Car handling --- Different online

  • Thread starter Sail IC
  • 294 comments
  • 35,309 views
To be honest I'm beginning to think it's largely due to our our preferred driving style (and thus the cars that we select) that we notice it more than others. If I try to avoid drifting and just follow the game's racing line and brake cues I don't notice much difference, except for some extra oversteer on powering out of the corner.

Then I did the same in my lobby which was a completely different experience. There's just seems to be less grip but at the same time more snapsteer. :confused: Laptimes where all over the place and didn't manage to lap under 2:10 before I got fed up with it :grumpy:

I worked my way through a set of tires and at no point I felt the handling was the same as in offline mode.

Anyway, I can't say which mode is "right", but offline is a lot easier to me and appears to be much more like the old versions of the game. In any case something is definately different online.

Glad you are liking the Alpine, I just love that little car.

Anyway, I spent about 2 hours today just trying to eliminate the SnapBack. This time I took a stock BTR and then added a customizable LSD and racing suspension. I used deep forest as my test track since Trial Mountain exaggerates this problem so much. I used sports soft tires Offline things were pretty good. The car responded pretty much as a classic-era Porsche does. Like the alpine (and like my real life cars), I could pretty much play "lift-rotate-counter steer-depress-come back around" all day long. I would get occasional SnapBack if I let her come way too far around. It was all maybe a bit too easy, but in terms of how the car felt and how it responded to real-world driving techniques, it was very close to reality as far as I can judge it.

Online of course everything went haywire. Even the tiniest amount of lift-off oversteer frequently resulted in catastrophic SnapBack. Additionally, the ability to squeeze the right pedal down, shift the weight reward, and regain rear lateral traction was completely (and I do mean completely) absent. Without this basic fundamental real-world handlng characteristic, driving this car as it was designed to be driven will be essentially impossible online. Nonetheless, I went about trying to get rid of the SnapBack. Nothing I tried worked. As mentioned above, it is possible to make the car respond at least somewhat like it should by adjust rear toe and LSD settings, but I tried everything I could think of and was unable to eliminate the SnapBack. So if anyone here comes up with something that works, please post it!

Then I went back to stock and played with it a bit. Try this: take one of these RR cars and pretend it is an FR car: go into a corner and instead of initiating turn-in by lifting the throttle slightly, start feathering the throttle as if you were trying to force slight oversteer on-throttle like an FR car. (Note: do NOT try this in a real-life classic Porsche because unless you are a driving GOD you will kill yourself if you actually manage to break the rears loose under power. This is NOT the way RR cars drive. Watch some old Top Gears. Watch Clarkson play drifter in all the FR cars. Watch Clarkson spin Porsches by trying to drive them like FR cars. Watch Clarkson hate one of the best handling cars of all time because he is too clumsy or too stupid to learn how to drive them.)

Anyway, if you do this you'll discover that the BTR when driven online handles like a poorly balanced FR car. It's almost as if nobody testing the online Physics at PD had any idea at all how RR cars handle, and thus made them handle like an FR car with a really heavy butt. But this isn't correct. Driving an RR car properly in a performance manner is vastly, vastly different than an FR car. I just don't get how they could have bungled this so badly.

At least for now I'm going to try and give up driving all my favorite cars online and try to enjoy some other ones instead. There is simply no way I'll ever be able to learn this completely wrong way to drive the RR cars; in every turn my real-life experience tells me the right way to do it and I don't think I could overcome that even if I wanted to.

I would be very interested to hear from anyone who has managed to tune-out the excessive online SnapBack in these cars. But ultimately, unless PD adopts a physics model that is the same both online and off, and which works properly with all types of cars across all tracks, then I think any chance the series ever had of being taken as a serious sim is in jeopardy. It's hard to defend a "sim" in which the exact same car handles two completely different ways...
 
Last edited:
I'd just like to say that it's very possible that there are two different physics engines for on and offline play, but I don't think it's likely.

I do think that the physic aspect of the game is very complicated, and it's much more likely that a couple parameters within the same physics engine have been changed for online play.

The contention that "cars can't roll over online, therefore there must be a different physics engine" is poor logic. If I were programming the game, and the programming team decided that allowing rollovers online would simply incite people to create the biggest, most spectacular smash-ups ever, the programming team would write an algorithm -- within the same physics engine -- to prevent rollovers. If the longitudinal axis of the car is defined as the Z-axis, then you simply prevent the Z-axis from ever rotating above 90 degrees, or below -90 degrees. The physics up to that point would remain identical. Re-write an entire physics engine to prevent rollovers? No way -- not cost-effective, and totally unnecessary.

Change parameters, yes, re-write the physics engine, definitely not from a programming and/or project development standpoint.

But there's no doubt that even one or two parameters can vastly change the way a car handles. I'd conjecture that a parameter has been changed for online play that exacerbates over-rotation, and that parameter really has a nasty effect on cars with the RR drivetrain.

On race day, tracks can get oily. I don't know if oil on the track surface is part of the simulation or not. But I'm very inclined to think it's a one or two parameter change within the physics engine that results in heightened oversteer online.
 
I pretty much believe that GT5 also simulates rubbering of the track in online mode, that can explain problems with RR cars, where it really stands out. Try free run in your lobby, do 30 laps and then judge.
 
If PD/Kaz would just come out and tell us what the deal is it would save us a lot of grief. Why are they so mysterious about all these huge problems/issues we have with GT5? Why not just be straight up with us and tell us what is going on with on-line physics? Vague nonsensical answers aren't cutting it.
 
In my testing, it feels to me as if the difference is that in offline practice the car is running with a weight as if there is no fuel in the tank, and the tires do not wear. Add 100-150lb to the rear and you should be a bit slower and looser.
 
Just ran my first GT5 online race last night....BIG difference from the test track....:crazy:

I will start tuning my online cars from my lobby....
 
BWX
If PD/Kaz would just come out and tell us what the deal is it would save us a lot of grief. Why are they so mysterious about all these huge problems/issues we have with GT5? Why not just be straight up with us and tell us what is going on with on-line physics? Vague nonsensical answers aren't cutting it.

You are going on the assumption that Kaz, or one of that crew has any clue what is wrong. I think that's a lot to assume given the level of competence they have demonstrated throughout the game. I seriously doubt anything will ever get "fixed", we will be given some "freebies" now and then to keep the kids happy, but problems with the physics are way beyond the capabilities of Kaz and the rest of the developers.
 
BWX
If PD/Kaz would just come out and tell us what the deal is it would save us a lot of grief. Why are they so mysterious about all these huge problems/issues we have with GT5? Why not just be straight up with us and tell us what is going on with on-line physics? Vague nonsensical answers aren't cutting it.

With some developers is not that easy, they tend to keep bugs or strange features hidden, especially Orientals ones. It's something I've seen thousand times and I don't think PD is different. I would like to know what's the difference between offline and online too, just to know where to go with my tunes with some certainty.

In my testing, it feels to me as if the difference is that in offline practice the car is running with a weight as if there is no fuel in the tank, and the tires do not wear. Add 100-150lb to the rear and you should be a bit slower and looser.

It could be, but if you've ever done an endurance you should know that tires consumption and fuel is active in those races and the car doesn't behave like online does, and that's mean, in my opinion, that there is something more.

You are going on the assumption that Kaz, or one of that crew has any clue what is wrong. I think that's a lot to assume given the level of competence they have demonstrated throughout the game. I seriously doubt anything will ever get "fixed", we will be given some "freebies" now and then to keep the kids happy, but problems with the physics are way beyond the capabilities of Kaz and the rest of the developers.

I understand the disappointment, because I am more than disappointed with the online (and some offline features, or lack of), but I've the feeling that we aren't talking of a bug that need to be discovered by PD, the physic engine is the same online and offline, with different settings for rear tires grip or car balance that brings a huge difference for the majority of RWD cars.

I just have the idea, at this point, that they've added a different car's handling online, just to make the game more challenging, without any relation to the "reality" and that's, i fear, the way many players, experienced and less experienced alike, think. Add the impossibility to have more than one setup per car and I don't see any value, at this time, of GT5 online.
 
Just ran my first GT5 online race last night....BIG difference from the test track....:crazy:

I will start tuning my online cars from my lobby....

I'm running in the Touring Car Championship online series and the difference is VERY noticeable. The car is MUCH loser online. Tire temp and tire wear are part of it for sure but I have not done enough testing to determine how to tune around those factors. All I know is you have to tune online if that's where you plan on racing because the difference is BIG.
 
Add the impossibility to have more than one setup per car and I don't see any value, at this time, of GT5 online.

Don't forget about the fact that you can't sort rooms by connection quality, or actually get new rooms by hitting "refresh list."
 
Don't forget about the fact that you can't sort rooms by connection quality, or actually get new rooms by hitting "refresh list."

You will get new rooms by pressing triangle and you see race and connection quality but is all very random.
 
Just ran my first GT5 online race last night....BIG difference from the test track....:crazy:

I will start tuning my online cars from my lobby....

You have to tune your cars in online lobby to accomodate tyre wear and temperature. Otherwise no big difference in handling of the car. I've tried Alpine 1600 and Yellow Bird on Nurburgring recently and failed to see any difference outside of different tyre model.
 
You will get new rooms by pressing triangle and you see race and connection quality but is all very random.


I get a cycle of 4 sets of rooms, always repeating the same ones. So, no, you DO NOT get new rooms, and it IS NOT random even though the game claims it is. :grumpy:
 
I have yet to find anyone who actually gets random rooms.


Keep hitting refresh and post up the number of rooms in your list each time. Dollars to donuts it repeats a set of 4.
 
BWX
If PD/Kaz would just come out and tell us what the deal is it would save us a lot of grief. Why are they so mysterious about all these huge problems/issues we have with GT5? Why not just be straight up with us and tell us what is going on with on-line physics? Vague nonsensical answers aren't cutting it.


Im saying, shoot I wish Turn10 could be PDs subcontractor for community relations/forums, then we would at least know they were looking at these issues.
 
I have a slightly different but related experience. I went to arcade mode, used the arcade-only SLS AMG '10, did not change anything in the car's settings, turned of all assists except ABS which is set to 1.

I did a free run in Rome circuit and the car was manageable, I achieved decent lap times. Afterwards, I did a single-race (still in Arcade mode, same settings) and the car fishtailed like crazy in every corner. What's up with that?

Can you guys try this and see if you experience the same thing?
 
Online of course everything went haywire. Even the tiniest amount of lift-off oversteer frequently resulted in catastrophic SnapBack. Additionally, the ability to squeeze the right pedal down, shift the weight reward, and regain rear lateral traction was completely (and I do mean completely) absent. Without this basic fundamental real-world handlng characteristic, driving this car as it was designed to be driven will be essentially impossible online.
You're right, and using weight transfer to get more grip at a particular tire is one of the fundamental tools used by a driver.

This keeps rattling around in my head, and it sounds like a bug in the software. Handling/grip at each corner will be calculated with the strength of the driving aids employed factored in. But if the driving aids are being mishandled online, we'd get screwy results. It just so happens that online is where the 'room' overrides the driving aid settings already set by each driver. My conjecture is that the room's overrides are causing driving aid settings to be calculated incorrectly. Where some aids are numerical from zero to 10, a miscalculation could cause the physics engine to treat a driver's aid as if it was set to a negative value. A traction aid with a negative value would cause some freaky and unpleasant results.
 
I was in a room last night which was nascar races and I couldn't stop it spinning out. In A-spec I never had this problem it is virtually impossible to spin the car out. I don't think it's a lag problem either as I took my car back to stock suspension settings and it was better but it was still a handfull. The room was set to all assists off which is the way I have it in game so it's not that and I don't think it's tyre wear as even on new tyres it was the same.
 
You're right, and using weight transfer to get more grip at a particular tire is one of the fundamental tools used by a driver.

This keeps rattling around in my head, and it sounds like a bug in the software. Handling/grip at each corner will be calculated with the strength of the driving aids employed factored in. But if the driving aids are being mishandled online, we'd get screwy results. It just so happens that online is where the 'room' overrides the driving aid settings already set by each driver. My conjecture is that the room's overrides are causing driving aid settings to be calculated incorrectly. Where some aids are numerical from zero to 10, a miscalculation could cause the physics engine to treat a driver's aid as if it was set to a negative value. A traction aid with a negative value would cause some freaky and unpleasant results.

I too keep hoping that this is some sort of bug that they can just fix. The problem is massively obvious in certain RR cars, which leads me to believe that PD put the physics engine together for offline racing, then made some tweaks for online mode, but failed to see if those tweaks modified the handling of all cars properly. The fact that certain tracks exhibit this problem more than others is also terribly frustrating because it means that tuning some of these cars for consistent online handling is basically impossible. I've played around with a huge number of cars and tracks trying to work through this, and one thing I've found is that for some reason PD has most of the RR cars modeled so that their rear traction breaks free under throttle much more easily than an FR or MR car of similar (or even greater) power. This is exactly the opposite of their real-life behavior, where the massive rear-traction of RR cars makes it more difficult to break their traction loose under power. Combined with the bizarre levels of snap-back and you end up with a complete mess (the most glaring example is on the cape-ring ramp where the Yellowbird can't even manage a moderate-speed steady-state sweeper).

I also wonder if PD didn't do something utterly wonky with the physics to make these cars "driftable" online (in the modern sense of exhibition-style drifting). If they made these cars handle like they do in real life, then it would be nearly impossible for people to play "drifty" online with them; their handling characteristics just aren't ideal for exhibition-drifting at all. They drift their own way, and when done correctly it's primarily for speed, not for style. Some people do exhibition-drift them, but man, I can't imagine how (my guess is that most of them do it with some extreme suspension mods).

Some other cars also suffer quite badly online, especially on the notably bad tracks, but so far nothing I've found becomes as nerfed as the RR Alpines and RUFs.

What it does remind me a lot of is ice-racing. When you are racing on ice with non-studded tires (not snow, ICE), it is essentially impossible to use weight-shifting. Squeezing the throttle or braking or letting off on the throttle just breaks traction between the tires and the ice: almost no actual weight-shifting occurs. So when ice-racing, you can't rely on a little bit of throttle to get the rear settled. Hitting the throttle at all is just more likely to spin you further. At that point all you have at your disposal is counter-steering to bring you back in-line. This is what's happening online to RR cars, even when they are on tarmac!

The best solution I've found so far is to run mixed-tires. I've got a very short thread over in the tuning forums about this, so I'll probably post the details there if I feel I've ever really nailed them. But the basics of it are:

Put 1 level harder tire on the front, lets say SM front and SS rear. That makes the car too stable however. So now you have to "re-loosen" the rear of the car until it is where you want it to be. There are a number of ways to do this with either the racing-suspension, the LSD, the brake balance, or best a combination of the three. This seems to work reasonably well. Far from ideal, but at least it is a good starting point. Also, the softer tire at the rear helps recover from those great gob-smacking amounts of snap-back you'll get.

Then, when you race that same car offline don't run mixed tires and you should be OK there too. You'll be a bit twitchier that before, but that's OK (these cars in offline handling mode were probably a bit too stable, at least they feel just slightly too stable to me...)
 
Last edited:
Offline a midget on racing tires with a fully raised suspension will tip over around high speed turns making for exciting racing, I was hoping to be able to have some fun midget races online with a buddy, but online its impossible to reach any kind of angle on two wheels, its extremely annoying, especially after I convinced my friend to spend 200k+ on a midget to have it not work
 
found this thread after running the lotus 111R rm last night at laguna seca. the same car offline handles on rails. online, corner speed is way down and the back end is much looser. i wish it was the same online/offline.
 
Just read the whole thread. And came up with one word : appalling (addressed to PD).

I recently published my first tune, and experienced the difference of online physics. During my tests, and reading this, I came up with a few WAGs (Wild A** Guesses).

The online physics ARE different. Cars are loose in the back, but not only. I will need to test this further, but the added weight in the boot seems to be moving around. Get your entry speed correct, and just when you should be able to floor it, your car is nudged further towards the outside wall.

Of course, tire wear & temp is accounted for, as well as the fuel weight (and momentum in the tank but gooeyfied). The first thing you learn when you drive container trucks (like firemen pumps) is : never drive with the container which is not 100% or 0% full.

Stock settings are correct offline, for a dry weight car. Online, the car gains immediately a poorly secured depleted uranium bowling ball in the back. Which means the settings become instantly wrong.

The solution could be to tune the car with 80 to 100 Kg more on the rear axle. Accounting for a fuel weight closer to 100 Kg would provide enough resistance to withstand the momentum. Just like you need to account for the dynamic weight transfers when setting up your springs rating.

I was tickled by the track bits that have a different effect on your car on and offline. When I was playing GT5P, mainly on Suzuka, I had to come to grips with weird behavior on several parts of the track. Namely the exit of the bank curve, the long right hander after the hairpin, and the last corner, after the casio chicane.

On these parts, my car would just lose grip entirely. It was my fault. Because I had set up my car for race tires, but downgraded to sport ones for an event without altering the core of the setup. What my old setup would have negotiated without a hitch, the new one gave up on. Because my suspension was asking my sport tires to do a race tires' job, and I had to go slower than I could have if the setup was made for sport tires.

That was a bit long but I'll get to my point on tricky track bits. Offline, your tune is correct (even the stock ones) and you can drive on those bits all day everyday without noticing any difference in track surface, banking or elevation change. Online, however, your tune is dead wrong and those parts will make sure you notice them, and generally introduce you to their concrete mate they call Wally.

As I said, those are only WAGs. I will put those to the test and report.
 
Last edited:
When I'm online and my car has fresh tires and less than 10L of fuel, it performs exactly the same as it does with a fresh tires and a full tank offline.

/thread, imo.
 
I will put those to the test and report.

It's nice to see that people are still working on this. I wouldn't spend too many hours on it though, simply because there isn't any way to fix it. We are at PD's mercy on this one. One thing you'll find is that the problems vary so extremely from track-to-track and car-to-car that you probably won't be able to find any consistency anyway. That's why some people have no problems at all and others have severe issues (try the Alpine 1600s on Trial Mountain offline and online. It's ridiculous enough to make you laugh, or cry, or both!) I've just slowly learned to avoid certain tracks completely online, especially if I'm in one of the cars which is most dramatically changed. Overall I like the online handling model better. It is much more dynamic in many of the cars. But get the wrong car on the wrong track and suddenly things go completely and totally screwball. I've yet to figure out why PD thinks RR cars break their rear-tires loose more easily than FR cars. They obviously don't really think this, because offline these cars remain fairly convincing. But they've really buggered something up there in online mode (I still think they may have done it on purpose to try to make the RR cars "drift-able" in the modern-sense of the term). At any rate it's got to be something that PD can fix if they just put their minds to it.

(It still makes zero sense to me that they would do this at all. Why in the world did they think people would want to learn 2 different ways to drive the exact same car? It's just such a bizarre situation...)
 
Last edited:
Just thinking, could it be possible that in online races, in addition to fuel consumption and load and tire temperatures, road condition differences are made more apparent? On race tracks the optimal driving line should have more grip, but on street circuits (which in general should have less grip) it should be often the opposite as traffic consumes and makes smoother the road over the years. This could explain the differences found in several tracks (however I haven't played GT5 in weeks, so I have not verified this yet).
 
Back