Tire Testing - Strange, but interesting results...

[calan_svc]

They just take real life numbers and gave the game cars the equivalent stock tires, without any regards to all the previously mentioned factors (contact patch, side wall stiffness etc).

I don't even think they did that; I think they just guessed. I can't remember now, but what tires do the 2010 Camaro SS and ZR1 come with stock in GT5? I don't think it is CM and SM though.

I'm not claiming anything here guys...this is just some interesting stuff (to me) that I'm playing with, in the hopes of tuning my cars to be as close as possible to IRL. Take it for what it is.

 

[Exorcet]

I've only glanced over the topic, but for a while now (since the beginning really), grip levels in GT have depended only on the tire. I'm not at all surprised that a GT ZR1 cannot turn any tighter than a GT Camaro. I'm pretty sure that in the Challenger seasonal event, my 1970 Challenger was pulling 1 g on the slow S-bend at the end of the track.

I don't think it has anything to do with contact patch, as cars with bigger tires are generally easier to launch off the line (great example is Challenger vs RM Challenger). I think the issue is that GT ignore suspension settings when factoring in grip. The suspension changes how your car is balanced and how it handles, but the extra normal force from stiffer springs and lower CoG is completely ignored. It's a major flaw and it should be fixed.

I don't even think they did that; I think they just guessed. I can't remember now, but what tires do the 2010 Camaro SS and ZR1 come with stock in GT5? I don't think it is CM and SM though.

I'm not claiming anything here guys...this is just some interesting stuff (to me) that I'm playing with, in the hopes of tuning my cars to be as close as possible to IRL. Take it for what it is.

It's a stretch to say they guessed. The 1970 Challenger came with Sports Hard. The 2008 Mustang came with Comfort Soft.

 

[hyyperactive]

I think it just indicates that you are able to drive a faster lap on a game, rather than in real life where money and life are at risk.

While it is easier to be faster in a game I don't think money and life at risk make any real impact tbh.. You can see people going off in big crashes in various low level motorsport events all the time just because they are pushing too hard.

Also while tires can easily make a big difference in lap times so can drivetrain coefficients, aerodynamic drag and downforce (or lift in most cars). So it can be many things that make the cars go faster and be easier to drive in gt5. Plus gt5 isn't even meant to be a proper simulator. More like a fun good looking game that sells millions

 

[CoolColJ]

Well the average/peak G numbers seem to vary from source to source

According to the InsideLine videos on Youtube, the ZR-1 only does 1.06G average, and the GT-R Spec V 1.12G etc

You also have to factor in the suspension setting in GT5 don't correspond to real life, no camber, lots of positive rear toe to create under steer, anti roll bars that are even front and back etc, and the cars don't roll as much as they do in real life. Which may mean they're too stiff.

In any case it backs up with what I feel. Sports medium seem to be what a high performance street tyre in the super cars should be in GT5. Sports hard just feels too much like on ice type feeling

 

[Griffith500]

GT5 had a section in the manual that stated which tyre compound gave the best approximation of the stock-car's OEM rubber. Interestingly, they were all ranges of tyres - e.g. "Sports Hard - Sports Medium" (or whatever they were called) rather than a single grade.

So, the tyre compounds presented in GT5 are not necessarily comparable to real compounds, as we should expect, and are probably averages of some sort. This means that most cars will have incorrect grip, even if the closest compound is selected - that is, unless, other things are edited (such as the weight balance or other properties of the tyre and suspension). So a '70 Challenger on Sports Hards is probably no more incorrect than the '08 Camaro on Comfort Softs. I guess it just depends which way they "round" up or down.

I am interested to learn if tyre width truly does affect grip, though. The Supra RZ, for instance, has always suffered terrible traction compared to other cars - I'd narrowed it down to the tyre compound only multiplying a "base grip" level that was set per car, or that simply the Supra had narrower stock tyres than other cars - which, of course, cannot be changed.

Further to this, the geometry of the tyre has a profound effect on the contact patch size and shape over various loads, as does the suspension. For example, if the bulging of the sidewall (more prominent on "high-profile", tall tyres) is not accounted for, the contact patch will be artificially too small, hence giving lower grip - a stickier compound will be needed to compensate. Equally, tyre pressures will have inconsistent and incorrect effects on the grip available for much of the same reasons, probably why they (and the ability to specify tyre sizes) are absent in the game.

It might seem, then, that GT5's tyre model is still quite basic. But, looking at the competition, a "complex" tyre model isn't automatically better. I think doing tests on cars with differing geometry tyres, but similar real-life compounds, (as suggested with the Ferrari 599), and seeing what differences fall out would be a good next step. 👍

 

[calan_svc]

In any case it backs up with what I feel. Sports medium seem to be what a high performance street tyre in the super cars should be in GT5. Sports hard just feels too much like on ice type feeling

But it varies wildly from car to car. With sports medium on the stock Camaro, I'm pulling numbers that are way too grippy...at least in this one aspect and with the limited conditions we have to work with. It performs identically to the ZR1 when they are both on the same tires, which is just not right. My ultimate goal is to find which set of tires provides the most realistic experience for each car (or at least the more popular ones).

You also have to factor in the suspension setting in GT5 don't correspond to real life, no camber, lots of positive rear toe to create under steer, anti roll bars that are even front and back etc, and the cars don't roll as much as they do in real life. Which may mean they're too stiff.

I was trying to ignore GT5's quirky suspension settings, but you bring up a good point. I think later tonight I may toss on the custom suspension and compare results to the first test. I'll probably just go with factory(ish) settings for camber and toe, and leave the other settings as is.

Further to this, the geometry of the tyre has a profound effect on the contact patch size and shape over various loads, as does the suspension. For example, if the bulging of the sidewall (more prominent on "high-profile", tall tyres) is not accounted for, the contact patch will be artificially too small, hence giving lower grip - a stickier compound will be needed to compensate. Equally, tyre pressures will have inconsistent and incorrect effects on the grip available for much of the same reasons, probably why they (and the ability to specify tyre sizes) are absent in the game.

Given everything else I've seen, I really doubt that GT5's tire model (and physics in general) is anywhere near this complex. My hunch is that Kaz and PD have stumbled across MANY approximations that give the illusion of a decent simulation of car behavior. Their target audience has obviously shifted away from racers and towards "gamers" even more than ever, so simple physics and making room for eye candy would fall within those goals I think.

 

[zk523]

Here's where it starts getting weird. The real-life Camaro comes equipped with Pirelli P Zero tires, and the ZR1 comes with Michelin Pilot Sport PS2 ZP tires. According to TireRack.com data sheets, both of these tires have identical speed rating (186+mph), tread wear (220), and traction rating (AA).

I dont have time at the moment to read all the replies, but I read your full post and glanced through the replies and didnt see anyone mention this. You are comparing tires from two different manufacturers and as such, everything outside of the speed rating is not comparable (so to speak).

Tread wear and traction ratings are only useful when comparing between tires of the same manufacturer.

Btw...I looked at getting PS2s for my car, but they're still too damn expensive and don't last that long. It's the price you pay, I suppose...

 

[CoolColJ]

Problem is custom suspension in GT5 lowers the car too, and you have no idea what the difference is vs the stock settings, ride height wise, unlike in past GT games....

Well you can put all the super cars on SM and they feel about right. Even a 370z will snap oversteer on SS, but others like the Mustang GT won't, so it's a bit puzzling in this regard

 

[RWB1213]

I find it a little disturbing that the numbers reveal an almost linear increase in grip. Nature and physics live in curves, not lines.

 

[calan_svc]

You are comparing tires from two different manufacturers and as such, everything outside of the speed rating is not comparable (so to speak).

Tread wear and traction ratings are only useful when comparing between tires of the same manufacturer.

Good point, and point taken. But they should be close enough to be in the same ballpark, especially as far as GT5's tire performance is concerned. At the very least, I don't think they are as far apart IRL as a CM and SM tire should be in the game.

I find it a little disturbing that the numbers reveal an almost linear increase in grip. Nature and physics live in curves, not lines.

Yep. That's why I think the tire type is nothing more than a simple grip multiplier...but I need to test more cars.

Here's my revised testing plan:

***********

1. Pick a car with a known skidpad value and factory tire type.

2. Test with CH tires (as described in the first post) and record the g value.

3. Test with RS tires and record the g value.

4. Find the scaling increment by subtracting the CH value from the RS, and divide by 8. Add this to the first value and each successive one to get a linear scale across all tire types. (I use an Excel sheet for it). In the case of my Camaro, the increment would be .06 and the table would look like this:


Comfort Hard CH 0.85
Comfort Medium CM 0.91
Comfort Soft CS 0.98
Sports Hard SH 1.04
Sports Medium SM 1.10
Sports Soft SS 1.16
Racing Hard RH 1.23
Racing Medium RM 1.29
Racing Soft RS 1.35


5. Find the closest g value that matches the real life number for that car, and equip those tires.

6. Run another test with the newly selected tires and see how closely the resulting g value matches the IRL value.

***********

The '02 Mini Cooper, Challenger SRT8 ‘08, and Veyron will be good test subjects, as they have varying skidpad numbers and good data from R&T. I still need to figure out how to normalize the suspension issue somehow.

This may turn out to be nothing, or it may follow a surprising pattern; either way it will be interesting. Someone feel free to test them also if you are bored, as it will be a while before I get time to do it.

 

[CoolColJ]

In Prologue there were bigger gaps between the 3 major types, and smaller gaps between the 3 variants in each group

Not sure if this is still the case in GT5.

 

[JTSnooks]

Good point, and point taken. But they should be close enough to be in the same ballpark, especially as far as GT5's tire performance is concerned. At the very least, I don't think they are as far apart IRL as a CM and SM tire should be in the game.



Yep. That's why I think the tire type is nothing more than a simple grip multiplier...but I need to test more cars.

Here's my revised testing plan:

***********

1. Pick a car with a known skidpad value and factory tire type.

2. Test with CH tires (as described in the first post) and record the g value.

3. Test with RS tires and record the g value.

4. Find the scaling increment by subtracting the CH value from the RS, and divide by 8. Add this to the first value and each successive one to get a linear scale across all tire types. (I use an Excel sheet for it). In the case of my Camaro, the increment would be .06 and the table would look like this:


Comfort Hard CH 0.85
Comfort Medium CM 0.91
Comfort Soft CS 0.98
Sports Hard SH 1.04
Sports Medium SM 1.10
Sports Soft SS 1.16
Racing Hard RH 1.23
Racing Medium RM 1.29
Racing Soft RS 1.35


5. Find the closest g value that matches the real life number for that car, and equip those tires.

6. Run another test with the newly selected tires and see how closely the resulting g value matches the IRL value.

***********

The '02 Mini Cooper, Challenger SRT8 ‘08, and Veyron will be good test subjects, as they have varying skidpad numbers and good data from R&T. I still need to figure out how to normalize the suspension issue somehow.

This may turn out to be nothing, or it may follow a surprising pattern; either way it will be interesting. Someone feel free to test them also if you are bored, as it will be a while before I get time to do it.

As far as the suspensions go, I wouldn't touch them. PD has to have accurately simulated the suspensions to some extent, otherwise all cars of similar weight would handle the same. I believe they have the correct specs for the stock suspension, they just don't give you the true values when you look at the suspension settings. When you apply the sports/racing suspensions it then defaults to different settings which may or may not represent the true suspension characteristics.

 

[calan_svc]

I did some more testing last night, with even more interesting results. I'm starting to think we have all duped ourselves into believing GT5 physics are a lot more complex than they really are.

Note: On the suspension issue, I decided last night to leave it stock as JTSnooks posted above. My reasoning is that even though the suspension values may not be accurate, they should at least all be the same relative to each car. Also, spring rates and dampers will have more of an effect in a dynamic environment where the chassis loads are changing direction; in a skidpad environment the forces are fairly constant, and changes to the suspension should have a minimal impact on sustained lateral acceleration when compared to tire grip.

For some variety, I tested the 4WD Gallardo, the squishy Challenger SRT8, and the FF Mini '02, all in completely stock form. Guess what numbers I came up with?

Somewhat unbelievably, every one of these radically different cars showed the exact same results as the ZR1 and 2010 Camaro; all cars were very close to .85 on CH, 1.15 on SS, and 1.35 on RS. Yes, you read that correctly... a stock '02 Mini Cooper pulls the same lateral G on any tire type as a ZR1 Corvette does on that same tire!

Thinking that this couldn't be right or that I was doing something silly, I ran all the tests again and got the same exact results. I then decided to try a car that I knew for a fact should pull a low skidpad number; the 1971 Cuda. (big, heavy, narrow tires, and low-tech...should be low 80's IRL at best). What I found was even more interesting than all the previous cars being the same; this one follows the exact same incremental pattern, but pulls numbers that are .10g lower across the board. CH tires tested at .75, SS at 1.05, and RS at 1.25. After seeing this I tested both the Camaro and 'Cuda again to be sure, and got the same results.

EDIT: I tested the 'cuda again a couple nights later and got numbers closer to the rest; .80 ish on CH, and 1.1 or so on SS. So I'm chalking the first test of that car up to beer or fatigue.

So now it's starting to look like the tires are in fact just a linear grip multiplier, although the range may shift slightly for different cars. (And conversely, the range may be the same for radically different cars as well). This also means that putting one certain tire on a certain "style" of car (i.e. sport hard on "sports cars") isn't the best idea, based on lateral g measurements.

So for the cars I've tested so far, I get these tire recommendations for as close to IRL as I can get:

2010 Camaro SS - CM
ZR1 Corvette - SM
'02 Mini Cooper - CH
'08 Challenger SRT8 - CH
Gallardo - CS

Using these tires, each of the cars feel about the way I would expect; the ZR1 is grippier than the others, the Challenger is much less forgiving but can be controlled, and the Mini feels like... well a Mini.

EDIT: Actually, PD would need a tire with even less grip than CH to get to the Mini's real life skidpad value of .79.

Lots more testing to do!

 

[RWB1213]

Great work. I guess the only good news as the linear scale slides a bit depending on the car. It remains curious (and disconcerting) that the Mini and ZR1 have matching grip scales. I'm not giving PD a pass on this one, but tire/traction modelling is an extremely complex exercise. There are so many variables that PD clearly had to simplify the calculations, I guess the question is did they oversimplify? I'm really disappointed that tire temps and pressure don't play a bigger part in tuning your car in GT5. Tire pressure adjustments are one of the most common tuning adjustments real race teams make.

 

[VBR]

Exellent work, bravo! Very interesting stuff indeed.


👍

 

[Barso]

That is so disappointing!
Tyres play such a large part in racing and simulation.
It seems GT5 has made alot of sacrifices to achieve that 16 car races and 1080p resolution.
It seems forza 4 tyre model is the way for Kaz to look into.
The fact that tyres are modeled this way, it makes the game nothing more than a automobile pokemon.

 

[JJ72]

A rather sad outcome, although very much expected when you see the unreasonable wear rate vs compound softness, they should have approach at least one tire company and get some real data.

 

[calan_svc]

It remains curious (and disconcerting) that the Mini and ZR1 have matching grip scales.

Assuming everything else was equal, I could almost explain this away as a weight-to-contact patch relationship. The Mini weighs much less and has a smaller contact patch, which equates to the ZR1's heavier weight and larger contact patch. But given that, there is no way that the Mini should be able to pull the same lateral acceleration number (especially on racing tires), due to the ZR1's suspension if nothing else.

Also, due to the considerable difference in contact patch sizes I would expect the results to diverge as grip is increased, but they don't. The ZR1 should have much more grip on racing softs than the Mini, relative to their amounts of grip on slippery tires. I think this is more evidence that the contact patch size (tire width) isn't being considered at all in calculating the grip.

 

[calan_svc]

Just tested the Ferrari 599 and Enzo, with both cars pulling the following numbers:

CH - .80
SS - 1.10
RS - 1.30

So once again the pattern holds, with each tire type equating to roughly .05g of lateral grip. This would put both the 599 and the Enzo on SH to match IRL skidpad numbers.

 

[Bluez_Freak]

Sorry for my ignorance about g-forces in general but does this all mean each car's grip equation relates only to tyre type??
I mean given the same weight, two different cars with same tyres have exact same grip?

 

[CoolColJ]

Also another thing - the cars don't handle the same in offline practise vs online.

From what others have deduced it's to do with fuel weight, try it again in your private online lounge area and see. Fuel weight isn't taken into account of with offline practise.

And the fuel tank in GT5 cars is 100 litres, so it's a lot heavier than in real life

 

[Bluez_Freak]

Also another thing - the cars don't handle the same in offline practise vs online.

From what others have deduced it's to do with fuel weight, try it again in your private online lounge area and see. Fuel weight isn't taken into account of with offline practise.

And the fuel tank in GT5 cars is 100 litres, so it's a lot heavier than in real life

Well that's OK. I think lounge practice is more realistic (tyre/fuel simulation) and switching between cars is quicker anyway than offline practice.

Not to mention that a friend could pop-in for a quick race too!!

 

[the Interceptor]

I wonder what track one could use for braking distance testing.

If we can find a track which offers distinct marks with regular distances, we might be able to gather data on braking distances depending on tires as well as brake controller settings. Maybe a real life track where we can measure a specific distance via Google Earth?

 

[jmke]

I see no improvement between tyre modelling in GT4 vs GT5, while this might make the game less realistic, it does make it a better pick up and play game, which doesn't require a wheel to be playable

 

[CoolColJ]

I see no improvement between tyre modelling in GT4 vs GT5, while this might make the game less realistic, it does make it a better pick up and play game, which doesn't require a wheel to be playable

GT5 feels different with a wheel, and with a controller.
Control pad = lots of hidden driving "aids"
Wheel = your on your own apart from super ABS brakes that stabilise the car

With a pad it feels pretty arcadey - dive into a corner, jam the brakes hard, flick the tail out, full opposite lock, and drift it out with massive slip angles without requiring any finese and you won't even spin.

 

[Rage0329]

Just an idea, I wouldn't trust what Road and Track provide as concrete. Check other sources too, MotorTrend, Car and Driver and Edmunds Inside Line.

 

[Big Ron]

As far as the suspensions go, I wouldn't touch them. PD has to have accurately simulated the suspensions to some extent, otherwise all cars of similar weight would handle the same. I believe they have the correct specs for the stock suspension, they just don't give you the true values when you look at the suspension settings. When you apply the sports/racing suspensions it then defaults to different settings which may or may not represent the true suspension characteristics.

Signed. I also think, they used real car parameters to create the physics...or why a BMW M3 feels like you would expect in real life? Or the classic muscle cars are worthless on race tracks?

When you start to tune your car, different parameters are set to standard to fine tune it for your wishes. It doesn´t mean, that those are the parameters PD gave the cars physics.

 

[Khoin]

GT5 feels different with a wheel, and with a controller.
Control pad = lots of hidden driving "aids"
Wheel = your on your own apart from super ABS brakes that stabilise the car

With a pad it feels pretty arcadey - dive into a corner, jam the brakes hard, flick the tail out, full opposite lock, and drift it out with massive slip angles without requiring any finese and you won't even spin.

This is very true. There's even a quite large difference between using an old Driving Force vs a new(er) DFGT...

 

[jmke]

GT5 feels different with a wheel, and with a controller.

yes, otherwise the majority would never play GT5

it's just that they had to make sacrifices to make the game appeal to a large crowd; they could have gone all-sim, but that would've cause significant less sales

 

[Battman23]

While I really like what the OP is doing here to inform us and to take a look into how PD developed their tire simulation you can't realistically expect PD to model every single car's suspension individually. They probably did this for most premium cars and then selected the closes matching car with a few tweaks here and there.

They don't have enough manpower to test the cars like the iRacing or other simulators do. And you see this in the MUCH smaller selection of cars for the PC simulators. GT5 is the best console racing semi-simulation that we have and there is no need for you to be disappointed in the liberties and assumptions that the PD team has made during development. In my opinion they should have focused a bit less on these areas and more on the online and user interface experience anyway.

Keep up the work, OP! Looking forward to seeing your results.