• Welcome to PiBoSo Official Forum. Please login or sign up.
 

GP Bikes beta20 WIP

Started by PiBoSo, April 06, 2021, 11:41:52 AM

Previous topic - Next topic

h106frp

Quote from: PiBoSo on April 06, 2021, 03:29:58 PMThe "lean help" is more of an input filter than a real helper. It is linked to the "Direct Lean" setting.

At the moment, at full lean input, the virtual rider tries to keep the bike at the max possible lean angle, at the edge of grip. When the track suddenly changes and less lean is possible ( or maybe there is just a bump ), there might not be enough grip to pick the bike up before it low-sides.
The stability helper is meant to save the bike from that kind of crashes, not to be a "catch-all".

Will the new calculation keep the virtual rider consistent over the full range of direct lean values? Historically I personally found that the problem with trying to tune the VR was that once tuned for a specific direct lean value, altering the direct lean degraded the rider tuning. Values above 50% DL appear especially troublesome.

Vini

Quote from: h106frp on April 21, 2021, 07:31:16 PMHistorically I personally found that the problem with trying to tune the VR was that once tuned for a specific direct lean value, altering the direct lean degraded the rider tuning. Values above 50% DL appear especially troublesome.
What happens when you tune it for 100% DL from the getgo?

h106frp

Tuning at 100% makes lower settings very sluggish due to (currently) having the problem of fighting the inside bar dropping on turns.

Vini

But if you tune the virtual rider in a way that lower DL values act like higher ones, then what is the point in low DL values anyway?
Being able to react quickly is precisely why I prefer 100% for example. Is essential on the brake especially.

PiBoSo

Quote from: h106frp on April 21, 2021, 07:31:16 PM
Quote from: PiBoSo on April 06, 2021, 03:29:58 PMThe "lean help" is more of an input filter than a real helper. It is linked to the "Direct Lean" setting.

At the moment, at full lean input, the virtual rider tries to keep the bike at the max possible lean angle, at the edge of grip. When the track suddenly changes and less lean is possible ( or maybe there is just a bump ), there might not be enough grip to pick the bike up before it low-sides.
The stability helper is meant to save the bike from that kind of crashes, not to be a "catch-all".

Will the new calculation keep the virtual rider consistent over the full range of direct lean values? Historically I personally found that the problem with trying to tune the VR was that once tuned for a specific direct lean value, altering the direct lean degraded the rider tuning. Values above 50% DL appear especially troublesome.

Different values of direct lean haven't yet been tested with the increased simulation accuracy.
Hopefully the virtual rider will be more consistent, though.
"Obviously your ambition outweighs your talent".

h106frp

@vini
But thats what makes trying to create a decent VR config that everyone is happy with so difficult.

I was hoping PB would modify the PID loop so that altering direct lean keeps the rider PID loop gain constant and avoids erratic behaviour in-game.

Vini

If in the end low DL and high DL settings feel the same (are equally fast), then there is no point in the setting. Maybe I misunderstand...

h106frp

April 22, 2021, 10:17:51 AM #22 Last Edit: April 22, 2021, 10:21:11 AM by h106frp
It should all work, but it appears at the moment that the VR model fights the DL input and produces erratic results. If the VR is expecting a steers input and working towards a point of maximum lean it would not be surprising that if a big lean is commanded that the VR loop might see a big oscillation.