GT5 physics model - how complex is it?

[sinbad]

I understand why downforce/drag is an issue, because it's something people notice, but for me on this topic, aerodynamics are a much smaller part.
When I look at a car game, the two things I judge the realism on are, tyres and suspension.

GT5 is a bit puzzling in that some cars are convincing while others are not. I find that suspension often appears too firm when the car is changing direction or speed, and yet too compliant when passing over bumps/jumps. Commenting on the tyre model is more tricky, but there is something "soft" about them. You don't get cars hopping as tyres struggle to regain grip following a greedy moment over a kerb.

In general I think it's pretty good. I just wish the guy responsible for the good cars (I'm guessing that one guy is some super enthusiastic physics expert or something ) had been signing off all the others too.

 

[robbo6]

I suppose GT is a bit far removed from reality in that sense, then. However, in that respect, I guess the real emphasis is on the progress; in being able to gradually amass more money and progressively afford "better" cars, akin to the item / abilities / gear "leveling" in most modern RPGs (which I suppose is where the "GT is an RPG" argument comes from).

In the same way that PC sims are a vehicle for experiencing the particular motorsport niche of interest, perhaps GT is more a vehicle for "experiencing" the individual cars in a range of motoring situations. Which is why the physics are so important, and also why, I feel, GT excels at the characterisation of cars (accurately or otherwise) and translates the virtual idiosyncrasies of fundamentally similar cars - an example immediately in my mind from a few shuffle races recntly: I drove a Celica SS-II (T230), a Clio Sport 2.0 (well, "Lutécia") and an EP Type-R on Eiger back-to-back, and each one was very distinct in the way it handled the different speed corners, down to the steering response and the way the engine "felt" running through the gears - you know, how they drove.

Again, I can't comment on how accurate the experience was, but I think characterisation itself must be heralded as quite an achievement.
I will concede the physics have their flaws, gaping holes even at times, and the racing framework is sub-par compared with other games, but those other games don't have the character or range that GT has, so it really is a case of swings and roundabouts.

I'm afraid I went off on a tangent, I was only having a go at the "driving simulator" comment, but in the context of the discussion I feel I shouldn't have. Physics wise I agree with what you say 100%. There are so many cars in GT5 that 100% accurate physics is impossible. I will also say that as a "driving experience" it's quite excellent....I'm just someone who wants more of a "racing experience" too and I tend to jump on the tagline quoters....but in this thread it actually made sense....my apologies to Dodzzz.

classic case of bringing a different argument into a thread where there's one common link, but in the context of one discussion is perfectly valid.

Pretty much a double facepalm and d'oh moment for me in retrospect!

 

[Exorcet]

That is how they are do going by this video:

It seems like you don't really understand how programming works, but that's OK.

In a program like GT5, you have a physics engine, and then you have parameters. The engine is just a big calculator. The parameters are what go into the calculator. In this respect, every sim in the world is identical. They all do this. GT5 should be able to handle customization as well as anything else. If it doesn't, then it's inaccurate in some areas.

The video confirms this.

 

[CSLACR]

Thats some seriously nice work there 👍

What I was moreso getting at however was when you did the speed test, you werent comparing the gear ratios, so a full max speed test wont quite add up regardless. In saying that, im not saying the physics model is faultless, because I could almost garantee that having more down force on Le Sarthe will be slower in real life compared to a lower downforce setup; something that is reversed in GT5.

IRL, C5R's were rumored to be aerodynamically limited to 165-170mph. Presuming that's a high downforce setup, a high speed setup may turn as high as 200mph, which is a 35mph difference between min and max aero, theoretically.
In GT5 both min and max aero C5R setups can both reach a top speed of less then 5mph difference.
Real world: 35mph difference.
GT5: Less then 5mph difference.

See the problem?
Gear ratios have nothing to do with either of these tests, when performed properly.

 

[Ronald6]

There are so many cars in GT5 that 100% accurate physics is impossible.

I always believed that the wide range of cars that GT provides increases the accuracy of the physics engine. In order for all of the cars to be modeled correctly the engine must be able to handle all the various variables.

I always pictured a programmer leaning over a testers desk saying, "Show me."...a 2CV laps..."$&!+, looks like we need to tweak the engine again."

 

[Exorcet]

The greatest F1 car ever never built

Two set ups

"Monaco"

2 element front wing
2 element rear wing (fixed main element, high AoA secondary element)​

"Monza"

1 element front wing
2 element rear wing (fixed main element, low AoA secondary element)​

Data break down
Total - Total force on car
F - Front wing
R - Rear wing

X - Side force, ignore
Y - lift (negative is downforce)
Z - drag (negative is drag)

Vortex wake - shows induced drag as air kinectic energy.

A note : clearly, my rear wing was ineffective, I probably need to make the second element bigger, or adjust the AoA of the main element. Either way it shows that drag does up faster than lift, as it must.



"Monaco"
Total
Integral Parameter Value X-component Y-component Z-component Surface Area [ft^2]
Force [lbf] 1036.14 -1.43198 -675.482 -785.693 295.94
F
Integral Parameter Value X-component Y-component Z-component Surface Area [ft^2]
Force [lbf] 600.961 0.00597609 -590.225 -113.086 15.9187
R
Integral Parameter Value X-component Y-component Z-component Surface Area [ft^2]
Force [lbf] 333.933 0.311238 -321.63 -89.8055 9.39765

Vortex wake

TOP SPEED (700 hp) - 213 mph


"Monza"
Total
Integral Parameter Value X-component Y-component Z-component Surface Area [ft^2]
Force [lbf] 831.906 -1.49984 -421.721 -717.089 291.856
F
Integral Parameter Value X-component Y-component Z-component Surface Area [ft^2]
Force [lbf] 308.696 -0.115179 -306.964 -32.6561 12.5432
R
Integral Parameter Value X-component Y-component Z-component Surface Area [ft^2]
Force [lbf] 256.635 0.13035 -254.805 -30.5943 9.46782

Vortex wake

TOP SPEED (700 hp) - 220 mph






A 3% top speed reduction, mainly from the front wing alone (which might be misleading because drag on the front wing can manifest itself as reduced drag on the body. The rear wing is pure loss when it comes to drag).



EDIT 11/12/11 - Top Speed was off, I had just thought it was due to low mesh accuracy in the CFD, but it turns out that I forgot to vary the drag when I was comparing it to engine power output. The tops speeds are now corrected.

Coincidentally, the top speed change from t he front wing is just under the change from front and rear predicted in GT5.

 

[Griffith500]

I'm afraid I went off on a tangent, I was only having a go at the "driving simulator" comment, but in the context of the discussion I feel I shouldn't have. Physics wise I agree with what you say 100%. There are so many cars in GT5 that 100% accurate physics is impossible. I will also say that as a "driving experience" it's quite excellent....I'm just someone who wants more of a "racing experience" too and I tend to jump on the tagline quoters....but in this thread it actually made sense....my apologies to Dodzzz.

classic case of bringing a different argument into a thread where there's one common link, but in the context of one discussion is perfectly valid.

Pretty much a double facepalm and d'oh moment for me in retrospect!

You were right about the requirement to race in GT games, and I agree totally about not latching onto a daft tagline.
My initial response was, I think, targeting the laziness of people (myself included) in learning about what games are like - we want to know in as few words (or numbers) as possible how "good" a game is, and whether it's worth our money. It's the same with books and film. People always say, "oh it's like such and such" or they try to classify things by genre, and really it's almost pointless for the individual. Game reviews are a great way to learn about games, but it's important to get a good cross section.
This all relates to the tagline "the real driving simulator", because it doesn't really tell you anything about the game. It's simply designed to sell the damned thing.

So yeah, it's meaningless. 👍

 

[kimi123]

It seems like you don't really understand how programming works, but that's OK.

In a program like GT5, you have a physics engine, and then you have parameters. The engine is just a big calculator. The parameters are what go into the calculator. In this respect, every sim in the world is identical. They all do this. GT5 should be able to handle customization as well as anything else. If it doesn't, then it's inaccurate in some areas.

The video confirms this.

I am saying the game is not designed that way just like collision physics and damage model has to be integrated into the game. Probably codemasters F1 game do better job in this regard.

They calcualte car based on weight distribution, engine power and other characterisitcs. GT5 is not a generic engine like ISI which can be modded to anything, there is RB X1 mod for rfactor and pretty much anything can be done by community let alone developers like Simbin which also uses gmotor engine for their racing games.

I personally do not mind it too much because in PC sims rather than driving most time is spend is trying to setup the car to no limit

 

[Exorcet]

I am saying the game is not designed that way

The game is designed the way I just laid out. That's because all of them are. How else is it going to work?

]They calcualte car based on weight distribution, engine power and other characterisitcs. GT5 is not a generic engine like ISI which can be modded to anything, there is RB X1 mod for rfactor and pretty much anything can be done by community let alone developers like Simbin which also uses gmotor engine for their racing games.

This isn't how simulation works. You don't "calculate a car". They ask the manufacturer for data, and then put that data in the physics engine. Done.

Logically, when you tune things, or whatever, you just change the data in the particular car file.

I personally do not mind it too much because in PC sims rather than driving most time is spend is trying to setup the car to no limit

The first time I tried GTR Evolution I hopped in and drove. I didn't need to spend time setting up anything.

And even if GT5 required hours and hours tuning, all you would need to do is ask someone on GTP for tunes and then you wouldn't need to spend time setting up anything. Even better would be a Forza system where you could get people's tunes in game.

 

[Exorcet]

EDIT 11/12/11 - Top Speed was off, I had just thought it was due to low mesh accuracy in the CFD, but it turns out that I forgot to vary the drag when I was comparing it to engine power output. The tops speeds are now corrected.

Coincidentally, the top speed change from t he front wing is just under the change from front and rear predicted in GT5.

I have to announce a rather careless error I made when doing this analysis that I left unchecked until now. The top speeds were off. Sorry for the false numbers, but they've been corrected now.

 

[TBR 427]

5 minutes with even something like foilsim and you should be able to get a wing that increased -lift/drag ratio as AoA increases. Sure it's not perfect, but as it's useful for things like model airplane stuff, it's good enough for now.

And could you post up some streamlines as they go over the 'sidepods' of that...thing. I'm half expecting to see some tornado-sizex vorticies coming off the nosecone, bouncing back violently at the top of the sidepods, then rolling over the top and straight into the endplates and/or rears.

Also, you didn't even manage a 2:1 efficiency? Heck, a Lotus 70 would destroy that thing around corners...No offense, but you'd be better off with an inclined plane and a gurney.

 

[Exorcet]

This isn't a real project, so it doesn't get high priority. Time spent on this is time taken away from other things.

There was no major turbulence over the side pods, I had the mesh spacing set pretty high (however I put a more dense mesh on the wings).

You can get better lift/drag with increased AoA when lift is small yes, because induced drag might be less than profile drag, but that's not going to happen in a race car set up for maximum downforce. I'm sure you're familiar with Lift curves and the peak that happens just before separation. And like I said, CDi is proportional to CL^2.

 

[MHPALA]

Implementing physics into a game is not hard for a programmer who knows physics. Its all algorithms 👍. To save myself from explaining a whole alot i'll let this document do the talking. Its the truth, not complex at all. What is complex? Adding the effects to the tires (rubber compound).

Link to the PDF http://tiny.cc/eyozr 👍

 

[silicon1138]

Implementing physics into a game is not hard for a programmer who knows physics. Its all algorithms 👍. To save myself from explaining a whole alot i'll let this document do the talking. Its the truth, not complex at all. What is complex? Adding the effects to the tires (rubber compound).

Link to the PDF http://tiny.cc/eyozr 👍

Yes, that doesn't look complex at all

 

[kimi123]

The game is designed the way I just laid out. That's because all of them are. How else is it going to work?


This isn't how simulation works. You don't "calculate a car". They ask the manufacturer for data, and then put that data in the physics engine. Done.

Logically, when you tune things, or whatever, you just change the data in the particular car file.


The first time I tried GTR Evolution I hopped in and drove. I didn't need to spend time setting up anything.

And even if GT5 required hours and hours tuning, all you would need to do is ask someone on GTP for tunes and then you wouldn't need to spend time setting up anything. Even better would be a Forza system where you could get people's tunes in game.

I never said they calculate physics for every car. They have a physics engine but they also modify it accordingly to make it behave like a real car. That is how did for SLS AMG the video I posted some pages back.

What I am saying for F1 car and aero especially they do not do as ISI that is why the car does not change according to the configuration of setups. It is very limited and not done like CTDP or RH mods on PC

 

[CSLACR]

The greatest F1 car ever never built

Two set ups

"Monaco"

2 element front wing
2 element rear wing (fixed main element, high AoA secondary element)​

"Monza"

1 element front wing
2 element rear wing (fixed main element, low AoA secondary element)​

Data break down
Total - Total force on car
F - Front wing
R - Rear wing

X - Side force, ignore
Y - lift (negative is downforce)
Z - drag (negative is drag)

Vortex wake - shows induced drag as air kinectic energy.

A note : clearly, my rear wing was ineffective, I probably need to make the second element bigger, or adjust the AoA of the main element. Either way it shows that drag does up faster than lift, as it must.



"Monaco"
Total
Integral Parameter Value X-component Y-component Z-component Surface Area [ft^2]
Force [lbf] 1036.14 -1.43198 -675.482 -785.693 295.94
F
Integral Parameter Value X-component Y-component Z-component Surface Area [ft^2]
Force [lbf] 600.961 0.00597609 -590.225 -113.086 15.9187
R
Integral Parameter Value X-component Y-component Z-component Surface Area [ft^2]
Force [lbf] 333.933 0.311238 -321.63 -89.8055 9.39765

Vortex wake

TOP SPEED (700 hp) - 213 mph


"Monza"
Total
Integral Parameter Value X-component Y-component Z-component Surface Area [ft^2]
Force [lbf] 831.906 -1.49984 -421.721 -717.089 291.856
F
Integral Parameter Value X-component Y-component Z-component Surface Area [ft^2]
Force [lbf] 308.696 -0.115179 -306.964 -32.6561 12.5432
R
Integral Parameter Value X-component Y-component Z-component Surface Area [ft^2]
Force [lbf] 256.635 0.13035 -254.805 -30.5943 9.46782

Vortex wake

TOP SPEED (700 hp) - 220 mph






A 3% top speed reduction, mainly from the front wing alone (which might be misleading because drag on the front wing can manifest itself as reduced drag on the body. The rear wing is pure loss when it comes to drag).



EDIT 11/12/11 - Top Speed was off, I had just thought it was due to low mesh accuracy in the CFD, but it turns out that I forgot to vary the drag when I was comparing it to engine power output. The tops speeds are now corrected.

Coincidentally, the top speed change from t he front wing is just under the change from front and rear predicted in GT5.

Now if you'll just tell us kindly how to relate real world lb figures into GT5's numerical system...

What I can take from this is that your setups both have more downforce on the front then rear, whether that's how they run IRL I have no idea, but it's certainly not how they run in GT5, so that makes it's relation questionable at best.

In any case, the downforce appears to be about 60% on the "Monza" setup of what it is on the Monaco setup.

Bit it's a fictional F1 car, so everything about it can be questioned which diminishes it's overall value.

It also doesn't show the improvements in cornering speeds that I can tell, unless that's the figure you told me to ignore, in which case I'd like to know why I should ignore it, it's critical to the subject.

In any case, very low downforce setups are never faster in GT5, not even at Le Sarthe, we don't need a graph to tell us that, our lap times do, which means no matter what the graphs here show, we know GT5's downforce system is not correct.

Did I mention downforce improves 0-60 acceleration times in GT5?

PS - Axle placement on the front of these pictures is just awesome.

 

[Exorcet]

Now if you'll just tell us kindly how to relate real world lb figures into GT5's numerical system...

Well, it came out to be a 300 lb difference at ~100 mph for the 908 between max and min downforce (check a few pages back where I provided the numbers). For the F1, it should be at least that much, but probably is much more.

What I can take from this is that your setups both have more downforce on the front then rear, whether that's how they run IRL I have no idea, but it's certainly not how they run in GT5, so that makes it's relation questionable at best.

It shows that increasing CL can greatly increase CD for wings. Also, you would change CL by varying AoA. The AoA change I made from one set up to another is reasonable considering the photos of the F2007 I posted. Based on that, the drag difference of the wings between the two configurations could be considered reasonable in terms of accuracy. I'm probably not off by an order of magnitude.

In any case, the downforce appears to be about 60% on the "Monza" setup of what it is on the Monaco setup.

Bit it's a fictional F1 car, so everything about it can be questioned which diminishes it's overall value.

About all that can be questioned is if the ratio of wing net force to chassis net force is realistic. Everything else is going to apply to any car. To first order, it probably is, because the dimensions are close to those that a real F1 car would have.

F = q*coefficient*Area. My q (dynamic pressure) is correct, my Area is pretty darn close, the coefficients are probably off though.

It also doesn't show the improvements in cornering speeds that I can tell, unless that's the figure you told me to ignore, in which case I'd like to know why I should ignore it, it's critical to the subject.

Maximum lateral g gains to first order would be related to downforce/mass of the car.

In any case, very low downforce setups are never faster in GT5, not even at Le Sarthe, we don't need a graph to tell us that, our lap times do, which means no matter what the graphs here show, we know GT5's downforce system is not correct.

Did I mention downforce improves 0-60 acceleration times in GT5?

PS - Axle placement on the front of these pictures is just awesome.

You're right. I wanted to show that it would be very easy to get bigger differences in drag than what GT predicts, and try to estimate what kind of difference you would see in a real car. It's also a good bit of modeling and CFD practice, though I usually don't get to spend a lot of time or put a ton of effort into side projects like this.

I don't remember why the front axles are like that, but I was probably going to try to do a more complete front suspension before I just said "forget it" and extruded a circle from one body to the other lol. This particular F1 car also happens to be amazingly flat, and features a roomy driver compartment.