2020 Virtual Digital Human White Paper

This notice reports that the first DHDC Digital Human Developer Conference (数字人开发者大会 ) was held in Beijing on December 3, 2020, and that the event included the on-site release of the 2020 virtual digital human development white paper (2020 虚拟数字人白皮书), which was first published by the AIIA Alliance General Group and the ZAI Alliance Digital Human Working Committee and positioned as a status-and-outlook document that reviews key technologies, surveys industry application progress, forecasts future trends, and offers recommendations for sustaining sector development. In the same setting, Luster LightTech (凌云光技术) participated as a co-organizer, discussed topics such as frontier digital-human techniques, applications, and AI, and debuted AI Motion, described as an in-house, multi-view, intelligent, markerless multi-person motion-capture system that uses ordinary video imagery rather than wearable capture gear and applies multi-view 4D information with global optimization based on joint research with Tsinghua University; the company also framed its longer-term work as a decade-long investment in stereo vision and an end-to-end digital-human production solution line used in film, gaming, and live production contexts, while highlighting ecosystem gaps such as limited domestic open datasets and compute access for smaller developers and calling for safer, more stable open-source communities, standards, and open service platforms to support AI-plus-digital-human industrialization.

Figure 12: The virtual human industry chain can be divided from top to bottom into the application layer, the platform layer, and the foundation layer.

The application layer sits at the top and groups virtual-human deployments by industry scenario, showing where end users encounter virtual humans and what roles they perform in each sector. It lists film and television as a domain for digital doubles, media as a domain for virtual hosts/anchors, games as a domain for digital characters, finance as a domain for digital employees, and cultural tourism as a domain for virtual tour guides and virtual explanation staff. The layer is illustrated with example logos corresponding to these sectors, including CCTV (中央电视台) and SMG (上海广播电视台) in the media area, Tencent (腾讯) in the games area and again in cultural tourism, and Bank of China (中国银行) and China Construction Bank (中国建设银行) in the finance area.

The platform layer is the middle layer and is presented as the industrial production and capability layer that enables virtual-human creation, deployment, and scaling, divided into three horizontal capability groupings. The first grouping is software/hardware systems, which the diagram breaks into a modeling system and a motion-capture system, indicating toolchains for creating the character asset and capturing movement data; this grouping is illustrated with vendors such as Luster LightTech (凌云光), Artec 3D, and Xsens. The second grouping is production technology service platforms, split into a rendering platform and a solutions platform, indicating packaged, service-oriented platforms that provide rendering/production workflow infrastructure and verticalized solution delivery; this block is illustrated with logos including Baidu (百度), Tencent (腾讯), and SenseTime (商汤科技). The third grouping is AI capability platforms, broken into computer vision, intelligent speech, and natural language processing, indicating the perception and interaction stack needed for virtual humans to interpret inputs and generate speech/behavior; this grouping is illustrated primarily with iFlytek (科大讯飞) and Baidu (百度).

The foundation layer is the bottom layer and depicts the upstream enabling infrastructure split into hardware and software, emphasizing that virtual humans depend on a broad base of devices, components, compute, and core software tooling. On the hardware side, it explicitly lists display screens, AR glasses, VR headsets, optical components, sensors, and chips. On the software side, it explicitly lists modeling software and rendering engines. The layer is illustrated with major upstream vendors whose products map onto these categories, including Meta, Microsoft, Epson, Sony, and NVIDIA on the hardware/compute side, and Autodesk, Epic Games, and Unity on the software and engine side.