Automobilista 2 Custom Force Feedback - Overview & Recommendations

Gunboat Diplomacy v1.2 is very very nice indeed. best so far for me. It has taken a lot of the harsh steering rattle away which also means there is more useful sound vibrating through my rseat cockpit so rear end scrub is so much more noticeable now as well. All I feel is missing now is a little more centering forces and resistance to turning the wheel. Brilliant job @NuScorpii

 

The anti-oscillation works very well thanks! sorry for the late reply. I'm gonna try to fine-tune it now, but for reference, I lowered :
(scrub_scale 0.3)
(kerb_scale 0.2)
(gyro_scale 0.6)
in-game: 65.35.50 I might increase scrub a bit.

 

Which file are you running? There's 3 gunboats now.

 

So I have modified even more now and to my taste, it feels great! I think is the diplomat?
I can really feel the weight transfer, especially with the Sigma, freaking awesome to drive now.
In-game 70.35.50 - if you try it keep in mind I'm on 100FEI.

 

For sure man we are on the same page - thanks for giving that a try and ruling this out to confirm that the files are indeed affecting our T300 wheel bases differently with the same file/settings. This is very good information to have, as I never expected the exact same settings to give the same experience, but hopefully similar enough that we might be able to provide a baseline for others to work from going forward with this file.

Thanks mate for the feedback! It does seem that my wheel base is more sensitive to LFB than others. I don't know if wheel base age might make a difference, as I know internal parts are typically revised through the lifespan of these products. I have a friend with a T300 that I'm passing my files over to and he's getting very good results with my settings, which again, is interesting. I do think trying to find a baseline for the T300 might not be worth the time investment, as the range seems to be too great between the same wheelbase.

 

Hi @NuScorpii
This is my first forum post here, saw your comments on Ermin Hamidovic YouTube video, and have just tried your profile on my SimSteering2 on version v0.9.6.0 using your Custom V1.02, great stuff THANKS

 

Hi Can you attach the v1 file as there is just updated one now which I found very "grainy' with too much engine and road noise. Thanks.

 

here you go

 

Hey guys! Huge breakthrough today in the feel of the FFB when at the limits of the output. (@Shadak You're going to love this)

I have rewritten the clipping code so there is no hard hit. Instead it is using sigma functions to gradually taper off towards the clipping point. Each range of output is clipped separately so even when reaching the limits of the output you will still be able to feel the subtle effects such as road noise, scrub, kerbs, etc. This really has made a huge difference IMO, enough to warrant a new name

"Heavy Messing"

V2.0

* Completely new way of clipping the signal data using Sigmoid functions. This will enable high output gain to be used with very little chance of encountering flat spots in the force feedback.
* Bumpstop code changed to have a heavily damped buffer so there are no hard hits or bounce back.

 

legend! let me try this out

EDIT: this is absolutely incredible, no more empty FFB through half of Suzuka in F-Ult. This was probably my only ffb issue I thought might not get fixed .. Are you secretely Heusinkvelds FFB specialist or something?

 

@Renato Simioni Dude, I am sure this post will be of much interest to you; it appears user @NuScorpii has uncovered quite the bug in the default clipping code (regards the IF statement) as well as adjusted for safer collisions for those with DD wheels....someone buy this effing guy a beer already.

 

Just to expand on this, the "if" statement is expected to work like this:
if <statement> <output if true> <output if false>

However that is not quite right, it actually is:
if <statement != 0> <output if true> <output if false>

So both positive and negative values equate to true, only 0 goes to false.

This is used in the clipping code to give 30% headroom, however this only does the limiting to max of 30% when gain = 0.7

So

(headroom ( - 0.7 gain))
(clipTo ( + 1.3 headroom))
(clipTo ( if headroom 1.3 clipTo))

For low values of gain such as 0.3 when using high strength wheels will actually result in a clip to of 1.7, not 1.3.

This could potentially lead to huge spikes of output for DD wheels. This can be fixed by using:

(clipTo ( if (max 0.0 headroom) 1.3 clipTo))

However, the sigmoid clipping is even better

 

Update on the big file:

* Merged NuScorpii's latest front load effect over.
* Avoid clipping (using weights for full control, since this is the large file anyways & this allows 100% use on higher end-wheels with no transformation on ffb).
* Effects based on weights.
* Possible to change the weights with no chance of introducing clipping (also allows engine vibes to fill more, relatively when needed).
* Enhancements and effects normalized.
* Internal adjustment to gain/fx to keep forces in range & not in clip.
* Alternative slider control can be enabled :FX slider from ONLY RACK to ONLY FX - Makes blending effects weights easier when setting FX at 100 in this mode.
* Less harsh LFB values on road details.
* Road & effects LFB according to FX slider, not LFB slider, which is reserved for the big rack force only.
* Removed lfb from rumble & road vibes.
* Adjusted to soften lfb on road details.

In this file, now the "weights" at the top are used to adjust each effect, so please make the adjustments there.

Custom ffb file attached "Measured Response":

 

I'm really liking the changes made to this file! The constant noise issues I was having has been alleviated and there does seem to be a much better balance between the LFB and FX effects now. Also, with the clipping limit changes I can now utilize higher Gain settings for open wheel cars. The combination of all of this means that for the first time in AMS 2 I've been able to run the V10 and have the same level of confidence in the car as with the GT cars. These files just continue to get better and better! Thanks again guys!

 

@NuScorpii @Karsten Hvidberg Good morning fellas. Just playing 'catch up' after a few days away. Looks like a ton of progress has been made. Do you guys ever sleep?

Quick question. Is 'Measured Response' an update to 'Heavy Messing' or are both files standalone?

Cheers!

 

They are each their series, but highly related.

The small files are "NuScorpii solo" files that he controls, trying to keep it simple.

The big files provide additional effects, multiple wheel type settings in the same file, adds "smart" slider control and all effects are weights based instead of absolute. Also tries to detect wheels off road and disable effects based on that.
The big file is a collaborative effort.

They are tweaked slightly different.

Just try both, look at the top where you can make adjustments and see what you like.

 

As Karsten says, they are standalone developments.

The series of which Heavy Messing is the most recent is designed to be simpler using only effects that make a positive and noticeable difference. This means it is easier to setup due to fewer variables to tweak and you are less likely to mask subtle effects with others. It currently has a much more advanced clipping algorithm that prevents the loss of details and effects even when using high gain with the F-Ultimate. This will probably make it's way into the other file at some point though.

At the end of the day they are different flavours, so try both.

 

Cool, you've merged some of my new developments

One of them is actually a huge breakthrough in what we can do with the custom ffb files so is worth detailing here for reference and for anyone else that is trying to do similar things.

The currently accepted thinking of how the custom ffb works is that every time it is run through it starts fresh, all variables are assigned and you can only operate on the inputs to create an output. i.e. a function that can be made fairly complex. This works for most effects where you are just influencing the shape or scale of the input. However for the load based effects I was working on you needed to have a reference weight of the vehicle otherwise the effects would be blown out of proportion between light and heavy vehicles (There isn't a static weight for the vehicles provided, just the real time load of the front wheels). One way around this was to use the smoothing function to give a kind of average over time, but we don't know how the smoothing function works and it leads to odd effects such as ghost force from previous high load situations.

What I managed to find out was how to persist variables across runs through the file. This allows you to create counter variables. That in turn opens up the ability to write custom averaging functions allowing a much more accurate and static average front load weight for a vehicle.

Down to the specifics:

If you declare or use a variable but don't initialise it, it will be auto initialised to 0.0 the first time it is used. So if you do this:

(new_var new_var)

The first run through the file will assign 0.0 to new_var and assign new_var to itself. You can use this to make a counter:

(counter (+ counter 1.0))

However, the variables do not get reset between different vehicles being loaded so a way is needed to reset it. This can be done by using the inbuilt variable et - elapsed time. This gives the elapsed time from the start of the current session. So to reset the counter we can do this:

(counter (+ counter 1.0))
(counter (if (max 0 (- et 0.5)) (counter) (0.0) ))

This will start counting from 0.5s into the current session. You can then calculate a running average of the front load weight to determine if the car is under heavy load or light:

(moving_avg (+ moving_avg 0.0))
(moving_avg (if moving_avg moving_avg f_load_avg))
(moving_avg (+ (* moving_avg (/ (- counter 1.0) counter)) (* f_load_avg (/ 1.0 counter))))

And then compare that to the real time FL/FR_loads.

 

Ignore. Double post.

 

Suggestion: Make these 2 file branches easily distinguishable by their naming, by following some sort of theme or convention.

Eg: All files derived from Gunboat Diplomat could begin with the word 'Gun' or have a boat reference in the title. Even a letter prefix like 'A' / 'B' or the Red / Blue file would be useful.

Loving your work guys!

 

What !
Seriously though , thanks so much to both of you geniuses for the hard work you have put in . Have really enjoyed this thread and testing out the fruits of your labour