People really aren't interested in wading through our back and forth banter.

You do bring some good information to the table and if you just stuck to that, instead of pretending you know better than the developers, and getting pouty and butthurt when someone steps in to correct you, you wouldn't find yourself being embarrassed when it does happen.

If you can stick to that, I'll look forward to your posts and will learn a few things myself from you and the discussion.

Please show me where I posted pics of COD use and cores. Seriously - I never saw this reply but you are a lier - find that post. As usual you make up whatever you want, I never posted any pics like that because I DO NOT OWN COD (yet) and could not of possibly posted any pics of it running on my machine. Over that the pic you are talking about displayed 8 virtual cores, I run a 980x which has 12 virtual cores.
Also DX10.1 uses basically the same shader pipelines and lighting effects that DX11 does, DX10 uses a lighting system closer to DX9 than DX11 which results in bad performance. Therefore the DX11 rescource managment which is grouped with dx10.1 are both better than dx10/dx9 which is shader model 2-3. DX11 builds on this improvement by harnessing dynamic graphics detail systems to further increase performance per effect than dx10.1 does (basically dx11 adds a few tweaks to the underlying unified platform, and a bunch of new graphical techniques).

Since you yet again lied, without any proof of your statement, I am going to call you yet again a name. Sorry, when you misquote me and bs I feel the need to say it and I will stop when you decide that in order to have a conversation, when you paraphrase what you think I said, you should probably have proof to back it up, at the minimum it makes you look lazy when you dont. I dont know better than the devs about the functioning of their own engine, but when it comes to solutions to problems that are not in the immediate engine, I feel I have enough knowledge to discuss it. If there werent so many rediculous problems, I wouldnt feel the need to comment at all, if the devs comments added up with what we see in the engine, I wouldnt feel the need to comment on the product, which is better to have fact based criticism and my money (as a customer) to fix problems then for me to not bother saying anything and not purchase the game at all. I also wouldnt be as blunt and outspoken as I have been if my comments were not eventually supported by the devs own comments months later (for example talking about the uses of DX11 features, and I wont link my comments and 4 months later the devs nearly point by point confirmation of what I said).

[Heliocon]

When you say that CryEngine 3 is a step back from CryEngine 2, I'll just leave you with this .pdf:

http://www.crytek.com/sites/default/files/Crysis%202%20Key%20Rendering%20Features.pdf

Just because they cut back some of the settings and numbers of dynamic lights etc etc for console versions, does not mean the engine is worse...


and for that matter, ALL respect to Oleg and Luthier for amazing skills in various areas of game development, but I'm unsure as to how many HLSL shaders they have written....

I never said the engine is worse, I said that in order to get it to run on consoles they had to spend four years working on optimizing features they already had, and cutting back on many of the Cryengine 2 effects. They have made some improvements graphicaly within confined spaces, as outlined in the document, but remember the range of view in crysis 2 is probably 1/4 or less than in crysis. Also note Chivas that a month or so back you argued that console gaming/multiplatform engines were not holding pc gaming back. But as the doc says - the lowest setting on the PC is equivalent or better than the console version (better I would say because you are guranteed to be running it at a higher res, and with more AA which is a big multiplier). They did great with optimizing the engine and cutting it down for consoles, but cryengine 3 in its current state is minimaly (aka most people wont see a real differance) better looking than cryengine 2 in the short distance with 3+ years of work in optimizing but lacks ability to implement the effects in anything but a confined space.

interesting article below
http://www.geforce.com/#/News/articles/battlefield3-interview