Separate X and Y mouse sensitivity

[mjr4077au]

Where was the sensitivity of 2:2 done for Raze? I'm dashing off to work but from what I recall, in GetMouseDelta (and excluding Raze's specific divisors for angle (16.f / 32.f) and horizon (16.f / 64.f)), given an axis input of 1 with the / 3.f divisor on the x axis would have yielded a (4/3):1 ratio.

[Graf Zahl]

Nowhere. That's the point. It took the prescaled values from the backend and passed them on - so the factors were baked right into the conversion math in InputState::GetMouseDelta, so I had to change those factors to account for a different default.

[mjr4077au]

Ah right OK, I think I misread your previous post sorry

[Graf Zahl]

Just pushed it. I'll leave the final tweaking of the factors to you,

[mjr4077au]

Thanks, I'll have a good play when I get home. Just built on my laptop (only have touchpad) but the horizon is pretty fast (comparatively) at the moment. I'm fairly used to the slow pitch given the old 4:1 pre-scaling code though.

[mjr4077au]

Curious, but you mentioned your mouse sensitivity was basically 1.7:1.7, is that how you played Raze before the uplift as well?

I know this is completely subjective, but with the x axis being matched to the y axis it feels terrible. Surely you'd want to turn quicker than you'd want to be able to look up/down? Again I only have the touchpad on me but I'm finding I'm boosting the x axis sensitivity up from 2.0 to 6.0 just to make it feel what I think is appropriate.

[Graf Zahl]

Actually, I played that last build of Raze with 2:1.4. Anything more on the x-axis and turning is so fast I can't control the game anymore.

The results here should be the same.

The constant factor in the old code was 1/6, all other scalers were 1 for me and the backend prescaled by 4, resulting in 4/6 = 2/3.
The new code omits the prescale, defaults to a sensitivity scale of 2 and multiplies by 1/3 so the end result would also be 2/3.

So why is it different? Or are you talking about Linux?

[mjr4077au]

This is on Windows and again it's just with the touch pad. I'm just going to leave it for now because I'm not on my proper gear but the horizon (I'll call it pitch since we convert it to pitch anyway) just felt a lot quicker than the angle.

[Graf Zahl]

Yeah, it was too fast for me, too, that's why I ended up using 1.4 for it.

[Graf Zahl]

'So, how does it look? Everything ok with Raze?

[mjr4077au]

I'll be home in 1.5 hours, I live 80km from the office and public transport in Australia sucks lol

[Graf Zahl]

I feel your pain. I have a 47 km way to work but there's no public transport options between my home town and my work place. With the car it takes 45-50 minutes, if I had to rely on public transport, I'd need 2 hours for that distance.

[Marisa the Magician]

Oh, this is much nicer after the latest commits.

[mjr4077au]

I feel your pain. I have a 47 km way to work but there's no public transport options between my home town and my work place. With the car it takes 45-50 minutes, if I had to rely on public transport, I'd need 2 hours for that distance.

I'd normally drive but I have a lead foot and unfortunately have a suspension until January. Just have to do what I have to do. Door to door is about 2 hours though, I'm already half way there which is why it's only another hour.

Preliminary opinion is I'll find the angle too slow out of the box compared to the pitch but will make a conscientious effort to not let personal bias overrule anything. It's been 4:1 for a long time though. I'm also in favour of replacing mouse_sensitivityx/y with an overall sensitivity slider that used to be there since m_yaw/m_pitch effectively do the same thing, just later in the piece.

Oh, this is much nicer after the latest commits.

Can't wait to have a proper test

[Graf Zahl]

since m_yaw/m_pitch effectively do the same thing, just later in the piece.

Not quite. The difference is that one scales the raw event which will can be intercepted by scripts in GZDoom while the other one scales how this value gets applied to that control. So considering that apparently the values being passed through here can differ quite a bit a sensitivity scale on the event level is needed.