• Fixed LookAtModifier3D ignores Time Based Interpolation when Relative Option is True #115584

LookAtModifier

• It was not correctly considering its own bone_rest (or bone_pose when relative=true), this bug is exist from 4.4.
  • In other words, while this issue can be reproduced in 4.4 when the neck and head rests differ (even if relative option is disabled), since the neck and head rests are usually the same in most cases, the bug likely remained largely unnoticed and unreported.
• However, when relative is enabled, it seems difficult to apply restrictions only to the rest (so it cannot work similar to JointLimitation in IK). I added a note to the documentation, but considering this, the relative setting for the LookAtModifier should probably be false. I assume the relative setting for the AimModifier should remain true (it was changed in Add relative option to LookAt/AimModifier3D #111367 (review)). cc @lyuma
  • When applying limitations to Euler rotations, it can be defined their range as a rectangle such as PhysBones - Polar. However, the problem is that flip detection and other functions such as dumping are based on the Euler rotation axes, making immediate fixes here difficult. Although, it might be possible to send a follow-up PR later as an additional relative option like “keep limitation rest”.

Before (left: relative = false/ right: relative = true):

After fix (left: relative = false/ right: relative = true):

AimModifier

• If target is node (added in 4.6), it was not accounting for the skeleton's own rotation, causing the vector pointing toward the target to break when the skeleton moved.

Before (left: AimModifier / right: LookAtModifier):

After fix (left: AimModifier / right: LookAtModifier):

aim_lookat_demo.zip