Automobilista 2 Custom Force Feedback - Overview & Recommendations

I love what you guys are doing with all the feedback, that's the power of community! Just got the Simagic Evo Pro, can't wait to try these updates!

 

"rFuktor ? experimental 3.9.2"

* Fixes the lightness of 3.9.1. A calculation was missing a value. It unf. was only very obvious in some cars and not others.
* Adjusts peak slip angle.

I feel the mission is closer to accomplished with this. No longer a slide show, since the peak slip angle is much better, and the slide feel will not happen until past that point.
This version strengthens the output quite a bit, so you will need lower ingame force than before. @PeterV This might work better for you, if you run your wheel base at 100%, then this version will allow lower ingame gain, which should help. Hopefully it wont't clip too bad. I did not have time to test well.

REMEMBER: Ingame "low force boost"/"LFB" is "compression" in these files. Start at 30% and go up and down a little to find how much strength you want on cornering vs driving straight.

(edit: if you downloaded early, a change has been made, so please download again, sorry for these happenings, but it's instead of wasting a version number)

 

Took me bloody ages
But been amending a profile for about 16hrs and almost there

feels dawn good at long last
I’ll post the profile after I’ve tested it for a while to be sure

 

Hi Daniel
Thanks for all the work on the file - I have the VRS 15M wheelbase. I enjoy the file very much other than - as soon as i turn the wheel more than a couple of degrees I get a vibration that I assume is scrub because it doesn't happen when going straight and increases when i turn more, i like the effect when actually scrubbing but it seems wrong when only turning the wheel for example 5 degrees, what setting would you advise i change? Thanks

 

I just had an insane race, I thought I had "rFuktor exp 3.9.2" but it was just my old edited 9.1 ... I think I'm struggling to get past that file ahahah the race was so amazing. The guy was the GT4 Spielberg Record holder with 2 cars, Audi and Merc, and he told me he used the V7000.

While testing FFB this whole week things have gotten risky, at some points I've ended up spinning when I tried to keep up my pace. But 3.9 just worked great for me and I had the most intense 25m of racing ever, because we spent 25m around 1 second apart the whole race lol.
I did try some moves but this guy was really good very hard to make it stick

I actually did the fastest lap of the race (by 0.030 lolol), but struggled a bit for consistency as well as him. I had 9 laps in the 33s and he had 8laps in the 33s but I had a bit less consistency of pace, I had some dips but interestingly, the most costly time lost for me was 0.9, almost 1 full second in lap1 due to cold tires.

 

I hate to be the bringer of bad news but I'm starting to think maybe I've been looking too deeply into these files and I end up optimizing "whatever version" it's being currently worked to "whatever car" I'm currently driving the most.
While this new 3.9 has many things that I really like on low DF cars like the GT4 and makes me perform at my best with them, as soon as I tried the C9 the file was unusable due to the huge downforce forces on top of the huge centering forces, trying to turn the car felt like trying to turn a building.

Also on a separate test which sadly I forgot to turn on OBS, I was able to match rFuktor 3.9 (edited by me) laptime (my previous personal best) using the Default+.

My biggest struggle with the Default+ is that it feels kinda light in the center compared to everything else, but when I add any sort of weight to it (via wheelbase or LFB) I start to lose detail, and I am less consistent with it.
(with the rFuktor I had that crazy race, but in the next race with Default+ while I was able to reach a close level of consistency also lapping in the 33s in the same combo, I had a couple of spins that felt like they came out of nowhere)

Another of my curiosities is, so many files are variations of and older rFuktor version?

 

This is the perfect one I think because I'm faster than before, I can trow the car into the corner and feel perfect the limits. I drove the Caterhams and P1 evo series and what a fun to drive with and so natural feeling, a masterpiece sir

 

For high df cars you need to lower ingame force and most likely increase the lfb(depending on how strong you want it to be on cornering).
Otherwise you get exactly as you describe

 

I was forgotten to mention that I tried the GT3 cars and now I can feel the weight of the GT3 cars especially when I was use to light cars, perfect job

 

"rFuktor ? experimental 3.9.3"

* Finetuned edge of grip adjustment further - Any feedback on "too early" or "too late" or "too litte"/"too much" on any car are very welcome! Trying to "nail it".
* Adjusted curb walls effect slightly down.
* Curb wall effect now follows FX slider, so it's somewhat lowish at 0% FX and peaks at 30%.

REMEMBER: Ingame "low force boost"/"LFB" is "compression" in these files. Start at 30% and go up and down a little to find how much strength you want on cornering vs driving straight.
High df cars: If too strong center at speed, lower gain until it is low enough & then adjust LFB further.

 

Karsten, I had a thought.... but your last V3.9.1 post had some info in it that might just show the same thing I was thinking about.....

I would assume that for the front tire 'force' - which would be made of self-centering forces and some others, such as the tire slip - you would want the forces to generally increase per cornering speed/load, UNTIL (* more info coming) the tires were sliding and thus force reduces.
BUT.... to be truly correct the forces would continue to increase as the tires even slid X amount because until they PASS the optimal Slip Angle they would be doing a better job than even for tracking true. So the maths would be something like 'All force factors, including reduction for ANY Slip Angle for now.... then that Slip Angle force 'curve' has an offset so that it is increasing force UNTIL passing , say, 6deg, where it then is SUBTRACTING force value.

And your bits of test of changes suggests that is EXACTLY what you have been looking into!! LOL
I had been pondering the Default+ and 'car weight' feel aspect of FFB, and decided that came down to LESS force reduction on sliding.... and also the GAIN of force UNTIL the optimal Slip Angle is passed.
Whilst I 'trained' myself on the greater information of MORE Rear Slide Feel, the cost of that is a car that feels LIGHT and SLIPPERY all the time! Even though it is not truly slippery at all!!

Boy am I glad I was SLOW to see, download, and try, the latest versions!! LOL
Skipping from V3.9 straight to V3.9.3 !

 

Yes, exactly right. We don't want the grip feeling to decrease until around peak slip-angle for the tire, even though there is sliding going on increasingly up until that point too.
In the older rFuktor files we also adjusted this aspect, and I want to make the new version work even better in this regard.
For the new version we just did not arrive at that bit until now, I guess.
3.9.3 just posted betters it one step more, and likely at least one more finetune iteration will be done, hopefully some comments will help adjusting it just right.
This is all related to the things that you have been talking about reg. sliding/not sliding feel etc, and is also why "going light" on oversteer vs "making oversteer more clear by more oversteer force" is a debate.
I tend to think we need the later, so the car feels solid in the front, since it really is.
Right now it goes only very slightly light on oversteer, and that could make sense too, even though that is a canned effect for sure.

 

Thanks, I was wondering if we were starting to "lose the plot" ahahah xD

I mean, we go deeper and deeper and at some point we get blinded or tunnel vision into the things we're trying to fix and we lose the "bigger picture".
I'm just saying it because it can happen with anything, car setups or even other topics.

I've also done so many tests... I feel like this week all I did was drive and test ffb files all the time.
In the end with default+ I've lucked out on a similar lap to the rFuktor file, but default+ seems way harder to be as "grounded" and consistent.
For some reason the ffb goes randomly light over some kerbs or due to some front tire effects in some moments where the rear needs correction, sometimes turning into a corner is enough to make the rear snap away and spin and that can bring inconsistencies.

At 1 point I thought... hell, if I could make Default+ work well with any wheelbase settings, I would really try to dig it, because somehow it feels the most consistent across cars...
But then instead of going from GTs to Proto or GroupC, I went into a Formula Trainer race... while I qualli first easily, inconsistencies creep up in the race, it felt surprisingly different, while there the rFuktor variants felt much more grounded like on the other cars.

I'm also starting to wonder about the "famous" 5.0.1.3 ? or whatever version is the "real" first "successful" rFuktor? because it seems everyone has custom FFB tunes based on that.

I found a folder with some of my old ffb files and it seems I really used to like the 5.0.1.2 probably the most advanced at the time. I tried it a bit and it seemed quite nice, I was actually also able to do an incredible qualli lap of 1.33.4 in an LFM lobby, which is very close to the WR, don't think I've ever seen such a fast laptime in this GT4 Lobbies.
But then during the race and variable grip and tire temps, I had again some unexpected inconsistencies but the pace was there, maybe I was pushing it too hard. I was pleasantly surprised.

I must wonder if I'm doing the right thing, we can never be too sure of ourselves since that will bring the arrogance that will blind us and give us ignorance over the things we fail to understand we do not know yet. And because of that I'm left scratching my head:
I enjoy many different files, in some cars some files are better than others etc...

I was surprised that the 3.9 made me as fast as the fastest guy in the world GT4s, now I gotta learn how to use the sliders to make that file work for all the other cars "that well"... since right now I can't even drive out of the pits in Group C with those same settings.

For the next few days I might not have this much time as I had this week, this was pretty intense spent all my time driving but eventually work comes my way.

I just wish we had "1 ffb to rule them all" ... ahaha something that just works the same for every car, regardless, for any base, all the time. I guess the engine is a bit "too complex" and variations are huge. In most other games the FFB has it's own characteristics, feels like "that" and has some variation inside "that" box. But no matter what I do in some older games, even if the ffb isn't that "great", it just works, more or less "samey" but all the time. No matter the settings ingame... but here in AMS2 even with same file, the FFB can feel like a new game from 1 car to the other. I don't think I've ever seen such extreme variations of FFB feel anywhere else.

 

V3.9.3..... Karsten....

OK, that prior post I had about WHEN the Slip Angle transitions from ADDING force to REDUCING Force is the exact parameter/maths that explains what I feel and aim to get, when I change the Grip Loss Feel to be greater (1.5).
BUT.... that value is NOT the correct fix to achieve the above!!
That value has NOTHING to do with the TRANSITION point.

As per some of those update changes comments you made, you ARE looking at that transition aspect.
But I just ran the V3.9.3 file and didn't get enough 'Grip Loss Feel' for that above valid outline case. eg The car feels solid and planted, much like Default+ 'weight', but the Grip loss when the tire slides is not enough - and also hard to tell if it is OCCURRING at the right 'transition' point - seeing there is not enough to tell that when driving.

I watched that car in replay, and because AMS2 allows LARGE Slip Angles to WORK, this means that the transition point where Slip Angle Force transition occurs would be something like '12degrees' !! Not around 6 to 8 degrees of a real slick tire.
So if a FFB file is made to transition at 8 degrees, then it is WRONG. Correct versus a real world car, but not correct for an AMS2 car! It neesd to match the GAME car, because when you race THAT is the Slip Angle value you can use for optimal grip and lap times, with NO reduction in Grip Loss Feel (force) yet.

I haven't looked a the FFB file specific maths area for that, but it will be the area that the Grip Loss Feel value is thrown into the mix.

A 'problem' of this aspect being done/used (and it should be!) might be that different cars - and more specifically their different TIRES - will have different optimal Slip Angles. So the FFB would need to be DIFFERENT per a TIRE type! If it was to be super accurate. Otherwise, maybe, a general value of '12 degrees' is just used and will be approximately correct for all cars in AMS2.

"12 degrees".... one problem in AMS2 is there is no way (???) to measure the true optimal Slip Angle.
You can DRIVE the car, and WATCH a replay, but that is not overly accurate/definitive as to what that angle is.
You can MOTEC log the car data, but AMS does not give Slip Angle information and whilst there is a 'shonky' way to derive it, with X amount of 'slack'/error, you then need to link that to the RESULTING GRIP... which again they do not give, so another 'error prone' lot of Motec values need to be used.
So in the end..... there is not a lot of accuracy in what Motec Logging will give.

Then the in-car driving and visuals you see..... again a very vague and subjective data source.

But I think the optimal Slip Angle value is overly large, versus real slick tires. Thus the approx 12deg range that seems to be what that is.

This is why I needed to increase Grip Loss feel to 1.5 !!
Whilst that is NOT the correct parameter to fix this 'error', it does it in (yet again) a SHONKY way..... it moves the entire Slip Angle Curve up in amplitude - not 'sideways' as per is really needed. But at least that means I GET SOME reductions before the 'wrong offset, lagging Slip Curve outcome' would have reached that range of reduction value.

This Fudge Fix meant you get a LIGHT CAR all the time - rather than the true Offset fix which will leave you with the base heavier car, but ONLY go light when it truly should!

So for now, I THINK the V3.9.3 (and all prior versions forever) does not, and did not, have this TRANSITION point correct. Or maybe not in there at all for some versions!!(?? earlier ones?)
And now you ARE working with that, BUT maybe you are acting on a base Slip Angle Optimal of 6 degrees... or 8 degrees... but the AMS2 cars are really optimal at 12 degrees!

Is it possible that you could add a 'user parameter' (like Grip Loss Feel is) for the Slip Angle TRANSITION point !????
You would set it as per what you have it now, or think is right..... but a user could easily edit that value to try to MATCH UP the real AMS2 Optimal Slip Angle to what that value should be in the FFB file !
It might really need a Front and Rear one in there because the tire physics data file could have very different front and rear tire factors - such as their optimal Slip Angles being different!

And even testing and separating the optimal Slip Angle from the Front or the Rear would be very difficult! Sigh.
To do that - as well as could be done - I would guess you would need to push the front end, understeer, WHILE making sure the rear end was ONLY driving it forwards. Thus you are testing the front end case.
And for testing the rear end case, the Fronts must be kept tracking for sure, whilst the rear is slid out - much like Drifting a car should give a fair reult from.

Or another way is just PERCEPTION. You race the car, to a truly optimal manner (if you are a very good driver!) and this means you are geting very fast lap times(!!) and then press on to OVER_DO IT on purpose - front understeer.... rear oversteer.... and FEEL if that FFB reduction (from exceeding the optimal Slip Angle) is appropriate in your GUESSTIMATION of the way it produces the FFB reduction - even across its curve as it grows in reduction amount per more excess Slip Angle.
This is effectively what I am doing to tune the Grip Loss Feel each time (but that being the WRONG fix for the issue , as mentioned). And I just gave up testing and used rear 1.5 as a commonly 'good enough' fudge that gave the extra FFB data (force reduction) required.

 

Somewhere in HERE I guess:
Working out what the maths does gave me a headache - too hard so far(!)..... LOL
Plus I have to 'guess' what the game supplied variables aer (if that is the correct name for them)....
(trace back to see if they were ever created in the FFB script or not...)
------------------------------------------------------------------------
(FL_sl_lat (/ FL_vel_y (max 0.01 (abs FL_vel_x))))
(FR_sl_lat (/ FR_vel_y (max 0.01 (abs FR_vel_x))))
(RL_sl_lat (/ (- RL_vel_y FL_vel_y) (max 0.5 (* 2 (abs RL_vel_x)))))
(RR_sl_lat (/ (- RR_vel_y FR_vel_y) (max 0.5 (* 2 (abs RR_vel_x)))))
(FL_sl_lat (hard_clip FL_sl_lat 1.0))
(FR_sl_lat (hard_clip FR_sl_lat 1.0))
(RL_sl_lat (hard_clip RL_sl_lat 1.0))
(RR_sl_lat (hard_clip RR_sl_lat 1.0))

(RL_sl_lat2 (/ RL_vel_y (power (+ (* RL_vel_x RL_vel_x) (* RL_vel_y RL_vel_y) 0.01) 0.5)))
(RR_sl_lat2 (/ RR_vel_y (power (+ (* RR_vel_x RR_vel_x) (* RR_vel_y RR_vel_y) 0.01) 0.5)))
(RL_slide_lat (* 50 (power (abs RL_sl_lat2) 2)))
(RR_slide_lat (* 50 (power (abs RR_sl_lat2) 2)))

(RL_slide_long (/ RL_scrub 15))
(RR_slide_long (/ RR_scrub 15))
(RL_slide (min 1 (power (+ (* RL_slide_lat RL_slide_lat) (* RL_slide_long RL_slide_long)) 0.5)))
(RR_slide (min 1 (power (+ (* RR_slide_lat RR_slide_lat) (* RR_slide_long RR_slide_long)) 0.5)))

(FL_slide2 (min 1 (abs FL_slide)))
(FR_slide2 (min 1 (abs FR_slide)))

(slide_factor 0.9)
(FL_sl (* 1.0 (min 1 (* slide_factor FL_slide))))
(FR_sl (* 1.0 (min 1 (* slide_factor FR_slide))))
(RL_sl (* 1.0 (min 1 (* slide_factor RL_slide))))
(RR_sl (* 1.0 (min 1 (* slide_factor RR_slide))))

(FL_sl (* FL_sl (if (max 0 (- FL_load 1)) 1 0)))
(FR_sl (* FR_sl (if (max 0 (- FR_load 1)) 1 0)))
(RL_sl (* RL_sl (if (max 0 (- RL_load 1)) 1 0)))
(RR_sl (* RR_sl (if (max 0 (- RR_load 1)) 1 0)))

#(slip_fl FL_sl)
#(slip_fr FR_sl)
#(slip_rl RL_sl)
#(slip_rr RR_sl)

(slip_rl (power RL_sl (power (abs RL_sl) -0.01)))
(slip_rr (power RR_sl (power (abs RR_sl) -0.01)))

 

Yeah, it's a true rabbit-hole and will eat your brain and never leave you alone at night!

The 5.1.0.2 - 5.1.0.4 were kind of where we left it at back then.
However, they had things like "static force reduction" and other slightly creative ways of getting the forces as we wanted them. Also they were focused more exclusively on "feeling real". They were also done sitting on older versions of the game, so might be slightly out of tune, I'm not sure.
In the new files there is no use of static for reduction, the rack is not a "special rack" at standstill, and many timings/latencies have been reduced by a lot. Also the newer files add caster trail/pneumatic trail, which should make it possible to tweak more realistic feeling in certain aspects. The new files have improved several single items, like for example suspension effect as well.
I THINK with the newer files, and also some of the tuned files others have done(based on the older versions), a newcomer would be able to race more safely close to the limit faster, than with the older files.

Hey Peter,

Better you write me directly if you want to discuss specific code pieces. It will eat the forum discussions to become for a niche group of us.

 

Oh... or maybe it is actually the Oversteer and Understeer sections really....
Or BOTH areas...
Too 'messy' for me to unravel!! LOL
Probably 'easy' (enough... relative....) once you get to know what is what.

 

The thing I noticed was in AC what I liked was that losing grip on the rear, or sliding, was communicated by vibrations and not by getting the wheel very light. Actually lightness was related to weight transfer. This in turn made it possible that you can feel counter steer to catch the car, even past the limit of the tire, the tire can be sliding but you get the "correct" amount of counter steer force.
Maybe even if it goes lighter on the rear during grip loss, it's by quite a small amount. I've also noticed that the front will go lighter on higher grip loss, but in initial grip loss it's not too noticeable.

It's easy to scrub off speed by turning the fronts a bit too much, and not really feeling it much on the FFB there. You just notice your apex speed wasn't as quick and you turned too aggressively, or you get a bit of understeer.

For me the most strange and odd think in AMS2 when I got back was actually having lower and lower feel on the wheel the more the rear end stepped out... this is something I can expect to feel in the corkscrew chicane at laguna, a bit of lightness but even then it's never "overdone".
But I think in AC lightness isn't used for the Front and Rear grip at the same time.

You can have 4 wheel drifts or slides into a corner or out of the corner and still have decent FFB counter steer to catch up and align up the car with the direction it is traveling.

Here what I noticed when I started complaining and asking for help, was in certain cars if I was already light in understeer and I had also oversteer and ffb went light both in front and in the rear, it's almost like I got near zero ffb, not only is this unrealistic, but it is undriveable.
In such a slide situation, you're supposed to be feeling the G forces (seat of the pants) effect to correct and align the car, try to find that grip back... if goes light on both sides? where is the grip?

Even if the lightness effect can be present, it must never be this strong. Some of the 4 wheel slides I've felt before in AMS2... felt like AC driving on the rain over puddles of water. The force drop off was that extreme.
I think that's something that I really enjoyed when I flashbacked to 5.0.1.2, being that most things were a bit more tame and not this extreme. And my 3.9 suggestions were trying to bring that "grounded feel" back in, while it ended up slightly overdone and also the file uses compression and it's a whole new thing and feel.

I'm trying to explain this part of the debate the best I can but this is truly difficult to express.
I would need time to record so much more, and I have so many more hotlap comparisons I should be editing.

I think also 1 big frustration was how much I've hotlaped at GT4 Spielberg, like at least more than 20h or 30h this week, I should've been able to find the time, get a new WR. I've never put so much effort into any hotlap in my life to come in close second and still struggle with consistency in the races.

This is what tells me that something must be very wrong with the ffb when I've had huge amounts of practice, my "potential" has increased but then I go race and my laptimes are all over or my TT "potential time" (when not bugged) is quite faster than my Pbest hotlap.

So far only 3.9 has been able to bridge that gap but I've spent so much time in 1 car and track that I'm afraid I'm optimizing this file for this combination... ahaha sorry for the huge rant.

BTW Peter I don't remember well what I did, but I mentioned I changed a few values inside the 3.9 file, but the file doesn't feel "slippery" or "light rear", the file I used felt very planted for the most part, but you do feel a stronger counter steer pull and hopefully it stays more during rear grip loss. Kinda like what I'm used to have in AC. In many cars including some GT4 guerrilla mod of the same Audi, I can use my inputs to induce the car into a drift and countersteer to catch it back during a corner. I also feel the car squirm for grip under braking etc. It's difficult to make the ffb feel light unless I'm literally flatspotting the front tires. The "lightness" variations there are a comparatively, smaller, you always feel more of the steering rack all the time.

 

I feel the steering going light in the front, as it should, I am not sure why you do not...
You might want it lighter still, and I could agree to that, I think it is borderline not enough right now.
There is no hardcoding of slip-angle degree in the code, as I am exactly trying to make it work for all tires.

 

OK

Is the are 'Slip Angle Curve' in there that does as mentioned?
ADDS force for 0 deg until the optimal value, and then transitions to SUBTRACT force from the optimal or greater?