Water tech *spawn

[Mobius1aic]

Thread spawned from 'desired next-gen effects' thread. Given ps3rocks was discussing water in two different places, I felt it warranted its own thread.
Both water discussions have been merged here.

One thing you mentioned that caught my attention was the use of FFT or Full Fourier Transform... and yes this is totally amazing that Cell is doing this and I've totally not seen any other processor even going to check this, let alone to be made for it. Processing wise, it can be put into perspective as the first game in the history of gaming that did dynamic wave-dispersion of water, when water is doing procedural wave generation as well... and is interactive for any number of individuals touching it... is Resistance-2. Entire game's water is done on FFT. Funny thing is that even Crysis hasn't touched it yet... so obviously no interactive wave dispersion in it(Crysis only does procedural waves, but neither dynamic dispersion, nor interactive generation blending).

For Crysis 2, we're hearing news of utilizing FFT. In Resistance-2 both small body and large body(ocean) water are rendered seperately but the core FFT algorithm remains the same. Be it 2 ft or 200 metres, it remains the same without framerate issues. That was the first where I was completely sold to Cell's potential even into the face of the 100s of cores containing modern GPUs.

I've seen it in action, both for small and large bodies of water, and it's pretty impressive. Problem is that at large scale it's somewhat silly as there is a complete lack of water separation (splash) as an object hits the water, you just get large waves. It's still quite a sight.

I would guess that it isn't FFT, but what are these games using then for their dynamic water simulation?:

Bioshock 1/2
Fallout 3
FEAR (yes the original)
Halo 3
Gears of War 2/Unreal 3.5

 

[Nesh]

The first Bioshock as well as Fallout 3 IIRC use some form of surface based dynamic water system. Don't know if it's FFT, but it looks damn good. Also, what about Gears of War 2's water? I think even the original FEAR had some sort of 3D water system going on.

The original Bioshock AFAIR did not have interactive water although it looked damn good. The water if I recall was a flat surface in disguise

 

[nightshade]

Bioshock 1/2 water isn't interactive. IIRC Gears 2 uses tessellated water, no idea about Halo 3 though. UC1 & UC2 are still my top fav. when it comes to water that looks & feels great [eventhough there's quite a bit of difference between both the games when it comes to how the game renders water]

 

[Mobius1aic]

The original Bioshock AFAIR did not have interactive water although it looked damn good. The water if I recall was a flat surface in disguise

Well I checked it out since it's been a while since I last played it, but it does indeed have interactive water, accept so far it's all based on the surface shader of the water, and it's not 3D. There is a 2D rippling effect from walking through as well as gunshots going into the water. Looks nice.

 

[TheWretched]

Resistance 2 had "nice" interactive water. It looked a bit strange (too synthetic), but it looked good, for a console game.

 

[Neb]

Funny thing is that even Crysis hasn't touched it yet... so obviously no interactive wave dispersion in it(Crysis only does procedural waves, but neither dynamic dispersion, nor interactive generation blending).

For Crysis 2, we're hearing news of utilizing FFT. In Resistance-2 both small body and large body(ocean) water are rendered seperately but the core FFT algorithm remains the same. Be it 2 ft or 200 metres, it remains the same without framerate issues. That was the first where I was completely sold to Cell's potential even into the face of the 100s of cores containing modern GPUs.

FFT is used in all Crysis games and some more PC games from RTS to FPS. FFT algorithm on/off barely taxes at all a dual-core CPU like mine (E8400). Water traingles blended range in the 100's of thousands of polygons.

 

[Neb]

I would guess that it isn't FFT, but what are these games using then for their dynamic water simulation?:

Bioshock 1/2
Fallout 3
FEAR (yes the original)
Halo 3
Gears of War 2/Unreal 3.5

AFAIk Bioshock, Fallout, Oblivion uses 2D sahder for procedural ripple effect on water. Same as Uncharted games do. FEAR has real 3D water deformation with physics. Dunno about Halo 3/GoW series.

Though some games without FFT like Anno 1404 yet their dense 3D water with complex physics is IMO more impressive than Crysis water. And not just how water is deformed with complex physics but also how it actually looks and behaves like real water with surface and underwater effects! :smile:

 

[BoardBonobo]

Are we talking Fast Fourier or Full Fourier? There is a difference.

 

[Neb]

FFT Fast Fourier Transform, FT Fourier Transform. Those game above listed as using FFT on both platforms uses FFT.

 

[patsu]

I've seen it in action, both for small and large bodies of water, and it's pretty impressive. Problem is that at large scale it's somewhat silly as there is a complete lack of water separation (splash) as an object hits the water, you just get large waves. It's still quite a sight.

Yeah I think for small scale water bodies, the physics-based ones are cool, especially when you see little critters charging at you in shallow water.

IMHO, the Kraken boss fight should have been more art driven, just like GoW3's Poseidon boss battle. That way, they can "bend" the shape of water unnaturally for drama (like appearing as charging horses in LOTR the movie).

 

[Mobius1aic]

Yeah I think for small scale water bodies, the physics-based ones are cool, especially when you see little critters charging at you in shallow water.

IMHO, the Kraken boss fight should have been more art driven, just like GoW3's Poseidon boss battle. That way, they can "bend" the shape of water unnaturally for drama (like appearing as charging horses in LOTR the movie).

It's the Resistence 2 Kraken fight I was specifically thinking of for the large scale FFT lol.

 

[patsu]

It's the Resistence 2 Kraken fight I was specifically thinking of for the large scale FFT lol.

Yeah, I was comparing R2's math waves vs GoW3's art waves. ^_^

 

[ps2rocks]

There is this quite detailed PDF by Insomniac about their water stuff in R2:
http://www.insomniacgames.com/tech/articles/0409/files/water.pdf

And this is a video showing some stuff of it:
http://www.youtube.com/watch?v=JyuY3-SLuDA
and a very small body one:
http://www.youtube.com/watch?v=Ii9Eek0CShM&feature=related

That is some of the most advanced water I've seen in a VideoGame. Ontopic... Unreal 3.0 can't do volumetric water, so yeah you are right in saying that both Bioshocks and gears2 had interactive water "surfaces" only(also that it was told in the gears 2 tech demo as well). That said, we have yet to see an UE based game to show volumetrics let alone FFT.

_______________________________________________________________________

Talking of something offtopic(other than CEll), as my name suggests, I'm a big BIG fan of PS2's technology and believe that a lot was achieved on it so early in its life-cycle that it leaves some next-gen games behind. I remember I was so shocked to see the volumetrics and caustics of ghosthunter:

see below:
http://www.youtube.com/watch?v=_AvEcX_ozwo&feature=related
http://www.youtube.com/watch?v=tn5-uZEp-l0&feature=related
http://www.youtube.com/watch?v=plvBXvMeS4k

But the point is... its definitely more elaborated by the systems/consoles/machines whatever... that are more CPU dependant. What I'm seeing here and on several occasions is an affirmation that the stronger your CPU is, the more 'simulated' or physics oriented your games/software would be. Something, sadly, GPU oriented machines have failed to cater the opinion upon. Yes, the tech-demos and benchmarks are really impressive of all these GPUs, but when you put it to a game(with A.I.s, NPCs, draw-dstances, portals, media etc.) all that collapses. I just have no clue why!

 

[Billy Idol]

One thing you mentioned that caught my attention was the use of FFT or Full Fourier Transform... and yes this is totally amazing that Cell is doing this and I've totally not seen any other processor even going to check this, let alone to be made for it. Processing wise, it can be put into perspective as the first game in the history of gaming that did dynamic wave-dispersion of water, when water is doing procedural wave generation as well... and is interactive for any number of individuals touching it... is Resistance-2. Entire game's water is done on FFT. Funny thing is that even Crysis hasn't touched it yet... so obviously no interactive wave dispersion in it(Crysis only does procedural waves, but neither dynamic dispersion, nor interactive generation blending).

For Crysis 2, we're hearing news of utilizing FFT. In Resistance-2 both small body and large body(ocean) water are rendered seperately but the core FFT algorithm remains the same. Be it 2 ft or 200 metres, it remains the same without framerate issues. That was the first where I was completely sold to Cell's potential even into the face of the 100s of cores containing modern GPUs.

I've seen it in action, both for small and large bodies of water, and it's pretty impressive. Problem is that at large scale it's somewhat silly as there is a complete lack of water separation (splash) as an object hits the water, you just get large waves.

Uh, I would be careful when demanding such things. Physics \not = Physics. Separation of water is an actual topic in research (people aren't even quite sure how one could describe the correct physics, let alone simulate the general case )
In my experience, reading stuff about the phyics implementation in games: it is all quite basic stuff (equations+methods) and nothing fancy, things our students typically get as "homework"!
So I would never call game physics...real physics , but of course I learned in this forum that games are smoke and mirrors!
As I said, the only instance the implementation impresses me is indeed the water in Resistance 2, although the physical model is quite basic and simple as well (the equations!), the equation solver can be classified as state of the art in research (IMO), and this is quite remarkable and it is impressive as well as FFT is a quite good match for cell!

Are we talking Fast Fourier or Full Fourier? There is a difference.

FFT means Fast Fourier Transformation

 

[patsu]

Do they do anything special when I'm in the water ? There was a segment where I needed to dive and swim across a river.

They should show comparable timing between an Intel CPU, a discrete GPU and Cell, and also different scales for people to appreciate the performance and speed up.

 

[ps2rocks]

although the physical model is quite basic and simple as well (the equations!), the equation solver can be classified as state of the art in research (IMO), and this is quite remarkable and it is impressive as well as FFT is a quite good match for cell!

This can be accounted as the best sum up of the FFT-Cell relationship as remarkably exhibited by Resistance2.

 

[Mobius1aic]

Uh, I would be careful when demanding such things. Physics \not = Physics. Separation of water is an actual topic in research

I figured as such, though I more or less meant an animated splash, not a procedural one. There is some for when the kraken descents below the surface, but not when the arms are hitting the water. Would it be difficult to calculate a specific "speed" that the animation bones are hitting the surface of the water in order to determine a high enough velocity to justify animating a splash? It's really cool to watch the water in action though. Almost mesmerizing.

 

[fearsomepirate]

I assume they're using something like the shallow water equations in Resistance 2, or just a dispersion relation of some kind.

I always find it interesting just how much math it takes to solve highly, highly simplified fluid models in real-time on what is essentially a very coarse grid. Yet at the same time, it's amazing we can do it at all. But yeah, we're years away from doing more sophisticated approximations in real-time, like large eddy simulation (still takes weeks of computation time to simulate a a minute or so of fluid flow using LES).

 

[Billy Idol]

I assume they're using something like the shallow water equations in Resistance 2, or just a dispersion relation of some kind.

If I remember correctly they just solve the wave equation, as they are mainly interested in the surface of the water (and some nice dispersion effects)!

I always find it interesting just how much math it takes to solve highly, highly simplified fluid models in real-time on what is essentially a very coarse grid.

Yes, but I was really shocked to learn that modern game devs use modern simulation technology for their games - cool! At the same time, it is interesting for me to judge what methods they use at the end: their highest priority is that everything runs in real time, whereas my focus is on choosing the methodology which produces the best (most accurate, most phyical) solution - although I share their interest that the computation should be efficient...

Yet at the same time, it's amazing we can do it at all. But yeah, we're years away from doing more sophisticated approximations in real-time,

The real problem is (at least in science), that we are years away from doing sophisticated methods (such as LES) for real technical relevant problems...and by this, I mean not even the real time aspect

like large eddy simulation (still takes weeks of computation time to simulate a a minute or so of fluid flow using LES).

I just talked yesterday to a colleague, who finally finished the evaluation of his last LES computation: the overall computation time was 10^6 CPU hours (1000 hours on 1000 CPUs)...that is why people nowadays get interested in CELL/GPGPU computing

 

[Nesh]

Well I checked it out since it's been a while since I last played it, but it does indeed have interactive water, accept so far it's all based on the surface shader of the water, and it's not 3D. There is a 2D rippling effect from walking through as well as gunshots going into the water. Looks nice.

I was talking about 3D interaction. I did not consider 2D rippling as interaction because I see it as a very basic visual feature that doesnt involve any special physics or calculations which existed for way too long