Hello there, I have noticed an issue when standing on the body parts of a skeletal mesh in ragdoll (like a dead enemy on the ground).
Even though the mover walking mode still recognises that we are standing on ground (CurrentFloor correctly finding an ImpactPoint), the DynamicBase calculation causes the capsule to move up/down by approximately 30-50cm while standing still.
I have managed to fix this behaviour by excluding bones that are simulating physics from DynamicBases, like so:
[Image Removed]
Do you think this is a legitimate fix, or should I handle standing on ragdoll actors differently?
This type of change is kind of game-specific and probably not one we would take into the engine directly. For your game, if the behaviour is that you want the capsule to be able to move onto/step up onto your ragdoll characters, then this change can be good. I would recommend subclassing the walking mode and overriding this function, though, so you don’t have to deal with engine contention during upgrades in case we make changes to the walking mode later.
Fair point, it looked like a bug to me, because it feels like the capsule-bouncing is exaggerating the tiny bone movements we are standing on.
Either way, I will look into a fix on our side. We are already overriding USmoothWalkingMode with our own implementation, but we might need to copy the entire inheritance hierarchy to reach UWalkingMode.