Digital Humans: Automatic character animation

数字人类：自动角色动画

• 批准号: 2446235
• 金额: --
• 依托单位: University of East Anglia
• 依托单位国家: 英国
• 项目类别: Studentship
• 财政年份: 2020
• 资助国家: 英国
• 起止时间: 2020 至 无数据
• 项目状态: 已结题
• 来源: https://gtr.ukri.org/projects?ref=studentship-2446235
• 关键词: Digital; Humans; Automatic; character; animation

Digital humans are a key component of animated television and film productions, and computer games. Traditionally, character animation is created offline by artists. However, interactive online computer gaming, and social virtual reality (VR) and augmented reality (AR) systems demand real-time and dynamically generated content. Automatically animated characters provide the feeling of natural communication on these platforms, leading to a more immersive experience. When combined with natural language processing and speech synthesisers, they can also perform autonomously as digital assistants or conversational agents.Historically, the field has been predominantly focused on animating the faces of digital characters along with speech. However, the non-verbal components of speech (face and body gesture) are key to both speech comprehension and for realistic character interaction. The main goal of the proposed project is to create a digital avatar that moves its body naturally and reacts appropriately to stimuli such as speech (speaking, laughing, coughing and listening), and changes in environment. Specifically, it will involve exploring the relationship between such stimuli and body motion. It will require the design, capture and analysis of a human motion dataset and the development of a suitable computational model for inferring body pose for changing inputs.

数字人是动画电视和电影制作以及电脑游戏的关键组成部分。传统上，角色动画是由艺术家离线创建的。然而，交互式在线计算机游戏以及社交虚拟现实（VR）和增强现实（AR）系统需要实时和动态生成的内容。自动动画角色在这些平台上提供自然交流的感觉，从而带来更身临其境的体验。当与自然语言处理和语音合成器相结合时，它们也可以自主地作为数字助理或会话代理执行。历史上，该领域主要集中在数字角色的面部动画沿着语音。然而，语音的非语言成分（面部和身体姿态）是语音理解和真实角色交互的关键。拟议项目的主要目标是创建一个数字化身，自然地移动其身体，并对语音（说话，大笑，咳嗽和倾听）等刺激和环境变化做出适当反应。具体来说，它将涉及探索这种刺激和身体运动之间的关系。这将需要设计、捕获和分析人体运动数据集，并开发一个合适的计算模型，用于推断身体姿势以改变输入。