Automobilista 2 Custom Force Feedback - Overview & Recommendations

Yeah, try 30 instead.

 

BTW in hotlaping ideal conditions I've set my Pbest and P2 TT of GT4 at Speilberg (1.33.440) with Danielkart V7000 Topspeed 1
(with some personal modification:
(over_steer_scale 2.99) #instead of 2.80
to help me drive the car and I was able to get quickly close to the limit, but then to drive on the limit I was very inconsistent and every LFM race I tried this file I had an accident or just very poor raw pace under less than ideal conditions, changes in temps and grip etc.

I've just tried Default+ again, I even forgot to change settings so I left 65 0 12 0 (like V7000 recommends) but I had time to adjust settings during the race on the wheel.

With some other default settings I had done 1.33.880 but I've been doing low 34s most of the race, which is far better consistency than before. I still have to test rFuktor in races but there are so many versions and variations that it's hard to check out everything.

This last one while the rear feeling can stay for countersteer, now there is no information about loss of grip or spinning like I'm used with AC where there is vibration going on. I think the rear can go light if there is less weight on it aswell in AC. But the countersteer feeling takes priority over everything.

What I've noticed is that after carefully adjusting Fanalab, I can feel more from the countersteer part of the FFB on the Default+ profile using all filters to zero, game forces at their 100% default but "Force Effect Intensity" toned down 30% to 50% can help to feel better other parts of the ffb.

Another thing that I noticed recently I did a P1 that became P2 hotlap with Mercedes C9 at Watkins GP for a league race and both rFuktor and V7000 were way too slow and too weak telling me anything about the rear of the Mercedes C9.
The only way I could drive it was with the Default+.
Sometimes with the custom files, I would be flatout 100% gas comming off a corner, I don't feel anything and the car has huge spins. This happened every other lap... no feel for any of that, you guys can try it out and then go on Default+.

I even spoke with another WR holder, hamtidamti, and he told me he uses Simcube1, with no filters everything off, and just changes steering ratio for his taste depending on what he wants (usually sharper steering) quite surprising to me.

I've also seen Mikk use V7000 to great success and I think he uses the recommended or some variation of the recommended settings but I can't really figure out how to feel certain things.

Edit: IDK what I'm doing wrong, I just used the V7000 file as usual but without my (over_steer_scale 2.99) and at every corner I was spinning out on entry, from turning too much. Neither under aceel or brake, I just trail finish the trail and I'm pointing to the apex thinking I have grip and I just spin. When I feel anything on the rear is way past the point where I could have reacted.

 

"rFuktor ? experimental 3.9"

* New way of calculating front grip - The idea is to feel more realistic/less slidey, while not losing anything.
* No loss of steering feel on oversteer. Instead trying to improve oversteer on its own first.
* Front tear scaled up again, since it seems like some people lose that signal when it's lower than this.
* Suspension feel scaled slightly down.

REMEMBER to use "low force boost"/"LFB" at 35 or so at first & go from there.
It means "compression" in these later files and is there to be used.
Lower will give bigger difference between driving straight and cornering, so since you likely adjust based on driving straight at first, it means that cornering will be more forceful the lower you go on the LFB.

 

Hi

in the ffb file (custom)

what’s the difference between tear & scrub
???

 

Tear comes earlier and is faster with less amp.
Scrub is on bigger slide and has more volume.

 

Ok brill thanks , more tear then with smaller amount of scrub I think then

thanks again

 

man this is so much better in this specific circumstance that I'm in I have a lot more consistency and control over the car. I did add a tiny bit of that reargrip thing. like 0.1 ahah

the center does feel very tight, like now you feel the car wanting to self correct earlier, but the scale of it goes right up quite quickly, I'm not sure if it is "overdone" or not. but it is quite nice, I'm also wondering if that "part" of the forces is clipping, meaning you reach the maximum countersteer force you can get and then if "flatlines" at the peak of what that value can achieve?

I'm not able to do as fast as before but I'm still getting used to it, my fastest was a 33.600 but the amazing part is that I can replicate times very close to this lap many times in a row, and even after quite a lot of laps when I start to have hot tires and struggle for grip, I'm not caught off guard. sometimes I can have slips but I catch the slides right away, I haven't hit the barriers any single time since I'm using this file.
gain 50 lfb 35 0 0 (instead of using 100% on the base I upped a bit ingame, somehow felt better idk how to explain)

 

I wonder why they would do such a thing. The game already performs well so i don't think it would be for performance reasons, or maybe that is why it performs so well? hmm.

 

rFuktor V3.9
Used in its 'raw' form it seemed better than all others after V3.3
Tweaking the LFB/Compression was important, to balance out the 'straight versus corners' feel. I ended up with LFB 20.
This causes a problem that 100% FFB Strength is weaker and not enough - but you then increase the Wheelbase FFB Strength

Whenever I go out lapping Time Trials, I am going for 100% every lap and with 2:14.6 best at Spa the majority of laps done are in mid low 15's because it only takes a MINOR 'not done best' area to lose a LOT of time. So I expect to do 2:15.5 typically, for the first or overall averages, but with some very low 15's or even 14's in there.
This was the same outcome with V3.9 but I did feel that I did NOT feel (LOL) the front or rear grip loss enough - as usual with the rFuktor default values. I only ever increase the REAR Grip Loss Feel value (to 1.5), but this time I did Rear 1.5 and Front 1.0......
What this did was make the overall FFB a lot lighter when cornering BECAUSE if you are running slip angles you will see those values have effect - and the greater the slip angles the lighter things get.

So... as you race around, you cannot truly know WHICH end of the car made it go lighter. But you also CAN. What happens is that as you corner, and feel the steering went lighter than 'last time' (or you can posssibly build up a feeling for what any corners would/should do really) it is telling you "You have more slip" and to tell where that comes from you have SOME visual information as to the cars YAW, but what you also do is CORRECT the Steering a MICRON - tighter and shallower - and the CHANGE in FFB strength tells you what was going on.
eg If you reduce steering angle and it get firmer FFB, then it was likely understeering. To reduce rear slide, oversteer, you need to do the opposite. Thus you can tell which was which. Well, there is also THROTTLE which you will alter a bit at that same time, to keep the car on its optimal line and ongoing path ahead.

So I decided 1.0 Front and 1.5 Rear was too much Front, and even maybe too much Rear in V3.9... I have to do changes and press on with what it changes/improves (or not).

But the big thing it showed me is how you WANT the Grip Loss that you know the Slip Angles you are running at, which then also tells you how far you can go, and how close to a grip LOSS (slide/spin) you are. It really helps fine tune that cornering maximum plus you can derive the balance of front to rear grip status (via the micron sterring changes to separate where it is coming from really).

Then in another test, that popped into my head randomly.....
Back in rFactor1 days there were close to ZERO FFB Wheels. At some stage I got a Logitech G25 and whilst IMMERSION felt really nice, the FFB DETRACTED from racing faster!!
This is because if you run curbs and it kicks the Steering, then it can affect YOU and make you mess up the steering 'accuracy', plus the fact that THEY moved the wheel a bit means it altered the car because of the FFB also!
So you could do way better lap times with no INTERFERENCE of FFB!!
You didn't need to feel the car at all as the visuals and repetition give you everything you need to be fast(est) anyway!

Thus I did a test with no FFB in AMS2.....
It feels really WEIRD!!! LOL. Horrible in realism terms!
But in no time at all I could lap as per with FFB anyway! It was quick to achieve because a lot of what you know about the track/lap is visual cues anyway, and I also have a head start on braking points, cornering speeds and manners, so you already know what the car can do in speed and in visuals - and that is enough to do the fast lap times still.
It just FEELS terrible, like you are playing some really bad 'no physics' arcade game!! LOL

I would never want to race like that - even though we did in rFactor 1 days. I would say the FFB of a G25 back then had lots of 'lousy ascpects' that meant there was as much bad as good from using that FFB. Thus nothing to lose if not using it. But these days the FFB, part software and part the mechanics (DD wheels etc) make it REALLY GOOD to have FFB !
But the test showed you don't HAVE to have FFB to lap just as fast!

When you have a poorly setup Custom FFB file - or you were experimenting and end up with some 'less optimal' settings, THAT is what you can get!! An 'arcade' feeling car! All the '50' mathematical factors used inside the FFB file - each 'module' of what it builds to give you - have to ENHANCE the outcome, and if even one is a 'degrader' it can ruin the whole outcome. Or mess that up at some % amount. Sometimes a BIG % amount!

So far (here) V3.9 does not match the car 'weight' of Default+, but that might be a mix of what LFB setting I have and the total FFB Strength used so far. I have a good balance, but it is a 'light' car. Though note that with the game FFB set for 100% it might be clipping too (probably not overly important track areas anwyay), and to need that level it suggests the average value of each 'module' is low, thus the output BEFORE the Gain stage is low. Maybe the entire rFuktor maths could cope with an increase and still not cause clipping???
Or maybe LOTS of modules in it are lowish outputs, but ONE is higher, and thus you cannot afford to increase the output gain or that will clip?
It is just that V3.9 means I am running 100% in game and will need very high Moza R12 FFB strength in there too. eg now 100% game 70% Moza, and that is too weak. I am likely to need 100% and 100% region!!

 

The center feel and the front stability are even better on 3.9. I prefer it on my T300 rather than the DD base. The rubber feel is great on T300 and it's more on the artificial side on my DD (specially at the center under loud). You can not go beyond 30 on LFB with the belt system or the sudden stress at high speed will cause spinning. I could hit 40 on my T818 base with no problems, but again I keep going back to my T300 after trying the recent files. The FX was set to 10 during testing on all bases and that's the sweet spot for me. FFB sits between 65 to 75 depending on the car. No clipping below 76. Anything above 75 will reduce the reaction time in terms of corrections based on my experience

 

FEI for a dd1 is best set to 20% because it covers the game's FFB. Increasing over_steer_scale to almost 3 will shift all the forces. FX is too low; more FX will give you a better feel for what the car is doing. But honestly, it's irrelevant to measure physics or FFB against world records or times. It's possible to be fast with any file or with the defaults if you spend enough time and always use the same thing. Also, always making an assessment or analysis with the same cars and tracks is very strange.

Note for everyone: there's more to Ams2 than just GTs

 

I would think that if the "FFB Maths" is correct (enough) then it would produce similar quality results across all car types. Maybe one thing it could show is if the 'aero maths' in the file is amiss, thus only shows up in aero type cars.

Too bad there are only the 4 Sliders in the game..... they (or having MORE!) could allow for 'modifying' key things that matter from car type to car type. Much as they are being re-assigned roles in FFB files now, but there is only really ONE 'free' one to steal, or two at most (3?).

The same track is fine.... Spa is like the best 'test track' to use. Though I guess you could find some more cases of special case 'feelings' at Road America or Laguna Seca.
If the car physics is right (right enough, again) it should deal with all track types correctly. And the same for a FFB file then - seeing it is just building off the car physics.
But if the FFB file was NOT doing something right, then it might show up at those more extreme 'cresting/dipping' type tracks.
But Spa is still a very good (optimal?) track to test all racing aspects at.

Hmmmm... Sebring and its mega bumpiness?????
Again, a 'correct' FFB file SHOULD operate as per reaslitic even there, but I guess it again (like Laguna etc) MIGHT show up some issues that were never seen at 'normal' track types.

 

There are some tracks whose physics is a bit too "strong," in my opinion; it might be better if Reiza toned it down a bit. Having more in-game settings would make everything even more complicated. Think about Raceroom back in the day; it was a science with so many settings, but they've corrected it now, which makes sense. You have the option with Ams2 with customs, so you can adjust everything yourself the way you prefer. Therefore, I don't see any advantage in making it even more complicated, especially with today's many different bases

 

rFuktor V3.9.....
After a lot of testing, I didn't like it. It seemed "OK" but it never felt totally right.
One thing in the list was the self centering 'sharpness'. You feel it in turns, but if you test it down a straight at high speed - by doing a short angle fast turn and then let the self centering bring it back into straight - on that return to straight it is fast and 'harsh', which MIGHT be 'right' but it also has a fair bit of overshoot. That part does not seem correct.
It doesn't oscillate by itself, but the forces involved help the driver cause that himself - seeing the pressure changes high, low, high, low, around that centred point.
It should stand out to anyone testing its responses doing that....

This would then tie in with ALL turning cases, thus means there is that same 'extra motion' included in those turning, and reducing steering/turning cases. eg you release pressure as the turn is reducing or over, and the self centering will do its additions. The self-centering forces were there for the whole time the wheels are turned and under a load (car moving forwards) but it is when they CENTRE (and can go a bit either side of centre) that it has the weird bit.

So, then I went back to my 'favorite' V3.3 (1.5 Rear Grip Loss'.....
The very first lap was FAST and 'easier', even though I didn't bother to adjust the game sliders to have what I had previously!! I just left them as per V3.9 came out best (LFB 34). It seemed V3.3 was more lenient to the sliders being 'precise'. FX 50 but the FFB Strength did need to be brought down off the 100 (back to 80 as per before) - I felt that immediately so went and changed that within 10 seconds of driving!
It felt like a SLIGHTLY different car, or that you had altered the setup a bit... which I guess is the amount of slide/slip feedback you are getting (or not getting). Eau Rouge became far more 'you will remain within limits' almost every time, versus with V3.9 it was notably tougher (and thus SLOWER). And I had done TONS of laps with V3.9, trying to tune it up for the best feel/information and that 'max speed, you are really asking for trouble limit!' up Eau Rouge could just never be achieved reliably and safely. No crashes really, but just over-running limits too often in V3.9 and pretty well never in V3.3.

It is just 'amazing' what these little extra bits of information - or difference in them - from FFB file to file (or even other creators FFB files), can make to what you 'know' about the car! Or even more so how it can make the car feel like a DIFFERENT car, or a different setup..... which can make one case (FFB file, settings etc) do something better than the other, as if it was actually a better, more capable car then!
Trying to 'learn' V3.9 to get the same ultimate lap times... well, that just did not eventuate. I could not get within about 0.3sec after such a long testing run. Though I would have THOUGHT it would be learnt and then match the times for sure. I think the NO FFB tests (only a handful of laps, 8?) ended up in much that same lap time region. But they were built up from repetition only really.

 

Running ingame force at 85-100 is not recommended for DD's.

Remember it's experimentals, so new stuff, like the oversteer change, is overdone, as @carloscepinha noticed, since I need to make sure it actually works first, and then we can dial it back and add some relaxation so it also kicks in just slightly later, possibly.
Also trying to work it away from that slidey feeling, when driving under the peak slip-angle of the tires, where it should not feel slidey.
3.9 is improved algorithmically compared to 3.3 in several other places, so I will continue to adjust algorithms and tune values towards a real improvement.

 

Yeah,

And I might actually be completely wrong.
As @carloscepinha posted he has looked in to it using telemetry and says it looks fine.
So I likely stand corrected.

 

If this is correct, can I ask why these lines haven't got the opposite settings?
tear_threshold 0.42) #how early the tear feel appears - the higher the later.
(scrub_threshold 0.12)
Because looking at those settings, wouldn't Scrub come in first, or does it not work like that? Thank you.

 

In the normal rFuktor files we have:

(tear_threshold 0.2)
(scrub_threshold 0.35)

Some rFuktor "tune" files have reversed them and changed their frequencies and volumes too.
Just a matter of taste which feel you like better at first and later.

 

So I've changed a few lines in the 3.9 to my taste (I think, since IDK what I'm doing ahahaha)

But I've not only managed to do my fastest lap there (GT4 Spielberg) but also gained a TON of consistency and control over the car.

I've even A/B last night with "rack force" again, because I know rack force gives the most immediate and snappiest and highest detail on oversteer control. I have to confess I would almost be more consistent with "Rack Force", because it's so naked, I can pull of saves for very aggressive driving and really sending it "too hard" and just catch the spins later, but due to it just being rack forces, there is nothing else there, I either overheat tires, overdrive, and sometimes I overshoot my turning point because it is too light in the middle of the wheel, instead of having precision the wheel is "too free", and maybe because I only have a F1 270mm wheel it turns quite fast.

Now with rFuktor experimental 3.9 I think these are the best results have had in quite a long while. Both in TT and also tried it in a race where it wasn't very challenging (the fast guys didn't enter it) but I went from last to first on a cold track doing laptimes that weren't so far off my Pbest while surviving and dealing with battles and traffic.

TBH with this file I had a very pleasing level of control. It's very tricky to have in the ffb say when you want to feel oversteer or not, because depending on car, that can change.
In the end you would have custom ffb for each type of car like each car class.

What I noticed was the "clean ffb" files while very good for "slow cars" that can be stable with much more aggressive driving stiles... that kind of FFB might not work on a really powerful car like group C.

This all ties up with the fact that my best result here from V7000 Topspeed was a version of the file where I heavily modified one oversteer setting to feel oversteer a lot more (like in the new 3.9 and the rack force files)

What did I change in 3.9 ? here:
I tried to feel the rack a bit better, everything else I can feel it pretty good already
(master_rack 0.91)
(master_road 1.15)
(master_effects 1.50)

(rear_grip_loss_feel 0.13) I do need a tiny bit of info on rear grip
(yaw_delay 0.002) less delay is more direct?
(front_drive_torque_feel 0.05)
(rear_drive_torque_feel 0.05) I find a tiny bit of this to be realistic
(balance_smoothing 0.02) IDK what I'm doing, less smooth means more direct?
(scrub_scale_rear 0.30) better rear feel?

btw rackforce best laptime (around 50 to 55 gain, messed around a bit w wheelbase settings)
1.33.573, 1.33.652, 1.33.441

rfuktor experimental 3.9 (edited) (50 40 0 0)
Wheebase (80 gain, "FEI":70%, zero filters default game forces)
1.33.340

While in many cases with just the rack force I'm able to consistently lap under 34s, while not trying too hard, when I try too hard is when I need to do big correction or inconsistencies appear, due to it being light, I've even tried some damping which helped make the steering feel more precise and less jerky over big kerbs.
I've noticed that going down a lot in "Force Effect Intensity" in my wheelbase settings, while it helps a bit to "tame" the effects and kerb spikes, it also removes some feel from the steering wheel.
I still think that this ffb rack force has a base is nice but there would need to be more stuff on top and it is a bit surprising when you do actually need a correction you get a surprising big spike in the forces.

In both cases it seems that I can kind of "drift" through the corners on my own will, or I can induce oversteer under braking at my own desired control, meaning in T6 and T7 (both left handers) in Spielberg I can choose how much I make the rear end step out and control it from there.

Before this, my LFM experience in races was horrible.
As soon as temps and grip were far from time trial ideal settings:

Sometimes I would lose the rear of the car on corner entry, during coasting!! I just turned too much for the speed I had and didn't feel anything on the FFB regarding the rear stepping out due to excess speed.
In these 2 files I've used, I can correct the car and even if I lose some laptime, go deep or even go a bit on the grass I'll come back without a spin, which is million times better.

Not only this, before, I couldn't feel the rear stepping out under trail braking, in some races I was fighting for position w rly fast guys but as soon as I'm trying to push pace or I just go into a corner a tiny bit too fast... I trail a millisecond too much, my car was in an uncontrollable spin, pack of cars going by I can't rejoin and my race is ruined.

With these 2 files these kinds of situations are mitigated.

 

"rFuktor ? experimental 3.9.1"

* Adjusted peak slip-angle precision for feeling front end slide. Hopefully better, but let me know whatever adjustment might be needed more.
* Adjusted oversteer slightly.
* Added relaxation right around slide 0 for rear grip/slide feel, so it does not kick in quite as much. Hopefully still enough, but let me know and we can adjust it.
* Adjusted values as suggested by @carloscepinha in prior post, thank you for posting those! I find them to make a lot of sense.

(edit: reuploaded, so if you downloaded right away, please fetch again)
(edit2: use version 3.9.2 instead, this one has a flaw that makes some cars go way too light)