How to get world-space matrix of any pose bone?

Question

Is there a command for getting world-space matrix of pose bones, same as obj.matrix_world for objects? I found pose_bone.matrix, but it gives weird result for me.

Answer 1

Yes:

As the documentation says, a pose bones matrix is in object space, this means - unless the object has no loc/scale/rot applied, you will have to apply the objects matrix.

matrix_final = obj.matrix_world @ pose_bone.matrix

This test script adds an empty at the pose-bone.

# Assume we are in pose mode with an active bone
import bpy
from bpy import context
pose_bone = context.active_pose_bone
we can get the object from the pose bone
obj = pose_bone.id_data
matrix_final = obj.matrix_world @ pose_bone.matrix
now we can view the matrix by applying it to an object
obj_empty = bpy.data.objects.new("Test", None)
context.collection.objects.link(obj_empty)
obj_empty.matrix_world = matrix_final

Answer 2

mathutils now uses the PEP 465 binary operator for multiplying matrices.

matrix_final = obj.matrix_world * pose_bone.matrix

becomes

matrix_final = obj.matrix_world @ pose_bone.matrix

---

Since 2.8 object are now stored in Collections, so :

context.scene.objects.link(obj_empty)

becomes

bpy.data.collections['Collection'].objects.link(obj_empty)

Here is the complete script updated:

# Assume we are in pose mode with an active bone,
# and a collection named collecion exist

import bpy
from bpy import context

pose_bone = context.active_pose_bone

# we can get the object from the pose bone
obj = pose_bone.id_data
matrix_final = obj.matrix_world @ pose_bone.matrix


# now we can view the matrix by applying it to an object
obj_empty = bpy.data.objects.new("Test", None)
bpy.data.collections['Collection'].objects.link(obj_empty)
obj_empty.matrix_world = matrix_final

Answer 3

Again with tip to tail offset.

Slight addition to @ideasman42's answer to enable globally locating to tip or tail of bone based on an offset. 0 is as above at fat end of bone (head), 1 at skinny end (tail).

This is to emulate somewhat the option in copy transform constraints to locate with offset along bone.

The vector

(pb.head - pb.tail)

describes the object space vector from head to tail A translation matrix is created to translate point of interest locally along the bone.

# Assume we are in pose mode with an active bone
import bpy
from bpy import context
from mathutils import Matrix
add empty half way along bone
offset = 0.5 # 0 for tail 1 for tip.
pb = context.active_pose_bone
T =  Matrix.Translation(
    offset * (pb.tail - pb.head)
    )
we can get the object from the pose bone
obj = pb.id_data
matrix_final = obj.matrix_world @ (T @ pb.matrix)
now we can view the matrix by applying it to an object
obj_empty = bpy.data.objects.new("Test", None)
context.collection.objects.link(obj_empty)
obj_empty.matrix_world = matrix_final