GP World analog sensitivity

[phasermaniac83]

GPWorld -preset 1 for me works great with the joystick.
Now I am trying to play with a steering wheel. I assume I should play without the -preset 1. It only react when i'm at +- 80% of the axis and it turns full to that side, not gradually

[DirtBagXon]

Sounds like your wheel has a long turn until it exceeds the "dead zone".

In Hypseus the Dead zone is hardcoded at a fairly usual:

const int JOY_AXIS_MID  = (int)(32768 * (0.75));  // how far they have to move the joystick before it 'grabs'

Maybe look at adjusting

Otherwise, as I don't own a Steering Wheel Controller, it's a WiP until:

• Someone buys, or donates, a Steering Wheel Controller to the project.
• An owner submits a Github PR with improvements (I can't test).
• You submit a PR to fix it, which again I cannot test.

Fingers crossed, adjusting the Dead Zone improves it.....

[phasermaniac83]

Thanks!
Just tested in another pc with regular pc controllers with analogs and with playstation2 controllers with usb converter, and have the same results. Maybe it's easier to test this way.
I went to windows 10 game devices -> configuration -> calibrate. Recalibrated to be sure it is fine. I can see the axis moving correctly without any deadzone.
Tested a bit more and gpworld with -preset 1 seems to work near perfect, but too much sensitive:
Regular controllers' analogs (as well as steering wheels) autocenters phisicallly itselves, when you are not aplying any force it should return to center ingame, like -preset 1 does.
But whe you press slowly to one side, it should turn a little, not top to that side.
const int JOY_AXIS_MID = (int)(32768 * (0.75)); // how far they have to move the joystick before it 'grabs'
can I change this myself? or is it in the sourcecode and it must be compiled?

[DirtBagXon]

Yes, you would need to recompile to alter that value, but it's far deeper than than after some thought and a look at the GPWorld control specs. The inherited control system has never needed to deal with this type of input.

Analog axis uses a 16bit integer with values from -32,768 to 32,767 : +/- depending on the direction left/right, the "dead zone" is a standard joystick process to stop "jitter" affecting moves. I assume Steering controls are no different.

So the current logic looks something like this in ASCII art:

|  -32,768           |            32,767  |
|        | -24,576   |     24,576 |       | 
|  turn- |   <-   dead zone  ->   | turn+ | 

However, GPWorld has a 4 phase turn on strength

( "SLIGHT RIGHT" )
( "MEDIUM RIGHT" )
( "STRONG RIGHT" )
( "FIERCE RIGHT" )

( "SLIGHT LEFT" )
( "MEDIUM LEFT" )
( "STRONG LEFT" )
( "FIERCE LEFT" )

This is "faked" in the limited control system inherited from Daphne. But what needs to happen for a steering wheel control is a four phased turn on left and right, so spitting -32,768 to 32,767 into 8 phases with a "dead zone" still in the middle to protect from jitter.

As a base, this is probably around 8192 value change per step in axis, but this possibly would need to be unevenly graded and (de|in)cremental steps for an authentic steering feel - minus a % of the value assigned as "dead" in the center. This is great in theory, but without hardware, untestable. Maybe someone can use this to form an idea and submit a PR.

For a reference on the control bits required for each strength step change, the MAME code provides some insight: MAME

[phasermaniac83]

Ok so its not that easy, 8192 values to change per step? I thought it was only a function. Shame.
Anyway I think if it works with a regular controller analog (stick) it should work for streering weel. so no need for extra hardware to test.
Many thanks anyway

[phasermaniac83]

No sorry, there's some misunderstanding, sorry english is not my language.
I say that steering wheel and standard controller stick (analog) have the same problem with this game.
Using -preset 1 works best in both but too sensitive