You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Using the same code in 3.4-beta3 presents different vectors! These are used for directing my AI characters so it is funny to see them get confused and drive off in apparently random ways, but obviously not ideal.
Here are a small collection of vectors from this function running in 3.3.2:
Here are the same ones from the same AI on the same level in 3.4-beta:
Chosen Direction: (-0.999867, -0.013116, 0.009671)
Chosen Direction: (-0.999867, -0.013115, 0.00969)
Chosen Direction: (-0.999866, -0.013215, 0.009691)
Chosen Direction: (-0.999862, -0.013517, 0.009655)
Chosen Direction: (-0.999854, -0.014096, 0.009661)
Chosen Direction: (-0.999841, -0.014996, 0.00967)
Chosen Direction: (-0.999821, -0.016237, 0.009685)
Chosen Direction: (-0.999797, -0.017688, 0.009686)
Chosen Direction: (-0.999768, -0.019232, 0.009716)
Chosen Direction: (-0.999737, -0.020774, 0.009761)
Obviously in order for my AI characters to perform as expected the vectors returned must be predictable. Attached is a screenshot of the path follow panel.
Steps to reproduce
Create a path on a 3d plane and place an object in the scene that passes its global position with:
getPathDirection(global_transform.origin)
Run this in 3.3.2-stable and compare the returned vector to 3.4-beta3 to see a different vector.
Minimal reproduction project
No response
The text was updated successfully, but these errors were encountered:
Godot version
3.4-beta3
System information
Windows 10
Issue description
The following code in 3.3.2 Stable works fine and returns sensible vectors:
func getPathDirection(position):
var _offset = _trackPath.curve.get_closest_offset(position)
_tackPathFollow.offset = _offset
return _tackPathFollow.transform.basis.z
Using the same code in 3.4-beta3 presents different vectors! These are used for directing my AI characters so it is funny to see them get confused and drive off in apparently random ways, but obviously not ideal.
Here are a small collection of vectors from this function running in 3.3.2:
Chosen Direction: (0.977867, 0.010528, -0.208961)
Chosen Direction: (0.97787, 0.010528, -0.208949)
Chosen Direction: (0.977907, 0.010628, -0.20877)
Chosen Direction: (0.978013, 0.010929, -0.208258)
Chosen Direction: (0.978215, 0.011506, -0.207275)
Chosen Direction: (0.978537, 0.012402, -0.205698)
Chosen Direction: (0.978995, 0.013633, -0.203426)
Chosen Direction: (0.979563, 0.015069, -0.200574)
Chosen Direction: (0.980218, 0.016591, -0.197225)
Chosen Direction: (0.980944, 0.018104, -0.193447)
Here are the same ones from the same AI on the same level in 3.4-beta:
Chosen Direction: (-0.999867, -0.013116, 0.009671)
Chosen Direction: (-0.999867, -0.013115, 0.00969)
Chosen Direction: (-0.999866, -0.013215, 0.009691)
Chosen Direction: (-0.999862, -0.013517, 0.009655)
Chosen Direction: (-0.999854, -0.014096, 0.009661)
Chosen Direction: (-0.999841, -0.014996, 0.00967)
Chosen Direction: (-0.999821, -0.016237, 0.009685)
Chosen Direction: (-0.999797, -0.017688, 0.009686)
Chosen Direction: (-0.999768, -0.019232, 0.009716)
Chosen Direction: (-0.999737, -0.020774, 0.009761)
Obviously in order for my AI characters to perform as expected the vectors returned must be predictable. Attached is a screenshot of the path follow panel.
Steps to reproduce
Create a path on a 3d plane and place an object in the scene that passes its global position with:
getPathDirection(global_transform.origin)
Run this in 3.3.2-stable and compare the returned vector to 3.4-beta3 to see a different vector.
Minimal reproduction project
No response
The text was updated successfully, but these errors were encountered: