Dr. Kazunori Yamauchi Gives Lecture on Gran Turismo's Driving Physics & Production
- Thread starter GTPNewsWire
- 232 comments
- 22,304 views
- 2,321
- Wales
- TSR_GTWelsh
So how do you determine if it has different temps for inside and outside of tyre and/ or carcass/surface temp? How do you know if the flex effects the tyre actions? How does slip angle effect grip, how is slip angle handled in general too that was a huge talking point in PCARS. You need more data than we have to really gain an informed opinion on these points, with just playing the game it's "best guess" .
Also this was already replied to, I got the wrong end of the stick and didn't realise past GTs were being talked about.
Scaff
Moderator
- 29,033
- He/Him
- ScaffUK
In some cases we can only make educated guesses in other areas we can far more information via testing.
I'm regard to tyre temps for example we can use the lack of three point temp readout, no tyre pressure adjustment or readout, along with the way tyre temp changes to indicate that the model is most likely quite basic in that regard.
The switch in tyre behaviour at around 18mph from standing starts is also an indicator of this, it's an evolution of what was found in past GT titles, but still fairly basic.
It's still early days and over time more will come out, such as for past titles the use of a single co-eff value. Or when extensive testing brought out the fact that the same compound gave the same lat g value around a know radius for every car, and that different compounds simply applied a multiplier to that value.
Honestly, never underestimate the ability of this community to get quite deep into the model of GT, regardless of how much we may get locked out.
Now it's much easier with PC developed titles and/or those that provide UDP output, as the data is handed to us on a plate.
It also how it's known that while still of course not perfect, these titles are modelling more, and in more detail (and we still find issues with them, such as PCs camber issues).
None of this makes any off these titles 'better' than others , but it does make some more accurate than others.
Done if us just live digging into this stuff, and I can assure you, we take the accuracy of the work very seriously, and are quite happy to say what is speculative and what is not.
Last edited:
- 5,161
- Massachusetts
I was getting confused from the back and forth. Has it been confirmed that Gran Turismo Sport calculates tire flex? Or is it just a visual effect?
Unless a big change was made and I some how missed it, which I doubt, Assetto Corsa currently calculates tire deformation - flex and expansion and thermal blistering to boot - however, they chose not to make that visible. I would guess it's probably a system performance decision. So, I don't want responses like "I've seen the tires flex!!". That's not a definite indicator of the Tire Model calculating squat. Has it been confirmed?
Unless a big change was made and I some how missed it, which I doubt, Assetto Corsa currently calculates tire deformation - flex and expansion and thermal blistering to boot - however, they chose not to make that visible. I would guess it's probably a system performance decision. So, I don't want responses like "I've seen the tires flex!!". That's not a definite indicator of the Tire Model calculating squat. Has it been confirmed?
Scaff
Moderator
- 29,033
- He/Him
- ScaffUK
In part I agree with you, however with a few caviats.
The first being that its possible to model a variable, but to do so incorrectly. As such simply modeling something isn't enough.
The second is that feel's right depends entirely what that persons frame of reference is. The steering weight, even at minimum, in GTS before the patch was objectively not right. That didn't stop a percentage of people feeling it was great, as such 'doesn't feel right' is not something that I personally hold in high regard unless its further supported by some reference to reality.
- 4,819
- Bulgaria
Take a car and try this: put 10 or 0 degrees of camber and make 10 laps,if the grip after 10 laps is the same as in the first several laps then there is no simulation of inside and outside temperature.Because with 10 camber the inner section of the tyre must overheat and lose grip the more laps you do and with 0 camber the outside section of the tyre must overheat and lose grip the more laps you do.
- 28,471
- Windsor, Ontario, Canada
- Johnnypenso
That should also have dramatic consequences in terms of acceleration and braking. I'd be curious to see someone do some testing and find out what actually happens. Had a lot of physics testing going on in GT5 and GT6, curious to see how GTS stands up and what secrets it brings forth, if any.
- 8,698
I like the aero maps. I had thought they'd been doing something like this, modelling the cars and using CFD to generate lookups for use in-game. Validation could be an issue, but PD are not shy about approaching the right people.
The confirmation that PD use photogrammetry is nice, interesting to see the comparison of the accuracy with laser scanning approaches. Alsace is clearly generated from photographs with the track and scenery (trees, houses etc.) popped on top - almost course-maker-esque...
In terms of the physics, I like the detailed approach that delivers an enjoyable (if not wholly realistic) experience, but I feel that PD have dropped the ball slightly as regards input / output and player interaction at its core. The DS4 controls are awful, for instance. Not as bad as Assetto Corsa, but still.
The use of empirical real world data to develop the physics is sound, but I think the mathematical basis is a little too long in the tooth now. The suspension kinematics took another step forward, and damping is a little better, but the overall rigid body dynamics have the same old GT uncannyness. The tyre model is a weak point, the low speed one in particular, but it is optimised to work with the drivetrain model, which is also in need of an overhaul. It is nonetheless driveable and communicative, but frustratingly non-responsive over the limit.
I think all of that means that a big change is required with all the interconnected systems, and it's a shame they didn't do it for Sport, because it would be odd to introduce such a large change to the overall driving experience through an update.
The confirmation that PD use photogrammetry is nice, interesting to see the comparison of the accuracy with laser scanning approaches. Alsace is clearly generated from photographs with the track and scenery (trees, houses etc.) popped on top - almost course-maker-esque...
In terms of the physics, I like the detailed approach that delivers an enjoyable (if not wholly realistic) experience, but I feel that PD have dropped the ball slightly as regards input / output and player interaction at its core. The DS4 controls are awful, for instance. Not as bad as Assetto Corsa, but still.
The use of empirical real world data to develop the physics is sound, but I think the mathematical basis is a little too long in the tooth now. The suspension kinematics took another step forward, and damping is a little better, but the overall rigid body dynamics have the same old GT uncannyness. The tyre model is a weak point, the low speed one in particular, but it is optimised to work with the drivetrain model, which is also in need of an overhaul. It is nonetheless driveable and communicative, but frustratingly non-responsive over the limit.
I think all of that means that a big change is required with all the interconnected systems, and it's a shame they didn't do it for Sport, because it would be odd to introduce such a large change to the overall driving experience through an update.
- 982
I doubt they are using high fidelity simulations but its good they are using actual engineering principles.
Really depends on the case, what turbulence model has been used, meshing etc but for external aerodynamics on cars good accuracy is commonly found in the engineering sector using models like RANS which are not computationally expensive.
You pretty much described 90% of sim racing enthusiasts who claim game x is not realistic by how it feels to them and who probably have vey little knowledge on how physics simulation works.
We see tire flex. Some claim its not physically modelled. Right now that seems to just a claim.
- 28,471
- Windsor, Ontario, Canada
- Johnnypenso
You have no idea what level of knowledge 90% of sim racers have. And there are plenty of people who have real life experience who also describe the physics simulation in GTS as flawed or incredibly simplistic in several areas.
Visual tire flex is not proof of it being modeled any more than the absence of visual tire flex is proof of it not being modeled. The absence of evidence isn't evidence.
- 338
- United States
On the contrary, having readouts when the car does not do what the readout informs also indicates something very basic and flawed. (a complex readout but a flawed basic model)
Scaff can you think of any recent simulators that has this?
speaking of tires i think all should be with same psi when racing competitively but for everyday gaming its fine.
Gran Turismo Sport is a very suitable game to participate in Esports, why because this is "Sport" Tom Brady is into "Sports". It was an accusation that he manipulated the air pressure in the ball to gain an advantage in his sport. All balls must operate at the exact same psi in the NFL. Racing others with different psi in the tires is all fun and games to find who has the magic numbers discovering who's car gets faster around the track. However, you should not be riding on those magic numbers with psi when you are ready to turn that game into a "Sport", just ask Tom Brady.
And those with real life experience has no idea how simplistic or complex the model is until they become a simulation developer themselves.
Last edited:
Scaff
Moderator
- 29,033
- He/Him
- ScaffUK
I'm sure you are going to tell us all, I do hope that it includes some proof.
The only problem with your utterly false analogy is that in every form of motorsport on the planet you are permitted to adjust tyre pressure, as oddly enough motorsport isn't football!
Utter and complete nonsense.
Tell me why I need to be a developer to state that GT's clutch model is utterly simple and innacurate?
- 338
- United States
What next Scaff?!? The game is utterly basic and flawed because you can't open the hood and see whats inside the engine?!?
It is obvious that you are just poking holes in every thing Gran Turismo Sport lectures with a list of weightless meaningless examples.
::sigh:: I guess my question was difficult to comprehend.
Well, look at it this way if someone can tell you tire flex is implemented in another sim but not show it to you visually. thats an indication of something flawed with realism. According to Scaff's beliefs it should be also considered very basic modeling.
Last edited by a moderator:
Scaff
Moderator
- 29,033
- He/Him
- ScaffUK
If you're not able to answer the question, then a strawman argument isn't a suitable alternative.
No I'm asking perfectly reasonable questions about the claims you're making.
Its a compromise that developer has chosen to make with the approach they have taken, just as PD have taken an approach that has compromises.
Oh and if you want to know my thoughts on something, then ask. Don't assume without at the very least doing me the courtesy of tagging me in the post or quoting me saying it.
However despite the discourtesy I'm going to be polite enough to answer you.
It a basic approach to visual modelling (very even if you like), as they have clearly chosen a route that compromises the visual modelling over the physics modeling. You do however seem to want to try and conflate the two, when they are clearly different approaches.
PD chose a bias towards visual fidelity that has resulted in a deficit in the physics modelling, other devs have chosen a different approach.
Seriously have no idea why pointing that out is 'poking holes in everything Gran Turismo Sport', when its actually a rather basic observation, supported by a quite large degree of evidence of differing levels.
However if you disagree the please feel free to support a counter argument with something other than strawman arguments, thinly veiled insults and distractions. You know like some actual examples that can be tested against reality.
Oh, and don't double post.
- 4,819
- Bulgaria
Whoever you ask who uses a shifter in GT will tell you that clutch model is innacurate.
Of course not!
Assetto Corsa says hello.
Scaff
Moderator
- 29,033
- He/Him
- ScaffUK
An argument to authority.
Its seems to be the day for logical fallacies.
Let me ask the question again, maybe you will answer.
Why should I need to defer to anyone to observe and demonstrate that the clutch in GTS isn't modeled accurately in regard to reality?
Both are just claims, that's the point!
You seem happy to accept that's enough to show its also modeled in the physics engine, and that its modeled accurately.
Others are not.
Now in this regard both GT and FM share a common problem that titles with origins on PC don't have, and that's the ability to extract data either from the code itself or via UDP sources to help validate or dismiss the point.
Neither GT or FM allow you to do that easily (past title have and in GT's case it shows that was not the case at all), however in the case of both PC, PC2 and AC it is included in the physics engine and tyre model. The remaining question then is how accurately is it modeled.
Last edited:
- 28,471
- Windsor, Ontario, Canada
- Johnnypenso
It's an indication of limited resources and a developer choosing to spend those limited resources on modelling what will appeal to their customers most. In one game it's the physics, in another it might be the visual. Different allocations of limited resources. Neither is an indication of flawed realism unless it purports itself to be something it's not.
- 789
- QLD
Scaff, don't get up me for the comparisons that will be in this post when you started it above.
My other argument is there are sims that do have adjustable tyre pressures and tyre temperature readouts but I think this stuff doesn't behave in a realistic way or the tyres themselves don't behave in a realistic way. And if anything I think the implementation in many sims is unrealistic and acts like a game. PC2 still has this stuff implemented in the same way they did back in the SimBin days and I personally don't think the way the game rewards balancing temps or the way you achieve this balance has ever been realistic. On the other hand GT Sport while having it's faults, for me so far responds to tunes in a very realistic way and I'd rather not have features at all then have features that are not implemented in a realistic way. I just wish GT Sport was like this in more areas because it does also have areas where they have included things I find unrealistic.
Actually this argument is similar to the old brake upgrade argument in older GT's. And my rebuttal is similar. Back then I pointed out that once you have brakes that can lock up at any speed and don't overheat there is no need for brake upgrades. In GTS we don't have dynamic weather which means there is an optimal tyre pressure that would apply to everyone. Now I understand this isn't as absolute as the brake argument because it is possible to combine an aggressive camber with a different tyre pressure to achieve a similar result but including it for that would be a waste of resources especially when you consider Ian Bell claims they PC2 uses 2 cores just for the tyre model. There is little return for going to that length in a Sim that doesn't have dynamic weather. Instead they can create a simpler more efficient for our needs but still realistic tyre model because we don't really need tyre pressure adjustment in this sim considering there is no dynamic weather.
My other argument is there are sims that do have adjustable tyre pressures and tyre temperature readouts but I think this stuff doesn't behave in a realistic way or the tyres themselves don't behave in a realistic way. And if anything I think the implementation in many sims is unrealistic and acts like a game. PC2 still has this stuff implemented in the same way they did back in the SimBin days and I personally don't think the way the game rewards balancing temps or the way you achieve this balance has ever been realistic. On the other hand GT Sport while having it's faults, for me so far responds to tunes in a very realistic way and I'd rather not have features at all then have features that are not implemented in a realistic way. I just wish GT Sport was like this in more areas because it does also have areas where they have included things I find unrealistic.
- 338
- United States
I don't see GTS to purport being something it's not?
GTS is not designed to be like another Assetto or PC. those games were designed to be similar to the old GT but attention is solely put in car physics control to end user. If you are really going to compare and contrast the games you should also consider the things GTS does that they don't do.
Forza is on a different level also. just because PC/ Asseto Corsa has more details in extracting data does not make Forza's system flawed. Ultimately it is how the physics transmit to the end user. The end user is not only you nor a racing professional but everyone who uses the ps4 and xbox consoles. if it can function well in the hands ears and eyes of players the data is not really needed. But you do need some type of data to help you through PC and Assetto. For an instance, If I had to deflate tire pressure in pc1 to increase heat rate then in PC2 I need to inflate tire pressure to increase heat rate; as an end user I will need some type of data to help guide me through that game for optimal tire temps otherwise it would only appear flawed by design.
Last edited:
- 27,385
- Toronto
- NewAesthetic
- SlipZtrEm
A clutch is weightless and meaningless? Same with adjustable tire pressure?
It's really odd that you've appointed yourself the arbiter of which criticisms are valid and which aren't. You just finished stating this:
Yet you have no qualms passing judgement on games that don't have the words "Gran" or "Turismo" in their names.
And, going by your own statement, how much weight should be placed in the feedback of the real-world racing drivers involved, at various levels, with the recent sims? The Hamilton brothers, Matias Henkola, Josef Newgarden... the GT Academy grads? They aren't sim developers, after all.
...
It should be noted that anybody being critical of the physics of GT Sport — or indeed any other games — is not the same as them telling others they can't enjoy the game. Because that would be preposterous. But that's just the nature of the genre; a sim is aiming to replicate something found in the real world, where everything is massively complex, and perfection simply isn't possible. It's not like an action/adventure game, which can construct its own rules. Enjoyment and realism are not (necessarily) related.
So why people take such personal offence from others' differing opinions is beyond me. I enjoy GT Sport, FM7, and PCARS2 (and all my other, less-new racing games), but I can find fault in all of them. And it really doesn't bother me.
Says who?
Or, as a more relevant example, GT6 tuning showing that a front right height that's higher than the rear is faster.
- 15,119
- Melbourne
- ScottPuss20
- CheetahsMeow
But all of this is really complex stuff, which could only be achieved with a significant graphical downgrade. I don't think PD want that.
- 28,471
- Windsor, Ontario, Canada
- Johnnypenso
I didn't say anything either way.
Scaff
Moderator
- 29,033
- He/Him
- ScaffUK
In regard to brake upgrades I agree, and first posted as such back in 2005 in regard to GT4.
However its not a similar argument.....
Tyre pressure isn't a tool simply used in wet or variable weather conditions.
Alongside camber it's one of the two basic tool used to ensure that the tyres get up to temperature for a given track. As no two tracks are identical in that regard, nor the loading on any tyre on a car for that track. As such even if the track temp is identical, the optimum cold pressure for a car is not going to be the same for a short oval as it is for the 'ring.
As such I can't agree that its a similar argument to brake upgrades at all.
Is it 100% accurate and realistic in PC2? Nope (nor in AC, iRacing, GTR2, etc.).
Should that give GTS a free pass given that it has objectively a more simple tyre model, that does model fewer variables? Well that depends how much you care about physics accuracy.
If you don't rate physics fidelity above other areas then feel free to do so; however if you value physics fidelity higher than other areas (as I personally do) then no it doesn't.
Aside from a certain marketing strap-line, I havn't claimed it does. Some GT fan however are a very different story.
I have, it was quite literally one of the key points of my last post!
That again depends entirely on what the end goal is.
If its something that people like the feel of and makes they feel like a 'driving god' (to steal a Top Gear line) then no it doesn't. If however the end goal is the highest level of physics fidelity, then that's quite a different story.
So on that basis are you suggesting that GTS removes all elements of car set-up?
As the exact same argument could be used for any area of set-up on any title.
- 789
- QLD
In relation to the weather on that day. What's the aim? The aim is to ensure the tyres are at the optimal operating range.
So in the same way that once you have optimal brakes you no longer need options, if you always have the optimal pressure given to you, you no longer need options. This can be of a massive benefit to the end user. PC2 uses 2 cores just for their tyre simulation, that's 2 whole cores that can't be used for other things. If you simplify the tyre simulation by instead of simulating the air inside the tyre and the effect on the carcass you start with an optimal result of this calculation that is a lot of processing power you free up. The point is that is a lot of processing power for very little return to the end user. Instead save that valuable power and give the user the result they would want to start with and now that power can be used for something that gives the user a better experience.
By the way, you forgot Toe.
This is key. Accurately modelling the result would be more accurate then modelling more inaccurately.
Opinions will differ on which has the better result.
Scaff
Moderator
- 29,033
- He/Him
- ScaffUK
Which would be fine if it was the only factor involved in ensuring tyres are at the optimal operating range.
A naturally aggressive driver (driver A) is going to get more heat into the tyres than a naturally less aggressive driver (driver B - Jenson Button is a good example of the later). This can result in either Driver A getting too much heat and the tyres falling off, or Driver B not getting enough heat and the tyres never getting up to temp, or both if the optimum is in between.
In these cases the ability to adjust static pressure to match the drivers particular style is one of the key tools of set-up, without it the window in which the driver has to operate has to be the optimal one for the that pre-set pressure. Which would limit the percentage of drivers able to make the most of the car and on balance make things more difficult that reality with no means of resolving it.q
Replace optimal pressure with any other set-up value and that argument ends up exactly the same. Camber, spring rate, damper, etc.
Give the physics engine the optimal value for these and you no longer need options for them either.
So why one and not the others?
You see to simply to the degree you are suggesting would indicate exactly what I have argued this entire time (actually it goes past it and out the other end). To have a single value or limited range of preset pressure vales used in the way suggested would undermine the entire tyre model to such a degree that any other part of the physics engine becomes limited by the final and key part. The vehicles only area of direct interaction with the track.
That's an assumption, and an incorrect one at that.
The principal purpose of adjusting toe is the balance between a vehicles desire to turn-in versus its straight line stability and ability to accellerate and brake optimally. It can affect temps, but very much as a second order effect.
And the evidence that GTS manages this is?
A good degree exists to indicate the opposite, I'm open to discussing how GTS manages to 'can' key elements of the physics engine and achieve a robustly accurate end result.
I do want to however check that the crux of this argument is that GTS can be more accurate by not bothering to accurately model a load of things and just using pre-set assumptions instead?
Last edited:
- 789
- QLD
The key difference is nothing else takes 2 whole cores to simulate so no the argument does not end up the same. You really have a habit of picking parts of an argument out of context and ignoring everything else said. The result is your replies are irrelevant because you are never really understanding what is said to you.
This is a rubbish statement because you mentioned camber. No one uses camber to primarily adjust tyre temps and toe can have a bigger impact on tyre temps than camber. The issue with camber is it changes the contact patch reducing or increasing grip and as a result effecting where the tyres are heated. An interesting thing is despite tyre temps not being user adjustable the optimal camber window in this game is wide. Think about why that might be. Better yet load up a sim you trust and test the camber performance window with a fixed temp and then see how much that window widens as you compensate with temperature adjustments.
I believe you have already answered this. If other sims are failing to provide accurate simulation my argument is it is better to not simulate those things. Go watch the Spottheozzie stream with Ian Bell if you want admission from a developer that in his opinion there are sims with simple simulation that are more realistic than the sims hardcore sim racers say are best. There is no way we can agree at the end of the day because we all have formed our own opinions. But I think the key is sim racers often claim certain sims are realistic and their primary background originates in sim racing and for many that goes back to the belief that the ISI motors are the epitome of realistic physics. On the other hand we have real life racers constantly saying the top sims are nothing like the real thing, that they are all too hard, and their foundation for comparison is real life not other sims. I agree with those. I have not driven a realistic sim but there are certain sims that are more authentic IMO and GT Sport is near the top of the pile in certain vehicles, not the VGT cars, not the rally cars. But many of the other cars I have driven (despite the grip issues) drive, behave and respond like real life cars, the driving experience is realistic. For me a key test is driving a marathon (a day of testing straight) in a game and then jumping into a real life car and going for a drive. If I go from iRacing driving a real car is bizarre, I have often joked that my real car is not a very realistic sim after a marathon in iRacing. GT Sport is the least odd feeling after this test for me. Interestingly despite my opinion that GT Sport is more like Assetto Corsa than other sims, PCARS is the next best performing sim for me in this test, then PCARS 2. And don't take that the wrong way, there are aspects of PC2 that I feel are great steps forward but overall in the cars I actually enjoy the most in PCARS, I think PCARS drives better. Of course the fact I own PCARS on PC and PCARS2 on PS4 might be a contributing factor as it seems PC2 is far better on PC than PS4.
I've already covered the first part of my reply to this earlier and that's why this part is at the end but there is a second part of it.
I have a theory about the choices in this game compared to other games like iRacing. iRacing has moved on from catering for just one type of person, it recognises that not everyone is a driver so it is no longer a sim for just the driver. iRacing is a sim for broadcasters, team managers, race engineers, artists and so on. I think GT Sport however is primarily trying to be a sim that focuses on driver development. Trying to take people who might have an interest or a dream of racing and turn them into real racers. Following that line of thought I think various aspects of the game have been designed in a way to help develop certain skills a race driver might need. A race driver doesn't need to be able to work on the car, he doesn't need to understand how every little thing effects a car but he does need to understand certain settings and their effect so he can communicate effectively to his team. I think for the most part the tuning in GT Sport can be effective in helping driver development in these areas. The understanding of tyre temps is not one of those areas because the driver for the most part would be unaware of what is going on with his tyres other than feel inside the car. His team would take care of analysis, changes and instructions about his driving. This is another key to this argument, the team dictates to the driver how he is to drive. Are you aware of how long GT3 tyres will last when driven aggressively? GT3 drivers have said they can destroy a fresh set of tyres in just a few laps but in races they sometimes double stint tyres.
Continuing with the driver development theory perhaps GT Sport is trying to develop drivers who drive a certain way. Kids have interrupted I'm done, not revising, not sure what I wrote. She'll be right.
