![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
Havok comments on GRAW
- Thread starter Sly
- Start date
for a bit of nastalgia here is a good readMrNasty said:OMG THE ARM...
I remember being so impressed by that - it wasn't exactly physics though... sorta quasi.
And you could make it go really weird angles... *ugh*... feeling queasy...
http://en.wikipedia.org/wiki/Trespasser_(game)
specifically about Physics
the type of physics used was pretty easy to tell the short comings..
something as simple as stacking 2 boxes would result in top box just jittering and then floating like it was on an air hockey table..
but the idea of the interactive world ala..
items are placed in the world and they take up real physical use..
eg.
in trespasser you would physically (with the wonder arm) pick up a baseball bat, which had certain physical characteristics and you could pummel a veloceraptor (Sp?) with it..
Oblivion, IMHO, comes close to this as well. All items , say things like cups, bowls weapons, can be picked up and maniuplated within the environment, but in oblivion its all for show, as in you don't pick up a weapon and use it. you pick up a weapon, it goes into your inventory, and that very same weapon is then converted to a "weapon" model that is then used instead of the same item you picked up.
magoo said:
I'm speechless. I can't believe there is really such a thing as pretentious posting. Stop dissecting the definition of "interaction" and amending it with your own prerequisites. The gravity gun example you gave is a perfect example of interactivity, however limited in scope. I assume since you do not have any programming experience you fail to understand how flinging a toilet at a wall is interactivity, but that point doesn't ellude the rest of us, who are trying to explain it to you. Before the world of interactive physics, if you "threw" (believe it or not, this is an example of kinetic interaction, one which you said only exists in the physical world. let me explain it to you. kinetics in games are simplified solutions to kinetic mathematical models, which are based on, and can predict to high accuracy, kinetic energy phenomena) a toilet seat at a wall, it would go straight through. Even something as simple as hitting the wall and bouncing back at an arbitrary (or even constant) trajectory can be defined as physics approximation. Ok? Whew, now get back on topic.
PPU is unproven tech right now. Since we're not wasting any money anyways, there's no reason not to examine future applications of it. I'm guessing that a lot of what is wrong with GR:AW's implementation has to do with latency hiding. A "review" of the PPU:http://www.hexus.net/content/item.php?item=5492
H-street said:
Wow that brings it all back - it was a first for so many reasons
This is particularly interesting:
3D Card Support
The game was developed at a time when the first generation of dedicated PC 3D rendering cards were taking off in the consumer marketplace. The designers had not anticipated adding support for 3D hardware, although it was successfully coded at a late stage in development. Nonetheless the engine did not make effective use of 3D acceleration; the mass of 2D sprites overtaxed the texture memory of contemporary cards, and as a consequence hardware mode uses low-resolution textures. It also disables bump mapping. With early 3D cards Trespasser's hardware mode ran slower, and looked worse, than software mode.
Sound familiar to a situation we have right here? lol - nostalgia 4TW!
I can say one thing about what I've learned today, as I've been called a liar,pretentious, and I guess ignorant.......there is a certain passion about physics processors in this forum that until now has escaped me.
Do I understand the laws of physics,yes.
Did I study physics in school,yes.
Do I appreciate the level of physics in the games we've discussed.yes.
Do I realize what you speak of as far as interaction,yes.
I cannot understand the enthusiasm by which you all throw yourselves at this topic. Aegia and Havok have the same concepts. Yes Ive read the sticky you posted.
Please understand I'm not some jerk that loves to argue, to the contrary, I understand everything I've read to this point. You do not seem to understand me. To make an example again.....the boxes you love in CellFactor.....can you pick one up? can you move them from side to side? or do they only react to an explosion or other massive force??? You can only interact with them in a guided fashion that has already been anticipated.
Thats all I'm saying. I wish to truely interact with physical objects in games.
Thank you for the fun afternoon.
I appreciate the discussion.
Do I understand the laws of physics,yes.
Did I study physics in school,yes.
Do I appreciate the level of physics in the games we've discussed.yes.
Do I realize what you speak of as far as interaction,yes.
I cannot understand the enthusiasm by which you all throw yourselves at this topic. Aegia and Havok have the same concepts. Yes Ive read the sticky you posted.
Please understand I'm not some jerk that loves to argue, to the contrary, I understand everything I've read to this point. You do not seem to understand me. To make an example again.....the boxes you love in CellFactor.....can you pick one up? can you move them from side to side? or do they only react to an explosion or other massive force??? You can only interact with them in a guided fashion that has already been anticipated.
Thats all I'm saying. I wish to truely interact with physical objects in games.
Thank you for the fun afternoon.
I appreciate the discussion.
magoo said:
I used the word "ignorant" and in quotaion, since I don't know any better english word for speaking about something you havn't the proper knowlegde or understanding of.
English is not my native tounge and I am sure there a loads of words I don't know, but the ones I know I seldom use wrongly.
And your first remarks is damn close to being yet another fallcy.
One called "Argumentum ad misericordiam"(Appeal to pity)
Then what is the problem?
I am not the only one telling you that your defination of "interaction" is off base?
And you could be a bloming racket sciencetist for all I care, that would still not alter the fact that you are either are using the word "interaction wrong, or as pointed out in another post tou dn't know waht the word "interaction means.
First off...this is the [H]ardforum.
We are (most of us I guees) in one form or another hardware(and software) enthusiasts.
This should come as no surprise to you?
But your argument is a fallacy:
A best: "Argumentum ad ignorantiam"
Agiea is offering both effects physics(none interactive via PPU) and dedicated hardware gameplay physics(interactivevia PPU) via their PhysX card.
Havock is offering gameplay physcis (interactive via CPU) and hardware accelerated effects physics(none interactive via GPU).
I suggest you go read the FAQ again.
At worst it's a lie.
I disagre, you are still clinging on to your misconception of the word "interactive"
Agreed, and I am not the only one judging from what other peoples post.
Yes?
Infact you can pick up a whole bunch via the "telekinectic" power you pocess in that game
Yes, look at the video?
You just described interaction.
A force that effects/interacts with objects.
Objects that even interact with each other?
Look a the video again please, the evidence is screaming at you.
If we where to use you example, the neither kinetic enery or gravity is interactive.
As they to follow the laws defined by the physics of the universe.
A club hitting me in the head and transfering the combined effect of it's weight, mass, fabric and speed woundn't be interactive by your "rules".
And that is simply very, very, very, very, very wrong!
1...2...3...4...5...6...7...8...9...10...*breathes out*
What do you call the interaction between player and objects...or between objects and objects in Cell Factor
"Scripted events"?!
You are running around in cricles, STILL insisting you your poor understanding of the word "interactin".
Hell, just turning the car in Need For Speed Most wanted is interaction!
This havn''t been fun for my part.
Having to repeat myself(and others) in order to get you too understad the meaning of the word "interaction" is not fun...it's sad.
University degree or not.
Terra - I don't have a university degree, but I understand the word "interaction" better than you it seems
D
Deleted member 133315
Guest
double post
D
Deleted member 133315
Guest
Sly said:
This is taken from the Aegia website:
In May 2006, Tom Clancys Ghost Recon Advanced Warfighter from Ubisoft, one of this years most anticipated titles, will hit the shelves with a wide variety of new knock your socks off PhysX features. The Microsoft Xbox360 version is already getting 9.9-class reviews with a big chunk of physics being driven by AGEIA PhysX software.
http://physx.ageia.com/faqs.html
so it looks as tho it is Ageia physics in the xbox 360 game and not havok...
EVIL-SCOTSMAN said:
Taken from Havoks website
http://www.havok.com/content/blogcategory/29/73?platformid=2
Taken from GRAW's website
http://www.ghostrecon.com/uk/newspost.php?id=11925
MrNasty said:Good question about which is better in software mode - Haven't seen anything on that yet. Might be difficult to put two different games that use the two head to head though. Maybe an SDK Iron cage? (they both have demonstration modes)
Looks like we're already seeing it. Now all we have to do is figure out how to force GRAW to use the PhysX API even without the PPU.
D
Deleted member 133315
Guest
Sly said:
Yea, i see that, but is that not just for the pc version ? as it only states that its uses havok for the pc version.
I am on about just the xbox 360 version using ageia, as why would there be a need to use 2 different types of physics software on the 360 when one would do, as there is no hardware to support, so i guess it would only need 1 type of physics software on the console version, as my copy of graw for the 360 has ageia plastered all over it, but there is no mention of havok.
The pc version however is a different story, those with hardware ppu's will use ageia while others who havent got the ppu will probably use havok ? i dont really know, but i dont think there is havok in the console version of graw.....
on the graw site it says this :
We are also using the Havok engine as a standard, so gamers not equipped with AGEIA chips can expect great physics too!
that leads me to believe that its only for the pc version, as there is no ppu in the 360, but physics effects can be purely software instead of hardware with ageias sdk, i.e there is no need to have a ppu in the pc to have ageias physics working.
There were 2 different teams making the graws, one for the 360 and one for the pc, so i really think that the 360 team only used ageia, but i dont know.
So i suspect that for the xbox version, it is only using ageia, and that havok is used only in the pc version ??? if havok was on the 360 version, then i would of thought it would be plastered everywhere like Ageia is on the 360 version.
Bottom line is, whoever made the 360 version made a real leet job, and whoever made the pc version should be shot.
for once, i can honestly say that the console version of a game is actually better than its pc counterpart, and the pc makers of graw have totally killed the ghost recon name with a lame game set in mehico. Mehico of all places
Ghost recon should be set in the wilderness, forests etc etc, but downtown mehico
Actually, can you post some articles that says that? I've been googling and i haven't found one specifically involving Ageia with the console version, always something about how the console can support PhysX but never saying if it's actually used on the console version of GRAW. Standard marketing lingo to never say anything definite the user can hold against them and leaving it up to the user to assume
Havok articles on the other hand says "We use Havok in GRAW in Xbox360".
I used these search keys
Ghost Recon: Advanced Warfighter +xbox360 +Havok -pc +physics
Ghost Recon: Advanced Warfighter +xbox360 +Ageia -pc +physics
I did find this:
http://www.mobygames.com/game/xbox360/tom-clancys-ghost-recon-advanced-warfighter/credits
Havok articles on the other hand says "We use Havok in GRAW in Xbox360".
I used these search keys
Ghost Recon: Advanced Warfighter +xbox360 +Havok -pc +physics
Ghost Recon: Advanced Warfighter +xbox360 +Ageia -pc +physics
I did find this:
http://www.mobygames.com/game/xbox360/tom-clancys-ghost-recon-advanced-warfighter/credits
D
Deleted member 133315
Guest
Well, am just going by the fact tha it says ageia physics on my graw box, and also ingame credits if i remember correctly.
my 360 is bust atm, power light is red and no power is getting to console, so its in a state of limbo atm.
but google graw xbox 360 ageia and there is numerous hits regarding aegia in graw for the 360, and with the advertising that i see in my version of the game, then i assumed it was ageia, as why would there be a need for 2 types of physics software in the console version when there is no physics hardware in it ? surely one or the other would do ? its not like a pc where if you havent got a ppu then you can rely on havok for physics, and if you have the ppu you get ageia's.
Me personally, i just cant see why there would be 2 different types in the console version especially as there is no ageia ppu in the 360. and if it says ageia physics on my game, then it "must" be in there to some degree.
http://ccfx.net/360era/index.php?p=newsfull&id=6989
it says in there that the physics that is implemented in the console version is supplied by ageia.
Also when you complete graw for the 360, it says that one of the tools used to make it was ageia physics. i never saw any mention of havok, it is maybe there, but i never saw it.
now whether it is or is not is no concern of mine, I am just happy the console version was made cuz the pc version is shite and grin have ruined the ghost recon name for the pc as far as i am concerned.
EDIT:
Also, the link you link to regarding the dev team, that is for the online content, which is probably why it has havok instead of ageia, ageia is probably used for the main single player game, and the less intensive havok is used for online play...
Maybe thats the reason that ageia say this :
The Microsoft Xbox360 version is already getting 9.9-class reviews with a big chunk of physics being driven by AGEIA PhysX software.
That is shown above in my first post, the big chunk could relate to the single player game ?
That comes from the ageia website, so no googling is really needed.
If that is the case, then i understand the reasoning behind having 2 different physics solutions in the game.
my 360 is bust atm, power light is red and no power is getting to console, so its in a state of limbo atm.
but google graw xbox 360 ageia and there is numerous hits regarding aegia in graw for the 360, and with the advertising that i see in my version of the game, then i assumed it was ageia, as why would there be a need for 2 types of physics software in the console version when there is no physics hardware in it ? surely one or the other would do ? its not like a pc where if you havent got a ppu then you can rely on havok for physics, and if you have the ppu you get ageia's.
Me personally, i just cant see why there would be 2 different types in the console version especially as there is no ageia ppu in the 360. and if it says ageia physics on my game, then it "must" be in there to some degree.
http://ccfx.net/360era/index.php?p=newsfull&id=6989
it says in there that the physics that is implemented in the console version is supplied by ageia.
Also when you complete graw for the 360, it says that one of the tools used to make it was ageia physics. i never saw any mention of havok, it is maybe there, but i never saw it.
now whether it is or is not is no concern of mine, I am just happy the console version was made cuz the pc version is shite and grin have ruined the ghost recon name for the pc as far as i am concerned.
EDIT:
Also, the link you link to regarding the dev team, that is for the online content, which is probably why it has havok instead of ageia, ageia is probably used for the main single player game, and the less intensive havok is used for online play...
Maybe thats the reason that ageia say this :
The Microsoft Xbox360 version is already getting 9.9-class reviews with a big chunk of physics being driven by AGEIA PhysX software.
That is shown above in my first post, the big chunk could relate to the single player game ?
That comes from the ageia website, so no googling is really needed.
If that is the case, then i understand the reasoning behind having 2 different physics solutions in the game.
Sly said:
We are seeing that the game's using Havok for all the important stuff and PhysX for the eyecandy - if we could hack it so that ageia handles the gameplay too then that would be a good comparison
And X360 GRAW uses only Ageia as far as I can tell - looking through the credits in game Havok isn't mentioned but Dolby, Ageia, and all other contributors are so I suspect they had nothing to do with it...
magoo said:
As the poster above me has answered, all of the questions you ask about the CellFactor demo are answered in the public video. You can do all of the things you question. The basic premise is that all of the objects in the demo (except player entities) are calculated as rigid-body objects, and can be moved, tossed, react to being bumped into, etc. When you describe reaction to an explosion or massive force, you seem to imply a scripted event. NOTHING IN THE DEMO IS SCRIPTED. Try to understand that. The reason they use "telekenisis" to move the boxes in the game instead of just having the player be able to run through them, is so that game could remain playable. Ever played CS:S? You can bump into, move, and get stuck between all of the sh:t laying around on the ground. It is really frustrating (but another example of "true" interactivity). If you want another way to truly interact with physical objects in the game, stick your _____ in the floppy drive.
And don't fling your degree around like it means something. It makes you seem pretentious (as I mentioned earlier). Your posts have a clear air of condescension directed towards us. And a lot of people went to college. You studied physics? So did 98% of everyone else who didn't go for literature or theater majors. And if you are having fun arguing about physics on computer forums, get out more and see how much fun the world can really be
.Obdicutmodo
n00b
- Joined
- May 4, 2006
- Messages
- 7
Pakotlar, Terra
I think you guys are making a mistake in your argument with Magoo.
He's right to point out that the physics are not truly interactive in a realistic fashion. He's right to point out the Ageia chip is an accelerator.
The Ageia chip does not simulate a real-world physics environment-- to do that in anything close to an accurate manner still takes a supercomputer. What Ageia's API does, like any physics API, is use a series of approximations, timed guesses, etc. in order to simulate the physics effects.
This does not allow for unplanned, unscripted interaction, and there's always going to be areas where the physics collapses and the validity of the interaction is lost. The characters in cell factor use some force-like power to pick stuff up and chuck it around, but not the character's actual hand.
You're right to point out this is a programming problem , not a hardware problem-- to script that circumstance, you'd have to program for it. But you're wrong to argue against the claim that the Ageia chip is just an accelerator-- it is. There is nothing inherent in the chip that has to do with physics-- it's a parallel processor with inbuilt math equations that fit approximations used in the API.
The question is whether it's needed, or even valuable, or if in the end it's rather limiting. Ageia's product is dual-- it's the actual chip, and the physics API for the chip. They distribute the API for free, which is why they can claim they support so many games-- they give it away for free in the hopes that people will start using the chip. But the API itself, the programming of it and on it, is still what determines the physics in the game, how good they are, how accurate, and how true.
I work in the video game industry, and I know that with talented programmers, you have no need of an ageia chip. By coming up with efficient, elegant algorithms, programming intelligently, and making good use of memory managmenet, you don't need 128 megs dedicated to physics. And if in fact the PPU was ever using its full capacity, outputting that information through a PCI or even PCI-E bus back to the CPU and to the GPU is a very, very bad solution to the problem.
What I like about the Havok FX solution-- which is, yes, "just" for effects-- effects are rather important, but never mind that for now-- is that it's resident on the GPU-- so there's no use of PCI bandwith. The physics calculations for the objects are done on the GPU and rendered by the GPU, so it can take advantage of the massive internal bandwith of the GPU.
As for "gameplay" physics-- the Havok physics API is what was used for halflife 2, Oblivion, etc. The interaction there, as defined by you, is quite well done, even if personally I think that oblivion could have used more time fine-tuning the physics. The solution is, as always, better programming.
Think about your daily life, or even a firefight-- how many actual physics interactions that really matter in terms of gameplay are there going to be at any given time? If you're driving a vehicle, you can simulate the whole thing with impressive fidelity with a couple hundred interactions-- the frame of the vehicle with the suspension, the tires with the suspension, the transmission with the gears with the axles, the tires with the road, etc. And that's it-- you don't need to simulate thousand and thousands of high-fidelity interactions.
Or, to put it another way: Since the physics in a game are always going to be approximations until we reach quantum computing, the argument should not be about the hardware, it should be about the software. The Havok software has been used in a very, very wide array of different games to excellent effect and in very interesting ways-- the Ageia API is relatively untested, and if you look at their list of games, the majority of them are relative unknowns. When they're giving it away for free but people still find value in buying Havok, that says very good things about Havok.
Finally, I called Ubisoft, and confirmed that the Havok physics code is used for the gameplay physics in the 360 version. The Ageia physics API is used for some of the effects physics, because there are some aspects of effects it handles better-- and its free. The Havok logo was left off the box by an oversight. Call Ubisoft corporate if you don't believe me.
I think you guys are making a mistake in your argument with Magoo.
He's right to point out that the physics are not truly interactive in a realistic fashion. He's right to point out the Ageia chip is an accelerator.
The Ageia chip does not simulate a real-world physics environment-- to do that in anything close to an accurate manner still takes a supercomputer. What Ageia's API does, like any physics API, is use a series of approximations, timed guesses, etc. in order to simulate the physics effects.
This does not allow for unplanned, unscripted interaction, and there's always going to be areas where the physics collapses and the validity of the interaction is lost. The characters in cell factor use some force-like power to pick stuff up and chuck it around, but not the character's actual hand.
You're right to point out this is a programming problem , not a hardware problem-- to script that circumstance, you'd have to program for it. But you're wrong to argue against the claim that the Ageia chip is just an accelerator-- it is. There is nothing inherent in the chip that has to do with physics-- it's a parallel processor with inbuilt math equations that fit approximations used in the API.
The question is whether it's needed, or even valuable, or if in the end it's rather limiting. Ageia's product is dual-- it's the actual chip, and the physics API for the chip. They distribute the API for free, which is why they can claim they support so many games-- they give it away for free in the hopes that people will start using the chip. But the API itself, the programming of it and on it, is still what determines the physics in the game, how good they are, how accurate, and how true.
I work in the video game industry, and I know that with talented programmers, you have no need of an ageia chip. By coming up with efficient, elegant algorithms, programming intelligently, and making good use of memory managmenet, you don't need 128 megs dedicated to physics. And if in fact the PPU was ever using its full capacity, outputting that information through a PCI or even PCI-E bus back to the CPU and to the GPU is a very, very bad solution to the problem.
What I like about the Havok FX solution-- which is, yes, "just" for effects-- effects are rather important, but never mind that for now-- is that it's resident on the GPU-- so there's no use of PCI bandwith. The physics calculations for the objects are done on the GPU and rendered by the GPU, so it can take advantage of the massive internal bandwith of the GPU.
As for "gameplay" physics-- the Havok physics API is what was used for halflife 2, Oblivion, etc. The interaction there, as defined by you, is quite well done, even if personally I think that oblivion could have used more time fine-tuning the physics. The solution is, as always, better programming.
Think about your daily life, or even a firefight-- how many actual physics interactions that really matter in terms of gameplay are there going to be at any given time? If you're driving a vehicle, you can simulate the whole thing with impressive fidelity with a couple hundred interactions-- the frame of the vehicle with the suspension, the tires with the suspension, the transmission with the gears with the axles, the tires with the road, etc. And that's it-- you don't need to simulate thousand and thousands of high-fidelity interactions.
Or, to put it another way: Since the physics in a game are always going to be approximations until we reach quantum computing, the argument should not be about the hardware, it should be about the software. The Havok software has been used in a very, very wide array of different games to excellent effect and in very interesting ways-- the Ageia API is relatively untested, and if you look at their list of games, the majority of them are relative unknowns. When they're giving it away for free but people still find value in buying Havok, that says very good things about Havok.
Finally, I called Ubisoft, and confirmed that the Havok physics code is used for the gameplay physics in the 360 version. The Ageia physics API is used for some of the effects physics, because there are some aspects of effects it handles better-- and its free. The Havok logo was left off the box by an oversight. Call Ubisoft corporate if you don't believe me.
pakotlar said:As the poster above me has answered, all of the questions you ask about the CellFactor demo are answered in the public video. You can do all of the things you question. The basic premise is that all of the objects in the demo (except player entities) are calculated as rigid-body objects, and can be moved, tossed, react to being bumped into, etc. When you describe reaction to an explosion or massive force, you seem to imply a scripted event. NOTHING IN THE DEMO IS SCRIPTED. Try to understand that. The reason they use "telekenisis" to move the boxes in the game instead of just having the player be able to run through them, is so that game could remain playable. Ever played CS:S? You can bump into, move, and get stuck between all of the sh:t laying around on the ground. It is really frustrating (but another example of "true" interactivity). If you want another way to truly interact with physical objects in the game, stick your _____ in the floppy drive.
And don't fling your degree around like it means something. It makes you seem pretentious (as I mentioned earlier). Your posts have a clear air of condescension directed towards us. And a lot of people went to college. You studied physics? So did 98% of everyone else who didn't go for literature or theater majors. And if you are having fun arguing about physics on computer forums, get out more and see how much fun the world can really be
You guys are really overwhelmed with yourselves.
Im not flinging anything anywhere. Nor am I sticking anything anywhere.
Your rude comments strip you of any air of intellegence you pretend to possess.
I am simply giving my opinion. Right or wrong as you see it.
The latin quotes are cute, thanks alot by the way.
Im really and truely sorry if Ive offended anyone. Its not my objective or intention.
This has degenerated into name-calling which is inappropiate.
tornadotsunamilife
2[H]4U
- Joined
- Jun 26, 2004
- Messages
- 2,492
MrNasty said:
Not wanting to sound ignorant but don't games like Halo for Xbox do this too?
digitalfreak
Limp Gawd
- Joined
- Apr 18, 2005
- Messages
- 356
I don't believe you've been called a poo poo head yet...magoo said:
POO POO HEAD
Obdicutmodo said:
Read my posts again. Never did I argue that the PhysX was anything but an accelerator. Of course it is. It is just a stream processor with a pretty good double precision flop throughput.
However, you are mistaken when you say that the Ageia PPU, or a CPU for that matter, does not simulate real world physics. Supercomputers are not inherently any different than your average consumer pc, just orders of magnitude faster at certain types of parallel ops. Both supercomputers and consumer pc's can approximate physics, the difference is the accuracy of the physics models. Wheras Havoc and Ageia's provide fairly limited accuracy, top-notch supercomputers can simulate to an almost arbitrary degree of accuracy even complex physics like fluid dynamics (though even these models are innacurate).
What I'm trying to get through to you is that the difference between interacting with a character's hand as opposed to "telekenisis" is completely arbitrary. As a programmer you should know this. If anyone actually tried to calculate force interaction of a character's hand on game world objects, the results would be devastatingly clumsy, but very possible. Trespassor did this 10 years ago. In Oblivion, you can "pick up" objects with your hand, and it allows completely unscripted, real-time interaction with the gameworld, however it is no different from picking up an object in cellfactor with "telekenisis", except for the visual effect.
"This does not allow for unplanned, unscripted interaction, and there's always going to be areas where the physics collapses and the validity of the interaction is lost"
Of course there will be areas where the physics collapse, but that has nothing to do with the validity of the interaction. What you are doing here is making an arbitrary judgment on the degree of interaction. My point isn't that the physics models are completely accurate, because they clearly are not. I am making the distinction between scripted interaction (aka pre-canned animation) and real-time physics approximation (aka ragdoll in Havoc). The degree of validity is a personal opinion based on what degree of representation you expect. The fact that it is an unscripted interaction is fact.
As far as PPU-CPU-GPU interaction, I agree that developers will have to spend a fair chunk of time hiding latency from PPU-GDDR3-PPU bandwith, PPU-CPU bandwith, and CPU-GPU bandwith. A better, but unlikely solution, would be a consumer pc version of a math co-processor, in a socket format, close to the CPU, with a fat bus.
What programming do you do? What projects have you worked on?
edit: Ah, I see that you did not state that you were a programmer, but "worked in the game industry". If you've had any physics classes, the mathematical models that you use in entry level calc physics (had to take some for my biomed eng. degree) are relatively gross approximations. Yet no one would argue that they are not describing object interaction, albeit roughly. In c++ at least, you can use similarly rough alghorithms for solutions to basic mass, velocity trajectory calcs. These are in every sense of the word "real" physics, and yet do not take into account many variables. However, that does not invalidate them. Game worlds do not inherently contain things like surface tension, gravity, mass, air density, etc. But now I'm going in circles.
My main criticisms with mangoo's posts were that he was mistaken in his understanding of what defines unscripted game interaction, and the tone of his posts.
Obdicutmundo - go and read some of the topics in this forum, as most of what you have said has already been discussed with excellent points by many people. PCI bus with not enough bandwidth etc all included and that part in particular has been debunked.
I especially like the way you inconvenienced yourself to phone ubisoft corporate headquarters to vouch for Havok's use in the XB360 version. Havok takes their branding very seriously so Ubisoft have some explaining to do it seems - they have a comment about that? Or don't they talk to random people phoning them about the technology they use in their games? Because, see, I just got off the phone to Forbes and they say that Bill Gates' investments are doing just fine, but they had a little problem with undeclared taxes earlier in the year.
Anyhoo - this all seems to have gone off-topic, so let's put it this way:
Physics |= Gameplay
Gameplay |= and is not dependant upon Hardware used
This subforum is about physics hardware, so go and monger your disappointment with programmers elsewhere or, better yet, don't buy any games and give your PC to someone who wants to play them.
Whether or not the gameplay in CF itself is "interactive" by your definition is neither here nor there. The style was not possible before Ageia's PPU came along, nor would it be with HavokFX. These are not opinions, these are facts. Facts based on the knowledge I have gained by extensive research into patents, websites, friends, colleagues and education - and you would do well not to criticize these lightly.
Havok still has little market penetration (38 titles over 5 years. Wow that's great), whereas Ageia have over 100 titles in developement (sifting out the crap, that's about 10). 10 in 2 years (technically 17 months, that's how long the SDK has been in circulation in the PhysX guise), versus 38 in 5. Tell me how you can claim Havok is an "established" SDK, or leave it at the door - what about Meqon and ODE? Or hadn't you heard of those?
You also hasten to add that HavokFX is just effects physics but as it uses your GPU that's OK. Well I got news for you - you may not notice effects or anyother kind of physics in firefights, but when they bottleneck your card you will. Time will tell if that happens, but as there has not even been an effective demonstration of HavokFX on an open system I think there has to be some work done on it yet, not to mention it costs extra, above and beyond, what the normal Havok SDK costs.
To round off, I am shocked by your ignorance in modelling. I work with CFD on a daily basis and am sorry to say you are completely wrong to the letter about everything you have stated to do with modelling dynamics. If that is any indication of your knowledge on other topics then by that fact alone your statements are at best apocryphal.
I especially like the way you inconvenienced yourself to phone ubisoft corporate headquarters to vouch for Havok's use in the XB360 version. Havok takes their branding very seriously so Ubisoft have some explaining to do it seems - they have a comment about that? Or don't they talk to random people phoning them about the technology they use in their games? Because, see, I just got off the phone to Forbes and they say that Bill Gates' investments are doing just fine, but they had a little problem with undeclared taxes earlier in the year.
Anyhoo - this all seems to have gone off-topic, so let's put it this way:
Physics |= Gameplay
Gameplay |= and is not dependant upon Hardware used
This subforum is about physics hardware, so go and monger your disappointment with programmers elsewhere or, better yet, don't buy any games and give your PC to someone who wants to play them.
Whether or not the gameplay in CF itself is "interactive" by your definition is neither here nor there. The style was not possible before Ageia's PPU came along, nor would it be with HavokFX. These are not opinions, these are facts. Facts based on the knowledge I have gained by extensive research into patents, websites, friends, colleagues and education - and you would do well not to criticize these lightly.
Havok still has little market penetration (38 titles over 5 years. Wow that's great), whereas Ageia have over 100 titles in developement (sifting out the crap, that's about 10). 10 in 2 years (technically 17 months, that's how long the SDK has been in circulation in the PhysX guise), versus 38 in 5. Tell me how you can claim Havok is an "established" SDK, or leave it at the door - what about Meqon and ODE? Or hadn't you heard of those?
You also hasten to add that HavokFX is just effects physics but as it uses your GPU that's OK. Well I got news for you - you may not notice effects or anyother kind of physics in firefights, but when they bottleneck your card you will. Time will tell if that happens, but as there has not even been an effective demonstration of HavokFX on an open system I think there has to be some work done on it yet, not to mention it costs extra, above and beyond, what the normal Havok SDK costs.
To round off, I am shocked by your ignorance in modelling. I work with CFD on a daily basis and am sorry to say you are completely wrong to the letter about everything you have stated to do with modelling dynamics. If that is any indication of your knowledge on other topics then by that fact alone your statements are at best apocryphal.
D
Deleted member 133315
Guest
MrNasty said:
LMFAO
quote of the week
Obdicutmodo
n00b
- Joined
- May 4, 2006
- Messages
- 7
McNasty-- I work in the industry-- I assume you do to. If you work in the industry, call Ubisoft, and want to know something that's relatively public information, you'll be able to find out.
I'm shocked by your manners.
I'm not disappointed in programmers. Programmers are driven by managers, who, at most companies, are driven by marketing, who sets unrealistic dates for game publishing. "We have to have this out by christmas!" is not a good way to get a game done competently.
Looking on Havok's site, they're used in more than 150 titles, not the 38 you claim. Where'd you get that number?
If by "modeler" you're referring to me talking about how physics is calculated using rough approximations, etc-- can you point out any ways that I'm wrong, or are you just going to keep making accusations?
I'm shocked by your manners.
I'm not disappointed in programmers. Programmers are driven by managers, who, at most companies, are driven by marketing, who sets unrealistic dates for game publishing. "We have to have this out by christmas!" is not a good way to get a game done competently.
Looking on Havok's site, they're used in more than 150 titles, not the 38 you claim. Where'd you get that number?
If by "modeler" you're referring to me talking about how physics is calculated using rough approximations, etc-- can you point out any ways that I'm wrong, or are you just going to keep making accusations?
I am all for the PPU. To be honest, I'm happy with UT2k4 graphics. I don't have money to burn like a lot of you do, so given the choice between better graphics, or gameplay that's in the Cell Factor video, I would definately take the gameplay.
Obdicutmodo said:
Why would you state that? Because I don't believe you took the time to bother someone at ubisoft's corporate HQ - the public answering section of which actually knows very little about games? If you work in the industry, why not phone the developers?
In addition, you can't be shocked by my manners, which are in and of themselves impeccable, but you may be shocked by the manner/tone/rudeness of my reply. If I have offended you severely without cause, please accept my sincerest apologies. If I have pointed out a flaw in your opinions, which you express as facts, and that offends you, then find solace in the fact that I am not attacking you as a person, just your method of expression and content thereof.
He said/she said arguments are very fanciful face to face, let alone on an anonymous public forum. People who take things for face value in these environments are quite simply gullible, and I take exception to people who capitalize on that.
Obdicutmodo said:
I agree with you completely and the comment to which you were referring was aimed at mrmagoo - I apologise for not making that clear.
Obdicutmodo said:
38 PC titles (you can count them, and they say as such in the article at the start of this thread) - the rest are ports to consoles or scratch games on consoles. You are quite right to point that out though - I was merely trying to point out that PhysX has 100 games in developement for their hardware, and used havok's PC SDK usage for comparison.
Obdicutmodo said:
I'd be happy to - in brief!
"to do anything in a close to accurate manner requires a super computer"
I ran a custom Fortran program that can model and solve laminar film instabilities with an accuracy of about 90% in about 12 seconds for a 30 second time period, this is on a small beowulf cluster that was current about 8 years ago. It doesn't even distribute on this cluster, and hence runs on one node. These are fairly complex fluids interactions using suitably modified Navier-Stokes equations. How accurate do want it to be? 90%? The power's certainly there. 95%? probably ok. 99%? Starting to be a little out there, but if I used a modern PC it could probably do it. These are just numbers however, anyone doing CFD will tell you that the accuracy required depends what you're modelling, how you're modelling it and what results you want from it.
I am excited about what a matrixed vector processor could do for my work, does that help?
"There is nothing inherent in the chip that has to do with physics-- it's a parallel processor with inbuilt math equations that fit approximations used in the API."
If it had "inbuilt math equations" that fit with those used in the API physics modelling would certainly be "inherent in the chip". As it is that statement is wrong - the driver and API supply the chip with a solver model for the problems that will be solved for during the course of its use.
This is as opposed to the HavokFX model which requires an entire code branch to be loaded via a shader programme onto the GPU which is then taken off again when its calculation is complete - that's my understanding of it, please correct me if I am wrong.
"By coming up with efficient, elegant algorithms, programming intelligently, and making good use of memory managmenet, you don't need 128 megs dedicated to physics"
"Coming up with" efficient elegant algorithms to model physics is a rather tremendous task. The engineering community would love to see how you define their problem as, well, merely thinking about it a bit. Or pulling a pink elephant out of a leprechaun's hat under a full blue moon with pigs in full flight on a cold night in hell.
Throwing memory at massively parallel problems puts you in a curious position - you can either take longer to solve the problem by allowing the storage of more transient results or you can allow for more complex solutions. Neither of which is relevant to your previous statement.
"so there's no use of PCI bandwith"
Bus bandwidth is rarely a problem when you're transferring numerical vectors representing xyz co-ordinates and velocities back and forth - this isn't a graphics card, you know, no textures here
And I could go on. On a lighter note, I'd like to apologise to you for perhaps being a bit too hard in my response earlier - mrmagoo's inability to express himself annoyed me and perhaps I let that cloud fall on you too. Your post was obviously made to the best of your knowledge and as someone from another industry I can respect the way you see this problem.
Obdicutmodo
n00b
- Joined
- May 4, 2006
- Messages
- 7
McNasty--
I assume you know that the games market for consoles is far bigger than for PCs, so having 112 titles on consoles, 38 on PCs sounds like a rather successful market penetration, especially for something you're charging for-- unlike Ageia. The games market isn't really divided up into PC and Console-- almost all companies make multi-SKU titles, and Ageia itself is claiming that it does support the 360-- so wouldn't it be fairer to compare their total supported titles? You also compared the Ageia "in production" titles with the Havok "Already out" titles-- Havok doesn't have a list of upcoming titles that I can find
With that said:
You're right that i wasn't nearly deep or accurate enough in my explanation of modeling of real-world physics, and I apologize.
If you're modeling something very specific, like fluid dynamics, tensile strength, spring-loading, etc you can use a relatively small targeted program and not a lot of memory. But if you're trying to model a complex, interactive system with fidelity-- like a ball bouncing off of multiple surfaces, the old "Where is a billiard ball after eight bounces?" problem-- you would need a supercomputer. And to get 100% accuracy of almost any physics model, no matter how simple, requires a supercomputer.
Specialized programs and computing cells for stress measuring, structural engineering, etc. already exist, as I'm sure you know, and they're not small fortran programs running on standard PCs. It's a niche market, and one that a hardware solution does fit, given the costs of failure.
I was also wrong to use the phrase "real world physics" anyway-- and you're right to call me out on it. We're not going to worry about modeling quantum reactions when all we need is newtonian output. we're not going to model brownian motion into our water in Oblivion.
"This is as opposed to the HavokFX model which requires an entire code branch to be loaded via a shader programme onto the GPU which is then taken off again when its calculation is complete - that's my understanding of it, please correct me if I am wrong."
I'm not an expert on HavokFX-- I saw it debuted at GDC and, well, it's like anything a company presents "Great, now I want to see how it actually works". But from what the presenter said, the code branch is not taken off the GPU when it's complete, it just gets directly rendered. The only things loaded back are very rare instances where you want one of those modeled objects to interact back with the player character or another CPU object.
Remember, this is just my understanding. For a better explanation, call Havok and set up an NDA with them-- I probably will, but then I'll be under NDA.
Their solution makes sense for all those times when you want to have mass destruction, trees swaing in the wind, tumbleweed, etc. Think about it: If you have a house, in-game, do you need/want each brick to be an individual object? No, obviously-- the polyc ount alone would be ridiculous, and you'd have to be using massively high fidelity modeling to keep it standing. So instead you construct it out of peices, as many as you dare without overloading your components-- CPU, GPU, sound card even. So then, when a player character blows up the door, you can send that door flying in fragments, using HavokFX, and render it, without ever going back to the CPU. The CPU sends out the call-- blow it up-- the GPU does the calculations, spawns the flying splinters, and uses pixel shaders to move and track them.
If you wanted those splintes actually doing damage, then yes, it'd have to come back to the CPU-- but to me that's one better than everything from the PPU having to go both to the CPU and to the GPU, anyway.
You said "Bus bandwidth is rarely a problem when you're transferring numerical vectors representing xyz co-ordinates and velocities back and forth - this isn't a graphics card, you know, no textures here "
It hasn't been a rpoblem in the past-- but in the past you haven't tried to pass 1000, 10000, 30000 XYZ numerical vectors every millisecond along the PCI bus. And the information being passed back from the PPU to the CPU-- if you wanted the flying bricks or splinters to wound the character-- would not just be XYZ vectors, they'd have to include fragment weight as well, so that the damage system could figure out and apply correct damage-- and there are probably other factors we're both overlooking.
The other massive, massive problem with the Ageia API/Chip combination is that it's an API/Chip combination. What if someone else comes out with another PPU brand? Since Ageia gives their API away for free-- which is probably why they've gotten the penetration they have, by the way-- their only profit motive can be selling PPUs. So they have every reason NOT to support cards other than their own. So basically they're trying to set themselves up as the only PPU company in existence.
Whether or not a PPU in some form or other is appopriate-- I'd welcome one tied directly to the CPU utilizing internal bandwith and API-agnostic-- or tied to the GPU-- the Ageia approach is, to me, all wrong.
Oh, and apologies accepted on all fronts, and I'm sorry for my somewhat slipshod statements earlier.
I assume you know that the games market for consoles is far bigger than for PCs, so having 112 titles on consoles, 38 on PCs sounds like a rather successful market penetration, especially for something you're charging for-- unlike Ageia. The games market isn't really divided up into PC and Console-- almost all companies make multi-SKU titles, and Ageia itself is claiming that it does support the 360-- so wouldn't it be fairer to compare their total supported titles? You also compared the Ageia "in production" titles with the Havok "Already out" titles-- Havok doesn't have a list of upcoming titles that I can find
With that said:
You're right that i wasn't nearly deep or accurate enough in my explanation of modeling of real-world physics, and I apologize.
If you're modeling something very specific, like fluid dynamics, tensile strength, spring-loading, etc you can use a relatively small targeted program and not a lot of memory. But if you're trying to model a complex, interactive system with fidelity-- like a ball bouncing off of multiple surfaces, the old "Where is a billiard ball after eight bounces?" problem-- you would need a supercomputer. And to get 100% accuracy of almost any physics model, no matter how simple, requires a supercomputer.
Specialized programs and computing cells for stress measuring, structural engineering, etc. already exist, as I'm sure you know, and they're not small fortran programs running on standard PCs. It's a niche market, and one that a hardware solution does fit, given the costs of failure.
I was also wrong to use the phrase "real world physics" anyway-- and you're right to call me out on it. We're not going to worry about modeling quantum reactions when all we need is newtonian output. we're not going to model brownian motion into our water in Oblivion.
"This is as opposed to the HavokFX model which requires an entire code branch to be loaded via a shader programme onto the GPU which is then taken off again when its calculation is complete - that's my understanding of it, please correct me if I am wrong."
I'm not an expert on HavokFX-- I saw it debuted at GDC and, well, it's like anything a company presents "Great, now I want to see how it actually works". But from what the presenter said, the code branch is not taken off the GPU when it's complete, it just gets directly rendered. The only things loaded back are very rare instances where you want one of those modeled objects to interact back with the player character or another CPU object.
Remember, this is just my understanding. For a better explanation, call Havok and set up an NDA with them-- I probably will, but then I'll be under NDA.
Their solution makes sense for all those times when you want to have mass destruction, trees swaing in the wind, tumbleweed, etc. Think about it: If you have a house, in-game, do you need/want each brick to be an individual object? No, obviously-- the polyc ount alone would be ridiculous, and you'd have to be using massively high fidelity modeling to keep it standing. So instead you construct it out of peices, as many as you dare without overloading your components-- CPU, GPU, sound card even. So then, when a player character blows up the door, you can send that door flying in fragments, using HavokFX, and render it, without ever going back to the CPU. The CPU sends out the call-- blow it up-- the GPU does the calculations, spawns the flying splinters, and uses pixel shaders to move and track them.
If you wanted those splintes actually doing damage, then yes, it'd have to come back to the CPU-- but to me that's one better than everything from the PPU having to go both to the CPU and to the GPU, anyway.
You said "Bus bandwidth is rarely a problem when you're transferring numerical vectors representing xyz co-ordinates and velocities back and forth - this isn't a graphics card, you know, no textures here "
It hasn't been a rpoblem in the past-- but in the past you haven't tried to pass 1000, 10000, 30000 XYZ numerical vectors every millisecond along the PCI bus. And the information being passed back from the PPU to the CPU-- if you wanted the flying bricks or splinters to wound the character-- would not just be XYZ vectors, they'd have to include fragment weight as well, so that the damage system could figure out and apply correct damage-- and there are probably other factors we're both overlooking.
The other massive, massive problem with the Ageia API/Chip combination is that it's an API/Chip combination. What if someone else comes out with another PPU brand? Since Ageia gives their API away for free-- which is probably why they've gotten the penetration they have, by the way-- their only profit motive can be selling PPUs. So they have every reason NOT to support cards other than their own. So basically they're trying to set themselves up as the only PPU company in existence.
Whether or not a PPU in some form or other is appopriate-- I'd welcome one tied directly to the CPU utilizing internal bandwith and API-agnostic-- or tied to the GPU-- the Ageia approach is, to me, all wrong.
Oh, and apologies accepted on all fronts, and I'm sorry for my somewhat slipshod statements earlier.
I'm glad the air's clear again - let the debate commence! J/K
There are still some things which you say that don't really apply - you don't need a super computer for particle collision modelling unless you use about 10^8 ish particles, and there are ways of optimizing those problems depending on what you require, as I stated earlier. There are some fun ones which demonstrate monte-carlo simulations alongside particle interactions of around 1000 (in a Java applet no less) - they're quite interesting and I'll find them for you.
I do still disagree with your statement about time-dependant problem solving, as these can be presented as a collection of instantaneous timesteps. They don't require extra memory per-say but if all prior timestep solutions are required the whole thing can get beastly - comparable to the memory requirement for a 30 second video clip compared to a 30 second real-time demo. And this is really the crux as to why physics modelling that's accelerated by hardware is a good thing - Going to bed soon so in-depth will have to be another time...
I also believe that there is a lot to be examined in the bus bandwidth issue - that has been discussed elsewhere in the forums and makes for good reading, I believe some hardware design person commented on it.
The real issue for me is that HavokFX relies on spare GPU overhead - an issue that still hasn't been addressed anywhere. Just like platform compatibility (which you point out is important to market penetration) and how open the platform will be - will other companies be able to release SDK's that use GPU's? or will Havok block such attempts? What does this closed standard mean for an OpenPhysics style API? Any games planning on using HavokFX yet? So many unanswered questions.
See, what I like about Ageia's PPU is that they are working on allowing other uses and therefore other API's for their card. It's not that I don't like Havok, it's that Havok reminds me somewhat of Glide with a pricetag - game programmers have a hard enough time producing games in the timeframe required with a budget, wouldn't an OpenGL for physics help?
Glad that the topic's on track again
J/KThere are still some things which you say that don't really apply - you don't need a super computer for particle collision modelling unless you use about 10^8 ish particles, and there are ways of optimizing those problems depending on what you require, as I stated earlier. There are some fun ones which demonstrate monte-carlo simulations alongside particle interactions of around 1000 (in a Java applet no less
) - they're quite interesting and I'll find them for you. I do still disagree with your statement about time-dependant problem solving, as these can be presented as a collection of instantaneous timesteps. They don't require extra memory per-say but if all prior timestep solutions are required the whole thing can get beastly - comparable to the memory requirement for a 30 second video clip compared to a 30 second real-time demo. And this is really the crux as to why physics modelling that's accelerated by hardware is a good thing - Going to bed soon so in-depth will have to be another time...
I also believe that there is a lot to be examined in the bus bandwidth issue - that has been discussed elsewhere in the forums and makes for good reading, I believe some hardware design person commented on it.
The real issue for me is that HavokFX relies on spare GPU overhead - an issue that still hasn't been addressed anywhere. Just like platform compatibility (which you point out is important to market penetration) and how open the platform will be - will other companies be able to release SDK's that use GPU's? or will Havok block such attempts? What does this closed standard mean for an OpenPhysics style API? Any games planning on using HavokFX yet? So many unanswered questions.
See, what I like about Ageia's PPU is that they are working on allowing other uses and therefore other API's for their card. It's not that I don't like Havok, it's that Havok reminds me somewhat of Glide with a pricetag - game programmers have a hard enough time producing games in the timeframe required with a budget, wouldn't an OpenGL for physics help?
Glad that the topic's on track again
EVIL-SCOTSMAN said:
Getting a straight answer from the marketing guys is like pulling teeth
Anyway, i've been researching on what the deal is since both companies released official conflicting statements. Looks like the guy in the other thread beat me to it
Terra said:
Looks like the thread raised a few heads. Hurrah for [H]ardforum!!!
http://www.firingsquad.com/features/ageia_physx_response/page3.asp
Now, would anybody mind explaining what their reply had to do with the quoted posts? My question wasn't answered, and MrNasty's wasn't clarified
Sly said:Looks like the thread raised a few heads. Hurrah for [H]ardforum!!!
If this promotes more full-flegde gameplay physics in games, then I am all for it
*L*I want more...this is [H]ard..not
*L*Dunno, could be that this a coy/paste between the trhead and then a separate statment from Ageia.
But I can only guess here, mail and ask the author?
Terra...
Terra said:If this promotes more full-flegde gameplay physics in games, then I am all for it
I want more...this is [H]ard..notoft
I'm still skeptical about that part. =P
Dunno, could be that this a coy/paste between the trhead and then a separate statment from Ageia.
But I can only guess here, mail and ask the author?
Terra...
Somebody get me some pliers. I'm gonna go over there and play 'dentist'
digitalfreak said:
Well maybe not but I guess I stepped in some.
You can call me right or wrong......but wow, I really learned a bunch from this, now I guess the "discussion" was worth it as it seems the parties involved have also decided to open the debate.
I enjoyed the thread. I did indeed come away with some very good technical information, even if I did seem to ruffle some feathers. I truely appreciate the expertise offered here.
Thank you for putting up with me. Again I apologize to anybody who still thinks I'm annoying,lying,cheating,ignorant,condescending,unrealistic,nearsighted,or just plain stubborn.
Lord of Shadows
2[H]4U
- Joined
- Mar 9, 2004
- Messages
- 3,322
Definition time! YAY \o/
Act
"The process of doing or performing something: the act of thinking."
Interact
"To act on each other: More than a dozen variable factors could interact, with their permutations running into the thousands (Tom Clancy)."
Both these taken from dictionary.com and funny enough one of the quotes being from Tom Clancy who if I'm not mistaken had a hand in the making of the Rainbow 6 games and GRAW.
Essentially what we're saying here is that interaction is allowing objects to manipulate each other such that the manipulation can alter the outcome of one or more of the objects interacting. In this sense the Ageia physics engine and dedicated PPU aid in setting up a virtual environment where by users can interact with objects within the world and those objects can then go on to interact with other objects.
The method of doing so is an aproximation of physics in the real world, but do not be mistaken this does not mean it is scripted in any way, it simply means the movement of the object in the virtual world will not exactly match the movement of one in the real world. Even with current physics technology (and by that I mean everything before ageia) does a pretty good job of allowing objects to interact with each other, they just cannot do it on a very large scale.
Some of you (magoo mainly) are appearing to others to know very little about the subject when in fact you may have a very good grip of real physics (I did an A level in physics at college) the problem is you do not understand the implimentation of physics in a game engine.
Magoo you state in your 2nd post:
You can absolutly do that with the current physics systems and since the ageia physics system can do everything the current generation can do, plus more that implies you can do that with ageias physics system as well.
The problem you seem to have is distinguishing the physics system from the game code or general game engine. The Ageia physics engine simply calculates the physical interaction of objects either on their own or together as one object effects another. The game code is what allows the gamer designers to allow the players to interact with this system. The very basic level of which is simply moving an avatar around the virtual world and simply knocking into these objects, but there is no reason why that avatar cannot pick up an object and throw it, that just has to be coded as an ability into the game and require the correct user input to do so.
The limitation here quickly becomes the engie coders ability to design new ways for the players to interact with the objects in the world, sure picking them up is easy, throwing them is also fairly easy. But the more you want your player to do the more you have to code into the engine and that becomes a bottleneck in production.
Because a player can pickup an object and throw it with telekenetic powers but not pick up the same object with his hands doesn't mean the system cannot manage this type of interaction, it simply means that the game developers have not coded this particular interaction between the players and the physics enabled objects in the world.
In the case of cell factor it is most likely due to huge number of objects the player will need some way of manipulating, sure you could pickup a crate and throw it at an enemy, or you could fling 1000's of crates at him with a cool effect causing crates to fly everywhere. Not to mention if you get stuck you can more easily become unstuck with this kind of ability. Another reason used for this is probably to allow players to manipulate objects which are clearly too heavy for them to physically lift allowing them to battle with massive metal containers rather than with just small.
As far as your comments go about ageias engine + hardware simply being an accelerator this is not true, it is true in the fact that it can process the phsyics interactions faster and therefore do more of the same. However it can do things that current engines and hardware are simply incapable of doing in real time, such as liquid physics, cloth physics which can tear and break.
Yeah sure these look cool but you only have to use your imagination as to how you can now make a much broader gameplay for players, you won't be able to imagine even half the clever tactics people will be developing in game with this stuff, as always the imagination of millions of gamers always outweighs that of the developers and before long we will all be doing stuff with the physcs systems which just weren't planned by the developers but are possible to the interactive nature of the engine.
Cell factor looks cool to me, I can't wait to grab some of those barrels of combustable liquid and run off into a indoors area and puncture the barrel and then carry it around dropping flamable liquid everywhere, then ignighting it when enemys run in.
Current games like GRAW don't make particuarly good use of the physics systems at the moment and that is surely expected as it's a new technology and needs to be slowly introduced. No game producer is going to be putting physics into games today which require the use of the PPU add on card because they decrease their target audience, so much so they would almost certainly make no profit.
Give it a few years and the price drop of the PPU hardware, as well as it being sold in prebuilt gaming machines and possibly sold in packages with other games, when the market accepts them as generally used hardware we'll see games using them for interactive objects more and more.
For now expect most of the uses of the PPU to be for effects which do no alter gameplay, but simply make things look more pretty.
Hope this helps clear up some of the misunderstandings around here, please correct me if im inaccurate anywhere as I am still learning about this technology myself.
Act
"The process of doing or performing something: the act of thinking."
Interact
"To act on each other: More than a dozen variable factors could interact, with their permutations running into the thousands (Tom Clancy)."
Both these taken from dictionary.com and funny enough one of the quotes being from Tom Clancy who if I'm not mistaken had a hand in the making of the Rainbow 6 games and GRAW.
Essentially what we're saying here is that interaction is allowing objects to manipulate each other such that the manipulation can alter the outcome of one or more of the objects interacting. In this sense the Ageia physics engine and dedicated PPU aid in setting up a virtual environment where by users can interact with objects within the world and those objects can then go on to interact with other objects.
The method of doing so is an aproximation of physics in the real world, but do not be mistaken this does not mean it is scripted in any way, it simply means the movement of the object in the virtual world will not exactly match the movement of one in the real world. Even with current physics technology (and by that I mean everything before ageia) does a pretty good job of allowing objects to interact with each other, they just cannot do it on a very large scale.
Some of you (magoo mainly) are appearing to others to know very little about the subject when in fact you may have a very good grip of real physics (I did an A level in physics at college) the problem is you do not understand the implimentation of physics in a game engine.
Magoo you state in your 2nd post:
You can absolutly do that with the current physics systems and since the ageia physics system can do everything the current generation can do, plus more that implies you can do that with ageias physics system as well.
The problem you seem to have is distinguishing the physics system from the game code or general game engine. The Ageia physics engine simply calculates the physical interaction of objects either on their own or together as one object effects another. The game code is what allows the gamer designers to allow the players to interact with this system. The very basic level of which is simply moving an avatar around the virtual world and simply knocking into these objects, but there is no reason why that avatar cannot pick up an object and throw it, that just has to be coded as an ability into the game and require the correct user input to do so.
The limitation here quickly becomes the engie coders ability to design new ways for the players to interact with the objects in the world, sure picking them up is easy, throwing them is also fairly easy. But the more you want your player to do the more you have to code into the engine and that becomes a bottleneck in production.
Because a player can pickup an object and throw it with telekenetic powers but not pick up the same object with his hands doesn't mean the system cannot manage this type of interaction, it simply means that the game developers have not coded this particular interaction between the players and the physics enabled objects in the world.
In the case of cell factor it is most likely due to huge number of objects the player will need some way of manipulating, sure you could pickup a crate and throw it at an enemy, or you could fling 1000's of crates at him with a cool effect causing crates to fly everywhere. Not to mention if you get stuck you can more easily become unstuck with this kind of ability. Another reason used for this is probably to allow players to manipulate objects which are clearly too heavy for them to physically lift allowing them to battle with massive metal containers rather than with just small.
As far as your comments go about ageias engine + hardware simply being an accelerator this is not true, it is true in the fact that it can process the phsyics interactions faster and therefore do more of the same. However it can do things that current engines and hardware are simply incapable of doing in real time, such as liquid physics, cloth physics which can tear and break.
Yeah sure these look cool but you only have to use your imagination as to how you can now make a much broader gameplay for players, you won't be able to imagine even half the clever tactics people will be developing in game with this stuff, as always the imagination of millions of gamers always outweighs that of the developers and before long we will all be doing stuff with the physcs systems which just weren't planned by the developers but are possible to the interactive nature of the engine.
Cell factor looks cool to me, I can't wait to grab some of those barrels of combustable liquid and run off into a indoors area and puncture the barrel and then carry it around dropping flamable liquid everywhere, then ignighting it when enemys run in.
Current games like GRAW don't make particuarly good use of the physics systems at the moment and that is surely expected as it's a new technology and needs to be slowly introduced. No game producer is going to be putting physics into games today which require the use of the PPU add on card because they decrease their target audience, so much so they would almost certainly make no profit.
Give it a few years and the price drop of the PPU hardware, as well as it being sold in prebuilt gaming machines and possibly sold in packages with other games, when the market accepts them as generally used hardware we'll see games using them for interactive objects more and more.
For now expect most of the uses of the PPU to be for effects which do no alter gameplay, but simply make things look more pretty.
Hope this helps clear up some of the misunderstandings around here, please correct me if im inaccurate anywhere as I am still learning about this technology myself.
All Ageia does is simulate gravity, friction etc, all physics in a semi-real world enviornment, it's up the devlopers to find uses for it.
Alot of people can be annoyingly short sighted. Just because you havn't seen a game where a character picks up a chair in throws it, doesn't mean it won't be done.
Alot of people can be annoyingly short sighted. Just because you havn't seen a game where a character picks up a chair in throws it, doesn't mean it won't be done.