GPU fluid solver videos: Water & Smoke

Well slap my ass and call me Nancy. Supergee ought to have a look at this. This is pretty much a resolution to a pretty lengthy discussion in that thread.

Edit: didn't see the link in the thread that I had posting in ::bows head in shame::
 
wow, that pretty much sums up the water and smoke problem.
happy to see smoke with 3 dimensions.
 
StevetheFox said:
Well slap my ass and call me Nancy. Supergee ought to have a look at this. This is pretty much a resolution to a pretty lengthy discussion in that thread.

Edit: didn't see the link in the thread that I had posting in ::bows head in shame::
Click to expand...
Already seen it. Look at it again. Nice but only the old rising smoke is something usefull in games. But need stil a lot of work. Because often its a after effect after a explosion. So big bang to fire to smoke effect. Or be optimised to a steam effect.

The glass fish tank tech demo make no ingame sense. looks cool but not usefull for games.
A destructable aquarium. Wich comes down to fluids flow after damage could be, but that not that demo. It looks more a way to kill fish by bugging them with Psi power by smoke in water. Only no fish.
The UE3 fluid demo is more practical for games.

Could be done with aquariums to. With fish in it. like in the movies.

Smoke could be interacting wenn player or Npc move to it. SMoke as part of a scene or explosion or fire or smoke-nade

It's time they put something in a Game or game demo.
 
SuperGee said:
Could be done with aquariums to. With fish in it. like in the movies.

Smoke could be interacting wenn player or Npc move to it. SMoke as part of a scene or explosion or fire or smoke-nade

It's time they put something in a Game or game demo.
Click to expand...

Agreed. The UE3 demo is more practical. These demos do look very polished though.
 
SuperGee said:
Already seen it. Look at it again. Nice but only the old rising smoke is something usefull in games. But need stil a lot of work. Because often its a after effect after a explosion. So big bang to fire to smoke effect. Or be optimised to a steam effect.

The glass fish tank tech demo make no ingame sense. looks cool but not usefull for games.
A destructable aquarium. Wich comes down to fluids flow after damage could be, but that not that demo. It looks more a way to kill fish by bugging them with Psi power by smoke in water. Only no fish.
The UE3 fluid demo is more practical for games.
Click to expand...


No in-game sense? Right. Then what is Aegia's demo for? Carwash Tycoon 2007?

Sucks for Aegia that nVidia's *first* tech demos clearly show a more polished product. Who knows what we'll see from nV and ATi once they've had an 8 month run at this thing.
 
xFlankerx said:
Messagewastooshortsohereitis.
Click to expand...
missed that part. then it just looks great. would have been interesting to see the box fill up, then the smoke would be exhausted, then let it fill up again.
 
SuperGee said:
Already seen it. Look at it again. Nice but only the old rising smoke is something usefull in games. But need stil a lot of work. Because often its a after effect after a explosion. So big bang to fire to smoke effect. Or be optimised to a steam effect.

The glass fish tank tech demo make no ingame sense. looks cool but not usefull for games.
A destructable aquarium. Wich comes down to fluids flow after damage could be, but that not that demo. It looks more a way to kill fish by bugging them with Psi power by smoke in water. Only no fish.
The UE3 fluid demo is more practical for games.

Could be done with aquariums to. With fish in it. like in the movies.

Smoke could be interacting wenn player or Npc move to it. SMoke as part of a scene or explosion or fire or smoke-nade

It's time they put something in a Game or game demo.
Click to expand...

You're obviously missing the point of a TECHNOLOGY demo.
 
Spewn said:
No in-game sense? Right. Then what is Aegia's demo for? Carwash Tycoon 2007?

Sucks for Aegia that nVidia's *first* tech demos clearly show a more polished product. Who knows what we'll see from nV and ATi once they've had an 8 month run at this thing.
Click to expand...

I disagree, go back to Aegia's website and watch the EU3 demo. You can argue about the carwash demo vrs. this one, but almost everyone agrees that the EU3 demo is the coolest liquid demo so far.
 
Hvatum said:
but almost everyone agrees that the EU3 demo is the coolest liquid demo so far.
Click to expand...

Disagree. The oil demo is still using about the same resolution for the liquid as all the other poor-looking, liquid demos. With the oil, you are able to clearly pick out the large beads that make up the substance.. where as with the water in the glass case, although not perfect, it's very difficult to spot distinct beads.
 
SuperGee said:
The glass fish tank tech demo make no ingame sense. looks cool but not usefull for games.
A destructable aquarium. Wich comes down to fluids flow after damage could be, but that not that demo. It looks more a way to kill fish by bugging them with Psi power by smoke in water. Only no fish.
Click to expand...

That's probably the dumbest thing I've ever read on these forums, and that's saying a lot. Congrats.
 
StevetheFox said:
You're obviously missing the point of a TECHNOLOGY demo.
Click to expand...
What point. this is apperently not about a discussion how bad or good those tech demo are from both topic threads. It's use for attacking and pissing on ageia PPU.
It not about this demo from the start of this thread any more.
It's PPU vs GPU(nV).

PhysX camp is already beyond techdemo's they are and have roling out games.
Your mis the point that green camp means that is the end of the war. Just by this Tech demo?
The Games deside. And till there are enough Havok FX games there are a lot more PhysX games. Where there is also more chance something good is between it.
Same as havok FX titles are there, some will be good and more bad or less ones.
In both camps some titles will stand out.

Ageia show only new Physics features at this stage. Deformable objects. Wich can be done better with each SDK iteration.
So maybe expecting a Metal sheet demo. Cloths aproach with other varable to emulate metal sheets.

Spewn said:
No in-game sense? Right. Then what is Aegia's demo for? Carwash Tycoon 2007?
Click to expand...
It's a tech demo and you know how I think about that. I don't care much about that. Ageia solution is passed that.
Because I Play CF:CT Graw BOS JTF. Where CF:CT stands out as a demo game.
Having the age-ing Unreal3.0 PhysX Game engine demo's show that it can be better then what CF show. But CF show mostly Gameplay PhysX.

It time for PhysX enriched and enhanced Games.
Sucks for Aegia that nVidia's *first* tech demos clearly show a more polished product. Who knows what we'll see from nV and ATi once they've had an 8 month run at this thing.
Click to expand...
Well I can Play PhysX now and more later and it only can get better. Well I wish you all fun watching just Tech demo's.
Waiting how all those games are turning out. Then we know more.

As for nV they only can put only Techdemo's vs games at this stage. So they must make some eye catchers. Ageia not, they depend on what the Game Dev's are doin with it.

I think UT2007 will look better then CF:R
But CF:R is more aimed at gameplay PhysX.
 
hughJ said:
Disagree. The oil demo is still using about the same resolution for the liquid as all the other poor-looking, liquid demos. With the oil, you are able to clearly pick out the large beads that make up the substance.. where as with the water in the glass case, although not perfect, it's very difficult to spot distinct beads.
Click to expand...

Fair enough, but the consensus still seems to be for the oil demo. Personally I find the oil demo much more interesting as it includes an entire environment, with the liquid actually dispersing through barriers and influencing other objects. Whereas the Nvidia demo is a bunch of water or gas in a box. Somehow it strikes me that such a simple environment easily allowes one to up the number of particles (beads as you call them) without sacrificing performance. From taking physics myself I certainlly find calculations involving strange angles, resistance and outside influence much harder then a bunch of stuff in a box.
 
Hvatum said:
Fair enough, but the consensus still seems to be for the oil demo. Personally I find the oil demo much more interesting as it includes an entire environment, with the liquid actually dispersing through barriers and influencing other objects. Whereas the Nvidia demo is a bunch of water or gas in a box. Somehow it strikes me that such a simple environment easily allowes one to up the number of particles (beads as you call them) without sacrificing performance. From taking physics myself I certainlly find calculations involving strange angles, resistance and outside influence much harder then a bunch of stuff in a box.
Click to expand...

Sweet christ you guys keep missing the point. It's about the liquid substance not the pretty box not the floating crates. It's the clear viscous fluid and their ability to calculate and render it in real time. Focus on the topic. The topic isn't:

"The Cool Backgrounds We See In Tech Demos."
or
"A Pixel for Pixel Rendition of What We Put In a Game"

It's:
"We are approaching a time where we can convincingly simulate some of the more complex and dynamic types of matter, and just imagine what else we'll be able to pull off. In doing so we can improve the realism and immersion of all future games."

This is a lot more impactful than:

"We made cool smoke for your smoke nades and cool fishtanks for your 'Goldfish Tycoon 2007' BECAUSE THAT'S THE ONLY REASON WHY WE MADE THESE DEMOS LITERALLY...for smoke nades and fishtanks in games."
[/rant]
 
I love this demo, very impressive - still not exactly stable (don't touch it for about a minute and you'll start to see the fluid elements go screwy - looks like rain) but very nice.


With in mind there are a few points:

We have yet to see this fluid model interact with an object (distinct from responding to a force - a point source (as seen in demo) is easier to model than a complex surface (a boot for example).

This could be difficult, as unlike particle based modelling you can't easily extract moments from continuum sims like NS - the easiest way is to ignore momentum etc and focus on the pressure element, which means your air/fluid interface has to be bang on.

You could do an approximation based on velocity of the fluid I guess, with a scaling factor for fluid density, but that kinda sucks after you've got a nice little NS sim happening realtime.

How large can the simulation be? My GF8800GTX hits a max of 60 fps at around 320x240. At 1920x1200 this starts to chug at 30fps. This is for a relatively small amount of fluid. What's the scalability like?
In my experience these sorts of sims get nasty exponentially based on grid size or res.

The smoke is very nice, but is not a fluid simulation, it's a 3D fractal rendering. Very nice and looks great - and WILL scale well from my experience and looks fantastic.

So to round up - some great strides but we'll have to see how this is actually implemented in game.

Another techdemo using it would be nice :D
 
MrNasty said:
How large can the simulation be? My GF8800GTX hits a max of 60 fps at around 320x240. At 1920x1200 this starts to chug at 30fps. This is for a relatively small amount of fluid. What's the scalability like?
In my experience these sorts of sims get nasty exponentially based on grid size or res.
Click to expand...
Even worse then you think.
A game have to render a full game besides those effects.
Also there are other Physics uses competing for resources.
So you could have a high detaild scenery
where there is lot of action and
mucho collisions
Cloths
Hair
other fluid rendering like blood, flame trower.
deformations
rigid body.
So in a real game this feature is part of a larger mix of physics features
the G80 or in that time G81 the g-card is already bog down by a heavy Graphics load. And then it have to share it resources with a mix of some light but also heavy Physics features. Knowing the top dev's and dev in general are more or less heavy fixated on good lookin GFX. the G-card is already busy.

To make the physix look good in behavior it has to use a decent fine grain solution to behave nore real but that needs also more resources for this to compute.

the upside is wenn those games come out G81 whil be out. I think it will have more power to it. And of course you can alway's use 2 G81 or G80. Or a midrange G8x by then, wich I would prefer. But then more likely a ATI solution. SM4 & a SM3 card.

Plus it have enough resources to do fine grain physics with it's 768 or 640 MB of Vid but also Physics mem.
So to round up - some great strides but we'll have to see how this is actually implemented in game.
Click to expand...
Yes while the waiting for PPU games feel long. HavokFX and it GPU partners, role out later so waiting for that is a bit longer.
Unless there first or one of there first games is a game wich make good use of HavokFX.
So I expect that we have to do with just demo for a while longer.
Another techdemo using it would be nice :D
Click to expand...
We have seen enough tech demo's. I like to see at least Demo's made in a full game engine with all the other bells & wistles so we know what can be done in a game.
Probaly a Game engine that uses HavokFX.
The Allan wake preview is just like that a physics demo use in a real game engine.

Better would be a Game that uses Physics in a good way.
 
Manny Calavera said:
I think its pretty impressive.Of course SuperGee is not impressed.... :D
Click to expand...

Let me explain I hate the movie twister. What has alan wake some stupid twisters of objects around the player. What Gameplay is that?
It's third Pirson. I Prefere the First person look.
I wonder what ALan wake has more in the game then that twister show.

I 'am more the kind of gamer that liked Crysis interactive realistic behaving jungle.
But loking at the vid's. It's still a grude aproximation. But at level never seen before.

And of Course Celfactor revolution. But I don't like the PSi powers I'am a demolition / nader player.
 
You must log in or register to reply here.
Back
Top