A Vector tuning theory for drift cars

[GhostZ]

Like i said before ...your tune is not perfect. Stop advertising it to be.

You should read my posts, and if you're really not understanding them (as it sounds like, you really aren't making much sense) try to understand them first before saying that they're wrong. Or just don't comment on them.

 

[Drift_Monkey]

You should read my posts, and if you're really not understanding them (as it sounds like, you really aren't making much sense) try to understand them first before saying that they're wrong. Or just don't comment on them.

well let me know where I said they were wrong and i'll fix it right away. And I understand fully but I don't think you understand anyone but yourself.

 

[GhostZ]

well let me know where I said they were wrong and i'll fix it right away. And I understand fully but I don't think you understand anyone but yourself.

I've done already, but I'll do it again for you.

I explained and defined what a perfect tune should be. If, in the hands of the ideal driver, a car's tune allows it to do what you want to the best possible result, then it is perfect.

Your claim that everyone has different styles, or (more accurately) different goals, doesn't mean the tune cannot be perfect.

...a car's tune allows it to do what you want to the best possible result...

Let's say you want to overtake a very particular car. If you know exactly how that car is going to move, you can make a tune that means, in the hands of the ideal driver, the overtake can be done as quickly and effectively as possible. Sure, OTHER tunes will allow you to do that, but there is only 1 tune, within the confines of GT5, that OPTIMIZES the result.

Remember that a perfect tune is not a perfect result. As I've said before, tunes are just a way for a driver to get the car to behave the way they want.

So how do you make the perfect result, even if you know the perfect tune for it? Driver skill. My tunes assume perfect driving ability. However, that's why I use comparative tuning which is my old method (before Vector tuning) that tuned different cars in relation to eachother, so that if a driver can do one thing in one car, they can just hop into the tuned car and do the same thing with the same success, even if the cars are very different. With those two in combination, you can bring down the "ideal" down and bring up the "comfortable tune" until you reach a point where the driver's skill is able to do that particular act, whether it is overtaking, cornering, tandeming, whatever, and execute it at the highest result.

Why do I think the 240Z-G is the best car? Because my goal is not to get the best result over a single corner, but the best result over 1000+ corners.

This method won't work if you are not an excellent driver. It will not work if you are limited to one 'style' that prevents you from improving. It will not work if you do not understand how to utilize the changes in the car. It will not work if you do not know what your ultimate goal is.

But if you do, then this tune method's job is to try and make a human driver, who is not perfect, perform as close as possible, without concession, to that ideal driver.

Do you understand what 'perfect' means now?

 

[Drift_Monkey]

your arrogance never ceases to amaze me

 

[GhostZ]

your arrogance never ceases to amaze me

That's because when I was taught to add numbers, I didn't say "But professor! There's no perfect way to add 1+1! Everyone is different!". I knew that 1+1=2.

 

[Drift_Monkey]

So you're the professor and I'm just the student? The arrogance continues

 

[Aquarelle]

Oh just stop it.

 

[SlipZtrEm]

Oh just stop it.

I think it's a valid statement. Comments like this:

But if you do, then this tune method's job is to try and make a human driver, who is not perfect, perform as close as possible, without concession, to that ideal driver.

Do you understand what 'perfect' means now?

Reek of arrogance. It's been brought up a few times that, quite frankly, this numbers-based tuning can't be completely accurate because GT doesn't even provide some of the numbers needed. It's a bunch of guess-work, pseudo-math, and self-fulfilling ego boosting. GT's incredibly simple tire model (with a completely silly selection of options that represent nothing more over other grades than a grip multiplier) means we can't get some of the figures Ghost claims to be using. This has been covered in the past in far more detail than I have time to explain - but a good read starts roughly here.

I'm all for in-depth tuning and strong reasoning behind it within GT5, but it could do without an attitude resembling this:

 

[GhostZ]

The only person I've reacted (really, overreacted) negatively toward is drift_monkey. Its because he's insulted me and called me wrong in many threads, but hasn't shown that he even understands what I'm doing, since his responses usually do not make sense or are trying to attack my personality, not whether my ideas are right or wrong. Really, I don't have a problem with that. Its not my place to make people agree with me. But if he just takes it for granted that he's right and I'm wrong because of things like "there's no perfect way to drift", and no actual reasoning or understanding about the idea, then he shouldn't be posting here to confuse other people and start an argument. If he DOES understand what talking about, he's not showing it for some reason. When you start throwing insults (such as condescending) around and making claims about math without backing them up, yes, I'm going to get defensive. I'm sorry for reacting the way I did. But you should understand that its not some general arrogance, its frustration I have with a single troll.

I want real, valid criticism, like 'would this calculate LSD numbers?' or 'how do you take into account different max steering angles' or even 'do you think grip increases at small slip angles in gt5 or just decrease?' these are ALL valid criticisms that help show where the limit s of thisodel are. But 'gt5 doesn't tell you your speed' or ' my idea of perfect is different than yours' are not. I want to improve the calculation and get better results, not argue semantics. That's why it bothers me that he's responding like that.

As for values in gt5 that you would need for this model, that just not a problem at all. As I said in the first post, you just need speed, weight and direction. Gt5 gives those, the rest is math. In fact, gt5's simple tire model actually makes this method more useful since I don't need to take tire deformation into account.

Or is there some other measurement you're thinking of that gt5 doesn't give that you think I might need for this? There are a few ways to make this model more accurate that would, but since most adjusents in tuning is down to a limited set of numbers (like camber, which only ranges 0-10 bt 10ths ofna degree) this model is precise enough to give the necessary values.

 

[Drift_Monkey]

I'm not trolling, I'm calling you out on your B******t. You can't expect me to know you're right when you don't give me a reason to believe you are. W've asked you before to provide certain figures which would change that and you continue to ignore it. And your arrogance is seen yet again in your last response by being so sure that your method is right. I've never said you were wrong. I asked you to provide some sort of proof or evidence that what you're saying is true instead of leading the forum to believe you based on blind faith.

If you don't want this to continue then all you need to do is provide the numbers we've asked for. I want the values you use and how you get them to find how much camber I need in the front of a 350z to get the most grip around a corner. How do you go about this in a mathematical procedure?

 

[GhostZ]

I'm not trolling, I'm calling you out on your B******t. You can't expect me to know you're right when you don't give me a reason to believe you are. W've asked you before to provide certain figures which would change that and you continue to ignore it. And your arrogance is seen yet again in your last response by being so sure that your method is right. I've never said you were wrong. I asked you to provide some sort of proof or evidence that what you're saying is true instead of leading the forum to believe you based on blind faith.

If you don't want this to continue then all you need to do is provide the numbers we've asked for.

(X + c)^2 + (y+c)^2 =(r+c+n)

R is radius of origin circle for the arc length of the inside corner boundary. N is difference in radius from the inside to the outside of the corner in width. C = N/(root2-1)

I have created for you a driving line. Put the equations into a graphing utility.

Now apply a third dimension to the graph as z. Z's value of x and y is the g force formula from speed. We know CH tires corner around .85 g, so for a given circle, the z value is max speed around it. As the circle gets wider, speed goes up obviously.

Create a vector for your point on the circle with weight, direction (as angle deviation from the tangent line, aka drift angle) and create a linear relationship between the slip angle and the g force. Now you know for any angle and any corner radius exactly what the lax speed that is possible is, for a one wheeled vehicle. It isn't hard to do four vectors (one for each wheel) related by their relative positions, and find ideal toe angles. Find cross sections of the contact patch and you can find optimal camber for those angles.

Of course, this does assume you're capable of driving the car that well. I would hope so.

I'm on my phone (and unfortunately making more spelling errors) so this is all from memory. I can give more and in more detail when in back home and with my notes.

EDIT: Also, your demand is too simple. Your optimal front end numbers depend on rear end numbers, instant torque output, the angle you want to go at, how wide the road is, width of the car, length of the car, your CoM, among other data. If you can calculate those (most are pretty easy, the rest I covered in other threads) this vector tuning tool will give you what you want.

 

[Drift_Monkey]

(X + c)^2 + (y+c)^2 =(r+c+n)

R is radius of origin circle for the arc length of the inside corner boundary. N is difference in radius from the inside to the outside of the corner in width. C = N/(root2-1)

I have created for you a driving line. Put the equations into a graphing utility.

Now apply a third dimension to the graph as z. Z's value of x and y is the g force formula from speed. We know CH tires corner around .85 g, so for a given circle, the z value is max speed around it. As the circle gets wider, speed goes up obviously.

Create a vector for your point on the circle with weight, direction (as angle deviation from the tangent line, aka drift angle) and create a linear relationship between the slip angle and the g force. Now you know for any angle and any corner radius exactly what the lax speed that is possible is, for a one wheeled vehicle. It isn't hard to do four vectors (one for each wheel) related by their relative positions, and find ideal toe angles. Find cross sections of the contact patch and you can find optimal camber for those angles.

Of course, this does assume you're capable of driving the car that well. I would hope so.

I'm on my phone (and unfortunately making more spelling errors) so this is all from memory. I can give more and in more detail when in back home and with my notes.

EDIT: Also, your demand is too simple. Your optimal front end numbers depend on rear end numbers, instant torque output, the angle you want to go at, how wide the road is, width of the car, length of the car, your CoM, among other data. If you can calculate those (most are pretty easy, the rest I covered in other threads) this vector tuning tool will give you what you want.

too simple? really? so how do you account for how wide the road is, gradient of the turn, front downforce, weight distribution, brake balance and suspension response while going into the corner? do you measure speed in mph or km/h?

 

[GhostZ]

too simple? really? so how do you account for how wide the road is, gradient of the turn, front downforce, weight distribution, brake balance and suspension response while going into the corner? do you measure speed in mph or km/h?

They are numbers. There are physics formulas that give them, and other formulas that use them. There is a reasom I have a massive excel document to give them to me off of car data. Looks up roll formulas, lift force formulas, down force formulas, and you'll get a good id era where I get them and how I use them. Like I said, covered in previous threads. I don't zee the need to do algebra for you if I can show you everything needed to do it yourself. This gives the best grip for cornering, combine it with comparative tuning, and you get tunes that are not quite ideal, but act like a car one is used to, to get the most successful tune. You really want something, submit an application to my tuning garage, or learn to use all of those physics formulas.

 

[Drift_Monkey]

do you really think real life physics and gt5 physics are the same? How do you take into account all these variables and still come out with a number that can be placed into gt5's tuning window? Where is the formula for dampers if gt5 has it's own model for dampening. You do realize that in real life these numbers aren't figures ranging from 1-10 but units based on weight etc. How would you get let's say the value from an equation to a value of 1-10 in the dampers column?

 

[Ridox2JZGTE]

I think Monkey should drive the car that is on share by GhostZ, see if the tuning method works on the car or not for drifting Instead of arguing the means, why not try the result No loss there, and I think Ghostz is doing fine, no arrogance detected at all - for me at least.

 

[Drift_Monkey]

I like that idea... you think he'll add me? what's your real psn ghost?

 

[Ridox2JZGTE]

I like that idea... you think he'll add me? what's your real psn ghost?

I am sure he will, here is his tuning garage thread :

Ghost Tuning Garage

His PSN is as stated in his profile, a member already tried his car :

So the day this thread came out I asked Ghost here to make me a drift car that I could use to compete with and drift on a daily basis. I choose the Z33 and I explained to him what it was I was looking for and he talked with me for a bit. I basically told him go nuts and use his imagination. Few days later I get my car and I LOVE it. It is up to Ghost if he shares the tune with others. I personally enjoy learning a new style. I feel it will help me greatly improve. Thank you, Ghost.

 

[GhostZ]

do you really think real life physics and gt5 physics are the same? How do you take into account all these variables and still come out with a number that can be placed into gt5's tuning window? Where is the formula for dampers if gt5 has it's own model for dampening. You do realize that in real life these numbers aren't figures ranging from 1-10 but units based on weight etc. How would you get let's say the value from an equation to a value of 1-10 in the dampers column?

This model deals with peak cornering ability. Dampeners affect how the corner peak is approached, whether slower or faster, which is different. But that doesn't mean you can't use reason to figure out dampener amounts. Since they affect the roll rate, it comes down partially to driver control, and partially down to the rate of change on the corner. If you hold driver control constant, than the rule becomes wider corners: more dampening force, shorter corners: less dampening force. You are right, I can't give you ideal numbers, but I CAN prescriptively tell you which direction to go with the numbers you have.

Thankfully, that corner model I showed you (with the circle formula) can be shown to you directly now that I'm on my laptop.

https://www.desmos.com/calculator/hf87yzndnw

I threw that up for you to play with. W is the width of the car. R is the radius of the inside of the corner, N is the difference between the two corners. The units are relative so that means that if you use the same values, the numbers all work out, provided its base 10 (metric).

Red circles are the front tires, black line is the center of mass over the front wheels. This simple diagram doesn't do rear tires yet.

This demonstrates a few principles:

1. As the width of the car increases, the radius of the path it must take decreases. (faster rate of change in direction)

2. As the width of the road decreases, the radius of decreases also, so the path changes.






In other words....

Wider cars on narrower roads need higher dampening force since change in velocity and direction happen in less time.

So that's if you hold driver skill as identical. But what if you hold corner radius as identical between Car A and Car B? How do you tune dampeners toward driver skill?

I can use comparative tuning to figure out the ratio of radii between car A and car B also. That means if someone is using car A at dampener setting 3 with radius 10, and car B does it in radius 13, then I need (assuming dampeners are linear, which I believe they are, if they are log scale or exponential, the difference only matters when switching by more than 2 points) then the dampeners that give the same "Feel" of driving, so that both cars respond similarly along their ideal path, is 3 + 3 x10/13, or (close to) 4. The difference should scale up.

 

[bread82]

A question for Ghost:

I'm interested in how you get tyre grip values for your model. As far as I know, the lateral (horizontal) force a tyre can provide depends on:
1) the downwards force (weight) on that tyre
2) The tyre load sensitivity (as you said above, the sideways force "increases almost proportionally as weight over the tire goes up")

(and obviously pressure and temperature, track surface etc, but let's assume those are constant for the moment.)

For tyre load sensitivity:
http://en.wikipedia.org/wiki/Tyre_load_sensitivity
Wiki says "In practice, the maximum horizontal force Fy that can be generated is proportional, roughly, to the vertical load Fz raised to the power of somewhere between 0.7 and 0.9, typically." Do you assume a value in this range, or have you found a way to determine one through testing in GT5?

The reason I ask is because I wondered a while ago if it was possible to get an idea of tyre load sensitivity in GT5. After reading criticisms of the tyre model in several threads I wondered if it could be investigated.

If Ghost or anyone wants to comment on the following method, I'd be very interested:
The furthest I got was to do "max G" tests on the speed test, using an FR car with known weight distribution, super stiff springs and dampers (I essentially wanted "no suspension", as an assumption I was going to assume only load transfer, not weight transfer due to CoG movement through suspension moving. Admittedly this is an approximation but I wanted to cut my teeth on something simpler to check my method was sound. I used ballast in different places to get a variety of weights and weight distributions in order to generate more data. I also used CH tyres - since I was assuming a rigid suspension I didn't want sticky racing tyres that would apply more force and make the suspension travel more.

I got a set of values for:
1) Static Front and rear weights (simply the weight of car with any ballast, using the weight distribution from the ballast tuning screen).
2) Max G from the speed test for all these ballast combinations.

Knowing: wheelbase, CoG height, Max G attained, mass of car and static weight distribution, I'm pretty sure it is possible to calculate the dynamic weight distribution i.e. as the car is "launching", when the max G reading appears on the screen, you can work out what the downwards load was on the front and rear tyres.

From the max G force and mass of car you can work out the horizontal force applied by the tyre on the road.

Knowing CoG height, and the horizontal force, you can calculate a couple.
http://en.wikipedia.org/wiki/Couple_%28mechanics%29

Then using wheelbase and this couple, you can calculate the load transferred from front to rear axles

You now know a set of vertical forces (downwards load on rear tyre), and horizontal forces (forwards driving force of rear tyre). Hence you should be able to deduce tyre load sensitivity

The only things not directly available from GT5 are wheelbase and CoG height (which I underlined above for this reason). Wheelbase could be guessed at from car length given in GT5 and a photomode pic (measure length of car, measure wheelbase). CoG height I'm not so sure about - I bunged in a real world value as I wanted to test my method rather than get an instant application to GT5. I think Ghost put his method for CoG height in another thread.

Apologies if anyone's eyes are bleeding. If people want more info on this I'm happy to dig out my sheet, or open another thread.

Cheers,

Bread

 

[GhostZ]

A question for Ghost:

I'm interested in how you get tyre grip values for your model. As far as I know, the lateral (horizontal) force a tyre can provide depends on:
1) the downwards force (weight) on that tyre
2) The tyre load sensitivity (as you said above, the sideways force "increases almost proportionally as weight over the tire goes up")

(and obviously pressure and temperature, track surface etc, but let's assume those are constant for the moment.)

For tyre load sensitivity:
http://en.wikipedia.org/wiki/Tyre_load_sensitivity
Wiki says "In practice, the maximum horizontal force Fy that can be generated is proportional, roughly, to the vertical load Fz raised to the power of somewhere between 0.7 and 0.9, typically." Do you assume a value in this range, or have you found a way to determine one through testing in GT5?

The reason I ask is because I wondered a while ago if it was possible to get an idea of tyre load sensitivity in GT5. After reading criticisms of the tyre model in several threads I wondered if it could be investigated.

If Ghost or anyone wants to comment on the following method, I'd be very interested:
The furthest I got was to do "max G" tests on the speed test, using an FR car with known weight distribution, super stiff springs and dampers (I essentially wanted "no suspension", as an assumption I was going to assume only load transfer, not weight transfer due to CoG movement through suspension moving. Admittedly this is an approximation but I wanted to cut my teeth on something simpler to check my method was sound. I used ballast in different places to get a variety of weights and weight distributions in order to generate more data. I also used CH tyres - since I was assuming a rigid suspension I didn't want sticky racing tyres that would apply more force and make the suspension travel more.

I got a set of values for:
1) Static Front and rear weights (simply the weight of car with any ballast, using the weight distribution from the ballast tuning screen).
2) Max G from the speed test for all these ballast combinations.

Knowing: wheelbase, CoG height, Max G attained, mass of car and static weight distribution, I'm pretty sure it is possible to calculate the dynamic weight distribution i.e. as the car is "launching", when the max G reading appears on the screen, you can work out what the downwards load was on the front and rear tyres.

From the max G force and mass of car you can work out the horizontal force applied by the tyre on the road.

Knowing CoG height, and the horizontal force, you can calculate a couple.
http://en.wikipedia.org/wiki/Couple_%28mechanics%29

Then using wheelbase and this couple, you can calculate the load transferred from front to rear axles

You now know a set of vertical forces (downwards load on rear tyre), and horizontal forces (forwards driving force of rear tyre). Hence you should be able to deduce tyre load sensitivity

The only things not directly available from GT5 are wheelbase and CoG height (which I underlined above for this reason). Wheelbase could be guessed at from car length given in GT5 and a photomode pic (measure length of car, measure wheelbase). CoG height I'm not so sure about - I bunged in a real world value as I wanted to test my method rather than get an instant application to GT5. I think Ghost put his method for CoG height in another thread.

Apologies if anyone's eyes are bleeding. If people want more info on this I'm happy to dig out my sheet, or open another thread.

Cheers,

Bread

Your max G test is exactly what I've done (you and I had a series of posts and PMs about this a while ago I think...) the Chassis swapping done shows that for almost all cars, wheelbase measurements are realistic and can be used from the production vehicles.

There is a graph for GT5 tires that another member created after testing, and while I cannot find it, you can see that for most cars the max cornering grip is as follows:

CH: .85-.9G
CM: .9.-.95G
CS: .95-1G
SH: 1-1.1G
SM: 1.5-2G

etc. The variation comes from different tire widths and suspension configurations mostly.

The question is now how much tire cornering force increases or decreases depending on slip angle.



Either way, when it gets down to it, this particular thread isn't about finding any of those values. It's about using them to find optimal toe and camber angles for particular cornering styles, so that when you're drifting at a given angle, your tires are pointed in a way that optimizes grip in the direction you want to go. Speed is given once you know how the individual tire's grip decreases as the slip angle increases.

 

[Drift_Monkey]

Your max G test is exactly what I've done (you and I had a series of posts and PMs about this a while ago I think...) the Chassis swapping done shows that for almost all cars, wheelbase measurements are realistic and can be used from the production vehicles.

There is a graph for GT5 tires that another member created after testing, and while I cannot find it, you can see that for most cars the max cornering grip is as follows:

CH: .85-.9G
CM: .9.-.95G
CS: .95-1G
SH: 1-1.1G
SM: 1.5-2G

etc. The variation comes from different tire widths and suspension configurations mostly.

The question is now how much tire cornering force increases or decreases depending on slip angle.



Either way, when it gets down to it, this particular thread isn't about finding any of those values. It's about using them to find optimal toe and camber angles for particular cornering styles, so that when you're drifting at a given angle, your tires are pointed in a way that optimizes grip in the direction you want to go. Speed is given once you know how the individual tire's grip decreases as the slip angle increases.

Where do these numbers come from?

 

[GhostZ]

Where do these numbers come from?

Just look at the bottom of the screen in most views. There is a g meter for you. Of all things to question, this is pretty obvious as fact, since GT5 gives you explicit numbers. All you have to do is start testing and recording, however, someone else already did that for you:

https://www.gtplanet.net/forum/showthread.php?t=160821

I found these results to be consistent with my own, the numbers were a little different since I was trying to remember (and the last time I tuned a non CH car was a while ago). But the source of the numbers doesn't really matter, or even the guesswork, since GT5 just gives you the number that you're cornering at if you do a skidpad test. Since the driving path circle formula I posted earlier is just a skidpad diagram, G numbers from that can be very easily used in equations for the path. After all, each corner is just a "part" of a circle, with varying radii before and after.

 

[Drift_Monkey]

Just look at the bottom of the screen in most views. There is a g meter for you. Of all things to question, this is pretty obvious as fact, since GT5 gives you explicit numbers. All you have to do is start testing and recording, however, someone else already did that for you:

https://www.gtplanet.net/forum/showthread.php?t=160821

I found these results to be consistent with my own, the numbers were a little different since I was trying to remember (and the last time I tuned a non CH car was a while ago). But the source of the numbers doesn't really matter, or even the guesswork, since GT5 just gives you the number that you're cornering at if you do a skidpad test. Since the driving path circle formula I posted earlier is just a skidpad diagram, G numbers from that can be very easily used in equations for the path. After all, each corner is just a "part" of a circle, with varying radii before and after.

Different cars will have different levels of lateral grip based on the tune. He tested a specific car. Those numbers won't be the same for all cars. And if tyre compound is a simple result of a grip multiplier then certain forumulas to find certain settings(camber and toe) be thrown off? What is the width of the tyre model in GT5?

 

[Ridox2JZGTE]

Different cars will have different levels of lateral grip based on the tune. He tested a specific car. Those numbers won't be the same for all cars. And if tyre compound is a simple result of a grip multiplier then certain forumulas to find certain settings(camber and toe) be thrown off? What is the width of the tyre model in GT5?

There is an in depth discussion about tire model at Forza 4 vs GT5 physics thread

Forza 4 vs GT5 physics thread


As far as I know, a CH on a wide tire car has the same lateral G with CH on a skinny tire car - when both cars have no aero and in stock condition.

 

[Drift_Monkey]

There is an in depth discussion about tire model at Forza 4 vs GT5 physics thread

Forza 4 vs GT5 physics thread


As far as I know, a CH on a wide tire car has the same lateral G with CH on a skinny tire car - when both cars have no aero and in stock condition.

yes we know, but the question is how would Ghost calculate certain settings if the game's tyre width is not mentioned? Surely tyre width is a big factor when it comes to camber.

Which brings me to what me and the mods try to address is that GT5 does not provide enough information to calculate settings using real life formulas.

 

[dice1998]

yes we know, but the question is how would Ghost calculate certain settings if the game's tyre width is not mentioned? Surely tyre width is a big factor when it comes to camber.

Normally, that would be true. But I don't think it is for GT5. Then again, it's all theories.

Unless someone done did illegal activities and knows exactly how the GT5 physics engine works.

 

[Drift_Monkey]

Normally, that would be true. But I don't think it is for GT5. Then again, it's all theories.

Unless someone done did illegal activities and knows exactly how the GT5 physics engine works.

Yep, A theory meaning it's a possibility that these methods work and there's a possibility it doesn't.

 

[anthaliscious]

Just Drift.

 

[Lock2Lock]

Just Drift.

Exactly. I miss the good old days.

I would never get away with drifting a Re Amemiya AsparaDrink RX7 again.

 

[Drift_Monkey]

Blatant arrogance grinds my gears