When searching to see progress on network multiplayer development, I saw it written on the FAQ that "[it] may be impossible or highly difficult to implement properly, given our physics engine complexity." I would normally ask exactly why before posing anything like a solution but I fear expecting a response from a Dev on the subject might be a little hopeful because... well ... who am I? With that said, I would like to pose a few suggestions to problems I can conceive of in an attempt at making an engine like this, work in multiplayer. My hope is that, short of me being helpful in anyway, a discussion like this might be useful for giving inspiration for potential solutions. and so... Say we have Client 1 and Client 2 connected to a game session. Wouldn't the only requirement for Client 1 to appear believable for Client 2, and vice versa, be that they send periodic updates of position, Orientation and velocity, like in any other game? This could at least give us "Ghost" of other players to play around with. I get that the physics engine does a lot of work during collisions and having updates for all of the events would likely swamp the network... but If we wanted to have collisions be believable also, say Client 1 crashes into Client 2, would it not be a case of the engine on respective client's sessions calculating the forces in a collision against their respective "Avatars" (perhaps there is a better term for other players on your client) which are necessarily less complex to simulate physically, and sending the information to be applied in the other client's engine? Using less complexity in the "Avatars" lets us have more players in a session and if the state of the "Avatar" (i.e. structural integrity, shape,) is updated from the server then there is no load on your client, barring the network traffic. I can see a situation where the physical state of a Client's car vs it's avatar would differ. In that case the Client's state would take precedence and and update would be sent. I'm not entirely sure what TCP and UDP are capable of, nor the size of packets that would be needed to transfer the relevant information but if need be, I'm sure we could deal with a little less fidelity to compensate for less frequent updating of force transference in a collision. If it means we can crash into our friends i'm sure most would understand. That being said is sending a collision update with relevant forces and "Avatar" state changes in high volume for short bursts really an issue? Isn't this what you might see in a game like, IRacing for example? Your thoughts would be appreciate. zorilya
The physics engine doesn't only do work during a collision, it is always calculating every single node of the vehicle at 2 kHz. There is no such thing as 'position, orientation and velocity', only position and velocity of each node of the vehicle. Currently there is no way to reduce the fidelity of an object, as it is a sum of it's parts. I see no way of online multiplayer being possible, and noone is asking for it. What we want is either local split-screen multiplayer or LAN multiplayer.
Yeah, search up multiplayer in the search tab. Probably 5-6 pages of multiplayer requests on there (I think idk)
your average home local network can barely manage to complete two pings between computers in under a millisecond let alone send any actual information in the microseconds required to do physics calculations between vehicles as it currently stands your dream is FLAWED
You are both right. But its all up to the dev team. And going from LAN multiplayer to full online isn't as big of a jump as you think.
nonononononono LAN multiplayer is a pipe dream online multiplayer is impossible unless we suddenly, magically, get ansible technology
I think that is a given. Each and every client on a network doesn't need to know what's going on with all individual nodes of other vehicles owned by other clients, just the general orientation of the object as a whole. Think bounding box. On your PC your driving around in a car with nodes that are being updated at 2kHz. Everyone else sees you as something which looks like your car but doesn't have the same CPU overhead, say... a cardboard cut out of your car. Still with the nodes intact to take information about deformations, but without the need to simulate forces on them, since that is done on the owner's machine. Just like other online multiplayer games, your PC would be telling the other clients, via the server updates, where your car is, which way it's facing, and (if you extend the same periodic updates to the nodes) what shape it is. Of course shape updates could run a lot less frequently than 2kHz and still be believable meaning an online multiplayer experience would more likely be achievable. ... hence the want for discussion on the topic. Are you suggesting that because you don't see a solution that there is no solution? It honestly seems that way with the kind of language you choose... your whole post seems aimed, less as a furthering of conversation on the topic, and more towards shutting down conversation on the topic... Hardly productive. I'd be careful who you claim to represent with your speech... "No one" means just that... Even just me talking about it here makes your statement false, and there are many others. Perhaps you don't want it but that says nothing about it's feasibility or other's desires for it. --- Post updated --- Just reading through this thread, I'd like to remind people that I intended this as discussion about potential solutions... It seems that some people are more focused on feeling good about telling people they are wrong. I would ask that you leave your ego where ever you enjoy visiting so you can retrieve it after you have contributed to the conversation in a meaningful and productive way. In the spirit of this forum, might I suggest a consideration for guiding your comments. If you see a problem, ask a question. In asking you will likely trigger people to think about the solution further, rather than attacking you. This can only be a good thing because at least one person will come out the other end with more understanding. I say "likely" because there are those who react to questions against their ideas in a very negative way and to those whom i'm referring, I would say you need to distance yourself from your ideas. Stop leaning on them, so to speak. It is my belief that you can't effectively look for a solution if you depend on (lean on) part of the problem, and lets face it... everyone can be wrong so sticking to a wrong idea because of your ego just slows the process. Your ideas don't shape you nearly as much as your work, so make your work positive rather than negative.
Like other people already said. A beamNG car has velocity and position information on every single node in the system. So in that sense each node is almost as complex as a whole player in a game with simpler physics. Which means a single car would be as server intensive as a full 50+ player server in a simpler game. The nature of BeamNG doesn't allow for any easy multiplayer intergration sadly.
Thing is if you use search, there's developer statements on it, 100 threads on it, not one of your ideas is new, and most of them are themselves flawed at root
Only way online multiplayer would happen is if they heavily optimized the game and still keeping the physics, graphics, and other features and little details we like, THAT would be very hard... at least in my opinion...
Could something like Teamviewer work to implement a multiplayer system? It seems to work just fine over Teamviewer to login to the computer remotely and then use the game. You should be able to at least have two players without a problem.
