The speed difference between the PAL and NTSC versions.

  • Thread starter RogsR34UK
  • 47 comments
  • 5,825 views
Ikari_San
I believe they adjusted the timing by 17% to match the speed of the NTSC (60hz) version, so they should be the same.

Seems i had them the wrong way around... PAL runs 50 Hz at 640 lines, NTSC runs 50 Hz at 525 lines. just wanted to end the confusion (primarily mine).

I don't think changing the timing (alone) doesn't necesarily make all things equal as i said in my earlier post. I have a feeling that the physics is different in the two as well.
 
nice test! try testing it with more cars, Ive always wondered if there was a difference btw pal and ntsc! thanks 👍


~~~~ Jase
 
I have no desire to test every car in the game. Someone asked for the Mercedes Mclaren SLR so that I wil do and probably a few more to make it worthwhile.
 
It makes sense that the clock display is a part of it, leaving all else out - it would be nearly impossible to make the clock display exactly the same time with different framerates being used. So if the PAL version draws 50 times per second and the NTSC version 60 times per second, the actual times are going to differ slightly. You can't display each 1000th of a second if your screen only refreshes about 50 times per second. In theory that means times would always show like 0.020, 0.040, etc. on PAL, and 0.017 for NTSC. In practice, there's going to be a tiny bit of variation depending on what instructions are being run under any given circumstance which determines at which point the time is 'snapshotted' and displayed on the screen. This can then be masked by additional tricks like random additions or whatever to give the illusion of a 0.001 precision. I think that these weren't in place yet in Prologue, where you could see that the times actually differed about 0.01666 minimum in the online db at db.gtrp.de.
 
arwin
It makes sense that the clock display is a part of it, leaving all else out - it would be nearly impossible to make the clock display exactly the same time with different framerates being used. So if the PAL version draws 50 times per second and the NTSC version 60 times per second, the actual times are going to differ slightly. You can't display each 1000th of a second if your screen only refreshes about 50 times per second. In theory that means times would always show like 0.020, 0.040, etc. on PAL, and 0.017 for NTSC. In practice, there's going to be a tiny bit of variation depending on what instructions are being run under any given circumstance which determines at which point the time is 'snapshotted' and displayed on the screen. This can then be masked by additional tricks like random additions or whatever to give the illusion of a 0.001 precision. I think that these weren't in place yet in Prologue, where you could see that the times actually differed about 0.01666 minimum in the online db at db.gtrp.de.

you are correct arwin. there is no way for the playstation to graphically display precision to 1/1000th of a second accuracy. in fact, it cant even display precision to up to 1/100th of a second! a closer look at the lap time and total lap time counters will partly prove my theory.
i did a test in b-spec mode with a number of different cars, and what immidiately struck me was the fact that the clock was displaying different figures between the total time and lap time on the first lap! whilst the total time was advancing 'seeminlyly' at 1/100th of a second, the 1/1000th of a second counter was stuck on zero. the lap counter was also 'seemingly' advancing at 1/100th of a second, whilst the 1/1000th of a second counter was stuck displaying a random number every time the car crossed the finish line!

so how does GT4 manage to display lap times to a precision of 1/1000th of a second? some of the theories i have managed to come up with suggest that the timings in GT4, may not be accurate or real-time, but calculated. my best guess though, is that there is a clock running in the background, separate to the actual displayed clock, and this clock is the clock that lap times are calculated from. this clock then displays the information on screen as the car crosses the start/finish line. it is a lot easier to display one frame with the correct time, than it would be to try and replicate 1/1000th of a second precision.

so what does that mean for the fastest lap times we set? compared to the real world, GT4, might seem to be inaccurate, but you must remember that afterall, its only a simulation. Whether the game is simulating car physics, distance travelled, or time taken to travel a certain distance, it is still a simulation, and simulations are open for interpretation. there is bound to be a margin of error for the time calculation regardless of the frame rate - all timing systems have a % error or % accuracy either in game or in the real world.

what does this mean for NTSC vs PAL lap times? comparing the times against each other is pointless, as they are both running separate time simulations. PAL lap-times maybe faster but equally, they could also be slower. this is partly because of the differing frame-rates. if the times database doesnt have a column for NTSC, PAL (and maybe SECAM), then maybe now would be a good time to include one.
i would be interested to find out how the time is displayed under SECAM (the french telivision standard).
 
There must be an "accurate" clock somewhere or how can you have a very close race where the winning time is less than a few thousandths of a second.

Here are some more stats:

Japan

McLaren Mercedes SLR - 20.825 seconds/263 kmh
Mine's Skyline GT-R N1 Vspec - 20.226 seconds/267 kmh
Mine's Lancer Evo VI - 23.158 seconds/233 kmh
Audi RS4 - 24.169 seconds/225 kmh
BMW 330i - 27.517 seconds/199 kmh
Honda Civic Type R - 27.94 seconds/200 kmh
Honda NSX-R - 24.091 seconds/225 kmh
Triumph Spitfire - 37.221 seconds/147 kmh


PAL

McLaren Mercedes SLR - 20.779 seconds/163 mph
Mine's Skyline GT-R N1 Vspec - 20.251 seconds/166 mph
Mine's Lancer Evo VI - 23.148 seconds/145 mph
Audi RS4 - 24.182 seconds/140 mph
BMW 330i - 27.509 seconds/124 mph
Honda Civic Type R - 27.989 seconds/124 mph
Honda NSX-R - 24.115 seconds/140 mph
Triumph Spitfire - 37.216 seconds/91 mph
 
at a cursory glance - the difference seems to be greatest in high powered one ended cars - where the physics differences between version would most likely be the greatest. i bet a fully tuned minolta would be even more different

if we could do a test on an actual circuit with the same skilled B-spec driver in the two versions then we would be getting closer to what's causing this, and how big the difference is in practice. Incidentally has anyone ever tested the actual effects of B-spec skill on lap times. Does a more skillful Berryl actually drive quicker than a lesser Benvolio?
 
There's a difference between keeping track of the time and displaying the time. While we only get time updates at a set interval (I.E each time a new frame renders we get a new time update for the display) the timer is constantly running and is most likely based on milliseconds.

So, the timing is the same in all versions of the game, but it's displayed differently based on framerate.
 
kungtotte
So, the timing is the same in all versions of the game, but it's displayed differently based on framerate.

sounds official, but it doesn't explain the differences in the way the physics are modelled
 
RogsR34UK
Triumph Spitfire - 37.221 seconds/147 kmh

:lol:

I'm surprised the Triumph travelled 1000 metres without breaking down!

On second thoughts, looking at the time it posted, it could well have broken down halfway up the run, and then got towed across the line.
 
I thought I'd better run something slow just to check the difference and the Spitfire came to mind as it easily purchaseable. It is intereting that the times between the two versions for it are so close.
 
RogsR34UK
...
Japan

McLaren Mercedes SLR - 20.825 seconds/263 kmh
Mine's Skyline GT-R N1 Vspec - 20.226 seconds/267 kmh
Mine's Lancer Evo VI - 23.158 seconds/233 kmh
Audi RS4 - 24.169 seconds/225 kmh
BMW 330i - 27.517 seconds/199 kmh
Honda Civic Type R - 27.94 seconds/200 kmh
Honda NSX-R - 24.091 seconds/225 kmh
Triumph Spitfire - 37.221 seconds/147 kmh


PAL

McLaren Mercedes SLR - 20.779 seconds/163 mph
Mine's Skyline GT-R N1 Vspec - 20.251 seconds/166 mph
Mine's Lancer Evo VI - 23.148 seconds/145 mph
Audi RS4 - 24.182 seconds/140 mph
BMW 330i - 27.509 seconds/124 mph
Honda Civic Type R - 27.989 seconds/124 mph
Honda NSX-R - 24.115 seconds/140 mph
Triumph Spitfire - 37.216 seconds/91 mph
I know this might be getting a bit pendantic, but can you convert the mph to kph (or vice versa)? This will show whether the difference is a timing thing, or an acceleration one.

If the differences truely is in the timing, then I think there should be no differences in the speed of the car. However, if the differences are down to different physics models, then the a higher speed may coincide with the quicker time?

There's a convertor here:

http://onlineconverters.com/costasjsuc.html

Sorry - can't be arsed doing it myself...

Does anyone else think this might be a feasible hypothesis?... :)
 
I still think it could be related to a difference in the total time taken from when you hit "drive - a-spec" to the launch countdown reaching zero in the different versions, which would easily lead to launching at slightly different revs in the different versions but always at the same revs in the same version. We're talking about tenths and in some cases hundredths of a second here. It's a tiny difference. The theories about the clock sound good, especially considering there is a history of timing peculiarities in the series. However, the countdown period is easier to test. Just need a stopwatch. Unfortunately, you also need both NTSC and PAL copies, which I don't have.

Just a thought. Shoot me down if I'm talking rubbish :)
 
rogsr34uk
There must be an "accurate" clock somewhere or how can you have a very close race where the winning time is less than a few thousandths of a second.

i believe that the actual displayed clock is only accurate to the nearest second, and nothing more. every time a car crosses a checkpoint or the start/finish line, the background clock is then referenced, and the results displayed as we see them in splits and lap times. this would go some way explain why i witnessed the odd occurances with the 1/1000th of a second counter. despite this, it would still be possible to have close races even down to a few thousandths of a second.
 
ZAGGIN
i believe that the actual displayed clock is only accurate to the nearest second, and nothing more. every time a car crosses a checkpoint or the start/finish line, the background clock is then referenced, and the results displayed as we see them in splits and lap times. this would go some way explain why i witnessed the odd occurances with the 1/1000th of a second counter. despite this, it would still be possible to have close races even down to a few thousandths of a second.

The display clock is accurate to 1/50th (PAL) and 1/60th (NTSC) of a second - 0.020s and 0.016s respectively.

To test this out, drive along and press pause.
 
Back