Game Development Reference
In-Depth Information
14 Using morph objects
In the preceding chapters we have looked at how to create and animate
a single mesh character. The techniques described are adequate for body
animation but using the same methods to animate facial animation can be
difficult to set up and often involves repetition. If only we could model a
frown, a smile and a scowl once and use this whenever the animation
required it. Model once and use many times means we need a way to
blend a mesh from one geometry to another. In many applications this
involves using morph targets. Morph objects are the easiest solution to
this complex geometrical problem.
Creating morph targets
You should have realized by now that my preferred CGI application
program is Lightwave. Version 5 came with a tool called 'Morph Gizmo'
that allowed the animator to create multiple versions of an object and
blend between them. Each version of an object must have exactly the
same number of points and if a point with index 100 was the left of the
mouth then the point with index 100 in any morph target must also be
the left of the mouth. The actual position of the point may change, but
its position in the topology of the mesh must be the same.
Lightwave 6 allows the animator to store multiple morph targets within
the same object file. These morph targets can be stored in groups,
perhaps a group for the Left Eye, Right Eye, Mouth and Nose. Each group
acts as a separate channel that can be handled independently. A real-time
application must be aware of the time taken to calculate the locations of
the target mesh. If you test the time taken to transform geometry and the
time taken to paint it, you will probably find that transforming takes only 10
per cent of the time while the final painting and rasterizing takes about 90
per cent of the time. This means that if transforming takes even twice as
long, it is only at most a 10 per cent hit in performance. As long as we do
Search Nedrilad ::

Custom Search