A lot of games are content to have the character nod and gesticulate, for example, Metal Gear Solid has no facial animation at all, despite the large amount of speech. If facial animation is used at all, it is usually achieved through one of two methods:

Texture map replacement:

This has been used notably in Ultima IX, Gabriel Knight 3, and Soldier of Fortune. This technique involves storing different texture maps for different phonemes. The maps are switched on the fly, creating the illusion of speech. This technique is the most similar to traditional animation techniques.

Advantages:

  • Quick to implement.
  • Easy for people used to traditional animation lip-sync to do.
  • Not processor intensive.
  • Works well on low polygon models.
  • Disadvantages:
  • Not automatic, somebody has to go through all the speech and animate it.
  • Can be memory intensive, as all texture maps must be stored in the graphics memory.
  • Not so good for high polygon models – it is apparent there is no real movement.

  • Movement based on sound level:

    Demonstrated in Half-Life, this approach is surprisingly effective. The amplitude of the speech is converted into displacement for the jaw, making the mouth move in time with the sound.

    Advantages:

  • Quick to implement.
  • Simply feed in a sound file and it works.
  • Good for games which are designed to be expandable by users – if they want to create new speech, they don't have to animate lip-sync to go with it.
  • Works well on low polygon models.
  • Disadvantages:
  • Not realistic – has a tendency to make people look like puppets.
  • Again, not suitable for high polygon models – as the poly count gets higher, people demand more realism.

  • Both of these methods work reasonably well for current games, but with the graphics potential afforded by Dreamcast, Playstation 2, X-Box and next generation PCs, new methods need to be used.

    Continue to Deciding On The Project.
    Home.