Metaverse News: Ali / Baidu / WiMi Gains Profits in Virtual Human Business


In recent years, with the advancement of computer graphics, deep learning, voice synthesis, brain-like science and other converging technologies, the rapid increase of people’s demand for digital life under the epidemic, and the accelerated promotion of the metaverse boom, the development of virtual digital human entered the fast lane, and the commercial value of virtual human was continuously tapped and released.

Virtual idols and digital avatars have been heating up with the release of movies such as Ready Player One and Free Guy, and reached the peak when the concept of “metaverse” boomed. It has become the “door knocker” for companies, organizations and platforms to enter the meta-universe. More and more companies started to launch virtual humans to improve the business layout of the metaverse. The virtual digital person has shown great commercial value and also highlights the strength of the head enterprises in the industry.

robot working with digital display
3d rendering robot working with digital display

Evolution of the virtual human industry

In the 1980s, avatars began to be created, but the degree of anthropomorphism was very low, relying mainly on hand-drawn creations, which only imitated humans in appearance. 2000 onwards, CG (computer animation) and motion capture technologies continued to innovate, gradually replacing the earlier hand-drawn creations. For example, the role of Gollum in The Lord of the Rings released in 2001, is created by CG and motion capture technology. This period saw the emergence of the first phenomenal virtual digital person: Hatsune Miku, a 3D virtual idol.

Hatsune Miku is actually a song synthesis software, when Crypton Future Media developed a sound source library based on Yamaha’s Vocaloid series of voice synthesis programs and used it to produce and release a virtual songbird character home series. Later in 2012, the female virtual singer “Luo Tianyi” from mainland China debuted, also based on Yamaha’s Vocaloid3 speech synthesis engine, and produced a virtual idol.

But whether it is “Hatsune Miku” or “Luo Tianyi”, we are still far from the virtual people we see now, because even with CG, motion capture, and other key technologies, they still can not achieve one of the most important functions: interaction with people. In other words, virtual idols have no personality. It wasn’t until after 2016 that avatars began to look more and more like “people”.

This year, Google AlphaGo defeated Go champion Lee Sedol, becoming the first computer Go to beat the professional Ku-dan, artificial intelligence began to come into the public eye. Voice recognition, natural speech processing, speech synthesis, speech-driven facial animation, and other technologies, as well as deep learning algorithm breakthroughs, provide technical support for the development of virtual humans.

Simply put, in the past to generate a virtual person, it may require animators, mod and elders to draw the character image at a stroke, frame by frame, after completing the original modeling, and then through real-time rendering, live-action capture, and other technologies to achieve, which consumes a relatively high cost of time and capital, mainly used in film and television entertainment industry, such as digital doubles. But after the development of AI technology, the demand side can directly use AI to complete the image modeling and subsequent drive, rendering, or the initial image modeling by hand, the subsequent drive by AI to complete the degree of anthropomorphism continues to improve at the same time, but also largely simplify the production process and reduce production costs.

For example, in early 2021, Epic Games, the developer of Unreal Engine, released Metahuman Creator, a tool for generating high-fidelity character images, which allows users to quickly create avatars with automatic mixing and manual adjustment based on a library of pre-made high-quality human faces, and small teams can quickly generate their protagonists with this tool. In this period, we see the avatar image of “Xiao Yang” who is resident in the host camp, “Hua Zhi Bing” who is a virtual student of Tsinghua University who can compose poems and music, and “AYAYI” who is a new master in selling goods.

Metahuman Creator

The commercialization model of virtual human

With the continuous emergence of virtual people, the outside world has given more optimistic predictions about the market prospects of this industry. Quantum bit in the “digital virtual human depth industry report” predicted that by 2030, China’s digital virtual human overall market size will reach 270 billion yuan. Behind the optimistic expectations, the first point that needs to be understood is, what value can virtual humans bring in the end? This determines the ceiling of this industry, as well as the industry chain-related participants on how to commercialize.

From the application layer, the current mainstream avatars can be divided into two types: service-oriented avatars and identity-based avatars. The former is mainly used to replace the real person’s services, and is a virtual representation of the service-oriented role in the real world; the latter is mostly a virtual IP/idol, intended to promote the production of virtual content, and can also be the second doppelganger of users in the virtual world.

For example, “Cui Xiaopan”, the employee who won Vanke’s Outstanding Newcomer Award, and “Xiao Pu”, a digital employee of Pudong Development Bank, are service-oriented avatars who appear to be able to help the company virtually; while the virtual idol group “A-soul” of Lehua Entertainment, Liu Yexi, who goes virus by a short video, and “Teresa Teng” who sings on stage with Zhou Shen, are all identity-based avatars.

Specifically to the application scene, and business model, these two types of virtual people have a large difference. Service-oriented virtual human is more oriented to B market, more important is how to empower enterprises through AI, the economic benefits do not lie in directly creating incremental income, but to reduce labor costs. Netease Fuxi marketing responsible person Tang Yinchen has publicly said that the logic of virtual people is essentially to reduce costs and increase efficiency, so that AI can better serve humans.

Because there are many simple, repetitive jobs in reality that can be replaced by virtual humans, and the chances of AI making mistakes are smaller than real employees, and are not limited by space and time. For companies entering the market, they can provide mature service-oriented avatars, one or more technologies, or personalized and customized solutions and services for different industries.

Identity-based virtual human is more oriented to the customer market, currently mainly focused on the game, media field, the realization mode to create virtual idols or virtual IP, first to expand the customer side fan traffic, and then through the flow of cash. For example, AYAYI has reached cooperation with Guerlain, Porsche and other international brands one after another after 1 month of debut in RED and became the first digital person employee of Alibaba in September 2021 as the digital mastermind of Tmall Super Brand Day; virtual person Ling Ling has also become a commercial darling, and has carried out cross-border cooperation with brands such as VOGUE, NAYUKI, Tesla, LEDIN and KEEP.

In other words, identity-based virtual people, the core of the business model is the fan economy, similar to real stars, through brand promotion, endorsement, participation in shows, live bounty, the release of music albums, selling peripherals and other ways to cash in, characterized by the ability to bring a high incremental income. But the requirements of such virtual people will also be relatively high, how to create a first-line virtual idol, and to ensure that it can stay in the first line for a long time, it is not easy, this test, in fact, is the brand’s long-term operation of IP, content capabilities.

Tech giants focus on metaverse segmentation

Last year, the concept of “metaverse” was a big hit, bringing fire to avatars, such as AYAYI. Many avatars were launched. Immediately afterwards, the industry set off a wave of investment. In a research report, Kaiyuan Securities pointed out that in the second half of 2021 alone, nearly 20 related companies completed financing, including star investment institutions such as IDG Capital, Sequoia China and Shunwei Capital.

However, the widespread popularity of virtual people is not only because of the “metaverse”, AI and deep learning and other technological breakthroughs have largely reduced the threshold for the production of virtual people, so technological progress is the fundamental thrust of the virtual people fire up. Especially for the virtual human that completely relies on AI-driven, the final effect presented is influenced by the combined effect of speech recognition (ASR), natural language processing (NLP), speech synthesis (TTS), and speech-driven facial animation (ADFA), and other technologies.

In other words, only with these technical elements, an avatar can see, hear, think, answer and present. Therefore, players who want to cut into this industry need to have the technical ability in one or several of the above.

The current digital avatar industry chain is divided into three layers.

1)the basic layer: mainly provide a modeling / rendering engine and other basic hardware and software.

Digital virtual human visual production part of the industrial process, covering the modeling, bone binding, motion capture, surface capture, rendering, material solving and other aspects, need to use display devices, optical devices, sensors and other hardware products, modeling binding, rendering engine and other software products. This layer does a better job of most foreign companies, such as the familiar King of Glory, Genshin Impact is created based on the United States Unity (U.US) company’s rendering engine.

2)Platform layer: providing software and hardware systems, production technology service platforms and AI capability platforms.

Most of the Chinese companies entering the market are located at this layer. We need to pay special attention to the leading AR plus AI provider WiMi Hologram Cloud (NASDAQ: WIMI), which relies closely on the three core advantages of data, algorithms and scenes, and continues to promote the deep development of business layout for the virtual-real integration of the real Internet, building up a business ecology of virtual people, virtual scenes and virtual marketing services intertwined to achieve rapid accumulation and incubation of the metaverse ecology.

As one of the first companies in the world to lay out the metaverse, WiMi has obvious advantages in creating classic IPs, linking the digital world, and providing resource contents corresponding to the real world, such as animation, e-commerce, and social entertainment. Up to now, it has nearly 5,000 virtual IPs, covering a wide range. These include 2961 holographic IPs for virtual education, 851 holographic IPs for virtual tourism, 739 holographic IPs for virtual art and entertainment, and 103 holographic IPs for virtual science.

With the imminent arrival of virtual human 3.0 era, WiMi will also explore more virtual human application scenarios, combine AR, VR and other cutting-edge technologies to create immersive interactive scenarios where reality meets reality, break the traditional live broadcast model with virtual anchors, explore the potential value of more virtual humans and drive business model innovation.

In addition, Baidu, Tencent, Ali, NetEase and other Internet majors have entered the virtual human track, they are based on their own precipitation in the underlying technology, hardware architecture, in the industrial chain layout will be relatively comprehensive, often providing the corresponding digital human technology service platform, as well as for different industry solutions for different industries.

For example, Netease’s Fuxi can provide solutions for the game industry, intelligent interaction solutions for space, virtual spokesperson solutions for cultural tourism, etc.; for example, Ali has an AI virtual model tower used in e-commerce, merchants only need to upload a flat map of goods, you can generate the model’s actual picture; for example, Baidu Intelligent Cloud launched an intelligent digital human platform Xi Ling, dedicated to building intelligent service-oriented & performing arts For example, Baidu Intelligent Cloud launched the intelligent digital person platform Xi Ling, which is committed to creating intelligent service-oriented & performing arts digital person, and also launched the intelligent interaction solution of virtual image based on real-time rendering of terminal.

3) application layer: the virtual human applied to various scenarios.

For example, the relatively mature film and television, media and game fields, as well as finance, cultural tourism, etc., which is also an important way to achieve commercialization of digital virtual human.



Data show that China’s virtual digital human market size is showing a high growth trend. It is expected that by 2026, the market size of the virtual human will reach 10.24 billion RMB, and the virtual digital human will be widely used in various scenarios. As you can imagine, the competition in this track will climb to a white-hot stage in the future, and people will see the figure of the virtual human in more and more scenes. It can be said that the virtual human industry is an opportunity to catch up or even overtake, but the companies that are already in this industry or will enter it need to think about whether they want short-term benefits or long-term digital assets, which will likely determine how far they can go in this industry in front of a new round of IP incubation and commercialization opportunities.