+- Forums (https://www.vdrift.net/Forum)
+-- Forum: Community (https://www.vdrift.net/Forum/forumdisplay.php?fid=3)
+--- Forum: Cars & Tracks (https://www.vdrift.net/Forum/forumdisplay.php?fid=11)
+--- Thread: Rollover testing (/showthread.php?tid=1273)
Pages:
1
2
Rollover testing - NaN - 04-19-2010
Tested on rouen 60/600Hz physics.
stable: Was not able to cause a rollover
stable?: Plausible rollover behavior? (was able to force a rollover)
buggy: Was able to flip the car.
broken: Car is unplayable. Will roll on turn.
360 buggy (has a light tendency to roll over the right side)
3S stable
AX2 stable (pulls to the left at high speed?)
C7 stable? (leaving track rollover)
CO buggy
CS broken
CT broken
F1 broken (suspension looks strange, mesh needs to be scaled down)
F1-02 stable (pulls to the left, aerodynamics?)
FE broken
FF stable? (leaving track rollover)
G-P34 broken (broken transmission)
G4 broken
GT stable
GTR stable
GTV6 broken
M3 broken
M7 stable? (has a very soft suspension, still hard to force rollover, interesting)
M8 stable? (see M7)
MC broken
MI broken
NS broken
RG buggy (has oval shaped wheel meshes!?)
RS2 broken
SB stable? (might roll over, maybe intended?)
SV stable? (rollover off road)
T73 broken
TC6 broken
TL broken
TL2 broken
XG broken
XM broken
XS stable? (forced rollover possible when leaving the track)
Z06 stable? (see XS)
350Z stable? (see Z06)
- NaN - 04-20-2010
trunk rev. 2672 update
- flo - 04-23-2010
Hey guys,
what's the thing with the new "rollover"-version?
Are there bugs in the car data which has to be fixed and do not work with the new version yet or is this just a test version in the trunk directory? And which is the latest "stable" version with all cars playable?
THX,
flo.
- NaN - 04-23-2010
It's the latest trunk.
A stable revision would be 2659 I guess.
Edit:
vdrift-data revision 486
- NaN - 04-27-2010
http://www.youtube.com/watch?v=gcZBNJbu3aQ
Start to watch from 1:10.
Do we have realistic physics or what? :lol:
- joevenzon - 04-27-2010
Luckily I haven't seen it do this:
http://www.youtube.com/watch?v=SFN_Gp1eHN0
- NaN - 04-28-2010
I think we could do this too. Just need to adjust the aerodynamics a bit. Haha.
- NaN - 05-26-2010
I've been bugfixing VDrift during the last few days and hit a bug in the wheel camber calculation today. There is still some work to be done regarding tire stiffness and overtravel, but the cars should be very drivable now(r2761).
The bug has been there for the last two months since my physics rewrite. :oops:
- joevenzon - 05-27-2010
Muuuuuch better!
I can still get the CO to roll over, but not without doing maneuvers that may cause the real car to roll over too. :-)
- stan.distortion - 05-30-2010
Quote:There is still some work to be done regarding tire stiffness and overtravelCan I get some feedback on the C7 if you have time? Overtravel is handled by ramping up spring rates to simulate bump stops (should never reach the travel limit).
Will have a look at a heavier car next, having trouble getting sidewall deflection to feel right and a shoebox like the C7 isnt going to put much strain on them
If the C7 seems ok then please suggest a car. Leaving the 360 and XS alone for now.cheers
- NaN - 05-31-2010
The comment was more towards the code not the car settings. Using a spring to deal with overtravel is a hack. At present I think it doesn't belong into suspension parameters.
I've added a hardcoded stiff spring back into the suspension code to deal with excessive car roll and wheels sinking into ground on harsh maneuvers. I think I observed it with one of the Porsches. But still, the proper way would be to use a constraint. I am looking into this.
- stan.distortion - 05-31-2010
Quote:Using a spring to deal with overtravel is a hackI would beg to differ on that, a rigid stop would soon destroy a road cars suspension and chassis components and in some classes of motor sports it is the only spring modification allowed (used to be the case in some WRC classes).
An upper limit should definitely be imposed but it should only be reached in extreme cases.
- NaN - 05-31-2010
Yeah. But thats another effect. We should keep them separate. There has to be something to stop the wheel going through the car. The correct way to do it is by using a constraint(calculate the needed impulse) imho.
I've noticed that you are running the C7 with very low rotational inertia values 0.75-1. I am surprised that it is drivable at all. I remember having problems with inertia below 6. But it was a heavier car I think. Have you tried bigger values for wheel inertia (>= 10)? Does it make any difference?
[edit]
With another effect I mean simulating a suspension bump(as stiff spring).
- stan.distortion - 05-31-2010
Quote:Have you tried bigger values for wheel inertia (>= 10)? Does it make any difference?Tried higher values and the damn cars wouldn't stop
Higher values did help reduce wheelspin and lockup to some degree but didn't seem realistic, I think a progressive delay (like the button delay) with fast release would work better as a driving aid.Values above about 2 where too high imho.
Something else I have to get around to trying is stdout'ing the load values on the wheels/suspension and graphing them. I know I was getting strange results from using very high spring rate values to simulate bump stops. If the damper values are too low the wheels will hop as the physics correct themselves, from the way the tyres are acting it may be there are huge forces oscillating, there may need to be some sort of averaging to cancel them out. I'll check that before the end of the week.
Another thing with the bullet constraints, the only way I could stop them from breaking was by using large max travel values and simulating a bump stop in the .car file. I'm sure there is a way to re-enforce the constraints in bullet but my brain was melting before I got that far with them. Not looked at them since, I wasn't able to work out a clean way of getting the wheel positions for the spinning car function.
- stan.distortion - 06-01-2010
Getting hysteresis, not sure if its coming from the physics or the tyre stiffness calcs, guessing the physics as increasing the stiffness is changing the amplitude and not effecting the frequency.
Loads fluctuate between 1800 and 2000 every 10 ticks in an n shape. Couldn't add an image, can email details if anyone wants them.
EDIT halving the stiffness to 1E5 has more or less got rid of the issues I was having, tyres are way too civilised with the settings I was trying to make them to behave more predictably
EDIT2 Guessing doing this is getting around the issue rather than fixing it (bodge
) but it's made a big difference. Tried it at 5E4 and hysteresis is less than 3% (is over 10% at 2E5) and the car feels a lot more positive. Suspension and grip values have to be raised by about 10% in the .car file for the same characteristics.