Processing Configuration

The processing configuration of SGX includes resource selections and various settings.

Character

SGX processing always requires selecting a character control file. This file contains all the information about your character’s muscles and behavior modes that SGX needs to drive the character. It is also used to inform the user about the available behavior modes and expressions.

It is critical to use the right character control file to produce animation for the intended character. Using the wrong file can lead to animation that is not well suited to the character rig.

Language

If transcripts are included in the input, you must select the appropriate language pack.

It is essential to use the right language pack for the language and dialect heard in the audio. See Language Support for more information, including the list of supported languages and dialects.

If processing without transcripts (audio only) you do not need a language pack. Audio-only mode works for any language, even fictional ones. It uses a universal model of the human vocal tract to map acoustics to articulation. See Input Options for a comparison of audio+transcripts vs audio-only methods.

Behavior Controls

Behavior can be directed in various ways by the user. For an overview, see the separate article on Behavior Controls.

Pre-Roll and Post-Roll

By default, each output animation has the same duration as the input audio file. However, if the audio file is cut sharply to a minimal duration with no starting or ending silence, then the resulting animation could start or end with the face in an active state, since the transitions away from and towards the resting state will take place outside the audio timeline. If you want to force the animation to start or end in the resting state, use pre-roll and post-roll.

If the pre-roll option is used, some padding will be added to the beginning of the animation to allow time for the transition from the resting state into the first syllable of speech. Likewise, if post-roll is used, some time will be added to the end of the animation to allow time for the muscles to resolve to the rest pose. There are three options for pre-roll and post-roll:

None - No time added (default option)
Auto - SGX automatically calculates the amount of time needed to transition from rest pose to animation for pre-roll, and from animation to rest pose for post-roll. This may result in different values for each event.
Fixed - The user supplies fixed pre-roll or post-roll durations to use for all events.

Note that when deploying animation, the start of the audio should be offset from the start of the animation by the duration of pre-roll.

Active Start and Neutral Finish

By default, every SGX animation begins in the neutral or rest pose (though the event may need pre-roll for this beginning to be seen) and it ends in the middle of the last nonverbal expression. So the normal progression is neutral start to active finish.

The rationale for an active finish is that returning to an expressionless neutral pose at the end of an expressive performance can look unnatural and strange.

However, at times you might want to change these default behaviors, for which there are two inverse options: active start and neutral finish.

With active start, the animation begins with the first nonverbal expression in the middle of its execution.

With neutral finish, all of the muscles return to rest at the end of the animation. (In case the audio is cut very close at the end, you can use post-roll to see the full transition to rest.)