Poll: rF1 vs. rF2 tyre-model

[Lazza]

I don't know why every issue becomes completely black or white. Now you're making it sound like rF2 tyres require an engineering degree to develop. Close enough can be good enough, especially when comparing to rF1 tyres which have more inherent limitations in their behaviour.

Part of the reason modders find the rF2 model difficult is because it's different to rF1, another part can be that the data they have on hand is performance based instead of construction based, and finally it's a longer process and trial and error therefore takes longer (much longer).

Many good rF1 mods weren't based on data at all, but trial-and-error testing with a few people checking how the end result 'feels'. You don't need an engineering team or classified information, though of course it would help.

Regarding the above list: some of those actually aren't modelled, some of those are in the [Realtime] section which means you can quite quickly narrow down to values that give realistic-looking results. The geometry is obviously a difficulty to get exactly right, but how close do you need to be for it to feel pretty good? Then we're back to a laser-scanned track vs non-laser scanned style debate, where again I would argue close enough can be good enough while for some people just knowing it's not 'exact' makes them unhappy with it. Once you bring the FFB discussions into things it's difficult to know how anyone can be completely happy with anything.

 

[Lgel]

I don't know why every issue becomes completely black or white. Now you're making it sound like rF2 tyres require an engineering degree to develop. Close enough can be good enough, especially when comparing to rF1 tyres which have more inherent limitations in their behaviour.

Part of the reason modders find the rF2 model difficult is because it's different to rF1, another part can be that the data they have on hand is performance based instead of construction based, and finally it's a longer process and trial and error therefore takes longer (much longer).

Many good rF1 mods weren't based on data at all, but trial-and-error testing with a few people checking how the end result 'feels'. You don't need an engineering team or classified information, though of course it would help.

Regarding the above list: some of those actually aren't modelled, some of those are in the [Realtime] section which means you can quite quickly narrow down to values that give realistic-looking results. The geometry is obviously a difficulty to get exactly right, but how close do you need to be for it to feel pretty good? Then we're back to a laser-scanned track vs non-laser scanned style debate, where again I would argue close enough can be good enough while for some people just knowing it's not 'exact' makes them unhappy with it. Once you bring the FFB discussions into things it's difficult to know how anyone can be completely happy with anything.

To have cars that feel pretty close to good, to enjoy a good race, alone or with your friends, you don't need the complexity of RF2 as other games have brillantly shown.

Cheers.

 

[lordpantsington]

...Anything that's done by one guy and release isn't worth the download.

I disagree. One person CAN learn it all, and build it all, it will just take time. Be inspired by this guy:
https://www.youtube.com/watch?v=SeUMDY01uUA
He taught himself to do that in the real world, everyone can do the same if they have enough motivation/drive. Virtual/Sim modding is no different-Just keep working at it, you only can't when you quit trying.

 

[Raintyre]

In my opinion it would be nice the possibility of choosing rF1 or rF2 model to drive. However, i think that rF1 model could be updated to some characteristics of the other. I am talking mainly of the temperatures, because Rf2 layers temperature distribution is much more interesting and realistic.
I hope we could drive a rF1 model with updated thermodynamics despite keeping the old 'mechanical' properties. Mechanical characteristics of Rf2 model, in my honest point of view, involve more difficulties to achieve desired handling and drive feeling, being especially hard to find an appropriate lateral/longitudinal grip balance (while RF1 can tweak long/lat grip with just a number).
What about introducing a lat/long grip multiplier for RF2 model ??? is it possible?

 

[Lgel]

May I remind you that the actual RF2 tire model is still not working after several years in development (try to tune camber by tire temps for instance).

Will it work someday (represent closely the behaviour of the real tire modelled) without requiring manual adjustments to have a good driving car?

There is nothing worse that a very sophisticated model that doesn't work, or needs data you have no way of determining precisely.

It reminds me of econometric modelling at a place I worked, a whole team of guys very well paid (all of them PHD in econometrics, one of them would won a worldwide prize of econometry) would work for a year using very expensive computer resources at the time (processing a lot of data).

The model would predict a growth of 2,1 % for the economy of a given country. The chief economist would consider this data, and replace it by a 2,55 % without any solid evidence (based on his feelings). This country would grow at a rate that was neither 2,1 % nor 2,55 %.

It went along for 10 years, I left, I suppose the same is still going on today.

Cheers.

 

[Korva7]

Lgel, Do you think there is lots of other things wrong with the tyre model than temps?

 

[Lgel]

Lgel, Do you think there is lots of other things wrong with the tyre model than temps?

Temps are consequence of suspension geometry, and how the contact patch evolves under loads, so it is an important point.

I am not a specialist of tires (it is a very complex matter), I just want to point that a modder lacks a lot of data when he wants to design tires for Rf2 even if the RF2 model was 100 % accurate.

The sophistication of the model, plus the lack of real data available to the modder on this matter, you end having a lot of probabilities to end with fantasy tires on your modded car.

As always I may be wrong, some RF2 cars feel very nice when driven, but so do Automovilista cars for instance with a different approach.

Cheers.

 

[TJones]

Tyre temperatures are looking a bit strange to me also.
Example: A few weeks ago i did a test with the BT-44 (beta) at Silverstone international.
I did a Motec recording of my stint (about 15 laps). DAMPlugin (thanks Lazza!) delivers following temperature readings each tyre: Surface temp. I/M/O, rubber temp I/M/O and carcass temperature. While Surface Temps are fluctuating heavy between 70°C and 100°C (FL Inner) within a lap, rubber temp (also FL Inner) only between 97.8°C and 98.3°C.. Considering the two meassuring points are pretty close to each other, i would expect a lot more heat transfer between both layers/points.

I dont know if the following is correct, but to my understanding "ThermalDepthAtSurface=0.0001" defines the point where the surface temps are meassured. While "ThermalDepthBelowSurface=0.0004" the point for rubber temp. So only 0.3mm between those.

EDIT:
The BT-44 seems to be a bad exaple as it is still in developement, but i'm afraid ther's not much difference with other tyres.
Recently did a test with Formula ISI at the same track. I did a few laps and instantly stopped the car to check the progress of tyre temps at stand still, so with very low air cooling.
There was definitely some thermal conductivity between surface and rubber temps, FL inner surface rises from 77° to 83°, and FL inner rubber drop from 99° to 97° while the car stand still for about 50 sec..
BTW surface temperature doesn't seems to be far off, considering this interesting Video.

 

[Lazza]

The tyres in Indianapolis 500 reacted wonderfully to pressure and camber changes. Perhaps we should return to a model that ran on 8MHz processors.

 

[Lgel]

The tyres in Indianapolis 500 reacted wonderfully to pressure and camber changes. Perhaps we should return to a model that ran on 8MHz processors.

The simulation software used by tire manufacturers wouldn't run on our computers.

Do you know for sure if a simplified model (comparing with tire manufacturer's software, still very complex, construction based like Rf2 is) will be able to accurately represent tire behaviour?

So generally speaking, I prefer a simpler model that works within it's design limitations, to a very complex that doesn't completly work (I don't say that RF2 model doesn't work at all, simply that very important features still don't work), needs data not available to the modder, and is finally tuned by feeling.

Cheers.

 

[Andregee]

The tyre degradation works fine with the rf2 Modell, the rest like temperature behavior is really poor and thats really disappointing after several years.

Gesendet von meinem SM-G920F mit Tapatalk

 

[Korva7]

What about this feature on latest build? "Added diffusive adhesion to tire model"
Is that a feature that will be implemented to new and updated cars or is it already working on everything?

Have you people seen difference in that temperature behavior when comparing old cars with the newest cpm cars like howston?
Also i would be interested to hear more specifically about the problem you have found.

About rf1 tires vs. rf2 tires, now after having spent good amount of time playing rf2 and automobilista i have end up to having a feeling that rf2 physics/cars are notably more immersive and interesting to drive. I'm okay with allowing rf1 tires on rf2 but i wouldn't be very interested on driving them personally.

 

[Lazza]

The simulation software used by tire manufacturers wouldn't run on our computers.

Neither does rF2's, as I'm sure you're aware. The building process takes many hours, producing lookup tables that are then used to run in realtime.

Do you know for sure if a simplified model (comparing with tire manufacturer's software, still very complex, construction based like Rf2 is) will be able to accurately represent tire behaviour?

So generally speaking, I prefer a simpler model that works within it's design limitations, to a very complex that doesn't completly work (I don't say that RF2 model doesn't work at all, simply that very important features still don't work), needs data not available to the modder, and is finally tuned by feeling.

A simpler model obviously is more tuned by feeling - whether specific tyres using that model, or the model itself shaped to suit the desired outcome. Even a very complex, detailed model, contains approximations. You can't avoid that.

Temperature is being concentrated on here, but my point is this: is not-yet-apparently-fully-working temperature a sign of a model that in practice is behaving less realistically, as a whole, than a model that produces realistic seeming temperatures (and effects of pressure, camber) but doesn't even try to simulate variations in contact pressure across the patch? I doubt any of us know the answer to that for sure, but obviously you can see 'wrong' temperatures while you can't see the contact patch or appreciate the fact there IS a contact patch being modelled, instead of a simpler model that just produces an inside-middle-outside temperature spread and compares that to an optimum value to produce grip.

If rF2 has portions of the tyre running 10° too cool, that has an effect on available grip (defined in the [Realtime] section). Is that inaccuracy worse than not modelling how much of the contact patch is in a static state, sliding state (adhesion, micro- and macro-deformation), and the pressure being applied to each of 1500 points across that contact patch? You could look at it the other way: a tyre being modelled to such a degree is producing temperatures that are close to being correct, which reflects how well the model is working.

 

[Lgel]

I agree with you, current car simulators need much more power than what is available on our desktop, and they just try to simulate one car, not a whole field of cars racing against each other.

I simply ask if this model is the better for a game (yes RF2 is a game, a sim game, but a game)?

How many man years of development are still needed?

How many times will modders be obliged to revise their cars to use the improved model if they want to implement the last improvements to the tire model?

Cheers.

 

[Jameswesty]

May I remind you that the actual RF2 tire model is still not working after several years in development (try to tune camber by tire temps for instance).

Will it work someday (represent closely the behaviour of the real tire modelled) without requiring manual adjustments to have a good driving car?

There is nothing worse that a very sophisticated model that doesn't work, or needs data you have no way of determining precisely.

It reminds me of econometric modelling at a place I worked, a whole team of guys very well paid (all of them PHD in econometrics, one of them would won a worldwide prize of econometry) would work for a year using very expensive computer resources at the time (processing a lot of data).

The model would predict a growth of 2,1 % for the economy of a given country. The chief economist would consider this data, and replace it by a 2,55 % without any solid evidence (based on his feelings). This country would grow at a rate that was neither 2,1 % nor 2,55 %.

It went along for 10 years, I left, I suppose the same is still going on today.

Cheers.

This guy is so right its not funny

People have to understand that a Model is Model and its only as good as its end result.

More complex does not by default mean better,more realistic,more or less usable.

Models are simply tools and its all about using the best tool for the desired end goal.

Now of course there will be certain utility that comes from the RF2 tire model with specific aspects some might want so in that case sure RF2 tire model is the best choice.

But when it comes to people driving at home against AI or online If the end result produces cars that handle and race well and ideally don't have weird setup exploits that damage the quality of racing then that's all that really maters , weather or not you are using Tire model 1 Tyre model 2 or Postman pat racing is irrelevant.

Commercially as well (which is generally important to sustain development of a sim) Even if RF2 tire model is all round better if it turns out modders simply cannot use it (despite better documentation)then it makes perfectly good sense to support the old models.

The worry I think people have is that they then think ISI is setting the president for accepting lower quality content, but if AMS has shown anything and following on from what I said above I'm not so sure this is a valid concern.

Separate to tries, I'd ague high resolution tracks ideally based on laser data has more effect to the end drive and perception of cars more so than the differences between a very well made car such as what you see in AMS vs cars in RF2.

 

[Miro]

This disscussion became absurd to a degree I can not explain myself. Now it is not about if it is a good idea to allow the rf1 tire model but if it is a better tire model.
Are we really discussing this? Really? Do you people even play this game? Do you ever race propperly within this game?
You are talking here about issues with temps being here for years so you want to swap this with the rf1 model? Im I nutts or is the world upside down?

Seriously they should shut down the rf2 tire model for 3 months ans swap it with rf1 tires and see how happy you really would be. I think it might be fairly easy to do.

 

[Euskotracks]

Both tire models should be better related one to the other.

For a given rf2tire model there should be one rF1model as an ideal. There should be a tool to make this translation in both ways.
rF1 model is used by the AI and they should be quite equivalent in order to have the fairest possible competition in offline races.

URD T5 series for example shows no difference between tire compounds for the AI. This tool would avoid things like this

Enviado desde mi ONE A2001 mediante Tapatalk

 

[TJones]

I think i have to relativate my post above a bit. I don't want to give the impression that actual rF2 tyre model is faulty. The fact that i dont understand temperature conductance between different parts of a tyre, doesn't mean there must be a bug. Maybe my assumption about meassurement points is wrong, maybe there is a other reason i can't see. This question is still open.
The base of the model is good enough for lots of years to come, there may be still some tweaks necessary, the full potential isn't exhausted anyway.

BTT:
Make old rFactor tyre model usable in rf2 sounds very interesting at first, as it make rF2 more accessible for other modders. handling and development is also easier and faster.
It also raises some important questions:
- What about RealRoad tech, rF1 tyremodel doesn't support it.
- What about Rain
- If you want to use rF1 and rF2 cars in the same event, there might be circumstances where one tyre was an big advantage or disadvantage in a particular driving state.
So maybe rF1 tyre model has to be raised to rF2 level.

Not to forget there is still room for improvement left regarding support and documentation of rF2 model. Just let Michael Borda continue his work at his interesting (but not easy to follow )Dev-Blog, helpful spreadsheets, filling the gaps in the list of example tyres and give some hints to tweak a tyre to your liking.

 

[stonec]

This disscussion became absurd to a degree I can not explain myself. Now it is not about if it is a good idea to allow the rf1 tire model but if it is a better tire model.
Are we really discussing this? Really? Do you people even play this game? Do you ever race propperly within this game?
You are talking here about issues with temps being here for years so you want to swap this with the rf1 model? Im I nutts or is the world upside down?

Seriously they should shut down the rf2 tire model for 3 months ans swap it with rf1 tires and see how happy you really would be. I think it might be fairly easy to do.

You can already drive the rF1 tire model in rF2 dev mode by setting "Old Tire Model" to true in devmode player.JSON.

I'm also confused if people seriously want the rF1 tire model back on ISI cars, as ISI made it clear it would only be for modders that find rF2 model difficult to work with. The rF1 model is inferior in so many ways, realistic flatspots and tire deformation for instance is never going to be the same on a mathematical model as on a physical model. The sliding / over limit behaviour is also so much worse on the older model that it's not even funny. I spent years working with the rF1 model in my league, and the handling when pushed over the limit never got even close where rF2 is today.

 

[Korva7]

You can already drive the rF1 tire model in rF2 dev mode by setting "Old Tire Model" to true in devmode player.JSON.

I'm also confused if people seriously want the rF1 tire model back on ISI cars, as ISI made it clear it would only be for modders that find rF2 model difficult to work with. The rF1 model is inferior in so many ways, realistic flatspots and tire deformation for instance is never going to be the same on a mathematical model as on a physical model. The sliding / over limit behaviour is also so much worse on the older model that it's not even funny. I spent years working with the rF1 model in my league, and the handling when pushed over the limit never got even close where rF2 is today.

+1 Especially to that The sliding / over limit behaviour thing.