8Ball
Registered
All ISI, S397 and mods have always had large variables in AI strength, aggression and AI gaps
From the very start first build I had tune Howston and Open Wheel.
Like I said though this depends upon someone's personal performance across all classes.
I am in no shape or form a classic racer can jump in anything and race well, far from it.
I older and like simple cars, steer throttle brake so I grapple with modern tech more then most.
Which puts you at a disadvantage in these type of cars which means you need slower AI to enjoy a race and see others cars.
F3 I can race at 120% some tracks because I have spent lot of time in them.
To use 120% with them you must lap for example Belgium 4:16 Italy 1:48 with no tow. Try that for staters Out of over 300 drivers frequented my room only 16 managed no tow 4:16 and less did 1:48 at Italy.
But take say Loch and Metro Toban others I am hopeless I just don't put in the laps needed so I run reduced AI.
So no matter what 100% is set across all vehicles your average person will need to lower and raise them according to pace in each car.
I mean you want the AI at a level they all enjoyable to race against.
Yes Mazda has fast AI but so do other mods, again what is hard for me may be easy for you.
Going below 100% casts no aspersions to your driving skill.
Some people run some car at 110% may crash out more then 100%
Every single F3 Eve race I do today is just as immersive to me as 10 years ago
I am not saying every car should not be tweaked better but it would accomplish little.
Really the beauty in rF2 for me is finding the combos that work for me.
Why do I want to spend hours struggling rather then enjoying.
But then again a twist is some people like hard AI so it seems impossible to finish up order.
The physics and AI learning are more dynamic then any other sim you can name hands down. That is why gmotor always put strain on CPU + car numbers because of the Realtime computations. i know you know just saying for anyone listening has never realized this.
I don't know how you could not if you have studied AMS2 ACC AC AI races. It is like chalk and cheese to me.
That is not to say ISIMotor is correct far from it.
Only way we will get better consistent AI is for studios to use same yet to made super engine. Different engines just breeds different behavior, no calculations are the same not even tyres.
imho we should be able to drive all sims all cars feel same, drive the same, they don't, they never will till industry does something about it. Like say going to new car dealer and having to drive 5 identical cars to see which one is better. lol
Why people saying S397 should just make rFactor 3 is a complete joke to me.
AS great as ISIMotor was for it's time it is simply flawed like all other engines were AI strength is the least problem.
But studios will keep dolling out the same crap over and over no vision for the future I can see.
ranted out lol
From the very start first build I had tune Howston and Open Wheel.
Like I said though this depends upon someone's personal performance across all classes.
I am in no shape or form a classic racer can jump in anything and race well, far from it.
I older and like simple cars, steer throttle brake so I grapple with modern tech more then most.
Which puts you at a disadvantage in these type of cars which means you need slower AI to enjoy a race and see others cars.
F3 I can race at 120% some tracks because I have spent lot of time in them.
To use 120% with them you must lap for example Belgium 4:16 Italy 1:48 with no tow. Try that for staters Out of over 300 drivers frequented my room only 16 managed no tow 4:16 and less did 1:48 at Italy.
But take say Loch and Metro Toban others I am hopeless I just don't put in the laps needed so I run reduced AI.
So no matter what 100% is set across all vehicles your average person will need to lower and raise them according to pace in each car.
I mean you want the AI at a level they all enjoyable to race against.
Yes Mazda has fast AI but so do other mods, again what is hard for me may be easy for you.
Going below 100% casts no aspersions to your driving skill.
Some people run some car at 110% may crash out more then 100%
Every single F3 Eve race I do today is just as immersive to me as 10 years ago
I am not saying every car should not be tweaked better but it would accomplish little.
Really the beauty in rF2 for me is finding the combos that work for me.
Why do I want to spend hours struggling rather then enjoying.
But then again a twist is some people like hard AI so it seems impossible to finish up order.
The physics and AI learning are more dynamic then any other sim you can name hands down. That is why gmotor always put strain on CPU + car numbers because of the Realtime computations. i know you know just saying for anyone listening has never realized this.
I don't know how you could not if you have studied AMS2 ACC AC AI races. It is like chalk and cheese to me.
That is not to say ISIMotor is correct far from it.
Only way we will get better consistent AI is for studios to use same yet to made super engine. Different engines just breeds different behavior, no calculations are the same not even tyres.
imho we should be able to drive all sims all cars feel same, drive the same, they don't, they never will till industry does something about it. Like say going to new car dealer and having to drive 5 identical cars to see which one is better. lol
Why people saying S397 should just make rFactor 3 is a complete joke to me.
AS great as ISIMotor was for it's time it is simply flawed like all other engines were AI strength is the least problem.
But studios will keep dolling out the same crap over and over no vision for the future I can see.
ranted out lol
Last edited:
8Ball
Registered
I mean I agree 101%
All AI combos should be graded so 100% reflects real lap time and 120% the fastest lap ever driven. Making all this data corelate is another thing.
To me it is the same as track rules, timing, penalties being all over the place across sims.
Imagine how much processing power IMSA or FiA need to do this.
We want desktop PC to do the same thing, not happening in my life
Even if you could design a engine each car had Traces, GPS, Transponders etc. you would never power it.
On top of that you have real time human intervention making decisions on the fly that most surely not all agree with.
" Oh they brought out that Safety Car way too early in my opinion " etc.
Hear it all the time real races.
You would need AI learning for stewards lol
All AI combos should be graded so 100% reflects real lap time and 120% the fastest lap ever driven. Making all this data corelate is another thing.
To me it is the same as track rules, timing, penalties being all over the place across sims.
Imagine how much processing power IMSA or FiA need to do this.
We want desktop PC to do the same thing, not happening in my life
Even if you could design a engine each car had Traces, GPS, Transponders etc. you would never power it.
On top of that you have real time human intervention making decisions on the fly that most surely not all agree with.
" Oh they brought out that Safety Car way too early in my opinion " etc.
Hear it all the time real races.
You would need AI learning for stewards lol
Last edited:
Kahel
Registered
Promising mod...
But go on a track like daytona and you'll know something wrong with the car at high speed, especially on banking... it's like the car keeps on balancing one side to the other... very hard to keep 'straight'...
And I'm not the only one noticing it... a lot of people I talked to in my league say the same thing (using various hardware...)
(Edit: one important precision... most people run their DD with heavily filtered FFB which could explain why some doesn't feel the issue... but everyone I know that run unfiltered signal... can feel something wrong for sure...)
But go on a track like daytona and you'll know something wrong with the car at high speed, especially on banking... it's like the car keeps on balancing one side to the other... very hard to keep 'straight'...
And I'm not the only one noticing it... a lot of people I talked to in my league say the same thing (using various hardware...)
(Edit: one important precision... most people run their DD with heavily filtered FFB which could explain why some doesn't feel the issue... but everyone I know that run unfiltered signal... can feel something wrong for sure...)
Last edited:
I love everything except a very important aspect in handling for me:
While trail braking, in the initial phase of weight transfer the car feels like iracing with even better ffb: big weight transfer and you feel the car can go easily in a unstable state if you aren't smooth on the input.
But then even if you brake too much while turning, the car will never oversteer, it's like past a certain slip angle the front tyres get less grip then back tyres, no matter how much you trail brake and how brutal you are on you inputs. So you never find the neutral steer phase where you can keep the wheels straight and turn with tje brake.
Couldnt get it better with setups. Is it just me because I don't see any other comment like that ?
While trail braking, in the initial phase of weight transfer the car feels like iracing with even better ffb: big weight transfer and you feel the car can go easily in a unstable state if you aren't smooth on the input.
But then even if you brake too much while turning, the car will never oversteer, it's like past a certain slip angle the front tyres get less grip then back tyres, no matter how much you trail brake and how brutal you are on you inputs. So you never find the neutral steer phase where you can keep the wheels straight and turn with tje brake.
Couldnt get it better with setups. Is it just me because I don't see any other comment like that ?
Nick9320
Registered
I agree on that one but at the same time the car react on the setup changes pretty good, so I think it depends on track and driving style.
E.g. we with my teammate had an online races on Laguna Seca and Croft recently with this car. At Laguna it was extremely easy to play with weight transfer in turn 6 and with a small tap on the brake pedal you could send the car in a small oversteer. At Croft it is possible to spin the car quickly in the last turn with heavy braking and quick downshift.
Also setup-wise I would recommend to play a bit with dampers, it affects the braking and cornering a lot. So you won’t get away from the understeer at the trail braking but at least you can minimize the effect and with less weight at the front of the car you can get a slight oversteer or at least put the car on a racing line pretty much quick
. Thanks again.I'll drop this here, from the MX-5 User GuideAlso setup-wise I would recommend to play a bit with dampers, it affects the braking and cornering a lot. So you won’t get away from the understeer at the trail braking but at least you can minimize the effect and with less weight at the front of the car you can get a slight oversteer or at least put the car on a racing line pretty much quick
David O'Reilly
Registered
Thanks, that was me. Standing next to my NC Race car.
David O'Reilly
Registered
Gents, thank you for your hard work on this mod. It looks and sounds great.
I race an NC MX5 Cup car IRL and was involved in the physics testing and setup of Marcoboosts NC1 MX5.
I hope you don't mind me giving some initial feedback on this mod after a few laps at Oulton Park.
I'm going to try some damper adjustments later but here's my first feedback.
-Basic Balance;
In my view it has a "hard wired in" tendency to oversteer on entry and understeer on exit. (this was present in the first version of the NC1 I tested too).
No matter how hard you stomp on the throttle the rear grip remains solid and this is witnessed by the rear tyre temperatures remaining much cooler than the fronts. This BTW was on the minimum diff preload. The result in the game is that any amount of oversteer can be cured by throttle application. The real life MX5 is much more balanced, its mid corner balance is more neutral and it can be teased into oversteer quite easily. The rear tyre temps are the ones you need to manage IRL. So it's my view that the balance needs some tweaking. I couldn't effectively change the balance because on corner entry I would want a stiffer front end (to reduce entry oversteer) and the springs are on maximum, then I would want to stiffen the rear for more realistic behaviour on throttle but the rear ARB was on maximum, preload on minimum and the rear springs are not adjustable. So I couldn't get the rear end moving (to help rotate the car on throttle).
Other observations: (some very minor)
The water temp was sitting at 20 degrees with oil at 88 deg.
The brake balance is biased too far to the rear so they lock up too easily provoking oversteer which exacerbates the above problem.
In real life these cars have on board brake bias adjustment.
IRL these cars don't have a pit limiter.
I hope you can take these as constructive points as I would love to see this car evolve into something special.
The real life BRSCC MX5 Cup and MX5 Supercup grid race a ND Miata in winter but currently use iRenting. I'd like to see them in RF2.
Cheers
David
EDIT:
Because a man with an opinion and no data to support it is a man with an opinion,
I went to the Nazareth oval to test damper changes. I set rear slow bump to stiffest, front slow bump to softest, also added some front toe. These changes maximised the potential for oversteer. Given at such an oval cars will tend to scrub the right front a bit with a road course style "neutral setup", it was still impossible to provoke oversteer on throttle. Again this is witnessed by the contrasting tyre temps (and wear). It should be possible with maxed out settings to get some heat into the right rear relative to the right front-I couldn't.
Ill attach two pics, one from cockpit after 2 laps, the other a motec trace from same laps.
I race an NC MX5 Cup car IRL and was involved in the physics testing and setup of Marcoboosts NC1 MX5.
I hope you don't mind me giving some initial feedback on this mod after a few laps at Oulton Park.
I'm going to try some damper adjustments later but here's my first feedback.
-Basic Balance;
In my view it has a "hard wired in" tendency to oversteer on entry and understeer on exit. (this was present in the first version of the NC1 I tested too).
No matter how hard you stomp on the throttle the rear grip remains solid and this is witnessed by the rear tyre temperatures remaining much cooler than the fronts. This BTW was on the minimum diff preload. The result in the game is that any amount of oversteer can be cured by throttle application. The real life MX5 is much more balanced, its mid corner balance is more neutral and it can be teased into oversteer quite easily. The rear tyre temps are the ones you need to manage IRL. So it's my view that the balance needs some tweaking. I couldn't effectively change the balance because on corner entry I would want a stiffer front end (to reduce entry oversteer) and the springs are on maximum, then I would want to stiffen the rear for more realistic behaviour on throttle but the rear ARB was on maximum, preload on minimum and the rear springs are not adjustable. So I couldn't get the rear end moving (to help rotate the car on throttle).
Other observations: (some very minor)
The water temp was sitting at 20 degrees with oil at 88 deg.
The brake balance is biased too far to the rear so they lock up too easily provoking oversteer which exacerbates the above problem.
In real life these cars have on board brake bias adjustment.
IRL these cars don't have a pit limiter.
I hope you can take these as constructive points as I would love to see this car evolve into something special.
The real life BRSCC MX5 Cup and MX5 Supercup grid race a ND Miata in winter but currently use iRenting. I'd like to see them in RF2.
Cheers
David
EDIT:
Because a man with an opinion and no data to support it is a man with an opinion,
I went to the Nazareth oval to test damper changes. I set rear slow bump to stiffest, front slow bump to softest, also added some front toe. These changes maximised the potential for oversteer. Given at such an oval cars will tend to scrub the right front a bit with a road course style "neutral setup", it was still impossible to provoke oversteer on throttle. Again this is witnessed by the contrasting tyre temps (and wear). It should be possible with maxed out settings to get some heat into the right rear relative to the right front-I couldn't.
Ill attach two pics, one from cockpit after 2 laps, the other a motec trace from same laps.
Last edited:
Nick9320
Registered
That's very helpful! Thanks for the feedback and some decent explanation!
Chris Lesperance
Registered
I'm assuming you may have all the info already. But just in case. Flis Performance was tasked with creating the cars. I did come across the homologation pdf. Will need to do research with the parts to put into the physics calculator.
https://flisperformance.com/wp-cont...p_Vehicle_Homologation_Version_2019_03_06.pdf
I thought they had more files available in the past. One crack I've found for a few of my projects has been to use the Wayback Machine. Using a site, click on the URL tab, and then search .pdf. You might be able to get lucky with some files that provide some good information.
The NC MX5 has been starting to take off as of late. I just attended an SCCA race this weekend and there were several Spec MX5 cars in attendance. There is an rF1/AMS1/GTR2 mod that would make for a good base I think.
https://www.overtake.gg/downloads/ams-2006-mazda-mx5-cup.25048/
https://esport-racing.de/hankook-mx-5-cup/
Lots of individual info here:
https://specmx-5.com/technical-information-for-cars/
I doubt I would be able to get enough time to convert it. But there is info out there for it if someone would want to get some motivation to do it.
https://flisperformance.com/wp-cont...p_Vehicle_Homologation_Version_2019_03_06.pdf
I thought they had more files available in the past. One crack I've found for a few of my projects has been to use the Wayback Machine. Using a site, click on the URL tab, and then search .pdf. You might be able to get lucky with some files that provide some good information.
The NC MX5 has been starting to take off as of late. I just attended an SCCA race this weekend and there were several Spec MX5 cars in attendance. There is an rF1/AMS1/GTR2 mod that would make for a good base I think.
https://www.overtake.gg/downloads/ams-2006-mazda-mx5-cup.25048/
https://esport-racing.de/hankook-mx-5-cup/
Lots of individual info here:
https://specmx-5.com/technical-information-for-cars/
I doubt I would be able to get enough time to convert it. But there is info out there for it if someone would want to get some motivation to do it.
J
JEmilio
Guest
I've just tried this mod and I must say it's fantastic to drive this little monster around the beauty of rFactor, so thank you very much for your job and dedication.
One thing I don't like is the brake pressure setting. I like to drive without ABS and without it I couldn't apply more than 40% of pressure without the rears locking up without warning causing me to end in the barriers. So it could be nice if one would be able to set less brake pressure in the setup than the minimum possible at the moment which I think is 95%, if I remember correctly.
Another thing I noticed is the water temperature staying very low, I don't know if it's wether a bug or the way is intended to function.
One thing I don't like is the brake pressure setting. I like to drive without ABS and without it I couldn't apply more than 40% of pressure without the rears locking up without warning causing me to end in the barriers. So it could be nice if one would be able to set less brake pressure in the setup than the minimum possible at the moment which I think is 95%, if I remember correctly.
Another thing I noticed is the water temperature staying very low, I don't know if it's wether a bug or the way is intended to function.
J
JEmilio
Guest
I agree that the ai is too strength compared to other mods and to official content. I did a trackday at Autodrom Most and I barely could beat the ai by half a second at 100% when I usually can beat them by 2.5-3 seconds on average, especially in such long tracks. Also considering the amount of effort I was putting should've been enough to pull at least 1.5 seconds.
Nick9320
Registered
Is that a bad thing really? Right now the AI gives you a proper competition on 100% level, could it be good for a proper challenge?
In my opinion if you're setting up the AI at 100% you can expect that you need to really beat the car-track combo to win the race which is, again, in my opinion how the things should be.
J
JEmilio
Guest
Absolutely not, it's just an observation. I mean, after all the ai level it's just a number