39 replies to this topic

[Dark]

You have restored my faith in force feedback sir! Nice work. Thank you very much  

[damianpanaggio]

thanks!

[fgubbels]

After several trials and races, my conclusions are that this add-on is really cool. The ffb is more realistic than before. Many thanks Petteri, continue to provide us nice tools like this

[Ken]

I tried it but I wasn't sure. It certainly gives a very clear feeling that the fronts have lost grip but it didn't seem to have any transition. I could feel resistance on the wheel, then suddenly non and it was like there was no ffb at all until the grip came back. With old ffb the damping seems to give that feeling of some transition from grip to no grip. Just my opinion. Peteri's new patch certainly felt very immediate I have to say.

[Larko29]

Anyone with a Logitech Black Momo tried this patch? What settings did you use? I tried the default but it felt like I was driving on ice, didn't have time to tweak much as I was racing online.

Keen to experiment more though.

Thanks Brr

[brr]

Ken, on May 4 2009, 01:44 AM, said:

I tried it but I wasn't sure. It certainly gives a very clear feeling that the fronts have lost grip but it didn't seem to have any transition. I could feel resistance on the wheel, then suddenly non and it was like there was no ffb at all until the grip came back. With old ffb the damping seems to give that feeling of some transition from grip to no grip. Just my opinion. Peteri's new patch certainly felt very immediate I have to say.

That sounds correct. If damping is preferred, the patch isn't too useful since it removes it entirely. The way damping is implemented, it will not give useful feedback about car behaviour which is why I prefer to get rid of it.

[freejrs]

This patch sounds perfect for me. I lost all my ffb settings in GPL and have been trying to get back to where I was before for a while now. Seems like I may not need to with this patch. Will give it a try as soon as I get GPL to run in 1920x1200 with the new rasterizers and at 60fps......

A long time then.

[charli181]

A bit off topic here but can you discuss how you got the physic figures from within GPL. I am looking into motion simulation with the x-sim project (http://www.x-simulator.de/forum/) and this info would be handy to getting the motion working for GPL... currently gpl crashes when trying to access the FFB whilst running x-sim.

regards

Charli181

[brr]

charli181, on Jul 31 2009, 04:28 AM, said:

A bit off topic here but can you discuss how you got the physic figures from within GPL. I am looking into motion simulation with the x-sim project (http://www.x-simulator.de/forum/) and this info would be handy to getting the motion working for GPL... currently gpl crashes when trying to access the FFB whilst running x-sim.

regards

Charli181

For motion simulator, you would be better off using similar techniques as Pribluda. Maybe this library by Tom de Muer would suffice (see the mini SDK):

http://users.skynet....demuer/gpl.html

I haven't used it myself so can't say anything about how to use it etc..

[Uwe Herrling]

brr, on Jul 31 2009, 06:59 AM, said:

...
For motion simulator, you would be better off using similar techniques as Pribluda. Maybe this library by Tom de Muer would suffice (see the mini SDK):

http://users.skynet....demuer/gpl.html

I haven't used it myself so can't say anything about how to use it etc..

Well, I did use Tom de Muers mini SDK for exactly the same purpose (retrieve realtime G forces from GPL), and it worked very well for me. Based on this I built a small lightweight server that is started up together with GPL on the same machine and is sending the retrieved data textbased to a network socket, so that you can link any network client to it. This eliminates all restrictions for the implementation of the actual motion platform software. Contact me if that might be a useful option for you (this of course applies to everyone )

The FFB patch.
I also felt that the FFB with the patch applied was very clear and gave a good grasp of what was going on with the car, but the overall expierience was not realistic IMHO: At least it lacks the stabilizing effekt of the rotating wheels: You can dash down a long straight at topspeed whithout feeling any forces at all, the steering wheel beeing completely loose. Which suggests that these forces are not modelled directly in the GPL Engine (?)

I think I will switch back to unpached FFB, but nonetheless: Thanks for sharing this!!

Grüße
Uwe

Edited by Uwe Herrling, Jul 31 2009 - 07:35 AM.

[brr]

Uwe Herrling, on Jul 31 2009, 11:59 AM, said:

The FFB patch.
I also felt that the FFB with the patch applied was very clear and gave a good grasp of what was going on with the car, but the overall expierience was not realistic IMHO: At least it lacks the stabilizing effekt of the rotating wheels: You can dash down a long straight at topspeed whithout feeling any forces at all, the steering wheel beeing completely loose. Which suggests that these forces are not modelled directly in the GPL Engine (?)

It looks like the GPL FFB is based on slip angle and lateral force. So driving straight does not generate a nonzero FFB force. However, the curve (a function of slip angle) which affects the forces seems to make the wheel unnecessarily loose near zero slip angle. I've experimented with modifying the curve, and got the wheel to feel much more steady near zero slip angle. The problem is to modify the curve consistently so that forces feel ok at all slip angles.

[Uwe Herrling]

Thanks for your reply!
Would it be possible for the patch to take into account the current speed/rotation of the front wheels, as a third parameter? If this was possible, the necessary calculations (correction of the curve) should be a simple addition of a second curve (maybe even linear as a first approximation), and I reckon it should improve realism considerably.
Grüße
Uwe

Edited by Uwe Herrling, Jul 31 2009 - 07:19 AM.

[brr]

Uwe Herrling, on Jul 31 2009, 04:03 PM, said:

Thanks for your reply!
Would it be possible for the patch to take into account the current speed/rotation of the front wheels, as a third parameter? If this was possible, the calculations (adjustment of the curve) should be simple, and I reckon it will improve realism considerably.
Grüße
Uwe

Rewriting the ffb is not technically difficult, the main problem is to figure out a better mathematical model. The current model is probably simplified so that nonlinear calculations can be replaced by a table lookup.

[Uwe Herrling]

brr, on Jul 31 2009, 03:14 PM, said:

Rewriting the ffb is not technically difficult, the main problem is to figure out a better mathematical model. The current model is probably simplified so that nonlinear calculations can be replaced by a table lookup.

I think one or two corrections (that is: addition) on top of the current calculation of steering wheel forces resulting from car movement would do. One correction for wheel rotation, one for the wheel friction (standing still, no rotation at all). Both effects add resistance to the steering wheel, and I think its a safe bet to assume that they are equal on both front wheels.

Correction for rotation: depending on wheel (mass, diameter) and speed (angular rotation of the wheels). Increasing with all three. I think it is even linear, so I would start out with a product of the three (with correction factors so that the "real" forces coming from the model dont get covered up, at least not at moderate speeds where most of the cornering happens)

Correction for friction: I would assume it as steeply and linearly decreasing with speed increasing. All in all not that important.

Only suggestions, of course, for some modder like you who has the knowledge where to get and set those values.

Grüße
Uwe

PS: Maybe the above is what "damping" is (was) all about?

Edited by Uwe Herrling, Aug 02 2009 - 10:21 AM.

[brr]

Uwe Herrling, on Aug 2 2009, 07:08 PM, said:

I think one or two corrections (that is: addition) on top of the current calculation of steering wheel forces resulting from car movement would do. One correction for wheel rotation, one for the wheel friction (standing still, no rotation at all). Both effects add resistance to the steering wheel, and I think its a safe bet to assume that they are equal on both front wheels.

Correction for rotation: depending on wheel (mass, diameter) and speed (angular rotation of the wheels). Increasing with all three. I think it is even linear, so I would start out with a product of the three (with correction factors so that the "real" forces coming from the model dont get covered up, at least not at moderate speeds where most of the cornering happens)

Correction for friction: I would assume it as steeply and linearly decreasing with speed increasing. All in all not that important.

Only suggestions, of course, for some modder like you who has the knowledge where to get and set those values.

Thanks for the suggestions. Currently ffb2 patch is done by sneaking a few bytes of code inside gpl.exe so nothing complicated can be done. But I'm writing a sort of "collected physics and other mods" dll, which allows me to write physics mods with C++. When I get around to adding the ffb2 patch to it, I'll try your suggestions then.

Quote

PS: Maybe the above is what "damping" is (was) all about?

The damping (one configured in the core.ini) is a simple proportional controller, i.e. it subtracts a portion of the previous value of the force from the current one.

Ffb2 patch removes two other types of damping: averaging of forces over one frame (originally this gives a half-frame delayed, smoothed force), and a non-zero damping value communicated to the ffb wheel driver (this damping cannot depend on GPL physics in any way). Inertia is also removed (this is also a non-zero value sent to the ffb wheel driver).