Introduction to Fixed-Function 3D Graphics and Hierarchical Modeling - Computer Graphics: Principles and Practice - page 145

Camel

Head Torso

Front left leg Front right leg

Front left
thigh

Front left
lower leg

Front right
thigh

Front right
lower leg

Front left
foot

Front left
shin

Front right
foot

Front right
shin

For readability,
only joint transforms
are shown here.

Rear left leg Rear right leg

Rear left
thigh

Rear left
lower leg

Rear right
thigh

Rear right
lower leg

Rear left
foot

Rear left
shin

Rear right
foot

Rear right
shin

Figure 6.49: Scene graph of a camel constructed without reusable components, allowing
individual control of each joint.

Camel

Head Torso

Front left leg Rear left leg Rear right leg Front right leg

Thigh Thigh Thigh Thigh

Lower leg

Foot Shin

Figure 6.50: Reducing the storage cost by reusing a lower-leg submodel, with no loss of
flexibility in joint control.

As the figure makes clear, each instantiation of the reusable LowerLeg has its
own knee joint transform, so there is no loss in flexibility—each knee joint is still
individually controllable. This DAG-based design is identical in functionality to
the tree-based design above.
Thus far, we have been focused on supporting a high-fidelity animation of the
camel's movement, but there other scenarios of interest. Consider a desert scene
in which hundreds of camels, seen from afar, are crossing dunes in a caravan.
The amount of processing needed to animate each hip and knee joint individually
might be considered not worth the cost, especially if the caravan is so far from the
viewpoint that such details would not be apparent to the viewer. In such a case,
we might choose a lower-fidelity motion in which the camels move in unison,