On this page

Customization Process

2026-03-20

This document describes the ZEGOCLOUD Digital Human avatar customization process.

1. Choose the Digital Human Avatar Type

ZEGOCLOUD provides multiple Digital Human avatar customization options.

Digital Human TypeDigital Human AvatarFeaturesRequired MaterialsMaterial Specifications
Video Digital HumanReal-person video Digital Human - green screen backgroundCan replace green screen with other backgrounds, can create transparent background avatars5-10 minute recorded videoVideo Digital Human Production Guide
Real-person video Digital Human - fixed backgroundCannot change background, more suitable for fixed scenarios
Image Digital Human2D real-person image Digital HumanProvide a high-quality photo of a real person to directly generate a real-person Digital Human avatar1 photo/image is sufficientImage Digital Human Production Guide
2D virtual image Digital HumanSupports cartoon, sci-fi, anime, and other styles. Materials must meet specifications

Important Information

  • Regarding the customization quality and effects of Video and Image Digital Humans

    • Video Digital Human: Based on the provided real-person video, AI training generates a Digital Human with expressions, movements, and facial expressions that rival real people — a "1:1 replica". Therefore, the quality of the video materials directly affects the customized Digital Human effect. Please ensure the quality of your recorded materials.
      If this is your first time recording, please strictly control the recording quality. Refer to the Video Digital Human Shooting Guide for video recording. Please carefully control the model's appearance, makeup, expressions, and manage factors such as on-site lighting, set design, green screen, and equipment to produce the best recording materials. If your model is not a professional, it is recommended to practice in advance by referring to the sample videos to ensure the best results during recording.
    • Photo Digital Human: Based on your provided photo materials, AI training can bring the image to "life" and generate a Digital Human avatar. Photos support real people, cartoons, virtual characters, and other styles. The generated avatar has clear speech, natural expressions, and supports a certain range of body movements, making it lively and vivid.
      To customize the best Image Digital Human, please strictly refer to the Image Digital Human Production Guide. Providing high-quality photos will produce better Digital Human results.
  • Regarding whether green screen background materials are required:
    If you need to change the Digital Human avatar background or generate a transparent background Digital Human avatar, please record in a green screen environment. If you do not need to change the background or use a transparent background, and only need to modify the person's lip movements, you can provide fixed background materials. Digital Humans with green screen backgrounds incur higher costs when generating content. Choose based on your actual scenario needs.

  • Regarding the output specifications of customized Digital Humans:
    Please note that the materials you provide directly determine the output specifications of your Digital Human. For example, if your Digital Human material is a 16:9 image, the content output when driving the Digital Human will also be 16:9. Outputting other specifications may affect the Digital Human effect.
    Therefore, before providing materials, confirm the subsequent Digital Human usage effect based on your UI layout (landscape, portrait, aspect ratio, resolution, etc. For detailed specification information, please refer to the Image Digital Human Production Guide and Video Digital Human Production Guide). You can use images first to replace the Digital Human to confirm the UI effect, ensuring the final customized Digital Human meets expectations.

2. Customization Process

To ensure the best results, ZEGOCLOUD currently only provides offline customization services.

1

Contact your sales/pre-sales/technical support and provide your customization materials

Please carefully review the material specifications before recording:

2

ZEGOCLOUD reviews the materials and confirms they meet requirements

  • Feedback within 1 business day after material submission
3

Train the avatar and output a sample Demo

  • Video Digital Human: 3 business days
  • Image Digital Human: 1-2 business days
4

Customer confirms Demo effect and provides feedback

  • Please provide feedback within 2 business days of receiving the Demo
  • ZEGOCLOUD evaluates adjustment plans and conclusions based on the feedback
5

Customer is satisfied; ZEGOCLOUD publishes the avatar

  • After the customer confirms satisfaction, ZEGOCLOUD will publish the customized avatar and provide the avatar ID information
  • The customer can also retrieve customized avatars under their AppID from the Get Avatar List API

After the avatar is published, you can call the Digital Human avatar for short video generation or real-time stream output.

Previous

Try the Demo

Next

Image Digital Human Material Specification

On this page

Back to top