Definition and classification
The underlying technology of virtual digital human is updated and selected rapidly. In the future, more virtual human with vivid and rich image and strong interaction ability will be applied to the field of e-commerce live broadcast
Five technical modules make virtual human have human characteristics
Virtual digital human is composed of five modules: character image, voice generation, animation generation, audio and video synthesis display and interaction. The character image module enables digital people to have human image, which can be divided into 2D and 3D according to the type of graphics; According to the image style, it can be divided into four categories: cartoon, anthropomorphic, realistic and super realistic. The voice generation and animation generation modules enable digital people to have language ability and action ability. The audio and video synthesis display module is responsible for combining voice and animation into video, so that digital people have complete expression ability. The interaction module includes two parts: recognition and perception and analysis and decision-making. Through voice semantics, artificial intelligence and other technologies, digital people have the ability of perception and logic, so as to realize the communication and interaction with people.
Development direction
The 2D virtual human produced by virtual splicing has poor simulation, but its low production cost and rapid online ability make it the choice of many brands. The core technology of making 3D virtual human is motion capture, digital 3D modeling and slag dyeing. These three technologies have high requirements for production equipment and post-processing, and the cost is also high. However, the image richness and interactivity of 3D virtual human are significantly stronger than 2D virtual human, which can better meet the needs of e-commerce live scenes. Therefore, more and more brands pursuing quality are willing to try 3D virtual human
At present, the three common sources of virtual human IP in the market are: brand self built virtual human IP, third-party virtual human IP and homogeneous virtual human. Brand self built virtual human IP helps to strengthen brand image and improve the conversion rate of live broadcasting room, such as ” Three Squirrels Inc(300783) “. The third-party virtual human IP refers to the cooperation with the virtual human IP established by the third party to carry goods live, which can quickly bring huge traffic to the search and broadcast room, but it does not have long-term effects, such as “luotianyi”. Homogeneous virtual human refers to the use of virtual human without obvious characteristics for live broadcasting and goods, mainly functional
At present, the virtual human in the field of e-commerce live broadcasting is mainly in the style of two-dimensional and 3D cartoon. Compared with 3D virtual human with high realism and real image, the production cost of two-dimensional paper human and 3D cartoon style virtual human is much lower. Secondly, the image changeable two-dimensional and 3D cartoon virtual people are more easily accepted by consumers and play the role of market education in the early stage of market development. With the continuous maturity of virtual human technology, super realistic real person images will continue to appear, such as “ayayi”
