GP Bikes beta8

[HornetMaX]

Yeh but I see Vins point he is saying it may just be visual and this vid supports that. If the physics was behaving like that the bike wouldn't go round a track?

It can't be only visual: the way the bike and its wheels are aligned with respect to the trajectory is not compatible with real life physics. So no, it's not only visual. Not by a mile. And the video shows it's definitely *not* only visual.

Then anybody is free to say it's more or less important than this or that other bug, but if one says this problem is visual only, then he's wrong.

The rider jump-off bug only started when Piboso implemented the knee slider contact and reaction to the knee colliding with the track. So yes and from the evidence provided in video by Vin and Blacky, the knee collision routine is the routine at fault for this bug in my opinion.

I am 100% sure of having seen it in beta3, on a very short french track (could be Issoire, can't remember for sure).
At the time the explanation was "something is scraping against the non-flat kerb --> sudden hard collision --> sudden spike in acceleration felt by the rider --> rider flies away".

[Vini]

@Stout
JamoZ described it well.
Now, what I was angry about was the fact that a bug that is potentially very easy and quick to fix and that is braking the game when you want to do proper laps, may not be getting fixed.
I know that you didn't say it like that but given PiBoSo's (lack of) reactions and now his new priority list, I worry that it might happen.

[PiBoSo]

At the moment it is the reason for the bike understeering in sloped-down corners.

Are you sure it is actually doing that, though? Have you calculated the maximum possible corner speed for a given corner and checked to see if GPB is simulating it correctly?
It still looks like the bike is understeering but if the corner speed (and lean angle) is correct, then it would be a visual bug and the physics would be correct.

No vin, it's a bug. And a terribly ugly one for a sim like GPB. I mean, if a Milestone game was doing that, we would be all here pissing on it (video posted by janaucarre here: http://forum.piboso.com/index.php?topic=3129.msg49169#msg49169, look the second half in particular):

It's not a bug. It's a limitation of the simulation model, that hopefully will be improved soon.
A simulation model can have flaws when the input data is outside its working range, but still be extremely accurate in all other conditions.
Example: the Pacejka formulas don't work well with very high camber angles, so a new set had to be developed specifically for bikes.

Milestone games, on the other hand, don't have a simulation model at all, so they are NEVER giving an accurate or even remotely realistic output.

[PiBoSo]

the rider's knee enters into the curbs (this is the bug)

I'm pretty sure it's the knee colliding with the bike geometry (like Hawk first said some weeks ago) instead of the rider colliding with the track.
If the knee/rider could enter (clip into) the curb, the bug would probably be fixed but the nicer solution would definitely be to allow the rider to clip into the bike (and keeping the sliding knee animation) or to automatically reduce the rider lean so that he has enough space (more difficult coding-wise, though).

The rider jump-off bug only started when Piboso implemented the knee slider contact and reaction to the knee colliding with the track. So yes and from the evidence provided in video by Vin and Blacky, the knee collision routine is the routine at fault for this bug in my opinion.

Hawk.

Again, not a bug, and not related to the knee animation, that is purely visual.
The behaviour is due to the new code to detect a low-side.
It's not bugged, because it works as it was designed to, that is to measure the distance of the rider from the ground.
Problem is: if the bike is leaning 65 degrees or more, and the rider is leaning further, and on the inside there is a very high curb, then the rider thinks it's a low-side.
And in some way, it is! Those mod bikes are probably leaning too much, the tracks are not smooth and accurate enough, and some of you expect too much from the physics...

A possible solution would be to make the rider automatically lean less when there is an obstacle on the inside.
However, you really cannot expect to lean at 70 degrees all the way around any turn of any track... Please check some real life footage!

[HornetMaX]

It's not a bug. It's a limitation of the simulation model, that hopefully will be improved soon.

Yeah, proper wording is as you said of course. For the average player, it's a just bug (even if it's not).
Glad to (re)hear it will be worked on.

Milestone games, on the other hand, don't have a simulation model at all, so they are NEVER giving an accurate or even remotely realistic output.

You don't need to convince me about that. The point was that if something as the behaviour under discussion here happens in a Milestone game, then it's kind of expected while if it happens in GPB it's a major issue, because it undermines GPB's strong point (physics accuracy).

[Hawk]

Thank you for the explanations Piboso.... Helped me understand more about what is going on..... Really appreciated! 

Hawk.

[Napalm Nick]

Yeh but I see Vins point he is saying it may just be visual and this vid supports that. If the physics was behaving like that the bike wouldn't go round a track?

It can't be only visual: the way the bike and its wheels are aligned with respect to the trajectory is not compatible with real life physics. So no, it's not only visual. Not by a mile. And the video shows it's definitely *not* only visual.

Then anybody is free to say it's more or less important than this or that other bug, but if one says this problem is visual only, then he's wrong.

  I think I am with you and I don't think it is *Only* visual either. What I was suggesting is that maybe the replay visually doesn't respond to the physics. Another example - the replay of the bike going into and around a corner - in the replay the front (or even back) wheel comes up floating in the turn. Sure is visual because the bike is not doing that in any live view and it certainly doesn't crash like the replay suggests you should?
But yeh in this case to say it is ONLY visual would be unlikely but replays definitely cant be trusted to see what goes on.

[Blackheart]

The offline replay is accurate, only the online replays (for the lag) did wrong movements. Mine vid was a offline lap 100% accurate 

So replays are perfect to notice the physical problems. 

[Stout Johnson]

@Piboso: thank you for taking the time to give some feedback here. Did you happen to have the time and look into the cause of the tyre temps/tyre grip/rider lean which I posted here?

[Vini]

A possible solution would be to make the rider automatically lean less when there is an obstacle on the inside.
However, you really cannot expect to lean at 70 degrees all the way around any turn of any track... Please check some real life footage!

A bit too much lean angle at some corners would still be better than completely unpredictable behaviour that ruins your lap or even your complete race.
Thanks for giving us that information, PiBoSo.
Could you list the low-side detection criteria of beta8?

[Mace-x]

The  bike understeering in sloped-down corners bug is present on mxb too, hopefully that can be sorted out in b5 since it´s a major limitation and looks super weird 

[BOBR6 84]

The  bike understeering in sloped-down corners bug is present on mxb too, hopefully that can be sorted out in b5 since it´s a major limitation and looks super weird 

Yes, exactly!  would really like to see this sorted or improved/fixed.. Even before all the online/netcode stuff (i know.. im alone in this opinion.) But now.. we have to wait for a GPB release with improved online/netcode so we can all play with eachother  before we get a realistic bike to ride with.

[HornetMaX]

Yes, exactly!  would really like to see this sorted or improved/fixed.. Even before all the online/netcode stuff (i know.. im alone in this opinion.) But now.. we have to wait for a GPB release with improved online/netcode so we can all play with eachother  before we get a realistic bike to ride with.

Not entirely true. On tracks with no to moderate banking, the bike is already realistic.
On the other hand, the stability and online issues happen on every bloody track

[Stout Johnson]

Yes, exactly!  would really like to see this sorted or improved/fixed.. Even before all the online/netcode stuff (i know.. im alone in this opinion.) But now.. we have to wait for a GPB release with improved online/netcode so we can all play with eachother  before we get a realistic bike to ride with.

Not entirely true. On tracks with no to moderate banking, the bike is already realistic.
On the other hand, the stability and online issues happen on every bloody track

Exactly. The netcode issue is really happening to everybody to some extent or much extent, when dynamic track features are on. So imo it is the single most annoying issue.

I agree though, that the understeering in sloped-down corners is something that should be fixed rather fast. But it depends on the track to which extent the limitation does effect gameplay. But many even don't seem to dectect it, so it definitely ranks lower imo...

[doubledragoncc]

Well I am sorry but IMO it is that you can ACUALLY ride the bike that is most important. If the front end wobble that is still VERY bad no matter what excuse you say, it is NOT realistic and kills ALL aspects of riding be it online or off. What point is there in netcode if you still cant really ride the bike hard.

Netcode is ONLY for those who want to race online(me included) BUT, ride-ability is for ALL of us.

The bikes ride-ability should be the foremost point especially to get new users and sales. The demo bike is putting people off buying a license, I know many have told me through my YouTube channel alone!!!

I try to promote GPB but cant show the 125 as it is washing out to much but it is the demo bike!!!

New customers want to see a ride-able bike and then there will be more money for P to work with to help us ALL.

DD