• Welcome to PiBoSo Official Forum. Please login or sign up.
 
June 27, 2019, 12:25:44 am

News:

GP Bikes beta15b available! :)


DS2 users

Started by Chris_Beeves, May 05, 2019, 12:11:21 pm

Previous topic - Next topic

Do you use Directsteer mode?

YES! YES! GOD YES! (DS2)
4 (36.4%)
No, I'm fine with default
4 (36.4%)
What is Directsteer?
0 (0%)
Tried it, hated it
3 (27.3%)

Total Members Voted: 11

poumpouny

The worst thing about DS is that it is way more difficult than riding a real bike, because of the lack of feedback and dedicated controller ........

poupou59890

That's why I am waiting to received my T300rs unit and make a quick rig to test DSA before making a full steel rig, just to be sure that will be usefull (and fun too, because that's still the main goal) Hope that the T300rs FFB feedback will be a little bit better... in order to acheive to make one turn whithout crashing. Is it me or just exiting pit is incredibly difficult in direct steer mode ?

Hawk

Quote from: HornetMaX on May 06, 2019, 11:28:34 am
Quote from: Hawk on May 06, 2019, 09:07:15 amSame with anything one does in life really, it's a bit like having played tennis right handed all your life and then being asked to only play left handed, it takes a lot of dedication and persistence to pull it off but it can be done as Klax has proved.  :)
The question would be: why would you do that ? Expect for the obvious "I've lost my right hand".
In GPB there's no benefit is playing with DST/DSA on a non-FFB device.
DST in particular makes zero sense on non-FFB devices (and DSA makes marginally more sense).

All I'm saying Max is that anything is possible if one really wants to do it; I agree that FFB would be an advantage but it can be done without it; it's amazing how the mind and senses can adapt with dedicated practice..... I'm sure that if someone had only used DS to control GPB and then tried to switch to using the common standard way of controlling GPB it would feel just as alien to them as standard control users first trying out DS to control GPB. :)

D4rw1n

May 06, 2019, 08:11:58 pm #18 Last Edit: May 06, 2019, 08:16:00 pm by D4rw1n
Please correct me if I'm wrong, as I'm still trying to understand some difference between DS1 (elderly called DST) & DS2 (elderly called DSA).

(I may repeat with different words what has been already said earlier on this thread, so my apologises in advance for the redundancy)

I'm not sure to understand all the things about DS1 & DS2 as I have never tried it with a handlebar, but will try some day with my xbox controller, *just for science*.

My current understanding is:
- DS1 gets as input a torque value from the game controller, i.e most likely a handlebar with some load cells or any alternative sensor able to measure the torque applied on the handgrips (presumably without force feedback, but see the * further).
- DS2 gets as input a steering angle from the game controller, i.e most likely a handlebar with force feedback feature.

Assuming the force feedback effect we usually talk about (like here and in any other racing game) is calculated from GP-Bikes and sent to the controller, and because the force feedback has a direct effect on the steering angle at anytime (turning resistance but also wobble effect), it would put pressure against driver hands on the handlebar and provide complete random and irrelevant values read from a load cell, therefore it would make no sense to use regular force feedback with DS1. Is that phrasing correct?

Assuming what has been stated above is correct, there is still one thing I don't fully understand and that I have read in multiple times on the forum: "DS1 can't be played with force feedback".

*
If we consider we use DS1 with a big&powerful stepper motor and optionally a rotary encoder (it's out of scope regarding the topic but still .. I would prefer a close loop though).

DS1 example would be:
You want to lean the motorcycle on the right, so you push on your right handlegrip and/or pull the left one. Call this moment A.
Automagically, your perfect calibration of load cells input is sent as a torque input to GP-Bikes and the physical engine makes its calculations and as a result turns slightly the handlebar (call this moment B) and lean the bike's chassis until you have reached expected lean angle. Call this moment C.
While this was happening, in parallel your steering handlebar has slightly moved as well in real life (moment B) and chassis has leaned as well (moment B too).
Then you release any pressure on your handlebar, and the absence of torque input lets the physical engine put the steering angle to 0 degree again (or close from that) (moment C).

In this scenario, I assume we would feel directly the force through the stepper motor, without using any regular "force feedback" input from the game, but actually a direct effect on the steering handlebar, isn't it?
I'm thinking as well such FFB mode would affect as well torque input, but that would be exactly like in real life.

I'm not sure I have been clear on the 2 different force feedback I'm talking about, as I have actually no precise clue how a regular force feedback is presented from a game to a game controller (int? float? array of float?). If I'm not wrong, I don't think such value is given through telemetry data shared from DLO memory.

EDIT: re-reading my post shows me it can be felt contradictory about the "regular" force feedback and the "stepper motor" feedback.
In other words (again): I'm wondering actually whether the regular force feedback given through DS2 mode is as accurate and as real as the one we would feel through DS1 with a stepper motor controlling the steering.
Hence my statement "therefore it would make no sense to use force feedback with DS1" ...


All that post to say that I have only a very small in-game practice and that I'm trying to see what path I should take to develop my own rig: DS1 or DS2?



Myst1cPrun3

I did try the DS method, both 1 and 2, and for me it just seemed a little weird, and un-natural, and quite frankly, I'm not willing to put the effort in to re-learn 10+ years of motorbike gaming controls when I can consistently (But slowly) get round without crashing.

Maybe if I get some handlebars on my TX base I'll look into it more, but yeah, don't even bother using it with a gamepad is my 2 cents, and that's as far as I can go with this topic lol

Sorry not much help but y'know XD

Chris_Beeves

Quote from: D4rw1n on May 06, 2019, 08:11:58 pmPlease correct me if I'm wrong, as I'm still trying to understand some difference between DS1 (elderly called DST) & DS2 (elderly called DSA).

(I may repeat with different words what has been already said earlier on this thread, so my apologises in advance for the redundancy)

I'm not sure to understand all the things about DS1 & DS2 as I have never tried it with a handlebar, but will try some day with my xbox controller, *just for science*.

My current understanding is:
- DS1 gets as input a torque value from the game controller, i.e most likely a handlebar with some load cells or any alternative sensor able to measure the torque applied on the handgrips (presumably without force feedback, but see the * further).
- DS2 gets as input a steering angle from the game controller, i.e most likely a handlebar with force feedback feature.

Assuming the force feedback effect we usually talk about (like here and in any other racing game) is calculated from GP-Bikes and sent to the controller, and because the force feedback has a direct effect on the steering angle at anytime (turning resistance but also wobble effect), it would put pressure against driver hands on the handlebar and provide complete random and irrelevant values read from a load cell, therefore it would make no sense to use regular force feedback with DS1. Is that phrasing correct?

Assuming what has been stated above is correct, there is still one thing I don't fully understand and that I have read in multiple times on the forum: "DS1 can't be played with force feedback".

*
If we consider we use DS1 with a big&powerful stepper motor and optionally a rotary encoder (it's out of scope regarding the topic but still .. I would prefer a close loop though).

DS1 example would be:
You want to lean the motorcycle on the right, so you push on your right handlegrip and/or pull the left one. Call this moment A.
Automagically, your perfect calibration of load cells input is sent as a torque input to GP-Bikes and the physical engine makes its calculations and as a result turns slightly the handlebar (call this moment B) and lean the bike's chassis until you have reached expected lean angle. Call this moment C.
While this was happening, in parallel your steering handlebar has slightly moved as well in real life (moment B) and chassis has leaned as well (moment B too).
Then you release any pressure on your handlebar, and the absence of torque input lets the physical engine put the steering angle to 0 degree again (or close from that) (moment C).

In this scenario, I assume we would feel directly the force through the stepper motor, without using any regular "force feedback" input from the game, but actually a direct effect on the steering handlebar, isn't it?
I'm thinking as well such FFB mode would affect as well torque input, but that would be exactly like in real life.

I'm not sure I have been clear on the 2 different force feedback I'm talking about, as I have actually no precise clue how a regular force feedback is presented from a game to a game controller (int? float? array of float?). If I'm not wrong, I don't think such value is given through telemetry data shared from DLO memory.

EDIT: re-reading my post shows me it can be felt contradictory about the "regular" force feedback and the "stepper motor" feedback.
In other words (again): I'm wondering actually whether the regular force feedback given through DS2 mode is as accurate and as real as the one we would feel through DS1 with a stepper motor controlling the steering.
Hence my statement "therefore it would make no sense to use force feedback with DS1" ...


All that post to say that I have only a very small in-game practice and that I'm trying to see what path I should take to develop my own rig: DS1 or DS2?




I think you pretty much nailed it with that description!

The stepper vs. DS1 I thought about as well. I discarded it as too complicated. DS2 works well enough for me to put my energy there. I'm looking forward to hearing more about this though!
I had to try..

HornetMaX

Quote from: D4rw1n on May 06, 2019, 08:11:58 pmPlease correct me if I'm wrong, as I'm still trying to understand some difference between DS1 (elderly called DST) & DS2 (elderly called DSA).

(I may repeat with different words what has been already said earlier on this thread, so my apologises in advance for the redundancy)

I'm not sure to understand all the things about DS1 & DS2 as I have never tried it with a handlebar, but will try some day with my xbox controller, *just for science*.

My current understanding is:
- DS1 gets as input a torque value from the game controller, i.e most likely a handlebar with some load cells or any alternative sensor able to measure the torque applied on the handgrips (presumably without force feedback, but see the * further).
- DS2 gets as input a steering angle from the game controller, i.e most likely a handlebar with force feedback feature.
Correct. So with a joypad stick, if you push the stick 50% right your sending to GPB:
  • 50% of the max steering torque with DS1/DST
  • 50% of the max steering angle with DS2/DSA

If you have a FFB wheel and you use DS2/DSA: you try to control the wheel angle with your hands, fighting the FFB that is sent by GPB to the wheel. If you think, this is exactly what happens on a *real* bike (and car too, if we forget about power steering).
GPB reads the wheel angle from the wheel and the virtual rider (a "scaled down" version compared to the virtual rider used without direct steer) tries to align the in-game handlebars to the angle your wheel dictates. This is not perfect but can be done reasonably well once everything is setup properly (e.g. reasonable FFB force compared to your arms').

If you have a FFB wheel and you use DS1/DST: this makes no sense at all. The position of your wheel is not at all related to the steering angle in-game. How realistic it is if the wheel handlebars you're holding in your real hands is not aligned to the in-game handlebar ?!

If you absolutely want to dictate to GPB the steering torque (DS1/DST) you'd need something complex, like a FFB steering wheel with a torque sensor on the steering axis. The device will send to GPB the torque, not the angle (the angle would stay unused), while GPB will send to the device the FFB signal, as usual.
What's complicate with that is the torque sensor sees not only the torque your hands apply to the wheel but also the motor torque (the FFB torque).
Theoretically you could take the difference between the torque sensor reading and the FFB signal to get the steering torque, but I'd expect this to be more a bit more triky in practice.

And even if you sort this out you have a final problem: the in-game steering angle is not tied to your wheel angle (as we said, the angle is not sent to GPB). So even if you manage to do all the savvy computation to align everything (assuming this is even possible) in terms of open-loop, without a feedback loop the thing will drift and your wheel angle will be different from the in-game steering angle. Imagine how fun that would be ...

I hear the question coming: so, why not making a feeback loop on top to somehow keep the two angles aligned ?
Answer: because in the end, that would be equivalent to use DS2/DSA, just much more complex, more expensive to build, way more tricky to fine tune and without even a gram of proof (or hint of proof) it will work any better (if at all).

Some will argue that with DS2/DSA we still have a virtual rider in the loop, so it's not really "as direct as it could be".
That's right but that would be the case in the above setup with DS1/DST too (in order to have the wheel aligned to the in-game bars), so the point is mooth, sorry.

Final words (aka TL/DR): car sims with FFB (including pro-grade ones) all work with DS2/DSA and do not even have the option to use DS1/DST. That should tell something ...
DS1/DST is just a curioisity and for the sake of not confusing future (MCGyver-style) users, I'd argue PiBoSo should take it away. You want FFB: do as Marcel, get a DD steering wheel and use DS2/DSA. From what I've seen of his rig, it works extremely well. And that's exactly what DS2/DSA has been made for.

poumpouny

What would perfectly mimic the real world with DST 1 (torque) will be :

- A handle bar with torque sensor (load cell) which you push when you want to lean, and which will sent torque input to gpbike
- Then Gpbikes calculate the appropriate steering angle resulting from that torque, and then re send it to a FFB motor in which is attached the handlebar. (the motor need to be powerfull enough to not receive "false angle" when you push on it ) this loop need to have the smallest latence possible. So you think your turning your handle bar when actually you just input torque.

- Then Gpbike add the virtual rider lean torque calculated for exemple from a Track ir attached to the rider and then lean the bike accordignly.

I don't know if GPbikes can separate the torque input and the steering angle output but even assuming it is the case, the real problem with Direct steering is (DST and DSA) : In real life you have the centrifugal force that tell you your leaning + off course your knee and what your eyes see. In Gpbikes DS mode you only know you reach your maximum lean angle when your tyres start losing grip, which is of course too late......
 

Chris_Beeves

May 07, 2019, 08:17:11 am #23 Last Edit: May 07, 2019, 08:24:34 am by Chris_Beeves
Quote from: poupou59890 on May 06, 2019, 02:56:07 pmIs it me or just exiting pit is incredibly difficult in direct steer mode ?

It is the hardest part! Especially on Victoria. I usually just pin it so I crash into the barrier. Holding reset from there usually spawns you on the main straight.
It's not pretty, but it works ;)
I had to try..

poupou59890

I really like this thread :)

I am totaly agree with HornetMax, I think FFb wheel must be only use with DSA/DS2 mode, as Marcel shown it on his video (and I can't wait to see his final rig), I am making same thing with T300rs, hope it will be enough precision and FFB on this base to have good feeling. I am using VR so Assuming that Virtual rider will be ok right ? Because the virtual rider will be dictate by my headset position even on Ds2 mode, am I right ?

For Chris_Beeves, Yes I do that too with Mugello, I don't know why, but when I start, the virtual rider is not center, a little bit on the right, so the bke tend to fall on the right (even with automatic virtual rider), So I need to balance on the left but at slow speed the bike react not very well (cbr 600 rr for exemple) then shake...and fall..again..and again...So yest I hit the pit wall and pray that I reset on straight line. Do the job for now.



D4rw1n

Quote from: HornetMaX on May 07, 2019, 07:32:59 amIf you have a FFB wheel and you use DS1/DST: this makes no sense at all. The position of your wheel is not at all related to the steering angle in-game. How realistic it is if the wheel handlebars you're holding in your real hands is not aligned to the in-game handlebar ?!
Exactly, hence my statement: it would make no sense to use regular force feedback (the one you would enable with DS2) on a DS1 mode.

Quote from: HornetMaX on May 07, 2019, 07:32:59 amIf you absolutely want to dictate to GPB the steering torque (DS1/DST) you'd need something complex, like a FFB steering wheel with a torque sensor on the steering axis. The device will send to GPB the torque, not the angle (the angle would stay unused), while GPB will send to the device the FFB signal, as usual.
What's complicate with that is the torque sensor sees not only the torque your hands apply to the wheel but also the motor torque (the FFB torque).
Theoretically you could take the difference between the torque sensor reading and the FFB signal to get the steering torque, but I'd expect this to be more a bit more triky in practice.


FFB on DS1 would be permanently a direct effect of physical engine telling exactly what should be the actual position of the steering. I'd dare to call this "Stepper Motor FFB" in this case.
It would actually not be a resistance effect given like the DS2 FFB, but just position as it should be.
Of course one could think the player would feel a huge/stall resistance through handlebar, as he is not controlling in any way the steering angle from that axis, but only with torque applied and read through load cells on handle grips that indirectly but very very quickly are computed by the GP Bikes physics engine.

So here for DS1 I'm not even thinking of getting anything more than a stepper motor + rotary encoder. There wouldn't be any torque sensor reading directly steering axis. you would actually only get a non-null torque value from the handlebar if the player fight against the steering angle given from the game.

Quote from: HornetMaX on May 07, 2019, 07:32:59 amAnd even if you sort this out you have a final problem: the in-game steering angle is not tied to your wheel angle (as we said, the angle is not sent to GPB). So even if you manage to do all the savvy computation to align everything (assuming this is even possible) in terms of open-loop, without a feedback loop the thing will drift and your wheel angle will be different from the in-game steering angle. Imagine how fun that would be ...
I'm not sure to understand your statement here.
Do you mean m_fSteer isn't reflecting the steering angle of the front wheel against the chassis?

In my case the rotary encoder on the stepper axis would just be a guarantee to be sure the stepper position is 100% correct, no matter a jump has occurred. This wouldn't be used to return the read angle of the steering to the game but would stay local to the stepper driver controller.


In TL'DR: my concern is to know whether DS2 + regular calibrated FFB would result in the same real effect than a DS1 + resulting "direct FFB from stepper motor".

Another way of asking the question: How is regular FFB values/behavior shaped usually in games? I know there are a lot of calibration settings on multiple sides: game &/or controller driver. But eventually what does that look like from the game to the driver? Is it just a scale? or are they different "profiles" of vibrations/effects?


But from an implementation point of view, I totally agree that DS1 mode looks like bringing much more challenge and less guarantee than going with DS2 & regular FFB, though.

poumpouny

For those who have trouble using DSA to get out of pit, please note that in DS mode, the virtual rider auto lean doesn't do anything, you are forced to use manual rider lean. I victoria, i take the bike out of garage by using the "push bike" command that i assigned with the wheel up and down button then when i'm aligned to the pit line i lean forward and centered and it is easier to Start. I'will make a thread with a video of me riding in DSA mode and explain to you that with a relative simple setup (in my case a G920 wheel + a ps3 cam that track my face with facetracknoir) you can enjoy Direct Steering ......

Chris_Beeves

I use Opentrack and a Ps3 cam to control the rider. It makes me smile every time the rider follows me when I tuck in behind the screen on the straights. :D
I had to try..

HornetMaX

Quote from: poumpouny on May 07, 2019, 08:06:42 amWhat would perfectly mimic the real world with DST 1 (torque) will be :

- A handle bar with torque sensor (load cell) which you push when you want to lean, and which will sent torque input to gpbike
- Then Gpbikes calculate the appropriate steering angle resulting from that torque, and then re send it to a FFB motor in which is attached the handlebar. (the motor need to be powerfull enough to not receive "false angle" when you push on it ) this loop need to have the smallest latence possible. So you think your turning your handle bar when actually you just input torque.
Your load cell will measure at the same time the input torque and the FFB torque, separating the two may not be as trivial as it seems.
Plus, for GPB to send the angle to your bar, the bar must have a feedback loop to track the angle provided by GPB.

So for your setup you'd have to have a motor with a torque sensor and a feedback loop in your hardware (unless PiBoSo wants to implement it in GPB, which I doubt) plus the logic to make the actual measure of the input tuorque, without the FFB torque.
The alternative is a plain FFB wheel with DSA.
Unless you're very convinced option one has any advantage, I'd stick to option 2.
Once again: car-sim people have been playing with stuff for quite a long time and, to my knowledge, DSA is *the* unquestionned answer.

Quote from: poumpouny on May 07, 2019, 08:06:42 amI don't know if GPbikes can separate the torque input and the steering angle output
Of course it can. The steering torque is always an input of the physical bike dynamic model (no matter if DST, DSA or default steering mode) while the steering angle is a state variable of that model.
The only difference is that in DST the steering torque is an overall input (i.e. the payer provides it to GPB), while in DSA and plain steering the steering torque is computed by the virtual rider (from the steering angle input with DSA, from the lean angle input with default steerig mode).

poumpouny

May 07, 2019, 01:36:11 pm #29 Last Edit: May 07, 2019, 01:42:23 pm by poumpouny
I dont undrestand what you mean by : "the bar must have a feedback loop to track the angle provided by GPB". Of course it need that feedback, like all normal wheel. the ffb motor will be a normal FFB motor like any wheel, it just need to not be "turned by the user by inputing angle, but torque, even if of course the handle bar will turn when you input torque".

Why whould be the FFB torque be a problem cause when it is static then the torque calculated is just the torque you apply on the loadcell witch is a separate input on the handle bar (nothing to do with the ffb motor or sensor). if the ffb return angle that may be caused by curbs or whooble then you just need to apply more torque, just like in real life .......


Edit : Car sim is not comparable, cause you turn your wheel way larger in angle than a bike handle bar. That's why apply in torque is more intuitive and natural then turning and applying angle .......

Edit 2 : It just like when you use head tracker to controll rider lean wich influance the bike lean ........just replace the ed tracker by the loadcell and the bike lean by steering angle .....