节点文献

超多视点显示中后向光线追踪的实时渲染方法

Real-time rendering of backward ray tracing in super multi-view display

  • 推荐 CAJ下载
  • PDF下载
  • 不支持迅雷等下载工具,请取消加速工具后下载。

【作者】 李婧楠魏嘉豪刘娟

【Author】 LI Jingnan;WEI Jiahao;LIU Juan;Beijing Engineering Research Center for Mixed Reality and New Display Technology, School of Optics and Photonics, Beijing Institute of Technology;

【通讯作者】 刘娟;

【机构】 北京理工大学光电学院北京市混合现实与新型显示工程技术研究中心

【摘要】 超多视点显示技术能够提供逼真的三维显示效果,但是大量视点图像的渲染合成过程使得三维显示的实时性受到极大的挑战。文章提出了一种超多视点显示中后向光线追踪的实时渲染方法,根据视点分配矩阵和子像素物理坐标,光线从视点位置出发逆向追踪到三维场景,并直接对视点合成图像进行子像素级别实时渲染。结果表明,在保证三维图像显示质量不变的情况下,随视点数目增加,渲染帧率始终保持在35FPS以上。该方法避免了渲染合成过程中的信息冗余,渲染速度相比传统的逐视点渲染方法提升了5个数量级。该方法为实现超多视点动态三维显示提供了新的途径。

【Abstract】 Super multi-view(SMV) three-dimensional(3D) display can offer immersive 3D viewing effects, but the rendering and synthesis process of massive viewpoint images poses significant challenges to real-time performance. In this paper, a real-time rendering method of backward ray tracing in SMV display is proposed. According to viewpoint allocation matrices and sub-pixel physical coordinates, rays are traced backward from viewpoint positions to 3D scenes, enabling real-time rendering directly on the synthetic image at the sub-pixel level. The results indicate that while maintaining consistent display quality, the rendering frame rate remains above 35 FPS even with an increasing number of viewpoints. The proposed method avoids information redundancy in the rendering synthesis process, achieving a rendering speed improvement of 5 orders of magnitude compared to traditional sequential viewpoint rendering methods. Thus, it provides a promising avenue for achieving super multi-view dynamic 3D display.

【基金】 国家重点研发计划(2021YFB2802301);国家自然科学基金项目(61975014&62035003&U22A2079)
  • 【分类号】TP391.41
  • 【下载频次】5
节点文献中: 

本文链接的文献网络图示:

本文的引文网络