1. Trouble with the game?
    Try the troubleshooter!

    Dismiss Notice
  2. Issues with the game?
    Check the Known Issues list before reporting!

    Dismiss Notice

Guys, come to talk about tires

Discussion in 'General Discussion' started by fufsgfen, Oct 26, 2018.

  1. vmlinuz

    vmlinuz
    Expand Collapse

    Joined:
    Mar 2, 2014
    Messages:
    2,409
    @estama DESTROYS BeamNG Tire Model Skeptics EPIC STYLE with FACTS and LOGIC while Need for Speed Fanboi WATCHES

     
    #21 vmlinuz, Oct 26, 2018
    Last edited: Oct 26, 2018
    • Like Like x 4
    • Agree Agree x 3
    • Informative Informative x 3
  2. Diamondback

    Diamondback
    Expand Collapse
    Vehicle Systems Lead
    BeamNG Team

    Joined:
    Apr 8, 2014
    Messages:
    1,957
    Yep that's unfortunately a very valid point. We get confronted by videos of people trying to go around corners at full lock all the time.
    There's only a very few people that we know of that can race properly with a gamepad, let alone a keyboard ofc.
    So the first step to properly race in beam is to get a quality wheel, FFB is a welcome addition but really the much more precise input is the actually important thing.
     
    • Agree Agree x 16
    • Informative Informative x 1
  3. atv_123

    atv_123
    Expand Collapse

    Joined:
    Aug 5, 2012
    Messages:
    1,710
    I find it amusing that this is a thread to discuss tires and then almost no one talks about the tires... people just say "they suck" and move on... no examples... no nothing. Come on guys, we are better than this...

    Edit: This first bit is now irrelevant as it took me so long to write this post that people have now actually talked about tires.... awkward.

    About the only thing I think the tires could be missing (well... other than temperature and wear, but we are all aware of that, so let's move on) is a lack of directional node grip. Now some of the people on here may not know, or remember, directional node grip, so let me explain.

    @estama, you might remember this from back in the RoR days. Basically it had to do with nodes being able to provide different levels of static and kinetic friction depending on the direction the force was applied in. So, basically, in terms of tires, side to side friction coeff could be different from forward and backward friction coeff. This, in essence, gave you the ability to model tire treads (kinda). Let's say you wanted to make a paddle tire. To do so would mean you would set the side to side friction coeff relatively low, and then set the front to back coeff relatively high. This would mean that whenever you tried to take a corner, you would slide sideways over the big paddles, but when you just wanted to go straight, you would have tons of grip.

    Now that is a very specific example and in normal road tires, these numbers would be fairly close to each other... in fact, I would imagine that slicks would be just about identical values in both directions. This is mostly useful in off road situations where tires would dig into something. Nevertheless, they could be tweaked just the same to help fine tune tire feel. I am just putting this out there as food for thought... you guys might have tried it already and decided that this just wasn't the way to go as it might have been seen as more of a cheat, and I am totally fine with that.

    Having said all of that though, I don't actually have much of an issue with the tire model at all. Is it missing things like wear and temperature? Well, sure, but this game isn't exactly released yet... things are still being developed... so on and so forth... basically I am fine with where it is at right now.

    I personally think the tire simulation in Beam is very good. It can accurately represent parts of a tires tread rather than just representing the whole tire as nothing more than a mathematical equation and some fancy model deformation. Beam doesn't have to fake it because it is the real deal. They don't have to do any fancy codework to get these tires to work well on uneven terrain or strange items like rocks or anything like that with multiple contact patches... technically speaking they have as many contact patches as they have nodes in the tire... no fancy code required.

    Moving onto road use, the tires in Beam react physically just the way you would expect a real tire to. If you push the tire to hard, it will lose grip... some types of tires have a smooth static to kinetic friction transition like normal road tires should. Other tires snap from static to kinetic friction just like a proper race tire does. This to me feels amazing as most games have just one tire profile or equation that they then tend to use across ALL of their tires... Beam stands alone in this department.

    Another thing to think about is that while it might not "feel" realistic to you while other games do... a lot of other games auto calculate your slip angle for you so that when you jam the joystick to the left or right, rather than you having to feel out where the slip angle is for you to turn at your best, Beam just lets you go lock to lock... just like real life... this will induce understeer, and it is up to you to sort it out. No driving help here, your on your own (other than the particular cars driving aids I suppose, but still... just like real life) The only other games I can think of that do this are Assetto Corsa and Project Cars... pretty much every other game will help you through the corner weather you want it or not.

    In conclusion, the tire model in Beam is very good... while not perfect in one specific type of driving... it is INCREDIBLY good in almost all other aspects of it. How many games out there can claim to use the same tire model for racing and rock crawling with literally no modification to it between the two... Not many... I can only think of one. (2 if we want to get cheeky)

    TLDR: I like the tire model. I think it reacts realistically to the environment it is in and holds up very well to very different types of vehicles.
     
    #23 atv_123, Oct 26, 2018
    Last edited: Oct 26, 2018
    • Agree Agree x 8
    • Like Like x 7
    • Informative Informative x 1
  4. MrAngry

    MrAngry
    Expand Collapse

    Joined:
    Sep 5, 2013
    Messages:
    612
    I agree with the approach of simulating real behavior that then applies to any situation in-game, however I have trouble putting confidence into BeamNG handling despite knowing the roads/tracks and cars very well. In other racing sims I know exactly when I lost grip and when I'm going to gain it back again which BeamNG somehow fails to communicate back to me.
     
  5. monobrau

    monobrau
    Expand Collapse

    Joined:
    Aug 18, 2013
    Messages:
    200
    Lol whoops. bad reply. I disagree. I don't have that issue. The way you input your controls to your vehicle should dictate how your vehicle reacts.
     
    #25 monobrau, Oct 26, 2018
    Last edited: Oct 26, 2018
  6. fufsgfen

    fufsgfen
    Expand Collapse

    Joined:
    Jan 10, 2017
    Messages:
    6,782
    I made a car that can understeer, even badly if you fail entry/turn too much, but that has also very easy to feel grip point, at least with my cheap wheel and settings (using 1080 degrees).

    In AC I bought Nordy DLC as it is so praised, I think that with that came E190 Evo II race car, ETKI has similar wing, so I made engine and config to have somewhat close to same engine torque. Then I adjusted suspension parameters and brakes a bit to my liking.

    It will plow straight if you turn too much, but it will turn really well with mild oversteer if you don't turn too much, might try to tweak bit more to get bit less understeer when coming into corner too fast, but if any of you would like to test there it is, all 10 000rpm of inline 4 roar.

    Tires in this seems to be quite easily causing plowing straight effect, might try Clockwise 533 and racing tires as those have been working really well for me before.

    This mod is nothing special, just changes torque and few other engine parameters, so don't except much, I just was curious to compare E190 of AC and ETKI in driving feel.
    --- Post updated ---
    Here is limiting steering lock which I talked about, set it something like that and you get full lock for maneuvering and limited lock for driving, it is also not changing at middle of turn, which would be annoying.
    upload_2018-10-26_22-58-5.png
     

    Attached Files:

    • Like Like x 2
  7. STEVIE_G4M1NG_YT

    STEVIE_G4M1NG_YT
    Expand Collapse

    Joined:
    May 26, 2017
    Messages:
    663
    Tires.. Yes I like tires.They stop my rims from getting scratched and sparking. :)
     
    • Like Like x 1
  8. Goosah

    Goosah
    Expand Collapse
    Global Moderator
    BeamNG Team

    Joined:
    Aug 30, 2013
    Messages:
    790
    I've been meaning to write a new blog for a long time, since that original one I did in 2014? Is pretty much laughable compared to the development since. We are not just sitting here with our arms crossed going "durr, its realistic" without doing any science ourselves.

    First off, racing sim tire models are at their core, a lot more simple and limited than you might know. They have a constraint to a point on the ground (or hopefully more than one), and compare angles and ratios of slip, which are used to pick some values for friction out of a lookup table or set of equations. The complexity comes in trying to layer all sorts of behaviors such as, how should the equation should change based on the angles and ratios, and how do they work together. But at the core there is no rotating tire there, just a point skipping along the ground at the physics frequency. And looking at various inputs, it changes the forces applied to the suspension hub rigid body.

    Advantage?
    Well, so far it's the only way people knew how to approach the problem, at least for a real time application. It fits in naturally with rigid body/constraint based game physics. One can pick whatever values they want for the tire, meaning whatever tire data they find, they can do the best to fit their equations to the data. So its the natural thing an engineer likes, they like to have absolute control in their model above all else. Now, as I said the complexity comes in adding all the layers of stuff on top of the rather simple concept. So there are values for temperature, pressure, etc all modifying the curves. And its not just about finding out how each variable works on its own, but how they complement each other. A change in pressure might do something different to the tire at a high temperature than a low temperature, the camber will affect the tire differently at different tire pressures, and so on. It becomes a crazy amount of work to figure this stuff out. And in the end, the latest racing sims get really close in terms of the pure handling aspect.

    Disadvantage?
    The model breaks at low speeds (because the slip ratio and slip angle get closer and closer to /0). Most racing sims switch the tire physics to some other mode below a certain speed to avoid the problem. The cars abruptly come to a stop, FFB often switches to another implementation to just provide a damper on the steering, etc. Another disadvantage is that there is no physical rolling tire shape, meaning for stuff other than racing on smooth surfaces, the model breaks down. As the tire contact point skips over the ground at higher speeds, it can skip right over detail or objects in its way. At high speed the model could drive right through a speed bump in the straightaway of a race track without registering a thing. A simple example that would foul these tire models is just rolling up to a curb so that the tire now has 2 contact patches. I tried this in various sims and what I saw happen was that the tire instantly jumps from being on the road to being on top of the curb.

    If we stuck a "standard" tire model under BeamNG cars, 90% of what people like to do in the game would immediately start hitting the limitations. It's just that racing sims hold a very narrow focus and don't really let you do the things we do. And for their purpose their models work well, you can find they match side by side videos on race tracks etc. But they are not designed to do that and also get an SUV driven through a moose test, off a jump, or over another vehicle, and so on.

    Our tire model is actually as close to a physical entity as one can hope for with current computing power. The tire exists in the same physics "universe" as the rest of the vehicle structure so there is always physical consistency and conservation of energy. It has an actual structure with distributed mass, internal forces, deflections, inertia. Right away the majority of the practical problems with the standard tire model are addressed without ever having to code anything tire specific. There is no low speed issue because the physics is not looking at high level things like slip ratios, instead forces come naturally as an end effect of the contacts and flexibility of the structure. The tire can contact anything and everything at multiple points. When a 4x4 falls on its side, it can spin itself around one the ground because the sides of the tires are contacting the ground, and sometimes the rim too, leading to spiral shower of sparks :cool: (Who knows what a racing sim here would do, with 90 degrees camber input to their model) The tire can travel at any speed and will always have the same resolution on the ground, so it will never skip over objects like curbs or miss detail in the road surface. All in all it's a very elegant and "honest" way to approach simulation. What we get is "completeness" that survives the most ridiculous possibilities one can dream up.

    In terms of a handling model, we have also an advantage: Once we match the tire model to known data in all the ways we know how, we can put trust in it to fill in the gaps in our knowledge. Basically, the tire will act like a tire, whether we understand what should happen in some case or not. In the real world, somebody changes the rubber compound mix, change the tire belts, etc, and pops out a new tire from the mold, and maybe it is good, maybe it is not. But in the end, the tire just is. It's up to you to measure it and learn what it does in testing, but it will just do its thing just the same without anybody understanding it. Creating things in our physics is similar. Rather than trying to distill a few different specific observations of reality into some curve fit equations, we distilled a structural simulation of the world, which the tire exists in. So in 2013, it is like we popped out our first ever tire from the mold, but we didn't measure how it performed, didn't know how it should perform, and we didn't have the tools to iterate the design.

    But since then we have. It's been a hell of a lot of hard work, solving simulation problems that nobody has touched before. We work with fundamentally different systems than any other simulator, so we are on our own completely.

    Keep in mind that no real-time model is going to hit all the behaviors of a real tire. The standard model may, by definition, match some measured data. But it can't fill in the blanks, every behavior needed must be added to the model. If one is not aware of the behavior, it can't exist. Our model fills in the blanks, but the challenge lies in calibrating it to measured data. So given that our model fairly low res right now, we will get 90-95% the way and then from there it is a matter of balancing design trade-offs. Of course as computers get faster we will inch closer and closer. Anyways, for now we have a tire model that excels in some areas compared to other sims, and in others it lacks. But we feel its the way forward regardless, as manually adding more and more complexity to the standard tire model is becoming scarier and scarier.

    That being said, let's look at some measurements of the front tire of the ETK800 in action on my tire testing rig. Its a 225/45R18 "Sport" tire (we've kept it pretty generic so far but this would be like an oem high performance summer tire one would get on a luxury car).


    This test shows the lateral force produced as the tire is steered, at 5 different loading conditions, and 0 camber. The horizontal lines are the measured normal force during the tests. So the tire here is peaking at around 7 degrees under light load and shifting over to around 9 degrees as the tire gets heavily loaded. The peak friction is also dropping relatively as the load increases. I happen to have some data on a 225/40R18 tire that we match really well here.


    This shows the aligning torque of the tire as it is steered, same conditions as above. This aligning torque, plus additional torque due to steering caster, is the primary force that ends up being felt at the steering wheel when turning. As in measured data, the values peak around 1/2 way to the peak grip slip angle. In this tire its a bit too early for my liking. As the slip angle becomes extreme the torque reaches 0 and starts to invert a little bit. This causes that dead feeling if you are understeering or didnt steer fast enough to catch a slide. Keep in mind the caster adds on to this and "fills" the curve out more to give some more consistent feeling.


    This shows the lateral force produced as the tire is steered with different camber (0,4,2,0,-2,-4). Negative camber here is increasing the peak force and also providing a shift to the line so that it no longer intersects 0,0. This means that a cambered tire "pushes" towards the direction it is cambered. I believe the term is "camber thrust". Doing pretty good here still. But, probably too much change in peak grip and not enough change in lateral thrust.


    Here is the lateral force measured at different speeds (14, 28, 42, 56, 70 m/s). Due to the velocity sensitivity of rubber friction our model predicts pretty high grip at low speeds (green line), as even when the rubber slides, its not being forced to slide very fast. As the speed gets higher, the grip falls away a little faster past the peak slip angle. The aligning torque hiding in the bottom corner starts out really high, drops off and gets shorter at higher speeds, because the tire contact patch gets smaller due to tire centripetal forces.


    This is the longitudinal force vs slip ratio at different loads. Braking to driving from left to right. If you compare with the lateral vs load curve, the tire is producing more force in longitudinal directions than lateral, though less in driving compared to braking. But what is interesting is that the node friction is not directional, these differences come out from the tire structure combined with the load sensitivity of the node friction. Basically, lateral vs longitudinal loads change the force distribution on the tire, not getting quite as much out of the nodes in the lateral direction as longitudinal. Also beware there is a bit of an offset here due to rolling resistance shifting things (I think).

    What remains to work on?
    We have some issues with the lateral location of the tire force center. When a tire is loaded more on the outside or inside, there becomes a radius arm trying to steer the wheel when applying torque. Happens in real life too, but ours is exaggerated. It explains the rather crazy FFB of the hillclimb cars as they have a crazy amount of power, wide tires and wide track. By the way I don't think this effect is even modeled in racing sims. In testing I've been able to solve this in certain ways with experimental tires but not in a practical way for release.

    Another issue is rolling resistance. Because the nodes repeatedly touch the ground as the tire rolls, there are some extra losses. We minimized this value a lot from around 25x reality in 2013 to around 2-4x. This impacts our top speeds a bit, and I think makes the handling suffer a bit at high speed because the tires are working harder to overcome that resistance. 4x sounds like a lot but we are talking a coefficient of 0.04 or so. It's not the primary contributor to top speed, air drag is. And honestly I wouldn't be suprised if racing tires have more rolling resistance than this anyways, though I have no data on that. I have succeeded in getting to matched rolling resistance but it tends to cause problems with handling. Recently I found that more complex tires improve the situation as the tire becomes rounder, but going this direction has some tradeoffs too.

    Well there you have it for now :) Both of the remaining issues I can see solutions for in the future with higher frequency and/or more tire resolution, but I still work on a regular basis to find a more practical solution for the now. But given the progress made so far I am actually amazed. I'm not the one who dreamed up these physics, I came in as an engineer to work on tires, and find that the physical model is already teaching me things, and leading me into papers about tire behaviors well beyond the typical literature.
     
    • Like Like x 35
    • Informative Informative x 4
  9. YellowRusty

    YellowRusty
    Expand Collapse

    Joined:
    Nov 9, 2016
    Messages:
    1,202
    If I'm completely honest, I'm very satisfied with the existing tire model, with a single exception. Cars keep grip until they shouldn't on dry asphalt, that point changes on wet asphalt, heavy-duty tires and bias-ply tires act differently than other tire types. It's all very well-put together.

    screenshot_00544.png

    But the way a car moves through deep sand or mud doesn't really look or feel very natural. Of course, some of it's due to the lack o terrain deformation, (and that's okay - Beamng is not Spintires, and adding that in would probably require completely rebuilding the game from the ground up. It's unfeasible), but some of it comes down to the fact that it's being controlled by a behind-the-scenes equation and not by allowing the Jbeam structure to react to conditions.

    Now, that's all fine and good - to a point. This equation works perfectly well for describing tires with fairly shallow tread or no tread at all. Slicks, low-profile tires, street tires, rally car tires, and even light-duty all terrain tires can all be described quite well via this system. Granted, that's the system that existed about 18 months ago, but I assume that it's very similar to the system in use today.

    The trouble is that it doesn't appear to account for the "paddling effect" of tires with very deep tread. As these tires sink into a soft surface like mud, loose dirt, or sand, the lugs sink in and grip with their edges, providing a surface for the vehicle to push off of. In soft terrain, it's not the portion of the tread that's tangential to the circumference of the tire that does the gripping, it's the surfaces perpendicular to the circumference. Probably the most exaggerated example would be the classic "tractor tire":




    Come to think of it, there aren't any actual mud tires in the game at the moment. Below is a picture of an All-Terrain tire (left) and a Mud-terrain tire (right). Note the wildly different tread styles, and which one bears stronger resemblance to what is found in the stock game. Even mods that insert them are largely constrained to dialing up the tread coefficient.


    There is one notable exception though. @Darren9 managed to simulate the proper behaviour of a mud tire by physically jbeaming the paddles. The "deep mud scoops" weren't quite perfect, but they did show promise as a proof-of-concept. If we really want to simulate proper mud tires, we may need to move beyond Jbeam structures that are simple cylinders.

    Now, bear in mind that I don't know much about upcoming content when I say this, but i'm starting to feel like off-road content is being pushed aside in favor of on-road content. Mud doesn't have it's own unique sound yet, it's actually quite hard to find a mudpit on stock maps, and off-road vehicle options are largely limited to rally cars, Gavril's light-truck platform, and the Hopper. It's very hard to find off-road themed content in the microblogs. It's probably not intentional, but with so much of the development process shrouded in secrecy, it's really hard to tell if this area of motorsports is even on the team's radar.

    I'm not angry about it - I just want to know if this is something the team has been considering or researching, or if a decision has been made to hold off on strange circumstances like this until the existing tire model is finalized.
     
    #29 YellowRusty, Oct 26, 2018
    Last edited: Oct 27, 2018
    • Agree Agree x 7
    • Informative Informative x 2
  10. Goosah

    Goosah
    Expand Collapse
    Global Moderator
    BeamNG Team

    Joined:
    Aug 30, 2013
    Messages:
    790
    We have been thinking on some changes to the ground depth to act more solid, so the tire tends to start on the surface and dig down rather than being more like a dense/viscous fluid like it is now. That would allow us to tune the behavior in a more convincing way. We can't get into storing deformation in the ground, that is tough given the high frequency of the physics combined with the huge amount of data in the map surface. We had some ideas to address the paddle tire issue too, as we noticed the CRD Monster Truck couldn't move forward in the water by turning its huge tires despite floating on the surface. The treadCoef was more intended for balancing normal-ish tires. We'll have to see what we can accomplish :)
     
    • Like Like x 14
    • Informative Informative x 4
  11. VeyronEB

    VeyronEB
    Expand Collapse

    Joined:
    Aug 5, 2012
    Messages:
    1,537
    Thanks for the detailed write up, really helped to explain a lot of things I didn't know about before.

    So it looks like its safe to assume the the tyre structure itself isnt going to be a limitation or bottleneck with this? or is it going to be something that may need a more complex structure in the future to get it the "last 5-10%"?

    Its great to hear its not nearly as much of a limitation as I had first thought though and the values are really impressive in accuracy.
     
    • Like Like x 2
    • Agree Agree x 1
  12. YellowRusty

    YellowRusty
    Expand Collapse

    Joined:
    Nov 9, 2016
    Messages:
    1,202
    Hey, thanks for that. It really puts my mind at ease knowing that the paddle tire issue is something that's being experimented on. Even if it never makes it in, I'd be satisfied knowing that it was tried. :)

    As for deformation in the ground, that's not as much of a concern for me anyways - I know that's not feasible anyway given the huge amount of computing power being taken up by vehicles. Beamng isn't Spintires, and I don't expect it to be.
     
    • Agree Agree x 5
    • Like Like x 1
  13. fufsgfen

    fufsgfen
    Expand Collapse

    Joined:
    Jan 10, 2017
    Messages:
    6,782
    I let you think about this for a while, 4 core CPU:
    upload_2018-10-27_8-56-21.png

    Physics or graphics is causing you not to have 11 cars real time 60fps in game?

    Answer is graphics, way that side of engine currently works relies on single core of CPU, which is not enough so you hit the wall of graphics before physics, with gtx 1080 that becomes very apparent.

    Physics have been optimized to insanely incredible level already, I believe now they are working on graphics side of things to get those physics into practical use.

    Of course compared to any normal game, physics are demanding and sadly hardware is not quite there yet for all the wonderful things we might want from tire / ground interaction, but people are having bit of misconception about what really is killing their CPU in this game, it is not really the physics anymore, physics are same if you run on lowest without shadows and AA.

    However for one, terrain resolution is not really giving much help to terrain deformation, you would need some additional layer of depth map with incredible resolution to have it physics based. To have that in convincing detail you would need less than 10cm resolution, while 2k terrain with 1m resolution is taking all the 16GB ram I have to open in Blender, 3DS max refuses to open, so how you are even going to work on such map, so there indeed are limits of hardware even on that side of things.

    0.0005 seconds is time where CPU has to finish all physics calculations, you can fit only that much to such short time, that is all nodes and beams, forces etc.
    0.333 seconds is blink of an eye, around there, roughly, so how many more things they really can fit to 0.0005 seconds?

    How it is even possible to have physics running for all those 11 vehicles of my screenshot in less than 0.0005 seconds?

    Then I read how tires have multiple contact points and it just blows my mind totally.

    0.01 seconds is what I can comprehend, anything less than that is starting to be something totally impossible to point that how small that time is.

    5/10000s, I think my DSLR can do 1/2000s, your CPU does all those thousands of calculations in hugely shorter time.

    2000Hz does not sound much, right? :D
     
    • Agree Agree x 5
    • Like Like x 2
  14. YellowRusty

    YellowRusty
    Expand Collapse

    Joined:
    Nov 9, 2016
    Messages:
    1,202
    o_O

    Okay... I mentioned in both posts that ground deformation isn't something I'm expecting at all, given the hardware limitations and total ground-up code rewrite that it would likely require.

    Somehow two different people walked away with the impression that it was the main thing I wanted to see going forward. It is not. I understood before posting that BeamNG is not going to concern itself with simulating terrain deformation (barring some sort of colossal, unexpected accidental breakthrough).

    With that said, I do hope that some sort of visual effect of mud/water/snow splashing outwards as the vehicle moves through the medium will eventually make it into the game, and that the mud groundmodel will eventually get some kind of overhaul to account for the different densities of dry, thick mud and wet, soupy mud. I also wouldn't be opposed to having more uneven terrain below the surface of the mud - but that's more of a map creation thing than anything else and not at all the topic of this thread.

    The big takeaway from my earlier posts was really intended to be that the current tire model isn't designed for tires with deep tread (like the kind you would find on, say, monster trucks and heavy equipment), and that it might be something worthwhile to tackle at some point in the future.

    I do apologize for any confusion caused by my earlier postings.
     
  15. fufsgfen

    fufsgfen
    Expand Collapse

    Joined:
    Jan 10, 2017
    Messages:
    6,782
    I thought that I did agreed on you, except for bit what actually is hard part, it is graphics currently, physics are surprisingly light for how intense they are in this game.
     
  16. Michaelflat

    Michaelflat
    Expand Collapse

    Joined:
    Jul 10, 2014
    Messages:
    1,543
    more importantly, why does the game have physics at 2000Hz anyway? Why not at the monitors refresh? The only reason i can see it needing 2000Hz is to do nice slowmo and maybe more accuracy.
     
  17. Ai'Torror

    Ai'Torror
    Expand Collapse
    BeamNG Team

    Joined:
    Aug 29, 2015
    Messages:
    1,547
    I'd guess it is for accuracy, since at 75 Hz stuff like suspension and tires could misbehave and you could end up with nodes of vehicles going though the cotris of the other vehicle if nodes were going fast enough to travel half of the thickness of the cotri during one refresh which is possible.
     
  18. fufsgfen

    fufsgfen
    Expand Collapse

    Joined:
    Jan 10, 2017
    Messages:
    6,782
    Well, 30 meters per second, that is 3000cm per second, which is 1.5cm per one physics simulation tick.

    At 60Hz that would be 50cm per each tick, you can always interpolate and do prediction math, but have you ever crashed a car at 220kph in BeamNG? That would be 100cm per each tick of physics @ 60Hz.

    So at 60Hz, instead of hitting rear bumper, you hit passenger's seat or something like that.

    Might get bit sticky, perhaps :p
     
    • Like Like x 1
    • Agree Agree x 1
  19. Michaelflat

    Michaelflat
    Expand Collapse

    Joined:
    Jul 10, 2014
    Messages:
    1,543
    oh yeah i forget how complicated it is :p I completely forgot about collisions, how in a game like this idk :p
     
    • Like Like x 1
  20. fufsgfen

    fufsgfen
    Expand Collapse

    Joined:
    Jan 10, 2017
    Messages:
    6,782
    I did set game to 120fps and boy does the FFB beat FFB of Assetto Corsa like hands down, so much better feel.

    I set hotlapping, then reversed and did go for a rolling start, managed to drive 8:29 lap, but that is whole full track of unknown scale and quality. Tourist track is some close to 1 minute shorter than historic, I think?

    Then I did lap in E190 EVO II race car in AC, except it stopped counting lap way before I reached full lap, then I even hit some barrier which appeared middle of road out of nowhere while I was doing close to 200kph. time was 7:49 or something that I already forgot and forgot to take a screenshot, but track is definitely different times can't be compared at all. Carousel is which I blew totally, almost ended up to wall there.
    upload_2018-10-27_12-28-32.png

    Way driving in AC feels for me is like if you would lean to stairs of G-force, one can almost feel those steps.
    It is really dull at low g-forces, it makes you feel like a champion driver at moderate g-forces and it is really weirdly static at excessive g-forces, but incredible easy to drift:
    upload_2018-10-27_12-38-0.png

    In BeamNG grip feels like it would be without steps, but it is much less forgiving, as there are forces pushing you to all kind of directions so you can't learn to lean into step like in other sims, but you have to drive the mass, it feels more natural to drive, but maybe that 10-15% could be what improves bit of that bite that was reduced some versions ago when loss of grip by increase of mass was introduced.

    I think something of that is where could be room for improvement, tires feel bit plasticy, like very cheap tires have that wonderful ability not to bite at rain, but here I think when loading tire by weight transfer, or bump etc. effect of bite is what could be perhaps bit more.

    It is very subtle balancing or finetuning that probably is needed, but will there be data to support that, I have no idea. I worry that some scientific tests don't account enough variables and bite effect might not be perfectly shown in data if testing is done using static force applied instead of natural dynamic behavior of tire in vehicle, but then again I don't know much about such stuff anyway, so I better shut up.

    I did put Clockwise 533 tires to EtkiEvo, bit more tuning too, but I feel that might want it bit stiffer, but I like how it is possible to dance that thing around the track.
     

    Attached Files:

    • Informative Informative x 2
  1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.
    Dismiss Notice