GT6 News Discussion

[Griffith500]

While I can certainly appreciate what Kaz is getting at with the talk of the "triangle", to simultaneously brush off the importance of getting detailed tire information in the first half of the paragraph, and then compare what they're doing to what RBR does seems... odd.

"Just" measuring tires isn't the point the critics are getting at with GT5's limited tire physics; it's getting them to react like real tires, to not have Minis matching the g's of Vettes on the same grade of rubber. That all said, the GT6 pods at JFF certainly seemed different from the Academy demo, so Yokohama might already be incorporating their data 👍

He's saying it's a black art; that they're (now) doing the best they can, assuming that the way RBR is doing it is also "best".

Building a computational model of something that even the experts don't fully understand can't be easy. He's saying that what will define whether it's a good model or not (as far as we, the players can perceive) are the choices they make in filling in those gaps of understanding.

For RBR, it's more about the simulated times matching reality and the behaviour up to the limit being correct (which is easier to measure); for GT, the times aren't nearly so critical, but feel is and so is a wide range of behaviours, especially at and beyond the limit. Presumably, Kaz / Nissan have been doing things the RBR way for their racing endeavours, so I'm sure Kaz appreciates the difference.

I don't understand how he's brushing anything off, he's just echoing what other games designers have said about tyre data - it's the player base that places the emphasis on measurements, when it's understanding that's lacking (what particular compound effects do those measurements actually show?)

Delineation of these compound effects against the "inputs" is very difficult to achieve in tyre measurement - see Dave Kaemmer's excellent post about it.
Computer models could potentially help understanding, but not without the ability to test the results of those models.

 

[Imari]

Delineation of these compound effects against the "inputs" is very difficult to achieve in tyre measurement - see Dave Kaemmer's excellent post about it.
Computer models could potentially help understanding, but not without the ability to test the results of those models.

True, but Dave also points out that empirical models can be generated, it's simply that the data ranges that sim models would most require are those least likely to be tested by tyre manufacturers and race teams.

Theoretical models are great, but at the very least you need some amount of real world data to test your theoretical model against.

Strictly, Kaz isn't right. There's not a limit to what you could get testing tyres. You could test tyres endlessly under many different scenarios and generate a highly complex and accurate empirical model. The problem, as Dave Kaemmer points out, is it's bloody expensive ripping up that many tyres.

So really, there's a limit to the budget (which is no news to anyone) and then after that they're forced to make do with educated guesses and trying to apply data from other fields like materials science and surface chemistry/physics. Which is the same as everyone else. But you can't reply to an interview question with "we just couldn't afford to buy a thousand tyres to destroy for the data we needed".

Most models can be arrived at by pure brute force if necessary, at least in an empirical sense. You may never get the theoretical knowledge, but you'll have a predictive equation that works, if that's what matters. The reason it so rarely happens is money. And time I suppose, but in business the two are largely interchangeable.

And that scientists and engineers are usually naturally inquisitive and start staying up nights trying to figure it out when they don't understand *why* something works, but that's a different thing.

 

[dr_slump]

In this pic, on Kaz's t-shirt, you see all the known companies involved with GT6, but we haven't heard anything about Recaro yet....?

So we'll probably get some fancy seats... ok...

 

[LMSCorvetteGT2]

So we'll probably get some fancy seats... ok...

Probably get to collect seats now like we do helmets

 

[TokoTurismo]

Probably get to collect seats now like we do helmets

What's funny is that what will be the point, not like we can see all cars interiors except for the premiums anyways.

 

[Samus]

Aren't they just real life partners of the Schulze team, that some of also happen to be GT partners?

 

[sparkytooth]

What's funny is that what will be the point, not like we can see all cars interiors except for the premiums anyways.

Exactly

 

[Ranman20]

What ever happened to the supposed scanned Alfa Romeo 155 DTM and Ferrari P4/5 competionize article?

 

[LMSCorvetteGT2]

Exactly

Someone got the joke 👍

 

[ZoidFile]

What ever happened to the supposed scanned Alfa Romeo 155 DTM and Ferrari P4/5 competionize article?

Nobody knows and we are forbidden to talk about it here.

 

[drummermaniac]

Fellow insiders , spill the beans

 

[Dmarc_Atl]

We need that space dude who leaked news about gt5.. PD probably had him silenced.

 

[Zlork]

When Jordan will post interview of JFF ?

 

[Griffith500]

True, but Dave also points out that empirical models can be generated, it's simply that the data ranges that sim models would most require are those least likely to be tested by tyre manufacturers and race teams.

Theoretical models are great, but at the very least you need some amount of real world data to test your theoretical model against.

Strictly, Kaz isn't right. There's not a limit to what you could get testing tyres. You could test tyres endlessly under many different scenarios and generate a highly complex and accurate empirical model. The problem, as Dave Kaemmer points out, is it's bloody expensive ripping up that many tyres.

So really, there's a limit to the budget (which is no news to anyone) and then after that they're forced to make do with educated guesses and trying to apply data from other fields like materials science and surface chemistry/physics. Which is the same as everyone else. But you can't reply to an interview question with "we just couldn't afford to buy a thousand tyres to destroy for the data we needed".

Most models can be arrived at by pure brute force if necessary, at least in an empirical sense. You may never get the theoretical knowledge, but you'll have a predictive equation that works, if that's what matters. The reason it so rarely happens is money. And time I suppose, but in business the two are largely interchangeable.

And that scientists and engineers are usually naturally inquisitive and start staying up nights trying to figure it out when they don't understand *why* something works, but that's a different thing.

Empiricism is ineffective for something as broad and varied as GT, though - it's not that it's too expensive, it's that it's just not tractable in any sensible time scale. Theoretical models aren't just "great", they are what allow real progress to be made. Imagine if GPS systems ran off an empirical model of position deltas (data) rather than the "actual" underlying mathematical model that an understanding of relativistic physics allows.

Another issue isn't just that data is scarce, it's also, as I said (and did Dave), indeterminate - you don't know what compound effects are causing changes in your measurements - delineating that is difficult, and would best be tackled (as it has in many different system in the past) with increasing complexity of physical models - i.e. hypothesise, test.

But, again, Kaz didn't say data is pointless, just that it's currently indeterminate given the lack of understanding (i.e. a model) of the underlying principles. And I did say any model would need testing - that implies (I did think about being explicit, but that requires more words, and my posts are nothing if not wordy) measurements of real systems to compare against it.

A better word is validation, and I had to do it all the time in a previous job for models that are based on real, "established" and understood theoretical physics for relatively simple systems (if exposed to "complicated" inputs). Namely because numerical models are fun systems in and of themselves, aside from the systems of equations they're built on - yet another layer of complication to get around for "just a game", theoretical model or otherwise (see divide by zero issue with slip ratio).

There will always be some level of empiricism for real-time entertainment, but that doesn't mean that's all that is involved - there's lots of room for more direct models to be used during production, or "offline" in the game itself. Real-time computer graphics, for instance, probably best characterises this approach.

 

[sems4arsenal]

When Jordan will post interview of JFF ?

He said there was nothing new in it anyway but probably by the end of the week.

 

[Foxiol]

What if they showed nothing just because they are working hard to implement the new sounds in the game and then surprise us all?

Yeah I know, I´m being way too much optimist but still it would be cool. But at the same time I won´t be expecting this at all any ways.

 

[LMSCorvetteGT2]

There is still Australia but I guess we'll forget about that.

 

[Samus]

What if they showed nothing just because they are working hard to implement the new sounds in the game and then surprise us all?

Yeah I know, I´m being way too much optimist but still it would be cool. But at the same time I won´t be expecting this at all any ways.

There are plenty of things for them to talk about, not just sounds.

 

[GTPorsche]

There is still Australia but I guess we'll forget about that.

Rumor is the Australia stuff will be announced sometime around the Bathurst 1000 weekend.

 

[Foxiol]

There are plenty of things for them to talk about, not just sounds.

Well but as something that can be kind of "shocking" to show...that will be new sounds for sure.

Livery editor, customization and such things are going to be there somehow...sounds are not expected any more since Kaz said "we are working in sounds but I don´t think they will make it for the release of the game...etc".

That´s why I put the "sounds" example.

Also you know well that I am one of those that want realistic sounds for once. (we all know each other here)

 

[LMSCorvetteGT2]

Rumor is the Australia stuff will be announced sometime around the Bathurst 1000 weekend.

Yeah I know but there is a good group acting as if this is it and then we wait for the game. Though I do criticize PD, they did say they'll say something and show stuff then and though we know what most of it is (or somewhat) let's wait and then pass judgement.

 

[Samus]

I don't need shock and awe, I just want to know what makes up the gameplay. So far we know.....nothing. How is A-Spec structured? Are the AI better? What about other events and gameplay features? Will there be licences? Championships? New Special events? They've told us a better course creator is returning but how does it work?

 

[CorvetteConquer]

PD- "They'll buy the game anyways...I mean, come on! It's Gran Turismo!"

If there isn't an embargo lift tomorrow, then this is a very underwhelming ride to GT6.

 

[fortbo]

PD- "They'll buy the game anyways...I mean, come on! It's Gran Turismo!"

If there isn't an embargo lift tomorrow, then this is a very underwhelming ride to GT6.

What embargo? Why is everyone talking about it,I need a source.

 

[ildd]

Apparently my definition of embargo and Gran Turismos definition is different...

 

[LMSCorvetteGT2]

Apparently my definition of embargo and Gran Turismos definition is different...

Yeah I agree, not sure where PD got there definition.

 

[pikes_peak]

What embargo? Why is everyone talking about it,I need a source.

This one?

For many more pictures embargoes, wait a few days.

 

[Samus]

So how has that translated to Monday exactly?

 

[amar212]

I don't need shock and awe, I just want to know what makes up the gameplay. So far we know.....nothing. How is A-Spec structured? Are the AI better? What about other events and gameplay features? Will there be licences? Championships? New Special events? They've told us a better course creator is returning but how does it work?

Before we get the game in our hands.. I doubt many of these questions will get an answer.

 

[jules283]

I remember amar-212 mentioning embargo.