Computer generated water
- Thread starter carpediem
- Start date
Mintmaster
Veteran
Holy crap that's a well done simulation. I wonder how much of a CPU hog that is to get the splashing water to break up into tiny droplets interacting with each other. I saw some decent stuff at Siggraph 2002, so I guess I should have expected this by now, but it's still really amazing.
Too bad this wasn't around during the filming of WaterWorld.
That is really incredible.
Going by the soon-to-be dedicated physics processors (Ageia's water simulation demos are already pretty good, btw) how long do you think before we see those kind of physics/effects in humble real time games? My guess is a looooong time indeed, but it's nice to dream...
Going by the soon-to-be dedicated physics processors (Ageia's water simulation demos are already pretty good, btw) how long do you think before we see those kind of physics/effects in humble real time games? My guess is a looooong time indeed, but it's nice to dream...
Chalnoth said:
The consumer POV: Perhaps they won't take off, but if Epic, Ubisoft et all are supporting the product in their upcoming titles, then surely it can't just be a fly-by-night thing. After all, aren't the majority lot of end-users finally accepting both multi-GPU (SLI and Crossfire) and multi-core CPU (X2, Pentium D) as feasable- if very expensive when high-end- solutions to irreputably make thier PC games run better one way or another? Whether these solutions will last for another 5+ years is another story...
On the other hand, and not to be arguementative, but do you not see the PPU taking off in a big way? Why not, do you only see accelerated physics as being additional non-interactive eye-candy only? I know from reading the script of Carmack's 2005 QC speech he feels this way, he does not see the dedicated PPU as 'the next big thing' or anything like it...
Mintmaster
Veteran
Indeed. How many zeros do you need to describe the number of CPU cycles per frame?hughJ said:
LVSeminole
Newcomer
That is crazy. And I thought the Far Cry, and Half Life 2 water models were great. LVS
Doing water "properly" is extremely computationally expensive. You need lots of particles to fully resolve the fluid (which is why the Ageia demos look so utterly utterly shite!) and with water being incompressible it really limits how large you an make your timesteps in your integrator.
Some colleagues and I have been playing around with some particle-based water simulations very similar to those shown in those movies ... they can take days to run on a single processor (granted they're not fully optimised at the moment, but still...). Our current efforts are focussed getting the code parallelised and running on an Opteron cluster. It's a bit of a bitch with such short timesteps, it seriously hampers scalability.
I'll post some movies once we have something polished to show
Some colleagues and I have been playing around with some particle-based water simulations very similar to those shown in those movies ... they can take days to run on a single processor (granted they're not fully optimised at the moment, but still...). Our current efforts are focussed getting the code parallelised and running on an Opteron cluster. It's a bit of a bitch with such short timesteps, it seriously hampers scalability.
I'll post some movies once we have something polished to show
Judging from how small the drops are, I think we won't see anything comparable to this in real-time for at least the next 5 years (hopefully I don't regret this estimate then ).
Amazing stuff! Especially the extremely large scale effects in the Shark movie - most fluid effects tend to work only well in small, closed areas and get _very_ slow once you bump up the grid size. Great!
).Amazing stuff! Especially the extremely large scale effects in the Shark movie - most fluid effects tend to work only well in small, closed areas and get _very_ slow once you bump up the grid size.
Great!Isn't a megalodon a kind of (extinct) shark?darkblu said:
Either way, still quite unrealistic.
Two points that is no good with the megalodon:
1) Too much cavitation. You won't see that many bubbles (if any?) behind a shark.
2) Propulsion? They zip through the water at great speed, but they don't move the fin(s) accordingly.
But other than that, the water is very realistic.
