Answers are grounded in this paper and cite their sources.
3D rendering and visualization have become integral components of numerous applications. However, rendering large-scale oblique photogrammetry models on terminals with varying GPU capabilities poses significant challenges. One of the primary issues is the efficient allocation of GPU DRAM to handle the rendering workload. Traditional methods often struggle to balance memory consumption, leading to performance bottlenecks and inefficient resource utilization. In this paper, we propose a novel approach to handle this challenge. Our method focuses on optimized 3D tile scheduling, specifically designed to adapt the GPU DRAM capacities of individual terminals. By canceling tile requests that exceed the maximum rendering memory allocation, we prevent unnecessary accumulation of the tiles within the cache, thereby maintains the valuable memory resources. The experimental results show that the proposed method performs well in saving memory consumption and visualizing large-scale photogrammetry 3D models across different terminal capabilities.
No takes yet. Share an insight, caveat, or question.
Shang et al. (2023) studied this question.