YGmang 这种想法在现有的物理逻辑下无法实现也了解了
This idea cannot be realized under the existing physical logic
Are you sure? You can use physics on a bone elsewhere in the hierarchy and copy the transform, eg:
This way the head physics will not react to the body movement, the physics movement comes from elsewhere.
It may be awkward to setup an extra bone and transform constraint for every bone where you need this. In that case, the head could be moved so it is a child of the root, not the neck. Next create a transform constraint to make the head look like it is attached to the neck. Now create a physics constraint on the head, but set the constraint order so it is applied before the transform constraint moves it to the neck. Now the character's head physics doesn't react to the movement of the body.
To control the head physics, rotate/translate/scale the head before the physics constraint is applied. The bone is awkward to modify directly because the transform constraint is controlling the position, you'd have to turn it off, move the bone, and turn it back on. Instead you could create another bone and transform constraint to move the head. The 3 constraints look like:
The neck-to-head naming means "copy the neck transform to the head". It's a little tricky to name, as you could think about it like "the head moves to the neck". It does indeed, but technically the neck transform is copied to the head to achieve that.
Here is a project showing this, move the head-physics-control for head physics and note in animations when that bone doesn't move, there are no head physics.
https://n4te.com/x/2061-spineboy-physics.zip
We tried having "source" and "target" bones for physics constraint, though still using world transforms. The effect is identical to the above. It was easy to have translation come from the source bone, but rotation and scale required more changes. Currently we are thinking that since transform constraints can be used to achieve the same effect, it is not worth adding a source bone to physics constraint.
If physics constraint could work on local transforms, a source bone may be more useful. However, I'm not sure that is needed. See what you think about the above, I'm curious if it can be used to solve where you wanted local transform based physics.
