Physics right or wrong?

[Frederic Schornstein]

Disclaimer:
First of all please don't make that a fanboy discussion. I always try to be objective and leave my personal preferences out of the way. I am always willing to accept, if something I don't like on emotional basis, is better than what I like on emotional basis.

Ok I hope I covered that part and this won't get into a "You are a fanboy of this and that thread".

Prequal:
Like most people I eagerly awaited the release of Assetto Corsa. I read about all the licensed content, loved the screenshots and everything. I didn't drive the tech preview, because I didn't want to get a false impression, because it was all still WIP. The day AC was released on Steam I bought it.
I started really testing it after the first update were the FFB was fixed.

I do really like the game and wanted it to be good and so far my only issue are the physics. Most people praise them and they get a lot of love and I do confess, that in external videos they look great.
The car dives in on the braking, you see all the weight transfer and everything. No other sim offers that!

Example:
But as soon as I drive it, it feels just so wrong to me. Just one example: I tested the BMW Z4 GT3 at Monza. I know the track very well and usually drive GT cars, so a perfect combination for me.
Despite all the massive understeer you get everywhere in AC even with setup tweaks, I wanted to point out something you can judge on a much more objective level. So I will talk about the braking distance.

Every racer knows how hard the first braking zone in Monza is. You have a low downforce car and arrive at 260+ kph and have to break down to 60-80kph. So if you look at a lot of GT3 onboards on youtube even with cars, that do have ABS and are a bit restricted due to BOP. They all break before the 200meter board.
That said, I can brake at the 300meter (270kph) board and have the car at a total stand still at 150meters. If I brake for the corner I can brake at or after the 150 meter board.
So we already have a 25% shorter braking zone compared to a real life car with equal or less power and ABS, which I didn't use.

Test conditions:
So now people would start arguing, that I maybe run more downforce and stuff and that is the weirdest thing. I tried to figure out the "worst" braking performance. I used hard tyres and removed all the downforce from the car for this test.

I don't want to get to much into the details of the cornering behaviour and the understeer and that you can just pull on tons of lock and don't get any turn-in oversteer.

Is my opinion qualified?!:
A lot of simracers don't have a lot of reallife experience and I didn't race a GT3 car in my life, but just a quick background:

I drove Race07, rFactor1 and now rFactor2 and I am a pretty good driver in rF2 especially in rear wheel drive touring or GT cars. Even in the new Civic I was racing for wins after 30mins on an unkonw track, so I can quickly adapt to new cars and tracks etc.

In my free time I did some kart races with friends and even on an unkown track with for me unkonw more powerful karts than the average rental karts on a bit cold track I got within 1,9 seconds of the track record within 15min and reduced it to 1,5 seconds in a further 15min.
Keeping in mind that real professional Kart drivers practice their and I had maybe 2 or 3 hours track time in karts ever, I would say I am not a bad driver.

So when I jump out of rF2 into a kart it just feels like home. I apply nearly the same technique and everything. When I jump into AC it all feels wrong. It is so hard to get wheelspin. You can turn the wheel so much, that you would end up in a wall in real life.

I also spoke to some guys with actual racing experience and they got the same feeling.

The end:
I really wanted AC to be a very good sim and I do love everything about it, but these physics keep me from driving it.
In a sim I don't want it to be easy, I want it to be as realistic as possible, but in AC you can apply some really bad driving technique to get quicker laptimes.

 

[Andrew Ford]

i love this "winning thread" things. Credit to Xavier, he is a winner in many ways because he was respectful and informative rather than rude and suggesting people shouldn't open threads to find out more info. Michael, I'm glad that you accept Xavier wins because he was much use to the thread rather than provocative.

 

[Frederic Schornstein]

I think we should close it here
@Xavier de Carvalho

 

[Matt Sentell]

Sorry Frederic, I know you called an end but you also brought my name into it.

I don't do forum wars anymore but since my name was mentioned and I've read a bunch of these AC vs <othersim> threads now, I do want to comment on something. I think it's a shame that none of these discussions ever go into the objectively observable differences between the approaches taken by the sim makers. In this case, thermal modeling of tires.

When you look at the modern, so called "next-gen" sims such as rF2, PCARS and iRacing's NTM, all of them have something in common when it comes to how they model tire temperatures. They've all moved beyond a single temperature layer. I'm not talking about the Inner, Middle, Outer temperatures across the width of the tire - that's been done for ages - but rather the distinction between surface temperatures, tread temperatures and carcass temperatures. They've all taken an approach of modeling the thermodynamic properties of these different components of a tire and, significantly, how the heat conducts between them.

Past sims were much more simplistic in this area. They modeled a single temperature layer which was usually considered to be a few millimeters into the tread, in an attempt to represent what people were accustomed to seeing as temperature measurements in real life when a car would return to the pits and have them measured by a pyrometer.

The problem is that with real tires the thermal effects on grip are occurring mostly as a result of the interaction between the tire's surface temperatures and the track temperature. And it turns out that the surface temps are changing quite rapidly, as can be seen in infrared tire monitoring such as this:

What isn't demonstrated in that clip or in any I can find is what happens when you really slide a tire and take it well beyond its slip peaks, but I think one can well imagine what happens just by observing how quickly the temps are changing during normal "limit" driving. I've never personally seen a test on a tire rig where this was examined, but I've been told by devs who have done them that the surface temperatures instantly skyrocket when the tire is taken to high slip angles. What constitutes "high" would vary by tire type, of course.

As near as I can tell, based on testing in the sim but moreso from Aris' comments on Kunos' forums, AC has not taken this next-gen approach to thermal modeling. It appears as if it continues to use a single layer of temperature modeling, and that in the case of their own content they've used the same sort of "several millimeters deep" temperature layer that they call the "core temp." By using the tire temp app, one can easily observe in the sim that the tire temps are not changing nearly as rapidly as in the video above. Aris does say that content can be set to model either the core temps or the surface temps, but it sounds as though it's a choice between one or the other. It would definitely be interesting to see how the sim would work if they used surface temps, but I suppose we'll have to wait for modded content to find out. But even in that event, if I'm correct in my understanding of their model, the complex effects that occur from the heat conducting between the other layers won't be present.

Those complex effects are mostly temporal (time-related) in nature. For instance if the carcass is very cool, as it would be on an outlap, then a momentary overheating of the surface - even if it's quite severe - will correct fairly rapidly. The surface temps will cool back down quickly, in other words. Conversely if you've been overdriving and sliding excessively for several minutes straight then the carcass will probably have overheated as well, and now even if you drive below the limit it will take quite a bit of time for the surface temps to recover into the desired range.

The net effect of all this is that currently AC's tires, in comparison to the other next-gen sims, are signficantly less thermally sensitive. It stands to reason then that they will be relatively more tolerant of overdriving, just as past sims were.

None of this has anything to do with the feeling of grip, is there too much grip, etc. There are many aspects of a tire model and thermal modeling is just one of them, although it's become increasingly apparent how important it is in recent years, which is why the other next-gen sim makers have tackled the complexity of it. And it's worth noting that it is indeed very complicated and difficult to model, in part because the understanding of what's happening in real tires is still incomplete. Some of the next-gen sims are closer to sorting it out than others.

As some of you might know I was with Papyrus in the 90's and directed GPL, among other things. I also did some real racing, and my greatest frustration (aside from Sierra's marketing group) was that overdriving in our sims was not only tolerated but often quicker than driving with good technique. At the time there were many theories as to why that might be. In recent years it's become fairly clear that overly simplified thermal modeling was one of the key reasons.

As I said, I don't do forum wars, so I've tried to stay to factual, observable information and not opine on anything. I will just restate that my assumptions about AC's thermal modeling are based on what I've observed in the sim combined with Aris' comments, and that they are consistent with one another. I see no observable evidence of complex thermal behavior in the sim, but that doesn't mean that it doesn't exist. Some will likely suggest that just because it isn't there now doesn't mean that it won't be added at some point. I would love to see that happen, although my dev experience tells me it's unlikely. Tire models being as complex as they are to develop, fundamental behavior like this tends to require integration at the foundational level. But of course anything is possible.

How much any of this matters to you, and if it matters at all, is for you to decide. I will just say that there is a lot of good stuff in AC and a lot that I like, some of which is better than, or non-existent in, any other racing sims. In fact there's one thing I love which we always talked about doing at Papyrus but still no one had managed until now - the physics-based sound effects of chassis and suspension squeaks, creaks and pops. It does wonders for immersion and adds to the character of different cars, and I've been disappointed at how little recognition it's received.

But perhaps it's because I know how complex this subject of tire modeling is that I find it a bit unfair that the efforts of the other sim makers in this area, and the potential importance of it, seems to go unnoticed. That said, I recognize that many will say they don't care and that this is a bunch of technobabble that makes no difference if the sim doesn't feel right. Believe me - 20+ years, I've heard all the arguments enough for one lifetime. As Marco Massarutto might say, this is intended for the people who like Ferrari's.

 

[Kristóf Kerekes]

Great post, from an expert who also has experience in developing games. The only thing I'd correct in the whole post is not every simplified-thermal modelling tolerate overdriving. For example in Live for Speed it is the simple type of tyremodel compared to rF2 and the other sims mentioned, yet overdriving the car will punish you hard by turning the tires into soap bricks in a matter of few turns, which is just... a bit overkill to what should happen imo. I'd compare AC's tire model with rFs currently. It tolerates high slip angles through every turn and doesn't punish you for it. Although I did not say they are the same! Before someone starts flaming, they are just similar, heat-wise.

 

[Stereo]

What I get from that, is that there are basically 2 parts to tire temperature.
1)

The tire's response to temperature - this is just made up numbers (inspired by this page plus Kunos staff mentioning the optimal ranges) normalized to a max of 100% ideal grip, because it's essentially impossible to find this data publically. At around 75-100C you'd have peak grip, and it drops off a little below that, and a lot above it (actually it's probably worse below 0 too). If you don't like this, the answer is simple - change the curve, and you can have a different response to temperatures. Don't know how much down you want it though; loss of 30% of grip seems like it should adequately impact cornering speed.

2) the part that's impossible to draw... how a tire temperature changes. Contact patch size, pressure, forces, air temperature etc. all contribute to it. And they affect each other too; temperature affects pressure, which changes contact patch size... This is where the sim authors have to work their magic. On this end all you can really do is say "at 1G cornering on a skidpad the tires should be hotter" or "tyres can't get hot enough - should be possible to melt them doing a burnout" and hope the magic has room for adjustment.

I'm not going to post inside info of AC, but both these components are present in some form in AC's car files; once modding tools are released it should be possible (maybe even easy?) to make the tyre model do what you think it should at least in broad terms. If you want slip angle/ratio to pour on the heat, you can do that. If you want tires with literally no grip at 150 degrees, sure. Then you just have to convince the biggest group of critics in the world that you're right. I think that's really the tricky part

 

[Matt Sentell]

I'm not going to post inside info of AC, but both these components are present in some form in AC's car files; once modding tools are released it should be possible (maybe even easy?) to make the tyre model do what you think it should at least in broad terms. If you want slip angle/ratio to pour on the heat, you can do that. If you want tires with literally no grip at 150 degrees, sure. Then you just have to convince the biggest group of critics in the world that you're right. I think that's really the tricky part

What you're describing is still a single layer system as opposed to the more complex multi-component systems. One could say, "well, why not just do it as a single-layer system but treat it as surface temperature, so it changes rapidly as in the infrared video?" Which I suppose is what you're suggesting.

This wouldn't produce realistic behavior because the temperatures would then fluctuate between the high levels caused by load, say 110C, and the ambient air temperature which could be something like 17C. That's far too great of a difference and would mean that if you've simulated the grip vs temperature to a fair degree of accuracy (sim makers do have this data from manufacturers as well as their own tire tests) you would have too little grip a lot of the time. In other words you wouldn't be able to get the tire up into what's called its "working range" and keep it there most of the time.

The thing that's really being simulated in these multi-component thermal systems is insulating properties. The surface of a tire can dissipate heat very rapidly into the atmosphere, whereas even a millimeter or two into the tread heat will dissipate much more slowly, and it does so mostly by going through the surface and into the atmosphere. The deeper you go the more insulative it becomes. Then you come to the carcass, which is made up of multiple components itself (in a real tire), some of which are metallic. This obviously has very different insulative properties to the tread block.

In simpler terms, the deeper you go into the tire the more heat is retained over a given span of time, and the heat is conducting throughout. If you want a simulated tire to have surface temperature response that mimics a real tire then there's really no other way to get it than to simulate, to some degree, these multiple layers.

 

[Andrew Ford]

We have a new winner lol

 

[Stereo]

Ah, I see what you mean...

Not really sure how many points the tire's temperature is being simulated at. The game is only reporting it at up to 3 points (across the profile) but that may just be a limitation of the app interface. Aris said over on the AC forums that those are at depth (so it's not reporting surface temp.) and that modders can use surface temp. if they want, but I don't know if that means every tire actually has both and only reports at one point, or just that the model is versatile enough that you can treat the single layer as being where you want it.

 

[David Dominguez]

Matt's posts regarding tire models are brilliant, it's hard to disagree with him. However, there is a huge "but"... Those tire models, as complex as they are, are extremely difficult to work with. And considering that AC is a game with modding being one of the cores of the game, probably that tire model approach makes more sense. Starting from the point that true tire carcass structure, thermal and chemical properties of the materials data will never be available for 99,9% of modders, it will b easier for them to work with a non physical based tire model. We can already see the problems of most modding groups with rFactor 2, in fact a lot of them gave up long time ago, and most of the current modders are just copy-pasting the tire files from some of the ISI's cars, hoping that they will somehow work.

 

[Joco Gorenc]

I agree, detalis like this simply dont play THAT much in the equation of grip and how a car handles. The differences between optimal grip and few degrees outside that are miniscule.

 

[Michael Hornbuckle]

EDIT: Nothing offensive now

 

[Austin Ogonoski]

iRacing and to a lesser extent rF2 still feel like inferior sims despite sporting a much more detailed tire model than AC.

I'm sorry but iR drives awfully, and rF2 is at least predictable but still has a ways to go.

And any sim where you can do this with the default setup loses a lot of credibility, no matter how complex the math behind the tires may be.

 

[DewCrew88]

The net effect of all this is that currently AC's tires, in comparison to the other next-gen sims, are signficantly less thermally sensitive. It stands to reason then that they will be relatively more tolerant of overdriving, just as past sims were.

As some of you might know I was with Papyrus in the 90's and directed GPL, among other things. I also did some real racing, and my greatest frustration (aside from Sierra's marketing group) was that overdriving in our sims was not only tolerated but often quicker than driving with good technique. At the time there were many theories as to why that might be. In recent years it's become fairly clear that overly simplified thermal modeling was one of the key reasons.

I disagree, overdriving on AC yields nothing for me but either being backwards or slower. Other than the fact that tire wear needs to be a bit higher, and we dont have brake fade (you can put the brakes to the floor every corner with no penalty) I dont feel that overdriving has made me any faster.

The sims with "more advanced" thermal sensitivity models love to make you slide everywhere and eventually out of control for the slightest slide of the rear tires. Watch any video from these sims, and the nervous twitchy feeling is just awful to watch just as it was to drive when I still played. It isnt more realistic just becuase it is more difficult. Infact the sims with "more advanced" thermal sensitivity are also notorious for extreme over driving being the fast way around.

 

[InsaneOzzie]

Ive said it before, and I'll say it again.
There is, and will never be a simulation either racing or flight that will EVER replicate what occurs in reality. This is why they are called simulations.

The fact is, we are all individuals and all interpret our experiences differently, what I feel in one sim may not be the same feel in another, take rf1 and gsc for example, same core engine, Gmotor2, but because gsc has been polished and tweaked it rates far higher in the simulator stakes for me than it's older brother rF1, yet they are the same game engine.

There are so many variables associated with this topic the mind boggles, in the end the OP will draw his own conclusions, based on the information or lack of, as we all will/have, as to wether AC is an accurate simulation of racing reality which I feel Xavier de Carvalho summed up in a nut shell, well done, and my thoughts exactly.

Enough said!!

 

[Matt Sentell]

Infact the sims with "more advanced" thermal sensitivity are also notorious for extreme over driving being the fast way around.

I have never seen a fast lap in rF2 achieved through overdriving. I'm usually near the top of the time sheets and I don't do it by overdriving, nor do I see anyone as quick as me overdriving.

iRacing, yes.

 

[Matt Sentell]

I agree, detalis like this simply dont play THAT much in the equation of grip and how a car handles. The differences between optimal grip and few degrees outside that are miniscule.

When you get a big slide on and temps go to 160C or more it has a very meaningful effect on grip, I assure you.

 

[Joco Gorenc]

Sure, but no need for a billion calculations to do that right? It seems to be working fine at this stage in AC, but one thing I find strange is that the tires heat up so quickly when going straight, and once at 100C they dont cool off as well...

 

[Michael Hornbuckle]

EDIT: Nothing offensive now

 

[Frederic Schornstein]

So after 7 pages of "discussion" how many people have quit playing AC and went to rf2 and vice versa?

Was never the point of the discussion. If you dont mind, please stop trolling around.

@Xavier de Carvalho If you dont mind, please close the thread before trolls start ruin a nice discussion, which contains valueable insights for every simracer

 

[Michael Hornbuckle]

EDIT: Nothing offensive now