GP Bikes beta8

[mikekimi]

Probably this http://forum.piboso.com/index.php?topic=3103.msg50698#msg50698

Try it at Brands Hatch and you will see true ugliness crab style. But in Beta 7 you couldn't get round the corners much, so its better to ride but looks awful lol.
Edit : Here is the full discussion http://forum.piboso.com/index.php?topic=3129.msg49169#msg49169

yeah!!!that is what i mean! thx nick!
!

[Vini]

https://www.youtube.com/v/8h89ZM5uI9Y

Not really important but could you shortly list the criteria for when a lowside is detected?

[PiBoSo]

Not really important but could you shortly list the criteria for when a lowside is detected?

It would be much better to create a thread in the bug reports section.
This bike mod seems to have wrong data for the tyres and torus radius.
The criteria has already been described: it's the distance of the physics rider from the ground. However, the rear tyre must also be touching the ground. Maybe this latest additional check is not needed.

[doubledragoncc]

@Piboso

Hi P, while your looking om the forum at moment can you please tell me if it is possible to reduce opponents bike volume as they are near you? Is there a way to reduce this in any way in the scl file or any files.

It is really a problem for me to hear my own motor during racing. It is not just me but others have this problem too and it causes crashes in the race, not core lol.

Much appreciation for any response, thanks

DD

[Vini]

The criteria has already been described: it's the distance of the physics rider from the ground.

Ah, sorry, I thought there was more to it.
There certainly should be because the rider distance to the ground is not always related to the bike lean angle (e.g. when leaning over curbs).

This bike mod seems to have wrong data for the tyres and torus radius.

Yes because it's completely unrideable otherwise.

[HornetMaX]

This bike mod seems to have wrong data for the tyres and torus radius.

Yes because it's completely unrideable otherwise.

Maybe but then you can't complain if things go wrong. A bad torus radius is pretty bad.
I don't even see how this could help in making it more rideable.

[Vini]

They go wrong anyway (with any bike).

If I'm not mistaking it's Manu's attempt of avoiding the rider eject bug as much as possible.
Obviously this is not a solution but at the same time it's also not the cause of the problem.
...Maybe I don't remember it correctly but ultimately it doesn't matter anyway.

GPB just needs a more logical lowside detection algorithm.

[HornetMaX]

OK so let's show it on the stock bikes, it just takes a bunch of factors out of the equation.

BTW, are we discussing "lowside detection" (I guess that means when to separate the rider from the bike) or the "rider gets ejected" problem ?

[PiBoSo]

They go wrong anyway (with any bike).

If I'm not mistaking it's Manu's attempt of avoiding the rider eject bug as much as possible.
Obviously this is not a solution but at the same time it's also not the cause of the problem.
...Maybe I don't remember it correctly but ultimately it doesn't matter anyway.

GPB just needs a more logical lowside detection algorithm.

Again, looks like some modded bikes lean too much.
If you watch real footage, they reach max lean ( around 65 ) only for a fraction of a second, and for sure not on corners with a steep, half-meter tall, kerb on the inside.
Are the laptimes of these modded bikes realistic?

[Vini]

Isn't it the same thing?
The rider ejection happens because a false lowside is detected if I understood PiBoSo correctly.

Stock bikes:
Second half of the first video
This video is followed by the discussion on whether it is actually a bug

@PiBoSo: I thought we have been over this. Maximum lean angle of the bike and incorrect/illogical lowside detection are separate problems.
The fact that the rider decides his bike has lowsided when he is running out of space (in his current body position) and consequently ejects off instead of moving closer to the bike simply makes no sense.
Also, real MotoGP riders only hitting the maximum possible lean angle for a very short period of time is simply the result of sitting on a real bike as opposoed to sitting in front of a computer screen.
The virtual rider that keeps the bike at it's maximum possible lean angle all the time (at full lean input) is the reason for this "unrealistic riding style" in GP Bikes.
Besides, I think extremely smooth riders like Lorenzo keep the bike at maximum lean for quite some time, especially now with the improved edge grip of the Michelin tyres.
Manu's bikes lean a bit too much at some corners but this only happens at banked turns which suggests that there is still something wrong with GPB uphill/downhill physics.
The result: Laptimes are pretty close on relatively flat tracks. Even Mugello is quite accurate when you have a "more realistic" riding style.

.....

[Meyer#12]

They go wrong anyway (with any bike).

If I'm not mistaking it's Manu's attempt of avoiding the rider eject bug as much as possible.
Obviously this is not a solution but at the same time it's also not the cause of the problem.
...Maybe I don't remember it correctly but ultimately it doesn't matter anyway.

GPB just needs a more logical lowside detection algorithm.

Again, looks like some modded bikes lean too much.
If you watch real footage, they reach max lean ( around 65 ) only for a fraction of a second, and for sure not on corners with a steep, half-meter tall, kerb on the inside.
Are the laptimes of these modded bikes realistic?

So true Piboso, bikes lean a bit too much over longer time than in real life.

On Mugello we go a we go a bit faster than they did last year with the Duc GP15. It is a few seconds, so not too much to be honest. I haven't tested on other tracks, but i could imagine we would be a bit faster than the MotoGP riders times there.

[PiBoSo]

Isn't it the same thing?
@PiBoSo: I thought we have been over this. Maximum lean angle of the bike and incorrect/illogical lowside detection are separate problems.
The fact that the rider decides his bike has lowsided when he is running out of space (in his current body position) and consequently ejects off instead of moving closer to the bike simply makes no sense.
Also, real MotoGP riders only hitting the maximum possible lean angle for a very short period of time is simply the result of sitting on a real bike as opposoed to sitting in front of a computer screen.
The virtual rider that keeps the bike at it's maximum possible lean angle all the time (at full lean input) is the reason for this "unrealistic riding style" in GP Bikes.
Manu's bikes lean a bit too much at some corners but this only happens at banked turns which suggests that there is still something wrong with GPB uphill/downhill physics.
The result laptimes that are pretty close on relatively flat tracks. Even Mugello ist quite accurate when you have a "more realistic" riding style.

It is physically correct to have more lean in banked turns.
It is possible to automatically lean the rider less. Experiments will be done in Beta9.
However, if there is a heavily banked corner with a tall kerb with opposite camber on the inside and the bike is leaning at the max, there is nothing to do.
Lowside detection can for sure be improved, but looks like there is ALSO a problem with bikes leaning too much on tracks that are not properly made.

[Vini]

It is physically correct to have more lean in banked turns.

Obviously but not to the degree it's happening at the moment with Manu's bikes. They reach around 6162.5° on a completely flat track with extrasoft (61 with hard) so if you think they lean too much in the banked corners, GPB physics is off.

It is possible to automatically lean the rider less. Experiments will be done in Beta9.
However, if there is a heavily banked corner with a tall kerb with opposite camber on the inside and the bike is leaning at the max, there is nothing to do.

But the bike doesn't care how much space the rider has. A lowside should only be the consequence of the bike losing grip. The rider should be completely irrelevant in the lowside detection.
Maybe something like if the front wheel is sliding for more than one second: Lowside. I realize that this purely physical detection is a bit harder to implement but including the rider in it is a dirty hack.

Lowside detection can for sure be improved, but looks like there is ALSO a problem with bikes leaning too much on tracks that are not properly made.

Sure but if you look at the pictures I posted, MotoGP riders don't have more space in corners with high curbs than the virtual rider had in this or this rider ejection clip (not that this should be relevant, anyway).

[PiBoSo]

It is physically correct to have more lean in banked turns.

Obviously but not to the degree it's happening at the moment with Manu's bikes. They reach around 61° on a completely flat track so if you think they lean too much in the banked corners, GPB physics is off.

It looked much higher than 61 in some shots, but probably it was the banking.
It very likely the physics rider is moving to the sides too much, and maybe the detection point is not correct, either.
Hopefully this issue will be solved for Beta9.

[Vini]

I corrected the value because I tested it with hard tyres (those were hard tyres btw). For extrasoft it's 62.5°. Yes, it is the banking.


It would certainly be a game changer to not be afraid of toggling the ejection seat when turning into the apex.