|
|
Post by CHILLI on Sept 1, 2015 14:01:07 GMT
Update: Finally found out what Roll Centre is most likely used for. One of my references can be found in the first post under RollCentreHeight.
|
|
FilippoM93
Member
Posts: 13 
Registered on: September 2015
|
Post by FilippoM93 on Sept 18, 2015 10:10:49 GMT
Broughy's spreadsheet is amazing with all the lap times, but I believe it has 2 lacks: - Stock cars lap times - How the car values (engine power, drag, top speed, cornering grip and straight line grip) change after it's fully upgraded. (Maybe it's in another file instead of Handling.meta) Handling.meta (basic) is about stock values of the cars. Choosing the right cars (to use in a stock race o to buy to upgrade and then race it) would be easier if we have those 2 missing things because we can see the relationships between Broughy's laps and value to the car (stock or upgraded) and then we can make comparisons between each car. This is only my opinion, I hope it's possible to do that even if it's a big commitment P.S.: Sorry for my bad english  |
|
|
|
Post by cloudmcshort on Sept 18, 2015 13:54:48 GMT
- Stock cars lap times - How the car values (engine power, drag, top speed, cornering grip and straight line grip) change after it's fully upgraded. (Maybe it's in another file instead of Handling.meta) Handling.meta (basic) is about stock values of the cars. Choosing the right cars (to use in a stock race o to buy to upgrade and then race it) would be easier if we have those 2 missing things because we can see the relationships between Broughy's laps and value to the car (stock or upgraded) and then we can make comparisons between each car. -Stock lap times: Including these means that Broughy would have to test EVERY single car again, and i don't think he wants to do that. It's so much time and effort for something that very few people will actually make use of. - Fully upgraded stats: No. It doesn't work like that. If it was available, it would already be included. I do however agree with you. It would be nice to have available. |
|
FilippoM93
Member
Posts: 13
Registered on: September 2015
|
Post by FilippoM93 on Sept 18, 2015 16:45:29 GMT
- Stock cars lap times - How the car values (engine power, drag, top speed, cornering grip and straight line grip) change after it's fully upgraded. (Maybe it's in another file instead of Handling.meta) Handling.meta (basic) is about stock values of the cars. Choosing the right cars (to use in a stock race o to buy to upgrade and then race it) would be easier if we have those 2 missing things because we can see the relationships between Broughy's laps and value to the car (stock or upgraded) and then we can make comparisons between each car. -Stock lap times: Including these means that Broughy would have to test EVERY single car again, and i don't think he wants to do that. It's so much time and effort for something that very few people will actually make use of. - Fully upgraded stats: No. It doesn't work like that. If it was available, it would already be included. I do however agree with you. It would be nice to have available. Yeah, it'd be a big effort, I'm not asking him to do it. The thing about fully upgraded cars data isn't so difficult if it's possible to prove that every upgrade increases the same value of different cars. For example, EMS level 4 increases acceleration of each car by an amount. If this isn't real it would be nice to have available that datas. |
|
|
|
Post by CHILLI on Sept 19, 2015 7:51:32 GMT
About the fully upgraded stats:
If we'd assume that the bars show the correct proportional difference we could approximate what the resulting values are. But with that info we'd only be able to guess 4 out of all of the stats which, knowing that some affect each other, still wouldnt end up being all that useful. For example we can see acceleration increase some amount when buying transmission but we still dont know how quick the gear changes become. So a car with slightly lower acceleration can still be faster if it has faster gear changes. Now imagine the same situation for more or less all the other stats...
And that's not mentioning the fact that the acceleration bar can extend beyond the 5 full bars so we have no idea how much higher it really gets.
Measuring some effects based on video footage wont give us the real answer either as, again in the case of acceleration, drag will slow the entire vehicle down overall and not just the top speed (assuming it works the same as the previous games). So the assumption that all parts give a percentual increase is good enough for me until someone figures out a way to extract all the data somehow.
In short testing for all of this, even based on footage, wont give us the real answers because of the amount of stats that affect each other. And the bars in the game only partially reveal 4 out of them all so that's not worth a whole lot... Also I've looked around in the game files time and time again in hope of finding some sign of what the upgrades change but I cant find anything apart from the models used. The model data/naming/structure also gives no indicators of what the associated upgrade(s) change...
|
|
FilippoM93
Member
Posts: 13
Registered on: September 2015
|
Post by FilippoM93 on Sept 20, 2015 9:10:59 GMT
You fully answered all my doubts, thank you  |
|
|
|
Post by CHILLI on Oct 3, 2015 21:03:56 GMT
I have added some more info to the first post: * " Math:" notes to some stuff * New math category at the bottom, below the screenshot The math category dives right into the deep end of the physics side of the game based on in-game testing aswell as coding my own physics that can be found on the General Board for those interested in that. |
|
Pardonias
Member
Posts: 138
Registered on: December 2014
|
Post by Pardonias on Oct 4, 2015 10:07:38 GMT
Hi CHILLI, I've read the updates and there are few points I am uncertain about. From what I've seen on PS4, the speedometer in first person is pretty accurate, so when it shows 112.5 mph, that will be the actual speed, not 150 mph. You can test that by travelling on a long straight with constant speed and by measuring the time and distance. I tried to check this, so I randomly picked the Jackal. Its drive force is 0.22; the acceleration bar on SC says 0.55. The brake force is 0.9; the braking bar is 0.29999998. Traction min and max are 2.45 and 2.6, which is far off the 0.7878782 of the traction bar, too. I'm not sure how you came to those results, but I can't seem to confirm them. It might still just be those values, but scaled by a different factor. How did you determine these formulas, i.e. do you have any way measuring an acceleration value on PC, for example? To me, the factors of 10 and 4 seem to be a bit random. Do they translate to a specific unit (Netwon, m/s²)? From a physical point of view, acceleration would basically be driveForce/mass, and at higher speeds drag would have more of an impact which reduces the acceleration. If you are more specific, rev values should have an effect on acceleration, too (higher revs means better acceleration), but the current gear should not have as much of an impact as in your formula. From my own testing, I'm also certain that the mass has an impact on acceleration in GTA in a rather realistic way which is completely missing in your formula.  Straight-line traction might play a part, as well. |
|
|
|
Post by CHILLI on Oct 4, 2015 11:51:18 GMT
From what I've seen on PS4, the speedometer in first person is pretty accurate, so when it shows 112.5 mph, that will be the actual speed, not 150 mph. You can test that by travelling on a long straight with constant speed and by measuring the time and distance. How did you determine these formulas, i.e. do you have any way measuring an acceleration value on PC, for example? To me, the factors of 10 and 4 seem to be a bit random. Do they translate to a specific unit (Netwon, m/s²)? From a physical point of view, acceleration would basically be driveForce/mass, and at higher speeds drag would have more of an impact which reduces the acceleration. If you are more specific, rev values should have an effect on acceleration, too (higher revs means better acceleration), but the current gear should not have as much of an impact as in your formula. From my own testing, I'm also certain that the mass has an impact on acceleration in GTA in a rather realistic way which is completely missing in your formula. Straight-line traction might play a part, as well. The reason I say that the speedometer only shows 75% is because I commonly see the fastest cars in the game naturally cap at 120 when the handling file claims 160 (or very close to). Also the ratio from KM/H to MPH is ~0.621 so the numbers will get really close if you measure it, you're right. I also noticed that if the ground isnt a perfectly flat surface the speed can go beyond the stated top speed very easily. If the car is upgraded then it will break this limit as if it's not even there. And the reason I multiply driveForce by 10 and brakeForce by 4 is because then the numbers will get converted into something that can be compared to the traction values to determine traction loss. I came to that conclusion when coding my own physics (I took their values and messed around with them until I got something that could be compared to a sort of "traction force target"). There could be a little more going on with how the traction limits behave but those assumptions got me really close when testing in-game. Also regarding how the mass ties in I mentioned at one point near-ish the end "multiply by the mass to get the final force(s)". If you compare the two Kurumas for example the armored one has a lot higher mass but the same acceleration value. In-game they're also very very close, which means that if the mass is taken into account it's not by much. From modding the previous games and looking at the values vs. what happens in this game it seemed safe enough to assume mass means nothing when evaluating the overall performance. Drag certainly plays a role but, again, that didnt stop the cars I tested from safely reaching their top speed. Also I did mention that I just eyeballed how the rate of acceleration decreases through the gears and I have yet to test this in my own simulation. I tried my best at tracking how quickly the speedometer needle moved through each gear and the difference got smaller and smaller between the higher gears. When checking what happens if you keep dividing by a value that only increases by 1 each time it made sense. 100% - 50% - 30% - 25% - 20%. So the first 2 gear changes show a huge difference and from that point on it happily accelerates at an average rate of 25%. The gear ratios are completely linear but the higher gears feel longer because the car is accelerating slower... I dont know if I should be disappointed or happy that it's as simple as this... There are no other factors stated in the handling file that directly change this outcome so this has got to be it. I should also mention that the RPM needle appears to be showing the resulting amount of player input that's later fed into the simulation. The reason the boosting works is because there is some inertia to how quickly the revs can change, so as one wheel rotates faster and lands it will try to accelerate at a matching rate for a moment before the RPMs resettle at their correct value. That said though I still havent nailed down why going over bumps prompts the revs to increase... Lastly I certainly think you're right about the LSC bars being multiplied by another uknown factor before being shown to us. I should probably mention that the relative differences seem to be kinda like that... from what I remember... Is the traction bar actually that low? If so then I should really pay more attention to it from now on haha. The acceleration bar is altered a lot to make it comparable against other vehicles. Only problem is that said factor keeps changing so it's near impossible to figure out. I do think the bar increase from each upgrade shows the correct ratios though. I'll edit the first post and remove that section because I completely forgot how much the bars vary from car to car despite having similiar values. |
|
Pardonias
Member
Posts: 138
Registered on: December 2014
|
Post by Pardonias on Oct 7, 2015 17:18:49 GMT
Right, so I had to think about this for a while and did some tests. The reason I say that the speedometer only shows 75% is because I commonly see the fastest cars in the game naturally cap at 120 when the handling file claims 160 (or very close to). Also the ratio from KM/H to MPH is ~0.621 so the numbers will get really close if you measure it, you're right. I also noticed that if the ground isnt a perfectly flat surface the speed can go beyond the stated top speed very easily. If the car is upgraded then it will break this limit as if it's not even there. I think a lot of confusion stems from the question of what the maximum velocity value actually is and what it does. I wasn't sure about it, either, so I went back to your first post and to the comments in the old handling.dat file. It states that the unit is km/h – contrary to what Broughy claims in the handling basic spreadsheet sheet (mph). This means that the value of a stock Adder, for example, is 160 km/h = 99.4 mph, even though it is quite easy to exceed this speed. The handling.dat file explains this value as "Max speed in top gear (used when configuring gears)", so I assume it is just a value which doesn't have any actual meaning in the game (i.e. it's not what you see on the speedometer at max speed), but it is used to tweak the gears of a car in the developers' test setup, and also has some effect on the top speed of the car (as tested by Broughy). What you see on the speedometer in the game is mph and it is pretty accurate. Also note that the value is named fInitialDriveMaxFlatVel, so performance upgrades likely increase this value. Also regarding how the mass ties in I mentioned at one point near-ish the end "multiply by the mass to get the final force(s)". If you compare the two Kurumas for example the armored one has a lot higher mass but the same acceleration value. In-game they're also very very close, which means that if the mass is taken into account it's not by much. From modding the previous games and looking at the values vs. what happens in this game it seemed safe enough to assume mass means nothing when evaluating the overall performance. I didn't think of the Kurumas – they're an excellent example. I got confused by the fact that the values are labeled fInitialDriveForce and fBrakeForce, when in fact these are acceleration and deceleration (as you mentioned in the OP: m/s^2). So, these do indeed seem to account for mass already, and multiplying by the mass would yield the forces. The reason I said mass has an effect is because of a test I did, motivated by Broughy's taxi passenger test. I measured the time it takes a Rubble to accelerate for a certain distance and then used real physics to calculate how much filling the Rubble with one ore two armoured Kurumas (because of the high weight) would slow it down (22% and 40%). The actual tests resulted in almost exactly those values, so, apparently, mass has the same effect on an accelerating car as in real life. Of course, changing the mass in the handling file would not have the same effect, as the acceleration is determined by the drive force value. Yet, the mass of the driver might play a small role, as well. Drag certainly plays a role but, again, that didnt stop the cars I tested from safely reaching their top speed. Also I did mention that I just eyeballed how the rate of acceleration decreases through the gears and I have yet to test this in my own simulation. I tried my best at tracking how quickly the speedometer needle moved through each gear and the difference got smaller and smaller between the higher gears. When checking what happens if you keep dividing by a value that only increases by 1 each time it made sense. 100% - 50% - 30% - 25% - 20%. So the first 2 gear changes show a huge difference and from that point on it happily accelerates at an average rate of 25%. The gear ratios are completely linear but the higher gears feel longer because the car is accelerating slower... I dont know if I should be disappointed or happy that it's as simple as this... There are no other factors stated in the handling file that directly change this outcome so this has got to be it. I found a test setup in which I can quite accurately measure drag and rolling resistance forces, free from other influences. In my model, drag is proportional to velocity squared (as in real life and as stated in the handling.dat file) and rolling resistance is proportional to the acting normal force resulting from the vehicle weight (thanks to blowntyre), and the test results leave no doubt that this is exactly how it's modelled in the game. I might create a separate thread some day to document this. If there are other influences, they are vanishingly small. These results lead me to believe that the game indeed models all those forces, so if your car reaches its maximum speed, it's because the sum of drag force and rolling resistance force equals the effective driving force (which means the drag is large). As to how acceleration works if gear shifts are involved, I'm not sure about that. My idea of it is that the shifting slows you down, and that acceleration is reduced under low revs, while under high revs there is increased drag, which might result in the sort of linearity you see. I'd probably start by analysing the acceleration characteristics of electric cars to eliminate the gear changes. |
|
|
|
Post by CHILLI on Oct 7, 2015 17:48:26 GMT
I found a test setup in which I can quite accurately measure drag and rolling resistance forces, free from other influences. In my model, drag is proportional to velocity squared (as in real life and as stated in the handling.dat file) and rolling resistance is proportional to the acting normal force resulting from the vehicle weight (thanks to blowntyre), and the test results leave no doubt that this is exactly how it's modelled in the game. I might create a separate thread some day to document this. If there are other influences, they are vanishingly small. Good point with the normal force. That's probably how the game gets the volume and pitch of the rolling sound aswell. For example going over the rumble strips on highways, bikes make next to no sound and trucks are really loud. So heavier vehicles would naturally have way more rolling friction/resistance... interesting indeed. Yea I didnt get to find an electric when I was testing acceleration vs. gear so I'll certainly take a look at that next time I'm on the game. The handling data states that a lot, if not all, of the electrics & hybrids use a "CVT" so their acceleration curve could look different for sure. And I'd be interested in hearing your take on it all if you were to make your own thread. This thread wasnt initially intended to delve into the math too much but more about how the values affect each other, so I'd say go ahead and start a new one. |
|
Pardonias
Member
Posts: 138
Registered on: December 2014
|
Post by Pardonias on Oct 8, 2015 8:52:06 GMT
CHILLI I have to apologise for claiming that the speedometers show the correct speed. While they did for many cars I tried (for example, Massacro, Comet, Insurgent, most of the super cars), the speedometer of the Sultan, for instance, only shows about 75% of the actual speed indeed. I assume that it's tied to the dashboard style, i.e. the speedometer of this dashboard is configured incorrectly and other cars with the same dashboard would have the same issue. Indeed, the Khamelion has the same interior and has the same problem.
|
|
|
|
Post by CHILLI on Oct 8, 2015 23:58:30 GMT
Pardonias Yea I think the angles of the needles are all the same so the textures will be out of alignment here and there. As far as I've found the majority of the cars show 75% on their speedos, hence why I hold on to that belief so strongly. For similiar reasons I dont care if my math isnt completely correct, hence why I dont look too far into it all because it's just too hard to get it nailed down perfectly. As long as it's consistently "close enough" in my tests, and tests suggested by others, I consider it worthy of being posted here. Also your apology is accepted. I think I've been a little defensive here and there in some of my more lengthy responses so I'll take this moment to apologize for that. I keep spotting mistakes like incorrect references and, as with the latest update, forgetting to include some stuff in my math examples. When I get my "Eureka!" moments I can very easily get carried away and forget things 
|
|
Nismo.
Member
Almost fully cross-platform scumbaggery
Posts: 52
Registered on: October 2015
|
Post by Nismo. on Nov 2, 2015 23:52:36 GMT
CHILLI I was fiddling around, trying to understand the effect offroads/tuners had on cars, and I went to check the sheet. It seems like all the vehicles which get a benefit from using offroads/tuners (smoother suspension) have a similar handling flag, always starting on 2 followed by other 4 digits, having a total of 5 numbers.  The flags are indeed read from the right to the left, and this can be confirmed by the Armored Kuruma. 8 on the 7th digit means "Armored", and in the handling file that 8 is the first digit. EDIT: After some searching, I was able to find the handling.cfg file which had the correct flag legend, and 2 in the 5th digit (counting from the right to the left) means "Tyres_Can_Clip".Now this explains a LOT, and it does make sense. With High Ends on a Feltzer you see some clipping, but it's only to the tyre and the game does know which part of the wheel is the rim and which part of the wheel is tyre, and it uses the same idea GTA IV did, it just did it's best to try to keep the tyre above the ground but it would eventually clip. This means, cars like the Comet and the Feltzer, which are almost the exact same on the paper (in terms of traction), because of that flag, are such different cars. The Comet doesn't get any benefit from offroads, while the Feltzer, and all the cars with that "2" on the 5th digit get all the same effect, and that is, tyres will clip under the ground for a smoother ride. Broughy1322 This might be fact finding worthy, but not just yet. I'm trying to find the file that contains the stats for the vehicle mods. No luck yet though... I've been using Se7ensins' forums to browse the X360's files, and it's quite confuse, and with a slow connection it's frustrating to navigate on the cloud. Any help is appreciated, but the PC side could find this much easier.
|
|
|
|
Post by CHILLI on Nov 5, 2015 15:01:37 GMT
EDIT: After some searching, I was able to find the handling.cfg file which had the correct flag legend, and 2 in the 5th digit (counting from the right to the left) means "Tyres_Can_Clip".. . . I'm trying to find the file that contains the stats for the vehicle mods. No luck yet though... . . . Any help is appreciated, but the PC side could find this much easier.
Yea, thanks for showing me that file Nismo. it explains a lot. For the longest time I've been wondering why cars like the Massacro, with next to no suspension travel, soak bumps like nothing compared to cars with longer travel like the Rapid GT but fly all over the place. No wonder why so many DLC cars take the top spots... And the vehicle customization data doesnt appear to be in a config anywhere. I've looked through the relevant folders on PC like 4 times in hope of finding it but all I come across is car color data in a different format that OpenIV cant yet open, aswell as all the model files. I'm really close to concluding that the car customization is hardcoded rather than generalized and stored in configs. I'll still keep my eyes open for the files, if they exist, but I'm having doubts.  These are the flags included in those files Nismo mentioned. The TYRES_CAN_CLIP flag is consistent with a lot of the cars that soak bumps, so there's little reason to doubt that this is correct. And maybe the REDUCED_MOD_MASS hints at armor, maybe amongst other parts, actually increasing weight? Dont be afraid of being slower though, as the vehicle should already be reaching the target performance regardless of mass. There are some other interesting flags to look at here, but as before I'll let you guys look into it as much as you want. The first post will get its old picture replaced with this one. |
|
Straight-line traction might play a part, as well.
