Cars Porsche 911 GT3 cup circuit + hillclimb [Deleted]

[rpmmod]

rpmmod submitted a new resource:

Porsche 911 GT3 Cup 2017 circuit and hillclimb - Two Porsche 911 GT3 Cup with complexe aero

INSTALLATION INSTRUCTIONS :


For circuit version :
- Extract the folder "ks_porsche_911_gt3_cup_2017_rpm" in content\cars.

- Copy from folder "ks_porsche_911_gt3_cup_2017" all the files and folders exept "data.acd" files and "ui" folder, say yes for merge folder "sfx".

- In sfx folder rename "ks_porsche_911_gt3_cup_2017.bank" to "ks_porsche_911_gt3_cup_2017_rpm.bank"


For hillclimb version :
- Extract the folder "ks_porsche_911_gt3_cup_2017_hillclimb" in content\cars.

- Copy from...

Read more about this resource...

 

[mclarenf1papa]

@rpmmod because, mathematically, you can resolve the ~20 wings you used into 5 that have *exactly* the same effect. Same deal with the fins - those can get resolved to 4 or even very closely by 2. There’s no CFD in AC - the “wings” are simply lookup elements for CL/CD - their meaning is nothing more than the numbers you put in - the chord length could be 2000 and it will behave the exact same as a chord length of 1 so long as you scale your values by 1/2000. The visual representation means nothing, the angle means only so much as your LUT files dictate, and the position is only as relevant as how you set the aero up. The way you did the aero is far from reasonable and generally far from the right way to do it. The effort that went into the idea is evident, but the idea is far misplaced.

So, perhaps before calling someone stupid, make sure you know what you’re talking about.

 

[rpmmod]

I know what I'm talking about, and you ?

With one wing on the front you can just have one support with a regular curve for change on angle of attack, and the center of downforce don't move.

In real when angle of attack change (on turn) the center of downfoce change and the downforce change with non regular curve.
If in addition the sides are not taken into account the result is a joke.

The difference is huge, flagrant, without appeal on rally cars, and on hillclimb if there is a lot of tight turns, mostly if the car have a lot of downforce.

 

[mclarenf1papa]

“I know what I’m talking about”
Evidently not...

Like I said, mathematically, everything you did could be resolved into 5 wings and 4 fins. The behavior, the numbers - it’d be the same.

 

[rpmmod]

That's why for ACC Kunos abandoned the wing system.

How are you doing so that the center of downforce at the front move whith only one wing on the front? (it moves longitudinally and laterally).

Compared to Iracing cars AC cars are bland, lifeless; whether the turn is tight or not the behaviour of the car il almost the same.

"Evidently not...", "Like I said", what did you do so great to have such an ego? where are the arguments ?

 

[mclarenf1papa]

That's why for ACC Kunos abandoned the wing system.

How are you doing so that the center of downforce at the front move whith only one wing on the front? (it moves longitudinally and laterally).

Compared to Iracing cars AC cars are bland, lifeless; whether the turn is tight or not the behaviour of the car il almost the same.

"Evidently not...", "Like I said", what did you do so great to have a similar ego? where are the arguments ?

Two wings on left and right sides of the front axle, two wings on left and right sides of the rear axle, and one wing for the adjustable rear airfoil. Those are the five I mentioned. Longitudinal CoP migration only needs two wings to be done, lateral needs two, and having both requires four.

Kunos abandoned the wing system for myriad reasons in ACC. One of them is that using an aeromap is simply a lot easier if you have the data for it. Additionally, you're overemphasizing the importance of lateral CoP shifts - the load variation doesn't change as significantly as you're insinuating and it's only worth trying to correlate if you already have data, which you'll be incredibly hard-pressed to find (not that you can model aerodynamic roll sensitivity correctly in vanilla AC anyway).

As for your last comment, I've done work with racing teams and manufacturers in the past and have been using AC to mod since before the dev tools were released.

 

[Michael Hornbuckle]

"Evidently not...", "Like I said", what did you do so great to have such an ego? where are the arguments ?

Considering that he has the respect of both Stefano and Aris, along with creating the physics of some of the best mods available for AC, not to mention the work he does for actual real world race teams...seems like pretty good reasons to have confidence (ego) to me.

 

[AccAkut]

Wasn't it also advised to use as little aero surfaces as possible for performance reasons? With every additional wing you add more calculations to the physics. (...or was this only regarding active aero and animation...)

 

[rpmmod]

"Two wings on left and right sides of the front axle, two wings on left and right sides of the rear axle"
I never seen a car with wings on left and right sides on assetto, your wings are all aligned; and this does not allow the front downforce to move longitudinally. See the front of z4 or ferrari 599XX EVO where is it pretty pronounced.

In addition the center of donforce move vertically because of side effect see the z4 gt3, the recent wrc, etc... where is it pretty pronounced.

 

[rpmmod]

AccAkut : now we have good CPU, for opponent you can use the kunos cars, and mostly I do hillclimb cars.

 

[rpmmod]

I would like to add, if the total downforce can get resolved to 4 or 5 wings, why the cars that have a lot of downforce have unrealistic coefficient of lift (kunos or mod).

 

[Kyuubeey]

Oh boy.

I believe you have the right idea but you've misinterpreted it a little bit.

Definitely if you want a more correct height behavior for aerofoil surfaces in roll, you will want two sides which interact with the ground. Roll cannot be simulated very well at all with the KS wing system, so I just ignore it.

Yaw is more complicated and I don't believe you can really do it with the wing system. At high yaw angles, we don't even know what is happening most of the time. Even then, the longitudinal movement component doesn't seem to be as dramatic as you insinuate it to be: unless you have some data you might want to share with us.

No, Kunos cars aren't as accurate as they could be aero wise. Most of them don't even have wing endplates as fins. Your method arguably isn't much more accurate either in the end: where are you getting all of this data from to be so sure that you have it right? CFD? Do you wind tunnel your cars? The yaw behavior especially changes a lot from car-to-car in my anecdotal experience.

Shader patch introduced some aero maps, maybe you will want to take a look. I think they would produce an overall closer result in the end given you have the data apparently.

If you only care about the final CL and downforce amount, and the position, and it doesn't have to be dynamic: yes, you can derive it with just one wing!

 

[Ghoults]

Wasn't it also advised to use as little aero surfaces as possible for performance reasons? With every additional wing you add more calculations to the physics. (...or was this only regarding active aero and animation...)

While optimizations are always important (things like body lift can be easily baked into wings with no physical inaacuracy) it is also important to remember that lots of these rules of thumb come from 2015-2016. And no doubt some of these limitations could be also somewhat based on console limitations. I don't know if the console versions used the same kn5s as the pc version (or if the shaders were the same etc..) but those rules of thumbs are also 3 years old at this point.

The average ac player pc has more processing power (can handle more drawcalls, tris and materials and wings) now than it had in 2016 so I would not obsess too much about getting under those 2016 limits. That being said they are still good targets to try to match and it is not good idea to go above unless you know it is the optimal solution to get higher quality result. And they are not hard limits to get under in most cases either. A sign of good artists is not just that he or she can work fast and create good looking stuff but the stuff is also optimized well. For example while a modern pc can run even non-lod cars it is not a good idea to release a mod like that. But getting under 45mb limit I'd not worry about too much. As far as I know that limit is based on some kind of calculation for minimum pc or console requirements and memory limits. Not some kind of ac engine limitation.

So while modern pc can probably do 20 wings I'd say that is also poorly optimized aero.ini. While it may sound like you get more accuracy if you can place all tiny winglets precisely to their positions in reality you are not getting any precision out of. For example if you take something like diveplanes on front which are not ground height affected at all really then using 6 wings for each 3 on both sides is total waste. 1 wing for all dive planes does just as well of a job.

On the other hand front splitter/diffuser does benefit from running 2 wings instead of one because the car rolls in corners and while ac does not simulate the effect of the roll-ground height relationship of the wing (there is no lut for roll angles for wings like there is for ground height and angle of attack) the roll itself still affects the ground height of the wings. And because the front splitter is heavily affected by ground height doing 2 wings does give you more precision and realism here. But going from 2 to 3 again does not give you anything really. Technically a car with no anti dive could nose down heavily under braking so adding the third wing in front might help catch that. But real race car designers are aware of those issues and unless you know the car has an issue like that then putting the third wing there could add something to the car it does not have. For example if you look at ford gt lm gte car it has raised front splitter in the middle to counter this issue.

 

[Kyuubeey]

The larger issue IMO is that the forces will, without a doubt, be incorrect in most cases with so many variables UNLESS you have very good data for many surfaces on the same car that you can cross reference.

I would really like to ask the OP what kind of data he is doing this with. I don't think there is any by looking at it, because if there was, he would understand that for example the area that can be considered the CoP of the windows, especially the rear window where the flow often detaches, is not where the actual "window surface" is. We're not placing surfaces here, just centers of pressure.

As an anecdote from myself, on a Porsche 911 with no spoiler for example, the point that can be considered the collective rear axle lift CoP according to wind tunnel and CFD data was so far rearward that I wouldn't have guessed it there without the data: but computationally it made sense when paired with all of the other data as well so I believe it. *shrug*

It's not particularly bad if you add 1 or 2 more wings, for example if you "add on" an underbody piece or a decklid spoiler. I've done it before. But the reason you do that isn't more accuracy in practice: it just depends how your data is presented and how you can best achieve plausible pitch and height sensitivity *for the end result* IMO. This is from a perspective of making roadcars without fancy aero maps or anything to work from most of the time.

 

[rpmmod]

kyuubeey : I never said my method is perfect but look at the total cl of kunos Z4 GT3 or 911 GT3 R, 2.5 it's absolutely unrealist, mine have a little more than 1.1 and the grip is roughly equivalent, so it's clear that the "traditional" method is far to be good.
The longitudinal movement component is important on tight turns with some cars like Z4.

I don't understand everything you and Ghoults said but I will take time and use google translate because the interact with the ground, the heigth is important.
I know I have things to learn, but it's clear that my cup it's more realistic than the original on tight turn and when drift, and for hillclimb it's very very important.

Ghoults and kyuubeey : thanks for the tips.

 

[Kyuubeey]

Aero really doesn't have a dramatic effect at low speed, of 50km/h or so as you're insinuating. There's so much wrong with the KS cars suspension wise that I'd rather look at that.

 

[grmblfx]

So after seeing you dissing one of the most appreciated AC modders on your first post, how high do think the probability is someone into the subject is going to download this cherry?

 

[rpmmod]

mclarenf1papa : I started the mazda 747b to compare and it seems more difficult than traditinal car with my method, no doubt you have good skill to do this kind of car.

I propose a challenge : doing the Z4 GT3 and / or the Audi S1 E2 with realistic total CL and "traditional" method.

 

[Kyuubeey]

Step 1: Make 2 wings
Step 2: Place wings over axles
Step 3: Put appropriate CL over axles

That's all there really is. If you wanna be more complicated, you can, and probably should.

I don't understand what part of "you can put anything you want into a lookup table" that you don't understand. You can achieve any CL that the engine will allow you to input.

 

[rpmmod]

Kyuubeey : I work on various rally cars and the rear / front balance of downforce is important for not understeer or oversteer, and the side drag is very important.

grmblfx : he's the one who attacked me, I defend myself, alone. I would prefer to have good relations whith experienced modder.