Handling in BeamNG compared to sim racers

This. Some much this.


To add to the discussion, been racing on tracks and drifting irl and sorry to say but its not as easy as it seems in most sims. If it was as realistic and unfiltered in most sims as it is in Beamng then most ppl would think it was unrealistic. Id say many sim racers are spoiled without knowing it, they want to think theyre hardore racers but theyre really just playing a game as where trying the same in Beamng youre handed the cold, hard truth.

Having said that, i do think for exmaple FFB has room for improvement in Beamng, or could be made more userfriendly. To find the perfect balance between FFB communicating information through the wheel and the wheel shaking around can be quite annoying. Have seen the shaking in other games though so its not only BNG, probably just hidden in other games.

 

I get hammering of wheel in BeamNG with some vehicles, I'm not sure if that is shaking you describe or if you get oscillation from side to side?

Raceroom had also annoying hammering, that was when front wheels started scrubbing, canned effect I believe, turned it off as it rattles whole table.

In BeamNG have to run higher smoothing with Piccolina track variants for example because of that, with some cars I can then run less smoothing and FFB is more detailed, but if I use too little smoothing FFB strength tends to die off.

New updates to FFB have fixed few issues though, but my cheapo T150 is not walk in the park with any sim I think, in AC it needs half update rate for example, in Raceroom many settings I had to turn to 0% and some pretty much opposite of what they were, really lots of fiddling until I can get really good feeling of cars in any sim or game.

Then also FFB is difficult subject in general as it is so subjective, some can do about as good driving without FFB than with it, one guy at Raceroom forums test and was surprised how he got same kind of laptimes even without FFB, does not work for every person though, we use senses different ways.


What you write about difference between real world and sims, that is spot on, what little race track experience I have, I would say exact same. Sure there are differences in g-forces and how you see distances etc. but still, I think what you wrote is exactly correct.

I don't like Raceroom because how realistic it is (it probably is not), but I like it because illusion and sensation it gives me, for sure IRL I would not survive Nordschleife in 1992 DTM car if I would drive like I do in game, I would be in fence or something, but truly it is incredible fun in Raceroom, it gives you satisfaction while not failing the illusion, Simbin has always focused on that illusion like a magician and now as Sector 3 they are really good at it.

Still it is same for everyone, so competition is fair, but it is not 1:1 with real world physics, like BeamNG is, which makes BeamNG incredible difficult.

BeamNG could work as a racing sim in some distant future with lots of additions, but it might be too difficult and indeed people might complain it being unrealistic, when they can't master it same way as racing sim, but it might work out as people learn more about it.

I still remember races in GPL, that was quite difficult one to get right, even hardware was quite bit lacking back then, perhaps not super realistic, but sure we all had lots of fun with it.

 

That hammering you describe is probably what i mean too, it jerks the wheel fastif FFB is set to high with too low Smoothing.

I think BeamNG could work as a racing sim now if it was possible to have more cars and/or multiplayer, also we would need to drive more carefully since making a shortcut in BeamNG has real consequences . I saw your videos in the earlier post and heres mine on Nurburgring. Its a 8 min race with the ETK K Coupe Track version, so a decent time that can be improved in alot of places:

 

That is a good run, even end is not the best, but still really good, imo.

UI changes, proper garage page to make adjustments without lagging heavily, timing system, online leaderboards, something like that would be needed for racing use.

Of course good pretty race tracks, we have so few, would need much more beautified version of that Nordschleife.

I think that ETK K-series trackday variant has bit weak config though, it is AWD but also it has rear wing that provides quite lot of downforce to rear, which make it very prone to understeer, aerodynamic balance is not very good in it, imo.

Also it has very low front camber compared to rear as well as LSD at the front, pretty much everything works against cornering in that stock setup, I'm not sure why it has been set up such way, ESC tends to work better with understeer, so does AI and then keyboard and gamepad driving might also be reasons for that, but I find that changing pretty much everything in that config is needed to get it to corner well.

Removing the wing helps a lot alone, also power to tire ratio is bit challenging, I think it also had narrower tires at front than at rear.

I guess there is some reasons behind choices they have made, but to get car really show potential of BeamNG, whole different approach to setting up is required, again, imo.

I think in that sense Jimmy is right, but it is hardly fault of simulation if setup is not performing optimally.

My current WIP setup is here if you wish to test how well or poorly it fits to your driving style (it's currently very WIP though):
https://www.beamng.com/threads/etk-k-series-trackday-for-simracers.60432/

 

Cant remember what settings i was running when i did the last lap time but ill happily try your setup. Ive been wishing there was better/more setups available from start since its a mess to keep up with and track of all mods showing up.

 

Just played this game and was searching through the internet as to how it compares to sim racing games. While I could not accurately compare them toe-to-toe, I would like to correct some misconceptions about racing sims. Let's also clear one thing out: Raceroom, when compared to Rfactor, iRacing, Assetto Corsa, and Project Cars, is arguably the least realistic of all with all the available grip that almost all cars have. It also doesn't simulate enough parts like tire pressure, tire temperature, road rubbering, etc.

So, with regards to handling, of course, race cars will tend to have more grip than normal cars. You can definitely go flatout on some corners with those aero packages. Understeering is minimal (at least when the corner is i.e Eau rouge, etc). So you can't entirely dismiss race sims and their physics. Saying sim racers are "spoiled" is a bit.....taking it too far. And btw, most of you might already know this, but real formula 1 teams use RFactor Pro as their simulators. With the physics slightly modded but with the same game engine.

Next, I've only been playing BeamNG for less than 3 hours and went straight to testing a track car (i forgot the name) and within just minutes I was able to adapt to the game's physics. All stock. No editing of FFB whatsoever. I would say this game lies somewhere in the mix of rfactor 2 and assetto corsa. I can;'t say at what level, probably more realistic than the two, probably in the middle, who knows. But definitely the handling does not deviate from what I've got used to. But i would say i can relate very closely to BeamNG when comparing to my not-so-long track experience. I would also like to commend the one who said that the tires lock up too quickly. Actually, as what fufsgfen said, you can probably just adjust the pedal settings and it's all good. That is actually my only complaint in this game.

I really wish the devs would have modeled another car that is based on a real race car (fia homologated) because as far as the car roster of BeamNG goes, we can probably only compare it to Assetto Corsa. It has a wide variety of cars in which some of the low powered road cars can definitely be used to compare to BeamNG.

 

Welcome to the forums, and BeamNG Drive!

 

where can i get that map? I would compare beam to maybe like project cars or the crew in terms of driving experience.

 

Could be this, was a while back that i downloaded it so not sure.

https://www.beamng.com/threads/nordschleife-reuploaded-need-help-to-improve-quality.35143/

Its so intresting to compare BNG to PC2 or ACC, when you actually know the different ways that BNG simulates tires compared to most other games (which simulate them more or less the same way with some individual extra tricks).

 

Yeah, I can agree with this.

This is true... but remember... If I were so inclined, I could take GMod and modify the hell out of the physics to make it the most realistic racing simulator ever. It would still be running on the same game engine, but the physics would be 100% reworked.

That being said, RFactor runs basically the same deal as most other sims... they run things either smoothed a little, or with a little higher grip than real life, or not 100% feedback from suspension forces... whatever, you name it... just to make it a little more enjoyable to play.

To modify it for real life use, you should be able to tune back all these factors and get the "full experience" out of the game I am sure. The nice thing about Beam is that the way everything is simulated, they couldn't even add in these fudge factors even if they wanted too while sticking with real life physics. Basically, if the car can do it in Beam, then there is probably a way in real life that you could design a car to do the exact same thing.

This is the same experience I had. The game is natural and easy to get used to if you have real life driving experience first. The cars react very realistically to their real life counterparts (owning a similar vehicle to each in the game helps) and in my opinion the most realistic aspect is the vehicles independently simulated suspension components and tires (unpopular opinion I am sure)

Most games don't simulate suspension properly (and by most I mean nearly ALL). They simulate the equations, sure, but no game I have ever played simulates bump steer... or even can for that matter. No game I have ever played simulates bottoming out from jumps (short of LFS interestingly). No game I have ever played simulates scrub radius (its minor on race cars, but on Baja trucks... it plays a BIG roll... Baja Edge of Control "might" have simulated this). But most of all... no other game I have ever played (again, except LFS) properly simulated tires... hear me out here though as I can already hear the pitchforks...

Tires in most sims are nothing but a set of equations with a contact patch. They are basically simulating another set of springs that happen to be in contact with the ground. There is no rolling. There is no inertia. There are no moments being calculated... It is all just a big mathematical equation that is done on all 4 corners of the car. Ever wonder why almost ALL sims stick to 4 wheeled vehicles? Even if that game clearly has a tendency to add in odd vehicles from all different years (I am looking at you Gran Turismo). It is because adding more than 4 or less than 4 wheels messes with their equations. Their system is set up for 4 wheels, and 4 wheels only. How can I prove it? Well take some exceptions to the rule. The Daihatsu Midget in GT4, the Morgan 3 Wheeler in modern Forza games, and the Reliant Supervan in modern Forza games.

These three are the exception to the rule... so how do they work? They have taken the easy way out that simulates a 3 wheel car fairly well while keeping the 4 wheel physics engine. That single tire is actually 2 tires just overlapped to look and act as one single tire. Give that tire and its suspension half the spring rate of the normal suspension and tire combo then slap it on top of another one with the same properties, boom. 3 wheeled car. This is how it is done in almost ALL games as it is just the easiest way to simulate a 3 wheeled car without completely rewriting the whole physics engine

What about 6 wheels cars? Well... take notice at how few representations we have of 6 wheeled vehicles in simulation games. There are cars that would fit the demographic of the games that would fit right in (more Gran Turismo than anyone looking at their affinity for odd historic vehicles in their games) with 6 wheels, yet there is not one game with them in. Take either one of these two for example...



The reason being that these are left out is because the physics engine just wouldn't be able to handle these machines. They could simulate the front set, or the rear set of tires, but not both. They could also simulate the tires by putting a fake tire in between the two set of visible tires... but this makes the vehicle handle strange compared to what it "should" feel like. Forza claims that they might be adding a 6X6 into Horizon 4... we will have to see how they go about making that happen.

Beam doesn't have this issue... since everything is simulated the same no matter where it is in the game, you can make a 23X37 wheel drive machine with 19 wheel steering if you feel like and it will be fully simulated exactly how you expect it would be simulated (if your computer doesn't crash from that many wheels that is). Every wheel is fully simulated all the way around, not just a contact patch, every suspension component is fully simulated, every shock and spring is fully simulated... and it all comes together to create a symphony of physically simulated greatness!

Also, going back to tires... other than LFS, no other game I have ever played truly, properly simulates the rubberiness of tires. Sure a lot of sims nowadays visually show sidewall deflection and what not... but they don't really "simulate" it... at least not fully.

What do I mean by this. Go in Forza, Assetto... whatever... pick your poison honestly... and conduct this little test. Find a car with high sidewalls in the tires to get the proper full effect. Some old muscle car should work nicely, or if you can't find that in the game, any car with high sidewall tires should exhibit this effect. Now if you can find a way to hit this car with another car from the side gently, then do that, otherwise we are going to have to make use of the handbrake. With a second car, go bump into the first car side on... just a bump... maybe about 3 to 5 mph... then back off... if the tire model is simulated properly, the car should deflect slightly and then spring back when you back off. This will result in a slight oscillation from side to side in the hit car of about 2 or 3 oscillations. The same can be done with the handbrake... take the car up just fast enough to lock the hand brake and slide the rear end so that the car comes to a stop about 90* sideways from your original direction of travel.... when you are almost fully stopped lock the brakes. When the car comes to a full rest (like, when the tires finally fully grip up) the car should jostle around a little as it settles down. This is from the sprininess in the rubber of the tires. Most likely, whatever simulation you have chosen to do this in (either that or they have gotten some pretty good updates since I last played them) will not exhibit this behavior... more than likely the car will just statickly come to a stop and just sit there... no rocking on the suspension, no bobbing about on the wheels, nothing... just completely static.

Now go do the same thing in Beam... In Beam, you can even cheat a little by using the node grabber to enact a sideways force on the car. Again, go grab something with high sidewalls (lot more options in Beam, but I suggest using something similar to what you used in your other sim of choice) and conduct the same test. Look at how the car moves, how the tires deflect (best seen in slow motion) without losing grip. Look at how they spring back and end up over shooting a couple of times... just like the real thing.

This sounds extremely minor, but it is IMPERATIVE to have a proper simulation like this to capture every aspect of a tires real life performance. This allows the tire to properly simulate losing and regaining contact on a bumpy road. It allows the tire to flow over the surface rather than skip. It allows for realistic amounts of turn in or delay in turn in from lower or higher sidewall tires. It can properly simulate sidewall loading... heck it even simulates the moment of inertia of the wheels and tires and just about anything else you can throw at it... it does it all.

I could honestly go on and on and on... but no one wants to hear about that... point is... Beam has nailed it like nothing else can even come close too even if they are still missing some features like wear and tire heat... and people without lots of real life track time, or time to study vehicle dynamics seem to really miss this point a lot as they just can't quite properly draw comparisons to the real world. (this is not pointing out any disrespect to any person... especially who I am quoting here... I am just pointing this out)

Yeah... I have had some thoughts about doing something like this myself... taking one of my real life cars and 100% 1:1 recreating it in game as best I possibly could... so that I could go gather some real life data on the vehicle and then translate that into the game to see how it handles it. As one can see though I never did end up getting around to that yet.

 

From my experience the only bigger things that beamng could improve are:

• FFB during "extreme situasions". For example sometimes when one looses the rear end I feel that there is a lack of some ffb effects, mainly sideways acceleration.
• FFB setup, takes a lot of time.

Also beamng is the only sim that i know of that sort of simulates power steering.

 

@atv_123 I'm curious to know how you came up with the conclusion that racing sims (I'm actually thinking of rFactor-2 right now) don't simulate suspension properly and when it comes to tires are "nothing but a set of equations on contact patch"? I will agree that racing sims almost only simulate 4-wheeled vehicles so we can put that out of the topic. Please pardon the fanboy tone in me I'm sure it is obvious. But I'm trying to keep an open mind. To be honest, after all the comments above I could go back to rFactor-2 and not feel the same way ever again.

PS: here is sidewall deflection in rF2. Is this what you are talking about? I hope I got it right:

and also this


This greatly affects handling mid-corner depending on your camber setup.

 

Let me start off by saying I should have given a bit more detail and specific examples, plus have stated that I have never played RFactor 2... the game you appear to be most familiar with... have most of the others under my belt, but not that one. So anything I have said or will say I am not splaying to that game as I don’t 100% know. I just don’t and I am willing to admit that. I also want to say that I am not knocking on any of these games other than coming from the purist of pure physics standpoint. My favorite series of racing games is the Gran Turismo series... but I think that most will agree with me that performance of the cars in most of their games is a bit exaggerated (can’t speak for Sport as I have never played that either). Still fantastic games though...

Now that I got that off my chest, moving on...

How are most games just a contact patch. I stated that tires are a complex set of equations that are just calculated down to just a contact patch with the road in most games. Well, we literally have formulas for just that. I linked a video as well that discuses it in great detail... I have seen another video before where they go even further in depth with how to specifically implement these things in a game, but I can’t seem to find that right now (it is 3 AM... so yeahhhh)

Also the other reason is because I took a vibrations dynamics class in collage and these calculations were literally 50% of my homework. That was about 5 years ago now, but I can remember all the times I messed it up pretty clearly

This was just the way you did things in a time when computing power just wasn’t up to the task of doing massive amounts of simple physics. Instead you would save computational time by using large equations to realistically simulate the properties of an object as a whole.

Anyways, onto some examples and how you can experiment to see the contact patch in most games.

Method 1. Clime something
Contact patch simulation almost never has the ability to calculate more than 1 contact patch per tire at a time. Sorry to bring up that multi tire thing again, but it’s the same deal... if your engine can’t handle more than 4 contact patches, then it ABSOLUTELY can’t handle more than 4 under any circumstance. So if you throw them a curveball that is a situation that you normally shouldn’t encounter during the games normal conditions, the system breaks down. A good way to test this is to go try and climb something steep in the game that will force the tire to have more than one contact patch. Best examples are rocks. Rocks have lots of sharp angles to drive over, and most times the contact patch for one tire can be calculated in two different areas. A good engine will try to transition smoothly from one contact patch to the next, a bad one will jitter about helplessly whenever you end up in this situation (4x4Evo2 anyone? Not really a sim but it illustrates the point) This behavior can even be seen in some games with as little height as driving off the back of a curb. If you can get a camera angle on a tire and slowly drive over a curb, there will be a point where the tire instantly drops to make contact with the grass surface below... this point is usually directly on the center point of the tire... if it’s a good engine it will flex the tire to not clip the curb, but the drop is still noticeable at very low speeds as it happens almost instantly.

Method 2. Roll an open wheel car and try to drive.
This can be a little hard and some engines can handle this, but if you roll an open wheel car (60s F1 cars are a good example, huge tires and small body) more than likely one of your upside down drive tires is touching the road. If it’s a good model, then the patch will flip around and still make the calculations, giving you drive. If it’s a bad model, the patch is always calculated toward the bottom of the car. This means the tops of the tires never have calculations done to them and can’t drive you.

Method 3. Incorrect animations
This one drives me round the bends, but I never see anyone talk about it... perhaps I am the only one who noticed, I dunno... this is Forza specific example, but now that you know to look you might pick up other examples in other games. Forza does t animate steering correctly. Take any car in any Forza game (at least back to FM4 as that’s the oldest I have) steer the car full lock to either side, then drive slowly. Look at the tire on the front of the car closest to the inside of the turn. It literally isn’t turned even close to sharp enough to make the turn that the car is doing... add that to the fact that Forza has calculated tire deformation and you can see that Forza is definitely a contact patch game.

This post is getting huge and I am tired, so I present you with this video... it’s really old, but still relevant.

If you would like to see the start of his discussion on vehicle physics, start at 12 minutes (ish)

If you would like to see the equations for tires to allow the the calculations of slip angle and things like that, start at 17 minutes for steering slip and 24 minutes for acceleration slip. Remember, these are industry standard formulas... every game (sim) and tire company that uses formulas like this basically has their own tweaked versions of all of these formulas.

At 30 minutes he brings up the issue that stenyak mentions about how you basically have to flip between formulas when speeds get too low.

43 minutes he talks about a more comparative way of calculating slip angles. It is basically steering slip angles and acceleration slip angles all combined into on singular formula.

At 55 minutes he starts talking about how the physics are done in RoR (the predecessor to Beam... this is an old talk... I don’t even think Beam had even been considered yet by this point... this is like... Forza Motorsport 3 time here)

That is a really good point. It slipped my mind at the time about the transition state. Simulators have advanced tremendously so I am sure that the sidewall springiness is probably being simulated properly now by some games... well... at least, like you said, while moving.

 

BeamNG tire simulation use polygons, making the vehicle shake left and right at very low speed.
--- Post updated ---
Also looks like non of the simulaton does simluate this

This physics, making the drive wheels always rotate faster than driven wheels while cruising at a constant speed.

 

Well, it's not like Beam's approach doesn't work, it's just that there aren't a whole lot of contact points per tire. You can easily make this issue go away by cranking up the number of rays in each wheel, but in doing that you end up putting tremendous strain on the computer to calculate all those extra nodes and beams. So the devs found a happy medium that keeps things realistic in almost every circumstance while preserving the ability to still run the simulation in real time on fairly low end machines.

Honestly, to get rid of the jostling you don't have to step very far above the number of rays they have now... I think standard vehicles use about 18 to 20 rays right now? If you bump that number up to about 30 rays, you can almost completely eliminate the jostling at low speeds... but again... at a rather severe cost to performance.

What he was referring too though is the fact that most tire models literally do not work at all at low speeds... the functions (which can be found in the video I linked in my last post) literally break down completely because you end up dividing by 0 which is no good.

Also... you point out that no simulation does what you are referring to with your little picture there.

If I am understanding your circumstances correctly, you are referring to the fact that while driving at a constant speed, there is a net force being enacted on the wheels that is necessary to keep that cars constant speed. This is required to overcome the aerodynamic drag, rolling resistance, and drivetrain losses (plus a slew of other things, but predominantly those 3). While the car is keeping up this little bit of extra force, it is going to inherently load up the rubber in the drive tires in one direction and load up the un driven tires in the other direction. The only time the drive tires and the undriven tires are loaded in the same direction is during braking. When the drive tires are under this constant force, it will deform the rubber so that some of the rubber bunches up on the front side of the tire, while the rubber on the back side of the tire stretches out... This ends up being unloaded on the contact patch and thus a little bit of built up energy ends up making the tires turn slightly more than they should have... am I right so far with what you're saying?

Here is an extreme example of what I am trying to describe... the deformation we are discussing here though is so minute that it would be almost imperceptible during normal driving situations.



Interesting thing about this is that the way that Beam calculates its physics, it doesn't have to program this into the game as some specific use case... it will just naturally manifest itself from pure physics... no need to make changes to the tire model to make this happen as it already does.

 

ok then btw i used this chat as proof beamng is better then iracing to a xbox person they said "SiM rAcInG iS bEtTeR!"