Video of a GDC 07 physics Demo on PS3

[chris1515]

http://www.gametrailers.com/player.php?type=wmv&id=17744

Kudos to Tiego on a french forum.

It is a benchmark of simulation of rigid body. But I want to know from which company this demo comes from.

 

[one]

Probably this demo
http://www.watch.impress.co.jp/game/docs/20070308/expo107.htm

SCEA booth
http://www.watch.impress.co.jp/game/docs/20070308/expo1.htm

 

[inefficient]

That chick in the gametrailers ad is hot!

These are Havok demos right?

 

[rounin]

I think they may be developed by IBM or Sony themselves, since there is no Havok branding or mention of the name. One problem I have with these demos is that they always show them rendering in slow motion and it gives the impression that the engine is slow and makes things look unrealistic. In the middle of the video, you get to see the engine work in "realtime" (near 1:16) and it actually looks quite realistic.

 

[slider]

It's difficult for a non-techie (like me) to know what to make of that given that there's no real context. Hmm.

Feel free to enlighten me...

 

[Shifty Geezer]

I think they may be developed by IBM or Sony themselves, since there is no Havok branding or mention of the name. One problem I have with these demos is that they always show them rendering in slow motion and it gives the impression that the engine is slow and makes things look unrealistic. In the middle of the video, you get to see the engine work in "realtime" (near 1:16) and it actually looks quite realistic.

These are in realtime from the looks of it. The calculation time in ms is shown in the corner info, up around 17 ms. The reason it looks slo-mo is unrealitic settings in the engine. The reason it speeds up is far simpler calculations. At the points it's animating quickly, there's hardly any collisions. When you get mountains of boxes piling up, all influencing each other, the calculations can skyrocket.

3000 boxes solved in 17 ms seems quite good to me, but I don't know have a PC benchmark for consideration (especially fo rthe specific physics engine). Also solving boxes is a doddle. It'd be nice to see mesh collisions if we want to get away from bounding-box physics!