I am already able to login to forum, post and manage my acount using mobile phone. The same way I can also log in to lfsworld and see some additional stats. Did we already become so lazy that we need to scan QR codes instead of typing username and pass?
What you propose, seems to me only like a waste of dev time for pretty much no improvement whatsoever. I'd much rather that devs spend their time on improving actual sim, which is exactly what they're doing right now, give or take some unplanned excursions due to ddos attacks.
The steering axis deadzone adjustment is tricky. You should be very careful with it, as it may have an opposite effect than what you're trying to compensate.
Some older versions of lfs had "remove deadzone" setting, but I never saw that work properly, or actualy do anything as a matrer of fact. In recent versions, it's completely removed for a good reason. However, not all is lost, you can use a 3rd part software like DXtweak2 to introduce some deadzone. Note that you can't remove them if they are implemented in the firmware of gamepad.
If I understood correctly, your problem is that thumbstick doesn't go back to the center and you would like to create a deadzone around the axis center, such that a small range of values around center will be set to axis value 0. This can be done with DXtweak2. I don't recommend you to do this, it is much more effective to correct it back with your finger to 0, as in very rare occasions while driving you really need wheel at 0 pos for a long time, like on some long straight, but not even there you wouldn't just let go of steering.
A much better option is to use DXtweak2 to recalibrate the center position of the thumb stick axis. The only side effect is that you will lose some sensitivity on one side. because you'll be capping off small part of axis range at end towards direction you shift the center to. What lfs had before, was a recenter option for steering axis, which did exactly what I described above. That indeed would be nice to have back.
Awesome, tnx so much for making it. I was eagerly waiting from Czecho-Slovak guys to start making newer škoda's, but it seems that you beat them to it. Hats off to you and happy modding.
You can thank inSim for that, as without it this kind of stuff wouldn't be possible.
I went through the thesis, it is well writen, nice work man. I'm still amazed that you used such a tiny neural network with only 6 inputs and 1 hidden layer and still managed to get good driving results.
I'm interested to see how would Gabor do on such type of tracks and if AI would benefit if trained by his data set.
Wow, the end result is very impressive for an AI that only got trained with your driving. You must give reinforcement learning a shot, although it will take an enourmous amount of time, because you have to do runs in real time allways. Awesome project for uni, I'm sure your teachers are proud.
I would be interested to know more details about your neural network. How many neurons do you have, how many (hiden) layers, how many inputs and outputs. Which kind of activation functions you used and so on.
You did twice what exactly? You have to press unlock button in lfs to unlock it. Forum reports correctly that you have an S2 licence, so there's nothing to worry about. We all get 2 unlocks per week, so you can give it a try after next reset, which should be every Friday if I'm not mistaken.
Yeah, I don't know if you have ever driven any of those karts in real life, but they are very twichy, the steering max lock is quite low, about 90deg or so, rear end is very lose, there is no differential so one has to be very careful with brakes.
Devs are aware of this and they recognize the power of single player. There were many improvements to AI recently, however further developement has delayed until new tire physics will be completed.
AI's were improved to the point where it makes no sense to do anything new with them, because updated tire physics will require to do that again = waste of devs time. Therefore, we'll have to wait a bit longer for some competitive AI, it's definately in the works, just not before higher priority updates that we're waiting for a couple of years. Devs are curently working very hard on those.
Not everyone wants to share their work with the rest of the world and that's fine. One can create something and share it with his close friends, without anyone being able to judge or complain (for example).
Wow, I'm amased that chat gpt came up with this text. I mean, I can see that it took and rearranged some of our posts, but man, we're living in the future already. It's scary how good of a useful tool it became. I wonder what will happen once it has no sources availabe that are written by humans.
I agree that it would add some immersion. However, I'm not sure if it's possible to set steering wheel rotation angle in the editor independently of the actual wheel lock angle. In that case, implementation would require quite a lot of re-work. Someone with more exp in editor pls comment.
Additionally, I see a potential problem with driver hands animation. Extending the range would also need some re-work in the available points for hand positions. Truckers usually have a small handle to help them to do many rotations using single hand, so that would also have to be taken care of somehow I guess.
All in all, a lot of work for devs and modders If they're up for it, I'll gladly accept it, but chances of seeing this improvement are slim at the moment.
Yes and they are using them. There are 3 LOD's, LOD1 is full car model, while LOD2 and LOD3 are reduced versions with progressively lower poly count, that you will see in distance, if multiplayer LOD reduction option is enabled in options/misc. As mentioned before, LOD3 the boxy one is used as physics for collisions.
Hi, yes, LFS supports only the DirectInput (DI) protocol for FFB. Xbox controllers work with XInput (XI) so you can't enable FFB in LFS for such controllers out of the box. Logitech gamepads like F510 and F710 are natively supported once switched in DI mode.
There are some 3rd party emulators that can do such conversions, what you need is to emulate an XI device into a DI one, not vice versa. There is one such program, although somewhat risky but you may give it a try. It's very nicely documented, so I suggest to do some reading first. Downside is that creator is Japanese and the program is in Japanese, but you can use google translate for the webpage. https://sites.google.com/site/0dd14lab/xinput-plus
At the moment the moto physics is not fully simulated in LFS and this help you are talking about is a workaround. It works like this3: user is giving the lean angle input instead of steering input and LFS does the rest, trying what ever it takes to achieve that lean angle. Once gyro forces on the front wheel are implemented, the handling will be more realistic, until then it's like this.