Random Online Physics

[bugsbunny]

I'm writing this post out of rage because I've had too many damn races ruined this weekend from random crappy online physics.

I can drive the same car, into the same turn, the same way every race but some races, the car will feel like it hits an oil slick and I spin. It also seems that the physics change dramatically (for the worst) with more people in the room.

I had some fun races just this morning with rooms that had 5-6 people. No problems. I get into 12-13 person rooms and all of a sudden, my car feels like its on ice and I'm doing everything exactly the same. It sucks and I wish PD would find a way to make the physics consistent all the time and make it so the offline physics and online physics are exactly the same!

Rant over.

the real kerb / outside grip setting

or

real tyre wear and fuel load

or

Skid recovery force

can easily change the physics

but the physics has nothing to do with the connection.

also physics in pratice and a spec are arcade so do not compare with online.


A ferrari enzo in gt5 understeers even below 100 kmh although the chasis produces the same downforce as few racing cars , almost half a ton of downforce at 124 mph (200 kmh) you couldn't make it to those speeds and it would understeer even way before that.http://auto.howstuffworks.com/enzo3.htm , when gt5 just came out the enzo was fine then updates came and its ruined.
in nurburgring nord in gt5 , the enzo will take off like an airplane even though it produces half a ton of downforce at 200+ kmh.

 

[208SpeedGeek]

Thanks for all the replies to my angry rant. I'm glad I'm not the only one that has been annoyed by this issue.

All the sudden the grip is gone, no matter how fast or how slow you are going you're gonna spin out for no reason at all, the breaks with abs 1 is like having no abs at all, the car just cant handle it self.
In free run is better but when you start the race ..... the damn thing is going nuts.

This. Exactly! It's cool that you guys actually ran a little experiment to see how crazy the physics actually got.

 

[Antineutrino]

👎
It's really dumb to have two different physics... If this is the 'real driving simulator', please PD choose only the one closest to reality and apply to both online and offline driving. Please....!!!

 

[DDAAVVIIDD82]

It's not so much to do with PD's programming, and more to do with internet connection strength. Online tunes do their best to cure problems that this causes.

 

[diptob79]

Do you mean, the faster your connection will close the gap between online and offline physics ?

Please elaborate

 

[DDAAVVIIDD82]

That and the connection of the host, for some peculiar reason a bad (or not great) connection increases the loss of traction during gameplay, or at least this is what I understand to be the case. I'm no expert, but this appears to be the consensus.

 

[diptob79]

Private lounge, alone, still different imo.

 

[DDAAVVIIDD82]

Still not a hard (offline) connection though. The lounge interface, and its ability to host other players, means that there will still be inconsistencies.

 

[MC_PEA]

That and the connection of the host, for some peculiar reason a bad (or not great) connection increases the loss of traction during gameplay, or at least this is what I understand to be the case. I'm no expert, but this appears to be the consensus.

I dont belive this is true because when you are online the connection strength only effects the loading of the track and the other cars on the track like when they are boucing around. And also if you open up a loddy your self and your the only one in it the online physics is still different to offline.

 

[Schwartz]

I think you are looking for this thread.
https://www.gtplanet.net/forum/showthread.php?t=233548
Please use the search function before creating a new thread about the same thing. It keeps everything on GTPlanet condensed and clean so there is not a mess of duplicate threads when you are looking for something. Thanks

 

[Juiposa]

I believe it has to do with RAM. In Offline there is less the PS3 has to process, and store in RAM. For Online there are addition scripts, with lobbies, voice chat, packets, and such. I don't know if it makes a difference, but if Offline took up all 3.4 GHz and all the RAM, then they would have needed to trim fat of the scripts to be able to run online.

Mind you this is my theory.

 

[diptob79]

@schwartz, that thread discussing differences within single online lobby


There are 3 types of physics I know in GT5. Some threads already discuss about it:
-offline physics
-online physics, free run mode
-online physics, race mode

The question on this thread is WHY?

And its good to see people came up with theories.

 

[Schwartz]

@schwartz, that thread discussing differences within single online lobby

There are 3 types of physics I know in GT5. Some threads already discuss about it:
-offline physics
-online physics, free run mode
-online physics, race mode

The question on this thread is WHY?

And its good to see people came up with theories.

Its the same thing (To me anyway). You are just splitting hairs. I'll let it be, and let the moderators do as they wish. After all, they are the ones who get to make the final desicions. Good luck!

 

[DDAAVVIIDD82]

I'll just go back to my original answer that it isn't a decision that PD made, to make vehicle dynamics differ between various methods of gameplay. That said, it could be a passive result of the decision to not do a better job of making the various methods similar. This is in no way a rant against PD or their decisions, I'm content with the game as it is now and I appreciate that it hasn't been "left for dead," it's merely a possible conclusion.

 

[diptob79]

^^combine it with the hardware issues above, so if we have a good connection plus if we can optimize PS3's performance (clear cache, no BGM, low resolution, etc), potentially we can gain valuable hundredths of second..

Interesting.. Will try it tonight

 

[Ealirendur]

@schwartz, that thread discussing differences within single online lobby


There are 3 types of physics I know in GT5. Some threads already discuss about it:
-offline physics
-online physics, free run mode
-online physics, race mode

The question on this thread is WHY?

And its good to see people came up with theories.

^^combine it with the hardware issues above, so if we have a good connection plus if we can optimize PS3's performance (clear cache, no BGM, low resolution, etc), potentially we can gain valuable hundredths of second..

Interesting.. Will try it tonight

Good question, and I'm glad you asked.

No, I have no real information either, but I'm happy to speculate!

In the game between two or more PS3s, there are many... factors... to consider. The wind.... *ahem*

So, ignoring things like tire wear and fuel mass, draft (free run has it not, and no dirty air effects therefore). Races will have draft and dirty air. If you believe in that stuff.....

It will have to do (at least partly) with the speed of light. Ya canna break the laws of physics, Cap'n! (or at least how long it takes electrons and photons to travel through appropriate media).

Local processing only has the resources of the PS3 to worry about; RAM, CPU, GPU, all that good stuff.

Introduce geographically separated PS3s with humans near them, much hilarity ensues. Even the protocols PD chose for transmitting between PS3s, right down to TCP/IP vs UDP, leaving aside star/mesh, might affect in a small way how we perceive it.


"I've got a joke about UDP.... but you might not get it".

When latency and some uncertainty enter the picture (bandwidth, latency, packet loss), the physics model has now to cope with the telemetry from the other cars on track not being current or all there; I presume from what I've experienced (and reasonable game programming) that we send and receive only a minimalist set of data that will allow our PS3s to show us the remote players' cars, since that will be much faster and less prone to error. Even so, gaps or dropouts in that data cause our view of someone to be jumpy, for example. And the collision detection mechanisms of the model have to cope with that (draft effects, too).

Lots of highly detailed cars close together at the start of a race with uncertain collision boundaries? Sounds like a guarantee of easily computable fun!

There's also how lag messes with our perceptual biases. Frame rate drops - and particularly if it applies to, say, force feedback, could make you feel different things than you are used to. This is a game where we care about tenths and hundredths, at least.

The Race Quality settings, and online in general, probably induce sets of default compromises. To what precisely, I couldn't say.

I'm guessing Kaz wouldn't like unecessary compromise, so the offline game is as good as he and PD can make it under those constraints (programming time, budget, local PS3 processing power) - why limit performance to an unknowably variable lowest common denominator?

 

[Scaff]

@schwartz, that thread discussing differences within single online lobby

Which fits in with and is a logical extension of the existing thread.

Merged.

Scaff

 

[diptob79]

Which fits in with and is a logical extension of the existing thread.

Merged.

Scaff

Thanks. 👍

Good question, and I'm glad you asked.

No, I have no real information either, but I'm happy to speculate!

In the game between two or more PS3s, there are many... factors... to consider. The wind.... *ahem*

So, ignoring things like tire wear and fuel mass, draft (free run has it not, and no dirty air effects therefore). Races will have draft and dirty air. If you believe in that stuff.....

It will have to do (at least partly) with the speed of light. Ya canna break the laws of physics, Cap'n! (or at least how long it takes electrons and photons to travel through appropriate media).

Local processing only has the resources of the PS3 to worry about; RAM, CPU, GPU, all that good stuff.

Introduce geographically separated PS3s with humans near them, much hilarity ensues. Even the protocols PD chose for transmitting between PS3s, right down to TCP/IP vs UDP, leaving aside star/mesh, might affect in a small way how we perceive it.


"I've got a joke about UDP.... but you might not get it".

When latency and some uncertainty enter the picture (bandwidth, latency, packet loss), the physics model has now to cope with the telemetry from the other cars on track not being current or all there; I presume from what I've experienced (and reasonable game programming) that we send and receive only a minimalist set of data that will allow our PS3s to show us the remote players' cars, since that will be much faster and less prone to error. Even so, gaps or dropouts in that data cause our view of someone to be jumpy, for example. And the collision detection mechanisms of the model have to cope with that (draft effects, too).

Lots of highly detailed cars close together at the start of a race with uncertain collision boundaries? Sounds like a guarantee of easily computable fun!

There's also how lag messes with our perceptual biases. Frame rate drops - and particularly if it applies to, say, force feedback, could make you feel different things than you are used to. This is a game where we care about tenths and hundredths, at least.

The Race Quality settings, and online in general, probably induce sets of default compromises. To what precisely, I couldn't say.

I'm guessing Kaz wouldn't like unecessary compromise, so the offline game is as good as he and PD can make it under those constraints (programming time, budget, local PS3 processing power) - why limit performance to an unknowably variable lowest common denominator?

nice explanation.
Bro, can you relate your explanation if we are alone in our privatelounge? Because it still has noticable different with online practice.

 

[GODfreyGT5]

I have had this same issue in the last three races of two different race series. In an S15 only race, I was lapped by the leader on lap 5, due to my car's sudden loss of grip and/or power. More recently, last Tuesday in BTCC, I was passed by the bigger half of 12 other cars on just the straight of Laguna alone! I believe this is related to the extreme frame rate drop I get during big-grid races.

 

[Ealirendur]

nice explanation.
Bro, can you relate your explanation if we are alone in our privatelounge? Because it still has noticable different with online practice.

...ignoring things like tire wear and fuel mass, draft (free run has it not, and no dirty air effects therefore). Races will have draft and dirty air. If you believe in that stuff.....


The Race Quality settings, and online in general, probably induce sets of default compromises. To what precisely, I couldn't say.

Most people don't practice or tune with tire wear and fuel consumption on. That's a biggie, since the tire temperature modelling is different (affecting grip everywhere), you're losing grip a lot sooner, and your car has ~71.1kg (IIRC) of extra mass to lug around, that has changed the roll center of the car some more. It's fluid, too, just for yucks (if they went that far).

The "default sets of compromises" I mentioned (high/standard/low, and being online at all) probably do cut out some bits of the physics model to allow for all the complicated communications (incoming and outgoing - Even if they're not happening yet, since you've probably got to load some code/libraries/APIs and such you didn't before just to be ready). RAM limitations may indeed play a part. It may adjust on the fly to cope with worsening conditions, and there's no guarantee it's going to always get it right.

Complicated software is complicated. It always has bugs. How one deals with non-deterministically located cars moving through space at a great rate and I WAS EXPECTING THOSE PACKETS THREE HUNDRED NANOSECONDS AGO, AND THE OTHER WAY AROUND are the sorts of things that make programmers nervous.

 

[diptob79]

Most people don't practice or tune with tire wear and fuel consumption on. That's a biggie, since the tire temperature modelling is different (affecting grip everywhere), you're losing grip a lot sooner, and your car has ~71.1kg (IIRC) of extra mass to lug around, that has changed the roll center of the car some more. It's fluid, too, just for yucks (if they went that far).

The "default sets of compromises" I mentioned (high/standard/low, and being online at all) probably do cut out some bits of the physics model to allow for all the complicated communications (incoming and outgoing - Even if they're not happening yet, since you've probably got to load some code/libraries/APIs and such you didn't before just to be ready). RAM limitations may indeed play a part. It may adjust on the fly to cope with worsening conditions, and there's no guarantee it's going to always get it right.

Complicated software is complicated. It always has bugs. How one deals with non-deterministically located cars moving through space at a great rate and I WAS EXPECTING THOSE PACKETS THREE HUNDRED NANOSECONDS AGO, AND THE OTHER WAY AROUND are the sorts of things that make programmers nervous.

Thanks for your explanation.
So now I'm curious to try the different between low, normal and high quality room settings.