Automobilista 2 Custom Force Feedback - Overview & Recommendations

This last ATMOS BASE + Partial NDef_Imm file will probably be the one that will surprise you the most for better or worse. It means that you will perhaps love it more than the previous ones (especially if you liked the predecessor file Initial NDef_Imm), you will perhaps love it less than the previous ones (if your tastes were probably oriented towards the Partial Old_D_Raw version) or perhaps you will ignore it completely because you didn't already like it none of them to date and so you don't want to waste your time, no problem, I enjoy testing with code and therefore I take this into account too.

The sensations will change further, there will be some new variables specific to each individual car, but above all a quantity and type of code will be implemented which is quite invasive, the cursors in game will have to be slightly recalibrated again but it will be a natural adjustment (just a few steps), so I hope to not having made any mistakes, the list of changes will be published later, I will therefore update this post.

Having said this, you should understand why up to now I have not dedicated a single minute to optimizing these files in advance, even if I risk criticism from the most demanding users, but it would have just been a useless job because there were too many changes in progress.
Now I have finished the development with this file and therefore from tomorrow (give me a few days' break) I will be able to find the best settings and apply them backwards from this last file up to the first at the same time.

ATMOS BASE + Partial NDef_Imm 1.5.7.X


CHANGELOG - BASE + Partial NDef_Imm
(only further changes compared to the previous BASE + Initial NDef_Imm version):

Implementation of the new variables necessary to obtain others specific customization for single car:
- vehicle_aero_load_reduction
- vehicle_aero_load_limit
- chassis_lat_accel
- chassis_lat_weight
- chassis_long_weight

New variables and code for signal compressor.....
- dry mix
- wet mix or vehicle_compressor_wet_mix
- threshold or vehicle_compressor_treshold

Implementation of code for "PARKING LOT FORCE" (Only in Rack Momentum and not in the Output)

New code for Dirt line;

New code for front_rear_diff, left_right_diff, load_diff, load_diff_reduction, lat ....... (chassis);

New code in Damping for decrese damping when front end lightens (front-rear diff).

-----------------------------------------------------------

Spoiler: CHANGELOG FOR BASE + Initial NDef_imm VERSION

CHANGELOG - BASE + Initial NDef_imm
(only further changes compared to the previous BASE + Partial Old_D_raw version):

Implementation of the new variables necessary to obtain specific customization for single car, in order to align the sensations for cars not only to physics but for how Reiza interprets them in its second "custom+" profile differently also called immersive (it is anticipated that, the content Of the individual variables and/or the values they correspond, they will not be the subject of divulation and/or public discussion, as we do not have vision, we see the immersive profile so the work of others will be strictly respecting).

Implementation of the Code necessary to introduce the "Linearly Decrease Cutoff Freq As Velocity Incaceses" for LP optimizations

Following the above, the file code was modified to prepare the implementation (in this initial version) of these first personalized characteristics:

- Vehicle_accel_mz
- Vehicle_braking_mz
- Vehicle_grip
- Vehicle_yaw_mult
- Vehicle_damping
- Vehicle_frication

- Vehicle_lp_cutoff
- Vehicle_lp_cutoff_min
- Vehicle_lp_Vel_scale

- Vehicle_Ngine_vibe
- Vehicle_flatspot
- Vehicle_scrub_mult_front
- Vehicle_scrub_mult_rear

- Vehicle_Rodnoise

Therefore the entire code of the previous file was hooked to the variables mentioned above.

If you are interested you can find the third file ATMOS 1.5.7.x BASE + Initial NDef_Imm version here:
Automobilista 2 Custom Force Feedback - Overview & Recommendations

Spoiler: CHANGELOG FOR BASE + Old_D_raw VERSION

CHANGELOG (BASE + Partial Old_D_raw version):

New code implemented in the slip:
-calculation of tires average lateral slip front & determination of the front lateral scrub breaking point;
-calculation of tires average lateral slip rear & determination of the rear lateral scrub breaking point;
-calculation of tires average longitudinal slip front & determination of the front longitudinal scrub breaking point;
-calculation of tires average longitudinal slip rear & determination of the front longitudinal scrub breaking point;
-calculation of tires slip_speed_rear_avg;

New code implemented in the yaw:
-allow yaw sense only on oversteer;
-calculation of the oversteer multiplier;
-calculation of reduce_load_norm;
-calculation of slide_lat_avg_rear_yaw;
-overall redetermination of the yaw following the changes indicated above;

New redetermination of the scrub effect, now following the changes in the code indicated above:
-insertion of the lateral break points (front and rear) into the scrub;
-insertion of the longitudinal break points (front and rear) into the scrub;
-insertion of the slip_speed_rear_avg only in the rear scrub, now the effect in the rear is reduced if sliding;

New redetermination of the tear effect, now following the changes in the code indicated above;
-insertion of the slip_speed_rear_avg only in the rear tear, now the effect in the rear is reduced if sliding;

New code implemented in the damping:
- new code for increase damping on oversteer with compensation;
- new code for damping on low lateral force and under 150 km (lateral and velocity compensation);
- new code for reduce damping if is sliding (sliding reduction).

If you are interested you can find the second file ATMOS 1.5.7.x BASE + Old_D_raw version here:
Automobilista 2 Custom Force Feedback - Overview & Recommendations

Spoiler: CHANGELOG FOR BASE VERSION

CHANGELOG:
What's new in how objects work (with or without new code in the file):
- New "tyre_resistance & timing" now calculated on 5 moments and not on 3 (the functioning and the formula was misinterpreted in the previous files because it was unclear);
- Center_Full low_velocity (0 Kmh) during acceleration morphs to high_velocity (250Kmh);
- Centering recalibrated and force_scale set to manual (small differences on both values bring great benefits so I didn't use the formula);
- The code relating to velocity and the gyroscope has been slightly modified, so now all velocity variables can basically start from a value of 1.00 (modify anyway if you like);
- Removed the front_grip_loss_feel (this was an error in my previous files the formula was misinterpreted and instead the rear_grip_loss_feel alone was enough to get what you want);
- The follow_weight dynamics are now 5 different (accel, brake, bump, road and suspension);
- The "front & rear drive_torque_feel" are now calculated differently or rather both in an increasing manner and as a function of the speed from 0 to 250 km where above we will obtain the maximum longitudinal forces;
- The RELAXES have all been disabled (OS, US and YAW) and the code inhibited, the car is very balanced without them (but if you want to re-enable them you can do it by deleting the # symbol both in the objects and in the code below relating to the Relax);
- Now you can also set the smoothness for textures and for both large bumps (not just for micro bumps);
- New scrub and tear values added, now with additional variables to change on both effects to your liking (we have hardness, damping and effect);

New code only:
- KART code move up before the dynamic code and enjoy this now this feature, it still to be tuned or WIP, we will do it later (code disabled if you want to do so, delete the # symbol in front of the respective lines);
- We have a new code "if isTRUCK...." i.e. I have implemented, if you need a different melody from the base, a specific code for the TRUCKS to be set to your liking, it still to be tuned or WIP, we will do it later (code disabled if you want to do so, delete the # symbol in front of the respective lines);
- Added rolloff, offTrack and offTrackRollOff code to many items of the code such yaw, scrub and road_effects (this will make more sense later with the other files that implement the reiza code);
- the Slide variables are no longer connected to the scrub (even only to the rear one) but only to the original Slide variables (this will make more sense later with the other files that implement the reiza code);
- New load_norm_wheel, now the value is no longer fixed at 250 but is variable based on the weight and mass of the car, this allows you to lighten the load (the wheel) if you are in the air, on the grass or if you slip , furthermore the variable value will influence the way you feel curbs and bumps (depending on the car) and enhances surface variations as happens a lot when sliding on dirt roads (RALLY is great fun) or bump roads and reduce that sensation of weight which is always perennial (in subsequent files this change will be more noticeable when we implement the REIZA code);
- The REAR has been removed in the ATMOS. Not because it is wrong but because the file already has the entire rear (slide, grip, textures, curbs, bump and all the effects) and does not need to be implemented as in the original file because only the front variables are calculated there.

If you are interested you can find the first file ATMOS 1.5.7.x BASE version here:
Automobilista 2 Custom Force Feedback - Overview & Recommendations
-----------------------------------------------------------

I just ask you to mentally start from the beginning by resetting your brain otherwise feedback of superiority or inferiority is absolutely useless to me if based solely on useless comparisons ... this is another file!!!

Info for Tuning:
- this version of ATMOS 1.5.7.x "BASE+Partial NDef_Imm" file does not have tuning optimization and is identical to the first "BASE", to the second "BASE+Partial Old_D_raw" and to the third "BASE+Initial NDef_Imm" to allow you to immediate comparison, if instead you have modified the first, the second or the third BASE file enter the same values to compare it better;
- however, if you are using it for the first time, I have no indications to give, tune it and adjust it as you see fit.

I remember that the file is a little pumped in the values and settings and therefore has a little excessive movement and the powerful DDs could have oscillations if guided without the hands, please focus on consistency and coherence, everything will be reduced to finished development ... I hope next week with the last file (for now) ... then i start the optimization or tuning.
Tested on the last AMS 1.5.6 and beta 1.5.8.4 versions

The next file??
For now we won't have any more files, so we will move on to optimizing them, just give me a little break, we will have time to optimize until the release of version 1.6 of the game and if everything goes well we will finish with the final files if necessary otherwise these 4 files will become definitive

Have fun, I hope so....

 

Even if @Stakanov is too obsessive to say that a file is okay, I once again underline to users that in these files it is very important to experiment and find new values on the cursors in the game, try more cars and more tracks, if you manage to find a correct balance between the cursors also the optimization will become superfluous or just a surplus, but 90% of the work would already be done... I myself haven't changed much, just some rather subjective values.

If you come from classic custom files, set the GAIN to your 70-80% of ideal even if the steering wheel is a little light for you, season with a bit of LFB even if you are on DD (obviously very little or 0 if you have 30NM steering wheels), raise the FX until you reach your ideal level or your personal taste even if you notice some small anomaly, now raise the DAMPING until the previous anomalies disappear, now fill the lightness or lack of power with the missing GAIN by going up (if necessary) until the moment that the anomalies or the untolerated oscillations return, if necessary go up again with damping but rest infomative...

Leave behind the mentality of the obsessive search for maximum gain because I have an expensive steering wheel and therefore I have to drag its neck... instead obsessively search for maximum information without necessarily having maximum gain (obviously remaining with a strong and full steering wheel) ... we spent our money to be able to have this possibility, not to treat our DDs like a T300 or a G27 which has no alternative to raise the gain as much as possible ...

My son will kill me as soon as he reads what I wrote and now he will want a DD too..

 

Simple ... if you read the file in the note I think they are clear...
to reduce the surface kerb go down 7.50 ... 7.25 ... 7.00 ... 6.75 ... search your value
the same is if you want reduce the tall curb 2.50 ... 2.00 ... 0.50 ... search your value

(kerb_surface_scale 7.50) #How much you feel the kerbs_surface effect. Suggested range-values: 5.00-10.00.

(kerb_bump_scale 2.50) #How big the tall curb bumps are. Suggested range-values: 0.25-0.50 or 2.50-5.00 (to taste but depending on how high you set kerbs_surface_scale & smooth).

 

To be honest I feel with your settings and the recommended ingame settings a little bit too weak for the alpha mini, I raised the gain to 100% to make it stronger, do you think I will loose the intended feelings you want to propose with your file or I'm fine?

 

Thanks for your work and your time.
What does the kerb_surface_smooth value do? I thought that's what cuts the curbs.
How does kerb_surface_smooth affect the other two parameters? Do you recommend leaving it as is?
Thanks

 

Good morning! I assume you mean increase in-game? Ultimately it's always a matter of taste and if you like it better then you can of course use your preferred settings. Everything is possible, in the end it's your taste that counts

 

The same for me with CSL DD 8Nm. I change it car to car and wheel to wheel, but I never went below 85% with 29cm prototype wheel (anything over 32cm is full 100% and the base is too weak).

 

From what you asked I think you need the answer to the question.
What is smoothing??

In statistics or mathematics (and therefore also in these files based on LISP language), the "smoothing" of any function consists in applying a filter to highlight the significant patterns and attenuate the noise generated by the artifacts of whatever nature they may be.

This process is useful but has its pros and cons, two very distinct aspects must be seen:

- Indirectly Quantitative. Smoothing improves the quality of the data by reducing the quantity of unwanted variations (especially in those functions or value trends with strong variability) and therefore the excesses are cut and a function that is too unstable is flattened. This process is useful until it becomes excessive (as well as counterproductive) because there will come a point in which the loss of significant data occurs which is no longer counted and therefore is no longer taken into consideration with the obvious loss of information and inconsistency.

- Directly Temporal. From what has been expressed above you will already understand that the smoothing filter is not really quantitative on the data (i.e. it does not only eliminate the peaks above a certain limit, that is another function), smoothing is also and above all a temporal update filter.
It is practically the answer to this question: with what speed or rather with what timing is the function free to update and move, in other words for how long must the value or result not be free and must remain unchanged?
The smaller the smoothing values are, the more the data remains free to update itself quickly and the values of the function are updated accordingly (removing only the values of some time fraction), while, the larger the smoothing values are, the slower the function is in updating the result (the value of the function will remain unchanged for a longer time), making it delayed.

Therefore choosing the appropriate Smoothing Parameter to influence the original data is crucial.

In the case of kerb_smooth (but you can apply it to the bump or any other function), if you raise the smooth value the variability of the effect over time will slow down, you will avoid some variations and you will slowly start to lose information, if you continue to increase sooner or later the effect will turn off because the data received from physics will be lost and not updated, on the contrary, if you lower the smooth value the variability of the effect will increase, the values will update quickly and you will not lose any information received from physics including unwanted peaks and the jolts. So if you want to intervene on the update speed of the effect then change the smooth... be careful because small changes in the value correspond to big changes in the function.

Link with the kerb_surface_scale: It is now clear that the scale is set according to the smoothness, if you turn off an effect with high smoothness values you will have to increase the scale of the effect to feel it again (provided you do not exaggerate with the smoothing and do not turn off the original function), on the contrary if you keep the "smooth" low you can reduce the "scale" because the variability is fast and high.

 

If it's too weak for you then simply increase the main switch at the beginning of the file (master_rack 0.70) to your liking

 

Thank you, yes I meant in game settings. I thought that changing them from your recommendations could break the purpose of the file. As I said on high gforces the wheel seemed too weak and also on curbs. I have the alpha mini paired with a gt neo, from the disclaimer into the setting file you recommend to add some inertia to the wheel into the base manager settings for lighter wheel, what would you suggest for my combo? At the moment I have the inertia set to 0 as per your recommendation, I was thinking maybe 15?
A thing that I like of I racing ffb is that when the car hoes above 200km/h the wheel becomes very hard to turn (which makes sense), and this is what I mean by high g forces. Also in Iracing when you get a high curb the wheel gets hit very hardly giving you more the feeling of loss of grip over a curb. How can I achieve that? Would increase in game FX work? (Now it is set to 25). Is there perhaps a way to increase the visual effect as well (like more bounce for example). For the rest the file seems perfect

 

Hey!
I drive with about 12Nm with a 32cm diameter wheel. You drive with 10Nm and a wheel (GT Neo) with a diameter of 30cm. Of course there are differences here.
Yes, you can increase the "inertia" on your base for your lighter and smaller wheel to simulate a heavier wheel, maybe 10 or 15 depending on your taste. As I said to "GoobMB" you can also increase (master_rack 0.7 ) in the file, maybe increase it to 0.8 or 0.9 for your 10Nm base. If you want more resistance at higher speed you can increase the value (velocity_res_scale 0.6). I wouldn't set the FX higher than your 25 on your Simagic Mini, otherwise the wheel could swing and clipping. For the curbs you can try to smooth less, i.e. reduce the value ( micro_bumps_smooth) to maybe 0.005 . Maybe this will help you a little, it's very difficult to give tips because every user has completely different priorities. And just as important as the file itself are the in-game settings and the base settings and the hardware you own

 

I just tried this file in the beta in the older F1 cars around Kansai GP and I’ve been getting an issue. When I turn through the esses. It seems to me that the effect of tyres gaining grip after losing it is too powerful so you easily lose the car. It feels very erratic

 

Thanks, I'll take a look at it as soon as possible, I'm away from the PC for a couple of days!!
Can you check if with the same settings and the same machines you have the same defect only by switching to the Reiza Default+ profile??
Which cars exactly, the retro ones. the classics?? ... Thank you again

 

It was with the classic G1 and G2, Retro G1 and G2. I will check with default+ tomorrow, but I remember having a similar experience in the F-U G1 around Azure. I had max DF and tried to turn quickly at the top of the hill. The front gripped up heavily and I hit the wall. I had a similar experience with the default+, but the custom file amplified it

 

Thank you for your insights, I was indeed explaining what effect I would like to have and how I can personalize your file to my needs in the way I can achieve what I need , there are other parameters to consider for high speed, high g corners, curbs?

 

Wow, thanks my friends.
Your file is amazing, I love it. I was wrong to set up the Fanatec panel; I used to set the EIF too high (now I keep it at 60/70).
Thanks for the work you are doing.

 

@Stakanov I also noticed the same attitude in Kansai at the exit of the S, with the Auto F Classic Gen 1 and Gen 2 (then i think the retro too) ... I improved a lot by modifying these two values relating to acceleration and with level the gain 1-2 step

(accel_feel 0.75)
(accel_feel_follow_weight 0.25)

Little by little we are fixing everything, we have a lot to adjust aspect by aspect, thanks to @Racinglegend1234

 

@Rintintin78 I also like the start but the difficult part begins now because we need to make it pleasant in a neutral way for as many people as possible and there are many things to adjust and level ... and as you can see, a value can be modified to change everything... when I am told that an aspect needs to be improved there may be something behind it infinite different reasons that I'm not aware of... but little by little we'll get there.

Ok thanks for the work in my absence, I'll take note

 

@Stakanov very clear and illuminating... but if the smoothing filter is a temporal update filter of the data, then eliminate the values below a certain timing and therefore smoothes the functions regardless of its values, what are the others functions that influence the values of the variable quantitatively and not temporally?

Mine is just curiosity and I'm a little fascinated, I hope I don't disturb you!!

 

For example, two quantitative and non-temporal functions are the hard_clip function and the soft_clip function.
The first clearly cuts the maximum (or peak) value that a function or variable can have (red line in the photo), the operator inserts the cutting limit and the hard_clip function does not make the result go beyond, which in fact always remains at the maximum until the value or result is higher.The pro or advantage of this function is that it leaves everything below the limit unchanged and therefore faithful to the original, but on the other hand it will always remain at the maximum value in all values positioned above the cut.
The second function, however, may seem similar to the first but implies big differences.It does not implement a clean cut but readjusts the entire curve so that the maximum is the one indicated by the operator (yellow line in the photo), therefore it has the advantage of not flattening out on the established maximum, but has the disadvantage of deforming a little the entire curve, including the underlying one, so that the maximum is the established one.

This image helps but doesn't show the deformation I'm referring to... I'll see if I can find another photo....



EDIT: Maybe here you can understand it better... I hope.