Messages

I understand what you say but in real life, especially with rear brakes only karts you must lock the wheels then quickly release the pressure to grab the best braking force. If you try to do that in KRP you'll spin all of a sudden.

I've done this in many karts from rental to DD2 as both front brake or rear brake rotaxes, from vega white to rock hard tires and i never spun all of a sudden. I think it was already mentioned by some people.

I have low pressure braking. But even in straight line if you lock the wheels you kart spins without warning.

The front end geometry may worsen for sure, but even in a balanced kart it still spins all of suden.

The handling is much better in beta 9! Much more enjoyable. THe braking is still not correct i think. the kart spins for no reason when you reach the locking point. That the rear axle becomes lights is okay but that you enter an uncontrolable spin is not. But overall much kudos to piboso! this is real pleasure!

Yeah i actually read that yesterday while browsing serveral research paper. This is very surprising..But maybe this is your chanceto be part of a new dawn in karting, try a set up a numerical approach to karting. You wouldn't be the only interested, for example university of turino and torvergata group did several numerical studies on both aero and chassis for karts.
They are now doing some FSAE project but i'm in touch with torvergata because they are very interested in aerodynamics downforce prospects for karts and research in general in those field.

...a ready made simulator could be a good tool for them to test their model!


Karting is fantastic, but in my opinion too much stuck in the past.

Actually this is the same problem in aviation really. Especially in military simulators. You can't get even the dimensions of a landing gear; Several plane manufacturers forced some amateurs to remove their planes from their simulators (sic!) and i can directly relate one case of a student that tried to do a school work on estimation of performance of one fighter plane's radar...he got the unpleasant visit of french rens services!
If you dare to ask for a CFD model of fighter jet my friend you're in deep trouble!

This is worsened by the fact modern fight simulators simulate in real time most of the systems contained in the planes (from tire's braking elastic limits to dials in the cockpit) thus the need for information is huge.
The only possibility is thus: either concentrate on older planes (but not too old) like WWII (many private collections fly them) or have a partnership with the constructor which is basically: You provide them professional services (You develop a version of your simulator for them) and they let you develop a public version with their plane giving you data.

In military flight simulators the most known company is Eagle dynamics from russia, they develop hardcore study sims. for example their DCS A-10C was developped because they developped the military version of their simulators that is used in USAF from recrutee training.
In car racing i've understood that kunos simulazioni survived by the exact same means and that assetto corsa will have lot's of licence cars because of their links with the professional industry.

Each time the partnership was done because of the physics engine. Professionals don't care about data, they have hundreds and hundreds of pages filled with coefficients measures in testing, what they want is a mathematically solid engine that because it is aimed a public is cheap...so they can buy it without having to invest millions asking a specialized company to do so.


IMHO as a simmer i don't really care if one kart is like the real counterpart. What i do care is that physics are respected, that is if i change to torsional stiffness somewhere it produces the same results. Now that a kart goes 100km/h in a sim while doing 110 in real is not the first priority for me.

About the understeer problem: I start to see a bit clearer now in kart dynamics (after reading many technical papers) and i tweak the set up and really my karts in KRP are not understeering that much and if the leaderboard is up to date, i'm far from being slow on the KRP tracks. However it is true that my approach is to recognize what is the problem (what the kart is doing) and apply the science i learned on the set up.

Now to resolve the problem it is needed to get into the detail of your physic engine to try to look out where is the problem; If you need my help i can only offer a "not too newbie" opinion because of my experience with physics engines and i'm not too afraid of maths and physics neither; I may not be able to pinpoint what the problem is but maybe i can give some ideas;
And then better real life drivers than me can help giving their feedback; But all of this require strict (and sometime boring) tests procedures! We're doing it flight simulators, it is not just a matter of taking the virtual kart out for some hotlap, we would need to test the same maneuver over and over, peharps a skid pad would be needed etc...

That is if you need my help! i don't even know your background and how KRP is done, except that i think it is a great thing to do a kart simulator and i feel KRP has strong potential.

So keep up the good work!

Well for example in WWI simulations clearly no, you don't have that much of experimental data since most of the data used nowadays was not used then. That's why the pre-render phase is very important. You need to get those data by calculations.

But i feel for you, it seems kart is a very secretive world!! Anyway to me while i'm looking for improvement in KRP i try to take the challenge as it is now until those changes happen and i'm having fun with it. I just try to see what works and what doesn't.

Keep up the good work!

Hi,

I understand your problem, the simulator i've been the most involved in is Rise of Flight (a WWI flight simulator). The lead physics engineer like often now in military simulators is a real flight dynamics engineer so he know what he's doing. And it still takes all of his scientific rigor to not break under public pressure to do this and that. After 3 years of existence this has proved successful but he has to constantly work that way:

He works on pre-rendered mathematical model to get precise data, then he feeds them into his more simplified real time model; When he can (with CPU power usage limits) he develops more complicated real time physical model. This is a constant evolution process.


As for set up, the main problem was highlighted in high speed corners like the last one before straight at Essay or at Escourse. Initially i had understeer at turn in, i modified the front to have more weight jacking but i also steer initially abruptly for very few degrees (that i do it in real life too) before being smooth on the steering wheel to be sure you have enough load transfert. After that the problem was an understeer on steady state that i resorbed by having a flex at the front but not a the rear. My reasoning is that you need flexing chassis to keep to real wheel lifted but either if the rear is flexible too the rear wheel may lift too much (and then you have understeer again) or..by making the rear flexing you may have losses by friction of the reactive force that makes the rear wheel lifted.

Here you have a very good example of a driver observation that is without sound scientific ground...I can't simply be sure of why this set up works....and from what i've read on the internet, in real life, nobody seems to understand that body flexing neither!!!

That's why a sound physic model (forced based, with viscosity and non solid body equations) would be itself provide the correct effects. The more you simulate and the base level the more you'll likely to get it right (but the more source data you need...)

Hello Piboso,

I' joined KRP just some weeks ago so i don't know the other builds; I've indeed seen some strange behaviors compared to real life and my initial feelings with the base setups were that understeer (especially on long bends) was too much (you really had to fight a lot). With set up changes it is much better now.

I'm not enough qualified to say where you have it right where you have it wrong however coming from other simulators fraternity (mainly military air planes simulators) and having been involved in alpha & beta testing with some teams i feel i have one good suggestion for you.

do NOT rely on observations to TWEAK your physics engine!! This is the best way to f^*ù up your simulator!! Why?

1/ People will tell you "this kart should oversteer there!" but why their kart oversteer there?? there could be a lot of reasons!
2/ People will sometime feel something that in reality has contrary manifestations for example "i oversteer there!" but in reality you can a fraction of second understeer; Sometime the driver has it completely wrong.
3/ People will invevitably say you that "a kart should do this in those conditions" and there they start to make a law of their own experience whithout knowing what going on.
4/You'll run the risk of tweaking some parameters that will have influence on the others and that will spoil the other points.

All of those points are called "the Hollywood effect" in military flight simulators I.E: people have a wrong idea of how a plane should behave in this or this situation.

My suggestion is that you should improve you mathematical model or tweak it in the basic parameters. If your physic model is good from a mathematical and source data point of view then when someone says that your physics are wrong, you can discuss with the person to see if he has a sound scientific approach or not and you can better see what it going on.

My post isn't there to say that the complains are not valid, and i'm sure you already know all of this but following the posts i'm under the impression that many people want some "sensations" back and  when you're under pressure to change your physics you run at risk to make fast tweaks that will please people because their driving technic will work but where everything is messed up in other department. Then another will say "i drive like that in real life! I'm a champion and it doesn't work in KRP!!!"..

Bai bai