A MetaHuman Identity asset represents a performer's likeness. You need one for each performer to solve facial animation from depth data using the MetaHuman Performance asset. The identity is created from video footage of the performer captured in a set of specific poses, collectively called the identity take.
The identity take consists of a neutral pose and, optionally, a teeth pose (strongly recommended). If you are using footage captured on an iPhone, a left view and right view are also required. You select individual frames from each pose when creating the MetaHuman Identity asset in Unreal Engine.
Capture the identity take using the same class of device you use to record the performance. If you are recording the same performer on multiple devices, or their appearance changes significantly (for example, changes in make-up or facial volume), capture a separate identity take and create a new Identity asset for each device and each appearance.
The framing, lighting, and environment guidelines for performance capture also apply to identity takes. The sections below describe the requirements for each pose.
Neutral Pose
The performer should have a relaxed face with lips closed and teeth touching, but not clenched. The eyes should be open and looking forward, with the head facing forward.
The following examples show an incorrect neutral pose.
 |  |  |  |
Looking down | Incorrect framing | Lips not closed | Head tilted too far upwards |
Left and Right View
If you are using footage captured on an iPhone, left and right views are also required.
As with the neutral pose, the performer should have a relaxed face with lips closed and teeth touching, but not clenched, and the eyes should be open and looking forward. The head should be turned approximately 30 degrees to the right, then to the left, to capture both views. Ensure all facial features are clearly visible, for example, both eyes should be fully visible in the frame.
 |  |
Left View | Right View |
The following examples show incorrect left- or right-hand views.
 |  |  |  |
Looking down | Incorrect framing | Head turned too far | Head tilted upwards |
Teeth Pose
A teeth pose registers the position of the teeth and considerably improves the animation solved by the MetaHuman Performance asset. Although optional, it is strongly recommended.
As with the neutral pose, the performer should have a relaxed face with the eyes open and looking forward, and the head facing forward. The upper and lower teeth should be touching naturally in the performer's normal bite position. The lips should be parted just enough to expose the teeth. This means a slight, relaxed parting rather than a wide grin or grimace. The jaw should not be clenched. The tips of the upper teeth and at least some of the tooth surface should be visible.
Keep the rest of the facial expression neutral, as it is more important than maximizing the visible tooth surface.
The following examples show incorrect tooth poses.
 |  |  |  |
Looking down | Incorrect framing | Teeth not visible | Overly exaggerated |
