Is the virtual human explosion in the era of the metaverse?

A virtual person is a virtual image of a person's face, features, and behavior that is displayed in digital form and presented through technology. As an important medium in the metaverse, virtual people can bring rich content and experience to the metaverse. Fake people can print, but how do you make a fake person a different "soul"? Let's start with the logic of business development.

Quadrant classification of virtual humans based on content design and implementation. In two dimensions of design, functionality and application, virtual humans can be divided into four categories: service-oriented virtual human (PGC mode of operation), virtual idol (PGC). IP value), avatar (UGC mode) and creative properties. . . (UGC IP value). The four types of virtual humans are technically similar in production, but have completely different development in their application.

virtual human state

Virtual Reality Status: Business and Community Integration, Operational Improvement and IP Value

Fake characters have recently gained popularity, and on October 31, Douyin-born fake character Liu Yexi received millions of views overnight. Liu Yexi isn't the first character to be fake. Hua Zhibing from Tsinghua University, celebrity Douyin Axi and goddess Jiyuanmei have many fans on the big platform. Nearly 400 million people nationwide follow virtual idols, and last year virtual commerce doubled to 200 billion people in two years. At NVIDIA GTC 2021, Huang Renxun focused on introducing toy-me virtual avatar and virtual humans, and the response was great.

What is a virtual person? Virtual people are in digital form with human appearances, characteristics and behaviors and trust the virtual images seen by the device. Stopped,A virtual person is a digital image that gives a user a personal understanding.. Now, virtual people can play many roles, such as virtual anchor, virtual idol, virtual reporter, virtual assistant, etc.

Virtual people are an important part of communication that can bring rich content and past experience to the metaverse.

The development and operation of behavior patterns largely depends on the operation of the expert group and the UGC content of the interest of the community, and the application scenario of the delusional behavior, mainly on the functionality and value of Intellectual property. In two dimensions of design, function and application scenarios, virtual humans can be divided into four groups: service-oriented virtual humans (PGC function type), virtual idols (PGC IP values), and digital avatars ( function type UGC). Creative support (UGC IP rate). The four types of virtual human applications are completely different and integrated with technology, but with some design differences.

Virtual human services provide social services to users.Compared to chatbots, digital services and digital people, the advantage of virtual people based services is that they can perform more important tasks due to high modeling and intelligence. From movies and TV to finance and games, virtual people can offer a wide range of services, providing smart and profitable services to users. Xinhua News Agency virtual news correspondent and CCTV presenter using virtual name fall into this category, are jobs that need human images and have become a good landing point for service pointers.

A virtual idol is to display a virtual image with a personal identity by means of.Virtual idols can also make movies, advertise in media, advertise products, participate in merchandise, and even make movies and TV shows. The virtual idol is made by the IP company and the content is created by a team of experts. For example, in 2016, a Japanese virtual idol "Gangai" was born, and the main creative team provided a cute and cute "artificial brain retardation", and in 4 months, the number of fans reached 400,000, mainly on platforms video. job. This type of 'V-Tuber' started to come to mind. .

The avatars and virtual idols service is developed and operated by a team of experts to enhance the interaction between the platform and the user. With the rise of MCN in recent years, these avatars will have more revenue and business models, leading to new "new fun". Like News and Douyin. Overall, these virtual human beings work fine as PGCs, but the ecosystem is also closed, and the production and operation process is done by their own team or outsourced. The “soul” of a virtual person is defined by the organization and created by the production platform.

The virtual avatar was created by the game "Pinch Face" to provide users with an intuitive experience.The game 'Pinch Face' which meets the user's personal needs is a game of 'The Elder Scrolls 3' and is now widely used in many games. . During the long development period, the tunable parametric data type of the "Pinch Face" system was very good, and its complexity gradually increased as the algorithm improved. In the future, virtual avatars, avatars, will not only be popular in gaming applications, but also encourage social interaction. A virtual avatar can enhance the user experience while meeting the requirements of the virtual application.

According to the designers in the community, the images and representations of these fake characters are not the same. The best-known creative medium "Hatsune Miku" is a soundtrack released by Crypton on the VOCALOID text-to-speech engine with anthropomorphic imagery. Due to the beautiful soundtrack of the original soundtrack and the printed media of the Krypton company, the Hatsune Miku community has grown enormously and great community work has been born. In 2009, the world's first holographic figure held a concert of virtual idols. . Later in China, "Luo Tianyi" was also introduced, and the producer bought the music of "Luo Tianyi" so that Luo Tianyi could sing his favorite songs.

Virtual human development: PGC and UGC integration, hyperrealism, and low-threshold dual development

The rise of virtual human tracking is not a one-time phenomenon, and application scenarios continue over many years of technology development. In the 1980s, manufacturers started trying to create digital images of themselves. However, due to the limitations of digital people at the time, the app was restricted as it was hand drawn in 2D. In the early 21st century, as technologies such as computer-generated (CG) animation, voice capture, and sound synthesis gradually developed, virtual humans began rapidly and virtual humans created with CG technology were widely used in appearance. Advances in information technology over the past five years have made it easier to create virtual people, improve their interactions, and enter rapid development. Now, the fine design of virtualization, capture, and AI interactions continues to improve, enabling virtual humans to achieve real-world integrity and the ability to express imaginative ideas and communicate. .

Although the creation of virtual humans continues to improve, the manufacturing cost is still lower. Driven by advances in technology, supported by technological education, deep study, computer vision and other technologies, virtual humans have become increasingly sophisticated in terms such as facial expressions, selling images, acting, speaking skills and connections, which affect the heights of the future. characteristics. At the same time, with the advance of AI algorithms and the emergence of production platforms, the beginnings of virtual human creation have slowed down. False images marked only count and divide surreal images against real images by modeling and modeling. Dynamic displays need to be modeled, animated, and sound-based, which often requires skill capture. Interactive virtual humans need intelligent intelligence to recognize and interact with creative users.

Virtual Human Production - Counting energy is king and the universe exists.

Currently, the most important method to create virtual people is to scan models.It can be divided into two types: static design and static lighting design, static electrical equipment is also important, and high-precision dynamic lighting field. Three-dimensional reconstruction technology appeared. In the future, dynamic light field reconstruction technology will be used in addition to the production of static virtual humans to improve the light and shadow of virtual people and the vision of people using it. the use of renewable energy sources will gradually decrease and achieve a noise difference. .

Platform-based devices support the creation of small, high-precision virtual humans.In early 2021, EpicGames developed MetahumanCreator, a tool for creating beautiful images, based on pre-built prototypes, allowing users to quickly and easily customize their own virtual human models. The function of these tools is to allow small groups and individuals to quickly and easily create the characters they need, improving visual quality and reducing production costs.

Virtual human drivers rely on providing human motion to the virtual human model. Migrating captured motion capture into a virtual human is the primary way to create 3D virtual human motion. Motion capture technology can be divided into optical, inertial and computational optical motion according to the different applications.

Optical capture and inertial capture are commonly used in manufacturing. Optical capture is mainly used in medical, sports, film and other industries, for example, in October 2021, Qingtong Vision and Huawei will provide the world's first two-dimensional 5G VR idol in media, perfecting the elegant dance of virtual idols. . . Inertial capture has also been widely used in film and television production to better represent 3D virtual idol images and interact with users. In the future, with the improvement of computer technology, it is hoped that motion sensors will replace poor sound recording and broadcasting technology and have become an important technology in the development of virtual actors.

Visual capture technology dramatically reduces initial usage. Visual sensing is commonly used in consumer applications, and face and body capture can be achieved by the phone's built-in deep sensing camera. As virtual idols are rapidly gaining popularity among young people, low-resolution solutions are expected to become the first choice for UGC developers to switch to virtual idols. For example, APPlivelinkface launched by Epic can easily capture user face movement and push it to the production platform.

Rendering technology can be divided into real-time and offline rendering. The former is fast and suitable for gaming or interactive scenes, and the latter is designed for scenes that require high power and precision. Real-time rendering refers to the calculation of time and output of graphics data, where each frame is counted against the frame against the actual ambient location, camera position, and whatever is not. The first real-time rendering technology was short-lived and used limited computer equipment, but the efficiency of processing speed and accuracy has been greatly improved with the improvement of algorithmic counting power and hardware level . Offline rendering technology image files are not produced by calculating the real time, the operation time is long, the quality is high, and the equipment includes a lot.

Create false symbols - counting power is king and the edge is up. Today, when creating virtual humans for the metaverse, we will focus more on edge computing. As mentioned earlier, the changes have focused on the integration of virtual and reality, offline rendering is insufficient, and the real-time rendering space requires high levels of power consumption. Although most energy calculations in the market focus on the cloud, we believe that the real workload must be solved at the edge of the terminal, which uses a lot of edge calculators. This architecture is different from the existing one. communication counts power. architecture. In the past, Unity has worked with Verizon to develop high throughput, low latency digital solutions ranging from entertainment apps to business devices. Communication and IT infrastructure service providers play a role, as independent companies cannot solve the performance problems of the Chinese side; Following. effect.

Virtual Human Interaction - A combination of virtual and real AI and real people

Human-to-human digital communication can be divided into AI driving and human driving depending on the driving environment. Real-time integration with message capture allows virtual people to interact with the audience in real time. Expert drivers can read, analyze and recognize the external input data of the intelligent machine, determine the human data after the release based on the analysis, and then drive the character model to create the sound and function. Make digital people interact with users. The development of virtual people is slowly opening up the market. With the development of advanced technologies such as deep learning, machine learning, computer science and natural language processing, virtual humans will gradually integrate more customer feedback, intelligent marketing and other operations. , and create a positive image for customers. point towards.

AI-powered virtual humans rely on a variety of technologies such as speech recognition, natural language processing, voice communication, facial speech, and more. In the field of voice recognition, home businesses like iFlytech, Baidu, Tencent, and Ali all have layouts. The development of cognitive understanding in natural language is slower and several times more difficult than fluency in speech. Good companies include Google and IBM. Text-to-speech is now widely used, but is often recorded and released as clips and is still far from true self-expression. The virtual human achieved intelligent communication within the context of spoken movement, which is important in creating synergy between text and audio output and visible data output. The main design idea is to train the model with written to speech or spoken animation files, obtaining models that can drive spoken animation by writing texts, and then use the model to intelligently create the image of human mouth. .

What is the cost of virtual human commerce?

1. Rapid service-oriented virtual human development is needed.

Along with advocacy, media will be a challenge in service-oriented virtual human development. On October 22, "14th Five Years" Research and development for radio, television and broadcast network, it was pointed out that virtual anchors and animated sign language should be widely used in media coverage, weather forecast, various broadcasts and research. and course. , and new procedures. Thus, service-oriented virtual people in the media have accelerated these efforts. Xiaozhen is a virtual journalist co-founded by Xinhua News Agency's National Core Institute of Media Convergence Production Technology and Systems and NExT Studios, a subsidiary of Tencent Interactive Entertainment. They were unveiled on June 17 and June 12 in Shenzhou. On the day of the pilot's release, the on-site chat got closer to the public. CCTV News showcases AI technology to create the first reality

The AI ​​signature anchor project will provide voice assistance to deaf people in China for the Winter Olympics.

The reality of the avatar service is now pushing the boundaries of Uncanny Valley, and application scenarios will guide its development. "Uncanny Valley Theory" shows that humans have a strong sense of human nature, and some of their immediate reactions are very disgusting. Even the slightest difference between a robot and a human can make it obvious and dazzling, and very harsh and scary. With the development of virtual machine users, the creation of virtual people with surreal performance makes the services provided by virtual people more. When technology pushes users to the edge of this unique valley, when virtual and real people will not know the difference in appearance, application scenarios will lead to greater room for improvement.

Virtual human services can modify many rules. By creating virtual people for special occasions, savvy users can improve themselves. Events include film and television, finance, culture, education, health and shopping.

2. Steadily Growing Virtual Idol Market

Currently, the virtual idol is the largest virtual people market, and the market is growing. Among them, the virtual client usually takes the form of a Vtuber virtual anchor, and the main reason is the low starting point, and the experience is simple and fast. In 2016, a virtual anchor "Jian Ai" was launched on YouTube and gradually began to grow. From January 2020 to June this year, the number of virtual anchors in Bilibili increased almost seven times. The catalyst is usually due to the following two factors: the impact of the spread: all growth in online business is new business. , need more. Leaving virtual agent IP: Leaving virtual agent idol group 'Holive'. Due to the insufficiency of the IP head, fans, industry and investors have been eager to find new constraints, offering some improvements to the new virtual anchors.

According to the revenue of the Bilibili virtual anchor in November 2021, the monthly total revenue of the virtual anchor reached 54.66 million yuan, and the revenue reached 255,000. The "Gircarol" virtual anchor, which is the first , earns 2.14 million yuan per month. Of these, the birthday party, which lasted 4 hours, garnered a total of 1,899,000 wins.

Virtual idols are not based on surreal virtual human creation technology, but high-precision virtual human production technology gives virtual idols a new way of working. With the advent of technology, surreal virtual idols such as Ayai and Michela, who find it difficult to distinguish between real people, may appear, and there may be many new developments in these graphic designs and marketing, and they can serve as beauty bloggers. , model, etc. The first virtual idols were usually two-dimensional figures, and the prototypes were music, animation, CG, etc. The content of these virtual idols is usually video entertainment.

Metaverse - Turning virtual people into "souls"

1. DAO will support community development.

The production and use of digital content is inextricably linked to the community. The famous "Hatsune Miku" is a classic example of community-driven, self-created content. The history of "Hatsune Miku" dates back to 2007. Music software developer CRYPTON created a Vocalid sound library that creates "Hatsune Miku" and uses a special script. As of September 10, 2007, it accounted for 30.4% of the Japanese music software market, four times the second place. The community is directly involved in creating value and sharing and publishing it online.

Many creators have presented their music as an embodiment of "Hatsune Miku", but now they are faced with the problem of losing the creator. 'Hatsune Miku' has written a lot of content in a variety of genres, and therefore, the musicians who are the service providers continue to take care of it, creating a win-win situation and a good circle. This is also one of the reasons for the long life of "Hatsune Miku". However, due to the recent retirement of the P-Master (founder) of Hatsune Miku, good content has been reduced and the impact of intellectual property has been threatened. The main reason is that there is no financial donation for the development of P projects for the community, and it is difficult to generate electricity with love.

The virtual idol ecosystem developed by the UGC community will partner with DAO to create new business models, allowing participants to participate in the development of intellectual property. Decentralized Autonomous Organization (DAO) is a form of global digital organization based on blockchain technology, in which organizational policies are drawn from a decentralized program to the satisfaction of participants and to achieve common goals. DAO has several characteristics such as transparency, token incentives, open source code, community autonomy, member ownership of the organization, independence and openness. The DAO-based token industry allows developers and fans to benefit from enhanced IP ecosystems.

(For more detailed information on DAO, see the previous report "Metaverse (VI): DAO in Metaverse Operation")

2. Virtual avatar and digital identity

Web3.0 has the characteristics of trust and distribution, and users have the right to manage their privacy. To use Web 3.0, a variety of technologies such as blockchain, intelligence, and the Internet of Things are needed. Among them, blockchain technology in particular is based on the vision of Web 3.0, also known as blockchain. One of the reasons for the “Internet of values”.

Decentralized digital privacy is at the heart of Web 3.0. The most important feature of Web 3.0 is the management by users of their personal information, the owner being able to access their personal information whenever they wish, without resorting to specialized service providers. At the same time, with the help of blockchain technology, it is possible to increase the security, independence and mobility of user IDs. For example, decentralized records and NFT avatars, digital identifiers, are in the hands of users and can be used throughout the project.

As a symbol of digital identity, virtual symbols can be recognized and used on many platforms in the future. Currently, Twitter, Discord, Xiaohongshu, etc. all support the NFT digital collection recognition function, allowing unique virtual avatars to be exchanged across multiple platforms and become the real characters of the digital world. Based on 3D model data of virtual characters, it should be open to many future project platforms to achieve true digital identity in the metaverse.

Over the next five years, we hope to see more and more fake characters appear, not only in game NPCs (non-players), but also with more information in the metaverse. It works with NFT and AiGC (AI design themes), and the working model has also been modified by professional teams in the community. At the same time, virtual humans have higher demands on computing power, and the inclusion (including counting, communication, storage, etc.) of events has become increasingly important.