​
Instructions

​
General

  • Duration: Your video recording should ideally be 4-5 minutes long, but at least 2 minutes.
  • Resolution: Ideally, your training video should be 1080p or higher, but at least 720p.
  • Record your video in .mp4 or .mov format. Avoid iPhone cameras.
For best results, use a professional camera or a good laptop camera for your recording. On Mac, you can use QuickTimePlayer to record 1080p videos, which work quite well in practice.

​
Setup

Before recording, make sure you setup your recording environment such that:
  • Your face is clearly visible in every frame of the video
  • Your face occupies a sufficiently large region of the video
  • There are no other human faces visible anywhere in the video
  • The lighting doesn’t change throughout the video and there is no shadow on your face
  • Your background doesn’t move throughout the video and has a different color from your face and hair
    • If possible, we recommend to record your video in a green-screen setup

​
Pose

Throughout the recording, try to maintain the following pose at all times:
  • Keep your head upright and stable
  • Maintain direct eye contact with the camera
  • Keep your hands below your shoulders and avoid excessive body language
  • Maintain a pleasant smile throughout the video
Your avatar will mimic the pose, gaze, and head movements of your recording

​
Content

  • Record one continuous speech without major interruptions or video cuts
  • Include distinct pauses (1-2s of silence) between longer sentences and fully close your mouth occasionally
  • Talk slowly and clearly, ideally in English
  • Use complete words and avoid spelling out the alphabet, numbers, or similar.
For best results, begin your recording with a quiet moment where you limit movements and just silently look at the camera for around 5 seconds

​
Known Issues

The following scenarios are known to be difficult and will likely not result in good avatars:
  • Long beards or long hair obstructing the face
  • Other people in the video
  • Moving around or strongly changing the position or lighting throughout the recording
  • Strong head movements that obstruct the face or otherwise occluding the face throughout the recording
  • Looking away from the camera, extensive eye rolling, or longer segments where the eyes are fully closed