GT Sport - Trailers, Videos and Screenshots

  • Thread starter sk8er913
  • 17,667 comments
  • 2,109,581 views
Incorrect. Physics calculations and the resulting FFB do not necessarily take place at the frame rate. Project Cars, for example, has a physics calculation rate of 600 Hz.

I was trying to explain why you might see more at 60fps than 30fps.

Even at a physics rate of 600 Hz... (I am surprised the PS4 could be running physics at 600Hz?) that wheel is on the 80cm square section for 11 physics ticks which is 24ms so assuming you really are calculating at 600hz the FFB (on a logitech wheel) has an initial delay of 9-10ms (http://www.isrtv.com/forums/topic/2922-interesting-info-on-torqueresponse-time/) and would then get approximately 8 updates before no longer being on the 80cm square section of kerb. According to that post the Thrustmaster T500RS has a 20-23ms delay for the first update, so unless they are calculating FFB of where the wheel is going to be you're over the kerb before the wheel has reacted. Sure, you might feel it, but my guess is that they would more likely fake something.

Bottom line which I was trying to explain was the higher anything is calculated at the better it is going to be for seeing and feeling what is going on... but we are still a fair way away from having response times that would reward you for scanning every millimetre of a track.
 
Last edited:
I was trying to explain why you might see more at 60fps than 30fps.

Even at a physics rate of 600 Hz... (I am surprised the PS4 could be running physics at 600Hz?) that wheel is on the 80cm square section for 11 physics ticks which is 24ms so assuming you really are calculating at 600hz the FFB (on a logitech wheel) has an initial delay of 9-10ms (http://www.isrtv.com/forums/topic/2922-interesting-info-on-torqueresponse-time/) and would then get approximately 8 updates before no longer being on the 80cm square section of kerb. According to that post the Thrustmaster T500RS has a 20-23ms delay for the first update, so unless they are calculating FFB of where the wheel is going to be you're over the kerb before the wheel has reacted. Sure, you might feel it, but my guess is that they would more likely fake something.

Bottom line which I was trying to explain was the higher anything is calculated at the better it is going to be for seeing and feeling what is going on... but we are still a fair way away from having response times that would reward you for scanning every millimetre of a track.
Good point. Assuming your calculations are correct, that would mean that the delay in force feedback is going to be much closer to lining up with inherent lag from a great many of the television screens that the game is played on.
 
Cockpit is very dark.. But the cool thing is that one can see laptime in dash. 👍
It's a good idea, now let's see it be a full multi function display with multiple pages of data as we have had for over a year in Pcars (not to mention having UDP output for external apps in Pcars and AC).

It's nice, but once again behind the rest of the competition.
 
Been testing at BH Indy and hit a 54.064 so started recording as I was sure I'd get a 53.9 anyway ended up with an identical laptime :( ah well... it gives me something to plug away at later 👍


edit: got a 53.9

 
Last edited:
I really hope they get in some better AF in final release - it's so low the grass textures becomes blurry at a length of a car.

After playing a lot 12AM "sunny" still looks wrong to me. The colors, contrast, bloom - it's like some wrong settings in game engine.

Other times of day like 19:00 on Willow or 9AM on Brands look excellent and with those I'm super happy.
 
It's a good idea, now let's see it be a full multi function display with multiple pages of data as we have had for over a year in Pcars (not to mention having UDP output for external apps in Pcars and AC).

It's nice, but once again behind the rest of the competition.
I think PD could put together a really cool app with lots of functionality on a tablet or smartphone for us to mount on our rigs. Like you say others do it well so no real reason not to expect in such a high profile game.
 
I really hope they get in some better AF in final release - it's so low the grass textures becomes blurry at a length of a car.

After playing a lot 12AM "sunny" still looks wrong to me. The colors, contrast, bloom - it's like some wrong settings in game engine.

Other times of day like 19:00 on Willow or 9AM on Brands look excellent and with those I'm super happy.
There is something wrong with the engine, I can't say what but the engine doesn't showcase the HD textures.
 
1.04
jg6jaH.png

1.05
8sl0qE.png

1.04
WuVfV9.png

1.05
LXxS9c.png

1.04
OUcaoT.png

1.05
iOaXnj.png
 
Last edited:
Been keeping an eye on this thread and seeing how the game has evolved. I must say I am very impressed in terms of graphics and sound. I am looking forward to playing this game when it comes out :)

I like your comment and I agree with everything you said but if you want any attention in this forum then you have to criticize and beat GT like when many people started losing their heads over 2D trees.To me, that is just silly.
Or people stating that the sounds are garbage or not good enough when this GT compared to GT6 it is not even in the same planet in the sound department (the sounds were the only reason I left GT6) but it is good to know that are people out there that still appreciate PD effort to make the best possible game, dispite not perfect, this GT is looking very sharp and polished.
 
Forget the trees,

Did you see the freaking detaild en awsome reflection in the cars paint from the cerbstones etc?!

Its in the little details!
 
So let's say you are travelling at a modest speed of 120 km/h which is 3333.33 cm/s so at 30 fps you would have travelled 111.111 cm's between frames... and if an FIA kerb was made up of 80cm square sections then you might not see (or feel) the wheel even touch a kerb section. Seems like you would want a minimum of 60 fps based on these calculations or your simulation is going to have to do things to fake it.

On that note, carrying on about having tracks scanned with millimetre accuracy would mean that if you were travelling at 120km/h according to my calculations you would need to have a frame rate and force feedback response equal to approximately 33333.3 fps to get the benefit of all that extra data...

Physics isn't just about where you are at each frame, but also about what happened in between the frames. If you hit a wall between two frames you're not sent ghosting right through it. The physics engine can tell that there was an impact between the frames and tell the car to behave accordingly.

As for seeing the wheel on the curb, that's not something you could do at 60 fps either. Reviewing the video frame by frame, yes, but in real time it's just too fast. It's not like you register gaps between the frames when you see something in 30 fps, the motions just seem a little less smooth.
 
I like your comment and I agree with everything you said but if you want any attention in this forum then you have to criticize and beat GT like when many people started losing their heads over 2D trees.To me, that is just silly.
Or people stating that the sounds are garbage or not good enough when this GT compared to GT6 it is not even in the same planet in the sound department (the sounds were the only reason I left GT6) but it is good to know that are people out there that still appreciate PD effort to make the best possible game, dispite not perfect, this GT is looking very sharp and polished.
The people who are criticizing happens to have been a fan from the GT series since the first irretation. You do not need to criticize the game in order to gain attention nor thats the reason people criticize.

One example: If you had a product from 10 years ago (Gran Turismo ps1) and buying that same one as of today (Gran Turismo Sport ps4) you would expect the product to get better right (since you do not want to deal with the same thing over and over again) You see some minor and some major improvement on the product and it seems to have gotten better and better over the years. Now compare that same product with another one from another place ("insert other racing game here") which seems to be a far better one (better details, sound, physics and can offer more things than your own product), would you not feel dissapointed about your product and start demanding for yours to get on the same level or better than that product?

That is the main reason people criticize the game in order for it to get better, If Polyphony Digital sees that nobody is saying anything they are going to think "Wow everyone likes the game the way it is!" and which clearly is not the case here.
 
Physics isn't just about where you are at each frame, but also about what happened in between the frames. If you hit a wall between two frames you're not sent ghosting right through it. The physics engine can tell that there was an impact between the frames and tell the car to behave accordingly.

As for seeing the wheel on the curb, that's not something you could do at 60 fps either. Reviewing the video frame by frame, yes, but in real time it's just too fast. It's not like you register gaps between the frames when you see something in 30 fps, the motions just seem a little less smooth.

I was trying to illustrate a fictional situation to share my thoughts... how do you know the physics is about what happens between the frames? How do you know how fast any physics engine is running at? Are you doing debugging on the PS4? How do you know they aren't calculating a few iterations of the physics every frame using physics substepping?

And the situation I was trying to illustrate was; if there is a frame with a wheel on the kerb, at least you have a chance to see it, if there was no frame, you are just imagining what's happening in the gaps between frames and you might imagine something different to what the PS4 calculates.

BTW since 600hz was mentioned as a reference to Project Cars, I had a read up about that, here's the explanation directly from the developer:

http://forum.projectcarsgame.com/showthread.php?26370-Project-CARS-On-AMD-GPUs-Clarification

• The MADNESS engine runs PhysX at only 50Hz and not at 600Hz as mentioned in several articles

• The MADNESS engine uses PhysX for collision detection and dynamic objects, which is a small part of the overall physics systems

So actually yes, in this case the Collision Physics (50hz) is being calculated less often than the screen refresh (60hz), so the car probably does appear to go through that wall if it's rendered before the Physics realise it has collided.

I hope this doesn't come across as argumentative, I'm just trying to contribute, I have been following various threads and frankly I have learnt a hell of a lot on here about driving, and game development, so I'm extremely grateful for being involved in these conversations and am happy to learn more, I'm just explaining how I think things work, I may be wrong.
 
Last edited:
Had a really fun Gr.4 race chasing the GT-R of @badaboum528 at Big Willow in the Alfa 4C. Shout out to @cKemc_PUR who fell victim to the grid bug but let the field go by.


Another fun race splitting up the Evo's of @scope_eye33 and WINNER_CIRCLE24 at Brand's Indy in the Lotus Evora, though I probably had no business going in front of WINNER_CIRCLE24, as he seemed to be faster than me once I got by. @MOPARbarrett5 kept the pressure on in the Audi TTS from behind by never leaving the rear-view.
 
Last edited:
Here's an onboard view of my best lap in the N300 Evo @ BH Indy.. I had to skip to lap 2 in replay mode as I wasn't recording live.. the set up is shown at the beginning of the video, I'm not sure if I took someone else's set up and fiddled with them to my liking but it's what I'm using anyway :)👍

The laptime is 53.923 and barely 1st in EU by a whisker.. until others use these settings and I wake up tomorrow in 10th place haha :D

Good luck guys and enjoy the car now! :cheers:

 
how do you know the physics is about what happens between the frames?

Because I read about it. There are essentially two forms of collision detection: discrete and continuous. Discrete detection only catches collisions that happen at each physics frame. Continuous catches collisions between physics frames.

Example: Catching the bullet. Say that you fire a machine gun at somebody's hand in a game like GTA. For collision purposes the hand is 0.05 m thick while the bullets are 0.02 m long. The bullets travel at 1000 m/s.

With discrete collision detection, most bullets would miss the hand, as they in frame 0 would be in front of the hand and in frame 1 behind the hand. In neither case the bullet would be touching the hand.

Even at 1000 fps (which is really expensive if you're going to calculate collision detection in each of these frames) the probability to hit the hand is 0.07, which is like 1 hit for every 14 bullets that you fire, assuming that you adjust your position between each shot - if not then odds are 13:1 that all bullets would miss.

With continuous collision detection, every single bullet would hit even at a frame rate of 1 per year.

Let's go back to the curbs. What triggers the rumble? Is it the wheel being ON the curb, or is it the wheel travelling over the edge of the curb? I think that we can agree that it's the edge, right? Now, what is the length of the edge along the dimension of which the car is travelling? Not more than a centimeter or two, right? How long (in the same dimension) is the contact patch of the tyre? Let's say 10 cm. If the car is travelling at 30 m/s, then the tyre would be in contact with each edge for (0.1+0.02)/30 = 0.004 seconds. That is 1/250th of a second. At 60 fps you'd be able to cath a little less than every fourth edge on average. Unless you happen to time the speed and the dimensions of the curbs so that for each frame the tyre is right between two edges, in which case you wouldn't feel a thing.

With discrete collision detection, that is. With continuous detection you'd get a rumble every time you drive over a curb.

How do you know how fast any physics engine is running at?

Any physics engine? Well, in Blender I can check the physics settings. What is your method?

Are you doing debugging on the PS4?

Why would I debug the console, and what has that got to do with anything?

How do you know they aren't calculating a few iterations of the physics every frame using physics substepping?

I don't and I never said that they aren't. I responded to a hypothetical situation where it was assumed that the physics are tied to the framerate (which they may or may not be). The questions were: 1. Is the physics the reason why you pick a certain frame rate? 2. Would 60 fps be better than 30 fps for the physics?

And the situation I was trying to illustrate was; if there is a frame with a wheel on the kerb, at least you have a chance to see it, if there was no frame, you are just imagining what's happening in the gaps between frames and you might imagine something different to what the PS4 calculates.

You can't see a single frame at 60 fps. All you see is that the motions are more smooth.

BTW since 600hz was mentioned (I'm assuming as a reference to Project Cars) I had a read up about that, here's the explanation directly from the developer:

http://forum.projectcarsgame.com/showthread.php?26370-Project-CARS-On-AMD-GPUs-Clarification

So actually yes, in this case the Collision Physics (50hz) is being calculated less often that the screen refresh (60hz), so the car probably does appear to go through that wall if it's rendered before the Physics realise it has collided.

Sure, but does the physics treat the collision as if it happened on the frame, or does it calculate the actual time and point of impact?
 
Thanks for this explanation, I need to read more about discrete and continuous, but if I understand what you said, in continuous they are treating collision areas as the whole length that the object has travelled during that physics "tick" whilst discrete they only do the collision checking during the "tick".

I think you will find that for a lot of bullet collision detection they use ray casting and don't bother with trying to calculate very high speed collisions as it is unreliable. But they still try to display something that looks like that happened.

I think displaying information to the user at 60fps vs 30fps is going to affect how they perceive the physics, so yes, I honestly believe that 60fps is better for the physics. Most games run at variable frame rates and the motion of a vehicle or projectile or another player is very hard to predict when the frame rate drops.

I guess I was also trying to say that no one really knows how Gran Turismo is doing this, they've been doing physics since before PhysX was a thing and had to roll their own, so we really don't know what they are doing.

Other developers are making claims like the physics runs at 600hz and other developers are claiming that the track is scanned to millimetre detail levels, all this is giving people a lot of what I believe are incorrect assumptions about what's going on behind the scenes in a racing game.
 
I like your comment and I agree with everything you said but if you want any attention in this forum then you have to criticize and beat GT like when many people started losing their heads over 2D trees.To me, that is just silly.
Or people stating that the sounds are garbage or not good enough when this GT compared to GT6 it is not even in the same planet in the sound department (the sounds were the only reason I left GT6) but it is good to know that are people out there that still appreciate PD effort to make the best possible game, dispite not perfect, this GT is looking very sharp and polished.
I like what I see. That is all :)
 
Nice win, I wasn't sure if anyone else was racing you until lap 3 :P, is the glare and sun effects as distracting as it looks in the video, or are youtube compression artefacts making it look worse than it actually is?

Sorry about the off topic conversation.
 
Nice win, I wasn't sure if anyone else was racing you until lap 3 :P,
I got a good start and the Lancer alongside me on the grid didn't, and held everyone up. The Ferrari got through to 2nd by Paddock, but I was half a second clear by then. He gained on me on lap 2, and got quite close up Hawthorn hill on lap 3, but I took the inside line and then he made a mistake at Sheene curve, which gave me the win.

is the glare and sun effects as distracting as it looks in the video, or are youtube compression artefacts making it look worse than it actually is?
It is quite annoying, although not as annoying at Brands as it is at Willow Springs, and I hate it when games do that as an artificial way to give you a hard time.
 

Latest Posts

Back