What can AIGC do for 'people' in the metaverse? ｜No.1 AIGC Season②

Author: Yu Sheng

The era of AIGC has come, how far is the Metaverse?

Recommended reading:

AIGC 「Generation Power」｜No.1 AIGC Season ①

The best way to overcome AI anxiety is to "join AIGC"｜AIGC Quarterly Issue No. 1

In 2023, AIGC will be "surge" all the way.

The AI ​​conversation model ChatGPT developed by OpenAI, an American artificial intelligence research company, ignited the spark of AIGC and set off a prairie fire in a short period of time. Subsequently, Baidu "Wen Xin Yiyan", SenseTime Technology "Japan" Various domestic large model products such as Nissin Sensenova have been launched one after another.

Not only that, AIGC also takes turns to introduce new products in various fields such as pictures, audio, and video. According to iiMedia Consulting data, the core market size of China's AIGC industry is expected to be 7.93 billion yuan in 2023 and will reach 276.74 billion yuan in 2028.

In comparison, the concept of "metaverse" that has frequently appeared in the public eye seems to have been left out. But in fact, AIGC and the Metaverse complement each other, and there is no trade-off between the two.

So, what kind of technical support can AIGC bring to the Metaverse? Can the current sense of "industry tremor" brought by AIGC be sustained, and can it be converted into an empowering effect in the construction of Metaverse content and application scenarios? How far are we from the metaverse?

With these questions, No. 1 interviewed Jing Maosen, director of the marketing department of Guangzhou Virtual Film Co., Ltd., and Jiang Yahong, founder and CEO of Hangzhou Youlian Times Co., Ltd., trying to talk about it from the perspective of "people" in the metaverse. Let’s talk about what AIGC can do for “people” in the metaverse.

Key breakthrough: Injecting soul into AI virtual humans

"Virtual humans are a very important concept in the future metaverse."

According to Jing Maosen of Virtual Cinema, whether they are digital avatars driven by real people or so-called NPCs in the metaverse, they are indispensable existences in the metaverse. However, in order to create a sufficient number of virtual humans for the normal operation of the metaverse, it is inevitably impossible to have all of them driven by real people.

The importance of AI virtual humans is self-evident.

Since 2018, many major Internet companies and media companies such as Tencent, Baidu, and Alibaba have launched digital virtual human services. Xinhua News Agency and Tencent's "Xiao Zheng", Alibaba's "AYAYI", Zhejiang Satellite TV's "Gu Xiaoyu", Mango Super Media "YAOYAO" and other virtual digital users are countless.

But the fatal weakness of AI virtual humans is that they have no soul.

In other words, most of our understanding of AI virtual humans currently on the market is still that it is a soulless AI robot that cannot understand human expressions, rather than a virtual person that can truly communicate with it. people. Jing Maosen also said that the current AI-driven virtual humans do have problems such as stiff movements, insufficiently agile expressions, emotionless voices, and low feedback efficiency, which to a certain extent limits the development of the virtual human industry.

However, after ChatGPT became popular, many people in the industry began to think that "ChatGPT injects soul into AI virtual humans."

Take GPT-4 as an example. As a large multi-modal pre-trained model, it can accept image and text input at the same time and give corresponding answers accordingly. Compared with the first generation of ChatGPT, GPT-4's problem-solving and communication skills have been significantly improved.

So, is the idea of ​​"ChatGPT injecting soul into AI virtual humans" feasible?

Jing Maosen believes that this path is possible.

Based on ChatGPT’s pre-training model and powerful knowledge base, virtual humans can quickly retrieve relevant information in the database after receiving corresponding instructions and give corresponding replies in a short time, thereby realizing ChatGPT and virtual humans The complementary advantages between them make the interaction between virtual people closer to the daily communication and expression state of humans.

On February 1st, domestic virtual technology service provider Shiyou Technology announced that its digital human business has used ChatGPT, the AI ​​"brain", and is using the digital human's own human background and other related data sets, and based on OpenAI to There is brain formation personalized model training. In addition, Yuanjing Technology, Cape Cloud and other companies also stated that the company's digital human-related business has been connected to ChatGPT to enhance and strengthen virtual digital human-related business capabilities.

On this basis, Jing Maosen predicts that the virtual human industry will usher in a new round of innovation and upgrades in the future.

Specifically, in the future, the virtual human industry will develop in two different directions: "qualitative" and "quantitative".

First, some high-quality virtual humans will continue to be optimized and improved, developing in a high-precision direction. For example, Virtual Pictures has been working intensively on the image creation and model production of virtual humans for a long time. The virtual human "Crane Chase" it created is rooted in the film and television field. It has participated in many film and television animation works and has an online reach of over 100 million people. .

The second is that the number of functional and applied virtual humans driven by AI will increase significantly and be fully rolled out. "Large model products including ChatGPT, as well as related technologies such as AI mapping and AI modeling, will reduce the asset cost of the virtual human industry. Many small and medium-sized startups can also obtain a considerable degree of success in this field. If there are development opportunities, then the entire industry will become more prosperous."

Comprehensive Empowerment: Time Acceleration to Rebuild an Earth

From the perspective of the media level division of information dissemination, the main areas of industrial layout of AIGC-related companies are: text, pictures, audio and video.

But judging from the related industry fields involved in AIGC, AIGC has actually been embedded in various fields such as information, games, media and film and television creation, e-commerce, and financial consulting, and has a profound impact on all aspects of our daily lives.

Similarly, AIGC will also fully empower the construction of the metaverse.

Jiang Yahong, founder and CEO of Youlian Times, started from his own vision of the Metaverse and elaborated on AIGC’s empowerment in the space construction, content generation, and experience scenarios of the Metaverse.

He believes that when talking about how AIGC empowers the metaverse, the first thing to think about is how "people" in the metaverse live, work and consume in the virtual world. "In the Metaverse, whether we are working with colleagues, socializing or entertaining, we all need to have our own digital avatars and be able to experience various application scenarios of the Metaverse without being limited by space."

As the basic construction of the Metaverse, real-life 3D digital humans have very broad application prospects, including Metaverse conferences, cultural museums, cultural tourism, universities, offline exhibition halls, film and television, game entertainment, brand promotion, etc.

Nowadays, the Youlian era has actual products in the cultural and tourism scenes, film creation, game entertainment, brand promotion, offline exhibition halls, etc., especially in the real-time generation of digital avatars, the Youlian 3D cloud array camera. This is a commercial-grade smart device for creating real-life digital people in the Metaverse. It can shoot and create in one second, and generate a real-life 3D digital avatar in as fast as 5 minutes. The cost is only 100 yuan, achieving a "consumer-level" breakthrough in the way of creating digital avatars. Jiang Yahong said that the development of AIGC will bring new opportunities in terms of the accuracy of real-time generative digital avatars and the development of application scenarios.

Specifically speaking at the level of "people" themselves in the metaverse, in addition to equipping virtual people with "brains", AIGC can also greatly improve the image drawing, model generation and construction of virtual people. Production efficiency can also bring about qualitative development in the flexibility and authenticity of virtual human expressions and movements, as well as the anthropomorphism of sound output.

For example, Sun Zhipeng, senior manager of the international 3D engine giant Unity China and head of cross-terminal transplantation technology, said in an interview with a reporter from the "Daily Economic News" that corresponding to AI painting, the 3D engine may realize "one-sentence modeling."

For another example, Jing Maosen specifically mentioned AI motion capture technology in the interview.

The investment and maintenance costs of the virtual human industry in the field of motion capture have always been very high. “Just building an optical motion capture studio requires an investment of several million, which is very difficult for many start-up companies. It’s a very high investment cost.”

The AI ​​motion capture technology can accurately identify and reproduce the movements of the characters in the video based on a captured video, and automatically generate the skeletal movement data of the virtual human. On this basis, by assigning this data to the 3D model of the virtual human, the virtual human's action drive can be completed.

In this process, neither expensive professional motion capture equipment nor specialized personnel are required to wear motion capture equipment to drive virtual humans. This reduces the cost of motion capture while improving the efficiency of motion capture, killing two birds with one stone.

Looking at it this way, AIGC’s empowerment of “people” in the metaverse is comprehensive.

On the one hand, AIGC can provide solid technical support for the scene construction of the metaverse and open up new space for various activities of "people" in the metaverse; on the other hand, the application of AIGC itself in the field of virtual human production It will also reduce the production cost of virtual humans, giving more people the opportunity to have their own digital avatars in the Metaverse.

In the end, the time to achieve the ultimate form of the Yuan Universe was accelerated.

Jing Maosen mentioned that for the Metaverse to be implemented in life, it is actually equivalent to recreating an earth in the virtual world, which requires a huge amount of engineering and assets. In this process, if there is AI assistance, the time to rebuild the earth will be faster.

Return to reality: the distance between us and the metaverse

"The era of AIGC is coming."

Jing Maosen said frankly that this was his first impression after experiencing the high flexibility, high accuracy and high feedback efficiency of ChatGPT.

There are rampant arguments such as ChatGPT will replace 80% of people’s jobs, and AI painting will replace the positions of middle and low-level original painters. At the same time, new positions such as ChatGPT researchers and algorithm engineers are being created, placing higher requirements on people's ability to use computer technology.

「"Space Opera" drawn by AI」

This is not a commonplace statement.

Generative AI integrates computer vision, data mining, machine learning, intelligent voice technology, natural language processing, knowledge graph and other core AI technologies, and can play a role in creativity, expression, iteration, communication, personalization, etc. Significant advantages. However, No. 1 learned during the interview that there was a contradiction between "ideal" and "reality" in the actual implementation of AIGC.

In July 2022, Baidu CEO Robin Li judged at the 2022 Baidu World Conference that AIGC will go through three stages of development: the first is the "assistant stage", where AIGC is used to assist humans in content production; the second is the "collaboration stage" ”, AIGC appears in the form of a virtual human that coexists virtuality and reality, forming a situation of human-machine symbiosis; the third is the “original stage”, in which AIGC will independently complete content creation.

Now, we are in the intertwining period of "collaboration stage" and "original stage".

For example, in the virtual human industry, the problem of "virtual humans without souls" can be solved to a certain extent by accessing large models such as ChatGPT. However, in the actual operation and application implementation process, natural language processing and conversion also need to be considered. , problems such as insufficient information feedback efficiency.

Only when the virtual person has sufficient authenticity and vividness, can the maximum value of the virtual person be brought out.

"Now everyone's expectations for AI are not that high." Jing Maosen analyzed that most people view the current AIGC technology with an experiential mentality, but in the real implementation stage of AIGC-related applications, those AI The hand that fails to draw well and the human needs that ChatGPT fails to understand are the key to determining the future development of AIGC.

It can be said that the current popularity of AIGC is just a prelude before the Metaverse has yet to enter a truly prosperous stage. We are still a long way from the envisioned "ultimate form" of the Metaverse.

Jiang Yahong also said that the current Metaverse is still in its infancy, and it will take at least three years for the actual implementation of Metaverse-related applications. He admitted frankly that the work of shooting and producing digital avatars that Youlian Times is engaged in is only part of the process of building the metaverse, but it is also a very important infrastructure of the metaverse. Real-person digital avatars have great prospects, and the market is waiting to explode.

In addition, another problem is that the related consumption and application scenarios of the Metaverse have not yet been fully opened. Take the virtual human industry as an example. The current application market for virtual humans is mainly in the media and entertainment fields. The hyper-realistic digital human "Mei Se Tian" created by Mandrill Pictures is mainly active in knowledge popularization, talk show performances, and fashion. Life, literary and artistic creation and other fields.

Most people only interact and communicate directly with virtual people as bystanders, not as participants or experiencers, so it is difficult to have a more intuitive observation and understanding of virtual people. Correspondingly, the same is true for the Metaverse. Only when enough people participate, the prototype of the Metaverse can be considered initially established.

In this regard, Jiang Yahong said that the core elements of the metaverse should include space, people, content and scenes. From a business perspective, the "people, goods, and places" in the metaverse should be able to be quickly reflected. . "On this basis, we can further unleash the economic value of the virtual human industry and reflect the true meaning of the Metaverse."

Conclusion No. 1

In October 2022, AIGC startup Jasper received US$125 million in Series A financing. It only took 18 months for Jasper to go from obscurity at the time of its birth to fame after becoming a unicorn company.

Including Jasper, there are countless companies that have taken advantage of AIGC to develop rapidly. It is foreseeable that with the massive influx of capital and the rapid expansion of market scale, the AIGC industry will usher in a new round of rapid development.

On April 19, scholar Yu Guoming gave a lecture titled "Metaverse, AIGC and Communication Revolution - From ChatGPT to the Future of the Comprehensive Intelligent Era", which systematically revealed the main focus of ChatGPT from 12 aspects. The AIGC is about to bring about a new era of intelligent interconnection, and the Metaverse is an inevitable product of the era of digital intelligence.

I have to admit that the popularity of AIGC has made us stop from fantasizing and discussing the floating metaverse, and we have begun to see the core technology engine that drives the development of the metaverse: AI and AIGC. Perhaps in the future, we can borrow the key of AIGC to truly open the door to the metaverse.

The above is the detailed content of What can AIGC do for 'people' in the metaverse? ｜No.1 AIGC Season②. For more information, please follow other related articles on the PHP Chinese website!