Updated on 2025-08-22 GMT+08:00

Concepts

Virtual Human

A virtual human exists in a digital space and has a realistic, human-like appearance and behavioral characteristics, as well as certain intelligence and emotional capabilities that enable it to interact with people and express itself. A virtual human is also known as a virtual image or virtual avatar. The core technologies behind virtual humans include computer vision (CV), computer graphics, motion capture and control, image rendering, and AI.

A virtual support personality is a hyper-realistic virtual avatar built on composited images based on deep neural networks. It has the following characteristics:
  • 2D model, which is generated after training a human video
  • No facial expression & skeletal data
  • AI-controlled only
  • Specified facial expressions and motions

Script, Image Layout, and Settings

Video compositing is required for video production and livestreaming. The following uses livestreaming as an example.

  • Video materials: as shown in area 1 in the figure, including templates, roles, backgrounds, overlays, videos, products, and text. You can use the preset materials or customize materials.
  • Image layout: as shown in area 2 in the figure. You can insert roles, backgrounds, overlays, text boxes, and videos into the image layout, and adjust the size and layer position of each material. All settings are visible in the image layout.
  • Livestream settings: as shown in area 3 in the figure
  • Script: Script design, including material preparations, layout design, and livestream settings, is required for video compositing to produce videos and livestream.
Figure 1 Page for editing a live room

Role

Virtual avatar model

Action

Action of the virtual avatar model