iRacing | 2019 Content and Features Review Video

[Paul Jeffrey]

iRacing recently released a rather cool video - looking back on the many achievements by the simulation over the course of the last 12 months.

Reviewing in video form the changes within iRacing that have happened since last December, the American racing simulation has certainly undergone some significant development this past year - adding many new features and pieces of content as they bid to grow and advance within the sim racing genre.

Literally too many changes and content releases have occurred since December 2018 to write down here in list form, but I'm sure many of you will agree that the improvement to weather and time of day changes have been welcomed, as has the addition of AI and a brand new damage model... but to get the full flavour of the changes, check out this awesome new video from the developers:

Got questions about the sim? Ask our community and start a thread at the RaceDepartment iRacing sub forum!

 

[Mike McCormick]

That sim racing paddock article is nearly 5 years old (March 31, 2015). It's not really relevant as the tire model on all of the sims is different now.

 

[mozzie]

Or not.
Also you seem confused about what TM popular sims use. pCars1&2 , AC, ACC, rF2 don't use Pacejka model or similar. rf2 also uses TM based on physical tire construction. AMS and Raceroom are based on Pacejka model, but it doesn't mean they are less accurate.
And using lookup tables instead of realtime calculations doesn't mean it's inferior by any means. It's actually the opposite, because you can factor in much more details and make way more advanced calculations.

The Pacejka model and lookup tables is fine until one goes over the limit, then that's when it falls apart, and then the dev's have to do a whole lot of fudging to make it work- this is one of the many reasons Dave K. wanted to take a different approach.
The other problem with using lookup tables and the Pacejka method is that the dev's have to use a whole bunch of what they call, "magic numbers" to make it work. These are numbers that have no meaning and are just thrown in to make the tyre model work. This is another reason Dave wanted to develop another Tyre model.
Do you have proof that AC, ACC and rFactor 2 are not using lookup tables, or a form of? I haven't heard or read anything to prove that they are not. Also how can you possibly know that using a lookup table is more accurate than a real time model? How can you possibly know that without being privy to Dave K.'s code personally?

EDIT: Wanted to add, another issue with a lookup table or Pacejka model is that as a dev you are totally dependent on the information that you are supplied with by whoever did the original measurements of that tyre. So if that person screwed up your entire model for that tyre will be wrong. Sh*t in, sh*t out. Even if it “feels” believable it’s still wrong and you will never get an accurate simulation from that.

iRacing take cuts of the tyre they are trying to model and send it to a lab to find out what properties are in that particular tyre they are modelling, (example: carbon black, sulphur quantities, etc...) they also look at how the ply and carcass is built up and then they are able to build their tyre model from that info. So, in other words, they are responsible for the entire process of data capture and building the tyre model.

 

[BrunoBæ]

iRacing take cuts of the tyre they are trying to model and send it to a lab to find out what properties are in that particular tyre they are modelling, (example: carbon black, sulphur quantities, etc...) they also look at how the ply and carcass is built up and then they are able to build their tyre model from that info. So, in other words, they are responsible for the entire process of data capture and building the tyre model.

EDIT: Urgent! You forgot that Dave also are measuring tire/tarmac stuff on the quantum level.

 

[mozzie]

https://www.iracing.com/the-sticking-points-in-modeling-tires/

Log in | iRacing.com™ Motorsport Simulations

members.iracing.com

I don’t think Dave ever claimed he was modelling/capturing on the quantum level. Please offer proof that he said this. And, here’s some smileys for you too. You seem like you need them in your life.

 

[BrunoBæ]

 

[mozzie]

I knew I knew you.

The thing with iRacing is, (unlike rFactor 2/pCars/RaceRoom, etc...) one cannot hide behind an anonymous handle. It's ones name and reputation bare for all to see. Some people find it hard to come to terms with the realisation that they are not as good as they had envisioned and therefore look for excuses, (especially when they claim they were racing Formula Fords in their younger days), "it's the crappy tyre model", "it's all these idiot people crashing into me", "it's the incident safety rating that's holding me back", so on and so forth... By the way, I've also met these types of people at real race tracks. They forget to realise that it's the same for everyone on the service- bad tyre model, or not. We all have to adapt and drive with the same physics model and drive with the same crazy drivers on the same piece of virtual tarmac.

Now I have better understanding of why you keep going on about iRacing the way you do.

"Elvis has left the building."

 

[BrunoBæ]

Hahaha.
I dont know what are you rambling about.
Hehe and you say that it is ME who have a problem with MY life.
If I wasnt so polite I would say that you need to have your head examined.

CatsAreTheWorstDogs: If you really want to check what BrunoB is capable of in a real sim then just check Racerooms leader bord for the FR90 series. Its free. Even for you. But like most people who have some kind of wacky agenda you are probably not interested in some simple facts.

Hahaha. I cant stop laughing - I have never heard about such a sick obsession. Just get help my good man

 

[avenger82]

The Pacejka model and lookup tables is fine until one goes over the limit, then that's when it falls apart, and then the dev's have to do a whole lot of fudging to make it work- this is one of the many reasons Dave K. wanted to take a different approach.
The other problem with using lookup tables and the Pacejka method is that the dev's have to use a whole bunch of what they call, "magic numbers" to make it work. These are numbers that have no meaning and are just thrown in to make the tyre model work. This is another reason Dave wanted to develop another Tyre model.
Do you have proof that AC, ACC and rFactor 2 are not using lookup tables, or a form of? I haven't heard or read anything to prove that they are not. Also how can you possibly know that using a lookup table is more accurate than a real time model? How can you possibly know that without being privy to Dave K.'s code personally?

EDIT: Wanted to add, another issue with a lookup table or Pacejka model is that as a dev you are totally dependent on the information that you are supplied with by whoever did the original measurements of that tyre. So if that person screwed up your entire model for that tyre will be wrong. Sh*t in, sh*t out. Even if it “feels” believable it’s still wrong and you will never get an accurate simulation from that.

iRacing take cuts of the tyre they are trying to model and send it to a lab to find out what properties are in that particular tyre they are modelling, (example: carbon black, sulphur quantities, etc...) they also look at how the ply and carcass is built up and then they are able to build their tyre model from that info. So, in other words, they are responsible for the entire process of data capture and building the tyre model.

You sound like iRacing fanboy(or at least Dave Kaemmer's) that fell into their the marketing completely without understanding tire modes. You're confusing Pacejka model with lookup tables, when it falls apart etc. Comparing Pacejka based model to lookup tables..., like they are something similar.
These are separate things, lookup table is just common optimisation technique, not something inherently wrong that breaks or limits an implementation.
LOOKUP TABLES DON'T MAKE SUCH TIRE MODEL INFERIOR. And BTW: Pacejka based models have advantages which you probably don't know about, that's why they are still popular.
"Despite using a ‘simpler’ model that basically has a base slip curve which is stretched or shrunken depending on conditions. It still has a considerable number of variables to set. There is enough information available for a dedicated user to recreate the model in a spreadsheet so you know exactly what the tire does at any slip angle, load, camber, pressure and temperature. "
It's explained in this video:

Of course making complex calculations work in realtime (especially something complicated as tire model) is very admirable from technical/engineering point of view. For instance I think even pCars 1 and AC use realtime TM calculations (also not based on empirical values).
But like I said, that doesn't make such tire model any better. If anything iRacing TM was and is still considered it's weak point by many. The "holy" NTM7 is just an evolution and still has significant issues. I'm not expert on tire models in sims and I'm not even sure if iRacing TM isn't using lookup tables for something.
But again, in fact it(realtime approach) has obvious cost - computational limitation. It cannot be too complex and can't simulate many factors, or even details, nuances as lookup based approach. Best example is rF2 TM. It's so advanced , takes much more variables, simulates so many things that you couldn't run it realtime on fastest NSA supercomputer. It uses a brush model(contact patch) in conjunction with "dreadful" lookup tables. It can take like a day (for their tTool) to compute all possible conditions into the lookup tables. Like in iRacing, rF2 tires are also based on physical tire construction.
"rFactor 2 has a physical thermo-mechanical tire model, based on real tire construction from first principles, with full carcass, thread and contact patch simulation, where peak slip angles dynamically change with temperature, pressure and camber, instead of staying static. And don't forget proper flatspots resulting in mass imbalances and some more. Such a tire model alone took many years of research and development effort. rF2 also has an updated collision system, with vastly improved wheel-to-wheel collisions for open-wheel cars.

It also has fully dynamic Real-Road and wet weather / rain with dynamic drying line based on where the cars actually drive.”

Yet it doesn't mean rf2 tires are necessarily most accurate, mostly because they need accurate tire construction data which is often not available, especially for modders.
I still think iRacing NTM7 (and even current NTM6) is great engineering achievement and honestly wish that David K. hard work it will really pay off(sooner than later). I would probably re-sub and invest more in iRacing.

 

[mozzie]

You sound like iRacing fanboy(or at least Dave Kaemmer's) that fell into their the marketing completely without understanding tire modes. You're confusing Pacejka model with lookup tables, when it falls apart etc. Comparing Pacejka based model to lookup tables..., like they are something similar.
These are separate things, lookup table is just common optimisation technique, not something inherently wrong that breaks or limits an implementation.
LOOKUP TABLES DON'T MAKE SUCH TIRE MODEL INFERIOR. And BTW: Pacejka based models have advantages which you probably don't know about, that's why they are still popular.
"Despite using a ‘simpler’ model that basically has a base slip curve which is stretched or shrunken depending on conditions. It still has a considerable number of variables to set. There is enough information available for a dedicated user to recreate the model in a spreadsheet so you know exactly what the tire does at any slip angle, load, camber, pressure and temperature. "
It's explained in this video:

Of course making complex calculations work in realtime (especially something complicated as tire model) is very admirable from technical/engineering point of view. For instance I think even pCars 1 and AC use realtime TM calculations (also not based on empirical values).
But like I said, that doesn't make such tire model any better. If anything iRacing TM was and is still considered it's weak point by many. The "holy" NTM7 is just an evolution and still has significant issues. I'm not expert on tire models in sims and I'm not even sure if iRacing TM isn't using lookup tables for something.
But again, in fact it(realtime approach) has obvious cost - computational limitation. It cannot be too complex and can't simulate many factors, or even details, nuances as lookup based approach. Best example is rF2 TM. It's so advanced , takes much more variables, simulates so many things that you couldn't run it realtime on fastest NSA supercomputer. It uses a brush model(contact patch) in conjunction with "dreadful" lookup tables. It can take like a day (for their tTool) to compute all possible conditions into the lookup tables. Like in iRacing, rF2 tires are also based on physical tire construction.
"rFactor 2 has a physical thermo-mechanical tire model, based on real tire construction from first principles, with full carcass, thread and contact patch simulation, where peak slip angles dynamically change with temperature, pressure and camber, instead of staying static. And don't forget proper flatspots resulting in mass imbalances and some more. Such a tire model alone took many years of research and development effort. rF2 also has an updated collision system, with vastly improved wheel-to-wheel collisions for open-wheel cars.

It also has fully dynamic Real-Road and wet weather / rain with dynamic drying line based on where the cars actually drive.”

Yet it doesn't mean rf2 tires are necessarily most accurate, mostly because they need accurate tire construction data which is often not available, especially for modders.
I still think iRacing NTM7 (and even current NTM6) is great engineering achievement and honestly wish that David K. hard work it will really pay off(sooner than later). I would probably re-sub and invest more in iRacing.

I've never claimed to be an expert or have any real deep knowledge of tyre modelling, and never will. I completely understand that lookup tables and the Pacejka model have advantages over a physically based model, I never claimed otherwise. What I was pointing out was that they are also flawed. The issue I had was with your comment claiming that an empirical model is more accurate than a physically based model without you having any data to prove your claim.

I am completely aware that the iRacing tyre model is not perfect and has many faults and anomalies to work out, also with V7. The thing that gets me is all these so called, "experts" that come out claiming that the iRacing tyre model is completely flawed compared to other sims, which just isn't the case.
For example, using rFactor2 (in the Oreca 07) it takes 30 seconds looking at a tyre heat graph in Motec to see that it's not simulating tyre heat correctly, increasing heat while driving in a straight line and dropping in temp in the corners!? In which reality does that happen?
Also, the wet weather characteristics in rFactor 2 leave a lot to be desired. The wet weather racing line is the same as in the dry, this is because the only thing rFactor 2 is taking into account is where the cars are driving, (yes it may be dynamic but it's also dynamically wrong), it's not taking rubber build up or pooling of puddles into the equation. It's a grossly simplified model where it's just percentage change of grip from a dry race track to wet. And then, on a wet track the dry line is just a percentage in grip difference over the wet part. Increasing tyre pressures on a wet tyre in rF2 also gives a complete opposite effect to tyre temp as it is in the real world. Try it for yourself and look at the data in Motec.

I completely get Dave Kaemmer's model is not perfect and is flawed in many ways, however, all the people claiming that other sims tyre models are far superior need to take a reality pill and look at the Motec data to find out that their favoured sims' tyre models are just as flawed.

 

[BrunoBæ]

You sound like iRacing fanboy

Fully agree

CatsAreTheWorstDogs: Thanks for the interview Krzysztof. Makes a lot better sense than blindly defending everything King Kaemmer is outpouring.

 

[avenger82]

The issue I had was with your comment claiming that an empirical model is more accurate than a physically based model without you having any data to prove your claim.

Where did I said that? I said:
“AMS and Raceroom are based on Pacejka model, but it doesn't mean they are less accurate.”.
Meaning you can have quite accurate tires in simpler Pacejka based model, and unrealistic in physical based and vice versa. It depend on implementation, particular car/mod, whether they used accurate real data etc.

As for tires (in any car) in rF2 dropping temps in corners and gaining on straight: I haven’t experienced that and didn’t even heard of something similar. Maybe you found a bug, but If that’s hard to believe no user found it. If it was common no one would play rF2. I think it’s more probable there’s something wrong with your interpretation of telemetry. Maybe you look at inner temps, which have some “lag”. In The Oreca 07 someone discovered bad aero, but it uses same tires as I.e. GTE cars.
Regarding rF2 tires in rain, yes I know they have significant issues, but I have too little knowledge about this.
In my opinion racing sim devs should focus on making tires realistic on dry tarmac, before spending too much resources on rain etc.

 

[maelstrom]

Meaning you can have quite accurate tires in simpler Pacejka based model, and unrealistic in physical based and vice versa. It depend on implementation, particular car/mod, whether they used accurate real data etc.

I agree. Saying that you're building a completely new tyre model which is based on first principles (not using lookup tables) doesn't necessarily mean that it's going to be any better than what already exists. Whilst it may, eventually, prove to be better (and there's absolutely no guarantee of this), the chances are that it will probably take some time to just match the status quo.

 

[mozzie]

Where did I said that? I said:
“AMS and Raceroom are based on Pacejka model, but it doesn't mean they are less accurate.”.
Meaning you can have quite accurate tires in simpler Pacejka based model, and unrealistic in physical based and vice versa. It depend on implementation, particular car/mod, whether they used accurate real data etc.

As for tires (in any car) in rF2 dropping temps in corners and gaining on straight: I haven’t experienced that and didn’t even heard of something similar. Maybe you found a bug, but If that’s hard to believe no user found it. If it was common no one would play rF2. I think it’s more probable there’s something wrong with your interpretation of telemetry. Maybe you look at inner temps, which have some “lag”. In The Oreca 07 someone discovered bad aero, but it uses same tires as I.e. GTE cars.
Regarding rF2 tires in rain, yes I know they have significant issues, but I have too little knowledge about this.
In my opinion racing sim devs should focus on making tires realistic on dry tarmac, before spending too much resources on rain etc.

What you wrote in yoir first post replying to mine, Quote: “And using lookup tables instead of realtime calculations doesn't mean it's inferior by any means. It's actually the opposite, because you can factor in much more details and make way more advanced calculations.”
The way I interpret this paragraph is you are implying that an empirical model is more accurate/advanced than a physically based model. How can you know that? Maybe I interpreted it wrong? Please correct me if I did interpret it incorrectly.

I think there are so many unknowns that neither you, or I, or anyone besides those that have any real scientific data to prove otherwise could possibly know the answer to this complex subject.

You seem like a level headed guy that has provided some good points. I have read your opinions and taken them onboard. I’ve come to realise, what’s the point of the discussion? It’s not improving the quality of your life, or mine. I respect every dev that is able to put some form of tyre model together. I think it’s awesome that there can be so many different approaches to a problem. The fact of the matter is, everyone has their preferred sim and that’s all good. Each to their own.

 

[avenger82]

I agree. Saying that you're building a completely new tyre model which is based on first principles (not using lookup tables) doesn't necessarily mean that it's going to be any better than what already exists. Whilst it may, eventually, prove to be better (and there's absolutely no guarantee of this), the chances are that it will probably take some time to just match the status quo.

NTM is for sure ambitious project and is impressive it's basically one man behind it(AFAIK). My subjective opinion is that NTM 6 already isn't bad, and of course NTM 7 fixes some issues . Maybe in another ten years it would already be objectively most realistic, who knows. But competition is not sleeping - as you'd say in my country.
IMHO currently rF2 most realistic tire model, but it requires detailed input data to create new tire, which is hard to get. Also has some long standing issues and it's development seem to stagnated.

Project Cars Revolution is already 200% better than pCars2, so when released will have ultimate TM

 

[avenger82]

The way I interpret this paragraph is you are implying that an empirical model is more accurate/advanced than a physically based model. How can you know that? Maybe I interpreted it wrong? Please correct me if I did interpret it incorrectly.

I think you still confuse empirical models with lookup tables. rF2 has physical based TM, but also uses lookup tables. Lookup tables allow you to perform some complex pre-calculations, which you'd never be able to perform in realtime. As I said it's programming optimization. So for a TM you can factor in more variables, use much more complex and time consuming calculations. Then when simulation is running you can just read some readily available pre-calculated results from RAM saving CPU power for some other things. In other words in realtime it's algorithms must be way faster, sometimes making shortcuts and simplifications.

 

[BrunoBæ]

Saying that you're building a completely new tyre model which is based on first principles (not using lookup tables) doesn't necessarily mean that it's going to be any better than what already exists. Whilst it may, eventually, prove to be better (and there's absolutely no guarantee of this), the chances are that it will probably take some time to just match the status quo.

In most other areas like in the real world then when somebody proudly does announce that he is taking a complete other approach than enybody else - even if we keep quantum theories outside - then the normal reception will be lets wait some reasonable(!) time to see the outcome.
Before we decide if we have a new Münchausen or a new Einstein here.

But if this proudly announced new revolutionary approach after 10-12 years not even are on level with the best of the competitive solutions in the same area - then its probably time to be a bit more sceptical when further outpourings about this revolutionary "new" approach keep comming from the same man.

CatsAreTheWorstDogs: Be aware Im not showing the same picture again about the man riding his imaginary cannon balls - this time.