Gran Turismo 7 Physics

[br1x]

100%

Polyphony should just license the physics and AI from Kunos, and then spend all their time doing the things that they do best - graphics and user friendly controls. It would be by far and away the best Gran Turismo game ever.

Don't forget ACC has a total of 2 car classes. It's rather easy to flesh out physics and AI that work for modern GT cars but how does it work when transfered to modern LMP cars, Group C, 50hp road cars?

Add to that that ACC is far from the holy grail in physics you guys seem to be making it out to be, it gets a lot of criticism for its tyre model, especially when driving "on the limit".
There are other sims with clearly superior physics in different areas, but they usually aren't as polished since ACC focuses almost exclusively on GT3.

 

[ScottPuss20]

It's crazy how the physics still have legacy quirks from 20+ years ago.

It's so cringy watching the nose of the cars pivot back and forth through high speed corners. The laughable collision physics. PD can afford to be lazy about such things because they have no legitimate competition.

GT with ACC physics would be a dream come true

Assetto Corsa is as close as you're going to get to a "realistic" GT game, especially with all the mods that are available.

 

[Imari]

Don't forget ACC has a total of 2 car classes. It's rather easy to flesh out physics and AI that work for modern GT cars but how does it work when transfered to modern LMP cars, Group C, 50hp road cars?

Add to that that ACC is far from the holy grail in physics you guys seem to be making it out to be, it gets a lot of criticism for its tyre model, especially when driving "on the limit".
There are other sims with clearly superior physics in different areas, but they usually aren't as polished since ACC focuses almost exclusively on GT3.

Assetto Corsa seems to work pretty well with a wide range of cars. You should try it, it's very cheap for what you get. Same developers, similar physics if an earlier iteration, which is why I said "licence from Kunos" not "licence the ACC system". The physics system as a whole seems to work fairly well with cars that are made correctly, and I include that as a caveat only because we've seen what bad mods look like. Even though the fundamental physics are good, it's not that hard to build a car badly and have it drive like a clown car.

The tyre model isn't the best in the world and we've all known this since NetkarPro, but I think it does a decent job of finding the middle ground between accuracy and computational complexity. I think that it would suit Gran Turismo and is simplistic enough to allow Polyphony to build the wide range of tyres that they would probably want relatively easily.

I'm not saying it's the holy grail, and I don't see where I've implied that. I'm saying that it's an easily available physics and AI system that is superior to Gran Turismo in every way that I can think of. I think it's the best fit I can see for a Gran Turismo game, given what we've seen modders achieve with it. If Polyphony wanted to use another system I wouldn't be averse to that either, there's any number of games out there that do physics and AI better than GT. Ultimately price would also be a factor, and realistically at this point I'd be happy enough to see Polyphony focus on their strengths and farm out the stuff that they're not as good at to other developers with demonstrated skills.

What physics system would you prefer they used?

 

[br1x]

Assetto Corsa seems to work pretty well with a wide range of cars. You should try it, it's very cheap for what you get. Same developers, similar physics if an earlier iteration, which is why I said "licence from Kunos" not "licence the ACC system". The physics system as a whole seems to work fairly well with cars that are made correctly, and I include that as a caveat only because we've seen what bad mods look like. Even though the fundamental physics are good, it's not that hard to build a car badly and have it drive like a clown car.

The tyre model isn't the best in the world and we've all known this since NetkarPro, but I think it does a decent job of finding the middle ground between accuracy and computational complexity. I think that it would suit Gran Turismo and is simplistic enough to allow Polyphony to build the wide range of tyres that they would probably want relatively easily.

I'm not saying it's the holy grail, and I don't see where I've implied that. I'm saying that it's an easily available physics and AI system that is superior to Gran Turismo in every way that I can think of. I think it's the best fit I can see for a Gran Turismo game, given what we've seen modders achieve with it. If Polyphony wanted to use another system I wouldn't be averse to that either, there's any number of games out there that do physics and AI better than GT. Ultimately price would also be a factor, and realistically at this point I'd be happy enough to see Polyphony focus on their strengths and farm out the stuff that they're not as good at to other developers with demonstrated skills.

What physics system would you prefer they used?

Oh trust me, I played Assetto Corsa for quite a few hours and it is a perfect example for physics system that doesn't work well across the different types of cars, mods or Kunos cars.

The easiest examples are the curbs of death that just want to make you spin for some reason across a lot of tracks with a lot of faster race cars, Porsche GT1 or RSS GT mod on Suzuka S-Curves or Spa for example.

I honestly like AMS2 physics the most right now, especially how the tyres feel after the 1.3 update, but here is the problem I mentioned earlier about being not polished. Some cars feel fantastic, some are absolutely horrible, GT3s are only OK for example which drives a lot of people away already being the most popular class for most.
The AI is also ranging from good to horrendous, sometimes LMPs not even being able to overtake GT3s in a multiclass race.

So its not just a game you can pick up and have fun with, that's the big strength of Gran Turismo.
So PD should IMO just expand on their physics but keep the base they have. Adding tyre pressures and temperatures should be the next step, especially with that Michelin partnership.

 

[Magog]

Haha no. Why do you think ACC runs at 30 FPS on PS4?

Irrelevant. The CPU is extremely weak in the PS4 compared to the GPU but Driveclub also runs at 30fps and is more focused on graphics than physics.

 

[05XR8]

Wonder with the World Tour participnts input, will there be changes.

I mean, the online racing is like real world race car testing. That knowledge/data feeds down to a company’s road cars. Maybe the data they gather from the FIA, trickles down to non-Sport Mode gameplay.

I understand latency are a big part of Sport Mode interaction. However, maybe car contact, contact with curbs/kerbs is dampened due to WT driver feedback. I don’t know.

 

[Magog]

Wonder with the World Tour participnts input, will there be changes.

I mean, the online racing is like real world race car testing. That knowledge/data feeds down to a company’s road cars. Maybe the data they gather from the FIA, trickles down to non-Sport Mode gameplay.

I understand latency are a big part of Sport Mode interaction. However, maybe car contact, contact with curbs/kerbs is dampened due to WT driver feedback. I don’t know.

Kaz has specifically stated that they are taking the pro players input to improve the physics in GT7.

 

[Dave A]

The CPU/GPU graphics v physics issue is pretty complex, a CPU is important to graphics, it just doesn't usually handle everything. You can bottleneck your CPU with graphical load, this is ypically more evident at 1080p or lower resolutions running at high FPS than 4k as at higher resolutions you're more likely to bottleneck the GPU first. But also pairing a high end GPU with a low end CPU could bottlneck the CPU too.

There are tasks a GPU cannot do that the CPU has to handle and the two talk to eachother. This takes time and adds load to the CPU.

How much a load a game places on one or the other depends largeley on the architecture of the game. A good developer will find where they have headroom and make use of that be it on the CPU or the GPU. When you are pushing a system there's always a tradeoff though. If you go for a higher refresh rate in your physics engine you will trade with something and that could be the resolution, FPS, HDR or other effects.

It's a complex issue and a lot comes downs to how the developers have designed thier code against the system specs and made use of what they have availalbe. What we must remember though with GT7 is it's cross generation and cross play, so unlike the graphics, the physics will likely have to be the same on the PS4 version as they are on the PS5.

 

[ThePotatoKing]

Assetto Corsa seems to work pretty well with a wide range of cars. You should try it, it's very cheap for what you get. Same developers, similar physics if an earlier iteration, which is why I said "licence from Kunos" not "licence the ACC system". The physics system as a whole seems to work fairly well with cars that are made correctly, and I include that as a caveat only because we've seen what bad mods look like. Even though the fundamental physics are good, it's not that hard to build a car badly and have it drive like a clown car.

The tyre model isn't the best in the world and we've all known this since NetkarPro, but I think it does a decent job of finding the middle ground between accuracy and computational complexity. I think that it would suit Gran Turismo and is simplistic enough to allow Polyphony to build the wide range of tyres that they would probably want relatively easily.

I'm not saying it's the holy grail, and I don't see where I've implied that. I'm saying that it's an easily available physics and AI system that is superior to Gran Turismo in every way that I can think of. I think it's the best fit I can see for a Gran Turismo game, given what we've seen modders achieve with it. If Polyphony wanted to use another system I wouldn't be averse to that either, there's any number of games out there that do physics and AI better than GT. Ultimately price would also be a factor, and realistically at this point I'd be happy enough to see Polyphony focus on their strengths and farm out the stuff that they're not as good at to other developers with demonstrated skills.

What physics system would you prefer they used?

As great as AC's physics are, i don't think they are a good choice for the demographic Gran Turismo aims for, which is more or less casual gamers.

Yes they are realistic, but they are far from beginner friendly. To even get close to being decently fast, it requires a great amount of track time, which translates to crashes and spuns, which will discourage your player to keep going and instead he will probably going to pick up something easier.

 

[Magog]

As great as AC's physics are, i don't think they are a good choice for the demographic Gran Turismo aims for, which is more or less casual gamers.

Yes they are realistic, but they are far from beginner friendly. To even get close to being decently fast, it requires a great amount of track time, which translates to crashes and spuns, which will discourage your player to keep going and instead he will probably going to pick up something easier.

I agree. I think they are solidly aiming for that sweet spot where the car physics feel real enough to immerse you but not so real they are painful to all but the most finicky purists. Considering how successful the series has been I think they've just about nailed it. They seem to be very proud of the atmospheric/weather simulation so it will be very interesting to see just how advanced it is.

 

[ImaRobot]

Yes they are realistic, but they are far from beginner friendly. To even get close to being decently fast, it requires a great amount of track time, which translates to crashes and spuns, which will discourage your player to keep going and instead he will probably going to pick up something easier.

Is that down to the physics themselves though, or the input peripherals and optimization for it? Games on the higher spectrum in simulation also have tendency to not put too much emphasis on optimization for pad users, for instance. If you're talking about a wheel though, I wouldn't be able to comment on that.

 

[br1x]

Is that down to the physics themselves though, or the input peripherals and optimization for it? Games on the higher spectrum in simulation also have tendency to not put too much emphasis on optimization for pad users, for instance. If you're talking about a wheel though, I wouldn't be able to comment on that.

I think it goes hand in hand. The more realistic the physics are, the harder it gets to make it well optimized for a gamepad.

 

[ScottPuss20]

I think it goes hand in hand. The more realistic the physics are, the harder it gets to make it well optimized for a gamepad.

Funny you say that because the games that are "well optimised" for a controller tend to feel like garbage - too heavy and indirect.

 

[ThePotatoKing]

Is that down to the physics themselves though, or the input peripherals and optimization for it? Games on the higher spectrum in simulation also have tendency to not put too much emphasis on optimization for pad users, for instance. If you're talking about a wheel though, I wouldn't be able to comment on that.

I would say it goes both ways. Higher-level simulators, such as Asseto Corsa and Raceroom for example, don't have the greatest pad optimization, according to my experience at least. However i'm willing to bet that the lack of proper pad optimization really comes down to the small size of the dev teams behind those games (you can't really compare Kunos to PD in that regard) so it's not that they can't, it's that they don't have the manpower and resources to pull that off. In short, i think that there isn't really a high-end simulator well optimized for pad in order to do this conparison properly.

Even without the pad optimization issue, high-end physics have one simple problem: there are just too many factors at play that can make you lose control of your car/ give you a hard time. such as tire temperature, inputs, brake locking, track evolution and conditions, bad-timed downshifts, dirty tires etc. Simply put, a beginner would have no way of knowing what each of these things are without looking them up or being told about them in some kind of tutorial.

 

[ShiftingGears]

I don't buy the idea that there will be significant improvements from PD getting feedback from tour players.

Unless the penny drops and PD hire or poach new people from smaller devs, there isnt going to be much change.

 

[05XR8]

That’s why I’m thinking what could any changes be? WT drivers suggest less sliding? More grip? Less understeer? Softer dampers? Less roll? Most of the cars used are Gr.3 and Gr.4. How do those apply to all the different cars?

I mean, a ‘65 Mini and ‘69 Boss Mustang, are different to an EK9 Type-R and 2020 Shelby GT350R. Can’t keep setting majority of cars with .5 negative camber front, 1.0 negative camber rear, anti-roll bar 7/4 f/r.

Particularly older cars like described below:

I’ve only got the Cobra Coupe to do it. No other classic.

PD haven’t replicated the ‘83 GTI characteristic inner rear tyre lifting off the ground. Add Clio R.S. to that list. Lots of work.

 

[LeGeNd-1]

As great as AC's physics are, i don't think they are a good choice for the demographic Gran Turismo aims for, which is more or less casual gamers.

Yes they are realistic, but they are far from beginner friendly. To even get close to being decently fast, it requires a great amount of track time, which translates to crashes and spuns, which will discourage your player to keep going and instead he will probably going to pick up something easier.

Again this is a classic misunderstanding of what "realistic" physics is by the general population. I've said it many times before, realistic =/= hard. Realistic = intuitive.

If you haven't already, do yourself a favour and try the LaFerrari in Assetto Corsa and compare it to GT Sport. Then do the same with the 458/Huracan/R8 LMS GT3/Gr.3. Come back and tell me which one is easier to drive

 

[Magog]

I don't buy the idea that there will be significant improvements from PD getting feedback from tour players.

Unless the penny drops and PD hire or poach new people from smaller devs, there isnt going to be much change.

What reason do you have to question them other than being a pessimist?

 

[Samus]

What reason do you have to question them other than being a pessimist?

They've been listening to actual pro racing drivers for years. Kaz himself has driven a race car. Why do you think some gamers, no matter how fast, are suddenly going to make a difference giving feedback? With all due respect to them, what do any of them actually know about underlying physics?

 

[Beezer215]

It's not even about these kids knowing underlying physics or any of that. PD has already made up their mind on how the game is going to be. It will be a little better but it will not be that realistic. They want to include newcomers and veterans alike. PD has rarely listened to any end users in the past and they have never drastically changed the game. Why would they do so now?

 

[Magog]

They've been listening to actual pro racing drivers for years. Kaz himself has driven a race car. Why do you think some gamers, no matter how fast, are suddenly going to make a difference giving feedback? With all due respect to them, what do any of them actually know about underlying physics?

I think they choose to make the physics approachable for all rather than a punishing sim. He's acting like they don't have the talent to make a punishing sim but I think they adjust the physics to make it fun and realistic in the balance they want to achieve.

 

[ThePotatoKing]

Again this is a classic misunderstanding of what "realistic" physics is by the general population. I've said it many times before, realistic =/= hard. Realistic = intuitive.

If that is really true, shouldn't you be able to drive decently in any high-end sim right from the get-go? Unless i'm misunderstanding something here.

Just imagine someone who hasn't played a racing game before pick up a wheel and Asseto Corsa and try to picture his first drive.

 

[ArR29]

Do people actually think that PD aren't capable of doing research and development for car physics themselves? That's totally absurd thinking. They are a triple AAA developer and have enough resources if they truly want to focus on physics alone and make a fully driving simulator. This is more like a design philosophy and being aware of your own market base. Kaz has stated before that he doesn't want his game to be a hardcore simulator. And let's be honest here, even those top tier driving simulators have flaws with their physics and will never reach 1:1 to a real thing.

 

[05XR8]

It would depend on track. Kaz says High Speed Ring was chosen as it’s a good Beginner’s track. Pick the easiest track for any Beginner, with no prior driving experience and see how they go.

 

[Samus]

I think they choose to make the physics approachable for all rather than a punishing sim. He's acting like they don't have the talent to make a punishing sim but I think they adjust the physics to make it fun and realistic in the balance they want to achieve.

II agree but what's that got to do with world tour players supposedly helping them? Nobody is acting like anything, we're skeptical what help they would offer when as said, the limited physics are most likely a choice.

Kaz has specifically stated that they are taking the pro players input to improve the physics in GT7.

BTW, when did he state this?

 

[LeGeNd-1]

If that is really true, shouldn't you be able to drive decently in any high-end sim right from the get-go? Unless i'm misunderstanding something here.

Just imagine someone who hasn't played a racing game before pick up a wheel and Asseto Corsa and try to picture his first drive.

They should, provided they drive the same as real life. Meaning following speed limits, not going beyond their limit/fear, etc.

If someone has never played a racing game before, they're gonna fly off the road no matter what game they play if they try to go fullspeed the first time. No matter if it's NFS, GT or AC. It's a big difference being in a 3D world with g-forces and sense of fear vs 2D with no other cues except the wheel in your hand. We've seen many examples of even pro racing drivers finding it hard to acclimatise to a sim (assuming they don't often use simulators).



Putting lay people aside, do the test I described first and then come back and argue

 

[Magog]

II agree but what's that got to do with world tour players supposedly helping them? Nobody is acting like anything, we're skeptical what help they would offer when as said, the limited physics are most likely a choice.


BTW, when did he state this?

On this site. He's only done like 3 interviews about GT7. I'm surprised you haven't read them.

New Gran Turismo 7 Details: PS4 vs PS5, Driving Physics, GT Cafe, and More Revealed in Kazunori Yamauchi Interview

A week on from the big Gran Turismo 7 reveal and launch date announcement, GTPlanet got the chance to sit down with Gran Turismo series creator Kazunori Yamauchi for an extensive one-on-one interview on the latest title. Although there was plenty of information to dig through after the grand unve

www.gtplanet.net

“In terms of physics, of course, it is something in the game that we have been evolving for the last 25 years. In GT7, I think you’ll find the precision has increased. I also think you’ll feel it to be more natural, even more-so than in GT Sport,” he explained.

Yamauchi also revealed that some of the best Gran Turismo players in the world have helped Polyphony Digital to help refine the physics engine in GT7. “Now we have new people like Igor Fraga and Mikail Hizal joining the family. We have feedback from them that has been very useful in tuning that.”

He also goes in depth on the weather simulation physics

“The time and weather changes are continuous, in real time,” Yamauchi confirmed. “You can also set the time progression for a certain race to a multiple that you want. The weather simulation is being done on a very large scale; it’s a shame to just call it a ‘weather change feature’ because it really is a full-on weather simulation.

“Everything from the temperature, to the humidity, to the surface temperature of the track, to the cars passing over puddles to spray the water off the track. All those new things are implemented and calculated within the physics .”

 

[Scaff]

What reason do you have to question them other than being a pessimist?

Because they haven't managed to do it over the course of the series!

FWD cars are still wrong, FFB on understeer is still wrong, to give just two examples.

Could PD make these areas actually accurate? I would hope so, but they seem to have little desire to do so.

 

[ShiftingGears]

What reason do you have to question them other than being a pessimist?

Yamauchi has more personal real life racing experience than almost all finalists and the AI still sucks and the offline racing doesnt resemble real racing at all. The physics still resemble a GT game so again I dont see that changing and frankly its part of GTs appeal to the fanbase (insofar as giving casuals something they can pick up and play).

The only revolutionary driving change in the last 15 years has been the audio which was from poaching the Forza guy (may vacuum noises rest in peace).

Otherwise the collisions still suck and AI still sucks. The driving physics themselves have been...managably consistent. So again I dont see any significant changes happening unless theres staffing changes as PD tend to just tweak around the edges otherwise, even if Yamauchi isn't honest or aware enough to acknowledge it.

 

[Samus]

On this site. He's only done like 3 interviews about GT7. I'm surprised you haven't read them.

New Gran Turismo 7 Details: PS4 vs PS5, Driving Physics, GT Cafe, and More Revealed in Kazunori Yamauchi Interview

A week on from the big Gran Turismo 7 reveal and launch date announcement, GTPlanet got the chance to sit down with Gran Turismo series creator Kazunori Yamauchi for an extensive one-on-one interview on the latest title. Although there was plenty of information to dig through after the grand unve

www.gtplanet.net

“In terms of physics, of course, it is something in the game that we have been evolving for the last 25 years. In GT7, I think you’ll find the precision has increased. I also think you’ll feel it to be more natural, even more-so than in GT Sport,” he explained.

Yamauchi also revealed that some of the best Gran Turismo players in the world have helped Polyphony Digital to help refine the physics engine in GT7. “Now we have new people like Igor Fraga and Mikail Hizal joining the family. We have feedback from them that has been very useful in tuning that.”

He also goes in depth on the weather simulation physics

“The time and weather changes are continuous, in real time,” Yamauchi confirmed. “You can also set the time progression for a certain race to a multiple that you want. The weather simulation is being done on a very large scale; it’s a shame to just call it a ‘weather change feature’ because it really is a full-on weather simulation.

“Everything from the temperature, to the humidity, to the surface temperature of the track, to the cars passing over puddles to spray the water off the track. All those new things are implemented and calculated within the physics .”

I did when it came out, I don't remember everything.

So "very useful in tuning that" doesn't exactly say much and again, no disrespect to Fraga and Hizal, but I doubt they have anything they can bring to to the table that PD don't already know. As I said, they've consulted with pro racing drivers as far back as GT3, Kaz drives a racing car occasionally, they presumably all drive road cars of various types. They know very well what is missing from their simulation and what isn't. Or should do.

Physics might improve for GT7, but it won't be anything to do with World Tour players.