PhysX

[max power 21]

I doubt they would change that. It seems like it would be much better for multiplayer if they only had to transmit whether the tree was up or down rather than sync its position and orientation. It seems like it would be a better idea if there's a possibility that many trees or other objects can be damaged or destroyed, that transmitting one bit per object instead of position / orientation / simulation state would help players JIP long standing games of CIT, etc.

[50.cal 10]

Ok then again about collision detection between human or NPC and vehicles, do you have any idea how it could look and work combined with damage and so on.

For example let's assume you're "pushed" by a car a very low speed will this make you slide just like it is know or is it more likely to make you fall like with euphoria physics?

Do you have any examples that demonstrate how it could look like in action? Not the euphoria physics that we all know but with "physx".

[james2464 176]

I doubt they would change that. It seems like it would be much better for multiplayer if they only had to transmit whether the tree was up or down rather than sync its position and orientation. It seems like it would be a better idea if there's a possibility that many trees or other objects can be damaged or destroyed, that transmitting one bit per object instead of position / orientation / simulation state would help players JIP long standing games of CIT, etc.

If the collision of explosions are synced then why worry about the position and orientation of the tree? wouldn't the simulation be the same locally for everyone if the explosion or collision is synced?

[.Taffy 10]

I doubt they would change that. It seems like it would be much better for multiplayer if they only had to transmit whether the tree was up or down rather than sync its position and orientation. It seems like it would be a better idea if there's a possibility that many trees or other objects can be damaged or destroyed, that transmitting one bit per object instead of position / orientation / simulation state would help players JIP long standing games of CIT, etc.

But what if one player is using the fallen treee as cover, meanwhile the tree is in a totally different position for the person shooting at him?

[PuFu 4600]

that is bound to happen if this is handled by physx since syncing position and orientation for every object in the game world would be a total lag-fest

[max power 21]

If the collision of explosions are synced then why worry about the position and orientation of the tree? wouldn't the simulation be the same locally for everyone if the explosion or collision is synced?

From what I understand, there is a certain randomness to physics calcs and so it is unlikely that the same thing happens on all clients.

But what if one player is using the fallen treee as cover, meanwhile the tree is in a totally different position for the person shooting at him?

That is why they use the current system, with scripted trees. It is always 100 percent the same for everybody, and all they need to is transmit whether the tree is up or down.

[ziiip 1]

It would be a shame if they didnt change the physics for knocking over thees at least in singleplayer. I'm tired of them falling through my tank.

Edited April 16, 2012 by ziiip

[dmarkwick 261]

It would be a shame if they didnt change the physics for knocking over thees at least in singleplayer. I'm tired of them falling through my tank.

This is probably why there is no collision object for fallen trees. I agree with you, I think the extra small bit of information that represents the direction of fall could be transmitted, MP data speeds have increased enough to accommodate this I'm sure :) the only issue might be that the trees are animated to fall in one direction, so that system would need a rewrite.

[Minoza 11]

Why not give it a try with interpolation?

[purepassion 22]

I think the extra small bit of information that represents the direction of fall could be transmitted

They actually said this will be the case :)

[maddogx 13]

Using something like 4D simplex noise even that would not be necessary (transmitting the individual direction for each tree, I mean). They could simply use the 3D position of the tree (which is the same everywhere) + a random seed number generated by the server at the start of the mission (and synched to all clients), and the trees would fall in the same direction on each client. This way only a single number would need to be transmitted at the beginning (the seed), making it more efficient in the long run.

Also, thanks to the random seed value, the trees would still fall in a different direction on every game (i.e. new seed, new direction). BIS is already using a similar pseudo-random noise method to generate the infinite terrain in Arma2, so perhaps this is even the method they are using. :D

[old_painless 182]

Hope they sort this sync out, because trees are one thing. Collapsed buildings from arty or tank rounds will make some nice ruins to fight in. And they better be the same damaged brickwall that i am covering behind that what others are seeing.

[max power 21]

They do that now with incremental rubble models.

[old_painless 182]

I was thinking if the collapsing is now run through the PhysX engine the outcome could be different than todays rubble.

[max power 21]

Strictly speaking, I think if something obliterated part of a wall, the pieces and crap could be animated by physx and not synched, because the exact way the wall falls apart is not necessarily important if the pieces are not large enough to affect the game and / or they are not permanent. I think that would still have to be a canned thing where set 'pre-destructable' pieces of the wall are being destroyed, and models replaced, though. I do not think we will see large parts of the game world destroyable using physx as a means to physically destroy objects dynamically, though. It will be interesting to see what they come up with, but I would not get your hopes up. As walker says quite a lot, it's an input / output problem, not really a technique problem. There is the possibility in ArmA to do quite a lot of destruction at once. The best bench pressing technique in the world is not going to help you bench press 20,000 kg, if you follow. Of course, these are just logical arguments, and I'm no programming expert.

[dmarkwick 261]

Using something like 4D simplex noise even that would not be necessary (transmitting the individual direction for each tree, I mean). They could simply use the 3D position of the tree (which is the same everywhere) + a random seed number generated by the server at the start of the mission (and synched to all clients), and the trees would fall in the same direction on each client. This way only a single number would need to be transmitted at the beginning (the seed), making it more efficient in the long run.

Also, thanks to the random seed value, the trees would still fall in a different direction on every game (i.e. new seed, new direction). BIS is already using a similar pseudo-random noise method to generate the infinite terrain in Arma2, so perhaps this is even the method they are using. :D

I think there is much to be said for a random seed system like this, but for trees specifically I'd still prefer a directional component upon destruction. But, I acknowledge that might simply be unnecessary for gameplay. What I defintately would like to see is the collision object component to be retained so that fallen trees can become a barrier. I'm always a little disappointed that even a copse of trees can be driven through with impunity once they've fallen :)

---------- Post added at 12:09 ---------- Previous post was at 11:57 ----------

Strictly speaking, I think if something obliterated part of a wall, the pieces and crap could be animated by physx and not synched, because the exact way the wall falls apart is not necessarily important if the pieces are not large enough to affect the game and / or they are not permanent. I think that would still have to be a canned thing where set 'pre-destructable' pieces of the wall are being destroyed, and models replaced, though. I do not think we will see large parts of the game world destroyable using physx as a means to physically destroy objects dynamically, though. It will be interesting to see what they come up with, but I would not get your hopes up. As walker says quite a lot, it's an input / output problem, not really a technique problem. There is the possibility in ArmA to do quite a lot of destruction at once. The best bench pressing technique in the world is not going to help you bench press 20,000 kg, if you follow. Of course, these are just logical arguments, and I'm no programming expert.

Yep, we should regard PhysX as an essentially SP/clientside eye candy application, and hope that BIS can wrangle a robust MP method for handling objects that have a large enough volume to have an MP impact, such as vehicles (towing/rolling etc) and large destroyed objects. As long as MP-sensitive objects have a robust sync system then it should all be good. Bodies, for example, should have their torsos synced as far as position goes, but clientside positioning of limbs etc can be done locally. I think the same general principal can be applied to building parts, large walls can be synced while small rubble be handled locally. (However I realise that building destruction effects will most likely be similar to previous ArmA games.)

[b00ce 160]

Yep, we should regard PhysX as an essentially SP/clientside eye candy application, and hope that BIS can wrangle a robust MP method for handling objects that have a large enough volume to have an MP impact, such as vehicles (towing/rolling etc) and large destroyed objects. As long as MP-sensitive objects have a robust sync system then it should all be good. Bodies, for example, should have their torsos synced as far as position goes, but clientside positioning of limbs etc can be done locally. I think the same general principal can be applied to building parts, large walls can be synced while small rubble be handled locally. (However I realise that building destruction effects will most likely be similar to previous ArmA games.)

I would be sorely disappointed if ArmA3 didn't have PhysX for vehicles/towing/ragdoll in Multi-player (I believe Ragdoll in MP has been confirmed anyway in one of the DEV blogs.) because PhysX is one of the major selling points of it.

[BobcatBob 10]

So I forgot, does the inclusion of physx mean it will be possible for modders to allow the player walk around in moving vehicles even if BI do not allow us to in Vanilla, because if such a basic (yet really cool) feature had been implimented I think it would've been spelled out by-now?

[neokika 62]

So I forgot, does the inclusion of physx mean it will be possible for modders to allow the player walk around in moving vehicles even if BI do not allow us to in Vanilla, because if such a basic (yet really cool) feature had been implimented I think it would've been spelled out by-now?

Hey,

Don't get me wrong, but what does that have to do with PhysX?

[Maio 293]

So I forgot, does the inclusion of physx mean it will be possible for modders to allow the player walk around in moving vehicles even if BI do not allow us to in Vanilla, because if such a basic (yet really cool) feature had been implimented I think it would've been spelled out by-now?

Hey,

Don't get me wrong, but what does that have to do with PhysX?

Maybe he is thinking now that the player entity will have actual weight, he will make contact with the vehicle and not just slide out of the cargo bay.

[dmarkwick 261]

So I forgot, does the inclusion of physx mean it will be possible for modders to allow the player walk around in moving vehicles even if BI do not allow us to in Vanilla, because if such a basic (yet really cool) feature had been implimented I think it would've been spelled out by-now?

I think PhysX will be an inefficient way to handle on-vehicle mobility, particularly when you imagine the lag involved during MP games. IMO the better way to handle such a feature would be to juggle the world context around a bit - make the vehicle a static "world" while animating the external world around it. I think this kind of happens already which is why you get the outside view jumping about when you're a MP passenger. But if this notion could be extended to allowing people to wander around a vehicle surface it might be a nice feature. However there's a sort of grey area with this idea when you think about large boats, or any open vehicle, what constitutes the player's local "world". It might simply create more issues than it would ultimately solve.

[Choki 10]

It would be a shame if they didnt change the physics for knocking over thees at least in singleplayer. I'm tired of them falling through my tank.

Yes I agree that urgently needs to be changed

[-=Grunt=- 10]

What if... ArmA 3 has this kind of destruction?

http://www.youtube.com/watch?v=rlVQ4EeeGkE&feature=player_detailpage#t=87s

[dmarkwick 261]

What if... ArmA 3 has this kind of destruction?

http://www.youtube.com/watch?v=rlVQ4EeeGkE&feature=player_detailpage#t=87s

Impressive... and desirable... but single-player only :)

[kimozabbi 1]

What if... ArmA 3 has this kind of destruction?

http://www.youtube.com/watch?v=rlVQ4EeeGkE&feature=player_detailpage#t=87s

You would need a PC powered by a nuclear reactor.