so if I accept the stated limitation, the best we can hope for with beam ng is what we have now.....a tech demo, with no real future....a few levels, a few cars to pick. and maybe in 20 years time we can have 20 ai on the map, racing like current games can I would prefer to know what the development plan actually is. (Is there a plan somewhere?) And the procedural road environment generation using real map data, should be not too resource consuming..... ps: please provide an explanation of why the physics are too taxing, and let's get into the nitty gritty, I am not convinced by you dashboard reference ....my pc has 8 cpu cores (apparently)' 12 gig ram, x2 GPUs. (3gig)....in this case how are the resources shared for the environment, drawing, physics for 1 car, n cars?
Calling the Estama, we need a technical explanation here. I can't explain it in technical terms, but he's right about what causes the physics to be taxing.
BeamNG, like the Brigade realtime path tracer or Outerra planet renderer, is intended as future-proofed experimental tech, that can be converted and modified as needed and adapt to situations and projects that demand it. Unfortunately, this usually means that it won't work well on today's hardware. Like a space elevator, we know how it will work, but we don't have the technology to see it through entirely. As far as I'm aware, the development roadmap from this point forward includes: More content (vehicles, props, levels) More interactive levels More variety among scenarios A possible upgrade to DX11 The ability to state-save damaged vehicles (for later repair, upgrade, etc.) Basically, they're going for more of a "Test Drive:Eve of Destruction-" or "Street Legal"-type gameplay setup, where damage to your vehicle is more permanent (in career mode), and you grow attached to your car via upgrades, repairs, and personal touches. Look at the endgame as more of a digital garage than a digital, destroyable city. In all likelyhood, further updates and additions will follow v1.0, and the game may even be development for years to come. Assuming Moore's Law doesn't bottom out in that time, your suggestions may eventually become viable. One other thing that is likely to be in development is the ability to put nodes "to sleep." This is a common feature in most game engines. Rather than have all physical objects physically active all of the time, you can have some that are moving slowly enough or are far away enough from all players disable their physics temporarily to save on power. The plan is to use this to make the environments more interactive (having some CPU-light road signs, cones, or barriers set around the levels), but it could be adapted to make some form of traffic possible. If only very, very light traffic. More than you seem to think. No, it's not the most improbable thing on this list, but it's still not easy. For starters, 'procedural' is a bit of a buzzword. It tends to mean dry, self-similar environments that lack an artist's touch (which you don't know the importance of until it's not there). Resolution data is pretty low-res, with the 30 meter dataset only just now being released, and while road data is easily accessible, groundcover and building data is not. In short, it's possible, but the levels would be flat, monochrome,and empty. The other major limitation is that Torque 3D, the graphics engine that governs the visual aspects of .Drive, only allows standard-resolution terrain up to 8192x8192m, about five miles long on one edge. You can string multiple terrains together, but Beam physics can only register and use one terrain object at a time, so you'd be limited to twenty-five square miles. Fair enough. If you understand the basics of soft-body physics, skip this section, as I've misconstrued your concerns. In most games that simulate cars, there are five independent bodies at play: the car's body, and its four wheels. Deformation is handled by either having an artist pre-construct the damage that they anticipate occurring into the car model (Unreal Tournament, Halo, Just Cause, Forza), or by using a dynamic deformation system (GTA IV, V, L.A. Noire, Burnout Paradise), which arbitrarily pushes the vertices of a model around as it receives impacts based on intensity and direction. BeamNG actually treats vehicles as a sum of a very large number of independently simulated points (nodes), which are attached by way of a series of lines (beams), possessing their own strength and bounciness. Thus, you have high-fidelity simulation, but it costs far more than the simulation of a regular game car. In the order of hundreds or thousands of times. I can't speculate much on the technical side as where the processing goes, so take the following with a grain of salt, but anything to do with visuals, naturally, goes to your GPU(s). Each car goes to its own CPU core, until all cores are filled up, and then the load is shared by another core. Currently, simulating a single vehicle across multiple cores is not possible, which is why the T-Series is not greatly popular. People with many cores can simulate several vehicles, but if they aren't powerful enough individually, the lofty T-Series can choke one out, and the others in turn.
thank you Mr Razor, the type of response I hope for. Outerra is another application, that I hope will give me some joy in my life time . I also agree procedural - algorithmically (..is that word right?) generated content can be bland and the artist touch can bring the gamer closer to the world. E.g. Outerra has no caves, the slope rises on the dirt has the same pattern smeering. elite dangerous, the systems and politics are no different to the previous. However, the generated content, it's variation does save on disc space......but a hand crafted Milky Way would be good )) regardless of of the implementation strategies, I would like a serious sim, with the scale of the newer "drive" games: gtv v, the crew...... I even find project cars driving too dull...flat . Strangely with beamNG I get a kick of staying on the rickty road.....the penalty of the mistake is my appeal. Now let's have some super large maps! Are you an informed fan of BeamNg or part of the project?
I very much like the way it looks, I'm just not sure how they plan to reach their endgame. Thus far, even collision physics aren't yet a part of it. But, vector deformation is a part of the roadmap, so perhaps vertical cliffs and caves are soon to be possible, if not already. We'll see. The largest maps I'm aware of are Desert Highway (not the prettiest map), Black Hills, Grooved Range (made for flying planes), and perhaps the upcoming Forest Hills. Informed fan. I've created a few maps and a basic vehicle or two, but, yeah, mostly just a fan.
You have created some maps...hmmmmm. Any chance you are up to the challenge of creating a small scale maps, procedurally generated of course.....well content added with a few variations based on "road type" (even city, country variations)....phase 5 of course . I can get access to map data. I also have some dev experience and test.
I'm not sure what you're asking me to do. Do you mean, create the map using the topographic and road map data, then fill in the remaining details? Even that wouldn't technically be procedural generation, but it would be easily possible. At least in the basic sense. But, after a collaborative project, a few personal projects I'm trying to work on, and a new job I'm starting next week, I don't think I can take on anything too heavy-handed.
Ok after reading this all of you are wrong. Most of what you guys are saying traffic "is impossible to do" Your right for this part. But your forgeting the fact that this is in alpha not finished or fully optimized. What I mean there is later on physics won't use as much CPU. There for making traffic possible. Look at the game "my summer car" it ain't released yet but it got all the feautures / physics and full in depth car building more than slrr including building the engine and linking the tranny all almost done by 1 person. Beamng is a company not a one man job, it will get done trust me i'll bet 5 bucks by 2016 sometime around that there will be traffic. And here's another reason if they're making a campaign you won't be the only one on the road, if so its pointless to play. And its a waste of time. So yes it will be done. Sent from my crappy transformer prime.
While, yes, there will be more optimizations to come, most of those will be pretty mild in effect (10-20% maximum), while very light traffic would require about 40-60% less overhead per vehicle minimum (armchair guess). Most of the vast optimizations have already been made, one during the major code overhaul between September 2013 and June 2014, and one more recently due to the x64 support. Estama estimates that there are more optimizations to be made, but they aren't going to make miracles happen. Tetris is a perfect analogy. There are more and less efficient ways to stack the blocks, but there is a finite limit as to how much content you can possibly put in a row.
