I'm going to fire up a second cup of coffee and read this again. Excellent work, but my brain just isn't alert yet.
I did have to quote this one just to ask you, where the hell did you get my kid's kindergarten drawings from? lol
Update 1.16 Physics Changes
- Thread starter Johnnypenso
- 222 comments
- 20,110 views
I did have to quote this one just to ask you, where the hell did you get my kid's kindergarten drawings from? lol
- 8,715
I'm not so sure there are damper changes, that should have appeared in the FFT plots, if I was right about them (and I'd have concluded the setups were different).
However, since the dampers deal with body movement (or, rather, its speed relative to each wheel), any change in that movement will work the dampers differently from 1.15 in the same situations.
Different people keep using different language to describe what must be the same underlying change.
What is still troubling is that not everyone perceives it. Is it isolated to certain cars, driving styles or even cognitive / sensory "types"? Some combination? Wasn't the MR fix equally contentious?
I know some people can't perceive the difference between 30 and 60 Hz graphics, I wonder if that's related.
I left it in its full glory really to show extents.
However, since the dampers deal with body movement (or, rather, its speed relative to each wheel), any change in that movement will work the dampers differently from 1.15 in the same situations.
Different people keep using different language to describe what must be the same underlying change.
What is still troubling is that not everyone perceives it. Is it isolated to certain cars, driving styles or even cognitive / sensory "types"? Some combination? Wasn't the MR fix equally contentious?
I know some people can't perceive the difference between 30 and 60 Hz graphics, I wonder if that's related.
Haha, I thought the same thing the first time I came across these kinds of telemetry plots. They make much more sense in motion, and if you window the playback to just a corner or two.
I left it in its full glory really to show extents.
- 7,765
- Yorkshire
- ScapeGoat4U
- WhippingBoy
- 2,643
- Finland
- OdeFinn
@Griffith500, thanks for describing things in charts for people "picture it, or it doesn't exist". Overall conclusion was same what I felt on first run after 1.16 update.
Thanks.
Thanks.
- 2,799
- Porto
- Burnout_PT
Drove a couple races in the weekend, and knowing what I was "looking" for I drove in different styles and cars and I see it everywhere now. But the more powerful/faster/harder the car, the harder it is to feel it. Still managed to do a proper race driving "normally" a Toyota TS030 and feel the better handling, but like I said, now I know it's there it's easier to spot.
Thing is, I can see it now in the replays, I see understeer, rotation and oversteer and it's just like I remember taking that corner in the race I just finished. The car looks to move more (suspension?) and you see the body moving with the driver inputs and weight transfer. It's brilliant to watch
I think you nailed it. Before we were fighting different forces in different moments, now it moves as a whole and any car turns in easier, and just rolls around the corner much better. The lighter/smoother/less grippy it is, the easier it can be felt, and the harder you drive it the more it shows, and the more rewarding it gets.
Except maybe FWD cars... Still can't get that liftoff oversteer to work properly.
- 8,715
I felt the same thing with FWD cars; I think it's the differential modeling under deceleration that is at fault. You can be pretty careless with rev matching and the penalty isn't as great as it should be with an open diff.
People may think that looking for the change is inviting confirmation bias (and that is true if you're not careful to be wise to it), but this is exactly what happened with the MR fix. Every update, people tested the "problematic" MR cars for changes, i.e. they were looking for changes they desired. The thing is, their looking for those changes did not stop those changes actually being there. Of course, often there were no changes (according to those of authority). This newer "fix" was neither strictly needed (/ desired) nor anticipated, so comparatively few people were looking for it before they noticed it.
I think that over-damping the suspension would mask the effect we're describing, and GT has always suffered from horrendous over-damping until GT6 - well, it's a marked improvement at least, but there's still room for improvement. This is in stark contrast to the MR problem, which was fundamentally a balance issue (might have been linked to the aftermarket wheel and pitstop bugs in hindsight), and was somewhat more in your face the first time it was changed. The second (argued) change was more subtle, related to perceived ease of control, especially with the "snap oversteer" that was rife at the time...
Incidentally, an over-damped state can be very easily reached by reducing the spring rate k (spring displaces further for given load, does so much faster -> higher damping force), reducing mass m (damping force dominates the mass' inertial force) or increasing the damper resistance c (obvious).
In addition, a stiffer (sprung) suspension does not allow the body to move as much in the first place; even though it technically contributes to a reduction in damping, the resulting movement is harder to perceive. In reality, the dampers are typically strengthened with the springs, compounding the reduction in perception.
People may think that looking for the change is inviting confirmation bias (and that is true if you're not careful to be wise to it), but this is exactly what happened with the MR fix. Every update, people tested the "problematic" MR cars for changes, i.e. they were looking for changes they desired. The thing is, their looking for those changes did not stop those changes actually being there. Of course, often there were no changes (according to those of authority). This newer "fix" was neither strictly needed (/ desired) nor anticipated, so comparatively few people were looking for it before they noticed it.
I think that over-damping the suspension would mask the effect we're describing, and GT has always suffered from horrendous over-damping until GT6 - well, it's a marked improvement at least, but there's still room for improvement. This is in stark contrast to the MR problem, which was fundamentally a balance issue (might have been linked to the aftermarket wheel and pitstop bugs in hindsight), and was somewhat more in your face the first time it was changed. The second (argued) change was more subtle, related to perceived ease of control, especially with the "snap oversteer" that was rife at the time...
Incidentally, an over-damped state can be very easily reached by reducing the spring rate k (spring displaces further for given load, does so much faster -> higher damping force), reducing mass m (damping force dominates the mass' inertial force) or increasing the damper resistance c (obvious).
In addition, a stiffer (sprung) suspension does not allow the body to move as much in the first place; even though it technically contributes to a reduction in damping, the resulting movement is harder to perceive. In reality, the dampers are typically strengthened with the springs, compounding the reduction in perception.
NCRthree
Premium
- 1,797
- Virginia
- ncrthree/GTP_ncrthree
Speaking of MR... I had driven the Premium Honda NSX a few times on CM tires/stock and found it a handful on lift-off. I think I ended up using CS just to stay on course. This was after the big MR fix. Yesterday I took it out and found a profound change in it's disposition. High speed lift-off instability was still present but in a more correctable fashion. I had a blast pushing it harder and harder through the esses at Suzuka. I was on CM's too.
Griffith I relate to how you drive in GT. I mainly enjoy the immersion factor, driving feel, intuitive ffb etc... Since PD tweaked the ffb and sound displacement recently, my enjoyment of GT6 has increased dramatically.
Griffith I relate to how you drive in GT. I mainly enjoy the immersion factor, driving feel, intuitive ffb etc... Since PD tweaked the ffb and sound displacement recently, my enjoyment of GT6 has increased dramatically.
- 3,383
- Some where near the Pub
- GTP_hasslemoff
All I can put it to is placebo effect, two replays I looked at had no difference with yaw, the car felt no different on the same combination. The car is drove on its limit in time trial mode with real track settings offline and as people say no aids except abs 1.
Mazda Roadster TC, SS tyres and no tunning.
Matterhorn Riffelse
Pre 1.15 = 1'20.250
1.16 = 1'20.366 (time lost last two corners)
Replay's will be made available to upload.
Underneath is the guide (red = 1.16 / black = 1.15)
I like to keep things simple so there should be no big words and not as technical
.This chart is the yaw for the 2 laps and as you can see they run very close together, there is slight variations and should be as you are not going to get a 100% equal lap so they should be slight variations. One thing to note is the rise in the black (1.15) line after 1min 10 sec this was down to my input as I made a slight mistake so this has an effect on the yaw.
The matterhorn riffelse track's first corner is a hide speed corner, with the apex been on the top of the rise so this should of shown a definitive answer I would of thought, but it shows minor differences which is expected.
This chart is the steering angle, again as you can see they again are very similar, around the 50 second mark there is what looks like a large difference, this again was down to me.
What im trying to say is the effect on the yaw and steering input was down to myself, all the slight errors are highlighted significantly when you start to review in close view.
If you have watched the two replays that were reviewed earlier there is a few errors which have contributed to the effect of yaw and steering input over the lap. Two replays I watched and viewed on the data log
There is a mistake for instance on corner 7 where the braking was late which in turn caused the car to understeer, before this point he was .3xx ahead of his 1.16 lap time and if the driver had braked at around the same time then that corner would of been very similar in inputs as 1.16.
To me overall the driver's input has had more of an effect then any subtle changes to the game, even when reviewing my own replays the differences have been when I have done something slightly different even been a few KPH out going into a corner produces more steering lock as you try to hit the apex.
If that was the case then most people would notice it I would assume ?
If it put the power down more predictably would it not increase the lap time as you can get on the power earlier ?
I did think you had touched onto something even though I cannot feel any difference. Then I sat down and watched mine and the uploaded replays and then looked at the data while comparing them to the replay's and its made me more unsure as driver errors have had more of an effect.
- 2,643
- Finland
- OdeFinn
@hasslemoff if you have any replay from 1.15 where you drive without ABS go and pick it and do new similar with 1.16 without ABS.
ABS makes people blind.
ABS makes people blind.
- 3,383
- Some where near the Pub
- GTP_hasslemoff
.
- 28,470
- Windsor, Ontario, Canada
- Johnnypenso
- 8,715
My argument was based entirely around the change in input. Do you not adapt your driving style to a change in response to the thing you're trying to control? This is all anyone is arguing: the cars feel different to control.All I can put it to is placebo effect, two replays I looked at had no difference with yaw, the car felt no different on the same combination. The car is drove on its limit in time trial mode with real track settings offline and as people say no aids except abs 1.
Mazda Roadster TC, SS tyres and no tunning.
Matterhorn Riffelse
Pre 1.15 = 1'20.250
1.16 = 1'20.366 (time lost last two corners)
Replay's will be made available to upload.
Underneath is the guide (red = 1.16 / black = 1.15)
View attachment 320573
I like to keep things simple so there should be no big words and not as technical
This chart is the yaw for the 2 laps and as you can see they run very close together, there is slight variations and should be as you are not going to get a 100% equal lap so they should be slight variations. One thing to note is the rise in the black (1.15) line after 1min 10 sec this was down to my input as I made a slight mistake so this has an effect on the yaw.
The matterhorn riffelse track's first corner is a hide speed corner, with the apex been on the top of the rise so this should of shown a definitive answer I would of thought, but it shows minor differences which is expected.
View attachment 320576
This chart is the steering angle, again as you can see they again are very similar, around the 50 second mark there is what looks like a large difference, this again was down to me.
View attachment 320619
What im trying to say is the effect on the yaw and steering input was down to myself, all the slight errors are highlighted significantly when you start to review in close view.
If you have watched the two replays that were reviewed earlier there is a few errors which have contributed to the effect of yaw and steering input over the lap. Two replays I watched and viewed on the data log
There is a mistake for instance on corner 7 where the braking was late which in turn caused the car to understeer, before this point he was .3xx ahead of his 1.16 lap time and if the driver had braked at around the same time then that corner would of been very similar in inputs as 1.16.
To me overall the driver's input has had more of an effect then any subtle changes to the game, even when reviewing my own replays the differences have been when I have done something slightly different even been a few KPH out going into a corner produces more steering lock as you try to hit the apex.
If that was the case then most people would notice it I would assume ?
If it put the power down more predictably would it not increase the lap time as you can get on the power earlier ?
I did think you had touched onto something even though I cannot feel any difference. Then I sat down and watched mine and the uploaded replays and then looked at the data while comparing them to the replay's and its made me more unsure as driver errors have had more of an effect.
Whether that's due to some trick of the screen "kinematics" or an actual underlying "physics change" is irrelevant and ultimately indeterminable. If the game is simply more enjoyable due to the change, which results in real changes in the way players interact with it, then this is indistinguishable from "actual physics changes" in its outcome. If laptimes are the only thing used to differentiate the two ("tricks" and "actual physics"), there is a great deal being overlooked. The player-game feedback loop is fundamental and final.
The mistake at the last hairpin has no effect on the yaw consideration, it's understeer: the steering and yaw rate are both relatively steady - in fact, it makes that corner entry look stable. Why do mistakes occur? What about how much faster the entry to the last corner was, with heavy trail braking (on CM tyres!)?
The first corner is a great point of comparison. The braking was much sooner in 1.16, but so was the downshift and the trail braking deeper, and yet the steering inputs are less erratic. The attempt at trail braking in 1.15 resulted in over-rotation. In fact, I showed that braking was generally initiated at greater yaw angles in 1.16 than 1.15. Why would that be? EDIT: look at the brake pressure histogram, also.
Notice that I didnt just rely on an assumed sense of intuition for the time-domain curves (however badly scaled in your case), I looked far deeper into it at the things a typical driver is not likely to have control over when matching lap times. Specifically, the spectral distribution of their control inputs, which you have neglected to include for your laps.
At the end of the day, these are just two laps, but they somewhat miraculously (it seems) show the effect that many have felt.
I'm a little put-out that you didn't bother to upload the Motec exports you're clearly working with, so that I could easily look for myself, and that you've been weirdly selective with your windowing / scaling on the steering graph posted. You're making me jump through hoops to see the points you ought to have properly illustrated yourself.
Not if ABS masks the effect. An obvious example being brake bias.
Last edited:
- 11,660
- GTP_Orido
The brake assist do mask a lot of things and the nature of the assist itself ( preventing brake lock up not by simulating ABS system ) means almost perfect deacceleration within physics calculation boundaries using the pre-programmed assist routines.
When braking without assist, the driver is interacting directly with the physics ( maintaining the hair line limit of tire traction without any interference ) and brake balance determine how brake force are distributed and applied to each axle. With interference from ABS ( brake assist ), the balance and force applied are being manipulated by the game assist routines. This is the reason why driver can brake safely with 1/10 BB on ABS 1 and have good weight transfer. Try do the same with ABS 0, and have fun spinning before the actual weight transfer completed.
So, what I'm trying to say is with ABS assist enabled, the rate of the weight transfer is always regulated by the game in ideal / close to perfect manner, in order to prevent lock up, while with the assist disabled, it is up to the driver to manipulate how much weight is transferred and how quickly ( by using front bias BB and higher BB + brake modulation ) Driving without ABS gives much more feedback to the driver about how well the car react to the weight being shifted to the front under braking. If the front axle can't take too much sudden weight ( due to spring or damper setup or car weight distribution ), a lower brake balance will be needed and the driver will have to be smoother and apply less brake input than possible. Enable the ABS, and this issue will be lessened even with different BB used )
I can never put a better words I recently took my KTM and Viper T/A replica on several lap run, and they are better car to drive now in terms of their feel when reacting to my input and road conditions.
When braking without assist, the driver is interacting directly with the physics ( maintaining the hair line limit of tire traction without any interference ) and brake balance determine how brake force are distributed and applied to each axle. With interference from ABS ( brake assist ), the balance and force applied are being manipulated by the game assist routines. This is the reason why driver can brake safely with 1/10 BB on ABS 1 and have good weight transfer. Try do the same with ABS 0, and have fun spinning before the actual weight transfer completed.
So, what I'm trying to say is with ABS assist enabled, the rate of the weight transfer is always regulated by the game in ideal / close to perfect manner, in order to prevent lock up, while with the assist disabled, it is up to the driver to manipulate how much weight is transferred and how quickly ( by using front bias BB and higher BB + brake modulation ) Driving without ABS gives much more feedback to the driver about how well the car react to the weight being shifted to the front under braking. If the front axle can't take too much sudden weight ( due to spring or damper setup or car weight distribution ), a lower brake balance will be needed and the driver will have to be smoother and apply less brake input than possible. Enable the ABS, and this issue will be lessened even with different BB used )
I can never put a better words
I recently took my KTM and Viper T/A replica on several lap run, and they are better car to drive now in terms of their feel when reacting to my input and road conditions.
Last edited:
- 2,643
- Finland
- OdeFinn
Yes it should, but wouldn't.
- 2,428
- England
- Lewis_Hamilton_
You guys are getting deep and technical into this, meanwhile I'm sitting here not feeling any differences ¯\_(ツ)_/¯
Did I die?
Did I die?
- 216
I can definetely feel something is different - I just can't put my finger on it!
- 8,715
You're welcome to contribute your own "unbiased" analysis.
This is the point of these forums, have at it.
We'll have three more laps to play with once @hasslemoff uploads the Motec exports, or when I get chance to copy the replays to my PS3, run the game, and export the Motec data for myself.
It would be a lot of work. If you wanted to be sure, get at least 20 people to send a before and after update lap time using the same track, car and aids and driver, this will reduce noise. The times for a person should be within a couple of tenths and half the post-update times should be slower than those people's pre-update times to reduce practice effect bias.
Now youve got 20 data points pre-update and post-update that you can do analysis on the sample with any variable you choose. But you need to do your analysis as repeated measures(stats software can do this) which means comparing each individuals pre-update to their post-update, this will remove individual differences bias. The statistical software will give you an R, the r is the percentage that you are observing is a real effect and not chance(noise).
From looking at the graphs in this topic, the amplitude of the noise is just as high as the percieved effect in most of those graphs. So Im doubtful you will find anything statisically significant. If indeed there is a physics change, the data will be to noisy to find it.
You can see now that it really is a lot of work and why I suggest you send an email instead to polyphony digital. You never know, they might reply.
recommendation for future updates:
------------------------------------------
If you collect data now from bspec using bob you will have a baseline for the next update and then you can test for any physics changes in the next update without any error or bias coming from involving a human driver in the data collection. But get a good sample size of laps from different tracks and cars because other confounding variables can creep in.
Last edited:
- 8,715
Yes, well it's a bit late for that, and now people know what we're looking for, they can attempt to skew the results easily one way or the other, especially with so few participants. We need at least 100.
It's something of a cop-out response, though; this is what we have to work with, and you claim the analysis is biased, but don't actually substantiate that in any way. If you're not capable of the analysis yourself, fine, but be a little more careful in your assumptions.
I know how confidence intervals work, but I also know how to analyse complex systems - I could design any number of experiments that would give a direct answer; we're not dealing with an inherently chaotic / "random" system, rather a pseudo-physical, purely deterministic one.
You could also just look into the game files, each update is its own changelog (much more reliable than PR).
But I'm not interested in reverting my game to 1.15, plus I'm supposedly biased.
How will Bob be able to show differences in the player-game feedback loop? Those changes are the most important kind of change. Collect the data by all means (good luck anticipating the kind of data you need to cover the spectrum of possible changes).
Yes, well it's a bit late for that, and now people know what we're looking for, they can attempt to skew the results easily one way or the other, especially with so few participants. We need at least 100.
It's something of a cop-out response, though; this is what we have to work with, and you claim the analysis is biased, but don't actually substantiate that in any way. If you're not capable of the analysis yourself, fine, but be a little more careful in your assumptions.
I know how confidence intervals work, but I also know how to analyse complex systems - I could design any number of experiments that would give a direct answer; we're not dealing with an inherently chaotic / "random" system, rather a pseudo-physical, purely deterministic one.
You could also just look into the game files, each update is its own changelog (much more reliable than PR).
But I'm not interested in reverting my game to 1.15, plus I'm supposedly biased.
How will Bob be able to show differences in the player-game feedback loop? Those changes are the most important kind of change. Collect the data by all means (good luck anticipating the kind of data you need to cover the spectrum of possible changes).
I am NOT going to get into an argument with you or anyone else in this topic. Just forget I said anything and ignore me.
- 8,715
Everytime I say something online I am forced to defend my position. People constantly tell me why Im wrong. I learnt research methods during doctoral training. Surely Im not wrong? People just like to argue on the internet. I dont like arguing, I like helping people.
- 2,643
- Finland
- OdeFinn
- 2,799
- Porto
- Burnout_PT
These things can be highly subjective, especially cause it's ultimately a game. Many people just play the game, adapt to the "rules" (physics model etc) and just have fun.
There's another bunch of other people who actually love the simulation, and like to see that seat of the pants driving from real life accurately replicated in game, but that doesn't mean everyone will or has to feel it.
I think this analysis is great to try and find out why or how the handling has changed, and I really enjoyed all of this in the previous posts/pages, but bottom line is:
- If you can't feel and believe it's not there, then great. Carry on playing as you were.
- If you do, great too, much better driving experience for you.
Anyhow, last night I remembered one thing I hadn't tried, drifting.
Now, I could not drift in GT6 if my life depended on it. Doing the challenges was a pain, and I'd mostly drop the G25 and use a controller, struggling to get to that gold in a highly tuned and "perfect" drifting car. Last night I took to 2 challenges I had left and got gold on 2nd attempt, just after getting used to the track on the first run. Using the wheel, the first car I came across, no settings teak, no extra hard tires.
I could not see how people could do those brutal scores before, something in the handling was making me screw it up all the time (my perception was skewed and my inputs wrong) and I couldn't understand what. Now? Now I know... I can feel exactly what I do, where I was wrong and how to improve and after re-running one of those challenges a few times, I got to double the score for gold. No setting changes, no change of tires, no riding the inside of the corner to get maximum points (what's with the 0 points when going more than a certain angle, or ridin the outside with the back close to a wall? that's quite the opposite of real drifting scoring! anyway, another argument for another thread). It was just pure fun drifting. Can't wait to get a proper car set up and go at it again.
So anyway, maybe drifting will provide better data for analysis?
