GT5's low-speed physics are still terrible.

[JTSnooks]

I searched the various physics discussion threads and never saw a specific discussion of the low-speed inaccuracies, so I figured I'd create my own thread. I apologize if it's been discussed at length before, but I couldn't find it.

Now let me just say that I'm not a serial complainer, and I love the game, but I was messing around on the TG Test Track this afternoon and was pretty disappointed that they haven't fixed some glaring issues with the tire/clutch/engine physics.


1. Why, if I dump the clutch at full throttle in 2nd gear in a Corvette Z06, does it bog down and not spin the tires? I also tried it in a Camaro SS on COMFORT HARD tires and it still wouldn't spin them!

2. I paid a lot of money for my wheel with a clutch pedal and H-pattern shifter, and the fact that I can't powershift is incredibly frustrating. If the tires are spinning at all, it won't let you change gear. Doing a burnout in first and try to quick-shift to 2nd? No dice, you have to wait until the tires stop spinning before it will select the next gear, even if you let off the throttle between gears.

3. Dumping the clutch at full throttle while in motion has no effect, you can't even chirp 2nd gear. Ridiculous.

4. The way the game handles low-rpm torque is terrible. In real life, if I'm in a corner going 45mph in 3rd gear in my GTO and floor it, the tail will jump right out on me. In the game, you can't get the tail to jump out at 45mph in 3rd unless you throw it back and forth to get it off-balance. Flooring it around a corner won't do anything at low speed/low rpm. Again, this testing was done with the 2010 Camaro SS.

5. I spun out in 2nd gear in the SS. I kept the throttle planted all the way through the spin, and rather than it continuing to boil off the tires after spinning 180*, the tires hooked immediately and the car bogged off with full traction. If you did that in real life you'd sit there bouncing off of the rev limiter for a few seconds with the tires going up in smoke before the car even started to move.

6. You can't do a proper burnout. Just try doing a brakestand. I'm guessing if you use the brake controller and set the rear brakes to 0 it might work, but I haven't tried because that's pretty pointless anyway since you don't need a line-lock to do a burnout in real life.

The fact that you can't do something as simple as a 2nd gear burnout is a huge deviation from reality in a game that claims to be a "driving simulator".

Someone please post some kind of rebuttal to this as it's really taken the wind out of my sails in regards to GT5.

 

[Boz Mon]

Do you have you traction control turned off?

 

[alexleighton]

I would love to post a rebuttal but I can't because you're completely true about it all.

Not being able to do a burnout doesn't kill the game for me but not being able to powershift or shift while the wheels are spinning is just stupid, no other word for it.

It's made even worse by the fact that the game doesn't give you any kind of audible or tactile signal that your shift hasn't worked, I shouldn't have to look at the gauge to see if I'm in gear or not.

 

[Rauwers]

Man you are 100% right!!!
I would really love GT5 clutch to work like forza3, the way it is now is just really frustrating and stupid. Looks like im playing a game in my ps1 10year ago.

 

[G.T.Ace]

While of course it's not perfect (cars definitely have too much traction when accelerating from 0), it's 100 times better than GT4.

 

[shirakawaa]

This is what I've written about low speed physics in another thread. Maybe it's about other aspects of this problem compared to what you addressed. Sorry for the shameless copypasting:

Yesterday I was playing around with some cars, and I noticed that low speed tire physics probably need some serious improvement. It's currently impossible to start a car slowly without making the tires slip a bit. This is so noticeable that it's possible to hear a slipping sound. Also, on slopes, the car always drifts (in varying amounts) a bit to the lower side of the road when starting, no matter how slowly. It's as if tires weren't grippy enough at very low speeds.

Has anybody else tried this?
This is probably one of the reasons why rolling over is so difficult even with tall cars.
When the car travels completely sideways, vehicle speed in the game is zero.
It looks like tire grip is proportional to game vehicle speed in some way.

I'm using a wheel.
By the way, now that I think about it, the faulty low speed tire dynamics are most probably the reason why many cars with skinny tires are unable to climb steep hills. It mostly happen not because of lack of power, but lack of traction!

If you use the roof cam with a car with a quiet exhaust it will be clear that at low speed, tire grip seems to be not enough, as the usual but light tire skidding sound will be able to be heard.

If you choose Special Stage Route 5 you will begin close to the long, slightly banked corner leading to the main straight. You will be able to quickly test there both that and the "skidding/lateral drifting" occurring when starting from a dead stop on a slope very slowly.

If you switch to the chase camera, you will also see visually that the car behaves weirdly.
Anyway, the more I think about it, the more I feel it's all "connected".

Strange low speed tire physics, extremely difficult rollovers, handbrake dynamics, little to no grip on steep slopes with skinny tires/old underpowered cars. It's probably all related to the low speed tire model. It looks as if grip has been intentionally limited when the relative speed between the road and rotating tires approaches zero, and this causes some "glitches" that most people wouldn't notice in most situations.

It would have been interesting to check if that E3 demo where the old Fiat 500 easily rolled over had the same low speed tire behavior.

I've tested this again a bit in GT5 and I found out that the strange things happen only at indicated speeds below 5 Km/h. Once the car speed reaches 5 Km/h, the behavior instantaneously disappears. You can even see (with the chase cam) the transition between the "low-speed physics" and the "normal physics".

There's something wrong and it looks it's there for a specific reason.
Interesting and disappointing at the same time...

Again, this is quite visible in the banked corner before the main straight in SSR5.

- Select a RWD car with a relatively tall 1st gear
- Start on Special Stage Route 5
- Select the chase camera
- Stop in the middle of the last corner of SSR5
- Turn the wheel to one side and start very slowly. The car should oddly drift sideways a bit.

Do you notice that the strange "low speed drift" immediately stops at 5 Km/h? This is also reflected in the wheel force feedback.

Oh, by the way:

6. You can't do a proper burnout. Just try doing a brakestand. I'm guessing if you use the brake controller and set the rear brakes to 0 it might work, but I haven't tried because that's pretty pointless anyway since you don't need a line-lock to do a burnout in real life.

I think the most probably reason for this is that the ABS implementation in GT5 prevents this from happening.
By switching the ABS off and setting a proper brake balance suitable for no-ABS braking (for example 4/1), burnouts are kind of possible.
I'm writing "kind of" because unless there's a huge imbalance in tire grip between the two axles, the car will still keep moving forward, although at a low speed (10 Km/h or so).

Other than this, I agree with all the rest you've written. The game needs better tire physics and proper drivetrain simulation.

 

[Eric.]

Forza did nail shifting. I never used a wheel with it, but even using the regular manual shifting there was that burst of torque from the hard upshifts. Newer, powerful cars with paddle shift transmissions could jump sideways mid-turn if you upshifted without stepping off the throttle a little. Cars like the Ferrari 458, 599XX, Mercedes SLK55.

And at least if you enabled the clutch in Forza and missed the shift, it would still happen, despite slowly and with a bunch of grinding. I believe the wheelspin issue is why a lot of people have trouble getting 2nd gear off the line.


Edit: Oh, and the clutch being an on/off switch is annoying also.

 

[doctorandrew]

I've always wonder what it takes as far as raw processing power to be able to compute REAL physics, accounting for all the formulas of mechanics and kinetics we know of. Why not just implement everything PD?

 

[tigermilk]

I noticed the low speed issues in the GT Academy demo from last year. It was a real breakdown of the physics modeling. Makes you question the entire coding IMO. I do wish for more tail happy cars. I'm trying to get gold on the Seasonal Event race with the Skyline. I'm just a couple of seconds away and it would help if I could get the massive understeer out or just be able to floor it in second gear around the tighter turns to whip the tail end out. Downforce set to a minimum, all assists off, but the tail end just won't budge. And even at higher speeds I suspect some major issues. The minimal downforce in there causes a constant yaw oscillation around the higher speed turns, like the tail wants to kick out but the poorly coded physics model says "Uh, no, that is verboten!" In Forza with all assists off I enjoy the tail happy cars.

I'd say it's not a generic issue but a problem with how particular cars are coded up. Some of those 60s American cars are loosey goosey when I want them to be.

 

[cicohipe]

I've always wonder what it takes as far as raw processing power to be able to compute REAL physics, accounting for all the formulas of mechanics and kinetics we know of. Why not just implement everything PD?

Not feasible on the PS3 or for a mass market video game. Physics engine's can take pretty much as little or as much processing power as is available, and that you want to model. You are unlikely to ever see a 100% accurate physics model though.

That being said, the problem with GT5's low speed physics is more a problem of incorrect algorithms I imagine, rather than lack of processing power or other.

 

[JTSnooks]

I'm glad I'm not the only one who noticed this, thanks for the input.

The thing that I don't understand is that in almost every other way, the physics are vastly improved from GT4 and are very realistic. Yet for some reason nothing has really changed regarding something so seemingly simple as doing a true standing burnout. It makes me wonder if there's an actual reason for it (difficult to program correctly, or would screw up some other aspect of the physics), or if it's simply a huge glaring oversight. Surely Kaz has done a burnout in real life and tried one in the game, right? When a teenager gets into a RWD vehicle what's the first thing they do? See if they can get the tires spinning and make the biggest cloud of smoke they can. Not being able to do this in the game is just baffling. Do they not do brakestands in Japan? Why would this not be made possible in the game? I just don't get it.

 

[Owen.C93]

It's not really a problem for racing, but it makes starts really difficult because getting a good getaway isn't really like real life. And doing a standing start on the banking at Daytona is hilarious.

 

[shirakawaa]

Well, in my opinion GT5 is not only about racing, but also enjoying cars at low speeds.
So low speed physics need to be addressed.

 

[tigermilk]

I've always wonder what it takes as far as raw processing power to be able to compute REAL physics, accounting for all the formulas of mechanics and kinetics we know of. Why not just implement everything PD?

Because they are spending all their computing resources on giving you picture perfect vehicles. It should and can be done. PC sims 10 years ago were doing it, and the PS3 has more power than a 10 year old PC.

I've done my fair share of kinematic simulations (mechanical/aerospace engineer by trade), coding from the ground up. Modeling the car doesn't take too many equations, and solving them is very fast. Just a simple F=ma problem with numerical integration.

I may have to download this package - http://www.racer.nl/ It's a 15+ DOF (6 for the vehicle CG (I expect it's at the CG), 2 for each wheel (vertical and roll), 1 for the engine) package that gets high marks. Some of the physics features:

PHYSICS FEATURES

* Full 6 degree of freedom motion of the car chassis.
* Independent suspension for all wheels.
* Suspension features: springs, dampers, anti-rollbar, rollcenter, anti-pitch.
* Wheel features: camber, wheel hop (the wheels have mass), toe.
* Tires: Pacejka tire model, relaxation length.

 

[JTSnooks]

Well, in my opinion GT5 is not only about racing, but also enjoying cars at low speeds.
So low speed physics need to be addressed.

Exactly. A game that bills itself as "the real driving simulator" should be able to simulate how my car will respond to something as simple as brake/clutch/throttle inputs. I would think a burnout would be a very simple physics calculation as there's no lateral movement whatsoever, only longitudinal and pitch (assuming a flat surface).

Maybe they do it to make it easier to drive with a controller? If you had to let off the gas every time you hit the upshift button it would be pretty difficult. You would think they could come up with a better workaround than handicapping every car in the game though.

 

[Stigsblkcousin]

You would think this would be something PD would have down pat by now, especially with Kaz talking about wanting to have a free ride mode where you had to abide by traffic laws and such. Maybe this is why that idea was scrapped.

 

[Alan_G]

GT4s traction/wheelspin was very, very primitive. With TC on zero, there seemed to be two modes; rolling, and spinning. So GT5P/GT5 is way ahead of that.

But you're surely right anyway. It's not the only problem with the physics (which is why I believe there are no leaderboards yet - physics update).

 

[jhietter]

I have definitely noticed the lack of feel at slow speeds. My mom had a mini Cooper s in real life and it was very zippy, lively, and fun to drive even at low speeds. In the game, the car seems to handle like a boat and has no feel below 70mph.

At first I thought that the game just portrayed the car inaccurately. I didn't understand why the car couldn't turn at such low speeds but when I looked at the speedo, I was going almost 50mph. Now knowing the speed, the car is behaving realistically.

The problem is that sensations of driving don't kick in until high speeds in the game but cars in real life provide feedback from 0.

 

[I6-4-eva]

I think the powerband may have something to do with it as well. The newer camaro and challenger in the game have a very peaky power delivery. Just take a look at the Max.Power/Torque chart in game of the 2010 maro and compare it to this http://www.dyno-charts.com/chart.php?whichchart=2010camaroSS.jpg .

 

[Zoom!Zoom!]

Because they are spending all their computing resources on giving you picture perfect vehicles. It should and can be done. PC sims 10 years ago were doing it, and the PS3 has more power than a 10 year old PC.

I've done my fair share of kinematic simulations (mechanical/aerospace engineer by trade), coding from the ground up. Modeling the car doesn't take too many equations, and solving them is very fast. Just a simple F=ma problem with numerical integration.

I may have to download this package - http://www.racer.nl/ It's a 15+ DOF (6 for the vehicle CG (I expect it's at the CG), 2 for each wheel (vertical and roll), 1 for the engine) package that gets high marks. Some of the physics features:

PHYSICS FEATURES

* Full 6 degree of freedom motion of the car chassis.
* Independent suspension for all wheels.
* Suspension features: springs, dampers, anti-rollbar, rollcenter, anti-pitch.
* Wheel features: camber, wheel hop (the wheels have mass), toe.
* Tires: Pacejka tire model, relaxation length.

I heard and read developers, like those at Iracing, claim that tire physics are extremely complex and hard to simulate so often compromises are made in sims. I read the same about flight sims that even today's PC are too slow to handle the complex physics of a plane.

 

[Ecchi-BANZAII!!]

Reverse is a bigger problem in my opinion.
When you do reverse it's lie you have PVC tires but you don't spin the tires, it's like he's spinning on the clutch.
And reverse has auto traction control? I hate it.

 

[CoolColJ]

One thing that has bothered me in every GT game, including GT5, is that you can just rev the car full throttle and dump the clutch and you'd would be fast most of the time, off the line.

In every license test that had a standing start, with ghost, it was always faster to just full throttle and dump the clutch off the line. Depsite the wheel spin, you were always faster than feathering the throttle and trying to launch properly. Especially in FWD cars.

Torque steer still isn't modelled in FWD cars as well. All you get is even wheel spin on both sides of the car. FWD cars feel the same way as they have since the first GT game.
And in real life FWD cars can be prone to lift off oversteer, but you can't do it in GT5

 

[Devedander]

I've always wonder what it takes as far as raw processing power to be able to compute REAL physics, accounting for all the formulas of mechanics and kinetics we know of. Why not just implement everything PD?

Well technically it can't happen because "real" physics has infinite resolution at infinite calculations per second.

But I think wht you mean is (for example) will games actually model the airfliow over a car and figure real liquid motion formulas to find air drag instead of just setting a drag variable and basing the effects off of this. For those who don't get why that would be important, when you just give a variable you loose the resukts of any of the related effects... Like when you drive close to a wall, the difference in air pressure on the wall side of you car may have effects on how the aero works etc...or the difference in effect from drafting off a larger vs smaller vehicle or vehicle with certain body shape and air distortion characteristics.

I think we are a long way off from seeing real physics calculations rather than just analogs. Look at advanced games like iracing and check out the data files for cars... When you look at the variable info for each car you see sinple terms like (lift) and (downforce) which mean they have precalculated these in generic conditions and the game is not figuring out what they are based on the cars shape and air flowing around it.

If iracing is doing it that way still I think we are very far off from models with enough data to actually calculate this off of and processors with enough power to do it on a reasonable scale.

 

[Exorcet]

I heard and read developers, like those at Iracing, claim that tire physics are extremely complex and hard to simulate so often compromises are made in sims. I read the same about flight sims that even today's PC are too slow to handle the complex physics of a plane.

It's true, however, not even NASA uses, or needs to use, the exact physics. What everyone uses are models of varying accuracy. It's my belief that PD simply does not want to make a full simulator.

 

[Eric.]

One thing that has bothered me in every GT game, including GT5, is that you can just rev the car full throttle and dump the clutch and you'd would be fast most of the time, off the line.

Not hardly. GT3 and GT4 had a huge problem with launching. If you didn't just give it enough throttle where the game simulated feathering the clutch up until it "engaged" at the actual car speed, you could do burnouts for a long time without moving anywhere. GT5 seems to get the traction of a launch down pretty well, but now we can't make that early gearchange into 2nd while the wheels are still spinning.

 

[CoolColJ]

Not hardly. GT3 and GT4 had a huge problem with launching. If you didn't just give it enough throttle where the game simulated feathering the clutch up until it "engaged" at the actual car speed, you could do burnouts for a long time without moving anywhere. GT5 seems to get the traction of a launch down pretty well, but now we can't make that early gearchange into 2nd while the wheels are still spinning.

This is with a DGFT or control pad, so no manual clutch changes.
9 times out of 10 it's faster to just rev the engine full bore at the start, you'll be faster than trying to feather it off the line

 

[Rdp616]

I can do standing burnouts, and i'm using a controller.

 

[Samus]

In every license test that had a standing start, with ghost, it was always faster to just full throttle and dump the clutch off the line. Depsite the wheel spin, you were always faster than feathering the throttle and trying to launch properly. Especially in FWD cars.

This really bothered me when doing the tests, especially the stopping test. Didn't matter what you do trying to feather the throttle and get a good launch, full throttle on the rev limiter is faster. Silly.

 

[shirakawaa]

License tests use Skid Recovery by default, and can't be turned off, by the way.
That's probably why the best results there are obtained by flooring the throttle pedal.

 

[JTSnooks]

I think the powerband may have something to do with it as well. The newer camaro and challenger in the game have a very peaky power delivery. Just take a look at the Max.Power/Torque chart in game of the 2010 maro and compare it to this http://www.dyno-charts.com/chart.php?whichchart=2010camaroSS.jpg .

You call having 340ft-lbs of torque at 2500rpm "peaky"? Trust me, the Camaro basically has a slightly larger version of the engine in my car and only weighs a little more. If you drop the clutch with even 50% throttle it'll spin the tires right up to redline all the way through the gear.

Also, CoolColJ, you bring up another point that I hadn't addressed. Their launching physics have always been terrible, although they were a bit more realistic in GT4. Stomping the throttle and burning the tires off is just about the slowest way off the line in real life, yet in GT5 it's the quickest. The Mini slalom license test was an excellent example. If you floored it and spun the tires off the line in real life, you'd be significantly slower than if you feathered the throttle/clutch to keep the tires on the limit of adhesion. In GT5, all feathering the throttle does is slow you down, another example of being completely unrealistic.

Again, I think the reason for this is just to make the game "easy" rather than accurate.