What sort of cornering G force does it pull? If its 1.2+ it will have a fall over tendency indeed.. The C6 with the semi slicks (1.07G) can already be flipped if you hit a kerb in some (very) unlucky way..

The skidpad is a nice test track (see rfactor central) though you'd have to disable the bumps in the main terrain.tdf to make it flat and the readouts stable.

I suppose the M3 pulls some 0.95G perhaps? dunno really..

Did you up the centre of gravity? 50cm from the ground is a good bet for a performance car, though the M3 might be a bit higher even; thats guessing..

It is the ride height value + the entered cog height (i.e. something like 0.12 + 0.38).

Should make a difference as well, lots more weight transfer

They all (rfactor / gtr2 / gt legends / race) use the same base engine, at least as far as 99% of the physics go. It all depends on the numbers they enter.

There is good and bad news, they all do a pretty horrible job of 'tire force combining'.. but at least the devs know this and plan to improve that.

Its the biggest limitation in the engine. At RSC they're now talking about brake duct lift.. When tire physics improve, so will the cars..

I'm not doing any GTR2 stuff, in fact, now that I know how poor the force combining is, I'll probably leave it after the z06 is done. It is certainly possible to get some fairly good handling, MUCH better than I ever thought possible with ISI, and in some cases more believable than LFS (especially below / at the limit).. but there are limits to what you can do.

Simracing physics are for a great extend stuck in the 90s. Go iRacing!

The C6 suspension is made with CarFactory. This means it will have the roll centres exactly where you want them, with exactly the anti dive / scrub radius / kingpin inclination / caster / roll camber rate you wish.

The result is 'too good' as it uses too long wishbones. A future version of CarFactory will incorporate 'limited space', making the suspensions not only look a bit more realistic, but behave a bit less 'good' as well.

Suspensions are too complex to simply look at and claim things; a good looking one could have a really unstable roll centre or some dodgy steering or..etc etc..

The M3 has a few problems that are far bigger. A 10cm too low centre of gravity and downforce are amongst them..

Quote :

Given that, there will always be a need to fudge the numbers to get a particular desired behaviour. The trick lies in identifying what the desired behaviour should be and then figuring out how to fudge the numbers to produce that result.

Thats a big risk you then take.. I agree neither ISI nor LFS has 'near perfect' physics.. but as soon as you leave the path of 'realish values' it becomes subjective.. 10 people making the same car will result in 10 probably quite different handling ones. The main things you can get close to and keep em that way. (weight, inertia, centre of gravity, aero lift, basic suspension properties).. Tyres can't really be realistic and probably always are 'fudged', but believable slip angles (well a degree or two higher perhaps) and grip levels should imo still be used. The main thing where I 'fudge' for handling is in the drop off of grip after the peak; imperfections in the force combining seem to make any tire too hard to drive or too 'uncommunicating' in ISI, especially over the limit.

Thing is, you can't really cure the faults, only influence some symptoms by .. i dunno.. increasing the yaw inertia.. but then you're just entering a random zone where you try to make it work.

I've found more realistic tire curves for a performance tire and it doesn't work too well in the (work in progress) Z06. I think its the limits of the tire physics, especially the combining of forces, that simply is too far from reality. I can get away with stretching the peaks one or 2 degrees outward. With some 600kg of vertical forces and 700 lateral, I can believe the bits flex a bit and the chassis might do as well.. I can also ease the grip drop off knowing that the base curves are not too representative when the actual final force vector comes from pretty poor force combining.. But at the end of the day you simply can't make a car handle really well if the physics engine isn't good.. No matter how you fudge.

Last edited by Niels Heusinkveld, Tue, 18 Sep 2007 06:47.

MoreCurvesThanYouCanShakeAStickAtMotor

Why would you use rollbars as finetuning only? I'd say a 5mm drop of the rear is probably more fine tuning than a rollbar adjustment..

While I agree that lowering the rear might help a bit, I doubt it will really solve the issue. Plus a downside of lowering, depending on how much the suspension generally moves, you may hit the bumpstops more often? I'd probably prefer a slightly softer higher car.

What about tires, and temperatures.. Perhaps the fronts get too warm or the rears stay too cold?

I'm really only talking out of my sim experienced arse, I have little practical knowledge of these type of cars..

Some things wrong with the BMW are:

- the centre of gravity is some 10cm too low (thats a LOT) bye bye weight transfer

- it produces about 65kg front and 128kg rear downforce at 200km/h.. the real car is neutral at best..

the tyres are too flexible..

ho hum below average.

It was a bad day, and as usual some ''slightly narrow minded'' people made some documentaries in their favour claiming all sorts of outrageous things. The real problem is that people listen to that crap..

There are probably a few valid questions to raise after the MS Flightsim Trained guys flew into the towers and there are always things the govt. should/could have done.. It was a sucky day for sure..

But the uber activists (in general not the 'lets hear both sides and use brain to verdict' type crowd) simply went conspiracy theory about 0.1 seconds after the first plane hit the building. Even if all the "terrorists" in the world would say "We did it!" complete with proof, these activist fools would still say its an inside job..

Scary people .. On average much more scary than some dude in a cave Afghanistan.

But more on topic.. Rick Ashtley (sp) rulez!! :P

Or just tell the 'real driver' to toss off in favour of real numbers.. Damn real drivers are partly responsible for the wrong tire grip curves that are floating around since the model T came out..

dxtweak, mess with the center value for the gas pedal. Setting center just abaove 'minimum' tends to work. For sims in general, controllers best avoid the Y axis because games tend to think that is a joystick axis that should have a center position.. So only half the axis works..

I love the ghost c... umm. tank

Its the usual ISI / Blimey that I wouldn't get too near too. Same old dreadfull skid sounds, slightly better tyre curves, 10cm too low centre of gravity, probably too much grip, silly bumps on the track...

As I hated rFactor / GTR / GTR2, this is firmly on the hate list as well. The Corvette I made for rFactor isn't perfect but quite surely a lot better than this effort..

And if BMW paid a six figure number, I think the few head screwed on ISI ''modders'' have full right to their fair share of that

That would make 4 that bought rFactor because of me? I'm really thankfull that the 'hardest crowd to please' (LFS players about rFactor) actually find it drivable and fun!

PS: here is the terrain.tdf file I use now for the Ring 2.0 (doesn't work on the 'fall' version)..

Backup the original one.. it goes in gamedata\locations\nordschleife_2007.. Even grip, much less bumps.. much easier and less nervous.


PS2: I'm actually enjoying LFS too a bit, Axus convinced me to go online in the Lx4 at Aston umm North? Was good fun..

Last edited by Niels Heusinkveld, Sat, 8 Sep 2007 12:08.

don't download ISI stuff.. rely on a handfull of rfactor people who are slowly making the 'difficult' physics engine act realish!

Last edited by Niels Heusinkveld, Fri, 7 Sep 2007 21:48.

I ran the new Nordschleife but using Dx7 mode (ugly..) to get the framerate up.. up up! (200ish, otherwise it just seems to stutter a bit).. However I changed the track surface grip to be even everywhere (100% aka 1.0) and I halved the sine bumps amplitude, which I may change even more.

The devs went a bit overboard with grip zones. The fall version has 10% less general 'road grip' which is a little over the top..

Anyway with some changes its decent! Layout is not good at some places, but that is acceptable. Less acceptable are rough camber transitions (sudden start and stop, sometimes also a rough transition from the 'inside' of the corner to a 'much flatter' outside of the corner, making some bits a bit harder than they should be if you're not perfectly on the inside at the right time, plus at the exit the camber may drop suddenly, making it a bit eventful. (the fast lefthander before Aremberg, the fast lefthander at Bergwerk before the
righthander going on to the loooong uphill bit.) The camber transitions sometimes make the flow a bit tricky

What do you call the next problem.. Road undulations? Going up and down a bit much. I have a feeling this happens too much after the 'jump' (another error) before Flugplatz.. the double righthander isn't that up/downy going in my opinion. A few more corners seem to have it a bit too much.

Hatzenbach has improved, I like the flow there from the exit of the first righthander.

Ah well, it represents the ring just enough and its just drivable enough to be very enjoyable to me.

http://youtube.com/watch?v=k95ITTKMK9I

Aris been working on a mod for GTR2, shows good promise as well

The whole 'real driver input' thing is so overrated! You're getting responses such as:
- it doesn't have enough grip
- brakes are bad
- it understeers too much

..

ok.. thanks.. umm..

There is probably less than a "newyears eve fireworks damaged hand with 3 fingers" hand full of real racers who would actually contribute significantly in detail instead of those general comments..

PS: Shotglass, try harder, cool that gfx card and run without vsync!

Last edited by Niels Heusinkveld, Thu, 6 Sep 2007 18:48.

The HOF reflects the average players enjoyment. It has nothing to do with accuracy and realism, which the average player doesn't seem to care much about..

Bit of nords action, not great.. normal tyres, z51 gears.
http://www.youtube.com/watch?v=OY5ssFnBTCI

As kinda expected (ISI is not likely to have such an obvious bug ) initially you don't notice understeer in the lateral G force but when you really apply too much lock, lateral g drops from 0.98ish to 0.83ish.

I'll do some skidpadding and see how much grip is lost when steering 34 degrees.. But first, labour.. bah

What I recall from LFS discussions a year or more ago.. Someone said that he got less G on the skidpad when he turned full lock,indicating 'a lot of grip drop off after the peak (laterally) .. And IIRC Todd pointed out that something else causes that..

Perhaps rFactor has a friction square instead of a friciton ellipse?

edit:Vsync is pretty bad regarding input latency btw.. And how can you fry your card at 300fps? A dx9 game stressing the card to 4fps will probably cause the same heat buildup.. :S

Glad you tried it Shotglass,

My main gripe with LFS is how easy you get power oversteer. While no expert, I believe that is kinda wrong. I didn't really do anything in parcitular to the Corvette car behaviour, they're just tyres that pull 0.985G laterally and about 1.1 longitudinallyish. There is a fair amount of longitudinal drop off as well. But indeed, its quite planted. I must say in 2nd gear if I combine a little lateral weight transfer with the push of the right foot, it certainly comes around..

I find the car quite sluggish and heavy in feel. I also find it a little lacking in 'attitude changes' when you let go of the throttle or when you steer 'more and more' on a skidpad.. The G force stays nearly identical when you do that and a bit of power overcomes the extra resistance. That feels odd, I expect to loose more lateral G when I have 30degs of lock on where 6 gets me optimal lateral g, or at least expect the friction to slow the car more.

So yes in general the reaction of the car to inputs is a bit muted though that is territory where I wouldn't really claim to know how it should exactly 'feel'. The data is as realistic as I can get it and for now this is how it handles. Once I start aiming for certain driving characteristics I'll leave the 'real data' path and then it just becomes a random car. So its either pretty real, or it isn't because of the data that feeds it, or it isn't because of the non perfect physics engine..

What I do like is the patience you must have. On a track like Barcelona, if you want to do consistent 2:06 laps, you have to be quite smooth on and off the controls and when you reach the limit sometimes (as you do with those laptimes) I find the car somewhere near neutral, slight understeer, slight oversteer and often 4 wheels sliding at the same time. Then I'm really not making big angles slides but nursing the car around this 'zone' of understeer/neutral/oversteer. That is what I like best. I do that with 18degs of lock and my wheel calibrated about 550 degrees.

In LFS I find it hard to drive in that 'zone' as it feels so small. I.e. you enter 'too much understeer' or ' too much oversteer' very quickly and this neutral zone requires VEEEEEEERY precise inputs. Then slightly too much inputs and you're catching a big slide. I find LFS hotlaps to look so odd at times. Driven near perfection but you know that a few degrees of extra wheel lock would have caused a big slide, or 0.2seconds sooner on the gas would have. In the C6 doing this wrong tends to get the car going wrong on 4 wheels a bit instead of having to be so ready to fight a big one.

This is dangerous territory to talk about, but I would be surprised if cars near the limit are so sensitive and will enter 'big trouble' as easily as it happens in LFS. The C6 announces big trouble and it requires fairly big stupidity and on purpose behaviour to really get into big trouble, even when doing quickish laps. Considering the relatively moderate real life track day casualty rate, I simply assume it might be somewhat more realistic than LFS in this regard.

The main gripe with the corvette is that tyres simply don't seem to have enough preference to stay in the 'rolling' direction, going sideways with zero lock versus appying perfect opposite lock doesn't change the yaw of the car enough. Opposite lock works, for sure, but just as the skidpad ''turn more and more'' without loosing grip or speed, opposite lock seems to be like that as well. When you apply perfect opposite lock (front tyres rolling exactly where the car is going) I simply expect that to be about 20x less lateral resistance compared to the tyres pointing straight, say at 25 degrees of where the car is heading.

Ah well.. a perfect world it ain't

If you ever go back to driving it i'd suggest trying the normal and sticky tyres at Barcelona and also at Road Atlanta.
http://rapidshare.com/files/51466266/RoadAtlanta_Beta3.rar (beta, texture detail needs to be at 'max' or sky looks odd)

Of course the usual applies, tripple check you get high framerates, no vsync, render ahead lowered, no dodgy low speed steering overrides, no dodgy head look, speed sensitivity etc etc etc but with you I believe the settings are probably fine

Last edited by Niels Heusinkveld, Sun, 2 Sep 2007 21:43.

lol!

Sadly the chances of some real BMW experts being involved is .. zero.. That means the chances of it being 'good out of the box' physics wise are pretty much zero as well. Unless a miracle happens

Hyper,
lift off oversteer is a good question.. What happens in a real car I don't quite know.. What causes lift off oversteer.. It confuses me. The weight transfer has two effects. A light side needs less lateral force before it starts to slide, but a heavy side has relatively a bit less grip due to load sensitivity. However, really, not much is happening when you let go of the throttle; its not like that causes significant 'braking G force' to my knowledge. The other thing is engine braking. Of course I could increase that so the tyres get a bit more longitudinal forces, leaving less lateral force to work with. But how much engine braking does it have.. I'm not sure if you'd really notice it a lot in a 50/50 weight distribution car that has no uber compression big engine that locks the rears as you let go of the throttle.. (any info on this would be appreciated) The c6 is likely to already have the performance springs; I found a few wheelrates and picked the highest ones.


About the new Ring, I 'found' a beta (only a few days old) which has a few issues, very likely still being in the release. Split times don't match. When I manage an 8:15 in the 'stock' corvette, just as a german magazine managed in their review, the first sector I'm 4.5 seconds too slow, and from the lowest point up to the hohe acht I make up 3.5ish seconds. Sure the car isn't perfect and the track isn't perfect but loosing 4.5 seconds in about 2 minutes is .. a lot.

And why on earth would you model the karrussel as sine wave bumps over a smooth 3d mesh? Rough mesh! that is exactly the karussell.. They very much overdid the sine wave bumps in my opinion, so it might be very much impossible to even get a 8:15 with the normal version. Some places still don't have the right layout, some elevation changes are either too great or too small (jump at flugplatz gives away its GPL track origins..)

However, when the TDF file is changed, and you ignore the few obviously bad bits, there is an 80% smooth and fairly realistic Ring to drive, which is a big step forward