Automobilista 2 Custom Force Feedback - Overview & Recommendations

Yes, it's on my short list of issues, not sure what it is yet as I didn't look in to it.
Might be related to the weird on/off feel when turning the steering at standstill, not sure.
So there are those issues and the exit pitbox issue, that I have not tried to reproduce.
And most importantly, to get the feeling on straight, on slow cornering and high speed cornering just right. That still needs some tuning, since 4.2 has not addressed that further.
Finally, the damping needs to be as good as possible and then the oscillations when letting go of the wheel as well. I might have missed something in this list, but that's on the top of my head

 

Guys you should mention which car and track you drove in time trial and also share some laptimes, that could be more helpfull than just "blank statements".

I've set some competitive laptimes with the custom ffb files and I will not only test how fast can I go using my driving tecnique compared to using default+ but also how consistent can I be.

TBH for me 3.1 has been the closest to perfection of ffb i've ever felt and while everyone wants to feel more of everything (also depends on which base anyone is using) in my case I want a snappy and precise feel just like good old AC that everybody praises it's ffb and it is still used by professional drifters to pratice for IRL due to the accuracy of the gyroscopic effects and the "sliding with precision".

For me in 4.1 the strenght overall is still much higher. But I think there must be a bug with this.

If I do Time Trial laps, due to tires starting at 50c and track being in a decent temperature, the ffb feels much heavier. Then I go for some LFM races and in qualli the tires start cold at 30c in the pits.
The ffb feels light, a lot lighter (normal due to less grip).
The problem is that when the ffb starts very very light, and I eventually warm up the tires, the ffb never becomes as strong as in time trial.
Then I adjust gain up to compensate the light ffb. From qualli to race I start on the grid with already warm tires... now the ffb is very heavy again and I need to adjust the car gain down.
IDK if this is related to multiplayer or LFM servers, I haven't tried to replicate this in a single session but I might try, using the ffb telemetry.
This is quite an odd behaviour.
I've tried to see if my wheelbase was changing gain or ffb profiles based on car but it didn't seem to be the issue.
With 3.1 it was lighter overall but I don't think this issue was happening or if it happened I didn't notice it as much.

I would recon that weaker wheelbase or nm users would enjoy 4+ versions more if they are this much heavier because they do feel a bit stronger in forces and some of the forces even feel a bit more compressed than before.

TBH I noticed the "stationary vibration" but as soon as the car starts moving it's gone, even with 15+nm of force I don't care about such vibration as long as I enjoy very fast or the fastest laptimes and specially consistency in replicating those.

Oddly enough yesterday a Brands Indy GT5 race had dynamic weather and it drizzled through 15m of the race and my lap times were almost as fast as in the dry, while in 4.1 I don't think I have the same edge for pure speed as in 3.1 I was surprised with 4.1 consistency in less than ideal grip conditions. But so far I haven't tested enough time to really have a honest and meaningful opinion on it.
And before I can even properly test 4.1 now we get 4.2.
I must say whatever happened in 3.1 I have no idea since I don't understand the ffb code, but 3.1 has something very very right in it that allows me to extract insanely competitive laptimes and keep a huge consistency throughout.
With v4.1 i did my fastest lap ever in online multiplayer race in LFM of 54.2, which was surprising.

 

V4,2
I thought the FFB strength was 'low' - per using the same prior game settings.
On looking at lat_weight and long_weight, they seemed to be base values that set the overall strength for each direction - lat and long
To test I upped lat_long to 3.0 (!!!)
All I could tell different was the weight of steering (maybe all FFB?) got stronger. As a spin off, that was what I wanted to raise the FFB that V4.2 was weaker than V3.1, lol. Racing around, it didn't seem much different than just that - so I turned down my game FFB strength from 90 to 80, which was about a good steering weight for the GT3.

But the front and rear FFB reduction, on slide, was also like the V4.1.... whereas V3.1 had the front really inform you of that front grip loss.
When you go out on cold tires, the V3.1 really lets you know you are pushing notably wide (understeering a lot) but V4.1/V4.2 does not.
It was only after the tires got warmer - to racing temp - that V4.2 began to show SOME front end FFB reduction on understeer. And the rear was hard to tell - I can't tell if I IMAGINE there is some, or not. It is so very low. The lat_weight did not alter that - it seemed to only alter the overall FFB level.

As far as I can work out, the lat-weight is ONLY used in the RL_Slide/RR_Slide formula(s), which sounds like it makes sense... BUT then why didn't my big change to 3.0 (up from 2.0) alter that outcome? (response from rear SLIDE)
And then why did it make the whole 'base' FFB get much stronger, as I would think RL and RR Slide are used later, combined to the totals, but they would only rise IF there was rear slide - but it seemed to affect strength at ALL times.

Also: It seems to me that a lot of these 'base' type values should be up the front area as variables to easily find/access and change.
eg the following at least:
lat_weight
long_weight
load_norm_wheel
slide_factor

Under some "misc variables" title/group. With, possibly, comment on where they are used later. And what they do - if they do ONE thing. A bit messy to outline if they are used in many places.

 

I agree that V3.1 was 'better' in real information terms. That will be from the better front end 'range' from grip to less grip. Though it hadd a flaw that gave ZERO rear end grip loss information! (LOL)

V4.2 has rear end information added, but it seems to do little - for now.

But with the FFB files, I have to agree fully that using ANY of these (V3.1, V4.1, V4.2) helps you lap a LOT better! That information supply so you aren't relying purely on parrot-fashion, repetition, sequences. Though the current Default+ was OK too. But if you got to choose, then one of those rFuktor beat Default+... with V3.1 leading the pack.
Plain Default is terrible! Severely lacking.

 

I have a Moza R5 and the ingame gain on 25 and FX on 20 but stil fighting to much with my steering wheel, with the default + I can have both on 50 and have good feedback and real good timings because of less fighting. When I go on throttle corner exit with the Mini the steering wheel jerks to much, so I think for the FWD cars this custom ffb won’t work as good as for the RWD cars.

 

rFuktor has parameters to do different for FWD and RWD.... though I am not sure how well they work, or are done. But some parameters could easily be tweaked to reduce torque 'kick' I would expect - IF it has that issue.
So first thing..... give both rFuktor V3.1 and V4.2 a try, and report what you think....

 

In experimental 4.2 the best I've done was 0.54.060 in the same TimeTrial conditions. And it took me quite some effort to pull it off. It's almost as if I feel less? or it's a bit more slugish? hard to describe.

Many of my attempts ended up in the 0.54.600s and 0.54.500s etc... quite hard to push with 4.2 with accuracy and confidence.

I changed back to default+ settings that I used to do the 3rd fastest lap of TT and I was instantly doing 53.900s at my first attempt. I also checked some first impression tests from yesterday since I forgot to mention but with v4.1 I was consistently lapping in 0.54.100s in my attempts and also had a 53.989.

Maybe there could be some acclimatization to the new 4.2 but for now with this level of inconsistencies and just how hard it is for me to feel if I can push more or not (not to mention the amounts of time I've spun under trail braking)
(I either turned too much carrying too much speed) (turned too much while pressing too much brake) (braking too hard?) it was hard to feel. maybe easier to feel the rear tires under power? but just that.

For me my best bet it's either on v3.1 or v4.1. Somehow in 3.1 I lost the rear end of the car less times but it's a low power GT5, but specially during coasting into T1 of Brands there is a weight shift that can catch many people off guard and they turn in too hard and just spin. Somehow with 3.1 I had the most control, and with 4.1 it seems to be a bit similar to that but I didn't have enough time to test it.

 

"Tear FX" uses an 'oscillator' function, which I assume is to add some form of rise and fall cycle (sine wave?) to formulas if you want/need that. It seems to be the only place where that is used. So maybe the 'rogue vibration' comes from there.....

The "Engine FX", which must create 'vibration' also, does not use that function - weird.... and I can't even work out HOW it creates a wave for the engine vibration!

I disabled the engine section and it still has the weird vibration and thus I suspect the "Tear FX" section. But that looks a bit messy to 'remove' to test that.

---
Oh! The 'oscillator' is also used in the canned curb effects (bumps) area. So it might even come from there.

 

I think you 'learn' some of the grip feeling from the TEAR response. The tire slipping/sliding 'grinding' FFB that is also there. So whilst the drop in FFB steering strngth is missing for the rear end, you still use the Tear information to help out.
But, V3.1 V4.1 and V4.2 all have much the same Tear information, so if you DID use that then they should all be the same.

The main information 'drop' for V4.x is the less drop in FRONT end grip loss (not as much FFB drop per front end grip loss level).

That rear end loss type of thing you mention is EXACTLY what the rear end FFB reduction, for rear end grip loss, would help a LOT with!! Even though front end grip loss AND rear end grip loss BOTH come through the steering wheel exactly the same, you learn that in THAT track area the front grip loss is not really going to happen (or much) thus the FFB strength reduction information you get 'must' be from the rear - as is expected to occur there.

 

I'm going back to 3.1 and adding only some few of the "improvements" of the later files back in, and trying to make rear loss felt more again. So 3.2 is on the stairs.
All information has been very helpful, and I really like doing these ping-pong sessions. That's my fun in all this

 

Karsten... is 'this' a problem......? :

(RL_slide (min 1 (/ RL_scrub 15)))
(RR_slide (min 1 (/ RR_scrub 15)))

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

FL_Slide2 and FR_Slide2 are used in the Pnu section, but FL_Slide and FR_Slide were never defined anywhere before.
Or are they inbuilt definitions/variables?

And if they are not inbuilt, then what value would they take on if not defined? 0? 1?
The whole file obviously works, so I would assume they are given some valid value somehow....

-------
I tried removing so many sections, to find the rogue vibration....
It comes from the Rack - but I removed tons of the sections that 'cascade' that (rack_momentum, then big_rack), with each addon function along the way, and nothing removed ever fixed it anyway!

 

FL_slide and FR_slide are provided values, yes.
Did you try removing all effects?

 

I copied the new V4.2 rear slide section into V3.1
I am pretty sure that block can just be put into the prior 'weeny/basic' section of the rear slide.
It all works - it seems - BUT it is still lacking enough steering strength reduction for rear slide. And as per before, I THINK I feel some - and that it is not just coming from the front slidiing a bit. It is hard to differentiate really!

I assume RL_Slide etc means the REDUCTION value when they do slide?
I can't really work out of they are negative values, or used later as a negative. I would assume so.
Maybe the Velocity aspect is also involved.... in limiting the value of change.
All that new 'complex' coding is giving me a headache to decipher the maths! LOL

 

Yes, everything removed EXCEPT the rack.... and the vibration is still there.
I THINK it reduced when I lowered the LFB... or was it the FX(?).... hmmm.
But anyway, it comes down the rack.....

 

"rFuktor ? experimental 3.2"

Subtitle: "Trying to improve on 3.1"

* Rear slide is felt. Now full slide on all 4 tires almost kills the ffb.
* More turning force than 3.1, both slow and fast.
* From 4-series: Tire effect, less lagged downforce.

Adjust how much the rear tires sliding should affect the FFB strength here:
(rear_grip_loss_weight 1.0)
(front_grip_loss_weight 1.0)

Adjust more/less cornering top force by adjusting these 2 lines:

(front_tyre_stretch_feel 0.275)
(rear_tyre_stretch_feel 0.275)

 

Oh... so when I INCREASED the lat_weight (and later long_weight) and that made the whole FFB much stronger, why would that be?
IF the slide portions are just to use slide values, with some maths, to REDUCE the FFB then, a greater lat_weight should make it all LIGHTER (weaker) not stronger. ???

The front end slip/slide via FFB still seems to be giving information - so if it was inverse I would expect that to get less (maybe it did?).

(RL_slide (min 1 (+ (* lat_weight (abs RL_slide_lat)) (* long_weight (abs RL_slide_long)))))
(RR_slide (min 1 (+ (* lat_weight (abs RR_slide_lat)) (* long_weight (abs RR_slide_long)))))

Oh yes....
(M_FL (* 10 M_FL (+ 1 (* (- 1 FL_slide2) FL_stretch2 front_tyre_stretch_feel) (* RL_stretch (- 1 RL_slide) (- 1 RL_slide) (- 1 RL_collapse) rear_tyre_stretch_feel))))

-1's......

So WHY did the FFB get stronger overall for larger lat_weight values? It seems they are ONLY used in the Rear Slide values.... which then are used as NEGATIVES..... (reduce FFB strength).
Hmmm... odd... not sure now.... seeing the script/maths says it would not get stronger (but it did!).

 

Yes, that sounds weird. Those weights are scales on lat/long, so the higher they are set, the sooner the slide feel should happen. In 3.2 I changed it back more similar to how it was earlier. There is a magic "50" scale used on the lateral part now:

(RL_slide_lat (abs (* 50 (power RL_sl_lat2 2))))
(RR_slide_lat (abs (* 50 (power RR_sl_lat2 2))))

Both that scale and the power of 2 can def. be adjusted. Very time consuming to get all those numbers just exact right, unf.

 

Thanks for the V3.,2 !!

BUT... the (rear_grip_loss_feel 1.0) use is commented out and doesn't truly do anything (in the V3.2)

(m_FL_load (/ m_FL_load load_norm_wheel))
(m_FR_load (/ m_FR_load load_norm_wheel))

#(grip_feel_rear (min 1 (/ (+ (* grip_rl rel_weight_FL) (* grip_rr rel_weight_FR)) (max 0.0001 (+ rel_weight_FL rel_weight_FR)))))
#(grip_feel_front (min 1 (/ (+ (* grip_fl rel_weight_FL) (* grip_fr rel_weight_FR)) (max 0.0001 (+ rel_weight_FL rel_weight_FR)))))
#(grip_l (max 0 (- grip_fl (* rear_grip_loss_feel (- 1 grip_rl)))))
#(grip_r (max 0 (- grip_fr (* rear_grip_loss_feel (- 1 grip_rr)))))

(grip_feel_rear (min 1 (/ (+ (* grip_rl RL_load_norm0) (* grip_rr RR_load_norm0)) (max 0.0001 (+ RL_load_norm0 RR_load_norm0)))))
(grip_feel_rear (smooth grip_feel_rear 0.03))
(grip_l (* 0.5 (+ grip_fl grip_feel_rear)))
(grip_r (* 0.5 (+ grip_fr grip_feel_rear)))

It would also be good to be able to tune BOTH the front and the rear independently.
And... mention a range of valid numbers. (even if just 0.2 to 1.0 etc - but maybe it can go to 2.0 fine ?)

Also note grip_l and grip_r with no front of rear portion. Expected, intended?

 

Ah, right, it's hardcoded in those lines to correspond to 1, will change the file so it's in there.

Updated the file in that post, just download again.
"front_grip_loss_weight" and "rear_grip_loss_weight"
are now the ones to change.

 

This sounds weird,
I def. will have to take a look soon.