摘要
近年来,生成式人工智能技术在各领域应用广泛,成效显著,推动内容生成从2D静态图像、3D静态资产演进至视频序列,4D生成作为新兴研究领域同步快速发展。4D生成技术核心聚焦于基于文本、图像、视频、3D模型等多模态用户输入,创建具备时空一致性的动态3D资产,其凭借更高的创作自由度与沉浸式体验优势,已成为人工智能与计算机图形学领域极具潜力的新兴研究前沿。本文系统综述4D生成技术的研究进展:首先梳理其发展轨迹与核心定义,明确其相较于静态内容生成技术的突破与创新;随后系统分类阐述4D表示模型、生成框架及面临的核心挑战;在此基础上,阐述当前4D生成领域的主流数据集与核心评价指标;最后,展望其未来研究方向及应用前景。
In recent years,generative artificial intelligence technology has achieved remarkable results in various fields,driving the evolution of content generation from 2D static images and 3D static assets to video sequences.As an emerging research field,4D generation has also developed rapidly.The core of 4D generation technology focuses on creating dynamic 3D assets with spatiotemporal consistency based on multimodal user inputs such as text,images,videos,and 3D models.With its advantages of higher creative freedom and immersive experience,it has become a promising emerging research frontier in the fields of artificial intelligence and computer graphics.This paper systematically reviewed the research progress of 4D generation technology:firstly,4D development trajectory and core definitions were outlined,clarifying its breakthroughs and innovations compared to static content generation technology;then,4D representation models,generation frameworks,and core challenges faced were systematically categorized and elaborated on;on this basis,the mainstream datasets and core evaluation metrics in the current 4D generation field were discussed;finally,its future research directions and application prospects were looked forward to.
作者
张若怡
张韬政
ZHANG Ruoyi;ZHANG Taozheng(School of Information and Communication Engineering,Communication University of China,Beijing 100024,China)
出处
《信息传播研究》
2026年第1期47-57,共11页
Information and Communication Research
关键词
人工智能
计算机视觉
4D生成
artificial intelligence
computer vision
4D generation
