I have always thought, and stated, that one of the better approaches for making BeamNG a true and engaging game, would be having a car design/testing/bench-marking/racing simulator (i.e. design your own car and then be able to test it/ race it against others' benefiting from the realistic physics and deformation ). I'm aware that it's been said already (merger with Automation, etc.) but I'm not asking for airbags, dummies or multiplayer so please don't throw me the first stone . What I propose it's a (probably difficult but maybe possible) designer module that leverages current vehicles/platforms (cab on chassis trucks/cars/vans, monocoque hatchbacks/sedans/sports cars) and allows, through an relatively easy UI, modification of parameters (wheelbase, wideness, height, etc.), body design, and components (engines, suspensions, transmissions, etc.) A little bit like automakers are doing now: platform MQB from VW serves as Golf, Passat, Touran, TT, etc.) I don't know if the idea it's Sci-Fi or not, and probably would not be very attractive for advanced users/modders that prefer a roots approach but I think it could be a good idea for gamers as there are other games that simulates well, or even better, the different game modes (racing, missions, etc.) that are currently on the table. Anyways just my own opinion/idea.
I get where you are coming from, but i think it's the sort of thing that would need to be designed/integrated from the very beginning, not added to a game half way through production. Games like Automotion and Gearhead garage both have the same draw back (other than no integrated driving model) is that design is always going to be limited to what parts are made by the dev team or modders. you can't possibly have it come up with your own design, unless they integrate some kind of actual 3D design studio. i'm not just referring to body design, but also even things like turbos that need to be designed to fit into an engine bay and onto the engine in such a way that it can be connected to the exhaust/intake system (plus routing all that custom pipe-work).
I love the idea, albeit it will be extremely hard to add to the game, however it makes sense, and is not totally out of this world, there is a small example in one of the tacky rice burner versions of Need for Speed from like 2008 - 2010, where you can take the body kits and drag and morph the pieces individually (im pretty sure it was NFS) but like you stated, this game is all about the root approach, all the cars are modded poly by poly and the coding is a completely, although connected, separate entity. But I like the idea at least
All good points, specially the planning (from the beginning vs. through production) one. I reckon it would take a lot of time/money/effort but also think that the final result could be a game changer. My, probably wrong, approach would be a compromise between making room for the turbo (fine) and have to route the pipe-work (too much): somehow arcadish in the design side but hardcore simulation in the driving side. It would be very fun, educational and creative, a little bit like Kerbal, but it's probably just a dream. - - - Updated - - - Thank you . Also "morphing" was the word I was looking for. Yeah, a little bit like NFS but centered in functional parts instead of tacky body kits. It would be like making modding accessible to dummies (like me) but in a meaningful way.
BeamNG vehicles are split into two major parts. There's the visible model which looks nice and pretty, and there's the underlying physical skeleton made up of nodes and beams tied to it. This skeleton is pushed to the game physics to be simulated. The nodes and beams of these skeletons are defined in .jbeam files, where each individual node is created/positioned by entering its coordinates in 3d space. When the game spawns a vehicle, it reads the node coordinates as they are and spawns the skeleton relative to a local origin. An Autosculpt-like feature would have to read multiple node parameters simultaneously and somehow change them on the fly, which isn't really practical (or even possible?) at the moment; changes to the skeleton requires the edited .jbeam to be saved, read from again, and the vehicle respawned. Moving small groups of nodes can affect the behaviour of some components such as suspension, or even compromise the integrity of the skeleton. The values that determine the physical behaviour of each beam will need to be adjusted for realism and stability, as beam length and function is directly tied to a certain, often very small range of acceptable beam values. Tuning beam values is difficult to do by hand, and would be pretty much impossible to do algorithmically due to the number and complexity of variables involved, such as whether or not the beam forms a part of a crumple zone, is key for correct suspension geometry, must remain stiff to keep the occupant cabin intact, etc. Changes to the skeleton also aren't tied to changes in the visible model on spawn. So if you scale the skeleton to 1.9m wide from 1.7m, the visible model will still only be 1.7m wide, meaning the vehicle will be able to collide with things that the visible model doesn't appear to be touching. With major scaling comes issues with skeleton resolution. Taking the rear half of a Grand Marshal skeleton and moving it back 5 metres to create a limousine will create a node deficit in the space between the front and rear halves which will compromise crash behaviour. Some community members have indeed created jbeam generators for very simple objects such as cloth and things with a uniform cross section, but cars almost never have a uniform cross section for any appreciable length save vans, buses, and limousines. Anything such a generator does for a skeleton will require modification. tl;dr it wont be fun to do. edit: didn't see the functional parts bit. Engines and transmissions and such are also defined in a .jbeam file. Torque curves and gear ratios are just plaintext in there. The physical skeleton plays no part in the performance figures; engines are usually a simple cube, and transmissions are prisms tacked on the end of it. The visible model can be changed to reflect different engines or trims or whatever. If you want to see piping making a difference, try CFD software, haha. If anything up there is incoherent, it's probably because I'm writing this at 3am
I've been busy but wanted to thank you for your solid and convincing response (even coherent at 3am ). You are probably very right, I can't say it because I lack the knowledge. Let's agree my first approach it's not feasible. Anyways I am little stubborn so here is another idea in the same direction: Official cars right now are somehow customizable (you can even drive them bare bones). How about producing one (or more) specific model/s that serve as a kind of platform. I mean without stretching/modifying, etc. the jbeams but with the ability to add different bodies, engines, suspensions, wheels, etc. It's very similar to what you can do right now but the difference would be in the parts. They'd be more diverse and the base model would have to be conceived more as a platform than an specific kind of car (pickup, van, full size sedan, etc.) Also it would probably work better with cab on chassis (old) american standard than with a monocoque approach. Does it sound feasible? If anything up there is incoherent, it's probably because I'm writing this at 23 pm .
