2023 Virtual Digital Human Industry Research Report

Chapter 1 Industry Overview

Virtual digital human refers to the existence in the non-physical world, created and used by computer graphics, graphics rendering, motion capture, deep learning, speech synthesis and other computer means, and has a variety of human characteristics (appearance characteristics, human performance ability, human interaction, etc.) capabilities, etc.) Virtual human can be divided into service virtual digital human and identity virtual digital human.

There are many ways to classify virtual humans: the first type is classified according to technology, and virtual humans can be divided into algorithm-driven (AI real-time or face pinching, etc.) and real-life driven (motion capture); the second type is classified according to visual dimensions, and virtual humans It can be divided into 2D type and 3D type; the third type is classified according to the structural composition, and the virtual human can be divided into digital type (users watch online) and holographic type (users watch with naked eyes on site); the fourth type is classified according to business model, virtual human It can be divided into IP types (KOL type, singing and dancing type, brand type, idol type launched by entertainment companies, star clone type) and non-IP type (functional type, academic type and identity type).

Table virtual person classification

Source: Asset Information Network Qianji Investment Bank

A virtual digital human should have the following three characteristics: first, the appearance of the owner, with specific features such as appearance, gender, and personality; second, the behavior of the owner, with the ability to express in language, facial expressions and body movements; It is the ability to have a human mind, to recognize the external environment, and to communicate and interact with people.

Figure Virtual human industry industry chain structure

Source: Asset Information Network Qianji Investment Bank Wind

Virtual idols are one of the earlier recognized subdivision tracks for virtual digital humans, and they belong to identity-type virtual digital humans. With the continuous growth of mass entertainment demand and the continuous iteration of network and video and audio technologies, my country's virtual idol industry has gradually entered the peak period of development. Data show that in 2020, my country's virtual idol market will be 3.46 billion yuan, a year-on-year increase of 68.8%; in 2021, my country's virtual idol market will be about 6.22 billion yuan, a year-on-year increase of 79.8%.

According to iiMediaResearch (iiMedia Consulting) data, the core market size of virtual idols in China will be 12.08 billion yuan in 2022, and it is expected to reach 48.06 billion yuan in 2025; the size of the surrounding market driven by virtual idols in 2022 will be 186.61 billion yuan, and it is expected to be 640.27 billion yuan in 2025 Yuan. iiMedia Consulting analysts believe that virtual idols have strong plasticity in terms of content and peripheral product output, and can continuously develop new explosive points according to the development of the trend of the times. Maintain a steady growth trend.

Figure 2017-2021 my country's virtual idol market size and growth rate

Source: Asset Information Network Qianji Investment Bank

The virtual anchor belongs to the representative service virtual digital person. Virtual anchors can reduce the cost of the anchor industry, reduce costs and increase efficiency for the stock market, and with its interesting advantages, the scale of virtual anchors has increased rapidly. At present, there are nearly 140 million anchor accounts in my country, of which virtual anchors account for 40%. According to statistics from the bilibili video platform, as of the end of 2022, a total of 230,000 virtual anchors have started broadcasting on station B, a year-on-year increase of 190%, and the number of live barrage interactions reached 1.42 billion. The annual broadcast volume of manuscripts reached 29.2 billion, a year-on-year increase of 112%.

Figure 1. Proportion of virtual anchors in my country

Source: Asset Information Network Qianji Investment Bank

Domestic virtual digital humans are also used in practical scenarios such as games, finance, cultural tourism, education, and smart services. For example, the popular girl group K/DA of "League of Legends" landed on the S8 final stage, and the audience can watch it through AR devices; Huawei Cloud Created the first virtual digital human, Yunsheng, and joined Huawei Cloud; Jingdong Import Supermarket launched a virtual recommendation officer "ViVi Zihan"; the first AI sign language anchor launched by CCTV News and Baidu Smart Cloud officially debuted; Baidu App launched Gong Jun Digital human voice assistant.

Typical application cases of domestic virtual humans in the following aspects:

(1) Games

• The popular girl group K/DA of "League of Legends" landed on the S8 final stage, and the audience can watch it through AR equipment

• "Glory of the King" launched the men's group Unlimited Kings

• In order to allow players to easily customize their own virtual images, in May 2019, Netease Fuxi AlLab and the University of Michigan invented the facial capture deep learning algorithm MelnGame, which can generate the shape and texture of a virtual face only by uploading a portrait photo of a character. The algorithm will also adjust the facial features according to the game age setting, which has been applied in games such as "Naishuihan" and "Eternal Tribulation".

(2) Finance

• HUAWEI CLOUD created the first virtual digital human, Yun Sheng, joined HUAWEI CLOUD

• Baidu created the first domestic bank "virtual employees"

(3) Cultural Tourism

• Jingdong Import Supermarket launched a virtual recommendation officer "ViVi Zihan"

• Netease Fuxi's first AI smart cultural tourism project landed in Hangzhou China Silk E. The project integrates the "Su Wen" genre virtual people in "Naishuihan" whose characters, skills and silk culture are highly related to the digital exhibition hall. Users can press the piano key Real-time interaction, "Su Wen" will dance with the change of melody

(4) Education

• The first Al sign language anchor launched by CCTV News and Baidu Smart Cloud officially debuted

• Xugu Future Technology launched the country's first virtual human in the field of education, Xiao Ai, who is set as a 12-year-old girl who likes to read and is good at sharing. It aims to stimulate children's interest in reading as a "squad leader" who leads reading companions. So far, Xiao Ai has participated in "Chinese Poetry Conference", "Time and Space Classroom" and other knowledge programs, and as a popular science blogger, he is active on multiple platforms such as Video Account and Douyin.

(5) Intelligent service

• The virtual host jointly released by Sogou and Xinhua News Agency shows a high-fidelity image, and can voice broadcast the news text input by the user

• Shanghai Pudong Development Bank's first virtual employee "Xiao Pu" can perceive the other party's emotions when serving customers, supplemented by facial expressions and gestures, and has the ability to actively learn and continuously improve service efficiency

• Baidu App launches Gong Jun digital human voice assistant

Chapter 2 Business Model and Technology Development

2.1 Industry chain value chain

The virtual human industry chain can be divided into basic layer, platform layer and application layer. The main users of the base layer and platform layer are 2B, while the main users of the application layer are 2C.

• The base layer is the tool layer, and software and hardware are the cornerstones of virtual human development. For example: 3D display devices such as VR/AR bring a sense of immersion and reality to users who use virtual humans, and modeling software can perform three-dimensional modeling of the human body and clothing of virtual digital humans. Representative companies at the base layer include Meta, Shadow Creation Technology, EPSON, Nvidia, Unity, and EpicGames.

• The platform layer is the provider of virtual human solutions. The platform layer uses technologies such as modeling, rendering, and motion capture to bring virtual humans to life. Well-known companies include Vicon, Tencent, Baidu, Sogou, SenseTime, Creative Technology, and iFlytek.

• The application layer adds a "soul" to virtual humans, creates and operates virtual human settings, and is ultimately applied to various scenarios. Leading companies in the application layer include BlueFocus, Mango Super Media and Next Generation Culture.

Figure virtual human industry chain map

Source: Asset Information Network Qianji Investment Bank

Figure virtual human industry chain panorama

Source: Asset Information Network Qianji Investment Bank

The virtual digital human system generally consists of five modules: character image, speech generation, animation generation, audio and video synthesis display, and interaction. Character images can be divided into two categories: 2D and 3D according to the dimension of character graphics resources, and can be divided into cartoon, anthropomorphic, realistic, super-realistic and other styles in terms of appearance; the voice generation module and animation generation module can generate corresponding text based on text The character's voice and the corresponding character animation; the audio-video synthesis display module synthesizes the voice and animation into a video, and then displays it to the user. The interaction module enables the digital human to have an interactive function, that is, to recognize the user's intention through intelligent technologies such as voice semantic recognition, and determine the subsequent voice and action of the digital human according to the user's current intention, and drive the character to start the next round of interaction.

Figure virtual human general system framework

Source: Asset Information Network, Qianji Investment Bank, Artificial Intelligence Energy Industry Alliance

2.2 Business model

(1) Virtual human + film and television

Film and television is a field that has the highest requirements for visual effects and has the greatest impact on society's perception of digital human images and brands. In recent years, China's film and television digital human special effects have achieved rapid development, and some special effects blockbusters have been recognized by the market. In 2019, China's film and television box office revenue exceeded 64 billion, an increase of about 8% for many years in a row, of which special effects films accounted for about 10%.

The country attaches great importance to the development of film and television special effects, and has issued a series of related supporting policies. In 2019, the Ministry of Science and Technology and the Ministry of Culture jointly issued the "Guiding Opinions on Promoting the Deep Integration of Culture and Technology", proposing to strengthen high-end cultural equipment such as laser projection, virtual reality, optical capture, video recording, high-definition production and broadcasting, and image editing. Independent research and development and industrialization. In 2020, the National Film Administration's "Several Opinions on Promoting the Development of Science Fiction Films" proposed that the development of science fiction film special effects technology should lead to the overall improvement of film special effects, and that financial and taxation support policies should be implemented. relief.

"Alita: Battle Angel", which was released in mainland China on February 22, 2019, is one of the typical application cases of the combination of virtual digital human technology and film and television. The heroine Alita in the play is a character made entirely with digital human technology. The film uses special facial capture equipment to accurately capture the face details of real actors, and then uses it as the basis for the movement of virtual characters in the computer, so that the movements and expressions of virtual characters can be as natural and lifelike as real people.

Special effects movies are widely recognized by the market, and supportive policies have been introduced intensively. The application of virtual digital human + film and television products represented by digital doubles is showing broad market prospects.

Figure 2013-2019 Chinese movie box office revenue and growth trend

Source: Asset Information Network Qianji Investment Bank

(2) Virtual Human + Media

The application of the media industry, represented by the virtual anchor, has well met the business needs of the media communication field for content generation, and has become a media weapon in the era of media integration.

According to statistics from the Qianzhan Industry Research Institute, the revenue of China's live video broadcasting industry reached 108.2 billion yuan in 2019, and 390 million people across the country are following virtual idols. Among them, station B, the largest two-dimensional activity community, has 11.4 million monthly active users, and virtual anchors account for the largest share of live broadcast camps. Receive 40%. During the same period, large-scale live broadcast platforms such as Bilibili and Huya also launched online and offline virtual anchor activities. For example, the number of viewers of BilibiliMacroLink series of VR live broadcasts has reached 6.6 million, and 100,000 "up hosts" (referring to people who upload video and audio files on video websites, forums, and FTP sites) can reach the revenue of 1 million live broadcasters.

In addition, in the field of traditional media, digital human applications represented by virtual hosts have also begun to enter the public eye. On the stage of the 2019 CCTV Network Spring Festival Gala, the AI ​​virtual host "Xiao Xiao Sa" based on Sa Beining's prototype appeared for the first time on the same stage as the prototype. This is also the first time in China that a human host and his virtual digital twin co-host a large-scale national cultural event, which has aroused strong attention from the industry and netizens.

(3) Virtual human + game

The game market is becoming increasingly fierce, and the demand for high-quality products is increasing, which may be the gospel of virtual digital human technology in this field. In 2019, the game market size reached 230 billion, of which RPG games that are more closely related to digital humans accounted for about 30% of the market. my country's domestic game regulatory environment and game version number approval policy are still tightening. The number of game users has only increased by 10 million compared with 2018. The market competition is becoming increasingly fierce, and the demand for high-quality products is getting higher and higher.

Virtual digital human technology can effectively simplify and speed up the game animation production process, and can enable more virtual characters in the game to have rich body movements and fine facial expressions at a limited cost, bringing players a more immersive game experience. For example, Netease Fuxi Lab has successfully applied virtual digital human technology to the production of animation scenes of multiple game plots such as "Naishuihan", and quickly generated animations without manual participation, which enables a large number of virtual characters to be rich in visual expression .

Figure 2015-2022 Chinese game market sales revenue and growth

Data source: Asset Information Network Qianji Investment Bank

(4) Virtual Human + Finance

Virtual digital humans can effectively help financial institutions achieve "cost reduction and efficiency increase", and major financial giants are vying to deploy "digital employees". The use of financial technology to achieve "cost reduction and efficiency increase" to deal with market competition has become a consensus in the development of the financial industry. The competition orientation with technology as the core has brought about an increase in financial institutions' information technology investment year by year.

According to light finance statistics, in 2019, the total investment in technology of 16 national banks exceeded 100 billion yuan, totaling 103.41 billion yuan. Among them, the investment of the four major banks of the Industrial and Agricultural China Construction Bank exceeded 10 billion yuan, and the investment of China Construction Bank was the highest, reaching 17.633 billion yuan. The virtual digital human has anthropomorphic expressions and movements, can conduct intelligent dialogues, and can be naturally combined with the financial industry with many service scenarios.

Many financial institutions are using virtual digital human technology to create "digital employees", which has become an important direction for technological innovation, cost reduction and efficiency increase. Taking the Industrial and Commercial Bank of China as an example, the digital human bank employees launched by it can display digital mimicry images on mobile terminals and large screens, and can be used in multiple business scenarios such as product marketing explanations, financial business handling, information broadcasting, and consulting questions and answers. Realize visual interaction with users, bring personalized services to users, effectively relieve users' concerns, improve user experience and dwell time, and truly make digital services "audible" and "visible" at the same time.

The figure shows the investment amount and growth rate of information technology of some leading international financial institutions

Source: Asset Information Network Qianji Investment Bank

Figure 1. The ratio of national total investment in science and technology to revenue

Source: Asset Information Network Qianji Investment Bank

(5) Virtual Human + Cultural Tourism

The digital cultural tourism industry has outstanding performance, and the application of virtual characters represented by virtual commentators may contribute vigorously to it. According to statistics from the China Tourism Academy, in the first three quarters of 2019, the operating income of cultural tourism was 6,218.7 billion yuan, an increase of 7.6% over the same period of the previous year. In particular, the digital cultural tourism industry performed outstandingly and became an important engine for the transformation and upgrading of the cultural tourism industry. Digital cultural content is developing in conjunction with new models such as Internet tourism, smart tourism, and virtual tourism, especially during the epidemic period.

In 2019, the number of domestic tourists was 6.006 billion, an increase of 8.4% over the same period of the previous year. In 2020, the number of domestic tourists will be 2.065 billion, a year-on-year decrease of 53.8%. In 2021, the number of domestic tourists will be 3.246 billion, a year-on-year increase of 12.8%. In 2022, the number of domestic tourists will reach 2.53 billion, a decrease of 716 million over the same period of the previous year.

Figure 2012-2019 National Domestic Tourist Number

Source: Asset Information Network Qianji Investment Bank China Tourism Academy

2.3 Technology Development

Figure virtual human technology threshold

Source: Asset Information Network Qianji Investment Bank

The early static modeling technology is mainly based on structured light scanning reconstruction. Structured light scanning and reconstruction can achieve a scanning reconstruction accuracy of 0.1 millimeters, but its scanning time is long, generally above 1 second, or even minutes, and it is not satisfactory in terms of friendliness and adaptability to moving targets such as the human body. Most of them are used in industrial production and testing fields.

In recent years, photographic camera array scanning and reconstruction have developed rapidly. At present, millisecond-level high-speed photographic scanning can be achieved (high-performance camera array accuracy can reach submillimeter level), which meets the needs of digital human scanning and reconstruction, and has become the mainstream method of character modeling. Internationally, IR, Ten24 and other companies have fully commercialized the static reconstruction technology, serving Hollywood large-scale film and television digital production, and the camera-type body scanning system produced by domestic companies such as Ling Yunguang has also been successfully applied in movies, games, and virtual anchor projects .

Compared with static reconstruction technology, dynamic light field reconstruction can not only reconstruct the geometric model of the character, but also obtain the dynamic character model data at one time, and reproduce the light and shadow effects of viewing the human body from different perspectives with high quality, which has become a key development of digital human modeling direction.

2D and 3D digital per capita has realized the intelligent synthesis of mouth movements, and the movements of other body parts currently only support recording and broadcasting. The underlying logic of the intelligent synthesis of 2D and 3D digital human mouth movements is similar. They are all related mappings from input text to output audio and output visual information, mainly for the collected text to voice and mouth video (2D) /Mouth animation (3D) data for model training, to obtain a model that can drive the mouth shape by inputting any text, and then intelligently synthesize it through the model.

However, the underlying mathematical expressions of 2D video and 3D lip animation are different. 2D video is a pixel expression; 3D lip animation is a vector expression of the BlendShape corresponding to the 3D model. Actions other than mouth shapes, including blinking, slightly nodding, raising eyebrows, etc. are currently achieved by looping pre-recorded video/3D actions using a random strategy or a script strategy.

Migrating captured motions to digital humans is currently the main way to generate 3D digital human motions, and the core technology is motion capture. Motion capture technology can be divided into optical, inertial, electromagnetic, and computer vision-based motion capture according to different implementation methods. At this stage, optical and inertial motion capture dominate, and motion capture based on computer vision has become a hot spot.

Advances in rendering technology and the emergence of new rendering technologies such as re-lighting have made digital human skin textures real, breaking through the uncanny valley effect. Before the emergence of PBR technology, all 3D rendering engines focused more on the realization of 3D effects due to the development of related software and hardware, which was not satisfactory in terms of realism. PBR is a collection of rendering technologies based on the simulation of imaging laws in the real physical world. Its key lies in the micro-surface model and energy conservation calculation. By more realistically reflecting the intensity of reflected and refracted light on the model surface, the rendering effect has broken through. Feeling plastic. The breakthrough of real-time rendering technology helps realistic digital humans realize real-time interaction, and the scope of application is rapidly expanding.

2.4 Policy Supervision

With the emergence of technologies such as 5G high-speed transmission, the Internet of Things, artificial intelligence, flexible display, and mobile high-performance graphics computing cards, the application of virtual reality technology has become one of the key development directions in my country. Looking back on the history of policy development, since the Ministry of Industry and Information Technology of my country issued the "Informatization and Industrialization Integration Development Plan (2016-2020)" in 2016, my country has gradually supported the technological breakthrough and innovation of virtual reality in the technical field; at the end of 2018, my country issued the first A policy document with the title of "Virtual Reality" shows that the Chinese government attaches great importance to the development and application of the virtual reality industry.

The 2016-2021 virtual reality industry policy is as follows:

• In November 2016, the Ministry of Industry and Information Technology's "Informatization and Industrial Integration Development Plan (2016-2020)" mentioned support for breakthroughs in virtual reality core technology and product and application innovation

• In January 2017, the Ministry of Industry and Information Technology's "Information and Communication Industry Development Plan (2016-2020)" proposed to give full play to the dominant position and leading role of Internet enterprises in innovation, and to drive the research and development and industrialization of virtual reality core technologies

• In December 2018, the Ministry of Industry and Information Technology issued the "Guiding Opinions on Accelerating the Development of the Virtual Reality Industry": it is the first document titled virtual reality in my country

• October 2019 "Industrial Structure Adjustment Guidance Catalog (2019 Version)": Incorporate virtual reality (VR) and augmented reality (AR) into the "encouraged" industries in 2019

• August 2020 "Guidelines for the Construction of the National New Generation Artificial Intelligence Standard System": Technical standards in key areas mainly focus on natural language processing, intelligent speech, computer vision, biometrics, virtual reality, augmented reality, human-computer interaction, etc. Intelligent applications provide technical support in the field

• The "14th Five-Year Plan" and basic electronic components industry planning in early 2021: pointing out to strengthen the integration of virtual reality technology and industrial applications

In addition to the virtual reality key promotion policies issued in 2016-2019, my country has also issued a number of related policies on the virtual reality industry in 2020-2021, mainly focusing on the in-depth application and industrial integration of virtual reality, such as May 2021 The "Notice on Carrying Out the Pilot Work of Publishing Industry Science and Technology and Standard Innovation Demonstration Projects" issued pointed out that it is necessary to strengthen the innovative application and research of virtual reality technology in the publishing field. The "14th Five-Year" Digital Economy Development Planning Notice issued in January 2022 pointed out that it is necessary to innovate and develop, deepen the integration of artificial intelligence, virtual reality, 8K and other technologies, and expand the application of social networking, shopping, entertainment, exhibitions and other fields. Promote the upgrading of the quality of life consumption.

Combined with the "Guiding Opinions on Accelerating the Development of the Virtual Reality Industry" issued by our country in 2018 and the related virtual reality policies issued by our country in 2021, the overall plan for the development of the virtual reality industry during the "14th Five-Year Plan" period in my country is summarized. It can be seen that the basic goal of my country's virtual reality is: by 2025, the overall strength of my country's virtual reality industry will enter the forefront of the world, master the key core patents and standards of virtual reality, and make a breakthrough in virtual reality and industrial manufacturing, learning and education, cultural and entertainment activities, foreign trade and business. and other aspects to strengthen integration and application.

Figure China's virtual reality industry development planning system

Source: Asset Information Network Qianji Investment Bank

Chapter 3 Industry Valuation, Pricing Mechanism and Global Leading Enterprises

3.1 Comprehensive financial analysis and valuation methods of the industry

Figure Industry History Comparison

Source: Asset Information Network Qianji Investment Bank Wind

Figure industry interval rise and fall

Source: Asset Information Network Qianji Investment Bank Wind

Figure operating income, net profit and growth rate

Source: Asset Information Network Qianji Investment Bank Wind

Figure PE multiple

Source: Asset Information Network Qianji Investment Bank Wind

Figure assets and liabilities

Source: Asset Information Network Qianji Investment Bank Wind

The virtual digital human industry valuation method can choose price-earnings ratio valuation method, PEG valuation method, price-to-book ratio valuation method, price-to-current ratio, P/S market-sales ratio valuation method, EV/Sales market-sales ratio valuation method, RNAV revaluation net asset valuation method, EV/EBITDA valuation method, DDM valuation method, DCF discounted cash flow valuation method, NAV net asset value valuation method, etc.

3.2 Industry Development

The development history of virtual human: The development of virtual digital human is inseparable from the progress of its production technology. From the earliest manual drawing to the current CG (Computer Graphics, computer graphics) and artificial intelligence synthesis, virtual digital human has roughly experienced embryonic, exploration, primary and There are four stages of growth.

Figure virtual human development process

Data source: Asset Information Network Qianji Investment Bank

In the 1980s, people began to try to introduce virtual characters into the real world, and virtual digital humans entered the embryonic stage. During this period, the production technology of virtual digital humans was mainly hand-painted, and its application was extremely limited. In 1982, after the broadcast of the Japanese animation "Macros", the producer packaged the heroine Lin Mingmei as a singer singing animation episodes, and produced a music album, which successfully entered the well-known Japanese music chart Oricon at that time. Minmei also became the world's first virtual singer.

In 1984, the British George Stone created a virtual character named Max Headroom. MAX has the appearance and facial expressions of a human being. He wears a suit and sunglasses. He has participated in a movie and shot several commercials. A well-known virtual actor. Due to technical limitations, its virtual image is realized by real actors through special effects makeup and hand-painting.

At the beginning of the 21st century, traditional hand-painting was gradually replaced by CG, motion capture and other technologies, and virtual digital humans entered the stage of exploration. At this stage, virtual digital humans have begun to reach a practical level, but they are expensive to manufacture, mainly appearing in the film and television entertainment industry, such as digital doubles and virtual idols. Digital avatars in film production generally use motion capture technology. Real actors wear motion capture clothing, and expression capture points are placed on their faces. The movements and expressions of live actors are collected and processed by cameras and motion capture equipment, and are given to virtual actors after computer processing. Role.

In 2001, the character Gollum in "The Lord of the Rings" was produced by CG technology and motion capture technology, and these technologies were later used in the production of films such as "Pirates of the Caribbean" and "Rise of the Planet of the Apes". In 2007, Japan produced the first widely recognized virtual digital human "Hatsune Miku". Hatsune Miku is a girl idol with a two-dimensional style. The early characters were mainly synthesized using CG technology, and the voices of the characters were synthesized by Yamaha's VOCALOID1 series , the presentation form is still relatively rough.

In the past five years, thanks to breakthroughs in deep learning algorithms, the production process of digital humans has been effectively simplified, and virtual digital humans have begun to enter the initial stage on the right track. During this period, artificial intelligence became an inseparable tool for virtual digital humans, and intelligently driven digital humans began to emerge. In 2018, Xinhua News Agency and Sogou jointly released the "AI Synthetic Anchor", which can display a virtual digital human image on the screen after the user enters the news text and broadcast the news, and the lip movement can be synchronized with the broadcast sound in real time. In 2019, Shanghai Pudong Development Bank and Baidu jointly released the digital employee "Xiaopu", which is also a virtual digital human made using artificial intelligence technologies such as natural language processing, speech recognition, and computer vision. It can provide users with "face-to-face" banking services through mobile devices Serve.

At present, virtual digital humans are developing toward intelligence, convenience, refinement, and diversification, and are entering a growth stage. In 2019, Doug Roble, head of the software research and development department of Digital Domain, an American film and television special effects company, showed his virtual digital human "DigiDoug" in a TED speech, which can capture and display real-time facial expressions and movements under the premise of photorealistic fidelity. This year, Samsung's STARLabs exhibited its virtual digital human project NEON at the CES International Consumer Electronics Show. NEON is a virtual character driven by artificial intelligence. and communication skills.

3.3 Driving factors

technological innovation

Virtual digital humans rely on multiple technologies. Therefore, technological progress is the main driving force for the development of the virtual digital human industry. According to the technical classification, virtual digital humans can be divided into real-person-driven and computing-driven, and there are differences in technical processes and details.

(1) Human-driven technical process

• Image design and modeling: 3D facial and body modeling based on IP design or original painting of real idols. Select a key.

• Modeling binding: map the identified key points to the model for binding. The number and position of key bindings affect the final effect.

• Performance capture: Use motion capture equipment or a specific camera + image recognition to capture key changes in body, expression, eyes, gestures, etc.

• Driving and rendering: Live actors (called Zhongzhiren in virtual idols) perform corresponding performances according to production needs, and drive virtual digital human performances in real time. In more sophisticated productions, retargeting will be required based on the distinction between live actors and models. And use different driving methods for actions, eyes, fingers, etc. Voice cooperation is required when necessary. Form a specific setting voice

• Generate content and interact: live broadcast, or record its actions to generate powerful content

Figure real-life-driven virtual human technical process

Source: Asset Information Network Qianji Investment Bank

(2) Computation-driven technical process

• Design images, scan real-life forms and performances, and collect driving data: use multi-directional cameras to scan general/specific models (full body or partial scans are available depending on final needs), and collect lip movements, expressions, and facial muscles when they speak Change details, posture and other data

• Image modeling, rigging: design the required model, or perform high-reduction modeling based on a specific real person. Perform keypoint binding. The number and position of key bindings affect the final effect.

• Training various driving models is the core step to determine the final effect: use deep learning to learn the potential mapping relationship between the model's voice, lip shape, and expression parameters, and form their own driving models and driving methods.

• Content production, based on input speech (or speech converted from input text), predicting parameters such as lip movement and expression: the core technical process is based on input speech, or first based on TTS technology (Text-to-speech, speech synthesis technology ), convert the input text into speech. Based on the voice, combined with the driving model obtained in step 3, and using the generative confrontation model GAN ​​to select the most realistic picture, and deduce the picture of each frame of digital human. Through the time stamp. Combine the voice and the digital human picture of each flight.

• Perform rendering to generate final content: Real-time rendering during live broadcast To ensure real-time low-latency rendering in a specific scene, technical issues such as the size of the computing framework and computing power supply will also affect the final life effect of the virtual digital human

• Additional steps: For the virtual digital human that needs to interact, the manufacturer will pre-set a question-and-answer database, knowledge graph, etc., and integrate it into the dialogue system of the virtual digital human.

Virtual human needs a variety of cutting-edge technologies to be greatly developed and applied. It is a super complex of various cutting-edge technologies, and advanced technologies fill its entire industrial closed loop. For example, its content production uses artificial intelligence and digital twin technology, its storage and authentication mechanism uses blockchain technology, its data processing uses artificial intelligence, cloud computing, and cloud storage technology, its network environment relies on 5G technology, and its virtual-real interaction and China Unicom Use human perception, 3D rendering, extended reality, brain-computer interfaces, wearables, robotics, and more. These technologies are currently in a stage of rapid development, and 5G, virtual reality, and augmented reality technologies are likely to be rapidly promoted and applied in the near future.

Diagram of computer-driven virtual human technical process

Source: Asset Information Network Qianji Investment Bank

Policy support

Since 2016, the prospect of my country's virtual reality market has been generally optimistic. At the policy level, the Ministry of Culture issued the "Opinions on Promoting the Transformation and Upgrading of the Cultural and Entertainment Industry" in September 2016, encouraging high-tech companies to use their own scientific research strength and technological advantages to actively introduce somatosensory, multi-dimensional special effects, virtual reality, and augmented reality. and other advanced technologies to enter the cultural and entertainment industry.

In December 2016, the State Council issued the "Thirteenth Five-Year" National Strategic Emerging Industry Development Plan, proposing to accelerate the development of innovative digital cultural and creative technologies and equipment such as virtual reality, augmented reality, holographic imaging, and naked-eye three-dimensional graphic display (naked-eye 3D). Development, enrich the content and form of digital cultural creativity, and promote the integrated development of games, film and television, animation and other related industries.

At the market level, the rapid development of China's Internet industry has cultivated a large number of user groups, aroused people's high enthusiasm for interactive entertainment, and an open attitude to new technologies and new applications, providing virtual reality industry with a C terminal market space.

Convergence needs of real world and virtual world

In the new round of information technology revolution, the combination of digital technology and the real world has become more in-depth. For example, technologies such as the Industrial Internet and the Internet of Things emphasize the digital control of production equipment and factories by digital technology; smart cities promote the intelligentization of urban governance and management; artificial intelligence provides digital auxiliary services such as automatic driving and human-machine translation, and users Requirements are directly transformed into data and algorithm products.

Figure virtual human domain application requirements

Source: Asset Information Network Qianji Investment Bank

3.4 Industry risk analysis and risk management

Table common industry risk factors

Source: Asset Information Network Qianji Investment Bank

(1) The development of virtual human technology is not as good as expected

Virtual digital humans are very dependent on the development of technology. If CG modeling, motion capture, AI and other technologies cannot be further upgraded to meet higher interaction requirements, some scenes cannot be truly implemented. In addition, the technical threshold cannot be lowered, resulting in high costs, and large-scale commercial applications may not be possible.

(2) Scattered management and unclear policy supervision

The management of virtual people is different from the management of real people and has diversity, involving cultural administration departments, education departments, public security organs, industrial and commercial departments, telecommunications management agencies, spiritual civilization construction steering committees, and corresponding group organizations.

However, in the specific management process, in addition to the cultural administration department and the public security department playing a leading role, other departments are mostly formal departments in virtual human management, and only in special rectification activities, such as Internet cafes and other Internet access service business establishments in recent years. The special rectification activities will be united under the supervision of the responsible department, and the various administrative agencies will be united. Usually, the management of the virtual society is basically neglected, so to a certain extent, it has caused the disorder of the management of the virtual society today. In addition, although there are so many functional departments in the form of virtual human management, there is no unified management department to coordinate, resulting in decentralized management.

(3) Ethical and moral issues

Due to the characteristics of openness and sharing of virtual humans, it is difficult for relevant government departments to conduct comprehensive supervision and management of online information, and problems such as information pollution have become increasingly prominent. At the same time, virtual people's excessive pursuit of individuality and uniqueness has caused some phenomena that violate normal ethics and morality, which has an impact on my country's mainstream thoughts, concepts, politics, ethics and morals, and has caused great damage to the network ecological environment of the virtual society.

(4) Legal issues

At present, the laws and regulations for virtual humans mainly include "Criminal Law", "Public Security Management Punishment Law" (2006), "Decision on Maintaining Internet Security" (2000), "Internet Information Service Management Measures" (2000) and other laws and regulations. The identification of cybercrime and cyberlaw in laws and regulations is too principled or general, and lacks certain operability. Moreover, most of the Internet-related legislation was made in 2000 (or even earlier), the pace of legislation could not keep up with the speed of Internet development, and the corresponding laws and regulations were relatively lagging behind.

In addition, due to the super-temporal nature of virtual people, under the principle of case jurisdiction definition, it is difficult to determine the jurisdiction of online cases, which will easily lead to jurisdiction disputes and affect the efficiency of case handling. The law is not yet perfect, and there is still a lot of room for improvement in terms of virtual assets, digital assets, and copyrights.

3.5 Competition Analysis

Porter's Five Forces Analysis

• Competitors in the industry: The number of virtualized enterprise registrations is increasing year by year, and there are many competitors in the industry, and the competition is also relatively fierce.

• Potential entrants: With the development of information technology, there is an increasingly broad market demand for virtual humans. At the same time, governments at all levels in China have proposed strategic plans to change modes and adjust structures, and vigorously support good support policies for virtual realization and virtual augmentation technologies, as well as China's increasingly open and good foreign trade environment has made more capital plans to enter the virtual human industry and obtain economic profits. The threat of potential entrants is greater.

• Substitutes: Whether it is an industrial market or a consumer market, the experience needs of real scenes have become an important driving factor for VR/AR demand. VR/AR products can meet higher-level requirements, and at the same time, their core functions are irreplaceable. Transformative interactive methods and immersion such as gesture recognition and motion capture determine their irreplaceable role in certain fields. need. So the threat of substitutes is weaker.

• Suppliers: The upstream of the virtual human industry supply chain is mainly raw material suppliers and service companies such as raw materials, machinery, electronic components, and industrial design companies. Virtual human product startups are in a relatively weak position among supply chain members and do not have strong bargaining power.

• Buyer: Technological progress will make products develop from bulky to light, and greatly improve user experience; market expansion will bring about a significant reduction in costs, which will lead to popularization of hardware and the emergence of disruptive products; capital support will drive the rapid development of the industry, The content is continuously enriched, the application field is continuously expanded, and a positive cycle of industrial development is formed. Buyers have strong bargaining power.

SWOT analysis

(1) Strength advantage analysis

With the continuous development of the mobile Internet and the continuous improvement of customer experience requirements, the current plane experience can no longer fully meet the user's requirements. The virtual human is constantly catering to the customer experience needs with the virtual reality experience, so as to achieve the real effect as much as possible: at the same time There are also diversified developments in the business model of virtual humans. Cooperating with film companies and game development companies or developing their own exclusive application software will become their own profit points: virtual humans will gradually be applied to all walks of life in the future, and its huge The application potential is still being developed.

(2) Weakness disadvantage analysis

At present, virtual humans are facing many technical bottlenecks that need to be overcome urgently: bottlenecks in hardware, image technology, data, etc., which also make it difficult for virtual humans to make further breakthroughs: the cost of VR is relatively high, which may lead to slow market penetration and some domestic prices. Low-level virtual human products cannot give users a good experience: virtual human applications are currently still focused on video and games, with relatively little content, resulting in low customer stickiness.

(3) Opportunity analysis

At present, the virtual human industry is in its infancy, but the future growth potential of the entire market is huge: according to the data of DigiCapital, by 2020, the global AR and VR market will reach 150 billion US dollars, and according to the statistics of the market research organization BusinessIntelligence, 2020 is only the beginning of the year. The market size of wearable VR hardware has reached 2.8 billion US dollars, and the compound growth rate will exceed 100% in the next five years. If you can seize the opportunity in the block market, you can bring unimaginable returns in the future.

(4) Threaten threat analysis

The virtual human industry will usher in an era of great explosion. This threat not only comes from the competition of domestic virtual human entrepreneurial teams such as Baofeng Mojing and Firework Workshop, but also from major global technology giants such as Facebook and Sony. , HTC and other threats to enter the VR industry, the virtual human industry in the future will be an era of hundreds of flowers competing for beauty and waves washing away the sand.

3.6 Important players of Chinese enterprises

Major Chinese companies include [002415.SZ] Hikvision, [002236.SZ] Dahua, [002414.SZ] Gaode Infrared, [603290.SH] Star Semiconductor, [688772.SH] Zhuhai Guanyu , [002008.SZ] Han’s Laser, [300390.SZ] Tianhua Super Clean, [688188.SH] Baichu Electronics, [301029.SZ] Yiheda, [688208.SH] Daotong Technology, [688002.SH ] Rui Chuang Micro Nano, [300296.SZ] Leyard, [0148.HK] Kingboard Group, [6088.HK] FITHONTENG, [1725.HK] Hong Kong Aerospace Science and Technology, [0327.HK] Pax Global, [ 3393.HK] Wasion Holdings, [0591.HK] China High Precision, [1206.HK] Tongfang Tide, [8073.HK] Xingye New Materials, etc.

(1) Hikvision [002415.SZ]: an intelligent Internet of Things solution and big data service provider with video as its core. Its business focuses on intelligent Internet of Things, big data services and smart business, and builds an open and cooperative ecosystem. It provides services for users in the public service field, enterprises and small and medium-sized enterprises, and is committed to building a smart city and digital enterprise with cloud-edge integration, IoT-information integration, and digital-intelligence integration.

(2) Dahua [002236.SZ]: It is the world's leading provider of smart IoT solutions and operation service providers with video as the core. Based on technological innovation, it focuses on the two core businesses of cities and enterprises, and continuously builds AIoT and IoT digital intelligence platform capabilities, around customer needs, comprehensively promote the digital intelligence upgrade of cities and enterprises, and provide one-stop intelligent IoT services and solutions for cities, enterprises, and families.

(3) Gaode Infrared [002414.SZ]: It is a high-tech enterprise specializing in the manufacture of infrared thermal imaging systems, and a world-leading professional manufacturer of infrared thermal imaging cameras. Since its establishment, based on independent innovation, Gaode has actively carried out the design and research of infrared optics, imaging circuit, image processing, artificial intelligence, mechanical structure and system engineering, and developed hundreds of infrared thermal images with complete intellectual property rights. System and high-tech photoelectric system, all technologies are leading domestically and internationally advanced, and have registered trademarks such as "GuideIR", "MobIR" and "Thermopro", which are well-known overseas.

(4) Hong Kong Aerospace Science and Technology [1725.HK]: relying on Hong Kong's status as an internationalization center and the industrial advantages of one of the four independent customs areas of the WTO, it will focus on satellite basic industries and satellite data applications to form an upstream and downstream industry serving commercial aerospace The ecological circle promotes the re-industrialization process of Hong Kong, opens up new industrial paths for the development of Hong Kong's social economy, and promotes the vigorous development of the global aerospace industry into the fast lane.

(5) FIT HONTENG [6088.HK]: FIT HONTENG is an industry-leading global solution provider. In order to make the world a better place, we are committed to realizing connectivity capabilities. The vision is to enable people to connect and share experiences using various devices through innovative technologies, products and services.

3.7 Global Key Competitors

Global non-Chinese major companies include Google, Facebook, IBM, [6981.T] MURATAMANUFACTURING, [APH.N] Amphenol (AMPHENOL), [006400.KS] SAMSUNGSDI, [GLW.N] Corning (CORNING), [ CAJ.N] Canon, 7751.T ]CANON, [6971.T] Kyocera, [6645.T] OMRON, [6762.T] TDK, [6506.T] Yaskawa Electric, [009150.KS] SAMSUNGELECMECH, [DLB .N] Dolby Laboratories (DOLBYLABORATO-A), [OMR.DF] OMRON, [IIVI.O] II-VI, [6976.T] Taiyo Yuden, [OLED.O] Universal Display.

(1) Amphenol (AMPHENOL) [APH.N]: Founded in 1932, it is a multinational electronics company headquartered in Connecticut, USA, which mainly produces and sells various electronic products such as connectors, cable assemblies and sensors. The company's products are widely used in communications, aerospace, automotive, industrial, healthcare and other fields.

Amphenol's product line includes various types of connectors, cable assemblies, antennas and sensors, among others. These products are widely used in communication, aerospace, automotive and industrial fields, such as mobile phones, tablet computers, aerospace equipment, automotive electronics and medical equipment, etc.

(2) Corning (CORNING)[GLW.N]: Founded in 1851, it is a multinational company headquartered in New York State, USA, and one of the world's largest manufacturers of special glass, ceramics and composite materials. The company's products are widely used in many fields such as electronics, communication, automobile, life science and industry.

Corning's product line includes a variety of specialty glass and ceramic materials such as optical glass, self-cleaning glass, heat-resistant ceramics and specialty glass fibers, among others. These products are widely used in communication, electronics, aerospace and automotive fields, such as mobile phones, tablet computers, LCD TVs, fiber optics and optical sensors, etc.

(3) TDK[6762.T]: It is a multinational electronics company headquartered in Japan. Founded in 1935, it is a leading manufacturer of various electronic products such as electronic components, storage media and sensors. The company's products are widely used in consumer electronics, information communication, automotive electronics, industrial automation, healthcare, energy management and other fields.

TDK's product line includes various types of electronic components such as capacitors, inductors, power supplies, sounders, magnetic materials, and sensors. These products are widely used in consumer electronics, automotive electronics, communications and computer systems and other fields. In addition, the company produces magnetic materials for storage media, magnetic heads and hard disk drives, among others.

(4) Yaskawa Electric [6506.T]: It is a multinational company headquartered in Japan. Founded in 1915, it is a leading manufacturer of industrial robots and automation equipment. The company has branches and sales networks all over the world, and its business involves automation control, motor and drive, motion control, robot, information technology and environmental protection technology and other fields.

Yaskawa Electric's robot products are widely used around the world, and its industrial robots are used in various fields such as automobile manufacturing, electronics manufacturing, semiconductor manufacturing, and food processing. The company's robot product line covers multiple fields, including SCARA robots, collaborative robots, Delta robots, 6-axis robots and other custom robots, etc.

(5) OMRON[OMR.DF]: It is a global high-tech company headquartered in Kyoto, Japan. Founded in 1933 as a manufacturer of electromagnetic relays, the company has gradually expanded its business over time and is now a company offering a wide range of automation, electronics and healthcare products.

OMRON's business covers many different fields, including industrial automation, electronic components, healthcare, environmental and social systems. The company develops and manufactures various electronic and electromechanical devices such as sensors, industrial automation systems, transportation equipment, medical equipment, home appliances, and game consoles, among others.

Chapter 4 Future Outlook

Virtual digital human is an emerging technology. With the continuous development and progress of technologies such as artificial intelligence, machine learning, computer vision and natural language processing, the future development prospects of the virtual digital human industry are very broad.

First, virtual digital humans will be widely used in many different industries and fields. In the entertainment and game industry, virtual digital humans have become an important tool for game and animation production. With the continuous development of technology, virtual digital humans will be able to simulate human behavior and actions more realistically, providing more real and immersive experiences for movies, games, virtual reality and other fields.

Second, virtual digital humans will also be widely used in business and industry. In the field of healthcare, virtual digital humans can be used to simulate surgery, teaching and training, etc., providing doctors and nurses with more practical tools and technical support. In the manufacturing and logistics industries, virtual digital humans can help companies optimize production processes, improve work efficiency and quality, and reduce production costs and waste of resources.

In addition, with the continuous development of virtual digital human technology, people will be able to better use virtual digital human to solve some problems in the real world. For example, virtual digital humans can be used in online education and training, remote work and collaboration, virtual tourism and cultural experience, etc., providing people with more flexible and diverse learning and lifestyles.

In short, the future development prospect of the virtual digital human industry is very broad, and it will be widely used in various industries and fields. With the continuous development of technology and the deepening of application, virtual digital human will become an important part of people's life and work, bringing people a more intelligent, efficient, convenient and comfortable life experience.

Cover Photo by Vitaly Otinov on Unsplash