world-wide havok
by Eric Forhan · in Torque Game Engine · 06/05/2001 (9:04 am) · 6 replies
How does the physics system in the V12 compare to such physics middleware as Havok (other than the lofty price tag, of course).
Also, can software such as Havok's be (easily?) integrated into the V12 environment?
Also, can software such as Havok's be (easily?) integrated into the V12 environment?
About the author
#2
06/06/2001 (8:42 pm)
There is no unified physics engine. All objects manage do their own physics, some more realistic than others. There are several reasons for this, not the least of which is trying to synchronize objects over the network. The vehicles have some rigid body code, suspension, rotating wheels, tire friction, etc. and is probably the closest thing to a traditional physics engine.
#3
First the word of caution - any object with nonlinear physics will require network updates in line with it's form of nonlinearity. These objects, or their potential nonliner interactions, should be kept to a minimum to preserve network performance. By nonlinear I mean any object which interacts with it's environment (other objects and world) in a way which cannot be predicted at the moment the object was created.
A slow moving projectile which simply moves along a straight line until it hits is a simple example which only requires a network update when it hits, and so is relatively trivial. However, a group of objects which continously exert forces on each other (magnets etc) are extreme cases. Bouncing objects come somewhere in-betweem, especially if they can bounce off other moving objects or terrain which can change.
However, a mortar round affected by gravity can have it's path predicted with it's endpoint (and explosion) being the only uncertainty (could hit someone mid-air etc.).
There are complex methods to overcome the errors which would accrue in predicting, rather than communicating, the state of such objects (such as communicating some global measure, and when this diverges sending further information to remove the divergence). However, I am uncertain of the validity of their use in fast action games as they can cause performance slow-downs at critical points in the game.
A great deal of believable 'physics' can be done with very simple (but well thought out) programming.
Good luck!
Doug Enkisoftware Limited
06/08/2001 (3:01 am)
As someone who was a research physicist in nonlinear dynamics for eight years, and then moved into making the computer game Avoyd, including programming low & high level network code, I have a word of caution and some advice about using 'physics'.First the word of caution - any object with nonlinear physics will require network updates in line with it's form of nonlinearity. These objects, or their potential nonliner interactions, should be kept to a minimum to preserve network performance. By nonlinear I mean any object which interacts with it's environment (other objects and world) in a way which cannot be predicted at the moment the object was created.
A slow moving projectile which simply moves along a straight line until it hits is a simple example which only requires a network update when it hits, and so is relatively trivial. However, a group of objects which continously exert forces on each other (magnets etc) are extreme cases. Bouncing objects come somewhere in-betweem, especially if they can bounce off other moving objects or terrain which can change.
However, a mortar round affected by gravity can have it's path predicted with it's endpoint (and explosion) being the only uncertainty (could hit someone mid-air etc.).
There are complex methods to overcome the errors which would accrue in predicting, rather than communicating, the state of such objects (such as communicating some global measure, and when this diverges sending further information to remove the divergence). However, I am uncertain of the validity of their use in fast action games as they can cause performance slow-downs at critical points in the game.
A great deal of believable 'physics' can be done with very simple (but well thought out) programming.
Good luck!
Doug Enkisoftware Limited
#4
06/08/2001 (9:23 am)
Thank you. Those are some very valid points I must consider in creation of this game.
#5
Anyone with any info?
09/26/2002 (6:06 am)
I was also wondering if the Havok Game Dynamics SDK could be adapted to work with the Tourqe engine.Anyone with any info?
#6
And with modified Marble Blast physics code I would think almost anything could be done.
09/26/2002 (8:12 am)
The answer is yes, but why? Torque's physics engine is getting better and 1000% cheaper. ;-)And with modified Marble Blast physics code I would think almost anything could be done.
Paul Robertson