Physx and Torque
by Albert Steckenborn · in Torque Game Engine · 06/26/2006 (5:29 am) · 37 replies
Hello Guys,
i want to implement the Ageia Physics engine into TSE.
I already know these Post
www.garagegames.com/index.php?sec=mg&mod=resource&page=view&qid=10258
I'm a beginner coder of c++ so it's a little bit too difficult for me to code it in alone. But may be we can do it together ? On the link (shown above) it's only implemented into the TGE but i need it into the TSE. The TSE terrain is working completely different.
I'm posting here because no one is replying there and i hope it's related that no one see the linked site. I do not hope that no one is interested in PhysX :-D
Greetings
i want to implement the Ageia Physics engine into TSE.
I already know these Post
www.garagegames.com/index.php?sec=mg&mod=resource&page=view&qid=10258
I'm a beginner coder of c++ so it's a little bit too difficult for me to code it in alone. But may be we can do it together ? On the link (shown above) it's only implemented into the TGE but i need it into the TSE. The TSE terrain is working completely different.
I'm posting here because no one is replying there and i hope it's related that no one see the linked site. I do not hope that no one is interested in PhysX :-D
Greetings
About the author
#22
I mean, this all assumes that there is an actual need to replace the stock physics other than for commercial purposes (ie, marketing bullet point for TSE).
My original point was that if there's going to be an official GarageGames effort made, it might be better to invest in utilizing a free and possibly open source solution and/or making it easier to plug in any phsyics engine you prefer. But I guess that's just stupid. *shrugs*
07/19/2006 (12:26 pm)
Quote:The other thing about non-free Physics are that you get support...How is it any better than the free solutions in terms of tech support? It's doubtful that ageia will be giving out free tech support. Thus, you're left with forums and community support like the free physics engines.
Quote:Think about it, how many people have integrated the *free* solutions? Not many.Well, I happen to be talking about GG implementing it with full and official support, not someone from the community doing it alone.
I mean, this all assumes that there is an actual need to replace the stock physics other than for commercial purposes (ie, marketing bullet point for TSE).
My original point was that if there's going to be an official GarageGames effort made, it might be better to invest in utilizing a free and possibly open source solution and/or making it easier to plug in any phsyics engine you prefer. But I guess that's just stupid. *shrugs*
#23
07/19/2006 (12:30 pm)
@James et al. Ok, whatever. You're clearly the authority on this matter, and I'm not. I'll get back to finishing more games now.
#24
Geez people, cut it out. This is in no way helping the situation and is probably going to discourage quite a few others.
Other than saying that, I'm out of here, don't want to be dragged into a bitch feast.
07/19/2006 (2:19 pm)
/me slaps everyone with a cold sticky slimey eel...Geez people, cut it out. This is in no way helping the situation and is probably going to discourage quite a few others.
Other than saying that, I'm out of here, don't want to be dragged into a bitch feast.
#25
07/19/2006 (6:50 pm)
Yeah, sorry. It was stupid of me to get annoyed. I'm a dufus sometimes. I'll get back to finishing more games now, again.
#26
All actual physics implementation projects are not updated since long time. And i need to know if it's necesarry to invest time to implement it by myself or should i have to wait ?
May be wen can make a CVS Project like nxOgre ?
For me PhysX will be the way of future. The possibilities you have with these physics engine is much more than with ODE or Newton.
Take a look into the Crysis Videos the you will see grass wich is moving when a player go through it or leafs of flowers are deform when a palyer go though it. Really goot desutruction effects. Really good cloth simulation and much more. For me you definitely cannot do it with ODE or Newton.
May be you can buy HuvokFX. But it's quit more expensive as PhysX and you are not able to get it free. If you implement PhysX in this way that your game will be much better with PhysX HW supprt that not then you will get PhysX for free !!
In the Discussion of HuvokFX and PhysX many people say "yeahr for Huvok you don't need a special Hardware". For me this is not correct because Huvok FX works as best when you are using two graphic cards in SLI mode. Then you have the choice to buy a second graphic card (nearly 300 ) or a PhysX card (230). Remember the PhysX card is low noise and need 35W and a second card is louder and needs more power and will be much cheaper in future and some next Graphic card will have it on board.
I've heared that Microsoft (may be) develops a new Api called "Direct Physics". This will be a directx for physics. I belive that many stuff of Ageia's PhysX will be implemented there because Microsoft uses PhysX by XBox 360 and is alredy a license holder.
It could be or not. Nobody knows it yet.
Greetings
07/19/2006 (11:00 pm)
It would be great to get an really official statement of GG how they want to go with physics in TSE.All actual physics implementation projects are not updated since long time. And i need to know if it's necesarry to invest time to implement it by myself or should i have to wait ?
May be wen can make a CVS Project like nxOgre ?
For me PhysX will be the way of future. The possibilities you have with these physics engine is much more than with ODE or Newton.
Take a look into the Crysis Videos the you will see grass wich is moving when a player go through it or leafs of flowers are deform when a palyer go though it. Really goot desutruction effects. Really good cloth simulation and much more. For me you definitely cannot do it with ODE or Newton.
May be you can buy HuvokFX. But it's quit more expensive as PhysX and you are not able to get it free. If you implement PhysX in this way that your game will be much better with PhysX HW supprt that not then you will get PhysX for free !!
In the Discussion of HuvokFX and PhysX many people say "yeahr for Huvok you don't need a special Hardware". For me this is not correct because Huvok FX works as best when you are using two graphic cards in SLI mode. Then you have the choice to buy a second graphic card (nearly 300 ) or a PhysX card (230). Remember the PhysX card is low noise and need 35W and a second card is louder and needs more power and will be much cheaper in future and some next Graphic card will have it on board.
I've heared that Microsoft (may be) develops a new Api called "Direct Physics". This will be a directx for physics. I belive that many stuff of Ageia's PhysX will be implemented there because Microsoft uses PhysX by XBox 360 and is alredy a license holder.
It could be or not. Nobody knows it yet.
Greetings
#27
[emo=:)]Can I have my 2 cents back? :-P[/emo]
07/20/2006 (1:19 am)
@Jason: Wasn't trying to be rude... sorry. I need to find the emotion tag [emo=:)]Can I have my 2 cents back? :-P[/emo]
#28
07/20/2006 (10:56 am)
NO! NOT EMO! EMO BAD! Emo like GOTH, only not as cool!
#29
Quite an interesting situation when you sit down and think about it!
Idealy, having a free software physics engine + PhysX geared for hardware, and being able to choose which one is compiled in, would be the way to go.
Some people may just need nice simple collision reaction, in which case RigidBody should be fine, but if you need a little more, then having something like NewtonDynamics implemented could be better (for, say, chains, joints, better vehicles), and then if you want to go overboard, PhysX for you.
A physics engine deffinatly needs to be shoved down Torques underpants someday soon, before its too damned outdated. Look at how long it took to even start getting up to date with shaders! And it still aint done.
So, physics engine in a couple of years? With milestones? (not that I'm trying to complain about TSE mind you)
07/20/2006 (3:15 pm)
Quick! Tickle him!!!! Quite an interesting situation when you sit down and think about it!
Idealy, having a free software physics engine + PhysX geared for hardware, and being able to choose which one is compiled in, would be the way to go.
Some people may just need nice simple collision reaction, in which case RigidBody should be fine, but if you need a little more, then having something like NewtonDynamics implemented could be better (for, say, chains, joints, better vehicles), and then if you want to go overboard, PhysX for you.
A physics engine deffinatly needs to be shoved down Torques underpants someday soon, before its too damned outdated. Look at how long it took to even start getting up to date with shaders! And it still aint done.
So, physics engine in a couple of years? With milestones? (not that I'm trying to complain about TSE mind you)
#30
At any rate, I think the technology is too young and unproven to jump on the PhysX bandwagon. And if PhysX were to become the standard and/or open source, then it would seem viable, but as long as they (Aegia) choose to go it alone and charge for licensing, I don't know how it will fare...
07/20/2006 (5:41 pm)
Let me throw in my $0.02 here: IMHO, dedicated physics processors are a solution looking for a problem. In other words, with the continued increase of processing power and addition of multiple cores of both CPUs and GPUs, is there even a need for a dedicated physics processor?At any rate, I think the technology is too young and unproven to jump on the PhysX bandwagon. And if PhysX were to become the standard and/or open source, then it would seem viable, but as long as they (Aegia) choose to go it alone and charge for licensing, I don't know how it will fare...
#31
I would love to see physics being used in games to some effect, but it seems that it is not currently feasible except in extreme cases and most of those are non-realistic situations. I find it interesting that the demos for Aegia, Tokamak, ODE, Newton, etc are more physics intensive (and hold more gameplay potential as even static movies) than most games utilize.
07/20/2006 (7:01 pm)
My problem with physics hardware is the same problem that I have with physics software: lack of use in gameplay. Sure there are examples like The Incredible Machine or Half-Life 2 which use them, but the vasy majority use them for visual bling, like shaders. Resident Evil 4 actually used ragdoll collision when knocking back enemies. I liked that. It was the first time that I remember ragdolls mattering in a game other than as explosion fodder.I would love to see physics being used in games to some effect, but it seems that it is not currently feasible except in extreme cases and most of those are non-realistic situations. I find it interesting that the demos for Aegia, Tokamak, ODE, Newton, etc are more physics intensive (and hold more gameplay potential as even static movies) than most games utilize.
#32
Specifically, I recall the physics in Motor City Online. It was incredible.
But - it wasn't so intense that it needed a co-processor or an insanely powerful box.
Still, it was pretty hot physics.
07/20/2006 (7:51 pm)
David - You are largely correct, but I disagree with you in one specific genre... racing games.Specifically, I recall the physics in Motor City Online. It was incredible.
But - it wasn't so intense that it needed a co-processor or an insanely powerful box.
Still, it was pretty hot physics.
#33
07/20/2006 (8:18 pm)
Yes, racing games have physics, often intensive ones...but they are often massively fudged for gameplay depending on the dependencies on "real" physics or "arcade" physics. I felt that Burnout had better physics than MCO (unless driving on a "grass" texture), but that may have had a lot to do with latency as well. I will definitely agree that racing games are a genre which has physics embedded into it, unlike most others which have a "good feel" and are now trying to capitalize on advanced physics without actually using them. Racing and to a much more limited extent flight-sims are physics oriented because of the nature of what they are. I think that this is one of the most daunting things about TGB. 2D engines are and have been a "feely" project rather than a realistic project and the interest in realistic physics and the problems with it obscure what people want to do with what they expect to do. And they often do not know where to go to modigy their needs according to their wants.
#34
If you use the physics engine for collision detection you must develop a scheme for moving the collision geometry to the physics layer. This is not trivial but definitely doable. You then step the collision engine and physics engine in the main game loop.
The ODEItem resource is a great starting point at using TGE's built in collision and passing the info to the physics engine for simulation. Look at this integration and set some breakpoints in the c++ code and watch how it steps through the loop. This will help you tremendously.
Integrating a physics engine with networking support is not easy but can be done. I'm sure with some dedication you will see quick progress. Good luck!
07/21/2006 (6:57 pm)
If you want to integrate a physics engine you can either use TGE's collision or the collision detection included with the physics sdk.If you use the physics engine for collision detection you must develop a scheme for moving the collision geometry to the physics layer. This is not trivial but definitely doable. You then step the collision engine and physics engine in the main game loop.
The ODEItem resource is a great starting point at using TGE's built in collision and passing the info to the physics engine for simulation. Look at this integration and set some breakpoints in the c++ code and watch how it steps through the loop. This will help you tremendously.
Integrating a physics engine with networking support is not easy but can be done. I'm sure with some dedication you will see quick progress. Good luck!
#35
So you spend $300 to get a phys-x card for your development rig,
code in a few measely "if PhysXHardWareFound = 1 then ....."
add a ton of spiffy hardware stuff, and not have to pay $50,000
So you'd be saving $49,700?
And you get to put the Ageia Phys-X mark on your project.
Please explain how that is not advantageous?
AGEIA HAS NEVER SAID THAT A PROJECT USING PHYS-X HAS TO REQUIRE A PHYS-X ACCELERATOR, just that it has to provide extensive support for it.
And freaking think about it, if there was an official GG phys-x integration than odds are good that all of the "extensive use of the phys-x accelerator" requirements would have been done for you, you'd just need to find a way to use it in your levels/missions/maps/intefaces/etc.
07/25/2006 (12:34 pm)
Okay, I'm not seeing how anyone could oppose phys-x integration?So you spend $300 to get a phys-x card for your development rig,
code in a few measely "if PhysXHardWareFound = 1 then ....."
add a ton of spiffy hardware stuff, and not have to pay $50,000
So you'd be saving $49,700?
And you get to put the Ageia Phys-X mark on your project.
Please explain how that is not advantageous?
AGEIA HAS NEVER SAID THAT A PROJECT USING PHYS-X HAS TO REQUIRE A PHYS-X ACCELERATOR, just that it has to provide extensive support for it.
And freaking think about it, if there was an official GG phys-x integration than odds are good that all of the "extensive use of the phys-x accelerator" requirements would have been done for you, you'd just need to find a way to use it in your levels/missions/maps/intefaces/etc.
#36
Just because you CAN get novodex for free in your game by supporting the physX ppu, doesnt mean it makes sense for your game.
Do we need advanced rigid body physics, particle effects, and ragdoll for a platofmring game?
Probably not, but i dont know.
Most cases, it's better to base off of the Torque rigid body and work it so it's more polished, as that was written with network effeciency in mind, as opposed to most 3rd party physics engines which are generally geared to a client-side-only or single player envionrment.
Are physX and novodex awsome? of course!
Should we support them? i dont see why not.
But the demand for the implementation is solely based on the fashion of games being developed here, which generally wouldnt make use of Novodex as intended.
As for HavokFX; first: client-side-only, it's desinged for special graphical effects like sparks, debris, leaves, etc.
anything that cannot have an impact on gameplay. second: you cant so stuff like standard havok in HL2, because it's totally client-side only. If it was implmeneted as such, it'd be atrociously sloppy and objects wouldnt line up. third: it *requires* an SLI configuration to use, so it'll ultimately be as cheap, if not cheaper, to just get the PhysX chip.
07/25/2006 (2:23 pm)
I'd like to take the time to point out that as awsome as PhysX and Ageia's physics engine is, most games being developed here on GG by the community are focused towards multiplayer/online.Just because you CAN get novodex for free in your game by supporting the physX ppu, doesnt mean it makes sense for your game.
Do we need advanced rigid body physics, particle effects, and ragdoll for a platofmring game?
Probably not, but i dont know.
Most cases, it's better to base off of the Torque rigid body and work it so it's more polished, as that was written with network effeciency in mind, as opposed to most 3rd party physics engines which are generally geared to a client-side-only or single player envionrment.
Are physX and novodex awsome? of course!
Should we support them? i dont see why not.
But the demand for the implementation is solely based on the fashion of games being developed here, which generally wouldnt make use of Novodex as intended.
As for HavokFX; first: client-side-only, it's desinged for special graphical effects like sparks, debris, leaves, etc.
anything that cannot have an impact on gameplay. second: you cant so stuff like standard havok in HL2, because it's totally client-side only. If it was implmeneted as such, it'd be atrociously sloppy and objects wouldnt line up. third: it *requires* an SLI configuration to use, so it'll ultimately be as cheap, if not cheaper, to just get the PhysX chip.
#37
Matter of opinion and implementation.
In Regards to HavokFX: requiring SLI is not a major requirement, it really isn't that expensive if you build yourself (less than 1,000 for a VERY good setup). Definitely not mainstream though. I also thought that the requirements were reduced to just a Shader Model 3.0 card? I could be wrong.
07/25/2006 (8:12 pm)
Client side based is the way to go (several major games have done it this way, Prey most recently). Just communicate events and conditions. Only anything of extreme priority really needs to be synchronized (being if it can obstruct or harm a client). There really is no reason for a rag doll or small explosion debris/dirt to be synchronized exactly, approximation is sufficient. Matter of opinion and implementation.
In Regards to HavokFX: requiring SLI is not a major requirement, it really isn't that expensive if you build yourself (less than 1,000 for a VERY good setup). Definitely not mainstream though. I also thought that the requirements were reduced to just a Shader Model 3.0 card? I could be wrong.
Torque Owner Tim Heldna