Physik Improvement Suggestions

[FastFabi]

I know there was a lot of discussion about that point, but for me this is the most important next step.

Things to improve (on my opinion):
- flipping over of some cars, e.g. the XFG flipps too fast
- collusion detection and impulses after collusions
- mechanical grip of the tyres e.g. the spinning on the FO8 in 4th gear

[Nitemare]

- mechanical grip of the tyres e.g. the spinning on the FO8 in 4th gearhave you ever driven a car, that weighs 600 kg and has 450 hp? i would think that it spins a lot :)

[snewham]

have you ever driven a car, that weighs 600 kg and has 450 hp? i would think that it spins a lot :)

The most important thing for me, that has to be fixed is the reaction of some solid objects when they are crashed into. Particularly red barriers and banners.

[Nitemare]

i think the problem is in physics rate... when two objects are approaching at high relative speed, in one frame they are close to each other, and in the next one they are intersecting.. that somehow causes a large force...

[R3DMAN]

have you ever driven a car, that weighs 600 kg and has 450 hp? i would think that it spins a lot :)

Ive driven a 500bhp mr2. that spun all the way in 4th and weighs shed loads more than 600kg.

[JeffR]

have you ever driven a car, that weighs 600 kg and has 450 hp? i would think that it spins a lot

Not once it's up to speed and downforce is in effect. Downforce equals weight of the car at around 115mph. The FO8 needs a much taller gear than the default setting, so that 1st redlines somewhere around 80mph to 100mph. F1 cars are 850hp, 600kg, and set 1st gear to redline around 100mph.

With the default final drive, try setting 1st gear to about 1.5, 2nd to about 1.35, 3rd thru 5th as needed for the track. The shifts take too long and/or the powerband is too wide to make using all 6 gears faster.

I have a Hayabusa motorcycle, 175hp, 350kg (with me on it), 1st gear redlines at 81mph, and the bike still lifts the front wheel if it's at full throttle and you approach 45mph where the torque gets strong enough to do this.

[the_angry_angel]

i think the problem is in physics rate... when two objects are approaching at high relative speed, in one frame they are close to each other, and in the next one they are intersecting.. that somehow causes a large force...
At the moment LFS does physics at 100Hz, iirc - but LFS is so detailed I'm not entirely sure how much more the devs could push it up without loosing the bottom end of their market - which would be a disaster.

[dUmAsS]

i think the problem is in physics rate... when two objects are approaching at high relative speed, in one frame they are close to each other, and in the next one they are intersecting.. that somehow causes a large force...

you need something called continuous collision to fix that. but it takes a bit of cpu time

[Bob Smith]

At the moment LFS does physics at 100Hz, iirc - but LFS is so detailed I'm not entirely sure how much more the devs could push it up without loosing the bottom end of their market - which would be a disaster.
Didn't Scawen say it's only the outer physics that runs at 100Hz, and the inner physics runs 20 times faster (at 2000Hz)? Trouble is he didn't specify what the inner and outer physics loops consist of.

@JeffR - if you want to be totally accurate, F1 cars weight about 440kg and the rest is ballast to meet the weight restrictions.

[the_angry_angel]

I'd assumed that the inner loop consisted of only sampling the tyres and what-not, rather than intersection checking as the latter would be too time consuming to do at 2KHz I would've thought.

[Rogz]

Didn't Scawen say it's only the outer physics that runs at 100Hz, and the inner physics runs 20 times faster (at 2000Hz)? Trouble is he didn't specify what the inner and outer physics loops consist of.


Hmm, cant recall to have read that, but then my memory has always been shaky :)

The_angry_angel has a point though!

[atledreier]

The inner loop being the entire car, with all the ground intersection, contact patch, suspension and all that. That runs at 2kHz, or with 20 subsamples. When those 20 cycles have run through, the car is moved in the 'real world', and any intersections with barriers, other cars and stuff is resolved.

At least that's how I read Scawen's post.

[Eowyn]

I don't think XFG flips too fast...

[Orthonormal]

I wouldn't be worrying about collisions or flipping until the basic handling dynamics were perfect for a car that's not colliding or flipping. You shouldn't be colliding or flipping during a real race. On the other hand, you should be doing a whole lot of handling...

[Gunn]

Remember that we are currently in an Alpha stage and we know that more development is coming soon. :)

[BWX232]

Things to improve (on my opinion):

- mechanical grip of the tyres e.g. the spinning on the FO8 in 4th gear


I think the tires may have enough grip, but once they lose grip, and you do what you should do to regain grip in a real car, the tires don't regain grip.. mainly the rear tires in a slow slide. That's my interpretation of "the bug".. :Looking_a

[MyBoss]

The most important thing for me, that has to be fixed is the reaction of some solid objects when they are crashed into. Particularly red barriers and banners.



And when you are rubbing someones side in a race and either you or him is lagging a little so one or both of you go flying, high high hiiigh over the sky.

[JeffR]

JeffR - if you want to be totally accurate, F1 cars weight about 440kg and the rest is ballast to meet the weight restrictions.

Yeah, if I remember correctly the miminum weight is with the driver in the car, about 600kg or about 1300 lbs.

[JeffR]

mechanical grip of the tyres e.g. the spinning on the FO8 in 4th gear

In the case of LFS S2, I think it's the slip angle versus force curve (breakaway grip) and/or the near instant hot spotting that makes the cars less controllable. In a lot of situations, the cars of LFS don't get "loose", they just spin.

If the breakaway grip were nearly the same as maximum grip the cars would be more controllable. Note that in real life, bias ply slicks have this quality that the breakway grip is about the same as maximum grip, making them much easier to control at the limits, so such a tire model would not be unrealistic.

In real life, suspension setups work because they cause one or more tires to slip more than the rest of the tires, setting up the understeer / oversteer characteristics of a car. None of this works if there isn't significant slippage by one or more tires. If the car is overall very close to or at the limits, then one or more tires is past the limits. LFS needs to be able to model this in a stable fashion.

[tpa]

Hm, very interesting JeffR. Think you might be onto something there... :thumbsup: