A lot of outright false stuff being said here, but it's a step forward. At least we have discussions nowadays. EDIT: I can rip into it a bit. 12:35, No. It depends on your wheel's peak output torque and what the sim is outputting. FFMULT needs to change based on your wheel's torque output, because every car will have a different "scale" of tie rod end forces. It's not as simple as "1.0 is 1:1" unless your wheel and car combination just happens to output the correct torque at 1.0. FFB TWEAKS "Output real steering forces to wheel" automatically does this if you have it setup right (Wheel peak Nm needs to be correct); but bear in mind that you would need to add a powersteering simulation for powersteering cars, otherwise you will get 20Nm+ torques out of a car that would have perhaps 5Nm assisted. FFMULT being set to a level where with 100% gain in driver and 100% gain in AC will result in a tiny bit of clipping mid-corner is not too bad of a reference point for most users. Typically from 1.5 to 3.0 or so. 15:50, No. Bumpsteer is when the steering angle of the wheel changes based on kinematic suspension travel. Although in some assemblies like struts where the spring exerts a high force onto the wheel when static, it will change the bumpsteer curve away from the kinematic one IIRC. However bumpsteer is not caused by tire forces per-se, it is a geometric thing based on the suspension travel. In AC it is entirely based on the suspension geometry. Bear in mind the SUSPENSIONS and TELEMETRY apps display wrong toe values in vanilla if the assembly has any caster; in CSP versions released lately (After mid-2022 or so) the setup screen toe value has been corrected and you should rather refer to that, using the "test_pad" map as it is flat. Or use a software like Suspension Analyzer, OptimumK, SHARK etc. Real front geometries usually exhibit some toe-out bumpsteer on compression (Perhaps 0.03 to 0.10deg per 20mm, but it is worse on some cars even in ideal geometry), and rear multilink/DWB geometries typically try to minimize it and prefer a toe-in bias. Don't worry about it *too* much because real cars will get a very large toe deflection in the magnitude of around 0.2 - 0.3degrees mid-corner due to elastokinematics, often towards toe-out on the outside wheel. Just get it within reasonable values and leave it; the real assembly will never be in design spec down to the 0.1mm anyway. 20:30, No. A big part of tire correlation is changing one little thing at a time. It's valid to just "add a bit of grip" and not touch anything else. In real tires if you change the material or construction of the tire there will often be more ramifications than just a small change to one aspect, so I understand what he means, but the point of parametrized tire models IS to be able to tweak individual aspects independently. If you are for example "converting" a 205mm width tire into a 255mm width tire you will need to "change everything" but it is a bit more nuanced. One real 205mm tire might have the same relaxation length and peak slip angle as another 255mm tire for example. Other things like load sensitivity are more often than not going to change. Tire springrate is also not nearly as important as it is made out to be, just do a ride frequency calculation comparing 200kN/m and 400kN/mm tires, or do an unsprung frequency calculation between them. However going way too low or way too high will affect bump compliance in extreme cases. It doesn't affect grip or heating or anything and you're not really going to feel it either. 26:30, No. AC is a brush model. No sim really uses Pacejka, nor is it really usable and there exist better models nowadays. 28:00, Correct, but every sim has this issue. Pressure is much more complicated than just "grip changes when you change pressure", it will affect camber sensitivity as well for example, to a quite large degree. No tire model, realtime or not, really does pressure right from what I know. However it would be nice to get a simple function for DX vs PSI based on another peak. For now you should just add roughly 10-15% longitudinal grip if you assume your longitudinal ideal will be perhaps 15PSI lower than your lateral ideal, run the car at a pressure where you will end up operating around that "longitudinal ideal" and then reset the car at the end of your runs. This is how I unfortunately have to do it for pro cars right now. 32:50, There is a whole suite of quite advanced physics features just waiting to be used but you can count the amount of people using them on one hand. Like he said though, userbase's knowledge needs to get up to the level to use them to begin with. More touched on at 38min or so. 36:50, Perhaps in the drift space, but in the serious sim space AC has never been better, and we know more than ever. However drifters are not terribly interested in an accurate simulation in my experience. 38:00, Antisquat, strut FFB etc. are already fixed in Cphys, you don't necessarily need to use Cosmic. It's probably worth just enabling the antisquat fix to be able to design in accurate rear geometries, even if you do nothing else. 55:50, DX/DY_REF is the MU value at FZ0. LS_EXPX/Y is the slope. I would avoid using these and really just use custom-input load curves. The issue with using the exponential load curves is that you have way too much grip at 0N and at beyond 6000N or so. It never produces an even moderately accurate shape. Pipeline has a simple tool to generate them based on some templates and I largely stick to it. 57:50, Load here means vertical load ie FZ. Not side load or anything. Load sensitivity of a tire will depend on its contact patch area/shape, construction/stiffness and to some degree the compound. Bigger tire, less load sensitivity is not a very crazy assumption. You can perhaps also use the load index as a rough guide. I find newer, sportier tires seem to have less slope on the longitudinal side compared to older tires just based on trying to correlate a bunch of cars. This is likely due to construction differences. 59:20, Most of my OEM tires are literally from the 80/90s. Most of my drift tires are from the early 2010s or so. The most modern tire I've made is probably the Kenda KR20A for a private client drift car and I estimated it to have about 8-10% more grip than similar older generation tires. Age of the tire does matter. However there is a good chance that the longitudinal grip in my tires is *still* not enough and the trend has been to increase it over time for better correlation, so it is a valid criticism that has a somewhat probable likelihood of being accurate. 1:06:20, There is a high likelihood that KS did not know what they were doing. I would be cautious of referencing anything too seriously. Some are really bad and used to correct large suspension issues.
This was a phenomenal video with so much to learn. I wonder if the DWG guys are interested in making pro drift cars. I wonder how they would compare to the VDC cars or the Tsujigiri cars.
Just wanted to say what's up, saw you were having a conversation with someone about tires in the comments to one of my videos. Cool to see a more in-depth discussion around it, watched as much as my goldfish attention span would allow, but really good video. Have been considering messing around with physics a bit on my own but know enough to realize that I'm way out of my depth lol. Especially after watching this, I'm just gonna sit here and wait on DWG to release that updated 350Z.
when i once inflated my tyres to abo9ut 3 psi more than standered to decrease rolling resistance my tures wore out very quickly on the middle of the tyre and had less wear towards the edges. but thats on flat bottom tyre some tyres have a slight rounded profile which means it would loose more grip per psi as it would stand up on the hump. some of the F1 tyres look like there rounded not perfectly flat , more psi would have a big difference to tyre contact patch per psi with a rounded profile tyre , and i agree with your tyre model you produce if someone alters it it cant really be classed as the same as the original authors. altering tires is a balanced science lots f numbers that affect other numbers. the tyre creator in content manager is good but when you make new tyres i don't know if it still includes the original author i haven't checked. and it can be complicated sometimes.
@@protomor Nah, I'm not talkative enough for that kind of stuff. Plus I don't know any of the super track makers that know a ton more than me and would be willing to do such a thing either. Hexor is the only one I know that might do that, but then you also have language barrier issues with that. Amazing video though, I watched the whole thing yesterday and got some good info even though I don't make cars in AC. Mainly for the reasons that were mentioned. I wouldn't be like others just taking work and slapping something together, I would be the guy that it takes a year per car to get done, not worth it to me
Nothing new here, but as i always used to say, it's all subjective, mostly. The era of diminishing returns has started. All cars feel the same to me, mostly, the basics of drifting is the same, dwg or whatever, tyres the same, i think most people get into way to many details, instead of actually drifting. It's all subjective, way to many variables to call anything "realistic". It will never work that way... And whoever thinks like that, is going to make mistakes. Honestly the dwg pack has weird steering, for me, not smooth lock to lock, tyres are to grippy in the front, i don't like them. Some other person will like it... It's all subjective, everything is moo at this point. More drifting, less chatting 😂
Damn, so this is why DWG cars are so good. The amount of sim x IRL knowledge that goes into this is crazy.
This interview is a hidden gem. More modders need to hear this.
We need that GS400 as a DWG car!!! 😻
And for something out the box, a Chevy Bel Air would be awesome too!
A lot of outright false stuff being said here, but it's a step forward. At least we have discussions nowadays.
EDIT: I can rip into it a bit.
12:35, No. It depends on your wheel's peak output torque and what the sim is outputting. FFMULT needs to change based on your wheel's torque output, because every car will have a different "scale" of tie rod end forces. It's not as simple as "1.0 is 1:1" unless your wheel and car combination just happens to output the correct torque at 1.0.
FFB TWEAKS "Output real steering forces to wheel" automatically does this if you have it setup right (Wheel peak Nm needs to be correct); but bear in mind that you would need to add a powersteering simulation for powersteering cars, otherwise you will get 20Nm+ torques out of a car that would have perhaps 5Nm assisted. FFMULT being set to a level where with 100% gain in driver and 100% gain in AC will result in a tiny bit of clipping mid-corner is not too bad of a reference point for most users. Typically from 1.5 to 3.0 or so.
15:50, No. Bumpsteer is when the steering angle of the wheel changes based on kinematic suspension travel. Although in some assemblies like struts where the spring exerts a high force onto the wheel when static, it will change the bumpsteer curve away from the kinematic one IIRC. However bumpsteer is not caused by tire forces per-se, it is a geometric thing based on the suspension travel. In AC it is entirely based on the suspension geometry.
Bear in mind the SUSPENSIONS and TELEMETRY apps display wrong toe values in vanilla if the assembly has any caster; in CSP versions released lately (After mid-2022 or so) the setup screen toe value has been corrected and you should rather refer to that, using the "test_pad" map as it is flat. Or use a software like Suspension Analyzer, OptimumK, SHARK etc. Real front geometries usually exhibit some toe-out bumpsteer on compression (Perhaps 0.03 to 0.10deg per 20mm, but it is worse on some cars even in ideal geometry), and rear multilink/DWB geometries typically try to minimize it and prefer a toe-in bias.
Don't worry about it *too* much because real cars will get a very large toe deflection in the magnitude of around 0.2 - 0.3degrees mid-corner due to elastokinematics, often towards toe-out on the outside wheel. Just get it within reasonable values and leave it; the real assembly will never be in design spec down to the 0.1mm anyway.
20:30, No. A big part of tire correlation is changing one little thing at a time. It's valid to just "add a bit of grip" and not touch anything else. In real tires if you change the material or construction of the tire there will often be more ramifications than just a small change to one aspect, so I understand what he means, but the point of parametrized tire models IS to be able to tweak individual aspects independently. If you are for example "converting" a 205mm width tire into a 255mm width tire you will need to "change everything" but it is a bit more nuanced. One real 205mm tire might have the same relaxation length and peak slip angle as another 255mm tire for example. Other things like load sensitivity are more often than not going to change.
Tire springrate is also not nearly as important as it is made out to be, just do a ride frequency calculation comparing 200kN/m and 400kN/mm tires, or do an unsprung frequency calculation between them. However going way too low or way too high will affect bump compliance in extreme cases. It doesn't affect grip or heating or anything and you're not really going to feel it either.
26:30, No. AC is a brush model. No sim really uses Pacejka, nor is it really usable and there exist better models nowadays.
28:00, Correct, but every sim has this issue. Pressure is much more complicated than just "grip changes when you change pressure", it will affect camber sensitivity as well for example, to a quite large degree. No tire model, realtime or not, really does pressure right from what I know.
However it would be nice to get a simple function for DX vs PSI based on another peak. For now you should just add roughly 10-15% longitudinal grip if you assume your longitudinal ideal will be perhaps 15PSI lower than your lateral ideal, run the car at a pressure where you will end up operating around that "longitudinal ideal" and then reset the car at the end of your runs. This is how I unfortunately have to do it for pro cars right now.
32:50, There is a whole suite of quite advanced physics features just waiting to be used but you can count the amount of people using them on one hand. Like he said though, userbase's knowledge needs to get up to the level to use them to begin with. More touched on at 38min or so.
36:50, Perhaps in the drift space, but in the serious sim space AC has never been better, and we know more than ever. However drifters are not terribly interested in an accurate simulation in my experience.
38:00, Antisquat, strut FFB etc. are already fixed in Cphys, you don't necessarily need to use Cosmic. It's probably worth just enabling the antisquat fix to be able to design in accurate rear geometries, even if you do nothing else.
55:50, DX/DY_REF is the MU value at FZ0. LS_EXPX/Y is the slope. I would avoid using these and really just use custom-input load curves. The issue with using the exponential load curves is that you have way too much grip at 0N and at beyond 6000N or so. It never produces an even moderately accurate shape.
Pipeline has a simple tool to generate them based on some templates and I largely stick to it.
57:50, Load here means vertical load ie FZ. Not side load or anything. Load sensitivity of a tire will depend on its contact patch area/shape, construction/stiffness and to some degree the compound. Bigger tire, less load sensitivity is not a very crazy assumption. You can perhaps also use the load index as a rough guide. I find newer, sportier tires seem to have less slope on the longitudinal side compared to older tires just based on trying to correlate a bunch of cars. This is likely due to construction differences.
59:20, Most of my OEM tires are literally from the 80/90s. Most of my drift tires are from the early 2010s or so. The most modern tire I've made is probably the Kenda KR20A for a private client drift car and I estimated it to have about 8-10% more grip than similar older generation tires. Age of the tire does matter.
However there is a good chance that the longitudinal grip in my tires is *still* not enough and the trend has been to increase it over time for better correlation, so it is a valid criticism that has a somewhat probable likelihood of being accurate.
1:06:20, There is a high likelihood that KS did not know what they were doing. I would be cautious of referencing anything too seriously. Some are really bad and used to correct large suspension issues.
appreciate all the corrections/context!
Dope talk.
This was a phenomenal video with so much to learn. I wonder if the DWG guys are interested in making pro drift cars. I wonder how they would compare to the VDC cars or the Tsujigiri cars.
Just wanted to say what's up, saw you were having a conversation with someone about tires in the comments to one of my videos. Cool to see a more in-depth discussion around it, watched as much as my goldfish attention span would allow, but really good video. Have been considering messing around with physics a bit on my own but know enough to realize that I'm way out of my depth lol. Especially after watching this, I'm just gonna sit here and wait on DWG to release that updated 350Z.
Thanks for the help btw! If you wanna collab or just chat about modding, lmk. I'm open to learning and spreading what I know.
fantastic video, every Assetto corsa drifter should watch this
thanks for the vid, very good information
when i once inflated my tyres to abo9ut 3 psi more than standered to decrease rolling resistance my tures wore out very quickly on the middle of the tyre and had less wear towards the edges. but thats on flat bottom tyre some tyres have a slight rounded profile which means it would loose more grip per psi as it would stand up on the hump. some of the F1 tyres look like there rounded not perfectly flat , more psi would have a big difference to tyre contact patch per psi with a rounded profile tyre , and i agree with your tyre model you produce if someone alters it it cant really be classed as the same as the original authors. altering tires is a balanced science lots f numbers that affect other numbers. the tyre creator in content manager is good but when you make new tyres i don't know if it still includes the original author i haven't checked. and it can be complicated sometimes.
Very informative video bro. I don't mess with cars in game and have no clue what most of that was lol, but very good info
You next? Maybe get a few more track makers?
@@protomor Nah, I'm not talkative enough for that kind of stuff. Plus I don't know any of the super track makers that know a ton more than me and would be willing to do such a thing either. Hexor is the only one I know that might do that, but then you also have language barrier issues with that. Amazing video though, I watched the whole thing yesterday and got some good info even though I don't make cars in AC. Mainly for the reasons that were mentioned. I wouldn't be like others just taking work and slapping something together, I would be the guy that it takes a year per car to get done, not worth it to me
@@almony8 Hehe understood. Lmk if you ever change your mind.
Btw I couldn’t find Shadow‘s Garage cars on Vosan or Race Department.
Are they paid mods?
He does the DWG cars. I think the shadow's stuff is paid mods.
This was way too informative. Almost need to listen to it in slomo
It's ok. just doors me and I'm good.
Nothing new here, but as i always used to say, it's all subjective, mostly. The era of diminishing returns has started. All cars feel the same to me, mostly, the basics of drifting is the same, dwg or whatever, tyres the same, i think most people get into way to many details, instead of actually drifting. It's all subjective, way to many variables to call anything "realistic". It will never work that way... And whoever thinks like that, is going to make mistakes. Honestly the dwg pack has weird steering, for me, not smooth lock to lock, tyres are to grippy in the front, i don't like them. Some other person will like it... It's all subjective, everything is moo at this point. More drifting, less chatting 😂