A 28nm Physical-Based Ray-Tracing Rendering Processor for Photorealistic Augmented Reality with Inverse Rendering and Background Clustering for Mobile Devices

📋 论文概要

该论文提出了一款基于物理的光线追踪渲染处理器，针对增强现实中的逼真渲染需求。通过引入BBOX交点评估器和三角网格交点评估器，将复杂3D对象分段处理，实现了线性可扩展性，平均速度提升13倍，内存开销仅增加5.6%。

💡 主要创新点

• 提出将复杂3D对象分段处理的光线追踪架构，通过BBOX交点评估器和三角网格交点评估器协同工作，在不损失光线追踪效果的前提下提升效率。
• 实现了线性可扩展的渲染时间，相比基线设计平均获得13倍加速，且内存开销仅5.6%。

🏷 关键词

📄 原文摘要

University of Minnesota, Minneapolis, MN includes a sub-group of user-defined objects inside. After the BBOX Intersection Evaluator (BBIE) detects intersection with TBBOX, the Triangle Mesh Intersection Evaluator (TIE) computes the triangle intersection, and the result is sent for shading computing. With this scheme, complex 3D objects could be segmented for RT processing without losing the ray-tracing effect. As a result, a linear scalability in RT rendering time and an average of 13× speed up with only 5.6% memory overhead is achieved compared with the baseline design. 1 2 As the applications of Augmented Reality (AR) or Virtual Reality (VR) expand rapidly with the growing demands on enhanced visual realism, photorealistic image generation and insertion has become an essential feature for the emerging AR applications providing real-time workplace/household visual assistance. Physical Based Ray-Tracing (PBRT) is

👥 作者与机构

Shiyu Guo1, Sachin Sapatnekar2, Jie Gu1

Northwestern University, Evanston, IL