Wasn't that the one with the completely fabricated curves that fall down after peak, then sort of go back up again for a few degrees worth of slip angle, only to continue falling down shortly thereafter?

Frank, are you serious or not?

The PCC07 mod *does* have the engine in the back. I seem to recall it also has the centre of gravity some 10cm too low and of course questionable tire curves.

Above that though, bumps in ISI sims are silly sine wave based which, at best, very poorly simulates about 10% of the real life bumps and for the other 90%, its rubbish.

And probably above that; the tires probably don't get close to the maximum achievable in rFactor, and sadly this maximum achievable isn't all *that* good to start with..

Niels: Well, the ISI engine is kinda a mess, so the input must be totally "crazy and unrealistic" in order to achieve reasonable real output.

And i`m not talking about sine waved scripted bumps but the "polygonated" bumps, although the sine waved gives those polygonated bumps a better feeling.

In most simracing games the tracks are pretty flat and if there are bumps on them they dont affect the car in a way it should.

Well, i am not taking the scientific approach to this issue, i just base my opinion on what i "feel", so yes...i`m a troll, clown and all that

I have yet to try an rFactor mod that is even remotely like real life (except of course Niels 'Vette ).

I'll stick with GPL - at least you know you have a decent physics engine underneath.

And no matter how flawed the physics engine, the same thing rings true - crap data in = crap data out. You can't fudge the input data to make it work properly. It might end up 'okay' in ONE scenario, but in a 10 lap race a simulator has to cope with thousands of different scenarios.

Maybe one day... but I'm not holding my breath.

Frank, the poorness of the ISI physics engine is mostly unfounded, its problems are, like any sim, tires. Suspension modelling is more advanced than LFS for example (dynamic roll centres, for one), it also has pretty detailed modelling of diffs, high and low speed dampers and a lot more aero modelling than LFS.

Of course it is not user friendly to change them but as far as the engine goes, its more advanced than LFS.. LFS however takes a more 'dynamic' approach in the tires. In my opinion still fairly wrong in places, but thats where it makes up for some of the other things.

Mind you thats a good thing, aero and dampers are not half as important as working rubber..

Since the base engine in ISI works well until you get to the tires, I'd say fudging and tweaking too much with the good bits is dangerous. Since the only thing that really isn't quite good (neither is LFS or RBR or any sim) are the tyres, that might be an area to fudge a bit..

My 2c anyway, its getting off topic.. Can't wait to see what the hopefully soonish patch for LFS brings.

Quote from Niels Heusinkveld :

Suspension modelling is more advanced than LFS for example (dynamic roll centres, for one)

where did you get that one from ?
lfs appears to have a full blown 2d suspension model which if done properly should include changing roll centre positions without having to model them explicitly

I heard and since then assumed that because of... umm something, LFS might not use roll centres. But yeah we know that assumption is the mother of all f...ups.. Might be wrong.

Quote from Niels Heusinkveld :

Since the only thing that really isn't quite good (neither is LFS or RBR or any sim) are the tyres,

I see you say this quite a bit, so I would like you to elaborate on your perception of the flaws in LFS tires. We know long grip is one, and "possibly" some force combination issues (but precisely what - noboday has ever said)... Why are they not good?

@Shot rF has 3d susp as far as I know... so bump steer etc...

Well I don't really know for sure. When I drive LFS there are some situations that I simply do not believe are realistic. When front wheel drive power oversteer was mentioned I kinda decided LFS currently isn't good enough. I don't understand the actual workings of tyres enough to really get into it. Todd has posted some ideas on what might be not quite good enough in LFS..

I don't think, back to on-topic-ness-ish that its the longitudinal curve per se. It is far more likely somewhere in the force combining, which I wish I understood better!

Quote from Niels Heusinkveld :

[FONT=Agrofont][FONT=Agrofont]I don't think, back to on-topic-ness-ish that its the longitudinal curve per se. It is far more likely somewhere in the force combining, which I wish I understood better![/FONT]

at least the force combined curves look very texbookish so they cant be completely off

speaking of which
is there any intuitive way to understand why the peak slip angle moves with longitudinal slip ?

Quote from Niels Heusinkveld :

Well I don't really know for sure. When I drive LFS there are some situations that I simply do not believe are realistic. When front wheel drive power oversteer was mentioned I kinda decided LFS currently isn't good enough. I don't understand the actual workings of tyres enough to really get into it. Todd has posted some ideas on what might be not quite good enough in LFS..

I don't think, back to on-topic-ness-ish that its the longitudinal curve per se. It is far more likely somewhere in the force combining, which I wish I understood better!

I'm not claiming LFS is perfect - it's far from it - but how do you know on-power FWD oversteer wouldn't work in real life? I mean seriously, does any real race team use setups like we do in LFS? If anything, having mechanical perfection (no bending or loose parts), indestructibility and far too fine-tuneable setups can be blamed for such odd behaviour, maybe longitudinal grip, maybe load sensitivity, but right now everything seems to be blamed on the mysterious force combining.

LFS certainly shows odd traits in certain circumstances, but there are far too many variables in the equation to confidently state "force combining is wrong". Every time I hear it said, for me it kinda comes with the notion of "I have no idea what it does and have no data or evidence to support that it's actually wrong, but Todd mentioned the term 'force combining' somewhere in one of his posts so I blame it for LFS' odd behaviours." No offence, but I think our time would be better spent actually researching and analysing* the facts and data we can gather, rather than parroting Todd's (admittedly extremely insightful) words over and over.

* Of course that's easier said than done, I understand that.

Edit:
For example I've now attached two graphs showing longitudinal (Y) versus lateral (X) grip. One is colour coded with slip angle (probably less interesting) and one with slip ratio. In these graphs you can basically "see" the force combining; first from the lateral curve, which is basically the left and right side of the "circle". The lateral curve has only one "line", representing that the lateral grip doesn't drop off, but rather stays the same when we go over it. Then we move to the longitudinal curve (top/bottom) part of the circle, which in this graph is mostly on the top as I didn't do much braking in the drive tests. As we follow the path from lateral to longitudinal curve, we can see the "main grip line" split up into two parts, showing that the longitudinal curve has a drop off. The very top and elongated part/line is the longitudinal peak, while the lower part is the grip we get at full dropoff.

For reference I've also attached a longitudinal grip curve and a lateral one. The test drive contained much parts with different slip ratios and angles and was more focused on getting lateral grip data rather than pure acceleration tests, so the longitudinal curve wasn't as clear, which is why I added the blue line there.

[Off-topic]

Quote from AndroidXP :

http://www.lfsforum.net/attach ... id=42593&d=1194975909

That is definitely a frontal projection of the human brain.
The green area is the frontal lobe, the black one is the temporal lobe that is next to the purple, parietal lobe. Even the occipital Lobe is shown (red).

[/Off-topic]

I am sorry for going off-topic, but those similarities are truly amazing.
Furthermore I can't contribute nothing relevant.

So LFS tyres are brains? Many a fanboy would agree with that.

/OT

It makes sense. Scawen has tire physics on the brain.

edit: damn mods always reading the forum like it's their job or something

So much effort and all I get is three off topic "lol brains" posts? :rolleyes:

Quote from AndroidXP :

So much effort and all I get is three off topic "lol brains" posts? :rolleyes:

hehe they are a tad hard to read
maybe 2 3d surface graphs of lateral and longitudinal forces over both angle and ratio would be easier to understand

Quote from Shotglass :

hehe they are a tad hard to read
maybe 2 3d surface graphs of lateral and longitudinal forces over both angle and ratio would be easier to understand

You do have to study those charts a bit... they do look like brain scans

One thing that comes to mind for me looking at the circle charts is tire deformation. Obviously, when in a corner, the slip angle is high. The sidewall is going to be flexed. But when you add power to the wheel, your slip ratio goes up, but wouldn't your slip angle go down slightly? Or at least change some?

The reason I ask is that looking at the charts, specifically the slip ratio chart and the 8° slip angle (red) it seems that it is possible that adding power would force the tire to a different shape and reduce slip angle. The tire can only flex and twist so much. The longitudinal stress of applying power has got to have some effect there. I kind of look at it from an aerodynamics standpoint where you have relative wind. This is the resultant vector airflow over an airfoil after combining wind, direction the aircraft is moving, prop/rotor wash, etc.

This is also where some robotic (AI driver) testing would be nice. Having something that could drive the same course in the exact same way multiple times would help with data analysis I think.

It also seems to me that if you used look-up tables for tire data, the results we are seeing in those charts would be more predictable and homogeneous than what it seems to be... but I might be way off the mark with that thought.

I appreciate the effort, those are really interesting to see!

I'd like to see this done on one of Biggie's hotlaps!

Android, what do you need / do to make those plots? I'd really want to try and make this for ISI physics as well. There is plenty of telemetry data available, and Motec allows for some configuring and maths..

To generate these plots I use two tools (that I didn't create), the first one is RAF Tyre Extract, which basically takes a .raf output (which you can make by watching a replay and hitting "Output lap data"), reads the data and creates a file containing extracted/converted/calculated data. Afaik, LFS outputs this data every 100th of a second, so each dot represents a tyre state snapshot for every 0.01 seconds of the whole lap. The data contains simple things like X/Y position and other more complex stuff like forces acting on the tyre, camber, etc.

This extract file is basically just a text file, the first line is the header which defines the data columns, all other lines contain the data. The second tool, called MultiDim, then reads this file and dynamically creates the plots. In the tool itself you can choose which data columns to use for X respectively Y axis, as well as apply certain constraints/filters on other columns (example: show lateral grip on Y axis, slip angle on X axis, but show only data points where the tyre load is between 3000 and 4000N).

How to get a similar output from rFactor? Sorry, absolutely no idea.

I think I can get longitudinal and lateral forces, loads.. I suppose you also need to know slipangle and slip ratio?

Slip angle and slip ratio are nice for visualization, but not necessarily needed

So you basically just look at 100hz what the lateral force and longitudinal force is and plot those points? (/me has a feeling this starts to get more suitable for a PM )

Quote from Niels Heusinkveld :

I think I can get longitudinal and lateral forces, loads.. I suppose you also need to know slipangle and slip ratio?

The longitudinal force is not available in rFactor internals plugin API (http://forum.racesimcentral.co ... php?p=3150984#post3150984). Unless there is another source for it, you would have to try deriving it from lateral force, steering arm force and suspension/steering geometry, but pneumatic trail seems to be also necessary.

Quote from AndroidXP :

...

which one of those 2 limits you in the data gathering ? (you mentioned something about limited raf sizes somewhere)