What @VeyronEB said pretty much. @estama spent (and is spending) an incredible amount of work and time to improve it to what it is now (compared to the very older versions, sticking is heavily reduced). It's not an easy thing, as you need to balance performance (yet estama still manages to squeeze out more performance from the engine). Think it as a constant work to a math problem, where the goal is reaching a more accurate result. Utterly old video: Compared to the recent versions (where we have more measures to against sticking), the improvements would be much more visible.
Yea it's good to see the progress. The mod author has tried to change some values to make it less sticky, but maybe the main game's own limitations just can't be surpassed, until the progress with those will improve. Anyway, this creates a lot of frustration, since most of the rounds end up into pileups with the AI cars, because they just simply stick together and can't get rid of it. The AI is still WIP, so that's not helping either, since they don't have the skills to try to ''twist'' their cars away from those situations. I myself can get away from those sometimes, but not always.
I'm sure that beam to beam collision is needed. I don't know how hard would it be to implement, but it would deffinitely improve collisions. For example, now you can stick two hay bales into each other (If you rotate one by 90° and place it on other one). Similar things may also happen between some parts of two vehicles.
This is something that we have been constantly been working all this time. Both collision stability (which you mostly see when stacking things) and collision accuracy (that tries to hold things from passing through each other while not destabilizing physics) have improved immensely in all these years. You only have to test a version of the game 5 years back and you'll immediately see the difference (i have done that). For example, the "anti-sticking" improvement from some time back, greatly dropped probability of sticking to a single digits percentage. There will also be more improvements on in the next update, as it is a subsystem that is near constantly being worked on. And if you wonder why it is taking that long. The reason is that we are not faking collisions. We are not "fixing" positions, velocities or anything like that. Everything happens through force transmission, and the code needs to be able to handle stacking complex structures like cars, and stopping a multi-ton car in 2.5cm without destabilizing the rest of the physics. To give you and indication of the effort that something like that requires, i could write several academic papers on the subject. I'm reading newly published papers on the subject, and at most i'll find a hint of one of the many techniques i'm already using in there.
Beam to beam collisions are something that we are already researching. For the time being a partial workaround to that is to increase the density of nodes on the edges of objects.
@MRcrash about sticking, observe how differently haybale and metal box are made: Metal box is not sticking much, because way it is made had more nodes at edges like estama hints. This same principle goes for any object that is supposed to touch something. So for example fork attachment and pallet, you would need to put slidenodes in rails to make dense node netting with bit easier work. Darren9 explained this technique well in mod support section: https://beamng.com/threads/rock-slider-concept.51970/ So when observing that metal box you can see how it has kind of web structure, all intesections of beams contain nodes so that it is not so easy to pass things trough it. Still it is lot of work to add so many nodes, but that is way to get nice gameplay at the moment, both pallet and fork would need to have this dense netting. Also sometimes it might work better to have shape of nodes and beams bit simpler than real object shape, for example I remember one mod had kind of mad max themed bumper that was sticking a lot, it might of worked better with simple straight, but dense node net. Kinda would be handy if there would be program that would add nodes and beams for low density structure making it high density structure with single click, but one can always dream. NBEditor without bugs would be nice too
I used this on wheel loader shovel but sometimes it still gets stuck into things. And another problem is that this sutructure is more unstable because of low nodeweight.
You would need to add more nodes to sides I guess, but then again weight would go up or node weight down, creating new trouble. Maybe weightless contacting nodes could help, but I guess then some other deforming type of issues would happen. Besides, figuring out how to have particles without a mass, would probably be sure way to get Nobel prize or something and in BeamNG world nodes are particles, beams are nothing, but energy holding particles at place
Agreed. The Moonhawk is roughly based off of the GM Midsizers from 73-77 (Malibu/Monte Carlo, Cutlass etc...) It's considered a midsize even though the IRL cars were known to be noticeably bigger/heavier than the typical US midsize (think Ford Granada). So the confusion is perfectly understandable. I actually used to think it was based off of the Aspen/Volare, until I realized those are Unibody (and smaller), the Moonhawk and the GM A bodies are BOF. I also can't wait until the cutting is implemented. Homemade convertibles!
I'm more curious about what actual new content we'll get in the next update. Will the Bluebuck be ready by then? And what are the chances they we'll get new props and parts/other options for the vehicles already in game? I rule out maps since WCA should keep us satisfied for a while, and Italy is evidently a long way from completion, unless there's something else that has yet to be revealed.
I'm honestly surprised we didn't get the Bluebuck in one of the latest updates, but I've heard that the person whos most involved with vehicle development is recovering from some kind of Illness. Hopefully wagon and convertible versions of the bluebuck are available. I'd expect powertrain options to be pretty similar to the barstow.
Ah, that is indeed longer. You could fit three Covets into space two of those big ones occupy. My GPU is already working very hard in BeamNG, CPU has cores and threads that are not doing much though. Somehow BeamNG is now in 0.11 version GPU heavy, rF2 has more CPU usage than BeamNG with single car, which I find bit amusing.