• Welcome to PiBoSo Official Forum. Please login or sign up.
 
March 28, 2024, 03:42:57 PM

News:

GP Bikes beta21c available! :)


Rider Ragdoll and realtime tyre deformation

Started by Manu, September 04, 2017, 05:10:40 PM

Previous topic - Next topic

Hawk

Quote from: passerBy on March 01, 2019, 07:16:02 AM
May I add that the tyre deformation would be even better coupled with a completely new tyre model? Are we going to use the "Magic Formula" to death?

And where would Piboso get the data from for this new tyre model you suggest?  :)

Tyre model data is strictly protected by teams and tyre producers.... The chance of getting hands on real data for such a thing is practically zero.

The "Magic Formula" tyre algorithm is the most efficient and comprehensive model available; it's just all about the accuracy of the data to put into it that can be an issue.  :)

passerBy

Quote from: Hawk on March 01, 2019, 12:06:38 PM
And where would Piboso get the data from for this new tyre model you suggest?  :)
Probably there where Scawen Roberts got the data for his? :) More than 15 years ago, might I add...

QuoteTyre model data is strictly protected by teams and tyre producers.... The chance of getting hands on real data for such a thing is practically zero.

The "Magic Formula" tyre algorithm is the most efficient and comprehensive model available; it's just all about the accuracy of the data to put into it that can be an issue.  :)
You are contradicting yourself there, Hawk. How come Pacejka's model doesn't need any "strictly protected" data, but brushes and other models do?
If you really suppose that the "Magic Formula" is that comprehensive and efficient, why even for cars you prefer rFactor 2? Though I might be wrong, but I'm pretty sure I saw you promoting rF2. I fully agree that it's a great sim. Mostly because of its tyre model.

I never said it was an easy feat to develop a new tyre model. It would take a lot of research. But judging by all the odd sideways sliding that still exists in GPB, I can't help but wonder if Pacejka's model really is suitable for bikes. I mean, it's not perfect even for cars.

In a way, a deformable tyre model should be easier to implement for a bike than for a car. Also it should be less taxing to compute real time.

Hawk

March 04, 2019, 02:33:32 PM #17 Last Edit: March 04, 2019, 02:40:24 PM by Hawk
Quote from: passerBy on March 04, 2019, 02:18:30 PM
Quote from: Hawk on March 01, 2019, 12:06:38 PM
And where would Piboso get the data from for this new tyre model you suggest?  :)
Probably there where Scawen Roberts got the data for his? :) More than 15 years ago, might I add...

QuoteTyre model data is strictly protected by teams and tyre producers.... The chance of getting hands on real data for such a thing is practically zero.

The "Magic Formula" tyre algorithm is the most efficient and comprehensive model available; it's just all about the accuracy of the data to put into it that can be an issue.  :)
You are contradicting yourself there, Hawk. How come Pacejka's model doesn't need any "strictly protected" data, but brushes and other models do?
If you really suppose that the "Magic Formula" is that comprehensive and efficient, why even for cars you prefer rFactor 2? Though I might be wrong, but I'm pretty sure I saw you promoting rF2. I fully agree that it's a great sim. Mostly because of its tyre model.

I never said it was an easy feat to develop a new tyre model. It would take a lot of research. But judging by all the odd sideways sliding that still exists in GPB, I can't help but wonder if Pacejka's model really is suitable for bikes. I mean, it's not perfect even for cars.

In a way, a deformable tyre model should be easier to implement for a bike than for a car. Also it should be less taxing to compute real time.

I think your misunderstanding passerby..... The Pacejka's model is an algorithm that needs data to give a resultant action; that data needs to be accurate to give an accurate action, you see what I'm saying? :)
Basically if someone puts guess-work data into it then it's likely you'll get unrealistic action from the tyre physics; that doesn't mean that Pacejka's model is inaccurate or not suited for the task.
Any bike author has access to alter the tyre physics data to suit what they feel is best.... hence some tyre-physics are not as realistic as they should be because it's mainly guess work and not based on real data cause real data is almost if not even impossible to get hold of..... so we do our best based ideally on real-life racing experience, a lot of testing, and feel.  :)

You can even try altering the tyre physics data yourself and see if you can come up with a better reacting tyre model..... would be interesting to see what you can do.  :)

passerBy

Quote from: Hawk on March 04, 2019, 02:33:32 PM
I think your misunderstanding passerby..... The Pacejka's model is an algorithm that needs data to give a resultant action; that data needs to be accurate to give an accurate action, you see what I'm saying? :)
Please stop the patronizing act unless you are deliberately trying to offend someone who might very well be older than you are. Also there's no need to state obvious things. With that out of the way, let's continue.

Any algorithm needs reasonable data to produce reasonable results. This is plain obvious.

QuoteBasically if someone puts guess-work data into it then it's likely you'll get unrealistic action from the tyre physics; that doesn't mean that Pacejka's model is inaccurate or not suited for the task.
Any bike author has access to alter the tyre physics data to suit what they feel is best.... hence some tyre-physics are not as realistic as they should be because it's mainly guess work and not based on real data cause real data is almost if not even impossible to get hold of..... so we do our best based ideally on real-life racing experience, a lot of testing, and feel.  :)
You can put the most accurate data into it and it will still fail to produce realistic result. Simply because it's a very crude approximation.
There is a good analogy. Put the exact data you obtained from wind tunnel testing into MSFSX. Will you get a perfect simulation? Of course not. Will you get a decent simulation? Still not really. Because MSFSX' approach to aerodynamics modeling is obsolete, to say the least. These days there are more advanced approaches using either Blade Element Theory or whatever Eagle Dynamics use in their product.

Now, please answer the question I have already asked previously: do you prefer rFactor 2's tyre model to that of, say, the original rFactor or not?

QuoteYou can even try altering the tyre physics data yourself and see if you can come up with a better reacting tyre model..... would be interesting to see what you can do.  :)
Why would I waste my time fiddling with the very ancient tyre model I'm so fed up with already? But if you are asking me to develop my own brush-based or an entirely different model, well, why not ask me to develop my own bike sim from the ground up instead? I'm not sure I'd have enough time to complete it, so I think I'd rather invest my time in other things. Maybe one of these days though...

Blackheart

Just for info also BSE will use the Pacjecka's magic formula.

passerBy

Quote from: Blackheart on March 05, 2019, 10:05:58 AM
Just for info also BSE will use the Pacjecka's magic formula.
I figured they would. The more the reason to implement something better and more advanced for GB Bikes.