I'm going to say something controversal. Personally, I don't like the campaign. It's a collection of scenarios with cutscenes. (The art is amazing). Career mode is what I am most excited for. Ever since I got this game a couple years back, I didn't care about multiplayer, or scenarios, I wanted career mode. The fun in finding an old, rusty, broken Moonhawk in a barn and fixing it up into a drag beast, then winning races until you buy a Sunburst. Keep racing, keep driving, keep driving that GM taxi, and finally buy yourself an SBR4 TT S2 and driving like you're the most important person in the planet. That experience is what I want in a game like this. I think when I get home I *might* right a storyline for the career mode, to show what feels right.
I think the biggest issue is that it feels empty. In any type of visual media words to the effect of "show, don't tell" are often used, don't tell the player things through cutscenes if you can show them those things through gameplay. Currently there isn't any (major) narrative and little is shown through the world. For example, if you are going to have a drag race, you are probably going to do it on a road that is well known for being used as a drag strip. So maybe some evidence of that being the case could be placed around the roads such as skid marks and maybe even a wreck at the side of the road. Things like this turn a sterile environment into a world that has stories to tell. Then there is also the narrative dissonance between the activities that you take part in and the story that they are trying to show through the cutscenes. Since almost every event causes the car to get wrecked and quite badly. Which narratively seems strange since the story hinges on them being broke and not being able to afford "nice things". Not entirely sure how they could practically overcome this one though, but I guess that is why they have hired someone for a gameplay and narrative design position within the company. --- Post updated --- Something that could be neat is a Midnight Club style mission start system. In midnight club races are not shown on the map at a certain location. Other drivers are shown instead. Then you drive to the opponent and flash your lights at them. At which point a skippable impromptu race to the start line begins seamlessly (so you flash your lights at the other driver and then step on it). Midnight club never explained how you knew other drivers were competitors. It was just "one of those gameplay things" for the sake of the game. However it could be quite easily masked these days as an illicit underground racing app for mobile phones that places other racers who have the app open on the map. So for competitive events against other racers you could use that, then for personal challenges (or Strava like leaderboards) you could mark them on the map. This would also allow for world building, for example public perception and knowledge of the app heating up. Stuff like that. But of course this may be miles off the devs vision for the game.
Wow. My dad expects me to be perfectly fine with grounding me for the smallest thing, and then extending it for SOMETHING I HAD NOTHING TO DO WITH(My brother made him mad, so he grounded him and extended mine. I asked him what I did wrong and he said "nothing.") He was actually surprised when I complained. Sorry for caps, but I'm mad.
I pictured that the difference between the career and campaign modes would, in the game's final state, not be so black-and-white. That campaigns would slowly introduce concepts, such as permanent damage and career earnings, and those features would be used toward the feature set of an emergent non-story or story-light open-ended career. At the end of the game's development, it would be at the behest of any campaign's creator whether the campaign would simply be a string of objective-oriented scenarios, or a full-on sandbox career. Also, though I know it's a bit petty, I really don't like the lack of a 'save progress' feature for the campaign mode. I can play through the campaigns in fairly short order no problem, but it just doesn't feel polished not to have such a standard feature in place. Interesting, I think that this is one of the only times I've seen a discussion about game design on the forums. I agree that more and better worldbuilding would be pretty neat. Doubly so if the story is meant to pick up in the face of something bigger or more far-reaching, such as police cracking down on street racing or a massive international motorsports event approaching. That, and the cutscenes aren't perfect (a few moments of uncanny valley, and why is there suddenly a crowd of 40+ people in the middle of nowhere to see two nobodies race). As for the Midnight Club formula, I'm not sure how willing the devs would be to have extra Jbeam objects in the level, if there were to be more than one race available at a time. Still, one of the big complaints about BNG is the lack of ambient traffic on the roads, so giving some of that traffic a gameplay purpose might provide a good, solid excuse to include it.
TOOBES! a 6CA4 rectifier tube two 6BQ5's and a pair of 12AX7's. Typical single endded tube amp design, in the middle of restoring my Eagle SA-100, probably going to replace all these tubes with new matched ones, or at very least the rectifier tube which has a short from heater to cathode.
Yeah, this is an issue, one that I am not entirely sure how they will overcome. But IMO they are going to have to overcome it somehow because an open world driving game with no other cars is kinda inexcusable in this day and age. A couple options that I see are: Have really sparse roads with pop in You could probably get away with having only 3 other cars on screen, however you would have to have them pop in within a couple 100 meters of the player Have AI cars that have no deformation Frankly this would work, but it would suck. The pro's and cons are fairly obvious so I won't detail. Have AI cars with simplified deformation using very simple jbeams Once again this would work, it could even be fairly effective. So long story short the idea would be to have jbeams that only have around 20 nodes in them. The jBeam would not handle any of the vehicle dynamics, that can all be handled by more traditional methods (same as anything else [even on rails would work, if primitive]), this driving rig would have 4 anchor nodes in the middle. You can then build a very simple jbeam around it in order to add simple deformation. Hot load the vehicles in run time This one is a lot more interesting! Essentially all the cars would be models of in game cars and nothing else, then whenever the game predicts the player will hit another vehicle it will hot load the jbeam in as fast as possible in the background, then you get full jbeam physics upon impact. This is difficult to do though. My guess as to how you would implement this would be: Load all vehicle jbeams in the scene into RAM without simulating them. When a crash is predicted you duplicate the data in RAM to create an instance of the jbeam, position it against the cars model, then enable simulation for that. Simulate for as longs as needed, then stop simulating it. Then depending on how much RAM is available you will need to start despawning simulated vehicles (that are no longer relevant) when they are out of frame. However that is a lot of algorithms (crash prediction algorithm, algorithm for duplicating data and starting the simulation (in less than 0.1s), algorithm for figuring out when to stop simulating, algorithm for deciding when to despawn the car to free up RAM)
Won't hot loading the jbeam cause massive lag in some PCs, especially if the player is about to hit multiple cars at once?
Depends. If it is already stored in RAM you can copy the data very quickly without waiting for a hard drive, if the loaded jbeam is only a couple hundred kb then it should be fine. Of course pre loading all the jbeams will make initial level loading times longer. Then you need to translate the coordinates which shouldn't take longer than 1 frame. Then you can start to simulate it. This could all be done on a separate CPU core to cause minimum interruptions to gameplay. I think it would probably be possible to do in under 5 frames fairly seamlessly, you could even do a burnout paradise style crash camera to buy yourself extra time. But obviously I haven't tried it, so who knows in this case. One other difficult aspect which I hadn't mentioned is syncing the two vehicles. So that when you do the switcheroo the suspension is in the same place and front wheels are pointing in the same direction.But this shouldn't be CPU intensive, its just time consuming code to write. For clarity, I should add that the main reason why spawning vehicles takes a lot of time is because the game has to convert the jbeam into data structures that it can understand. They will have to parse a JSON file in order to do this and then may also have an intermediary step in the process where they run lots of error checking. Then it also has to copy textures from the cache to the GPU which takes a while. If you do all of this when the level loads you could spawn vehicles super fast.
I've used code that can reposition and reset vehicles at runtime in a few scenarios. The trick would be to make it work only in the player's vicinity, and using random respawn points dynamically placed along roadways. You'd also have to consider the player's speed, heading, and the posted speed limit, so you can position vehicles mostly ahead of the player if they're moving faster than the traffic or behind if they're moving slower. I actually did a little thinking about how many traffic vehicles I'd consider the minimum to make the maps feel like living places, sans pop-in. Utah is a pretty remote place, so three or four traffic vehicles over the entire map would probably work well enough. JRI would need only two or three plus a good reason to be someplace so abandoned. ECA would need at least seven, though, and WCA an ungodly large number without pop-in.
I think WCA would need at least 100 or so, with most of them located in Belasco...that would be so hard to run I'm quite sure nobody would be able to play the career mode at playable FPS, if at all.
Indeed. With pop-in, however, one could probably make due with three to five traffic cars. Archetypal NYC traffic is unlikely to be possible for a long while, but with the exceptions of the freeways and central business district, most of the streets of Belasco could viably be sparse at near any time of day. Edit: I have a little project that I'm working on, but I need some assistance from a trusted forum-goer that knows engines. Physically, visually, and how to translate the real-world specs into BNG performance. Any recommendations?
Well yeah. That's why you don't run any of them You just run mimics until the player gets within touching distance. When you initially load all the vehicles (when loading a level) you would load them all as fully as possible, meaning placing the jbeam is super fast. You wouldn't need to change the visible mesh of a vehicle either, you just slide the jbeam deformation maths underneath it. So you would only be swapping in the physics skeleton alone. So you would be able to do it in the same or less time than it takes to reset a vehicle in game (as in pressing the 'i' key reset), which is already very fast as is. --- Post updated --- Depends what you need really. I am more than willing to help you out with stuff that I know, providing it is helping you out and not writing it myself . My knowledge of engines is very limited to be frank, but my knowledge of making stuff in BeamNG is fairly ok.
Sanded and polished up one of the output transformers from this tube amp. Looks pretty compared to it's brother.
Maybe. Still, I think that even new instances of the same Jbeam have an effect similar to loading in a new Jbeam. Spawning in a Moonhawk, and then spawning in another Moonhawk of the same configuration still take just as much time each. To my knowledge, at least. Mostly understanding torque curves, as well as building the 3d model of an engine that isn't some bastard of mechanical misconceptions. No idea why the forum is splitting up your quote.
That wouldn't be an issue at all. AFAIK, For each vehicle that you load, BeamNG loads the vehicle as a brand new one from the hard drive as if it has never seen it before. Doesn't matter if you already have one spawned. If it had already fully loaded the vehicle into RAM, and was already displaying the mesh on screen, there is basically nothing left for the game to do in terms of loading.
I'd live to help with your project! I am experienced with engines, and I think I've been here long enough to be trusted. Torque curves are my specialty
