GT6 Top Speeds Are Not Realistic, Please Fix PD!

Discussion in 'Gran Turismo 6' started by jlmcmillan1978, Jan 16, 2014.

  1. Polsixe

    Polsixe

    Messages:
    1,188
    Location:
    Canada
    True, but some cars are still "pulling" at 5280 ( or 5278 for your track) feet so the top speed observed is lower than max possible.
     
  2. Polsixe

    Polsixe

    Messages:
    1,188
    Location:
    Canada
    But which would Polyphony Digital model against?
     
  3. Polsixe

    Polsixe

    Messages:
    1,188
    Location:
    Canada
    The Motec traces aren't too bad, compared to the R&T article which also provides time/distance plots. One big difference is the speed drop on changing gears, the R&T traces show a large downward spike whereas GT6 gear changes are almost invisible as if the transmissions were CVT.
     
  4. Ridox2JZGTE

    Ridox2JZGTE

    Messages:
    11,660
    One key problem that often forgotten, weight, not one car in GT6 has accurate weight/distribution, if wanted to compare to real life performance, at least get the weight as close as possible. Some cars in GT6 can be lighter by several hundreds of kg vs real car running weight.
     
  5. eran0004

    eran0004

    Messages:
    9,269
    Even if you reduce power by 50%?
     
  6. MclarenGT

    MclarenGT

    Messages:
    226
    I think they use european versions.
     
  7. RedTailBoa

    RedTailBoa (Banned)

    Messages:
    203
    i get a good laugh when I see Supras doing 330mph on SS Route X
     
  8. Polsixe

    Polsixe

    Messages:
    1,188
    Location:
    Canada
    I only used the SRT Viper SRT-10 Coupe '06. Wikipedia has this car at 510 HP and a top speed of 192.6 mph which gives a calculated (effective) Cd = 0.329, seems about right.

    Using the GT6 versions, horsepower, top mph, Cd:
    900 HP, 215 mph, 0.417
    800 HP, 207 mph, 0.416
    700 HP, 199 mph, 0.409
    600 HP, 189 mph, 0.410
    500 HP, 178 mph, 0.409
    -----------------------------
    500 HP, 175 mph, 0.430
    400 HP, 164 mph, 0.418
    300 HP, 149 mph, 0.418
    254 HP, 141 mph, 0.418

    That's as low as I can go with 50% power limiter. Note even at 254 HP setting the Viper is still accelerating at one mile. Cd formula is consistent, shows PD used a fixed number, given the speed values are visually observed only.

    Now compare 450 HP versions:

    Sarthe 2009 no chicanes, 191 mph, 0.298
    SSRX km 6 downhill, 208 mph, 0.231
    SSRX backside, 195 mph, 0.280

    So it looks like PD used a CD about 10% lower than actual, 0.298 instead of 0.329. SSRX seems wrong if Sarthe is right.
     
  9. eran0004

    eran0004

    Messages:
    9,269
    Try this track: https://www.gran-turismo.com/gb/gt6/user/#!/friend/eran0004/course/1344297/
     
  10. MclarenGT

    MclarenGT

    Messages:
    226
    The Mclaren test was done in poor weather condition and tested car was 30kg heavier than European version.
    By the way I have reached 335mph with Mclaren F1 before applying any updates,updates just ruined the 6th gear.
     
  11. Polsixe

    Polsixe

    Messages:
    1,188
    Location:
    Canada
    Ah yes, a good 6 mile straight. Thanks for that. 2 minutes of pressing.

    First the Viper '06:
    Viper 06 - 202 mph, ~6000 rpm, 1:51.884 secs, 508 HP, 0.284 Cd
    0.74 * 3.07 gears , 49/51, 1550 kg

    So this looks close to SSRX back straight, speed too high as Cd is too low.


    Corvette '06 - 210 mph, ~7000 rpm, 1:50.989 secs, 400 HP, 0.190
    0.74*3.42 gears, 51:49, 1445 kg

    Very low unrealistic Cd, faster than Viper with 100 less HP.


    McLaren MP4-12C '10 - 236 mph, ~8500 rpm, 1:36.085 secs, 617 HP, 0.218
    0.816*3.60 gears 45:55, 1238 kg

    Very low unrealistic Cd, very high speed

    Bugatti Veyron - 266 mph, 7200 rpm, 1:25.144, 1223 HP, 0.299
    0.795*2.73, 46:54, 1888 kg

    I've seen 0.348 listed as Cd, seems PD is making every top car more slippery for enhanced albeit artificial arcade effect.
     
  12. eran0004

    eran0004

    Messages:
    9,269
    What makes you think that? I think it's more likely that the new CFD aero model isn't accurate enough.
     
  13. Polsixe

    Polsixe

    Messages:
    1,188
    Location:
    Canada
    I'm assuming they used a fixed Drag Coefficient for each car model, then apply a drag formula to determine car speed. With all the other processing needed for the game, visual and car placement/orientation, then Cd should be as simple as possible, with standard conditions. Wind is not even a factor. Even if they bothered to use CFD on every car models, is a lot of work, I t still comes down to a coefficient.

    Now they must have tested and realized the cars are much faster but chose to release anyway. This was to appeal to the more arcade element of buyers more impressed with simulated top speed.
     
  14. eran0004

    eran0004

    Messages:
    9,269
    The point of CFD is that the drag coefficient is calculated by the fluid dynamics. So it would not be a preset value, but an output of the CFD simulation.

    All we know is that it's wrong, but we don't know why. Your guess is as good as mine, although I have a hard time believing that it would be deliberate when all they talk about is how they want to improve the simulation:

    "They chose to release anyway"? They may have known it wasn't correct, but that is not the same as it being incorrect on purpose. Some things you simply can't fix within a given timeframe.
     
  15. Polsixe

    Polsixe

    Messages:
    1,188
    Location:
    Canada
    Agreed, but I think that GT blurb on CFD was embellishing a little. F1 teams, the big auto makers, aerospace industry use CFD to find that last nth, but a video game? Maybe they used a simplified freeware software to generate a Cd but someone got their sums wrong.
     
  16. eran0004

    eran0004

    Messages:
    9,269
    The last nth is usually found in wind tunnels. CFD is mainly used during the initial design process, because it's cheap and fast, but limited by its use of approximations.

    Its use in a video game would be similar to any other model that's used in a simulation. When you build a physics model you can use two different approaches: use preset values or let the model calculate its own values. Take an engine model for instance: you can use a preset torque curve or you can build a model that calculates the torque based on a number of inputs (like, air temperature, air density, air-to-fuel ratio, throttle position, etc.)? At some point you need to use preset values because you can't let the computer calculate everything, it would be absurd to count the number of molecules in a petrol tank for instance.

    In the case of aerodynamics you can use a preset drag area value (based on empirical evidence), or you can let the physics model calculate the drag area for you.

    The benefit of a fixed value is that you can base it on real life data, and make it very accurate. The drawback is that the value is fixed, so if you need the value to change when the conditions change (like, if you increase downforce or if you change the ride height or angle of the car) you need to guesstimate.

    The benefit of letting the model calculate the value is that it allows for changing conditions - no need to guesstimate. The drawbacks are that it's hard to create a model that produces an accurate output, that the model requires more calculations to be made than the preset value, and that a complex model has a greater risk of bugs.