Overview

Before customizing a virtual avatar image on Huawei Cloud MetaStudio, you need to shoot a video of a human by referring to Shooting a Human Video.

Please note that the quality of virtual avatar image modeling is closely related to your recordings.

To create better virtual avatars, please read the following virtual avatar video shooting standards carefully.

**Table 1** Overview of virtual avatar video shooting standards
Preparation Item	Preparation Subitem	Shooting Standard
Shooting Specifications	-	The total video duration is 5 minutes, the resolution and frame rate are 4K/25 FPS or higher, and the format is MP4/MOV.
Site layout	Background	Use a green-screen background with even color and without damage or wrinkles.
	Site Lighting	Use even and stable illumination and standard daylight color temperature. Ensure that the model is well-lighted and there is no shadow on his/her face. The light does not change significantly during the recording.
	Camera Installation	The camera is aligned with the person's eyes and focuses on the face area to ensure that the face is clear. It is recommended that you take photos vertically.
	Suggestions	Keep the recording environment quiet and free from noise interference.
Model	Facial Requirements	Avoid light reflection caused by excessive oil on the face. Ensure that there is no scattered hair on the face. To avoid light reflection, do not wear glasses. Ensure that the facial outline is clear and the model is enthusiastic.
	Dress Code	Do not wear clothes whose color is similar to the green screen, or clothes with green patterns. Do not wear metal earrings, bracelets, or watches that may reflect light.
	Model Performance	The model should smile and act naturally (including the head). After the action is complete, the hand should be put back to the initial position. The model should keep the mouth closed when he or she is not speaking.
Script	-	Use the provided Chinese/English script. A teleprompter is recommended to ensure smooth reading. Errors in reading will not affect the final delivery. There is no need to pause the recording when an error occurs. If possible, use a piece of script that you are familiar with.
Video Submission	-	The original voice of the video for training must be retained, and the audio and video must be in sync. Do not clip the video. Ensure that the narration, silence, and gestures of the video are exported to the same video.