• Welcome to PiBoSo Official Forum. Please login or sign up.
 
March 28, 2024, 08:29:56 PM

News:

GP Bikes beta21c available! :)


front wheel alignment - 2

Started by Stout Johnson, February 04, 2014, 01:24:09 AM

Previous topic - Next topic

Stout Johnson

February 04, 2014, 01:24:09 AM Last Edit: February 04, 2014, 01:53:49 PM by HornetMaX
well I was thinking about that front/rear wheel alignment bug I already reported here http://forum.piboso.com/index.php?topic=410.0 it does look bad in replays and might also be an indicator of some inconsistent physics... (I was thinking it might even be connected to the bad physics on camber corners...)
    -----------   WarStout Kawasaki Team   -----------

HornetMaX

Quote from: Stout Johnson on February 04, 2014, 01:24:09 AM
well I was thinking about that front/rear wheel alignment bug I already reported here http://forum.piboso.com/index.php?topic=410.0 it does look bad in replays and might also be an indicator of some inconsistent physics... (I was thinking it might even be connected to the bad physics on camber corners...)
Saw the video and to me there's nothing really disturbing.

Even when the front wheel is firmly pushed against the track, the front wheel can (and will) slide laterally and hence can "point" towards a direction that is not colinear to the "overall direction of the bike" (things between quotes are easy-wording of stuff that would be a bit tedious to word in the correct manner).

In particular, when you see the bikes in the chicane (from 2:19) there's nothing wrong IMO: in a quick left/right the sudden change of lean will lower the load on the wheels (due to centrifugal force) and if you have the throttle open, this will affect more the front wheel which will become very "light", hence the seemingly unnatural movements of the front wheel.

Do you have an example where that is more visible/disturbing ?

In the one in your topic, the only thing that is slightly annoying is that the "animation" is not as fluid as we'd like it to be, but that's normal (server ping and delays) and it is even more visible as the video is in slow motion.

MaX.

C21

I think that Stout is complaining about the front wheel.
It´s not on the surface of the track.
Look carefully at 2:13-2:19, fullscreen and slowmo.
You can see the front wheel not flat on the surface but a bit up in the air.
And that on a specific point on the track where definitve is load on the front, so the front wheel should be down on surface. ;)
# Member of the CAWS Racing Team #


HornetMaX

Quote from: C21 on February 04, 2014, 11:26:18 AM
I think that Stout is complaining about the front wheel.
It´s not on the surface of the track.
Look carefully at 2:13-2:19, fullscreen and slowmo.
You can see the front wheel not flat on the surface but a bit up in the air.
And that on a specific point on the track where definitve is load on the front, so the front wheel should be down on surface. ;)
In his referenced post he says that:
QuoteWhen watching replay, I noticed that sometimes the front wheel alignment does not correspond with the direction of where the bike is headed.
Doesn't sound like a wheel-track contact thing.

MaX.

Stout Johnson

Quote from: C21 on February 04, 2014, 11:26:18 AM
I think that Stout is complaining about the front wheel.
It´s not on the surface of the track.
Look carefully at 2:13-2:19, fullscreen and slowmo.
You can see the front wheel not flat on the surface but a bit up in the air.
And that on a specific point on the track where definitve is load on the front, so the front wheel should be down on surface. ;)

well, that is not really what I mean - but I think that those 2 problems are connected...

Quote from: HornetMaX on February 04, 2014, 10:01:00 AM
the only thing that is slightly annoying is that the "animation" is not as fluid as we'd like it to be, but that's normal (server ping and delays)
well I am aware of that :) that has nothing to do with what I try to describe...

Quote from: HornetMaX on February 04, 2014, 10:01:00 AM
In particular, when you see the bikes in the chicane (from 2:19) there's nothing wrong IMO: in a quick left/right the sudden change of lean will lower the load on the wheels (due to centrifugal force) and if you have the throttle open, this will affect more the front wheel which will become very "light", hence the seemingly unnatural movements of the front wheel.
am also aware of that (not a physics noob :D)... what you describe is realistic behavior and it should be there. but I try to point to sth else...

Quote from: HornetMaX on February 04, 2014, 10:01:00 AM
Do you have an example where that is more visible/disturbing ?
I will try to make it more visible...
    -----------   WarStout Kawasaki Team   -----------

HornetMaX

Quote from: Stout Johnson on February 04, 2014, 12:05:25 PM
am also aware of that (not a physics noob :D)...
I know, but I doubt we can say the same for many others, hence the details :)

OK, let's wait for your video (or replay, it's fine too).

MaX.

Stout Johnson

February 04, 2014, 12:58:19 PM #6 Last Edit: February 04, 2014, 01:13:57 PM by Stout Johnson
yeah, I will try to make a video in order to make it more apparent.

in the meantime, maybe it will also be clear if I try to explain the problem I observed it in more detail:
to make it not too scientific from the start  ;) --> this is how a 2 wheel vehicle and its wheels behave while turning



in GPB it rather looks like this


This would only be possible if you would oversteer/powerslide (Stoner style)... but in the replay there is no powerslide visible. It rather looks like the bike is just being pushed by an invisible force towards the inner of the turn, sort of "sliding" over the front wheel which just does not correspond with rl physics.  ???

As a long shot for explanation: maybe GPB is computing powerslide behaviour realistically, but it is not represented graphically?! because I think that this bug does only appear (or at least is most noticeable) in long, fast corners with medium/much lean and under rear load - just guessing....

Has it become more clearer what I tried to point at? if you watch the video from 2:13 to 2:19 keeping in mind what I just posted, maybe you will also notice the problem...
    -----------   WarStout Kawasaki Team   -----------

Arvoss

I'm not really following the posts about this bug BUT remember that this is a replay footage. It's pretty laggy so maybe that could explain some weird movements.

HornetMaX

Quote from: Stout Johnson on February 04, 2014, 12:58:19 PM
yeah, I will try to make a video in order to make it more apparent.

in the meantime, maybe it will also be clear if I try to explain the problem I observed it in more detail:
to make it not too scientific from the start  ;) --> this is how a 2 wheel vehicle and its wheels behave while turning
Please do make it more scientific from the start: it's the only way out, otherwise we'll be talking about feelings and perceptions :)

Concerning the bicycle video (I'm saying this for the benefit of the others) notice that the motion represented is with zero lateral sliding of the tires. Also, the overall geometry may be different from the one of a bike (thin tires !) and finally that the lean angle is tiny. All these things do affect what the front wheel does in a turn.


Quote from: Stout Johnson on February 04, 2014, 12:58:19 PM
This would only be possible if you would oversteer/powerslide (Stoner style)... but in the replay there is no powerslide visible. It rather looks like the bike is just being pushed by an invisible force towards the inner of the turn, sort of "sliding" over the front wheel which just does not correspond with rl physics.  ???

As a long shot for explanation: maybe GPB is computing powerslide behaviour realistically, but it is not represented graphically?! because I think that this bug does only appear (or at least is most noticeable) in long, fast corners with medium/much lean and under rear load - just guessing....

Has it become more clearer what I tried to point at? if you watch the video from 2:13 to 2:19 keeping in mind what I just posted, maybe you will also notice the problem...

Not sure, let's see: essentially you're saying that in the video from 2:13 to 2:19 it seems like the bike (the 1st one I guess) is globally (i.e. not only the front or the rear) sliding towards the center of the turn. Is that right  ?

I don't see it that much in the video, I'd say thee's some hint of it but it could be easily be generated by the banking of the turn or maybe even by camera positioning/fov/etc.
Also, if I watch it a 3x the speed (which is probably still below its original speed) the thing disappears totally (at least for me). I'm tempted to say that it's just something that tricks our perception, most likely the slow-motion combined with the laggy replay (brain is very sensible to lag, choppy animations etc because it's always trying to predict the future trajectory).

This is a good slow-motion, for comparison: http://www.youtube.com/watch?v=-YAiXqOwkmE. Notice that to have that quality, they had to use high speed cameras, otherwise, normal footage played that slow will just look bad.

If the replay is still around, that would be ideal to dig into it.

If you try (and manage) to reproduce, keep the replay file at hand.

Other note for the "others": GPB does not compute powerslides or other behaviors (e.g. high-sides) explicitly. Everything comes from the physical model, with no "hardcoded" stuff.
For example, other games just have something that, oversimplifying a bit, says "if you are in 4th gear and more then 45deg lean and more than 70% throttle open, then slide the rear". GPB does nothing like this: if the bike slides it's because the physical model says so (and not because Piboso or whoever has decided it should and coded accordingly).

MaX.

P.S.
I've splitted the topic, so that we can discuss it here fully.

Stout Johnson

Quote from: HornetMaX on February 04, 2014, 02:23:00 PM
Concerning the bicycle video (I'm saying this for the benefit of the others) notice that the motion represented is with zero lateral sliding of the tires. Also, the overall geometry may be different from the one of a bike (thin tires !) and finally that the lean angle is tiny. All these things do affect what the front wheel does in a turn.
true MaX, but it was the best visual explanation and the easiest and fastest way to describe what I (and a few others) see. Obviously it is not obvious like


And the variables you describe all affect the leaning and the behaviour of a bike while turning - but the overall principles stay the same... you can't slide inwards over the front wheel in a turn. But again, it seems not to be that obvious - if one of the other members knows what I mean, please feel free to join this discussion ;) otherwise it looks like a pointless conversation between geek and nerd  (you choose what you are MaX ;D)

Quote from: HornetMaX on February 04, 2014, 02:23:00 PM
I'd say thee's some hint of it but it could be easily be generated by the banking of the turn or maybe even by camera positioning/fov/etc.
Also, if I watch it a 3x the speed (which is probably still below its original speed) the thing disappears totally (at least for me). I'm tempted to say that it's just something that tricks our perception, most likely the slow-motion combined with the laggy replay (brain is very sensible to lag, choppy animations etc because it's always trying to predict the future trajectory).
if you see a hint of it - you might already see it. And in normal speed it isn't all that abvious. It also very much depend on camera angle to see it. I first saw it in replay when watching bikes in the turn approaching Lucky Heights in Philipp Island.

Quote from: HornetMaX on February 04, 2014, 02:23:00 PM
If you try (and manage) to reproduce, keep the replay file at hand.
It's not a matter of reproducing it. It is always there - it is rather a matter of having a corner where it shows and looking at it from a camera angle that makes it more obvious.

Quote from: HornetMaX on February 04, 2014, 02:23:00 PM
Other note for the "others": GPB does not compute powerslides or other behaviors (e.g. high-sides) explicitly. Everything comes from the physical model, with no "hardcoded" stuff.
yeah, GPB wouldn't be a sim if it was differently. But the physics of GPB, as jaw-dropping and advanced as it is, is still not near perfection.  Just take high-sides.... the beginning is always rl-behaviour, but the way the bike spins and turns afterwards is often very unrealistic. So, the fact that nothing is scripted (or hard-coded as you call it) is not a proof that there is no unrealistic behaviour ;)

to make my case I will try to make video that makes it more obvious. But I presume that if you haven't been able to spot it yet, you might still not see it and the discussion will revolve around whether someone clearly sees it, sees a 'hint' of it or nothing at all  :-\ again, if anybody could second my case here, I'd be thankful ;)
    -----------   WarStout Kawasaki Team   -----------

HornetMaX

Quote from: Stout Johnson on February 04, 2014, 03:40:41 PM
But again, it seems not to be that obvious - if one of the other members knows what I mean, please feel free to join this discussion ;) otherwise it looks like a pointless conversation between geek and nerd  (you choose what you are MaX ;D)
Easy call, as I'm both  8)

Quote from: Stout Johnson on February 04, 2014, 03:40:41 PM
if you see a hint of it - you might already see it. And in normal speed it isn't all that abvious. It also very much depend on camera angle to see it. I first saw it in replay when watching bikes in the turn approaching Lucky Heights in Philipp Island.
As it is a "trajectory thingy", viewing from up above should help.

Quote from: Stout Johnson on February 04, 2014, 03:40:41 PM
It's not a matter of reproducing it. It is always there - it is rather a matter of having a corner where it shows and looking at it from a camera angle that makes it more obvious.
Well it should then be even easier to find one particular replay and one particular camera that show this at its best.
Advice: don't do this online, use offline (replay is likely to be smoother).

Quote from: Stout Johnson on February 04, 2014, 03:40:41 PM
yeah, GPB wouldn't be a sim if it was differently. But the physics of GPB, as jaw-dropping and advanced as it is, is still not near perfection.  Just take high-sides.... the beginning is always rl-behaviour, but the way the bike spins and turns afterwards is often very unrealistic. So, the fact that nothing is scripted (or hard-coded as you call it) is not a proof that there is no unrealistic behaviour ;)
All agreed, was just to be sure people don't start thinking that GPB does weird stuff.

Quote from: Stout Johnson on February 04, 2014, 03:40:41 PM
to make my case I will try to make video that makes it more obvious. But I presume that if you haven't been able to spot it yet, you might still not see it and the discussion will revolve around whether someone clearly sees it, sees a 'hint' of it or nothing at all  :-\ again, if anybody could second my case here, I'd be thankful ;)
All I'm sayinmg is that visually we may get that impression, but in reality it may not be there. The idea behind reproducing is that, if the thing is very visible, I'll be able to reproduce it too (same cornet, same trajectory etc) and have GPB spit out some telemetry data with which I may be able to double check if "we see what we see" or not.

Notice that with the HUD plugin and the lean indicator, the black line (somehow related to bike sliding) may show something of interest.

MaX.

Stout Johnson

February 04, 2014, 06:48:56 PM #11 Last Edit: February 04, 2014, 06:53:46 PM by Stout Johnson
Quote from: HornetMaX on February 04, 2014, 05:54:01 PM
Quote from: Stout Johnson on February 04, 2014, 03:40:41 PM
But again, it seems not to be that obvious - if one of the other members knows what I mean, please feel free to join this discussion ;) otherwise it looks like a pointless conversation between geek and nerd  (you choose what you are MaX ;D)
Easy call, as I'm both  8)
LOL!

Quote from: HornetMaX on February 04, 2014, 05:54:01 PM
All I'm sayinmg is that visually we may get that impression, but in reality it may not be there. The idea behind reproducing is that, if the thing is very visible, I'll be able to reproduce it too (same cornet, same trajectory etc) and have GPB spit out some telemetry data with which I may be able to double check if "we see what we see" or not.

Notice that with the HUD plugin and the lean indicator, the black line (somehow related to bike sliding) may show something of interest.
that's a good advice! I will try to reproduce it, try to capture it visually and come back to you on that ;)
    -----------   WarStout Kawasaki Team   -----------

Stout Johnson

there are always 2 sides to classify a problem: 1) the problem probability 2) the possible extent/risk the problem can cause

I reported the behaviour I described essentially because I though there might be some physics bug laying undiscovered underneath. I particularly think that it might be connected to the bad behaviour on rough surface or camber corners --> so point 2 affected 

But in terms of point 1) I have problems making the problem visual (not much time atm) and it seems like I am the only one reporting it... so I'll leave the situation as it is. In all likelihood, I will come across that problem again sooner or later when watching replays and will report it again.
    -----------   WarStout Kawasaki Team   -----------

HornetMaX

OK Stout, no need to waste your time on that, if you stumble on it again just grab the replay.

Now that I think about it, I think I recall cases where I noticed something strange too: uphill corners with maybe even some banking (negative IIRC), the bike seems to slide a lot towards the exterior of the turn.
So it's kind of the opposite of what you were pointing. It was very visible in 3rd person view.

MaX.