Hello BeamNg, A great find for me to discover this game, climbing a hill in first gear, it hooked me. I hope the game can be finished. I want to share some thoughts to improve the experience... - TrackIR - when I have climbed that hill I want to check the view out - Improve steering feeling - feels very light atm. - Traffic - Open world, I want to drive all day from A to B (and not crashing). ---- [dream game] use real maps*, use real traffic data*: then procedurally generate the decorations/environment. - Car radio - linked to Spotify? I really hope you succeed. Bringing the crash into a car game, is a winner for me. Regards, Bob Hoskiins * if this appeals I may be able to help.
Radio would be a nice touch. For me though a real world map populated with real world traffic, would be a wonderful experience. I could then listen to radio in the traffic jam And speeding in busy traffic thrilling.
Minus the fact that a) its a PC, you can just run a media player in the background and b) no computer can cope with running enough simultaneous vehicles to achieve anything other than extremely light traffic. TrackIR and rift I want to see. We did briefly have rift, but it was problematic. If trackIR can be bound to a joystick axis in software somehow then you could use it already. Traffic, open world, radio, suggested before, although a and b cover 2 of those already. Real maps, if not for the map size limitations is technically possible.
- - - Updated - - - Ok, let's think about technical limitations.... procedural environment, based on current location and map data. This allows for worlds to be created, not a limitation. Just some algorithms to keep the environment not too static.... traffic, well look at current games, gta v, gives the impression of traffic. Project cars, 35 cars on one track.... the traffic has to be local to player, not the whole map. there are ways.... ultimately, the open world goal (in any game) tries to reproduce the real world (as far a technically possible and gameplay allows). i want a busy large map, with beamNG physics. please.
But few computers can cope with as much as 4 or 5 cars. even 4 or 5 cars local to player would be super sparse traffic.
couldn't the cars be not loading physics, but then when you get close the physics on the cars around you spring to life, and when they are far away they switch off their physics
How is it possible for people to be so oblivious. You can easily have more than 5 cars moving in the very near vicinity of you during normal or even light traffic.
NASA Computer = ITs possible with more than 10 cars, i can barely do 5 cars with I5 and AMD 7700 HD lol. (Not even stable fps then..)
if ya got $15000 you can get like 36 cars https://www.youtube.com/watch?v=25FfiS9JWD8 but i doubt anyone would spend 15k just to play beamng
6677 has a point but I think that the core of that Idea is feasible (if slightly over the horizon). I'd say that rather than having the traffic be dictated by the physics engine, have it be non deforming objects locked the a 'track' up until the moment that you hit them. Obviously this could still cripple a computer if you took a t75 or something and smacked every car you saw, but it would be a step forward. As for traffic density I think that it wouldn't be a huge deal with our current maps, as they are all pretty rural.
Ok we would like the same thing. And, can agree a clever resource utilization is required. some smoke and mirrors. also, 1 thread per vehicle seems a tad restrictive especially for the non visible traffic. can we agreed though 'open world' real maps with traffic is a great goal?
Some CPUs don't have the single-threaded performance to run one vehicle per core - now you're suggesting adding more vehicles per core? The game already does this when you run out of cores, but expect the framerate to plummet.
We can agree that is a great goal, but that doesn't make it any more feasible. As for 'real maps,' I've given those a go a few times: French Quarter, New Orleans, LA LaSalle Street Canyon, Chicago, IL As you can see, I never got very far. And aside from a select group of real-world racetracks, there's B25 Mitch's Riverside Expressway, which is based on a real-world freeway in Brisbane.
I'm working on a city map, and it is going to be beta released in hopefully a week. Check the "city" thread at the bottom of this post.
I am suggesting: cars outside the players locale do not need a entire thread. Multiple cars can be handled, x and y and z and speed and planned route, this can be handled for multiple off screen cars, every (I guess for example) 0.05 milliseconds. 0.00 car A is here 0.05- car B is here 0.10 - car A is now here (is this possible?) then when a AI car is candidate for collision and in player view, car is promoted to its own thread for drawing and collision. .... If a traffic jam is not possible, occassional traffic on country road is good.
You are right about the cars, but traffic cars do not need so much realism to them, they need just to respect traffic rules and have a proper crash physics, for example they're interiors don;t have to be fully working, they won't need other things to them, hell, they could be even car shaped boxes on wheels that are propelled by a thruster in a way they respect the laws, and they deform if you hit them as far as the physics care. So they (the traffic cars) can be a simplified version of the car you're driving, so the pc won't have to struggle that much.
I hate to be blunt, but nothing in this comment suggests that you understand what causes the physics to be so taxing. Interiors don't take up much processing; they aren't even physically simulated in the current vehicles. You could remove the interiors altogether and it wouldn't help almost at all, especially for the CPU. Powered by thrusters instead of an engine? The engine and transmission make up less than 2% of a JBeam. The 'deforming at all in a way that is impressive or feasible' part is the one that is difficult to simulate. You may be able to make traffic vehicles less populous with nodes and beams in order to save on some overhead, but: A. That would mean creating all and fine-tuning new JBeams for each vehicle (not a quick process), and B. They wouldn't deform as well, voiding the very purpose of using Beam physics to begin with.
