Is it me, or is this the coolest thing in a long time?
IMO, there's works on suspension to do .
--
Sbastien TIXIER - Game Developer
Overall GPLRank -42.97
Monster GPLRank -106.02
Undrivable? It has problems with the bumps in the original track (carrera)
, however I suggest you try it on Surfers Paradise and you will be VERY
surprised at the difference.
Ashley
| well, i think it's almost undrivable for now, but the idea is GREAT.
|
| IMO, there's works on suspension to do .
|
| --
| Sbastien TIXIER - Game Developer
| Overall GPLRank -42.97
| Monster GPLRank -106.02
|
|
| > Is it me, or is this the coolest thing in a long time?
|
|
>Undrivable? It has problems with the bumps in the original track (carrera)
>, however I suggest you try it on Surfers Paradise and you will be VERY
>surprised at the difference.
I won't do any work on suspensions though, I'll stick to the sim
programming part now that several people are working on the cars (and
tracks).
Ruud van Gaal, GPL Rank +53.25
Pencil art : http://www.marketgraph.nl/gallery/
Free car sim : http://www.marketgraph.nl/gallery/racer/
speaking of tweaking; you know how hard it is to get any good Pacejka
numbers, but could you just include a simple multiplier (say,
Pacejka_factor ;) ) to multiply the Pacejka results (lateral,
longitudinal force and aligning torque) with? The default should be 1,
but to, say, increase the rear wheel grip (different spec tyres rear to
front) you would put this simple factor in, as something like 1.05 would
already change things dramatically? What do you think, to much of a
fudge?
-Gregor
> Yes, although I never drive there actually, there's some huge
> differences between some of the cars (some of them are indeed
> undrivable). However, from experimenting with almost all of them I can
> see that it's in there (the physics engine) and just needs the right
> numbers. Which is ofcourse an art in itself. The Evo6 is very nice, as
> is the Ferrari 312. Still, it has that GPL touchy feeling, perhaps
> even more than GPL itself. :)
> The bumps in Carrera will only get better when the polygon subdivision
> is improved I'm afraid; it is a little TOO coarse. But it's a nice
> track to test on until you move on to bigger ones.
> And indeed, suspension is important, as is CG and roll centers. The
> Evo has really soft suspensions and drives ok on Carrera.
> I won't do any work on suspensions though, I'll stick to the sim
> programming part now that several people are working on the cars (and
> tracks).
> Ruud van Gaal, GPL Rank +53.25
> Pencil art : http://www.marketgraph.nl/gallery/
> Free car sim : http://www.marketgraph.nl/gallery/racer/
From what I have seen so far of Racer, it's turing out to be something
very close to GPL and with the constent updating I believe it'll far out do
GPL in the not to distent future. (Can't wait for the multi player code:-)
Mike Barlow
>speaking of tweaking; you know how hard it is to get any good Pacejka
>numbers, but could you just include a simple multiplier (say,
>Pacejka_factor ;) ) to multiply the Pacejka results (lateral,
>longitudinal force and aligning torque) with?
And what do you think, multiply directly after calculating Fx/Fy/Mz or
after your slip combination method?
Ruud van Gaal, GPL Rank +53.25
Pencil art : http://www.marketgraph.nl/gallery/
Free car sim : http://www.marketgraph.nl/gallery/racer/
Have you tried the Controller Peeker? With this you should be able to
write down the numbers and be able to use them (as the code is largely
the same in Racer as for the peeker). So if you turn left, and see
'583' at axis X, and at center you see '10' (highly unusal numbers but
for the idea), enter:
steerleft
{
min=10
max=583
Same with right steering, and probably you'll find min=10 as well
(otherwise there would be no center) and max=-583 probably.
AI & Multiplayer will be the toughest nuts to crack. Fortunately I
think, a lot of cheaters won't be playing the game, since they'll all
be playing NFS. This would make multiplayer life quite a bit easier
(the first version will be oh so simple and have lag, syncing problems
and the whole forest of problems probably hehe).
But thanks for the compliment! :)
v0.4.8 will probably have caster, thanks to Gregor (I should subtitle
it Veble Racer, lol), mr. Formula Wizard. That's something GPL doesn't
have. :) (but N4 does, ah well) And let's throw in some toe, should be
easy enough, an offset.
Ruud van Gaal, GPL Rank +53.25
Pencil art : http://www.marketgraph.nl/gallery/
Free car sim : http://www.marketgraph.nl/gallery/racer/
I tried other tracks and other cars, it's easier to play, but IMO it's far
from GPL "gameplay".
Attention, i'm not saying your sims is craps ! i try it since some month,
because, i'm doing car physics simulation for my job, and i saw lot's of
improvement.
I just tell my opinion to help you improve.
Another question . do you have any setting to draw longitudinal and lateral
forces forces and grip like there was on older version ? it was very
intersting.
I have the feeling that a main problem in your simulation is that you can
block tyre too easily when breaking. and the main consequence is that you
loose lateral grip on front tire and just can't turn. I try brake smoother
with my guillemot FF, but din't manage not to brake too hard :o( Badly, i
didn't find where to set up them in the carlab :o(
Do you plan to make a replay ? have consider to record driver actions or
cars position/rotation ?
king regards,
--
Sbastien TIXIER - Game Developer
Overall GPLRank -42.97
Monster GPLRank -116.28
> >Sebastien,
> >Undrivable? It has problems with the bumps in the original track
(carrera)
> >, however I suggest you try it on Surfers Paradise and you will be VERY
> >surprised at the difference.
> Yes, although I never drive there actually, there's some huge
> differences between some of the cars (some of them are indeed
> undrivable). However, from experimenting with almost all of them I can
> see that it's in there (the physics engine) and just needs the right
> numbers. Which is ofcourse an art in itself. The Evo6 is very nice, as
> is the Ferrari 312. Still, it has that GPL touchy feeling, perhaps
> even more than GPL itself. :)
> The bumps in Carrera will only get better when the polygon subdivision
> is improved I'm afraid; it is a little TOO coarse. But it's a nice
> track to test on until you move on to bigger ones.
> And indeed, suspension is important, as is CG and roll centers. The
> Evo has really soft suspensions and drives ok on Carrera.
> I won't do any work on suspensions though, I'll stick to the sim
> programming part now that several people are working on the cars (and
> tracks).
> Ruud van Gaal, GPL Rank +53.25
> Pencil art : http://www.marketgraph.nl/gallery/
> Free car sim : http://www.marketgraph.nl/gallery/racer/
if not, this could be another reason why the car is so understeering and
it's VERY diifcult to keep the car on road when you manage to oversteer the
car.
We saw at office that differntial is VERY important for the car handling...
especialy on 4WD cars !
king regards,
--
Sbastien TIXIER - Game Developer
Dynamics and Car Physics
Overall GPLRank -42.97
Monster GPLRank -116.28
> >Sebastien,
> >Undrivable? It has problems with the bumps in the original track
(carrera)
> >, however I suggest you try it on Surfers Paradise and you will be VERY
> >surprised at the difference.
> Yes, although I never drive there actually, there's some huge
> differences between some of the cars (some of them are indeed
> undrivable). However, from experimenting with almost all of them I can
> see that it's in there (the physics engine) and just needs the right
> numbers. Which is ofcourse an art in itself. The Evo6 is very nice, as
> is the Ferrari 312. Still, it has that GPL touchy feeling, perhaps
> even more than GPL itself. :)
> The bumps in Carrera will only get better when the polygon subdivision
> is improved I'm afraid; it is a little TOO coarse. But it's a nice
> track to test on until you move on to bigger ones.
> And indeed, suspension is important, as is CG and roll centers. The
> Evo has really soft suspensions and drives ok on Carrera.
> I won't do any work on suspensions though, I'll stick to the sim
> programming part now that several people are working on the cars (and
> tracks).
> Ruud van Gaal, GPL Rank +53.25
> Pencil art : http://www.marketgraph.nl/gallery/
> Free car sim : http://www.marketgraph.nl/gallery/racer/
I don't think it matters where exactly the multiplication takes place,
as the result is the same. Do it as to keep the code as elegant as
possible :). Basically, I believe the same should be possible by
altering those Pacejka coefficients that deal with available grip, but
just making a common factor would be much easier, and it would keep the
'hard' (to get ;) ) Pacejka coefficients intact.
-Gregor
> Hm, pacejka_factor.fx/fy and mz probably then. A nice idea, although I
> wonder whether just scaling a2/b2/c2 isn't enough. Probably isn't
> because why would the others be there. Hm. Interesting idea, a fudge,
> but a good fudge probably for practical use. I found that editing
> Pacejka numbers quickly gave me some inconsistent skidding sounds and
> all that.
> A Pacejka viewer/editor is still needed, but the factor might be in
> the next version. Thanks for the idea. :)
> And what do you think, multiply directly after calculating Fx/Fy/Mz or
> after your slip combination method?
> Ruud van Gaal, GPL Rank +53.25
> Pencil art : http://www.marketgraph.nl/gallery/
> Free car sim : http://www.marketgraph.nl/gallery/racer/
I got the same problems some month ago, the dampers created force so big
that they lifted the car, and that's IMPOSSIBLE, dampers can only void a
suspension velocity , they can't negate them.
I remember Gregor helped me a little with the formulas but the main idea was
to predict if the force created by the dampers could negate the suspension
veolicty, if so i decrease the damper force to only void the suspension
velocity.
hope that can help
cheers,
--
Sbastien TIXIER - Game Developer
Dynamics and Car Physics
Overall GPLRank -42.97
Monster GPLRank -116.28
> >Hi Ruud,
> >speaking of tweaking; you know how hard it is to get any good Pacejka
> >numbers, but could you just include a simple multiplier (say,
> >Pacejka_factor ;) ) to multiply the Pacejka results (lateral,
> >longitudinal force and aligning torque) with?
> Hm, pacejka_factor.fx/fy and mz probably then. A nice idea, although I
> wonder whether just scaling a2/b2/c2 isn't enough. Probably isn't
> because why would the others be there. Hm. Interesting idea, a fudge,
> but a good fudge probably for practical use. I found that editing
> Pacejka numbers quickly gave me some inconsistent skidding sounds and
> all that.
> A Pacejka viewer/editor is still needed, but the factor might be in
> the next version. Thanks for the idea. :)
> And what do you think, multiply directly after calculating Fx/Fy/Mz or
> after your slip combination method?
> Ruud van Gaal, GPL Rank +53.25
> Pencil art : http://www.marketgraph.nl/gallery/
> Free car sim : http://www.marketgraph.nl/gallery/racer/
what the hell is this toe-thing, anyway?
--
Gunnar
#31 SUCKS#015 Tupperware MC#002 DoD#0x1B DoDRT#003 DoD:CT#4,8 Kibo: 2
myk, trygg og god.
There's a problem with collision detection that keeps bouncing the
velocity off the surface if it stays IN the surface. Just worked on
that and now it's possible to scrape the surface more, or well, it
doesn't collide when you are LEAVING the geometry, so collisions are a
lot better.
But it may be the dampers that you saw...
Hm yes good tip! :) I'll see if I have bogus situations which lead to
this. Must be near Chris' method of testing energy conversation in the
shocks. Sometimes the wheels seem to get stuck in the upper compressed
situation (which is probably what you mean) and only at a certain
point they plop back again. Might have something to do with it. But in
the mean time, they do press against the body in such a way that it
seems to be able to fly or something.
Thanks,
Ruud van Gaal, GPL Rank +53.25
Pencil art : http://www.marketgraph.nl/gallery/
Free car sim : http://www.marketgraph.nl/gallery/racer/
Yes, no offense taken, I've got some tough skin. :) Most criticism is
right, although a lot ofcourse I already know and is on big tasklists
here.
Yes, turn on debug.ini's stats.show_tire_forces. It is off by default
now because more people start just racing it instead of being a freak
of simming. ;-)
Carlab is really only used in my case for the car cameras. As for
braking, this may be just the braking torque, which you can set in
car.ini (data/cars/fer312/car.ini for example). Some cars can brake
too hard so you're too quickly locking the fronts.
Then there's a 2nd problem; I interpolate with negative slipratio's
(from 0..1) the resulting force from purely lateral/longitudinal to
the slipvelocity direction. Because when locking your tires ofcourse,
the force direction = -slipVelocityDirection. However, I think this
grows as you brake and brake harder, but perhaps I'm overdoing it and
this is just an effect which appear at the limit of braking
performance (say, slipRatio=0.9..1). I don't know.
So basically it may start giving up steering control a bit too soon
when braking.
Yes, replays are very important, just to review the great moments, and
for a ghost car too. I'll record car positions, since the system is so
complex that I'd rather just store the pos/rot variables. Esp. if you
change the car physics a bit (like inertia values), your replays would
become unusable if you just stored driver inputs.
Cheers,
Ruud van Gaal, GPL Rank +53.25
Pencil art : http://www.marketgraph.nl/gallery/
Free car sim : http://www.marketgraph.nl/gallery/racer/