No collapse, no rest, and low cost, can digital humans subvert live streaming e-commerce?

In the future, what role will digital people play in the e-commerce industry?

This year, digital people have become completely popular in the live streaming e-commerce industry.

With the large-scale application of digital human anchors, a "hundred-model war" has been staged in China, and digital humans with a price as low as one hundred yuan have become a hot business. On the other hand, the other side of digital humans is the uncertain effect of use, the platform's increasingly strengthened supervision, the followers who are cut off, and the market's doubts.

Why has digital human become a trend? What is the future of digital human development? Where will the digital human industry develop in the future?

1. Digital Human Enthusiasm and Diversified Presentation

This year, digital people have emerged on short video platforms.

The virtual characters in short videos often appear in an exquisite virtual scene, building a cool content universe and exploring new ways to present short video content through technology. The most representative one is the beauty and makeup expert "Liu Yexi" who can catch monsters. One of her videos contains a variety of popular content categories such as suspense, plot, beauty and post-production special effects. Various secondary creations such as imitation makeup and cosplay around Liu Yexi exploded on the platform, and she once became the most talked-about virtual character at the time.

At present, Douyin and Kuaishou are the platforms with the most digital human applications, but the two platforms have different attitudes towards digital humans.

Douyin has some control over digital humans. According to the "Platform Specifications and Industry Initiatives on AI-Generated Content" released by Douyin, the use of AI technology to assist in creation is permitted, does not violate platform specifications, and does not restrict virtual human live broadcast technology. Kuaishou currently has no official statement on digital humans, so many digital human suppliers will focus on the Douyin live broadcast track.

From the perspective of content, the field of digital human live streaming is divided into two directions, one is entertainment live streaming, and the other is e-commerce live streaming. E-commerce live streaming is further divided into local life live streaming and regular live streaming with goods depending on whether express delivery is sent.

The carrier behind entertainment live broadcast is the emotional connection between people. At present, it is difficult for 2D digital human anchors to meet this demand, but 3D virtual digital humans have appeared in cases such as A-SOUL. In the field of e-commerce live broadcast, local life live broadcast rooms are more electronic redemption coupons, and users place orders in the live broadcast room and then go to the store to consume. The live broadcast of goods mainly focuses on the live broadcast room of popular products with a relatively single category.

At present, the driving modes of digital human anchors are mainly divided into "pure intelligent driving mode" and "semi-real driving mode". "Pure intelligent driving mode" mainly relies on keyword awakening and online Q&A to achieve user interaction. For common high-frequency problems in live broadcasts, keyword awakening does not require the digital human anchor to reply, and user problems can be solved. Online Q&A is to answer questions after the digital human product explanation is finished. Although there is a certain delay, it does not have much impact on user experience.

The "semi-real person driven mode" combines virtual human technology with real people to achieve interaction. The front end of the semi-real person driven live broadcast room is a virtual image, and the person behind it is the real person. It can achieve real-time online interaction, which is closer to the viewing experience of a real person live broadcast room. In addition, another form of semi-real person driven live broadcast room is to use a central control, and the operator controls the live broadcast room from behind.

For example, last year, Hu Bing staged a virtual fashion show on Douyin with his own virtual image, continuously showing the selling points and features of the clothing in constantly changing scenes, and through in-depth involvement, presenting an immersive live broadcast of selling goods.

Image source: Tik Tok screenshot

It is worth mentioning that according to the current development of technology, digital human live broadcast rooms can achieve real-time interaction without real-person assistance by calling large language models, but due to platform rules, this type of live broadcast room is not allowed.

In the field of short videos, the application scenarios of digital humans are relatively wide. For example, in the field of TOG, some state-owned enterprises and central enterprises have a large amount of content that needs to be promoted through videos, but in fact, many places are prone to repetitive work. If digital humans are used, there is no need to use a site construction team, nor do they need photographers, editors, directors and other personnel. When modeling, a large amount of content can be output in batches to improve work efficiency.

Digital people are also suitable for bloggers with personal IP. After the blogger copies the digital clone, he can quickly follow the hot spots without being restricted by the shooting environment.

For the two scenes of live broadcast and short video, 2D digital people have different charging standards. Xiong Da, the founder of Yinyi Future, said, "The usage fee of digital people used in short video scenes by our company is 8,000 yuan per year, which includes real-life image cloning and one year of usage rights." However, the digital people in Yinyi Technology's video scenes are charged by time. There are 500 minutes of free time in the first half of the year, and 20 yuan per minute in the second half of the year.

In the live broadcast scenario, the cost of using digital humans has increased. Generally speaking, the anchor images of digital human suppliers are divided into public and private libraries. If the template anchor uses the public library, it costs 60,000 yuan a year. However, the anchor in the public library will be used by multiple customers. Once the platform detects that the same faceID appears in different cities or live broadcast rooms, the digital human live broadcast room may face blocking or flow restrictions.

Image source: Tik Tok screenshot

In order to avoid the risk of having similar faces to other anchors in the live broadcast room, many customers will spend 8,000 yuan to clone their images. This belongs to the customer's private library and cannot be called by other customers. The annual usage fee is basically 68,000 yuan.

2. Cost reduction is the core reason for institutions to enter the market

Cost reduction is one of the most significant values ​​of digital humans. The application of digital humans first saves personnel costs and secondly saves part of the training costs.

For example, in Hangzhou, a live streamer’s hourly fee is 300 to 500 yuan. Adding the operation and site control, a single live studio costs at least 50,000 yuan a month. If you include the cost of purchasing lighting, cameras and other equipment, renting venues and scenery, and other materials, the cost of a live studio will be even higher.

The digital human live broadcast room has greatly reduced the cost. The digital human supplier outputs a signal that includes the scene, anchor, and speech. The live broadcast room captures the signal through a window and pushes it to platforms such as Douyin, Kuaishou, or Taobao using OBS, and the live broadcast can start immediately. The monthly price is only about 2,000 yuan.

In terms of live broadcasting capabilities, digital humans can achieve performances that exceed the average level of real people. However, the prerequisite is that high-quality computing power is used as technical support, audio and video can be synchronized, and they can interact with users. In addition, the role of operations is also very critical, and the art of speech is also a factor that affects the live broadcasting effect of digital humans. High-quality computing power plus mature operations can form a high-quality live broadcast room.

Liu Yang, vice president of Luban No. 7, said in an interview: "Compared with real-life anchors, the cost of a digital human live broadcast room is only 1/25 of that of a real-life anchor. In essence, as long as the effect of a digital human live broadcast room can reach 1/25 of that of a real-life anchor, it is not bad." And actual tests have shown that the effect of digital humans can reach 1/3 or 1/2 of that of real-life anchors, or even better than real-life anchors. This is an opportunity at the moment. It solves the problem of businesses being unable to broadcast or broadcasting poorly at a lower cost.

It is worth mentioning that Luban No. 7 has also set up its own local life live broadcast room, with daily sales reaching tens of thousands of yuan. Although the write-off rate is only 20 to 30 percent, multiplied by the commission of about 15%, a live broadcast room can also earn thousands of yuan a day, and a monthly income of 30,000 yuan. In fact, its investment cost is only the usage fee of a computer and software. Overall, the digital human has a high cost-effectiveness.

At present, digital people are mainly engaged in the role of brand advertisers. For the demand for digital people to bring goods through live broadcast, brand advertisers have stronger purchasing power, higher acceptance, and lower communication costs. For some brands, live broadcasting has become a daily sales channel. There is no need to enter the live broadcast room of the top anchors. Ordinary anchors can meet the needs of brands, and digital people can play a role as a substitute.

The value of digital humans is also reflected in the enhancement of anchor careers. For top anchors, digital avatars can fill in the anchor's rest time and further improve time utilization.

With the support of digital human technology, the scenes in the live broadcast room can have greater imagination space. Usually, due to physical conditions, the space that the anchor can display is relatively limited, and the scenes are relatively simple. However, the digital human live broadcast room can switch between near and far scenes at will, and the scenes are more diversified, giving users an immersive viewing experience.

In the short video scene, digital humans can greatly save time costs, and its tool value has been verified. Liu Run's video account and Focus Media's Jiang Nanchun have already used digital human technology. Compared with real-life recordings, the creation time of a single short video by digital humans can be reduced by about an hour. In terms of restoration, after sound cloning training, you only need to enter text to replicate the voice of a real person. It is worth mentioning that at present, when making digital avatar videos, real people need to record a short audio clip to achieve better audio and video synchronization.

3. The balance between virtual and real is still not balanced

Although the value of digital humans is being verified, the impact of digital humans on all industry participants cannot be ignored.

From the perspective of the anchor, in the live streaming industry, the anchor is in a relatively advantageous position. Regardless of the sales volume of the live streaming room, the anchor's salary is guaranteed to a certain extent. If the digital human copies the anchor's image, the cost can be reduced for the company, but the anchor may face the risk of reduced income or unemployment. Therefore, mature anchors will not actively participate in modeling.

Specifically, for top anchors, although digital avatars can continue to occupy the user's attention and give full play to the anchor's personal influence, under the trend of removing top anchors, top anchors generally strengthen the influence of the live broadcast room by cultivating assistants to avoid the risks brought by a single anchor. For mid- and low-level anchors, industry insiders mentioned that "small anchors have just entered the industry and hope to achieve greater development through their own efforts. They are more careful about their reputation and basically do not use digital avatars."

However, in the field of live streaming, some live streaming rooms have authorized celebrity portraits to sell goods. Some netizens found that in the live streaming rooms authorized by celebrities such as Calvin Chen and Zhu Zixiao, the celebrities themselves did not appear on screen, but the digital images of the celebrities were used for display, which also caused dissatisfaction among many netizens. Some netizens said that "it is relatively easy for celebrities to make money, and this is equivalent to lying down and counting money, which is really unacceptable." It can be seen that the application value of digital avatars is feasible in theory, but it can only play a practical role after being tested by the market.

Image source: Tik Tok screenshot

From the perspective of the platform, once digital human anchors are used on a large scale, the quality of the platform's content will be greatly reduced. The style of digital human live broadcast rooms is stereotyped, and even many anchors have the same image, making it difficult to guarantee the quality of the platform's content. If the platform's content is not attractive, it will cause user loss, and both the platform's advertising and e-commerce business will be affected, and revenue will be even more difficult to guarantee. Therefore, short video platforms will impose certain restrictions on digital human live broadcast rooms, mainly targeting low-quality live broadcast content. When the live broadcast level of digital humans is below the average level of natural humans, it is easy to touch the platform's bottom line.

From a technical perspective, the production of digital humans is not difficult in the industry . Many large companies at home and abroad have open source codes for digital human models, and digital human suppliers only need API interfaces to use them. In order to enhance its own competitiveness, the industry has begun to move towards 4K resolution to achieve more realistic effects.

However, it is currently difficult for digital humans to establish a strong emotional connection with users. This is mainly because the interaction between digital human hosts and users is one-way. Although digital humans can answer users' questions, such responses are mechanical. Digital humans cannot think and ask questions from the user's perspective, so it is difficult to generate sustained interaction.

From the perspective of the industry, some industry insiders said that some digital human manufacturers are not serious about making products and may be suspected of cutting leeks. They take advantage of some people's knowledge blind spots, package the digital human business as a wealth code, and then use the form of developing agents to collect agency fees of 500,000 or even 1 million from "trend chasers" and sell digital human products to some people who have never engaged in short video production and live broadcasting. In essence, it is a game of "pass the parcel".

After paying, many agents could not find customers, or their customers had no experience in live streaming and found it difficult to apply, and they asked the agents for a refund. However, the agents found that the digital human business was not that easy, and they thought they were deceived by the digital human manufacturers, so they took the path of defending their rights, which also tarnished the industry.

In fact, digital people seem to be low-cost and do not require too many resources, but they still require a high threshold. Digital people's live broadcast rooms also require the matching of "people, goods, and venues", and there are various ways to play. For people without industry experience, it is relatively difficult to create an account through a live broadcast room. If there is no traffic, the existence of the live broadcast room is meaningless.

In addition, there are no popular digital human live broadcast rooms on the market. The number of people online in many digital human live broadcast rooms may always be around ten, which is not conducive to the long-term development of the industry. Because digital human manufacturers need to have high-quality cases as support when expanding the market, and customers will be skeptical of digital humans if they cannot see the expected results. Some customers even said that they do not expect to make much money, and whether they can recover the investment cost is the first consideration. The difficulties in promoting digital humans have greatly affected the development of the industry.

An industry insider revealed that "many digital human suppliers believe that the market is mature and have expanded their teams to prepare for a big fight, but the reality is that they are faced with many obstacles and have to make strategic cuts. The team size of a leading digital human supplier has been reduced from more than 600 to more than 300."

More service providers are needed between digital human suppliers and customers. Whether they provide agency operations or training services, service providers can solve difficulties in application and are also a key link in driving the industry forward.

4. Rules and policies still need to be formulated by multiple parties

The application of digital humans is an increasingly obvious trend. On the industry side, more and more stakeholders are involved, and digital humans are becoming more diversified. For example, Silicon Intelligence and Qianxun have established a joint venture. Silicon Intelligence, as a technology provider, and Qianxun, as a live e-commerce operation service provider, have to some extent solved the problem of the separation between digital human technology and actual operations.

Silicon Intelligence has also established a joint venture with Lehua to engage in celebrity virtual image licensing business and invest in the field of live streaming. At present, some artists under Lehua have tried the water. You can also see some live streaming rooms authorized by celebrity virtual images on Douyin.

In the field of 3D digital humans, Alibaba Damo Academy and JD.com have mature technologies and are more focused on local life and cultural tourism. In the field of 2D real people, silicon-based intelligence is the leader, with more applications in food and life categories.

The biggest challenge facing digital humans is the rules of the platform . For example, platforms like Douyin, Taobao , and Video Account have relatively more rules and constraints, which actually limits the exploration of the value of digital humans. For example, virtual human live broadcast rooms must be officially whitelisted, otherwise they will most likely be restricted or even banned. This is also a problem that many uneven digital human suppliers in the market will encounter.

Although digital people are "anchors", their nature is completely different from that of anchors. The essence of digital people is SaaS software, and the standard of evaluation is the ability of digital human suppliers to provide stable services, and they are not responsible for sales. The evaluation of anchors is closely related to indicators such as sales volume, sales revenue, and audience views.

And digital humans cannot solve the problem of the entire link of live broadcasting. Because the "people, goods, and venue" in the live broadcasting room is a very complex scene, digital humans only solved the human part this year, the goods tray and the operation strategy of the live broadcasting room. Each team needs to show their own skills and solve their own problems.

In general, digital humans seem to be very popular, but the application scenarios are relatively fixed, the demand market is still immature, and the industry is still in its initial development stage. In addition, digital human anchors are detached from real life, have no background support and distinct personality characteristics, and it is difficult to establish a trusting relationship with users. It is difficult for the live broadcast room to stand out and expand its influence. In the future, can digital humans make innovative breakthroughs in interaction? What role will they eventually play in the short video and live broadcast industry? These are all worth looking forward to.

Author: Aries, WeChat public account: TopKlout (ID: TopKlout)