Game physics

[Pyronick 21]

When you are supposed to be able to take cover behind a body the position of legs and arms matters quite a lot.

Let's assume we are talking about a very complex ragdoll physics system which realistically portrays the anatomical restrictions of our bodies. (tendons, muscles and bones make sure we cannot fold into positions you often see in other games with ragdoll physics)

It can assume the correct position and very accurate pose only with a grid location of the pawns' fictional absolute center (grid location on the map), azimuth and the three dimensional Cartesian position and angles (XYZ and ɑβγ) of the wrists, head and combined feet relative to that absolute center.

Angles ɑβγ predict the pose of connected limbs by using anatomical restrictions of the human body and gravity.

The data of the pawns pose can be synced without using much data. Legs have a small angle of freedom, so you can see both feet as one position, ragdolls doing splits aren't realistic anyway and you want to avoid them so this is the best solution. The head is synced in the game already afaik.

That only leaves the wrists. A human wrist restricts movement of the entire arm, there is only a little bit of freedom. So the XYZ + ɑβγ of both wrists, combined with that of the head and the gravity gives you a very accurate pose of the upper body and already a pretty accurate pose of the lower body.

Here's an example:

A pawn is placed at position XYZ, azimuth 45° off magnetic north. His left wrist is a relative XYZ away with alpha, beta and gamma as direction. His right wrist is a relative XYZ away from absolute center with ɑβγ relative to absolute center. His left wrist is a relative XYZ away from absolute center with ɑβγ relative to absolute center. The camera raytracer gives the head's position and direction. The combined positions and directions of both wrists and the head and gravity create a very limited amount of possible poses so you now have a fairly accurate pose of the upper body.

Edited April 16, 2009 by SgtH3nry3

[oneoff 0]

Very valid arguments from all sides , but maybe i can give you a few parameters to complete your debate.

there wont be any ragdoll in Arma2 because there is not even enough cpu time left for community scripters to run per frame checks in the free boundaries they had in ofp and arma .

read

"in order to maintain smooth frame rate in real-time content, time limit for all scripts in each frame is enforced by the engine in ArmA 2. Generally speaking, in case of more demanding scripts, be prepared that their result may come way later and also there probably can suffer from significant latency. It is under evaluation if and how possibly allow user scripts to change how much time they may take from the CPU in every frame."

so there will have to be some massive dev in the next two months if we are to even have DAC ,mando missiles Spon_ etc in arma2 never mind space for ragdolls.

and on the other side if we negate the above statement , Ragdoll can be local , but you would have to forget any permanent identity of the body , so no taking cover behind it , unless your happy for bis to set a rule of where body falls it will be set in a pre determined final pos with geo added on all clients ,but that just wouldnt be right.

[SaBrE_UK 0]

Although an excellent system for death animations is welcome in future games, (whether it be ragdolls, inverse kinematics mixed with animation, Euphoria, whatever) it's a little far-fetched that it will make it into for ARMA2.

For now I'd either like to see a few death animations for each stance, so not as many dead bodies in the same position, or if that's asking too much, just animations that flow better: Remember the death-while-running animation in OFP- that was very satisfying, and they didn't come to a dead halt to fall over like in ARMA.

[froggyluv 2130]

This maybe a stupid question, but is ragdoll necessary to ensure that gravity has an effect on the soldier? That's really all I care about as I just hate seeing dead soldiers laying rock-hard straight off the edge of a watch tower,

[PuFu 4591]

"in order to maintain smooth frame rate in real-time content, time limit for all scripts in each frame is enforced by the engine in ArmA 2. Generally speaking, in case of more demanding scripts, be prepared that their result may come way later and also there probably can suffer from significant latency. It is under evaluation if and how possibly allow user scripts to change how much time they may take from the CPU in every frame."

can you provide the source of the above statement please?

[xeno 230]

Added by Maruk last saturday to Biki:

http://community.bistudio.com/wiki/ArmA:_Editing#Forward_Compatibility

Xeno

[walker 0]

Hi all

As Xeno points out when you develop you are dealing with constrained resources, there is a cost for every event in a program whether it be CPU time, memory, hard disk space, virtual memory, graphic memory, shaders, bandwidth etc.

Even if computers were capable of doing the calculations for complex ragdoll needed for ArmA in real time for SP, we want the soldiers hand gripping the rifle and their finger on the trigger and we do not want the robo movement of limited ragdoll that is all other games so far do. We also do not want to go back to the OFP days of clipping. And we want large entity counts.

And the half assed robo looking ragdol we get in games now is just not as good as motion tracked animations.

The key constraint on Ragdoll is transmission time on the net. That particular bottle neck is not going away soon.

Kind regards walker

Edited April 17, 2009 by walker

[Scrub 0]

Hmm. I'm actually excited by the per frame limitation. That could very well mean performance noticable to the user/multiplayer is paramount, and there must be so many details crammed in (multiprocessor requirement noted) that resources are closely monitored? I'm liking this more and more - until I try to make my typical bloatware ubernoob doall script. In the future, a feature to dedicate a core to scripts/functions would also be nice, as core counts are increasing at a fair pace.

[Zerst0ren 0]

Can't we just implement the physics engine to work with dead bodies which have already completed their death animation?

Btw, have they fixed the utter horrible vehicle physics? I mean every car goes up to 120 km/h in 5 seconds and holds it steady there, even a fucking bus. Then when you run into a bush (example) you slow down to 0 km/h instantly, and then speed up again.

It's like having m = 1 in basic equations like E = mv^2/2

Edited April 17, 2009 by Zerst0ren

[PhilippRauch 0]

Why are those ppl so keen on Deathanimations?? I dont care if someone flips around like a ragdoll just to satisfy my morbid dreams...

Several mocapped (get it? ;) ) death animations, randomly choosen are far more than we, sorry I, really need...

See some real snuff movies from warzones around the world and you will see they aint that pretty or varied anyway... ppl mostly just drop dead, thats it! Or get torn into pieces, anyway not much fun, so why care about a technology that is just not realistic...

I would rather have more fun with better collision detection and more animations inside/on/with vehicles/weapons/animals/things!

Thats my take on this ragdoll juice, although i like the math, i.e. how quickly something sums up to big values. :)

PS:

I second that we should/could have a CPU core for each major part, i.e. one for AI, one for physics, one for scripts etc... could be nice...

Edited April 17, 2009 by PhilippRauch

[S!fkaIaC 10]

My personal vote:

- fix collision detection of objects in general NOW (means correct acceleration considering mass and, if feasible, internal structure [object inside object e.g. for vehicles])

- add in 2035 ragdoll when every kid owns a supercomp :D , ArmA-54 MUST have ragdoll

- death animations? dead is dead, even an animation during that a dying unit makes a salute while dying is ok for me.

Edited April 17, 2009 by S!fkaIaC

[PhilippRauch 0]

My personal vote:

- fix collision detection of objects in general NOW (means correct acceleration considering mass and, if feasible, internal structure [object inside object e.g. for vehicles])

- add in 2035 ragdoll when every kid owns a supercomp :D , ArmA-54 MUST have ragdoll

- death animations? dead is dead, even an animation during that a dying unit makes a salute while dying is ok for me.

hehehe, yeah and then when on the floor a flagpole pops out of the corpse with the clantag/side flag of the one who killed it... (Like in that marsians movies scene with the president) :D

[Nigelwow 10]

you need to have proper physics...

driving from a mountain isnt the same as the road...

[Zerst0ren 0]

My personal vote:

- fix collision detection of objects in general NOW (means correct acceleration considering mass and, if feasible, internal structure [object inside object e.g. for vehicles])

- add in 2035 ragdoll when every kid owns a supercomp :D , ArmA-54 MUST have ragdoll

- death animations? dead is dead, even an animation during that a dying unit makes a salute while dying is ok for me.

How in the name of science are you able to think that collision detection has anything what so ever to do with acceleration and mass physics?

[mr.g-c 6]

Added by Maruk last saturday to Biki:

http://community.bistudio.com/wiki/ArmA:_Editing#Forward_Compatibility

Xeno

Is that good or bad? Sounded rather bad to me....

Why can't it be programmed that when you have more than 2 cores (ie between 3-8 threads/cores) that the scripting engine gets its own, or must only share it with sound-system, whatever....

[11aTony 0]

Maybe ArmA dont need ragdoll but you cant say that it is totally not realistic either. It is most realistic of what we know. Look at GTA4. When you hit a person with a car it looks very realistic. Same goes for car-car crashes. Remember what happens in arma when you hit somebody or crash in another car?

[dunedain 48]

Red Orchestra has also realistic Ragdoll and collision between vehicles, that's the most realistic i've seen in a 5 years old engine, i'm sure they can do much better now, it seems that masses idea of ragdoll is BF2 ... -_-'

Over here it's like people love to watch bodies and ammo shells levitate, go through walls .. rigid death animations and those perfect collisions we all know which didn't change a lot since i bought Ofp in summer 2001 ...

Is that taboo to ask up-to-date features ? i mean we don't want ragdoll to satisfy our "morbid dreams" judging my every morning h*°d on, that's not what i'm dreaming of personally, i won't repeat all advantages last 5 pages are full of..

Edited April 18, 2009 by dunedain

[NeMeSiS 11]

Red Orchestra has also realistic Ragdoll and collision between vehicles, and that's the most realistic i've seen in a 5 years old engine, i'm they can do much better now, it seems that masses idea of ragdoll is BF2 ... -_-'

over here it's like people love to watch bodies and ammo shells levitate, go through walls .. rigid death animations and those perfect collision we all know which didn't change a lot since i bought Ofp in summer 2001 ...

Is that taboo to ask up-to-date features ? i mean we don't want ragdoll to satisfy our "morbid dreams" judging my every morning h*°d on, that's not what i'm dreaming of personally..

If BI manages to implement it in a nice way (Like GTA4, most amazing stuff i have seen in a game) without affecting the gameplay in *any* negative way then of course it would be appriciated. However most of us are afraid that in order for such features to excist the unitcount or other things must be significantly lower, or other changes must be made, and IMO (And apparantly alot of other people are with me on this), that is just not worth the amount of eyecandy gained.

EDIT: A solution that would work for me as well is just to play the animation, then when it has finished activate ragdolls for a few seconds so units dont stick out of the ground/trough walls anymore, and then disable ragdolls again. It would give some weird effects (Units popping out of walls if they fell trough because of the animation), but it would be better then nothing and shouldnt be very hard on the CPU. (Unless you have dozens of AI dying withing seconds of eachother. :p )

Edited April 18, 2009 by NeMeSiS

[metalchris 0]

If there are still performance problems with real physics- engines like havoc, why not make it an optionable feature for future systems?It seems possible in VBS2 with PhysX. I mean , there is the possibility to have 10km viewdistance in full details , but the computer has yet to be built to run on these settings.

And physics are not just eyecandy , like a well implemented IR/NVG shader they add greatly to the immersion into the game (which is a part of beeing realistic isn't it).

Edited April 18, 2009 by metalchris

[Kristian 47]

ArmA 2 is basically same as ArmA 1. I believe that ragdoll wont be added. It isnt that simple: "this addragdoll" :D

[dunedain 48]

Yes, obviously that can't be achieved for Arma2, we know that. But they could release a major expansion, like Resistance was for Ofp.

Edited April 18, 2009 by dunedain

[sic-disaster 311]

A different matter of physics: how about hadding proper physics to grenades?

A great new feature would be the ability to roll the grenade across the floor of a building, but it would require good physics of the grenade to actually be usefull.

Right now, grenades inside buildings are totally useless as the only thing you will blow up with the thing is yourself.

We need an underhand roll and good grenade physics.

[S!fkaIaC 10]

Rolling and rebouncing.

In ArmA they sometimes glue at the wall instead of rebounce.

[PhilippRauch 0]

Yeah, somehow too bad they got rid of PhysiX support in ArmA2, especially since you dont need a specialized card for it anymore...

[dm 9]

Yeah, somehow too bad they got rid of PhysiX support in ArmA2

Was there anything to suggest it was ever IN ArmA 2 to begin with?