GTR 2's Wrong Coefficient of Friction

[Menrva]

It's amazing that 17 years after release, nobody has ever noticed that a specific parameter in the .HDC file for cars is responsible for weird collisions with track geometry we usually get to experience in-game.

I made a quick conversion of Recaro Raceway Park from GTR1's King of Ovals expansion (materials are rather broken so there are visual glitches). With the Saleen S7-R, I had unrealistic behavior when driving from the flatter inner oval to the heavily banked track; the car kept getting stuck at the beginning of the banking in a very rough way to the point that I couldn't get onto the track reasonably. In other cases the car would suddenly start flying after touching the slope change in some way at high speed.
As a test, I also had the conversion of GTR Evo's Saleen S7-R from another mod to compare it with GTR2's one. Surprisingly, GTR Evo's one worked like a charm! At first I thought it was the difference in the collision GMTs causing this, or the .TYR or .SP data, but I was so wrong. It's all due to a parameter in the .HDC file:

GTR2's Saleen S7-R: UndertrayParams=(262500.0,11600.0, 4.0) // spring rate, damper rate, and coefficient of friction
GTR Evolution's Saleen S7-R: UndertrayParams=(100000, 5000, 2.0)

With GTR Evo's values, it finally felt a lot more natural, I could get onto the banked track without any issues like weird jumps/crashes, getting stuck or losing much speed on elevation changes. This also explains why I lose much speed on Spa's steep Raidillon while the AI is obviously unaffected; such high friction coefficient makes the bodywork and tyres contact too harshly with the elevation change. The difference is like night and day!

The last value, the coefficient of friction, is the one that affects this behavior the most. Other gMotor games use lower values. For instance, default rFactor content uses a value 2.0 as well as SimBin's Race 07 content. Discussing this on GTR2's Crew Chief plugin Discord server, other users have shown that stock content on Automobilista even uses lower values than 2. Damper and spring values are also worth looking into to better understand what difference they make.

This excessive value of 4 is also in the first GTR game, while GT Legends uses a value of 3. I highly suggest that the value be brought at least to 2. I've been using GTR233's HQ 2022 physics and the UndertrayParams are the same as the stock content. I cannot tell about other physics mods yet, but I really suggest this tweak to avoid weird contact between the surface and the bodywork and tyres of a car.

Here below two quick videos I recorded about it, showing the difference the coefficient of friction makes:

Recaro Raceway Park is an extreme example, but the issue can be replicated under certain angles and speeds even on stock GTR2 tracks. Apologies for the rendering glitches of the track, I'm just learning how to convert tracks to GTR2 and I'd appreciate any help.

 

[Menrva]

Just in case anyone wishes to fix it for countless mods, you can use Double Commander to quickly standardize the UndertrayParams line in all HDC files in an instant. Here's how:
Go to Search... (ALT+F7) and put the following data in:

Filter file: *.hdc
Search text in file: (?-s)UndertrayParams(.*,.*,.*)
Replace with: UndertrayParams=(250000.0,12500.0,2.0) // spring rate, damper rate, and coefficient of friction

Enable the Regular expression flag or it won't work obviously.
I took the spring and damper rates from the WTCC cars of Race 07. You can experiment with different values of course and see what feels more natural to you.

 

[jgf]

This is also useful for editing many cars at once-

GTR2 MultiEdit Tool

Again, not my work. Just making it available again. "It basically works this way: You mark base GTR2 folder, it scans for all CAR and HDC files, you then select several CAR or HDC files and in Change Key Value window you set which value you...

www.racedepartment.com

 

[dwightyone]

Of course been reading your posts about this in Crew Chief's GTR2 discord - will have a fiddle myself in due course and can't wait to see other peoples opinion of this.

 

[Speednut357]

Thank you for writing up your findings so I can learn.

 

[Menrva]

The info on this site may explain why SimBin chose a value of 4 but it's a mistake: https://vehiclephysics.com/blocks/tires/

I believe a more realistic value would be between 1 and 2. Some Reiza content uses 0.35 which seems too low. I'll probably test AMS' spring and damper rate values with a coefficient of friction of 1.35 or 1.5: UndertrayParams=(150000,2500,1.35)

 

[jgf]

Every sim of my experience in the last quarter century has fudged the physics in some way to "improve gameplay". Even the venerable GPL made some tracks wider and smoother than reality for "better racing".

The last decade has seen a shift in gaming, most players today have neither the interest nor patience to learn anything, with a race sim they do not want to take time to learn how to setup and drive the car, they just want to download setups, hop in their favorite car and win every race, then move on to the next "latest greatest" product. The NASCAR Racing series was killed when Vivendi wanted to dumb down the sim to attract a larger audience and Kaemmer refused. In FlightSim there's no need to learn engine management or navigation, just hop in your favorite plane, take off, and there are pointers around the landscape to show where everything is located. I once tried a farming "sim" that didn't even have seasons.

There was a time when "sim" implied a realistic depiction of something done in real life, conveyed as accurately as the computer would allow. Today the term merely means doing something on the computer that is done in real life, and realism be damned. Add in a world of "achievements", points, ranking, and the sim is pushed further from reality.

Be wary of comparing values from different sims, even if both ostensibly use "real world" figures it's not definite that their individual physics process that data the same way.

 

[Menrva]

Be wary of comparing values from different sims, even if both ostensibly use "real world" figures it's not definite that their individual physics process that data the same way.

I'm well aware of this and I hope that my posts do not spark an uncostructive debate or flame war like the NAP physics mod did at its time (it's not even comparable, this is just about three values).
I felt it was worth pointing out, especially since many physics mods for GTR2 and also mods adding other types of cars to GTR2 seem to neglect UndertrayParams by using stock GTR values which in my view do not feel natural at all compared to what I feel in other gMotor games. It's up to users whether to consider if GTR2 deserves tweaks to this. I know there are purists who love GTR2 as is, others who don't, but I'm not interested in that kind of discussion.

 

[jgf]

Lol, I well remember the NAP controversy. Purists form two groups - those who advocate maintaining everything default, which is good for open online leagues, and those who want realism via any amount of tweaking. I'm thoroughly in favor of any experimentation that improves our knowledge of the physics, but don't be surprised if I question how someone arrives at their conclusions. Often feel, with any sim, I spend as much, maybe more, time tweaking as driving.

 

[Durge Driven]

Wouldn't it be better to make sets of HDC you activate with a GME to suit the track situation.

You could make a dozen, they take no space to speak of and activate 1 click.

 

[JB57]

It's amazing that 17 years after release, nobody has ever noticed that a specific parameter in the .HDC file for cars is responsible for weird collisions with track geometry we usually get to experience in-game.

I made a quick conversion of Recaro Raceway Park from GTR1's King of Ovals expansion (materials are rather broken so there are visual glitches). With the Saleen S7-R, I had unrealistic behavior when driving from the flatter inner oval to the heavily banked track; the car kept getting stuck at the beginning of the banking in a very rough way to the point that I couldn't get onto the track reasonably. In other cases the car would suddenly start flying after touching the slope change in some way at high speed.
As a test, I also had the conversion of GTR Evo's Saleen S7-R from another mod to compare it with GTR2's one. Surprisingly, GTR Evo's one worked like a charm! At first I thought it was the difference in the collision GMTs causing this, or the .TYR or .SP data, but I was so wrong. It's all due to a parameter in the .HDC file:

GTR2's Saleen S7-R: UndertrayParams=(262500.0,11600.0, 4.0) // spring rate, damper rate, and coefficient of friction
GTR Evolution's Saleen S7-R: UndertrayParams=(100000, 5000, 2.0)

With GTR Evo's values, it finally felt a lot more natural, I could get onto the banked track without any issues like weird jumps/crashes, getting stuck or losing much speed on elevation changes. This also explains why I lose much speed on Spa's steep Raidillon while the AI is obviously unaffected; such high friction coefficient makes the bodywork and tyres contact too harshly with the elevation change. The difference is like night and day!

The last value, the coefficient of friction, is the one that affects this behavior the most. Other gMotor games use lower values. For instance, default rFactor content uses a value 2.0 as well as SimBin's Race 07 content. Discussing this on GTR2's Crew Chief plugin Discord server, other users have shown that stock content on Automobilista even uses lower values than 2. Damper and spring values are also worth looking into to better understand what difference they make.

This excessive value of 4 is also in the first GTR game, while GT Legends uses a value of 3. I highly suggest that the value be brought at least to 2. I've been using GTR233's HQ 2022 physics and the UndertrayParams are the same as the stock content. I cannot tell about other physics mods yet, but I really suggest this tweak to avoid weird contact between the surface and the bodywork and tyres of a car.

Here below two quick videos I recorded about it, showing the difference the coefficient of friction makes:

Recaro Raceway Park is an extreme example, but the issue can be replicated under certain angles and speeds even on stock GTR2 tracks. Apologies for the rendering glitches of the track, I'm just learning how to convert tracks to GTR2 and I'd appreciate any help.

Hello,
I tested with two of my favorite car (setting 2.0). I can say that i have a better feeling of the road now, and my ffb are much subtitle.
Thanks for your post.
Best Regards Joseph from Alsace.

 

[Need_For_Speed]

It's amazing that 17 years after release, nobody has ever noticed that a specific parameter in the .HDC file for cars is responsible for weird collisions with track geometry we usually get to experience in-game.

I made a quick conversion of Recaro Raceway Park from GTR1's King of Ovals expansion (materials are rather broken so there are visual glitches). With the Saleen S7-R, I had unrealistic behavior when driving from the flatter inner oval to the heavily banked track; the car kept getting stuck at the beginning of the banking in a very rough way to the point that I couldn't get onto the track reasonably. In other cases the car would suddenly start flying after touching the slope change in some way at high speed.
As a test, I also had the conversion of GTR Evo's Saleen S7-R from another mod to compare it with GTR2's one. Surprisingly, GTR Evo's one worked like a charm! At first I thought it was the difference in the collision GMTs causing this, or the .TYR or .SP data, but I was so wrong. It's all due to a parameter in the .HDC file:

GTR2's Saleen S7-R: UndertrayParams=(262500.0,11600.0, 4.0) // spring rate, damper rate, and coefficient of friction
GTR Evolution's Saleen S7-R: UndertrayParams=(100000, 5000, 2.0)

With GTR Evo's values, it finally felt a lot more natural, I could get onto the banked track without any issues like weird jumps/crashes, getting stuck or losing much speed on elevation changes. This also explains why I lose much speed on Spa's steep Raidillon while the AI is obviously unaffected; such high friction coefficient makes the bodywork and tyres contact too harshly with the elevation change. The difference is like night and day!

The last value, the coefficient of friction, is the one that affects this behavior the most. Other gMotor games use lower values. For instance, default rFactor content uses a value 2.0 as well as SimBin's Race 07 content. Discussing this on GTR2's Crew Chief plugin Discord server, other users have shown that stock content on Automobilista even uses lower values than 2. Damper and spring values are also worth looking into to better understand what difference they make.

This excessive value of 4 is also in the first GTR game, while GT Legends uses a value of 3. I highly suggest that the value be brought at least to 2. I've been using GTR233's HQ 2022 physics and the UndertrayParams are the same as the stock content. I cannot tell about other physics mods yet, but I really suggest this tweak to avoid weird contact between the surface and the bodywork and tyres of a car.

Here below two quick videos I recorded about it, showing the difference the coefficient of friction makes:

Recaro Raceway Park is an extreme example, but the issue can be replicated under certain angles and speeds even on stock GTR2 tracks. Apologies for the rendering glitches of the track, I'm just learning how to convert tracks to GTR2 and I'd appreciate any help.

Good observation. This was also changed in my GET REAL physics mod so it uses GTR EVO values but for coefficient of friction only. I dont know about spring and damper rate here GTR EVO values seem suspicious, while GTR2 are more specific judging by the numbers. But in addition to this I have also corrected undetray values so they better match physical car dimensions.

 

[Menrva]

Good observation. This was also changed in my GET REAL physics mod so it uses GTR EVO values but for coefficient of friction only. I dont know about spring and damper rate here GTR EVO values seem suspicious, while GTR2 are more specific judging by the numbers. But in addition to this I have also corrected undetray values so they better match physical car dimensions.

Hey there! I've been trying your physics mod and I'm truly impressed, it's the best I ever experienced so far! My thanks to you for an incredible physics mod! AI lap times are so correct. On my end I tweaked Road and AI grip values for each track's .gdb, IMHO the values you gave are proper for the AI, while for the player they should always be set to 1.0 and 0.75 for Wet. I mention this because some tracks are too slippery like Anderstorp and I was lapping several seconds slower compared to other tracks where 1.0 is for both AI and the player.

Back to the topic, I'm currently using these UndertrayParams values: (250000.0,12500.0,1.0).
I took spring and damper values from Race 07's WTCC cars, they also seem to be closer to GTR2's values compared to GTR Evo cars. Friction is set to 1 and it feels very nice, I can feel curbs and lose grip on them without any weird contact like on stock GTR2.

 

[Need_For_Speed]

On my end I tweaked Road and AI grip values for each track's .gdb, IMHO the values you gave are proper for the AI, while for the player they should always be set to 1.0 and 0.75 for Wet. I mention this because some tracks are too slippery like Anderstorp and I was lapping several seconds slower compared to other tracks where 1.0 is for both AI and the player.

The reason for some tracks having lower coefficents is because of lower grip levels IRL. Not all tracks had the same grip obviously so having a coefficent of 1 would be unrealistic. Wet works different however because it multiplies dry track grip with 0,75 so it should always be that number. You have mentioned Anderstorp and yes it was by far the lowest grip track on calendar because of its nature. It was a very old track with worn out asphalt. I dont know if that is still the case but in 2003 definitely.

 

[Menrva]

The reason for some tracks having lower coefficents is because of lower grip levels IRL. Not all tracks had the same grip obviously so having a coefficent of 1 would be unrealistic. Wet works different however because it multiplies dry track grip with 0,75 so it should always be that number. You have mentioned Anderstorp and yes it was by far the lowest grip track on calendar because of its nature. It was a very old track with worn out asphalt. I dont know if that is still the case but in 2003 definitely.

I thought as much and I don't question why you did it, it makes a lot of sense. I wonder what else I could do car setup wise to get better grip when it's so low. No matter what I do with the setup, I've tried a lot of settings, but on Anderstorp I can't seem to have a fair fight against the AI, they are several seconds faster. For instance, with the Saleen my best was 1.34 but the AI could make 1.29 or 1.28, which are realistic times for them. With RoadGrip to 1.0 I could do a 1.30, and that matches more closely the same performance I achieve against the AI on tracks with higher grip.

That being said, I'm nitpicking. I love your mod and there's nothing else I'd change. Don't take my post negatively. It's most likely I'm bad as a driver and at getting setups right.

 

[Need_For_Speed]

I thought as much and I don't question why you did it, it makes a lot of sense. I wonder what else I could do car setup wise to get better grip when it's so low. No matter what I do with the setup, I've tried a lot of settings, but on Anderstorp I can't seem to have a fair fight against the AI, they are several seconds faster. For instance, with the Saleen my best was 1.34 but the AI could make 1.29 or 1.28, which are realistic times for them. With RoadGrip to 1.0 I could do a 1.30, and that matches more closely the same performance I achieve against the AI on tracks with higher grip.

That being said, I'm nitpicking. I love your mod and there's nothing else I'd change. Don't take my post negatively. It's most likely I'm bad as a driver and at getting setups right.

No problem, I just wanted to point that grip is not a constant value but changes from track to track.
Also new update is coming to my mod which will bring AI and player laptimes closer in general, due to revised tire data for AI and player tires. Currently AI peak slip values are to low and that is why they are able to lap faster. New update will close the gap while retaining realistic lap times as well as making front tires perform even better then before.

 

[jgf]

took spring and damper values from Race 07's WTCC cars, they also seem to be closer to GTR2's values compared to GTR Evo cars

The three sims by default depict different series so you would expect different values for the parameters. Using the settings from one in another may "feel" better but is it actually more accurate? (Not saying it is or isn't, but without seat time in the real cars involved it is quite subjective.)

 

[Need_For_Speed]

The three sims by default depict different series so you would expect different values for the parameters. Using the settings from one in another may "feel" better but is it actually more accurate? (Not saying it is or isn't, but without seat time in the real cars involved it is quite subjective.)

I think that it doesn't matter what king of series there was, since all racecars have flat bottom made of the same material beeing WTCC or GTs. I would be surprised if that was not the case.

 

[Menrva]

The three sims by default depict different series so you would expect different values for the parameters. Using the settings from one in another may "feel" better but is it actually more accurate? (Not saying it is or isn't, but without seat time in the real cars involved it is quite subjective.)

Everything can be subjective. Explain me why GTR Evo's cars use UndertrayParams=(100000, 5000, 2.0) while GTR 2 counterparts of the same cars use UndertrayParams=(262500.0,11600.0, 4.0). GTR Evo ones seem much easier and more floaty, hence I proposed the other values SimBin applied to WTCC cars, which are very close to GTR2's stock string and damper rates. All GTR1, GTR2 and GT Legends cars share the same spring and damper rates, but we're talking of cars of different generations, GTs from the 60s compared to early 90s/early 2000s ones. Not to mention that cars in GT Legends are very varied, from a sleek Porsche 906 to an Abarth 1000 TC. Even the devs didn't seem to have a clear clue as to which UndertrayParams values are more accurate and realistic, they just chose to stuck to something equal for everything.
To tweak those values doesn't hurt as they do not alter major physics properties, I'm just inviting people to experiment more rather than to accept things as they are. I'm not claiming my choice is correct nor that it makes everything extremely realistic. It certainly makes the cars behave closer to what I get in other gMotor games. As NFS pointed out, we're talking about the bottom of the cars and I don't expect a difference in materials between WTCC and GT cars. It should be a different case with high-end open wheelers.

 

[jgf]

Explain me why GTR Evo's cars use UndertrayParams=(100000, 5000, 2.0) while GTR 2 counterparts of the same cars use UndertrayParams=(262500.0,11600.0, 4.0).

Because those sims are representing different series, with different specs and regulations for the cars?