'Developer Section' Response Thread

Really, I didn't know that, better check it out then. Maybe it could be more smoother by default ?

 

They could easily do this - very easily - just as other games do it.
Normally, vehicles are processed at a mind-boggling-but-awesome 2000hz (2000 times a second!). This compares very favorably to GTA V's 20hz, which is why vehicles there glitch out quite a bit especially when you have a vehicle pinched by a train between some immovable things like a tunnel entrance (vs a long truck or tanker, heehee, admit it, you did it!). While this is totally awesome, processor technology isn't anywhere even close enough to run this in real time. It would be one frame every minute or two! It's 100-fold the increase, and that does nothing to account for things like pedestrians etc. I don't ever expect peds in here as vehicle to pedestrian action would violate the ratings on this and would stray from the mission statement. That being said, on with what I started saying: (m = meters here - 32inches / 2.66ft for US folk)
Say you have 5~8 vehicles in the immediate vicinity (25~40m). Lets run those at full speed! No, no, our 4790k, 5770C, or 6700k CPU would go 20fps there-abouts... okay, HALF speed until we're about to hit (one car length) THEN go the full speed, okay, 35fps. That works.
What about the rest of the cars? Lets run them at 100hz, 5% processor use of normal, since they're further away, they might get stuck in something with low-hz physics (we can check for this per-vehicle in the code triggered by the vehicle not moving after X amount of tics, so it doesn't negatively affect performance). This doesn't detract from our realism here, because we're not near them, this is how other games do it. When we get near enough to make a huge pileup, the hz-rate will increase, and we will still be able to make realistic accidents. So basically, even with the vehicles running at 5% our realistic simulation speed rate, it won't do much to our processing power even if there are a bunch of them, and if they hit while far away, even if they get lodged in one another (not likely at low speeds conducive/typical to in-city AI), code will check for this and spawn them back enough centimeters or a half-meter after not moving so long, unsticking them.
Of course, in a perfect environment, this 'falloff' for physics, would be adjustable (slider in options for this!) for your PC with the FPS target (could be affected by other things like ability to render the scene itself) or the distance (better, user-adjustable here).
There's little reason to want to render those vehicles on the other side of the map at full speed except for the very short time they crash off the road or hit another vehicle.

*most times, when I've said render, I mean rendering the physics, not the car model itself.

So yes, entirely, this game could 100% have a city full of AI, like many triple-A games like GTA or Saint's Row or Mafia series.

--Just my 2¢ --cheers!

 

Wouldn't that be able to be modified by ratios, or fractions, I mean, you wouldn't really have to run the vehicles suspension etc or engine physics (aside of temp) much at all this far away? I mean, there has to be a way to do this at engine-level, other games do it.

 

That would make it more complicated, you need to have a fully working AI system with cars that don't crash no matter what kind of setup they have, a realtime distance to player calculator with ratios for the fysics (ratios could be calculated) and something that makes the transaction smooth.
Quite hard I guess

 

That was extremely informative and more of an answer than I ever expected thanks! (and yes I of course iv'e tried that!) although Diamondback's reply as you saw states that this sadly isn't possible with the physics required here.
So wouldn't just a simple spawning/ despawning when out of range/ sight of randomized AI vehicles be possible to implement?

 

There's a lag associated with spawning or removing a vehicle, more-so on spawning. They'd have to be spawned by another thread, in a manner that wouldn't 'hold up' the physics in other words, causing the game to lock for 1~2 seconds when a vehicle is spawned (times however many vehicles are coming in, would cause a slide-show if not). You'd only have very sparse traffic at best as most consumer cpus can only handle max 4 vehicles and some not even that, whereas 4xxx series i7's can handle up to 7 vehicle instances before fps drops below 20fps periodically.

When the person above says it's not possible to run a vehicle below a certain hz without suspension etc going bonkers, you just wouldn't run various things far away, or run a generic, much simpler version of it in place of a real suspension (etc) until it gets close enough to let full-cycle physics take over the entire vehicle. The actual AI routines aren't a lot of work in comparison for the CPU vs the physics of the car itself, especially for 4xxx and newer i5~i7 cpus.

 

I think that only way to get traffic would be having traffic cars as simple objects instead of beam objects.

SLRR has some deformation and it certainly is lot simpler method used, I would guess that it is same as with less static world, some challenging problems needs to be solved.

 

Okay, I stand corrected, especially considering I haven't really looked at the code much. I am glad you are not 'other' games. Regardless of the fact, with CPU core counts increasing, and the cpu-wars one month exactly away from heating up again (AMD RyZen public availability) it may mean some more progress here. So at the very least, get your physics engine ready for some serious workload in the coming year.

 

I see a whole lot of you can't do this, you can't do that, that's not how things work. When all I think in response is that anything is possible and sometimes you have to change how things work instead of accepting it. While both are surely a factor, I believe it's more so a problem with the overwhelming amount of work it would take in order to implement some of the aforementioned features (given this games duration of development and current technologies), as opposed to technical limitations.
In short, I think when the Devs say that certain things are impossible, they don't necessarily mean said things are literally unachievable; but instead that the amount of resources it would take could be astronomical at this point in time and thus there's no justifying doing such things. It's much easier to tell people that it's not possible as opposed to giving them the entire spiel as to why it's not realistic for them to attempt. Then again, maybe it is literally impossible, what do I know really? I just like to think optimistically every now and again as life and recent events have a way of robbing me of optimism.

I understand where they want to go with this game and I think sometimes peoples expectations are just a bit too high. Which I understand, I too want that ultimate game that has a huge open world/universe like we've seen with the Dual Universe technology; where like thousands of players can interact in one world. Ideally that game would have the incredible immersive detail that GTA has, the simulation capabilities of BeamNG and KSP, sandbox capabilities like Space Engineers and Star Citizen, military simulation similar to Arma or Battlefield, roleplay comparable to what we see in the Sims (GTA applies here also), even more obscure stuff like Cities, Arc Survival, and Rust. Ideally this game would be VR compatible . I want you guys to just imagine that for a second, all of your favorite games rolled into one. You wouldn't even have multiple games, just this one near all encompassing game that you started up in order to play whatever you wanted. There doesn't only have to be this one game but wouldn't it just be amazing to have that level of interaction between players? Perhaps it's impracticable, but if Moore's law holds true then I for one would love to see something like this. This would of course require cooperation between game studios for the best results. So in regards to this, what I've been contemplating recently is what the team plans to do after BeamNG is complete. What happens after that, will they successfully license out this technology? Will they begin another game? Who's to say?

 

I know it can be done but its gonna be time consuming and difficult

 

Italy is improving. Wonder if the car will behave differently on the cobblestone than on the asphalt road.

 

If cobblestone is a mesh, then most definitely, if cobblestone is a texture then it can be just a grip difference, but if bump effects are added to textured cobblestone road, it can behave again more like if cobblestone would be a mesh, but perhaps with less performance impact.

I would say that in BeamNG, cobblestone will be more than just a texture.

 

Yeah, but so far I would say from the screenshot it looks more like just a texture.

 

I'm actually wondering if challenging part out of way in Italy means that CPU load from number of objects with shadows enabled is going to increase, because at the moment that is definitely holding back of how much stuff map can have or how big map can be. Would that mean that objects will have individual visual distance in a future?

 

I do hope the buildings will get more color in the final version; yellow brick gets old pretty fast. Perhaps they're looking into a way to add colored plaster into the material? A material mask feature?

 

If only the guardrails were deformable