Update 1.16 Physics Changes

[hasslemoff]

The lateral G Force is different in both versions, and the oscillations at the end could point to a difference in physics. But the difference is small and the oscillations at the end could be just a fluke. To rule out the small input differences, i would like to ask @hasslemoff to redo the X-Bow test on both versions 2 or 3 times, just to be sure how big the differences are for multiple runs for a single version?

Thanks for the testing @hasslemoff!

EDIT: corrected some misspellings

Ill try and get some more done, im thinking another track might be better and give more of a consistent result, maybe indi speed way with its wide start and a wall to hit, ill get some with full throttle and without as well.
If the DS3 is still showing interference ill try and get time to test on the wheel as it means moving a bit of stuff to move the ps3's .

Yes it was on auto transmission so there was no interference from me.

 

[tarnheld]

If the DS3 is still showing interference ill try and get time to test on the wheel as it means moving a bit of stuff to move the ps3's .

No need to hurry, i can not access a Windows computer for i2 over the weekend. I will try to add some Indy replays on 1.16 myself.

If someone in the community is still on 1.15, it would be cool if you could do the @Bhowe83-method on Indy 2-3 times, preferably with a wheel, but DS3 is ok too. 👍

 

[hasslemoff]

No need to hurry, i can not access a Windows computer for i2 over the weekend. I will try to add some Indy replays on 1.16 myself.

If someone in the community is still on 1.15, it would be cool if you could do the @Bhowe83-method on Indy 2-3 times, preferably with a wheel, but DS3 is ok too. 👍

Went back to motegi as there was not much difference in size and as a ive already got some can use them as well .

 

[Lewis_Hamilton_]

Just in case it makes a difference, make sure that pre and post 1.16 tests are done with the same controller steering sensitivity setting. I've only been skimming through the thread but I haven't seen anybody taking it into consideration.

 

[hasslemoff]

Just in case it makes a difference, make sure that pre and post 1.16 tests are done with the same controller steering sensitivity setting. I've only been skimming through the thread but I haven't seen anybody taking it into consideration.

Checked that they were both on 0 sensativity also the same controller, cheers for pointing that out as I never put it in the post .

 

[hasslemoff]

I hope all these help , ill get to looking at them over the weekend.
There only 10 sec replays on Motegi oval, some of the replays would not transfer to motec saying there was already a copy on the usb stick even switching them did not help. So a couple are missing, dont know which ones though.

I did 4 replays for each version 0 and 1.16 with each car in the previous post.

• Two replays applying left and accelerator in each version
• Two replays applying left only in each version

4 replays are missing though due to above problem.

The name T.Test is version 0's data. Sorry you have to download them all to see the relevant information I got a bit sick of hitting buttons to change the titles.

Again all the same settings as previous with the same cars but on Motegi oval and instead of running the full lap I quit not long after hitting the wall. Barring 4 replays from the other day which I ran the full lap.

 

[Haitauer]

Hi! Cant really say i have followed the conversation, But i did notice something from the visual point of view.

Havent played gt6 for A short while and returned to driving A very familiar car/setup/track combination for A photoshoot. I did notice the car felt Somehow more logical - more balanced in A realistic Way ( meaning not faster But more realistic feel). Also felt like the car felt A Lot more natural while drifting. ( not a drifting Expert though)

But them I fired up the replay and noticed A Lot more body roll - especially car nodding under breaking. That was an instant "what?!" Kind of observation at the very first corner of the replay. Then i Opened the photo mode at A moment when the car was drifting, and noticed a very natural looking body roll. And to be honest I have watched A Lot of replays With this car/setup.

Might be nothing - But that was just my gutfeel. And one should also note that i made those observation before i had any knowlege of the possible changes in 1.16.

 

[Mike_grpA]

For the last month and a half I've been only driving lower powered stock street cars on CH's offline. No aids. A couple days before the update I was running a Scion FRS at Tsukuba. The first combo I used after updating was this. I noticed nothing different, my lap times were exactly the same. I was really trying to pay attention to the car's dynamics but it felt the same to me. A BMW 2002 felt the same too.

I'm usually pretty sensitive to any new physics changes so I was surprised that it felt the same to me. I always drive in a "realistic" manner, no flicking or overdriving the physics engine so that may be the reason.

This. I have tested the same car, same exact set up, same track for ages before the update, and ever since the update. I felt no difference, and didn't expect any difference. I read everyone talking about it in the undocumented changes thread, and it surprised the hell out of me. I would have noticed any changes, as I spent ages tuning that car for my tastes for an upcoming series, which is about to start. If anything changed at all, I would need to change the tune.

I also don't believe polyphony would go to the effort of making physics changes that, as some claim, make the game feel a lot more realistic, without telling everyone about it. I would imagine the first physics change they'd make would be fixing the camber and suspension model lol.

 

[Rotorist]

A great change has happened in the way the vehicle reacts under heavy braking and especially when on the limit to brake the grip or just over it. Understeering was VERY profound before 1.16 but now it is much more responsive to steering than before. So, the change isn't so much about grip levels but most in the way the grip is gained again much faster and allows you to save the day. That is why the best laptimes aren't so good in proving the changes. Weight tranfer, suspension response, grip recovery should be what changed in physics imo.

 

[jimantonic]

My two cents worth:

Apologies but I've not had time to read past page two of this thread.

I too can feel a difference in physics which would suggest a change to weight behaviour.

A recent club race set at Silverstone GP using the Lotus Europa on CS tyres, the Lotus Carlton on SH tyres and the Elise 111R Race Car on RH tyres, all stock except oil change and BB adjustment.

Although I didn't have time to amass a great deal of mileage on each of the cars, I certainly noticed a difference, more notibly on both the MR cars.

Pre update the Europa was very easy to drive, trail braking was no issue at all, you could downshift mid turn with no problems.
Post update and in the very slow speed turns it would over rotate more than before, trail braking became an issue to the extent I dropped the BB from 5/5 to 4/2, and if you were to down shift whilst mid turn/just after turn in the rear would step out dramatically.

Pre update the Elise had pinpoint accurate grip, went wherever you wanted it to and stayed there.
Post update I noticed a big increase in understeer, and also a little more edgy mid corner but not nearly as pronounced as the Europa.

The Carlton on the other hand felt like piloting a river barge both before and after the update.

 

[Mike_grpA]

So of the people claiming a big change to physics, we have some saying it's made the bumps more noticeable, some sayinig it's removed understeer, some saying it's made mr cars easier to drive, some saying it's made mr cars harder to drive, some saying it's made body roll more pronounced, and probably a lot more claims that I just haven't bothered to read.

I know what could cause all of these things to happen. Placebo. If Polyphony changed the physics, especially something so big, they would have: a) fixed well known issues with the physics engine, which they haven't, and b) told everyone about it, since it's not like it would be an insignificant amount of work, they would want people to know how hard they're working.

/my2c

 

[Griffith500]

So of the people claiming a big change to physics, we have some saying it's made the bumps more noticeable, some sayinig it's removed understeer, some saying it's made mr cars easier to drive, some saying it's made mr cars harder to drive, some saying it's made body roll more pronounced, and probably a lot more claims that I just haven't bothered to read.

I know what could cause all of these things to happen. Placebo. If Polyphony changed the physics, especially something so big, they would have: a) fixed well known issues with the physics engine, which they haven't, and b) told everyone about it, since it's not like it would be an insignificant amount of work, they would want people to know how hard they're working.

/my2c

This is assumption borne from ignorance. Don't take offence, be honest: if a change can be made right now, why wait until a harder change has been made before implementing it?

PD have a track record of keeping schtum about even larger changes.


My two penn'orth: driving feel is more important than "correct" camber, roll bar or spring-rate setting adjustment.

One sure fire way to prove if there's been a change to the physics or not is to analyse the update .pkg itself.

 

[Rotorist]

So of the people claiming a big change to physics, we have some saying it's made the bumps more noticeable, some sayinig it's removed understeer, some saying it's made mr cars easier to drive, some saying it's made mr cars harder to drive, some saying it's made body roll more pronounced, and probably a lot more claims that I just haven't bothered to read.

I know what could cause all of these things to happen. Placebo. If Polyphony changed the physics, especially something so big, they would have: a) fixed well known issues with the physics engine, which they haven't, and b) told everyone about it, since it's not like it would be an insignificant amount of work, they would want people to know how hard they're working.

/my2c

Placebo in this case would have been if we waited for others to tell us about a change. Most of us here found out about the change immediatley after the update and didn't even suspect that there would be a change, nor have we any anxiety to check every update for physics changes in GT6. So, your theory is utterly busted, in my case at least.

 

[Ridox2JZGTE]

I also have made changes on some cars, usually higher ARB at the rear where previously a lot lower in 1.15 to contain oversteer. Damper changes also necessary on some cars to make them more planted, example my C7 Z06 Corvette replica, with 654 lb-ft torque and CS tire.

 

[Bhowe83]

I'm not sure if I feel physics changes in weight, but I definitely feel a difference in grip levels from Lap 1 to Lap 2 (Cold tires to fresh tires)

I noticed this right off the bat on the update across the board.

Weight transfer though again is hard to prove as it has close to no effect on lap times when it is changed in small amounts, which is the claim here.

 

[Johnnypenso]

Placebo in this case would have been if we waited for others to tell us about a change. Most of us here found out about the change immediatley after the update and didn't even suspect that there would be a change, nor have we any anxiety to check every update for physics changes in GT6. So, your theory is utterly busted, in my case at least.

There are people convinced after every update about changes to physics, camber, sound, paint, features etc. Much of the time they are proven wrong.

 

[Rotorist]

There are people convinced after every update about changes to physics, camber, sound, paint, features etc. Much of the time they are proven wrong.

I speak mostly for myself and I know about GT from 1st iteration, being in sim racing for many years now and having a wheel too to campare effects and feedback and all. I immediately feeled the weight transfer being more profound after the update. And with 6axis, I could verify the change to less undesteer on or over the limit especially when heavily pressing acceleration control. Not even 1% for the game's physics engine to be the same as in 1.15. Grip levels are the same as they depend on tires only. Tendency for under or oversteer is changed and makes drifting easier and predictable.

 

[tarnheld]

Back analyzing @hasslemoff's Twin Ring Motegi data:

The variance for Howe's method on a single version is really small, except for the G Force measurements, they can vary a lot between runs of the method. I am not quite sure why, but i have no clue what exactly is measured by 'G Force', is it fixed to the car including suspension interaction, or without suspension? As PD has to somehow integrate these values twice to get the trajectory and orientation of the car along the track maybe they don't include suspension and calculate it later for replay rendering? But as they stand they are not a good indicator of physics changes, unlike vehicle/wheel speed, as you will see later.

Here is the comparison of two runs of Howe's method in the XBow-Street '12 , version 1.16, from start of race until hitting a wall:

You see only minute Speed differences, but big lateral G Force difference.
Here are two runs with only stick full left but no throttle, in the FR-S '12, version 1.00:

Here the differences are barely noticeable in the diagrams, including G Forces, even while the Steering input is a little different at the start.

And now a 1.00 vs 1.16 comparison of the FR-S, this gets really interesting (black is 1.00, same as the color diagram above):

You see the steering input has a different ramp as it goes to max. angle in 1.16. The throttle is cut different too, it goes down much smoother in 1.00.

Lateral G Force is also different, but that could be just variances between runs.

But the vehicle and wheel speeds are telling a different story: As throttle is cut, rear wheel speeds go down in 1.00, but up in 1.16. They meet again after throttle is 0%. Then in 1.16 there is auto-blip while downshifting, engine RPM goes up while clutch is pressed, the wheel speeds go down 5 km/h in this moment. Don't know if this is realistic, may a drive-train expert can chime in. In 1.00 no auto-blip, RPM goes up only after shifting, rear wheel speeds are not really affected. The evolve also rather differently without considering the shifting, front wheel speeds and vehicle speed drop much faster in 1.16 than in 1.00.

The wheel speeds and the vehicle speed change in a significant way, and taken together with the comparison of different runs with a single version i would call that a solid evidence for a physics change. 👍 What exactly changed should be discussed by the respective experts, but i leave it at that. Attached you find some more diagrams of other cars tested, they show essentially the same.

For a definite answer of the 1.15 vs 1.16 question we would need replays like that for 1.15, but i doubt that anyone is still on 1.15.

 

[OdeFinn]

@tarnheld, answer why values differ: first run car body was straight, second run it was wrecked/bended. Funny how this Motec stuff will prove one long argued thing to reality, and opens truth of it; yes car will suffer damage from collisions and that changes car "physics values" more understeer and less grip, until body is fixed. Damages will stay until fixed.

To prove above you may go and test same runs again with two new cars, one run per car and you'll get identical values.
Test after body fix and again getting identical values. After wall hits you may have identical values from both cars, probably not, what values hit changes is mystery to me, some cases it feels like it changes vales near to hit area, sometimes not sure. Near hit area means: hit front corner and you'll have modified steering and front grip, hit rear and you'll loose grip easier from rear. Some hits are side dependant, affecting only on other side values.

There is a great mystery now to solve, and troll about, @Johnnypenso I'm inviting you now to start your favorite thing.
Tools to solve my words are in hands of all you, just go and do your loved Motec analysis and answer will pop on front of you.

 

[super_gt]

delete

 

[killerjimbag]

There is one man who can do that:

Good place for him.
I personally don't feel any changes.
I'm assuming if the Motech data works like KW suspension, Yokohama tire model implemented by PD it probably doesn't work properly either.
This will all be laughable come April, when we can drive a real Sim.

 

[Johnnypenso]

There is one man who can do that:

Man?

 

[TDZdave]

We did the CORE N24 the weekend after the update, used the same setup prior to the update in the race, nothing has changed, nothing in the car, the view, the roll, pitch etc, not even the sound, and if there was a change there would have been a noticeable difference in our lap times, people ran their pace that they had done in 3 3 hour races prior and constantly matched their average,

 

[400kphMarker]

i used the two LM Vettes for GT3 and LMGT/GTE , but focusing on the ZO6 The Specs i Used are 480-490HP and 1290KG - Aka Full Ballast in the Car, Before 1.16 it felt like i placed a 200kg Rock in the Middle of my car , it felt wrong like the weight wasnt Pushing down on all 4 wheels , like it was Extremely tailheavy and i was sliding out a little bit on corner exit but now it feels as if the car came with 1290KG it just feels even , my Car doesnt Slide and lose grip anymore and i havent really noticed untilll i ran a Multiclass Race with a Buddy on sarthe , Thank you PD

 

[tarnheld]

@tarnheld, answer why values differ: first run car body was straight, second run it was wrecked/bended.

I'm sure @hasslemoff has used courtesy cars for the tests on both versions. So unless you have insider knowledge that PD made courtesy cars pre-damaged in somewhere from 1.00 to 1.16 i think this explanation lacks something.

a real Sim.

I chuckled at the play of words. Is this the opposite of Virtual Reality?



As the 1.15/1.16 question has piqued my curiosity, i tried to apply @Griffith500's FFT analysis (really good idea! 👍) to @hasslemoff's and @Ridox2JZGTE's replays from 1.15/1.16, only with a twist. I didn't try to get an FFT from the G-Forces of the complete lap, but only from selected parts, where there was a sudden excitation of the system like throttle input or going over curbs.

The FFTs show the frequency response of the system, sometimes showing clear notches where frequencies are damped. If the notch pattern would show significant change, this would be and indicator of physics changes, even without keeping the input completely the same, as the notches are like a signature of the underlying spring-mass system black box.

I tried hard to find any significant signs here, there were a few interesting spots here and there, but in the end i really only found a clear damping in all G Forces at about 6Hz irrespective of driver input, car, track and version. This could be an effect of the the underlying physics computation method. So for now this means: "don't know" for me.

You can PM me if you want to know more details about this. I want to spare you from diagram spam this time.

 

[OdeFinn]

'm sure @hasslemoff has used courtesy cars for the tests on both versions. So unless you have insider knowledge that PD made courtesy cars pre-damaged in somewhere from 1.00 to 1.16 i think this explanation lacks something.

Nice to be sure for someone other than your self.
On seasonal races where you use courtesy car provided by event damage is permanent during time you're in seasonal, as it counts kilometers too. When you exit to seasonal menu, menu where all seasonals are and go back then body is restored.
So how sure you want to be? Are courtesy cars working same, permanent damage until exiting back to main menu? Was run first thing to do on that car or was it already smashed before recorded test?

 

[Griffith500]

I'm sure @hasslemoff has used courtesy cars for the tests on both versions. So unless you have insider knowledge that PD made courtesy cars pre-damaged in somewhere from 1.00 to 1.16 i think this explanation lacks something.



I chuckled at the play of words. Is this the opposite of Virtual Reality?



As the 1.15/1.16 question has piqued my curiosity, i tried to apply @Griffith500's FFT analysis (really good idea! 👍) to @hasslemoff's and @Ridox2JZGTE's replays from 1.15/1.16, only with a twist. I didn't try to get an FFT from the G-Forces of the complete lap, but only from selected parts, where there was a sudden excitation of the system like throttle input or going over curbs.

The FFTs show the frequency response of the system, sometimes showing clear notches where frequencies are damped. If the notch pattern would show significant change, this would be and indicator of physics changes, even without keeping the input completely the same, as the notches are like a signature of the underlying spring-mass system black box.

I tried hard to find any significant signs here, there were a few interesting spots here and there, but in the end i really only found a clear damping in all G Forces at about 6Hz irrespective of driver input, car, track and version. This could be an effect of the the underlying physics computation method. So for now this means: "don't know" for me.

You can PM me if you want to know more details about this. I want to spare you from diagram spam this time.

I think the cars are still the same cars, and that all that has changed is the precision / depth of the inertial modeling (at most - it could still just be a feedback thing). As such, the overall ("average") behaviour will be broadly the same. The peaks and troughs due to the "mechanical system" will be in the same place in the frequency spectrum: they're the same cars. There's more than one way to arrive at the same eigenvalue.

It seems to me that there could be a subtle change in the spatial "degrees of freedom", causing a slight redistribution of the nodes and anti-nodes of the body-suspension system (chiefly in the rotational modes), certainly in terms of their individual amplitude (gain) if not their absolute frequency. Damping comes into that as well, however damping always strongly affects the frequency as well as the gain.

So changes in gain (observed in yaw rate), but not frequency (not observed in yaw rate), are not due to a change in damping alone.
Also, I'd question exactly what it is that a lateral g force FFT would actually be showing. You can technically build yaw without building lateral g; therefore, dynamically speaking, lateral g is a bit misleading (it measures translation parallel to a specific axis, rather than rotation about a different axis, for a start). For steady-state cornering, it's fine - but that's not where a change has been felt.

I'm currently focusing on exploring possibilities for hard evidence.

 

[GTP_CargoRatt]

Man?

Going against your own values/teachings there, aren't you Johnny? Off-topic post with the sole and only purpose being is to take a shot at a banned member. Just wow. 👎

On the subject, I have done a lot more testing and tuning since the update and still haven't seen/felt any changes to physics. Everything still feels the same to me as it did before the update.

 

[tarnheld]

So how sure you want to be? Are courtesy cars working same, permanent damage until exiting back to main menu? Was run first thing to do on that car or was it already smashed before recorded test?

Good questions! 👍 Hmm, let's see ... I have multiple replays from 1.00 that don't show significant difference in all logged data from the beginning to the hitting of a wall. I have multiple replays from 1.16 that don't show significant difference either. I compare any two replays from 1.00 and 1.16 of the same car, and find significant difference in wheel/vehicle speeds. All replays show typical signs at the start, indicating that they where not cut from a longer replay where damage could be applied in laps before logging.

The explanation you describe would require one replay was done with a damaged car, and the other one with a pristine car. If this happened accidentally, wouldn't there be differences in the replays done on one version -- one replay with damaged car and one with pristine car? There are none i could find. And wouldn't i find one combination of versions that don't show differences -- when both cars where pristine? There are none either. That leaves only fraud -- all replays on one version were done with only damaged cars and the all replays of the other only with pristine ones. I don't think that @hasslemoff would do this.

This leads me to the conclusion that the differences found are significant, and can not be explained by car damage.

 

[nickg07]

Due to having no internet i am still on 1.15 but don't get to play as my ps3 is at my parents house. ..