HiGS achieves up to 15.8x faster real-time 3D Gaussian Splatting by running partitioning at coarse macro-tile scale and rasterization at fine tile scale, issuing work proportional to Gaussians per macro-tile.
arXiv preprint arXiv:2505.18764 , year=
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
fields
cs.CV 2years
2026 2verdicts
UNVERDICTED 2representative citing papers
VkSplat delivers the first fully Vulkan-based 3DGS training pipeline with 3.3x speed and 33% VRAM reduction over CUDA baselines while preserving quality.
citing papers explorer
-
HiGS: A Hierarchical Rendering Architecture for Real-Time 3D Gaussian Splatting
HiGS achieves up to 15.8x faster real-time 3D Gaussian Splatting by running partitioning at coarse macro-tile scale and rasterization at fine tile scale, issuing work proportional to Gaussians per macro-tile.
-
VkSplat: High-Performance 3DGS Training in Vulkan Compute
VkSplat delivers the first fully Vulkan-based 3DGS training pipeline with 3.3x speed and 33% VRAM reduction over CUDA baselines while preserving quality.