How Shift 2 Really Drives (with videos)

[RomKnight]

I'm not seeing a phone catching up everything on the screen... Its better to save a replay in-game (change the camera view for anyting but TV ) review it frame by frame

Is it possible to that that in-game?

[edit]
I realized that the game might save the video using 24/25 fps which would alter the values...
[/edit]

 

[pgagoober]

The problem with the replay is that it won't have my wheel input. The framerate of the video is not important anyway. It just needs to be fast enough to accurately identify the key moments of input vs game reaction. The onscreen game timer handles the realtime measurements.

 

[Imari]

I've got a DFGT, and it's nowhere near as bad as Pgagoober's video shows. My experience has been much more like JJ72's. I had to tweak the settings a bit, but I used my settings from Shift 1 which I'd got to a point I was pretty happy with and then adjusted a little more.

 

[pgagoober]

Imari. Would it be possible for you to create a similar video to mine? It might help us nail it down further. The more info we have the better our chances at honing in on a workaround if such a thing exists.

 

[BryantFC]

I've got a DFGT, and it's nowhere near as bad as Pgagoober's video shows. My experience has been much more like JJ72's. I had to tweak the settings a bit, but I used my settings from Shift 1 which I'd got to a point I was pretty happy with and then adjusted a little more.

Agreed. I did a little test over here too on a DFGT and my test car was a decently tuned (PI 800) Audi S4. The track i used was Spa. I tested AWD's settings first starting with no assists. This was all done on ELITE mode. My consistent track time was roughly around 2:58 with no assists except for ABS ON. With Assists ON (ABS,TRAC low, STABILITY ON), i was able to do 2:56.

With JJ's settings I clocked consistently 2:54-2:55. This is with assists and without assists, just like the test i did with AWD's settings. I'm not the greatest driver but i was able to drive aggressively and still keep a good line.

Did it all within cockpit mode and the input steering wheel was a bit bothersome since it did look like it lagged about half a second or so. As for feel, both settings felt similar. I honestly really couldn't tell the difference. If i did have to choose, AWD's settings were a bit more comfortable meaning you can still swing the car around and power through the corners but if you give it a bit too much steering, it's over. You also do feel the tires kinda giving out/understeer or w/e you get from the force feedback getting a bit lighter when you take a corner hard.

I also changed the steering lock within the car tuning settings to 30. Not sure if this would help anyone but these are the best results I have gotten from using the two settings i felt were the most useful so far on GTPlanet. Still not up to par to GT5's wheel settings or w/e, but it has finally made the game feel a bit more enjoyable with the wheel. Furthermore, I don't think it would be that much of a difference considering that I'm using a DFGT and they're using a G25. I would've made a video, but i'm too lazy for that lol.

 

[Imari]

Imari. Would it be possible for you to create a similar video to mine? It might help us nail it down further. The more info we have the better our chances at honing in on a workaround if such a thing exists.

I'll see if I can borrow a semi-decent camera off someone. All I've got is a 4 year old phone camera and that won't tell us anything.

I've been trying to take more notice of lag, and some things stuck out to me. First, it feels different between cars. This could just be me, which is why it'd be good to film and measure it. Second, stuff like gear changes and braking seem to be absolutely instant. There's no hesitation at all when I change gear.

I want to believe it's something in the physics system that's designed to be that way. That would suck, but it would mean that correctly tuning a car would probably remove it. I'm reminded of the stereotypical American town car, something like a big Cadillac. In things like that the steering is massively disconnected from the road and it takes ages for the suspension to load up enough to turn the car. We're driving sports cars so we *shouldn't* be seeing any of that, but if they've overexaggerated the suspension effects that may well be what it is.

 

[Delta-64-]

I've made this little video, sorry about the quality and I do hope people will pay attention to how the game can be a lot better than most think when dialed in.

"There seem to be many complaints about shift 2's handling which imo isn't warranted, so I made this video to show that you can have responsive control in Shift 2, while also try to help out those having issues."

thanks weasel

Thanks for the vid. Those settings worked for the DFGT as well. Thank god, it makes the game alot more fun.

 

[pgagoober]

@Imari : Very interesting thought about the suspension introducing additional lag. I just posted a GT5 test in my other thread and the result was .2 seconds faster than S2U. If that gap can be closed with suspension mods in S2U I'm back in the game. Let us know what you come up with!

 

[HeadsoupBob]

^ I wonder if some of that difference is just shift having tyre deformation and the time it actually takes to transfer the steering input to the wheels (as Imari said)? Try an old heavy muscle car and see if the delay is the same or worse maybe?

 

[advan031]

I think there is a major flaw with shift 2's physics or steering wheel input! I don't want to fiddle with a bunch of settings so that cars will handle normally.

While I enjoy playing it because of the Porsche cars...I pretty much dislike everything else.

 

[drumc0de]

^ I wonder if some of that difference is just shift having tyre deformation and the time it actually takes to transfer the steering input to the wheels (as Imari said)? Try an old heavy muscle car and see if the delay is the same or worse maybe?

Doubt it Bob, considering the same is experienced in the modified GTR you drive when first starting your career. A car that should have razor sharp steering response.

 

[Imari]

I think there is a major flaw with shift 2's physics or steering wheel input! I don't want to fiddle with a bunch of settings so that cars will handle normally.

While I enjoy playing it because of the Porsche cars...I pretty much dislike everything else.

The thought is that if it's the physics, it's not a flaw as such, it's just the way the cars are meant to handle. If you take a GT3 car, they're very, very tight to drive. Even most works cars are very responsive.

If it's just a fault in the steering code or something that they've added in artificially, yeah, it's pretty messed up.

We can test whether it's input lag or within the physics system as well, using the telemetry. If the telemetry shows immediate changes but the car takes time to move then at least we know it's not an input problem. As long as the inputs are getting to the physics system in time, we can then work to find out how to finagle the physics system so that the cars respond better.

As with every game, there are good bits and bad bits. It's just whether the good bits are good enough to bother working through the bad bits. I feel that Shift 2 has potential once we solve it's problems. And they do seem solvable at this point.

 

[JJ72]

@Imari : Very interesting thought about the suspension introducing additional lag. I just posted a GT5 test in my other thread and the result was .2 seconds faster than S2U. If that gap can be closed with suspension mods in S2U I'm back in the game. Let us know what you come up with!

interesting, I thought about that too. So i've gone some extra mile and show you the telemetry.

apparent there is lag from the steering to the car responding, but if you look at the steering slider (lower left) and the tire load graph, the input is fact transmitted to the game almost instantly, however it does take some time for the car to react.

In the first part I tried larger steering angles, the lag is quite evident. But when I use smaller adjustments the lag is much smaller, so maybe it is how the tire physics are coded that produce this sort of behavior?

I've also loaded up GT5 for comparison, and it is quite clear that the reaction in GT5 is quicker than in shift 2.

in fact I will go to a race track next week, they have track prepped MX5 for rental and I might just try this test on it too lol.



also included is a full lap on ebisu in a lotus.

 

[Imari]

It turns out that I can make my car do the delayed input thing as well as seen in pgagoobers video earlier, I was just doing it wrong before. However, I found something interesting. Turn telemetry on. Most cars will have some front toe in, so when you're going in a straight line you can see the forces on both the front contact patches pointing inwards. You can play with the steering while watching this, and if you use small movements you can identify the range in the middle of the steering where there's no change in trajectory because of the toe angle. This is what front toe is meant to do, it stabilises the car by requiring larger steering angles to deviate from straight.

You can also see while doing this that there seems to be very little lag between steering inputs, and the reaction of the telemetry. The car doesn't move if you make small steering movements, as it shouldn't. But the telemetry reacts instantly.

Before the tin foil hat brigade points out that they've fudged the telemetry, it'd be a lot hard to make a fudged telemetry system than it would be to just make a decent physics system in the first place. I suspect what we're seeing is a real effect of the physics.

Now, the obvious test after that is to see if toe out makes the car more responsive (as real life physics would lead us to believe). Put your front toe all the way down to zero, a major toe out setting. You'll now see the forces on the front wheels pointing out when you go straight. Do it with the back toe too if you want super responsiveness. Now try and turn. Feels faster, right? It feels faster to me. It's not immediate, but it's not in real life either. It takes time for the wheels to turn, for the suspension to load up, for the tyres to bite and travel. You're never going to get instant response. What we're looking for is a response time that is suitable for a sports car.

So, temporary solution, add toe out. That's the same solution you'd prescribe for a real car which had sluggish handling, which is promising.

 

[CAG_Theory]

I've made this little video, sorry about the quality and I do hope people will pay attention to how the game can be a lot better than most think when dialed in.

"There seem to be many complaints about shift 2's handling which imo isn't warranted, so I made this video to show that you can have responsive control in Shift 2, while also try to help out those having issues."

thanks weasel

Im not impressed.

 

[spapadillion]

That video illustrates the problems w/ input lag that plague the game more than anything else.

 

[Imari]

That video illustrates the problems w/ input lag that plague the game more than anything else.

It's definitely not input lag. The inputs are received by the game instantly. If anything, it's "physics" lag. What now needs to be determined is whether it's an arbitrary amount of lag that they've introduced for whatever reason, or if it's simply a consequence of the physics system.

Remember that all normal road cars also have a certain amount of delay between you turning hard from straight and the car actually veering off too. You need a very stiff race car on very low profile tyres with very specific suspension setups to approach anything like an instantaneous response. It's just that most people don't notice this on the road because a) you rarely find yourself veering sharply off from a straight, and b) you're so familiar with how your car operates that your brain just takes it into account. It's how the car works, so it doesn't seem weird. This seems weird because it's different to GT, which is what people are used to.

It may end up being physically inaccurate, but for the moment I'm simply judging it as an interesting effect until we can nail it down more. Tuning the suspension does seem to help, if you're having problems with it. Toe and caster are the main ones to work with.

 

[JJ72]

That video illustrates the problems w/ input lag that plague the game more than anything else.

I take it you are not the kind of person who sticks to a fixed opinion and just come in for a quick jab.

so what stands out wrong in the vid? I am not the best driver but I do believe I had shown control of my car even when it steps out, and I can place it where I want it. And I don't think I am driving in any unnatural manner, if you watch some motortrend shootout video on willowsprings, their test driver also steers ahead of the car and actively throws it around like I do with the Z06.

If the control is truely that bad, as a sim racer for over a decade myself I would have noticed straight away and certain won't cope with it in a matter of days.

 

[HeadsoupBob]

Remember that all normal road cars also have a certain amount of delay between you turning hard from straight and the car actually veering off too

Exactly! I cannot believe how many people think response should be instantaneous in a production car!

Which reminds me of another point. Watch a replay of GT5 from the cockpit and watch the suspension response, it responds instantaneously to the bumps and curbs, no dampers, just up and down. Then watch Shift 2, cars seem to 'float' because the suspension absorbs the bumps. I never noticed it until I watched the comparison video on Nords with Shift 2/Forza 3/GT5, then it became VERY evident. Not to mention the track and horizon bump around in GT5 cockpit view instead of staying steady while the car moves around, like the camera is attached to the chassis instead of the driver...

 

[Imari]

Which reminds me of another point. Watch a replay of GT5 from the cockpit and watch the suspension response, it responds instantaneously to the bumps and curbs, no dampers, just up and down. Then watch Shift 2, cars seem to 'float' because the suspension absorbs the bumps. I never noticed it until I watched the comparison video on Nords with Shift 2/Forza 3/GT5, then it became VERY evident. Not to mention the track and horizon bump around in GT5 cockpit view instead of staying steady while the car moves around, like the camera is attached to the chassis instead of the driver...

This is one of my most liked features of Shift 2. Hitting a curb seriously upsets the car. Doing anything more than brushing a wheel up against it really alters your line, your weight distribution as the car is thrown around, you run the risk of hitting bump stops and putting the car into the wall, all sorts of things.

Prime example, the jumps in Shanghai. You cannot hit them full speed. You land too hard for the suspension to absorb, and unless you're really lucky your car won't be flat so you crash.

Same thing in GT5 at Cape Ring, 95% of the time you can hit the jump as fast as you like and your only concern is going too fast to make it around the next corner. You will not crash because of the suspension bottoming out.

There are severe differences in how suspension is modelled. I'm not surprised that they react differently to inputs. I do think that Shift's delay is overdone, but not by as much as some would say. And it's certainly very controllable when you're not yanking the wheel, some might say it was encouragement to drive smoothly and plan your moves ahead like they teach in every driving school.

 

[conath]

Seems that PS3 version has a lot greater input lag then PC ver.



I`m really thinking about switching to PC version. There`s already a mod that tweaks FFB and reduces input lag even further, everyone who tries it is like "wow, now i`m loving this game"

http://www.nogripracing.com/forum/showthread.php?t=246577

 

[RomKnight]

@imari, by reading your comments, there is a possibility that the physics are very good indeed (maybe just to good for what we're use to...)

Another thing i can "read" is that it seems the game is made for you to mess with car settings since there are poorly configured ones at stock (like a BIG Cadillac? )

Have I misread something here?

On a final note, I have to thank you for the effort your're making and share it with us. 👍

JJ72, great thread you have here

 

[LarsDeLuxe]

So this would actually mean that Shift 2 has a MORE realistic physics engine than GT5? The sad thing is that when you are used to instantaneous response from the input, it takes a while to adopt...

 

[spapadillion]

JJ72 and Imari:

Just look at the delay between the real life wheel and the virtual wheel on the screen. I'm not taking a jab at anyone or anything. I own the game AND like it, but fact are facts.

 

[Spagetti69]

Has EA actually acknowledged this phenomenon?

 

[spapadillion]

I'd love to hear their thoughts on the subject.

 

[RomKnight]

JJ72 and Imari:

Just look at the delay between the real life wheel and the virtual wheel on the screen. I'm not taking a jab at anyone or anything. I own the game AND like it, but fact are facts.

True, but it seems the telemetry got it right. Its something AFTER the input readings. This can be caused (from what i understand) by:

- the game engine is somewhat messed (fixable by SMS)
- the car setups are messed (also fixable even by user with a bit of time)

BTW, it seems the car answers to the screen wheel instead of the player's wheel as spapadillion said so i really think its one of the options i said above.

The good news is a patch can solve that. The bad news we need to wait...

@imari
we need a work around

 

[Imari]

JJ72 and Imari:

Just look at the delay between the real life wheel and the virtual wheel on the screen. I'm not taking a jab at anyone or anything. I own the game AND like it, but fact are facts.

The virtual wheel is even slower than the car to move. I wouldn't read too much into that wheel if I were you, there's definitely something funny going on with that. See how it jitters when you're going straight? It's definitely not linked directly to the player's controller.

I know. It's too much. It's a problem. It's just that a lot of people seem to be under the impression that it's a completely wrong effect. It's not incorrect, it's just overdone (as far as I can tell). What's causing it to be overdone, I'm not sure yet.

Basically, my position at the moment is that yes, it's wrong, but not for the reasons that most people seem to be saying (input lag). It's not input lag. It's something else.

 

[HeadsoupBob]

It may even be a PS3 version thing, I have Shift 2 on PC and there's no recognisable lag whatsoever...

 

[spapadillion]

Basically, my position at the moment is that yes, it's wrong, but not for the reasons that most people seem to be saying (input lag). It's not input lag. It's something else.

The argument that it's not technically "input lag", while it may be correct, is splitting hairs. There's a delay between controller input and the onscreen response. As you said, it boggles the mind as to why this is the case, but the term is more than accurate enough for me.