Journal of Frontiers of Computer Science and Technology ›› 2023, Vol. 17 ›› Issue (2): 263-278.DOI: 10.3778/j.issn.1673-9418.2207067

• Frontiers·Surveys • Previous Articles     Next Articles

Review of Real-Time Ray Tracing Technique Research

YAN Run, HUANG Libo, GUO Hui, WANG Yongxin, ZHANG Xincheng, ZHANG Hongru   

  1. School of Computer, National University of Defense Technology, Changsha 410073, China
  • Online:2023-02-01 Published:2023-02-01



  1. 国防科技大学 计算机学院,长沙 410073

Abstract: Ray tracing has been regarded as the next generation of mainstream image rendering technology for a long time because of the authenticity of its rendering effect, and it is a hot research point in the field of computer graphics. In recent years, academics and commercials have extensively researched real-time ray tracing. To promote the research of real-time ray tracing, this paper reviews, analyses, and summaries the related literature. Firstly, the concept, algorithms, and classification of acceleration structures are introduced. Three commercial graphics processing units (GPU) supporting ray tracing are introduced and the differences between them are compared. This paper summarizes the optimization of ray tracing from six aspects, ray packet, stackless traversal, ray reorder, wide BVH, denoising techniques, and real-time ray tracing combined with the artificial?neural network, and expounds on the advantages and disadvantages of the relevant specific methods. Based on the acceleration of the algorithms, the hardware acceleration method on GPU and the dedicated architectures are summarized. Finally, this paper makes a brief summary of the content, points out the difficulties that real-time ray tracing is still challenged, and looks forward to the future development direction. It can help researchers systematically understand real-time ray tracing status and provide follow-up research ideas.

Key words: rendering, ray tracing, graphics processing unit (GPU), domain-specific architecture

摘要: 光线追踪因其渲染效果的真实性,长期以来被视为下一代主流图像渲染技术,是计算机图形学领域的热点研究方向。近年来,学术界和商业界对实时光线追踪开展了广泛研究。为促进实时光线追踪的研究,对相关文献进行归纳、分析和总结。首先阐述了光线追踪的概念、算法、加速数据结构等理论知识;介绍了三款支持光线追踪商用图形处理器(GPU),并对比了之间的差异;从光线束遍历、无栈遍历、光线重排序、多分支BVH、降噪技术、与神经网络结合的实时光线追踪这六个方法综述了光线追踪的算法优化工作,并阐明了相关具体方法的优缺点;在算法加速的基础上,对使用GPU优化加速和采用定制化设计的硬件加速进行了归纳分析;最后对文章的内容进行了总结,指出了实时光线追踪仍面临的困难,并对未来的发展方向进行了展望。可以帮助研究人员系统地了解实时光线追踪的研究现状,为后续开展相关研究提供思路。

关键词: 渲染, 光线追踪, 图形处理器(GPU), 领域定制体系结构