Quote from jtw62074 :

Quote :

is it possible this effect is similar to the traction circle concept, where if the rubber is using more elastic energy to counteract the increased air pressure, there is less left to counteract the load of the car?

Not sure I understand the question. Perhaps you might be overthinking it?

perhaps, i'm really not sure if what i was thinking makes sense.

if we take the Avon data as correct - and i don't see why we shouldn't - then there is a logical explanation for the tire deflection being greater even with more air pressure. of course this only happened in a specific case, but if true it highlights a complexity in how grip relates to tire pressure and load that would be missed in a linear model.

my thinking is that there is some maximum force that is shared among more than one work. in a similar way to the traction circle concept where the maximum grip has to be shared between longitudinal and lateral forces, the elastic force of the rubber in the tire could be shared between resistance to air pressure and resistance to deformation due to load.

we know that a tire carcass in and of itself has some ability to support a load. even without air in the tire, the rubber resists deformation. when it is deformed, it stretches. we can say that the more resistant the rubber is to stretching, the more load it can hold. this makes sense because a harder compound could hold more weight for a given deflection.

when we add air pressure we get a complex relationship. on one hand the air inside the tire presses out against the rubber and acts to hold up the load. as the air resists compression, so it resists the load deflecting the tire. however it also expands the rubber of the tire and as it does this it makes the rubber stretch.

since increasing air pressure uses some of the rubber's elastic resistance to stretching, there is less of that resistance available to support the load.

--

having said all that, it's only a conjecture so i invite people to tell me i'm wrong or what the actual cause is, or that the data is wrong. i also think momo has a valid point that we have to wonder if creating a "non-linear contact patch vs air pressure and load model" gives enough bang for the buck compared to other things that could be advanced in the lfs physics.

for one thing the complex behaviour only happened in a specific case. i think the narrower a tire is, the more the sidewall works to support the load which might explain why that case showed the behaviour.

we have to ask ourselves "did Avon specifically choose tire dimensions that would make use of this complexity to affect handling during transitions in cornering direction and during turn-in?" if the front tire contact patch is actually larger than the rear under light loads, this could affect e.g. how the car balance changes as the inside tires under cornering see increased load during a transition from turning left to turning right.

many people have a complaint that lfs setups are unrealistic. better modelling of tire contact patches vs load could lead to more realistic spring and damper tuning. i say could. i'm no expert!

i'm pretty sure the current lfs model uses a linear approach. this makes sense because S2 had to get off the ground and every little complexity can take a looong time to code, debug and properly test. we know for instance that in lfs the spring rates are constant with respect to spring deflection. there are a lot of little simplifications like that which were used to save time and computational requirements and i think Scawen did a great job of prioritising which things to do in a complex way. we definitely get more challenge and enjoyment out of having many independent points of heat storage in each tire than we would from having a more complex spring rate vs spring deflection model.

what do you think?

weird i thought the rift was built using standard lcd panels. almost the entire industry uses 60fps hardware at least for desktop monitors but i guess smartphone/tablet tech is better? does anyone know if the rift itself can do say 144 hz?

can you get > 100fps? instant buy if so.

Quote from Racon :

NNs are highly suited to reducing complex interactions between numerous inputs into relatively few outputs.

yes. my argument is that the outputs generated by neural nets are not the floating-point vector results of complex physical interactions.

hello, i'm having too much fun not to share

good set allowing aggressive driving. steps out a bit coming off the apex but easy to catch with linear response. has that nice decreasing radius characteristic under trail braking.

i'm trying to make an all-around set so of course it isn't wr material, just plain fun to drive.

ps no locked diff

did 27 laps of so chicane with no heat issues at maximum attack. have fun

EDIT: made to go with 540 deg setting of G25/G27!

Last edited by CarlLefrancois, Fri, 31 Jan 2014 22:25.

Quote from Mountaindewzilla :

This question was addressed in a previous post. Check this out for more information.

i don't think you understand what neural nets are for.

just because google gives you search results with "neural net" and "nonlinear" in the same article doesn't mean what you think it means.

Quote from DANIEL-CRO :

You may find this thread interesting : LFS tyre model?

i remember reading that thread Daniel. unfortunately, i don't think LFS could implement that. also although there is a difference it is so small the driver doesn't notice it?

one of the strong points of LFS is the client-side prediction model and i don't believe that model could work with some variables affecting the tire radius. currently you can use tweak to change your tire parameters but all racers on the same server have to have the same parameters, right?

you have a point momo. if adding better tyre load response costs more CPU than adding better aero response, then aero would come first since both have a big effect.

i think we can agree that a properly simulated multibody suspension is pretty far away. does LFS even have a dynamic roll center? it seem obvious it must but the previous S1 had no dynamic camber or tyre heat modelling!

if there is a concept like the traction circle at work that makes the impact of tyre pressure on the contact patch more complex than just "patch_size = weight / tire_pressure", then having this added to the tire model would be like the difference between the grip model respecting the traction circle vs allowing max lat and long grip at once.

people, neural nets are good at recognising patterns. who ever suggested they were good at computing results to non-linear functions?

a good neural net learns whether to return 0 or 1 when asked "is there a letter A in this bitmap data?". how is that related to "what is the resultant force vector when the following objects come together under the following conditions?"

i want to agree that there are other priorities...

however, in a quote from the article "If we look in the middle range at a load of 553 lb, the calculated contact patch area is 67.42mm. At the same time, the front tire at almost the same load has a contact patch area of 67.35mm. This is almost no change at all, and in fact within a reasonable error limit it is probably sufficiently accurate to say that the wider tire has the same contact patch area as the narrower one does at around 500 lb load.

At very low and very high loads the wider tire does have a larger contact patch size, however"

this kind of behaviour should have a big impact on fine tuning the handling around the limit.

but i still want to agree that LFS needs to put a priority on what will affect our online races.

la faute est nullement avec le simulateur qu'il voit une "mort à petit feu" comme tu dis. la faute est avec les consommateurs qui ont choisit des produits inférieurs. si nous ne sommes pas capable de voir ce qu'il y a de meilleur et de rester avec ça par quoi par manque de nouveauté? c'est notre perte si ça fini là.

i just came across an interesting article about the effects of tire pressure on contact patch and other characteristics like deflection. half way through the article i was becoming convinced the data were just wrong or the engineer had mislabelled the data columns because for example lowering the tire pressure under certain loads actually decreased the tire deflection. (i.e. the front tire radius for 550lb load is 308.3mm @ 28PSI vs 310mm @ 18PSI.)

is it possible this effect is similar to the traction circle concept, where if the rubber is using more elastic energy to counteract the increased air pressure, there is less left to counteract the load of the car?

does anyone know how accurately this complexity is modelled in lfs?


http://www.performancesimulati ... ct-or-fiction-tires-1.htm

Quote from OTone :

"Simple" things like day-night system would revive LFS massively.

you're right day/night would be awesome in LFS. i disagree that it's simple, and not because it couldn't be done simply, but because LFS devs wouldn't accept the quality of realtime lighting that you would get from a simple solution, and maybe they aren't convinced the average graphics card they are targeting can handle it yet.

it's a matter of perspective and like it or not i think the devs are very happy with the result so far, and hopefully will keep going exactly like they have. day/night in LFS would be awesome because it would simply be above and beyond day/night in another sim.

yeah the devs must really regret the way the announcement of scirocco + rockingham failed to deliver when they expected.

+1 to the complainers on that one.

but did you think? the next step for LFS has to be a big one. a decade of development could be ok. i think people who say 5 years is too long don't accept why LFS is so much better than other projects, it's because the devs stick to their vision and their method of development. it's because the result when it does arrive doesn't have big holes in performance. it's because it's sexy

if you want to go wheel to wheel with someone and really feel them bump against you at the limit and just save it with a wheel on the grass, LFS is there for you right now.

that was fun those bumps on the blackwood straight remind me of the nurburgring

i'm trying to get in.. what's the secret club word?

i hope by classic layouts you don't mean actual S1 physics too does anyone still have any good setups for that (no high nose)? either way it'd be fun to grace the rear of your server's pack

or you could really drive the car.

and lose against hackers

funny, this is the sickness LFS gives us. it makes us wish all our other racing game content used the LFS physics

but... LFS physics wins because devs decided to enforce very strict priorities. no fancy shit. Euro truck simulator succeeds because devs spent ages adding candy.

you can't have 50+ ai cars driving around an endless open map with LFS physics. at least not in 2013 on our crappy computers!

Quote from Flame CZE :

I think what you mean is caster, not inclination, isn't it?

you're right. i think inclination isn't even in the LFS setup options anymore.

no sense in having more than one open source racing simulator!!!!!!1

http://en.wikipedia.org/wiki/TORCS

the graphics are done, the physics are .. physics! needs a little net code. all the people who are sure they could make LFS better if it was open source can make something better than LFS using TORCS.

my hunch is there is no better solution to vsync than "wait for refresh". in a single-threaded program that is literally a CPU wait.

this post is for anyone who is undecided about buying one of these pricey screens. if there was no difference in feel i would be very regretful of my purchase.

Quote from shimon-ifraimov :

there is a way to t
Reduce wheel weight ?

LFS gives the force from the front wheels. you can try to lower the inclination in the car setup. most setups have this on the maximum and it makes a lot of resistance to steering.

i'm not sure, but i think inclination adds camber as you steer more, so less inclination makes more understeer :S

ppl - drifting and racing is done for the same reason: it's fun to control a vehicle at its physical limits

Quote from Woz :

My work around is when you callibrate start half way and only do the lift and press so only about 1/3 of range is calibrated. This then becomes the area the clutch works over. Sort of simulates the range bit transitions. Adjust range to suit

i may be wrong but i think he wants the size of the range where the clutch is only partly engaged to be bigger. this calibration trick is good to position the bite point where you want in the range, but makes the engage zone smaller, right?