How does the camera works ?

[tchemi]

Hi riders, I was wondering if there was a way to improve the 1st P camera.
I know, this subject has already been discussed, But Today, I am ready to get my hands out of my pocket and really work on it. I'm not only complaining, I want to improve it.

First of all, here is for me, the main problem.

In 1st P, the camera is binded to te bike tank with a gyro stabilizer for the leaning axis, with a tilt factor (iirc. something like, cam_lean_axis = 1 - bike_lean * tilt_param;  where tilt is lower than 1)
But, there is no other stabilization for any other axis. This is the for me a real problem because it gave me sickness, and I don't talk about VR, that I have purely stopped since it makes me sick in GPB.
If there is any vibration, bump, spring movement, wobbling or whatever you will see it on the screen. The camera doesn't focus on the horizon, what your eyes/brain would do in reality.
For instance :
-wheeling : it goes to the sky
-hard breaking : you will see the ground and not the track ahead of you
-wobbling : the cam goes from left to right very quickly

Lets make it simple, Your brain can filter barely all parasite movements (and vibration, and shake and whatever) and your eyes will lock the horizon. You will see the bike move in your peripheral vision.

This has been done in assetto corsa with the "real head motion" mod. It emulates a trackir to move the camera according to the brain filter (sort of compensation). It works like a charm but I think there are a lot of easier way to achieve this.
As there is already a filter on the lean axis, why would it not have other filter for the other axis ?

What I was thinking of :
In 3p, the camera doesn't shake. It keeps the horizon. why ? how ? What are the rules ? Is there a target point on the horizon ? For me it's like the camera is focusing on a point in the horizon determined bay the bike actual movement.
This point always has the same distance with the ground, the camera never lean. The is only 3 DOF for this cam, x and y translation and z rotation.
This is exactly what I would like for the 1P. A target point for 1P based on bike actual direction.

Do we have any control of the cameras in config files ? How could we explore it ? How could we make a mod for the cameras or change their behaviour ?
Is the logic and code behind the cameras a real secret ?

It may be not look as a real priority. But as GPB is intended to be a realistic sim, I think that the cameras (The main tinterface between us and the sim) should be appealing and realistic too.

[HornetMaX]

Separated by less than 24hrs we have h106frp asking for more visual feedback (camera shaking when the rear is hopping, here) and you asking for less visual feedback. PiBoSo has a tough life

I'm tempted to side with you: there's something in the 1st person view that is odd.
Each time I've tried GPB's 1st person view I've always ended up thinking there was something "wrong" or, at least, something that confuses me after a few minutes.

Surprisingly, I don't seem to have the same problem with MXB. Can't say why ...

Also, I'd need to test a bit more but it seems to me that when I use my EdTracker, some of the weirdness of 1st person view goes away.

P.S.
EdTracker wireless seems to be almost ready ... tempted ...

[h106frp]

Great news on EDTracker - love mine and it has proved to be very reliable in the wired version

I do think one of the issues people have in 1P without tracking is the fixed view as its unnatural not to be able to look where you want to go.

[Warlock]

My EDtracker is on the way (wired) 
I always ride in 1st P  , we'll see what difference i can found. Now im pretty comfortable with lean heading, feels very nice.

But this is different to what Tchemi is suggesting, and i agree with him, i've discussed a few times about this. Rider head shouldn't shake like the bike with bumps for example.
The point of wiew should remain almost the same (dampered), inertia, and you should see the bike taking the bump under your point of view.

[tchemi]

Rider head shouldn't shake like the bike with bumps for example.
The point of wiew should remain almost the same (dampered), inertia, and you should see the bike taking the bump under your point of view.

Yes ! That's exactly what I wanted to say/illustrate.

This example is quite extreme but I think it illustrate well the difference between hard mounted and stabilized cam

https://youtu.be/VjK-asaqDMU?t=547

[HornetMaX]

But this is different to what Tchemi is suggesting

Yeah I got that, it's just that using EdTracker makes 1st person view somehow more bearable for me. Could be as h is saying (more natural if you can look around, even for tiny movements).

What i really can't explain is why in MXB I don't have the same bad feeling in 1st person view. I was expecting it to be even less bearable (lots of jumps, whoops, bumps) ...

The point of wiew should remain almost the same (dampered), inertia, and you should see the bike taking the bump under your point of view.

Hmm ... but the distance between your head and the tank does not change, I'm not sure you "should see the bike taking the bump under your point of view".
If sight is compensating the pitch movement of the bike, maybe you should just see that (e.g. bike climbing on a ramp, you eye-sight pitches down to compensate that): is that what you mean ?

[tchemi]

Hmm ... but the distance between your head and the tank does not change, I'm not sure you "should see the bike taking the bump under your point of view".
If sight is compensating the pitch movement of the bike, maybe you should just see that (e.g. bike climbing on a ramp, you eye-sight pitches down to compensate that): is that what you mean ?

In fact, the distance between the tank and your head will change. Your torso will get G force from up or down. I allready experienced that, your chest get crashed on the tank when you land after a jump, even a little one. And you are pointing it right with the ramp example. Even if your torso would keep a constant distance from the tank, if you have like an axis between your hips, you can see the front of the bike going up or down. Think about a wheelie, you look at the sky but in front of you and the dashboard come in your vision. Same for a stoppie, you still look ahead and the bike disappear beneath you

[tchemi]

Double post !!

Yes, as you can see in this example, the bike is shaking like hell. The vision is shaking too, but far less. When it land after the first jump, you can see the torso going down, like if you were going to eat the bar (don't laugh, I know someone that plant his tooth in a steering bar).

https://www.youtube.com/watch?v=8236-ET7Wjw

I think that the 1st p cam in mxb is way more stabilized, but when you look closely, not that much in fact...

[HornetMaX]

In fact, the distance between the tank and your head will change. Your torso will get G force from up or down. I allready experienced that, your chest get crashed on the tank when you land after a jump, even a little one.

Not that we have too many jumps in GPB on tracks like Victoria though ...

And you are pointing it right with the ramp example. Even if your torso would keep a constant distance from the tank, if you have like an axis between your hips, you can see the front of the bike going up or down. Think about a wheelie, you look at the sky but in front of you and the dashboard come in your vision. Same for a stoppie, you still look ahead and the bike disappear beneath you

OK, so it's "just" a matter of compensating pitch/yaw movements with sight.

I think that the 1st p cam in mxb is way more stabilized, but when you look closely, not that much in fact...

On the MXB forum PiBoSo has recently explained that the rider in MXB is modeled with a bunch of "spring/dampers" as links (arms, legs) between the torso and the bike.
Could it be that the camera position in MXB is linked to the torso (or head) and hence benefits from some filtering ?

But then I'm not really surprised as an MXB rider standing on the pegs is way more "flexible" than a motogp rider sitting and with his torso on the tank ...

To me it's more down to what you said in your 1st post (image stabilisation done by the brain).

[Warlock]

In fact, the distance between the tank and your head will change. Your torso will get G force from up or down.

Exactly.

[tchemi]

So, I think that a good starting point could be to import the MXB 1P cam into GPB for testing purpose.
No ? What do you think ? MAx, You are the boss of pluggins, Do you think it is manageable ?

[HornetMaX]

I think we need to hear PiBoSo's opinion on all that.

To be honest I'm not very supportive of a solution like the mod for Assetto Corsa (which uses a plugin that reads telemetry data, does some filtering and then controls a virtual input device that dictates the head movements).

[h106frp]

Without bike movements 1P could become disconnected like the ones in the Milestone games where you just feel like your watching someone else ride a bike 

[HornetMaX]

I don't think anybody wants to totally remove bike movements in 1p view.

[Warlock]

Not at all, bike movements must remain there but they shouldn't be 100% transfered to the rider head.
In that video of mxb , you can see the bike moving without affecting the view excessively . I think it still have the inmersion required.