#来点儿干货#
The intelligent transformation brought by GPT is promoting the rapid development of the virtual digital human industry.
Enabled by AI technology, augmented digital virtual humans already have multi-modal interaction capabilities.
As a key entry point to the all-immersive interconnection, digital humans combine multiple AI technologies such as computer vision (CV), natural language processing (NLP), automatic speech recognition (ASR), text-to-speech (TTS), and interactive intelligence, and are promoting a more comprehensive integration of the online and offline worlds.
The virtual digital human industry in mainland China is showing an accelerated growth trend. According to qubit data, the market size of virtual digital humans in mainland China has exceeded 200 billion yuan and is expected to climb to 270.3 billion yuan in 2030. Among them, the personality virtual digital human is expected to occupy a dominant position in the future development, accounting for 64.60% of the market size in 2030, while the functional virtual digital human will account for 35.40%.
Pay attention to Leqing's industry observation and gain insight into the industrial pattern!
Overview of the virtual digital human industry
A virtual digital human is a virtual character that exists in the digital space and has the appearance, behavior, and characteristics of an anthropomorphic or real person.
Due to the highly anthropomorphic characteristics, users are more likely to have a sense of intimacy and trust in the virtual digital human, and recognize its role and positioning as part of the "human" in reality, rather than just treating it as general digital content. This allows virtual digital humans to take on more social functions, which is different from the general virtual role.
Virtual digital humans are generally composed of five modules: character image, voice generation, animation generation, audio and video synthesis display, and interaction.
Its core technologies can be summarized into three major processes: modeling, driving and rendering, mainly covering computer graphics, motion capture, image rendering and AI.
From the perspective of commercialization and application scenarios, virtual digital humans can be divided into three categories: content/IP type, functional service type, and virtual avatar type.
Content/IP-based virtual digital humans are mainly used in scenarios such as film and television, entertainment, and marketing, while functional services are mainly used in industry service scenarios, such as e-commerce, finance, education, medical care, and cultural tourism. These two types of virtual digital humans are more oriented to B-end users.
For 2B virtual digital humans, the key elements include scenario-based data, technical support, and business operations, and channel and scenario selection are the key to success, and the first-mover advantage is particularly important here.
In addition to B-end users, virtual avatar digital humans will also have the needs of C-end users to create personal avatars and agents in virtual spaces, such as on virtual social platforms.
For 2C virtual digital humans, content creation, technical support and business operation are the three core elements of identity-based virtual digital humans. Among them, content is the most critical factor that determines the success of a virtual digital human.
The development of the virtual digital human industry is closely related to the breakthrough and progress of the underlying technology. AI technology has reduced the production cost, shortened the cycle, and lowered the threshold for digital humans, while achieving a more similar appearance and action effect to real people.
According to the breakthrough of AI technology and its infrastructure capabilities, the development of virtual digital humans can be divided into four stages: embryonic stage, exploration stage, primary stage and growth stage.
At present, digital humans are developing in the direction of intelligence, convenience, refinement and diversification, and have entered a stage of accelerated growth.
At present, the mainstream method is to use NLP capabilities to drive text-driven, that is, to use AI technologies such as ASR-NLP-TTS to build a closed loop of perception-decision-expression to drive digital human interaction. At the same time, it is necessary to set up relevant knowledge graphs or Q&A libraries in advance to connect with the dialogue system of digital humans. However, the current capabilities of NLP in general scenarios need to be further improved.
With the continuous improvement of AI technology, the perception ability of digital humans (such as the development from the current text-based language understanding to multimodal input), thinking ability, and content output ability will be significantly improved, making them closer to real people in thought, language, and behavior, and achieving a higher degree of intelligence.
Sort out the virtual digital human industry chain
The industrial chain of virtual digital humans can be divided into three main levels: the basic layer, the platform layer and the application layer.
The base layer, also known as the tool layer, provides basic software and hardware support for virtual digital humans.
In terms of hardware, it includes display devices (such as mobile phones, TVs, projections, LED displays and other 2D devices, as well as glasses-free stereoscopic and AR/VR 3D equipment), optical devices (used to manufacture visual sensors and user displays), sensors (used to collect the original data of digital humans and users), and chips (used for sensor data preprocessing, digital human model rendering and AI computing).
The basic software mainly includes modeling software (used to model the human body and clothing of digital humans in 3D) and rendering engine (used to render lights, hair, clothing, etc.). Representative companies in the base layer include Meta, Filmtech, EPSON, NVIDIA, Unity, and Epic Gmaes.
The platform layer is a provider of virtual digital human solutions, covering software and hardware systems, production technology service platforms, and AI capability platforms.
This layer provides technical capabilities such as modeling, motion capture, rendering, and voice, and provides comprehensive technical support for virtual digital humans. The main players include vendors, general/Internet technology vendors, specialty AI vendors, CG vendors, and XR vendors. Representative manufacturers include Vicon, Tencent, Baidu, Sogou, SenseTime, NetEase Fuxi, Creative Fantasy Technology and iFLYTEK.
The application layer is a virtual digital human combined with actual application scenarios to form industry application solutions and empower various industries. This layer is mainly responsible for creating and operating virtual digital human personas and applying them to various scenarios. The main layout manufacturers in the application layer include BlueFocus, Mango Supermedia, and Next World Culture.
Domestic related manufacturers are mainly involved in the platform layer (such as motion capture, artificial intelligence, etc.) and the application layer (such as virtual idols, virtual anchors, etc.).
There are many application scenarios of downstream virtual digital humans, and the market size is long-tailed. Among them, the main subdivision track is virtual idols, whose growth rate has always remained above 60% in recent years, and is still improving.
Virtual digital human industry chain and some representative manufacturers:
Source: Chinese Artificial Intelligence Industry Development Alliance
epilogue
At present, with the increasing maturity of virtual digital human theory and technology, its application scope is constantly expanding, and the industry is gradually forming and enriching. At the same time, the corresponding business model is constantly evolving and diversifying.
With the rapid development of multimodal AI models, digital human creation has entered the AIGC era, and the digital human industry has also entered a period of vigorous development. Combined with the research results of Tencent Research Institute, AIGC can not only achieve pipeline production of "good-looking" digital human images, but also continuously promote the development of digital humans in the direction of having an "interesting" soul, thereby significantly shortening the creation cycle of digital humans. #Virtual Digital Human##Digital Human##Media##Game##Artificial Intelligence##Multimodality##Article Launch Challenge##财经新势力#
The emergence of virtual digital humans can not only greatly reduce the cost pressure of enterprises, improve the efficiency and coverage of live broadcasting, but also provide more personalized and unique live broadcast content to meet the diverse needs of enterprises and users. This will help drive the enterprise live streaming industry towards a more efficient, easy-to-use, sci-fi and immersive direction.
Pay attention to Leqing's industry observation and gain insight into the industrial pattern!