Quote from ORION :

You said it m8
Baisically, you have absolutely NO CLUE

Physics in a game are simply not the same as in real life. Just think about how freakin awesome and complicated LFS is. This shows clearly that Scawen can program as hell. Do you really think he can't understand m1*v1 = m2*v2 ?
Are you serious?

Wasn't trying to suggest that he wasn't a very good programmer, I just don't believe the current programming has reached the limits of the internets capability...

Surely if suggestions on how to improve the game were to result only in 'go and make a new game yourself' we wouldn't have a suggestions forum?
(not that we're in it of course... yes, go ahead, blame me for dragging this off topic )

Last edited by Blowtus, Fri 3 Feb 2006, 04:58.

Quote from thisnameistaken :

Imagine you're doing 100mph (or approx. 47m/s), and you've got around 50ms latency. You could travel two metres between each snapshot the server sees. The car next to you is doing the same, there's still potential for you two to get entangled by about a metre before the server will be aware of it, and that's under pretty good network conditions.

How do you resolve that situation? You'll have to backtrack, work out the point of collision, then determine the result of that collision and skip back to the present to teleport the cars to where they ought to be. This is going to be a jarring experience for the players.

your example is an easy to understand one, but fairly pointless, as the result of a 100mph collision is always going to be rather jarring for both parties. If you have both cars doing 100mph *next to each other*, and they slowly drift into each other, (ie, perhaps a 5mph collision) there is no potential for penetration of that depth (assuming 50ms latency)

And what's more jarring... being launched skyward 100 metres, flipping end for end, taking out the field... or having the car skip a little unexpectedly?

Quote from Hankstar :

Poor old computers need all this basic information input in great, mind-numbing detail, and a car collision contains many thousands of variables even when modelled simply.

thousands of variables for a non elastic car collision? come off it
2 vectors and relevant masses would do a decent job of providing a framework for the collision.

Last edited by Blowtus, Fri 3 Feb 2006, 04:28.

Quote from Gunn :

The distance between two cars is calculated after receiving packets of data from a client. A slight delay in receipt of a packet of data may cause slight inaccuracies. Prediction can only be accurate to a certain degree. A longer delay, or some other network anomoly (even congestion) may contribute slight inaccuracies of car distances especially when network latency (ping) is high. This is not a limitation of the software this is a limitation of the world-wide network of servers and cables that we call The Internet.

Do other games experience this level of collison randomness? I understand roughly how collision detection works, it just seems to me that it needs an extra layer of calculation. At the point of collision, calculate the vectors of each car over the last 5 packets or something. It's only an internet limitation if you view the current collision system as unchangeable...

Quote from thisnameistaken :

That's going to be really hard to eradicate though.

Don't see why. I'm no programmer, but working out a vaguely accurate, non elastic collision is not difficult.

the lfs multiplayer still has the fairly large problem of random, violent collisions, upon very soft contact between players... still needs a bit more polishing imho

I disagree... you don't feel exactly the same things, but I don't find it any more difficult to understand what the car is doing. The effects may never be the same, but they never are between different cars anyway.

the thing with jet ski / snowboard / other games, is that you can't accurately simulate them because a huge part of the discipline is body movement. The drivers body movement really has no effect in a car, it's all about the movement of the controls, so it can be much more easily simulated...

Quote from george_tsiros :

Debating whether LFS is realistic or not is useless. Stop wasting precious time and go do some laps in a go-kart (a real go-kart. not the MRT).

Why is it useless? One of the reasons I enjoy motorsport (participating, not spectating) is an appreciation / understanding of the physics involved, and how my actions fit in with those constraints. Understanding where LFS gets it wrong is of interest to me, much the same way understanding how the poor head design of my bike affects it's running is, etc. Knowing where LFS is wrong would seem to be part of the task of fixing it. No doubt Scawen understands the consensus opinion by now, but if none of these threads ever appeared, one wonders if he'd have as much reason to be looking at the physics, or would improve them as easily.

How would real tracks help you find that out? Cars don't respond differently on different tracks...

don't see what 'real' tracks has to do with anything myself. Quality tracks however, that's important. Each to their own I guess. Knowing I'm simming on a 'real' track design doesn't aid my experience in any way, it's all racing.

Quote from micha1980de :

despite the fact lfs is a sim, it souldn't mean it's got to be complicated.
so +1 here Woz.
On the other hand, show me any professsional real driver that screws on his car, instead of talking to the mechs.
I like the concept cause, I don't know very much about car-specs and "what " to change to get "this" effect.

there's a lot more folk that race who set their car up than those that don't... lfs doesn't simulate the 'professional' side of things at all as far as I'm aware

I think it'd be more suitable to have a section in the tutorial dedicated to very basic setup, ie the rollbars, the final drive, and maybe tyre pressure. It's really not very difficult to learn how to adjust these things, and they can change the nature of the car enough to suit particular driving styles and race tracks.

Sorry Todd, somehow I misinterpreted your post as referring the skid pad numbers and suggesting lfs should have similar values. You're right, I was being irrelevant...
You did do a lot of typing to say that tyre modeling can cause large changes though

skid pad numbers are not an accurate representation of 'max' g values, for real world driving / racing. They're just something for U.S car mags to write about

I'm pretty sure 'real life' g's are always higher than skidpad tests...

a lot of the setups available for download are hotlap setups, designed for on the edge driving at a very fast pace. Most people create more forgiving race sets, I know I sure do. Being able to create a set that you're able to drive fast with, not destroy tyres, and drive consistently, is a very handy skill... keep at it

it works best for me when I set it up for the most grip, then reduce front camber until flipping is less of a problem.

you really don't sound like you have a temperament suited for hotlapping... it's a fairly frustrating way to play lfs. Why don't you do some leagues / racing instead of worrying so much about times???

lfs is a racing simulator... you can join whatever league you like, you can hotlap, or you can race casually... do you really need the game to tell you what to do???

I don't see how anyone who likes driving cars could fail to see the potential of the game after playing the demo. If they don't like driving the demo I'd suggest they won't like the full thing either...

Eastsiderz, LFS probably doesn't shock you with goodness (sound being the major problem here) straight up like GTR, you can't really add much in the way of mods etc, but it provides a racing experience that others are unable to match, imho.

what point would that really serve though?

how much accel dif locking do you have? drop it down if you're having trouble with throttle spins at low speed.

sense of speed is one of the problems that causes this 'low grip' feel. Was doing a corner on fe gold rev the other day that felt terribly slow, but when I thought about a similar corner I'd done on a track 'in real life', I was going about 20kms an hour faster in the rb4, than my rx7. Real life doesn't feel slow, but the game does at times Obviously that's hardly a scientific measure, but I'd be pretty confident the problem is not with outright grip levels, but with the way the grip 'feels' when you push the limit. There is also the issue of putting power down while cornering, which definately has a way to go...

spend some time learning to set the car up to your liking. if the rear is sliding out, a few things can help. lower accel value on lsd (high values aren't very important in the fox because it's so low powered and grippy) less rear roll bar, lower rear tyre pressure, more rear camber, more rear downforce.
A setup you can drive well will get you far better times and consistency than a setup someone else is fast with. You should be able to get into the 1:10's at least without a setup that's hard to drive.