Dr. Kazunori Yamauchi Gives Lecture on Gran Turismo's Driving Physics & Production

7HO
So much yes. Besides the way people confuse interpreted physics I'm constantly amazed how simple comparison to real life seems beyond so many people. I haven't found a realistic sim yet, they all have their own issues. But I prefer some based on how I interpret the experience to be more authentic or how I prefer the end result over the end result of another.

Here's something I like. PCARS2 gets talked about a lot for its tyre model but the grip as arcade like in that game so everything the tyre model does is lost due to poor implementation. The tyres go from comically high grip to comically low grip and very rarely represent something realistic, PC2 gets around this ultra grip in the dry with comical handling, the result is something close to the difficulty of driving a real car in the dry but not a realistic driving experience IMO, in the wet with standing water it's just comical. Now GTS on the other hand lacks Longitudinal grip and has tyre heating issues but gets around these with driver aids like traction control. I don't like the grip issues in GTS but the result is a superior driving experience IMO, other people like PC2 better for whatever reason.

Each sim has its own issues but most people don't seem to be aware of the various issues each sim has and still these people argue over which one is more realistic or more Simcade. Much of it comes from people basing their expectations based on their conditioning to other games they have played but physics is based on the real world so any driving experience should be compared to what we find in the real world. At the end of the day no one is getting this right yet so our preference is simply our preference.

Great explanation ! I agree with you.
 
7HO
Well I wasn't as I'm sure you are aware and I hope no one else does based on my comment.
Referring to other titles (or any element of them) as arcade isn't a good way to go about things if that's not you intention.

We have threads to discuss 'vs', please use them. The staff have already had to spend a considerable amount of time cleaning up this thread ans we would rather not have to do so again.
 
Grip is added as we know when looking at car files (which we are not supposed to know about)
.ini files have many lines of hexadecimal code and one is the grip coefficient, which of course can be lowered or raised to match data from the real car
Grip coefficent shouldn't be a fixed value for a car.

The coefficent of grip should be varying based upon the tyre make-up, its load, the current slip angle and percentage, pressure and a whole host of other factors.

The are in reality represented by a rather complex series of interlinked graphs that allow the co-eff to be arrived at for the combination of factors, and then adjusted again by the Mu factor of the final value for the tyre and the track surface.

That's how tyre models are build in reality and I would hope that GTS is not still using a single value for the car, as the rest of the sim industry moved to tyre curves quite a while ago.
 
7HO
So much yes. Besides the way people confuse interpreted physics I'm constantly amazed how simple comparison to real life seems beyond so many people. I haven't found a realistic sim yet, they all have their own issues. But I prefer some based on how I interpret the experience to be more authentic or how I prefer the end result over the end result of another.

Here's something I like. PCARS2 gets talked about a lot for its tyre model but the grip as arcade like in that game so everything the tyre model does is lost due to poor implementation. The tyres go from comically high grip to comically low grip and very rarely represent something realistic, PC2 gets around this ultra grip in the dry with comical handling, the result is something close to the difficulty of driving a real car in the dry but not a realistic driving experience IMO, in the wet with standing water it's just comical. Now GTS on the other hand lacks Longitudinal grip and has tyre heating issues but gets around these with driver aids like traction control. I don't like the grip issues in GTS but the result is a superior driving experience IMO, other people like PC2 better for whatever reason.

Each sim has its own issues but most people don't seem to be aware of the various issues each sim has and still these people argue over which one is more realistic or more Simcade. Much of it comes from people basing their expectations based on their conditioning to other games they have played but physics is based on the real world so any driving experience should be compared to what we find in the real world. At the end of the day no one is getting this right yet so our preference is simply our preference.
Do you mind me asking which other sims you've tried? You mention Gran Turismo Sport and Project Cars 2. Have you driven a car in rFactor 2? Live For Speed? Heck, even Assetto Corsa? I only has because a sample size of 2 doesn't support a good argument, in my eyes.

I sort of agree with the second passage I highlighted. Each sim does have it's own issues, however, I believe the issues GTSport has are much more significant than the issues rFactor 2 has when it comes to simulating the driving experience on a consumer grade computer. Remember what the great George Orwell told us. “All sims are equal, but some sims are more equal than others.” :sly:
 
Last edited:
Referring to other titles (or any element of them) as arcade isn't a good way to go about things if that's not you intention.

We have threads to discuss 'vs', please use them. The staff have already had to spend a considerable amount of time cleaning up this thread ans we would rather not have to do so again.
Well that's just silly and it is why context is important. Every game has arcade or gamey or unrealistic elements, that's been one of my key arguments against the word Simcade, that word is too vague. Anyone who reads my entire post can easily see that the point of my post is that people wrongly argue one is better based on subjective experience. My point is all sims are flawed and ultimately people are just trying to justify their preferences. I have my preference, I know people disagree, I might find it difficult to understand why they prefer a particular thing I don't really care for but I understand that we all have different taste and taste is neither right or wrong, it is simply a preference.

Coming back to the start, context is key here. Everything in that post should be first taken as a response to the posts below and then looked at in it's entirety. If someone is to pick out single words and misapply them then that person is at fault not me.

why they dont implement physics in gts insteed of talking about it? gt-s is poor like all other gt´s.
It's not a PD's problem if some people are unsensitive and confused about physics and physics feel in simulators.
7HO
So much yes. Besides the way people confuse interpreted physics I'm constantly amazed how simple comparison to real life seems beyond so many people. I haven't found a realistic sim yet, they all have their own issues. But I prefer some based on how I interpret the experience to be more authentic or how I prefer the end result over the end result of another.

Here's something I like. PCARS2 gets talked about a lot for its tyre model but the grip as arcade like in that game so everything the tyre model does is lost due to poor implementation. The tyres go from comically high grip to comically low grip and very rarely represent something realistic, PC2 gets around this ultra grip in the dry with comical handling, the result is something close to the difficulty of driving a real car in the dry but not a realistic driving experience IMO, in the wet with standing water it's just comical. Now GTS on the other hand lacks Longitudinal grip and has tyre heating issues but gets around these with driver aids like traction control. I don't like the grip issues in GTS but the result is a superior driving experience IMO, other people like PC2 better for whatever reason.

Each sim has its own issues but most people don't seem to be aware of the various issues each sim has and still these people argue over which one is more realistic or more Simcade. Much of it comes from people basing their expectations based on their conditioning to other games they have played but physics is based on the real world so any driving experience should be compared to what we find in the real world. At the end of the day no one is getting this right yet so our preference is simply our preference.

As you can see here only the first post is provocative. oneloops reply was on point and relative to the thread. My reply expands on his and uses specific examples while pointing out they are not isolated and that every single sim has faults and unrealistic elements and ultimately the way these are combined and then interpreted through FFB and graphical representation and settings is how we interpret the experience. When we evaluate physics I have not found an absolutely realistic sim yet. What we are left with is preference. That is in direct rebuttal to user3392345 saying GTS is poor like all other GT's and claiming GTS does not implement physics. He made an absurd statement that if he has even played the game is based on interpretation and in context it can be argued that all sims are flawed, we just have our personal preference.

Nothing about it is us vs them, it is the exact opposite illustrating that us vs them is stupid. Any implication that claims it is us vs them must be based on either not reading the post in it's entirety or a prejudice of some kind most likely one stemming from feelings certain words provoke in certain people. But that isn't my fault or the words fault, if that is the case it is only the fault of the person who responds that way.

Do you mind me asking which other sims you've tried? You mention Gran Turismo Sport and Project Cars 2. Have you driving a car in rFactor 2? Live For Speed? Heck, even Assetto Corsa? I only has because a sample size of 2 doesn't support a good argument, in my eyes.

I sort of agree with the second passage I highlighted. Each sim does have it's own issues, however, I believe the issues GTSport has are much more significant than the issues rFactor 2 has when it comes to simulating the driving experience on a consumer grade computer. Remember what the great George Orwell told us. “All sims are equal, but some sims are more equal than others.” :sly:

I'll start off by saying we have been told not to let this degrade into an us vs them. So I'll only answer the part you want to know without crossing a line. I have played racing games for over 25 years. I learned to drive in a simulator originally, Race Drivin'. I don't have Live for Speed. I have played all the Simbins, both rFactors 1 and 2, Assetto Corsa, both PCARS, I have multiple years experience as an iRacing member and R3E. I feel like I'm leaving some out. I've played plenty of other games too from the Need for Speed series, Burnout Paradise and earlier Burnouts, GT's, Forza's, Dirt series and these are just the ones that have left the biggest impression on me as I have many more car games than those, I tend to have only really focused on car games over the years but more recently they all started to lose their shine. Gt Sport has been the only one that I've been really enjoying despite the things I don't like recently. I'm not going to go into what puts me off RF2 especially in this thread and I've done so plenty of times in the past (I'm not sure if I have here), I'll just say that I have never been able to understand the hype of both RF1 and RF2 and always preferred other sims. If others enjoy them fine, I can absolutely explain the way I feel about those but this isn't the place as we have been told. At the end of the day view it as a preference based on qualities we each consider more critical to our end experience. Neither of us is wrong.

I don't like Ian Bell but I think something he talked about is key to this when he was discussing PC2 on the Spottheozzie stream. He discussed how in his opinion the technology used in the simulation is more advanced than what other sims are using but he pointed out that doesn't make a sim better or more realistic to the end user. Ultimately what he was getting at is the experience of the end user is more important than how you get there or how many features you have and I agree with that. The problem with the end user is we are all individual with different priorities and preferences.
 
Erm you do realise an honorary doctorate does not give you the title Dr.

There was a news story recently of someone misusing it, can't find it though.
 
@7HO Thank you for the detailed response. I asked you those questions, not to be snarky, but to see if you have had any experience outside of GT/PCars and you have a considerable amount. Nothing bothers me more than someone offering an opinion on a title if they haven't even played it yet. Believe it or not, I've run into quite a few of these people. BTW, I'm with you on rF2...it doesn't do it for me either.

I would have loved to been able to sit here and watch a full presentation just like the one Kaz gave, if it weren't for the necessary translations. In fact, I would love to learn as much as I can about every mainstream title available today. There's no doubt the lecture would be very insightful.
 
Kaz should've given a demo of the 2015 Honda Civic Type R at Brands Hatch Indy to demonstrate the superior physics simulation of GT Sport such as extreme understeer etc.
 
Kaz should've given a demo of the 2015 Honda Civic Type R at Brands Hatch Indy to demonstrate the superior physics simulation of GT Sport such as extreme understeer etc.
Or how about the oversteer of the RWD cars? Of course a FWD car is going to understeer when the tyres lack longitudinal grip, that's the real issue.
 
7HO
Or how about the oversteer of the RWD cars? Of course a FWD car is going to understeer when the tyres lack longitudinal grip, that's the real issue.
No FWD car that is worth its salt (besides a Point A to Point B car like a Yugo) would understeer like that as such low speeds. Not one.
 
IIRC, Kaz knows English, but he prefers to speak in his main language. He might not be as comfortable speaking English, or something.
It is just a usual sign of japanese perfectionism. Even if one knows english good, he/she never uses it in any serious or professional appearances. At least without _good_ reason.
A lecture should not be affected by the lecturer’s language skills, and there are professionals to do the translation.

Interesting to speak about 100% simulation or not.
Even if they’ve got a nearly perfect model of each subsystem, they can’t use it in a game. It is sold to the wide public, and not for those (37 :) ) individuals, who can handle the fact of setting up and tweeking for a month before the car runs its first meter on track. On a TV screen.
Just remember Richard Burns Rally. It was awesome, but easily get lost in parameters if you don’t know what they are.
Another question is that why we are speaking about simulation if the users’ huge percent uses the controller? What car is that, in everyday life, that is controlled by a d-pad and a few buttons...

Don’t get me wrong, I hate that “rally” feels like having fun with a hovercraft. But compromises always needed. A game is not simulating perfectly - a perfect simulation is not usable by most of us. And not fun. If you just struggling with simulated parameters, you will not think you are better than S.Vettel, W.Röhrl or S.Blomquist in 2 minutes.
If we have a 100% simulation, and a champions exact setup, we will get 2mins on a 1.5min track. Because of lack of feedbacks. No pressures and hits on our back - just to mention one.
 
Last edited by a moderator:
It is just a usual sign of japanese perfectionism. Even if one knows english good, he/she never uses it in any serious or professional appearances. At least without _good_ reason.
A lecture should not be affected by the lecturer’s language skills, and there are professionals to do the translation.
I worked for the Renault Nissan alliance, all of my Japanese colleagues who could speak English did so when addressing native English speakers or in situations when English was the predominant language in the room.

Its a preference he has; well either that or the Japanese First Lady shouldn't be considered serious or professional in this appearance.

 
I worked for the Renault Nissan alliance, all of my Japanese colleagues who could speak English did so when addressing native English speakers or in situations when English was the predominant language in the room.

Its a preference he has; well either that or the Japanese First Lady shouldn't be considered serious or professional in this appearance.


OK, call it a preference. A traditional one.
 
that sentence grab my intention too like: oh! Kaz seems to mean that if PD uses the realistic tyre model they managed to get, the game would be too hard for most of the customers so let's make it less " simulation like " to have an understandable game
#the real sweetened simulator :D

Only if real cars are undriveable.
He said that they are realistic in terms of lap times, nothing more.
 
Interesting to speak about 100% simulation or not.
Even if they’ve got a nearly perfect model of each subsystem, they can’t use it in a game. It is sold to the wide public, and not for those (37 :) ) individuals, who can handle the fact of setting up and tweeking for a month before the car runs its first meter on track. On a TV screen.
Why can't they use it in a game?

Driving a car is not difficult, millions do it daily. Driving a car up to the limit is not difficult, finding the limit is not difficult, with some basic training correcting minor issues at the limit is not difficult. I've taught all of these things.

The difficult part is finding the limit for every corner, in changing conditions and balancing the car at the point just over that limit, where the last vestiges of grip reside, consistently over a lap. For lap after lap.


Just remember Richard Burns Rally. It was awesome, but easily get lost in parameters if you don’t know what they are.
Another question is that why we are speaking about simulation if the users’ huge percent uses the controller? What car is that, in everyday life, that is controlled by a d-pad and a few buttons...
Which is why no title (interestingly, apart from RBR) uses 1:1 controller mapping for steering input.

Its quite possible to have input filters that 'smooth' rough inputs for controller users, and remove them for those that use a wheel.

That was you get the best physics engine you can, with the appropriate level of support for the input device being used.


Don’t get me wrong, I hate that “rally” feels like having fun with a hovercraft. But compromises always needed. A game is not simulating perfectly - a perfect simulation is not usable by most of us. And not fun. If you just struggling with simulated parameters, you will not think you are better than S.Vettel, W.Röhrl or S.Blomquist in 2 minutes.
No one has said that a perfect sim exists, but that doesn't mean it shouldn't be the aim, as the argument that it would not be usable for most of use is simply not true.

Personally a title that allows you to be as quite as the worlds best in 2 minutes is neither fun, nor doing its job as a sim very well (well unless you are that person).


If we have a 100% simulation, and a champions exact setup, we will get 2mins on a 1.5min track. Because of lack of feedbacks. No pressures and hits on our back - just to mention one.
Again that depends on the input device you are using and the rig you have.

As already said you can have input filters to allow for controller users and quite a bit can be done with rumble, as well as audio and visual cues to let you know whats going on (why do you think that many titles have exaggerated tyre squeal) .

Those fortunate to have wheels, etc can then still have FFB, tactile feedback and in some cases motion.

No good reason exists to not aim for the highest level of fidelity on a sim if your aim is a true sim. Now that's not to say that the market doesn't have a place for the 'hero fulfillment' titles, but it also doesn't mean that aiming to be the most accurate recreation of reality you can isn't fun. As fun is utterly subjective.
 
I think that's a far cry from a 2015 Honda Civic Type R in the dry though lol
You said no FWD car drives like that and I'm pointing out that you are incorrect and when there is no grip any FWD car with decent power drives just like that. As I keep saying the main problem with the physics is the lack of longitudinal grip. The wheels spin too easily.
 
Kaz should've given a demo of the 2015 Honda Civic Type R at Brands Hatch Indy to demonstrate the superior physics simulation of GT Sport such as extreme understeer etc.

In this video this Lemans driver says that he is very dissapointed because of terrible understeer in that car (sorry it's in french, I'll try to find another in english, but we can see the poor handling of the Civic specially at 3'48" and his face is not happy at all :odd: ) Maybe it should be drive with a little less understeering in GTS but it has a lot of understeering in fact.

 
Last edited:
I would argue that utilizing CFD for aero map generation is an extremely difficult task. Typical engineering turbulence models (i.e. simplifications) therein have significant uncertainties, and more detailed computations (where uncertainties may be a bit lower) are far beyond the reach of any sensible computational resources. If they were to thoroughly validate the maps with regard to experimental data the situation would be much better. While they may have a lot of pertinent data, this is surely not the case for the vast majority of cars and/or dynamic conditions (e.g. rake, slip angle).

These sentiments really extend to many other places such as the tire model. Increasing complexity to the nth degree (such as in a certain other available product) is in no way a guarantee for functionality. Ideally the models should also be subjected to reference test cases where detailed experimental data is available (such as those in aerodynamics, e.g. https://www.cfd-online.com/Wiki/Ahmed_body ). To date, I'm not aware of good quantitative studies regarding tire model functionality in the contemporary racing sims.
 
In this video this Lemans driver says that he is very dissapointed because of terrible understeer in that car (sorry it's in french, I'll try to find another in english, but we can see the poor handling of the Civic specially at 3'48" and his face is not happy at all :odd: ) Maybe it should be drive with a little less understeering in GTS but it has a lot of understeering in fact.


And all of the other FWD cars in GTS?

I've driven the Type-R, Megane and the ST and none of them understeer to the degree that they do in GTS.

No is the issue just around the understeer, but the physics engine slow reaction in regard to correcting it, with even the most severe lift-off returning grip ti the front end and getting these cars to start rotating in the way they should.
 
And all of the other FWD cars in GTS?

I've driven the Type-R, Megane and the ST and none of them understeer to the degree that they do in GTS.

No is the issue just around the understeer, but the physics engine slow reaction in regard to correcting it, with even the most severe lift-off returning grip ti the front end and getting these cars to start rotating in the way they should.

Yes, in GTS those cars should have less understeer as I said for the Civic. PD should do better with that issue. I hope they will solve it in next updates...
 
Why can't they use it in a game?

Driving a car is not difficult, millions do it daily. Driving a car up to the limit is not difficult, finding the limit is not difficult, with some basic training correcting minor issues at the limit is not difficult. I've taught all of these things.

The difficult part is finding the limit for every corner, in changing conditions and balancing the car at the point just over that limit, where the last vestiges of grip reside, consistently over a lap. For lap after lap.



Which is why no title (interestingly, apart from RBR) uses 1:1 controller mapping for steering input.

Its quite possible to have input filters that 'smooth' rough inputs for controller users, and remove them for those that use a wheel.

That was you get the best physics engine you can, with the appropriate level of support for the input device being used.



No one has said that a perfect sim exists, but that doesn't mean it shouldn't be the aim, as the argument that it would not be usable for most of use is simply not true.

Personally a title that allows you to be as quite as the worlds best in 2 minutes is neither fun, nor doing its job as a sim very well (well unless you are that person).



Again that depends on the input device you are using and the rig you have.

As already said you can have input filters to allow for controller users and quite a bit can be done with rumble, as well as audio and visual cues to let you know whats going on (why do you think that many titles have exaggerated tyre squeal) .

Those fortunate to have wheels, etc can then still have FFB, tactile feedback and in some cases motion.

No good reason exists to not aim for the highest level of fidelity on a sim if your aim is a true sim. Now that's not to say that the market doesn't have a place for the 'hero fulfillment' titles, but it also doesn't mean that aiming to be the most accurate recreation of reality you can isn't fun. As fun is utterly subjective.
You are quite right. I think mostly the same, maybe my english is not food enough to summarize with the proper expressions ;)
I just wanted to point out that
- the enjoyable experience is subjective and has a wide wide range in a PS title’s users.
- 100% simulation is hardly achievable because of the limitations of platform. Another example: you can turn your head and have a periferical sight of 360 degrees IRL, while in a top rig you can have lets say 3 monitors - which is far from reality. Tactile feedback can be nailed but what about your other parts besides your hands? And that’s why...
- these titles are always about the balance of modelled reality and “playable”, usable methods - with the widely available rigs / circumstances, as I think. I can imagine that making those filtering and making your model “dumber” than it could be is a pain for the developers, but a necessity in terms of marketing and having a “succesful” product on market.

As a sidenote, let me mention Norbert Michelisz, as a real WTCC driver and a sim rig developer, and active user of racing factory simulators, not games. He use to say “quite close” but never “is the same”.
 
Back